2018 Book AdvancesInStructuralIntegrity

Raghu Prakash
Vikram Jayaram
Ashok Saxena Editors
Advances in
Structural Integrity
Proceedings of SICE 2016
Advances in Structural Integrity
Raghu Prakash Vikram Jayaram
•
Ashok Saxena
Editors
Advances in Structural
Integrity
Proceedings of SICE 2016
123
Editors
Raghu Prakash Ashok Saxena
Department of Mechanical Engineering Department of Mechanical Engineering
Indian Institute of Technology Madras University of Arkansas
Chennai, Tamil Nadu Fayetteville, AR
India USA
Vikram Jayaram
Department of Materials Engineering
Indian Institute of Science
Bangalore, Karnataka
India
ISBN 978-981-10-7196-6 ISBN 978-981-10-7197-3 (eBook)

https://doi.org/10.1007/978-981-10-7197-3
Library of Congress Control Number: 2017958620
© Springer Nature Singapore Pte Ltd. 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore
Foreword
I am delighted to note that the first International Structural Integrity Conference

and Exhibition (SICE 2016) organized by the Indian Structural Integrity Society
(InSIS) along with Indian Institute of Science, Bangalore and Indian Institute
of Technology Madras, Chennai had made a modest beginning by a young pro-
fessional Society. It was also convened as the Inter-Quadrennial conference of
International Congress on Fracture. It was a grand success with more than 130
technical papers and 6 keynote lectures been presented during the 3-day event. I am
delighted to release this Lecture Notes in Mechanical Engineering brought out by
Springer Publishing consisting of selected 57 papers that were chosen after peer
review by a panel of experts in the domain.
Structural integrity and its longevity studies in our country have found India-
centric as well as generic solutions that can be applied worldwide. The fact that
India provides engineering solutions to many multi-national companies dealing
with energy, transportation, refining, and materials development is a testimony to
this; there is always a quest for knowledge in this domain considering that India has
embarked on “Make in India” mission for major infrastructural projects like
advanced ultra-supercritical power plants. Apart from dealing with conventional
mechanics, material characterization, and defect analysis, the field of structural
integrity has expanded to structural health and condition monitoring, reliability and
risk analysis, etc., over the last three decades since the 6th International Congress
on Fracture conference held at New Delhi in 1984.
The conference proceedings covers many aspects of structural integrity—
Fatigue and Fracture Mechanics; Creep–Fatigue Interactions; Structural Health
Monitoring; Failure Analysis; Experimental Techniques; Computation Methods;
Metals, Composites, and New Generation Materials. Many novel concepts have
been applied to practical situations in diverse domains such as bioengineering,
aerospace structures, civil engineering structures, power plants, and transportation.
As a professional involved in this domain, I experience and find that the quality
of papers presented in this conference is of high caliber and would serve the
v
vi Foreword
immediate and future needs of researchers and engineers working in this domain.
I congratulate the Indian Structural Integrity Society, the organizers of this con-
ference, and the editors of this volume for establishing this tradition and wish many
more such events to happen in the near future for the purpose of pursuit of
knowledge dissemination and enhancing collaborations.
Bangalore, India Baldev Raj

Preface
This book contains selected papers from the presentations made at the first inter-
national Structural Integrity Conference and Exhibition (SICE 2016) held at Hotel
Le Meridien, Bangalore, during July 4–6, 2016. The conference, exhibition, and
pre-conference workshop were organized jointly by the Indian Structural Integrity
Society (InSIS), Indian Institute of Science, Bangalore, and Indian Institute of
Technology Madras, Chennai. The timing of the conference was particularly
appropriate, given the critical role of total structural integrity in the new manu-
facturing initiatives of the Government of India, such as the “Make in India”
program. Of the 134 oral and poster presentations, a total of 57 were finally selected
for publication after a peer review.
The conference attracted nearly 220 registrants from seven countries (India,
Japan, USA, Russia, Ukraine, Switzerland, and UK) apart from the 56 registrants
for the 2-day pre-conference workshop held during July 2–3, 2016 at IISc,
Bangalore, and BiSS Research, Bangalore. SICE2016 was also convened as the
Inter-Quadrennial Conference of the International Congress on Fracture (IQ-ICF).
The SICE 2016 conference focused on all aspects relating to Structural Integrity,
including Fatigue and Fracture Mechanics; Creep–Fatigue Interactions; Structural
Health Monitoring; Failure Analysis; Experimental Techniques; Computation
Methods; Metals, Composites, and New Generation Materials. The range of
applications covered bioengineering, aerospace structures, civil engineering struc-
tures, power plants, and transportation. The conference spanned 3 days with 6
plenary lectures, 23 invited lectures, and 93 contributed papers delivered by
researchers working in the domain of structural integrity in three parallel sessions.
There were 12 poster presentations, mostly from research scholars.
About 85 authors were invited to submit full-length manuscripts after the con-
ference deliberations, and the same was reviewed by a team of 60 domain experts
from various organizations. The high quality of the final selection of papers is due
to the hard work put in by all the reviewers and volunteers, and we acknowledge
their support. We hope this volume will serve as a good reference for practicing
researchers in the domain and as a good source of material for budding engineers
and scientists.
vii
viii Preface
The conference was sponsored by several agencies such as Department of

Science and Technology, Defence Research and Development Organization, and
Atomic Energy Regulatory Board of the Government of India; BiSS Research,
Bangalore; MTS Systems Corporation, USA, along with its local distributor Techno
Testing Instruments; ABS Systems, Chennai; Altair Corporation; Honeywell
Corporation; and Hysitron NanoTechnology India Pvt Ltd, Thiruvananthapuram.
Springer provided support in the form of merit prizes for the best poster papers
presented at the conference and is proud to be associated with bringing out this
collection of selected papers as a book in Lecture Notes in Mechanical Engineering.
This conference provided a platform for practicing scientists and engineers to
disseminate their knowledge and learn from one another’s experience. We hope that
this tradition will continue and that the Biennial SICE / IQ-ICF conferences will
play a key role in advancing knowledge in this critical domain of structural
integrity.
Chennai, India Raghu Prakash

Bangalore, India Vikram Jayaram
Fayetteville, USA Ashok Saxena
Contents
High-R Fatigue Crack Growth Threshold Stress Intensity

Factors at High Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Stuart R. Holdsworth and Zhen Chen
Part I Aerospace
Fracture Mechanics Analysis of Fuselage Structural Joints
with Cracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
H. Ramesha, N. C. Mahendra Babu and H. V. Lakshminarayana
Effect of Silica Nanoparticles on the Fatigue Life of a Glass Fiber
Reinforced Epoxy Composite Under an Aircraft Spectrum Load
Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
N. Jagannathan, K. Sakthivel, Ramesh Bojja and C. M. Manjunatha
Fatigue Life Estimation of Typical Fighter Aircraft Main
Landing Gear Using Finite Element Analysis . . . . . . . . . . . . . . . . . . . . . 39
T. Sivaranjani, D. V. T. G. Pavan Kumar, C. M. Manjunatha
and M. Manjuprasad
Designing for Strength: ASME’s DBA Guidelines and Beyond . . . . . . . 53
Kotur S. Raghavan
Premature Fatigue Failures in the Hot Zone of Low Pressure
Turbine Rotor (LPTR) Blades of Aero-Engine . . . . . . . . . . . . . . . . . . . . 65
M. Madan, R. Bharathanatha Reddy, K. Raghavendra, M. Sujata
and S. K. Bhaumik
Fatigue Life Prediction of Composite Airframe Panel . . . . . . . . . . . . . . 77
P. K. Sahoo, Shriram Gujar and M. Manjuprasad
Integrity Evaluation of Feature Level Test Specimen
of an Aircraft Primary Composite Structure . . . . . . . . . . . . . . . . . . . . . 93
S. Venkatesh, S. C. Lakshminarayana and Byji Varughese
ix
x Contents
A Study on the Effect of Crack Stopper for Enhanced Damage

Tolerance Behavior of a Fuselage Stiffened Panel . . . . . . . . . . . . . . . . . 103
G. Akshitha, A. Shailesh Rao, N. Srinivasan and M. Mohan Kumar
Failure Analysis of a Squirrel Cage Bearing of a Gas
Turbine Engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Swati Biswas, Jivan Kumar, V. N. Satishkumar
and S. N. Narendra Babu
Part II Civil Structures

On-line Deformation Monitoring of Building Estate Above Mining . . . . 129
R. Tsvetkov, Igor Shardakov, A. Shestakov and V. Yepin
Mechanical Properties of Nano Concrete . . . . . . . . . . . . . . . . . . . . . . . . 137
Ajaysinh R. Vaghela and Gaurang R. Vesmawala
Estimation of Critical Microcrack Length in Concrete by
Considering Interfacial Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Keerthy M. Simon and J. M. Chandra Kishen
Fracture Analysis and Remaining Life Assessment of
Ultra High Strength Concrete Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
A. Ramachandra Murthy, N. R. Iyer and B. K. Raghu Prasad
Part III Composites

Rupture in Abdominal Aortic Aneurysm . . . . . . . . . . . . . . . . . . . . . . . . 169
Qusay Alfaori, Ashok Saxena, Hanna Jensen and Morten Jensen
Damage Tolerance Assessment of Laminated Composite Plates
Subjected to Ballistic Impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
H. L. Vinayaka and Shivashankar R. Srivatsa
Prediction of Mode II Delamination Onset Life Under Spectrum
Fatigue Loads Using Equivalent Strain Energy Release
Rate Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
A. R. Anilchandra, M. Seshagirachari, Ramesh Bojja, N. Jagannathan
and C. M. Manjunatha
Dependence of Crack Velocity on Stress Intensity Factor
in PMMA Using Single-Edge-Notched Clamped Beams . . . . . . . . . . . . . 205
G. R. Rahul, V. Jayaram and S. Bose
Effect of Wrap Thickness and Ply Orientation on Cracking and
Failure of FRP-Wrapped Columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Pradeep B. Kodag and Gaurang R. Vesmawala
Contents xi
Studies on Fracture Features and Fracture Toughness of CFRP

Composites Fabricated by Resin Infusion Technique . . . . . . . . . . . . . . . 225
K. Panbarasu, M. Suresh Kumar, V. R. Ranganath and R. V. Prakash
Numerical and Experimental Analysis of Double-Sided Stepped
Lap-Repaired CFRP Laminates Under Tensile Loading . . . . . . . . . . . . 235
Matta Seshadri and M. Ramji
Part IV Computational Mechanics

Lower and Upper Bound Estimates of Material Properties
of Pristine Graphene: Using Quantum Espresso . . . . . . . . . . . . . . . . . . 253
T. Chaitanya Sagar and Viswanath Chinthapenta
Studies on Behaviour of Steel Tubular Compression Members
Subjected to Accelerated Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
A. Cinitha, P. K. Umesha and G. S. Palani
Effect of Delamination Defects on Buckling Behavior of
Wind Turbine Blades . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Vinodkumar Boniface
Analysis of the Performance of the Cantilever-Type
Piezoelectric Generator Based on Finite Element Modeling . . . . . . . . . . 291
A. N. Soloviev, I. A. Parinov, A. V. Cherpakov,
V. A. Chebanenko, E. V. Rozhkov and L. V. Duong
Energy Absorption Characteristics of AA7075-T6
Tube Filled with Aluminum Foam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
S. Vignesh, C. Lakshmana Rao and Simhachalam Bade
Strain Intensity Factor and Interaction of Parallel Rigid Line
Inclusion in Elastic Matrix Using FEA . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Prataprao Patil, S. N. Khaderi and M. Ramji
Part V Creep-Fatigue
Creep-Fatigue Damage Assessment of Reactor Hot
Pool Components During Crash Cooling . . . . . . . . . . . . . . . . . . . . . . . . 325
Rosy Sarkar, S. Jaladeen and K. Velusamy
The Influence of Prior Plasticity on the Creep Resistance
of Two Elevated Temperature Power Plant Steels . . . . . . . . . . . . . . . . . 341
Valliappa Kalyanasundaram and Stuart R. Holdsworth
Structural Integrity Mechanics and Creep Life Prediction
of 304HCu Austenitic Stainless Steel Under Multiaxial
State of Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Kanhu Charan Sahoo, Sunil Goyal, P. Parameswaran, S. Ravi and K. Laha
xii Contents
Stress Analysis for Integrity Assessment of High-Energy

Hot Reheat Pipe Bends of 210 MW Coal-Fired Unit . . . . . . . . . . . . . . . 369
Rajesh Daga and Mahendra Kumar Samal
Part VI Energy and Transportation

Structural Integrity Study for Additional Piles on an Existing
Jacket Structure in Western Indian Offshore . . . . . . . . . . . . . . . . . . . . . 385
Praveen Bhat and Bakul Master
Structural Integrity Assessment of a Reactor Pressure
Vessel Using State-of-the-Art Methodologies—A Case Study . . . . . . . . . 395
V. Chaudhry, S. M. Ingole, A. K. Balasubrahmanian
and U. C. Muktibodh
Sodium-Free Level Fluctuations: Concern on Structural
Integrity for the Pool-Type Sodium-Cooled Fast Breeder Reactor . . . . . 407
V. R. Chandan Reddy, R. Suresh Kumar, Anil Kumar Sharma,
K. Velusamy and P. Selvaraj
Monotonic Fracture Studies on Bi-metallic Pipe Weld
Joints Having Circumferential Through-Wall Crack . . . . . . . . . . . . . . . 419
G. Raghava, S. Vishnuvardhan, M. Saravanan, P. Gandhi,
Suranjit Kumar, P. K. Singh, I. A. Khan and V. Bhasin
Part VII Experimental Techniques

Assessment of Strain in a Corrosive Environment of
Structural Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
A. Cinitha, P. K. Umesha and K. Kesavan
A Test Stand for Studying Subcritical and Critical
States of Full-Size Reinforced Concrete Structures . . . . . . . . . . . . . . . . 451
Igor Shardakov, Irina Glot, Aleksey Shestakov,
R. Tsvetkov and V. Yepin
Mode I SIF Determination of Orthotropic Laminates
with Double-Ended Cracks Using a Single-Strain Gage . . . . . . . . . . . . . 461
Debaleena Chakraborty, D. Chakraborty and K. S. R. K. Murthy
Mechanical Testing of Elevated Temperature PMC,
Metallic, and CMC Coupons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
Erik A. Schwarzkopf and Michael J. Shepard
In Situ Measurement of Deformation Under Tension of
ABPBI and Its Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
Abhijeet Dhiman, F. G. S. Wasim, M. Neergat
and Krishna N. Jonnalagadda
Contents xiii
Determination of Effectiveness of Viscoelastic Dampers in

Controlling the Seismic Response of Piping System by Shake
Table Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
R. K. Verma, P. N. Dubey and G. R. Reddy
Experimental Study of Deformation Processes in Reinforced
Concrete Beams Strengthened with a Carbon Fiber Sheet . . . . . . . . . . . 503
Anton Bykov, Igor Shardakov, Aleksey Shestakov and Irina Glot
Partial Delamination Detection and Quantification in
Composite Laminates Using Laser Doppler Vibrometer . . . . . . . . . . . . 517
Yashdeep P. Nimje and Gangadharan Raju
FEA and Experimental Analysis on Buckling and Post-buckling
Behavior of CFRP Composite Panel Using Digital Image Correlation
Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
Muhammad Shuaib NK, Ramji M, Naresh Reddy Kolanu
and Gangadharan Raju
Part VIII Fatigue

High Cycle Fatigue Strength of Spring Steel with Small Scratches . . . . 541
Yoshiro Nishimura, Masahiro Endo, Keiji Yanase, Yuichi Ikeda,
Susumu Miyakawa and Nobuyuki Miyamoto
Estimation of Load Sequence Effect on Fatigue Crack
Growth Life According to Various Prediction Models . . . . . . . . . . . . . . 549
A. N. Savkin, A. A. Sedov, A. V. Andronik and K. A. Badikov
SHM—Prognostic Analysis of Tapered Attachment
Lugs Under Fatigue Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
K. Bharath, B. V. Sravan Kumar, L. Chikmath and B. Dattaguru
Experimental Study on Fatigue–Vibration Interactions of
Cyclically Loaded FRP Composite Cantilever Beams Made
of Glass/Carbon-Layered Arrangements . . . . . . . . . . . . . . . . . . . . . . . . 579
R. Murugan, R. Ramesh and K. Padmanabhan
Investigation of Cyclic Stress–Strain Behavior in FCC
Single Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
Nipal Deka and Krishna N. Jonnalagadda
Analysis of the Plastic Zone Size Ahead of the Crack Tip
in Notched Components and Its Application to Fatigue Problems . . . . . 599
Masahiro Endo and Keiji Yanase
xiv Contents
Effect of Loading Frequency on the Fatigue Crack Growth

Rate of Type 304L(N) Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611
Mukhar Sharma, Punit Arora, P. K. Singh, D. K. Sahoo, Gargi Choudhuri
and V. Bhasin
Fatigue Life Characterization of Additively Manufactured
Acrylic like Poly-Jet Printed Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623
J. A. Suresh, Gurunathan Saravana Kumar, Palaniappan Ramu
and Jayaganthan Rengaswamy
Part IX Material Processing

Influence of Machining Process on Surface Integrity and
Fatigue Life of a Turbine Rotor Blade . . . . . . . . . . . . . . . . . . . . . . . . . . 635
Benudhar Sahoo, S. K. Panigrahi and R. K. Satpathy
Spark Plasma Sintering Process as a Tool for Achieving
Microstructural Integrity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
P. Parameswaran, T. Ezhilarasi, Srinivasa R. Bakshi, V. Thomas Paul
and E. Mohandas
Part X Structural Health and Condition Monitoring

On Increasing Quality of Measurement Results at Testing
State of Construction Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
I. P. Miroshnichenko and I. A. Parinov
Damage Growth Study in Unidirectional CFRP Composites
Using Infrared Thermography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673
Sukanta Das, Naresh Reddy and Gangadharan Raju
Online Damage Detection in Metallic Cylindrical Shells
Using Guided Wave Modes by Time Reversal Method . . . . . . . . . . . . . 685
P. Jayesh and Christudas R. Bijudas
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
Subject Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699
About the Editors
Dr. Raghu Prakash is currently working as Professor

in Department of Mechanical Engineering, Indian
Institute of Technology Madras (IIT Madras); he
specializes in the areas of fatigue, fracture of materials
(metals, composites, hybrids), structural integrity
assessment, remaining life prediction of critical com-
ponents used in transportation and energy sectors,
apart from new product design. He has more than 25
years of professional experience in the field of fatigue
and fracture and has more than 80 journal, chapter
publications,100 conference publications and has
edited 2 book volumes. He has developed test systems
for use in academia, R&D, and industry during his
tenure as Technical Director at Bangalore Integrated
System Solutions (BiSS) Research, Bangalore, and
teaches courses relating to Fracture Mechanics, Design
with Advanced Materials, Product Design, Design for
Manufacture and Assembly (DFMA) at IIT Madras.
He is a voting rights member of the ASTM
International (Technical Committees, D-30, E-08, and
E-28) and has won several prestigious awards (Binani
Gold Medal, Indian Institute of Metals), scholarships,
and Erasmus Mundus Fellowships. He is the recipient
of Distinguished Fellow of the International
Conference on Computational and Experimental
Engineering and Sciences (ICCES) 2015.
He received his bachelor’s degree in Mechanical
Engineering from College of Engineering, Guindy,
Madras (now Chennai), in first class with distinction.
He obtained his master’s degree (by Research) and
Ph.D. from the Department of Mechanical Engineering,
xv
xvi About the Editors
Indian Institute of Science, Bangalore. He is a member

of several technical societies (Indian Structural Integrity
Society, Society for Failure Analysis, Indian Institute of
Metals, to name a few).
Vikram Jayaram is Chair of the Division of

Mechanical Sciences and Professor in Materials
Engineering at the Indian Institute of Science,
Bangalore. He received his Ph.D. in Materials
Science and Engineering from Stanford University,
USA, in 1985 and his bachelor’s degree in Natural
Sciences from Cambridge University, UK, in 1978.
His current research covers the mechanical behavior
and testing of systems at small-length scales, including
fracture and creep. He also works on the
low-temperature processing of ceramic composites.
Earlier, he had worked on the non-equilibrium pro-
cessing of ceramics and the fabrication of ceramic–
metal composites through infiltration methods. He is
an elected fellow of all the major science and
engineering academies in India, Indian Institute of
Metals, and of the American Ceramic Society. He is a
recipient of the JC Bose Fellowship of Department of
Science and Technology of Government of India and
has authored over 150 publications.
Dr. Ashok Saxena most recently (until January 3,

2017) served as the Provost and Vice-Chancellor of
Academic Affairs at the University of Arkansas,
Fayetteville, USA. He is currently working as
Distinguished Professor of Mechanical Engineering
and Dean Emeritus. He also served as the Dean of
Engineering and the Raymond and Irma Giffels’ Chair
at the University of Arkansas.
He previously held the position of Regents’ Professor
and Chair of the School of Materials Science and
Engineering at the Georgia Institute of Technology in
Atlanta, USA, and was a Fellow Scientist at the
Westinghouse Research and Development Center in
Pittsburgh, USA. He also served as the Vice-Chancellor
of Galgotias University in India for a 2-year period
between 2012 and 2014.
About the Editors xvii
He received his MS and Ph.D. degrees from

University of Cincinnati in 1972 and 1974, respec-
tively, in Materials Science and Metallurgical
Engineering and his B.Tech. degree in Mechanical
Engineering from the Indian Institute of Technology
Kanpur in 1970.
His area of research interest is mechanical behavior
of materials focusing on linear and nonlinear fracture
mechanics and fracture in materials at high tempera-
tures under the conditions of creep and creep–fatigue.
He is the recipient of numerous awards and recogni-
tions that include the George Irwin Medal (1992) from
the American Society for Testing and Materials
(ASTM) for his pioneering contributions to creep–
fracture mechanics, the ASTM Award of Merit and
Fellow (1994), Fellow of ASM International (1996),
Fellow of International Congress on Fracture (2009),
and Georgia Tech Outstanding Research Author
Award (1993). He is a recipient of the Wohler
Fatigue Medal from the European Structural Integrity
Society (ESIS) in 2010 and the Paris Gold Medal from
the International Congress of Fracture (2017) and is an
elected Fellow of the European Academy of Sciences
(2016). He is the author/co-author/editor of 8 books
and over 250 research publications.
High-R Fatigue Crack Growth Threshold
Stress Intensity Factors at High
Temperatures
Stuart R. Holdsworth and Zhen Chen
Abstract While knowledge relating to the determination and practical application

of fatigue crack growth threshold stress intensity factors for defect assessment is
relatively well established for many circumstances, this is not the case for materials
and conditions which are sensitive to time-dependent mechanisms. There are two
well-established international standard procedures for the determination of this
fracture mechanics parameter, although their respective crack growth rate criteria
differ by an order of magnitude. Unfortunately neither specifically addresses
determination of the property for very high-R (Kmin/Kmax) ratios, under conditions
when the environment can be influential, and DKth can be even more sensitive to the
da/dN(DK) criterion adopted for its determination. In addition to a general state of
knowledge review, particular attention is paid to circumstances concerning high-
R DKth in power plant steels at high temperatures for which oxide-induced crack
closure and creep cracking can be influential. Evidence for low-alloy 1%Cr,
martensitic 9%Cr and austenitic 17%Cr steels is examined.
Keywords High-R High temperature DKth Oxide-induced crack closure

Creep cracking
Nomenclature
a Crack depth
A Constant in Paris mid-K regime power law
B Specimen thickness
CT Compact tension (specimen)
CTOD Crack opening displacement
da/dN Fatigue crack growth rate
DCPD Direct current potential drop (electrical crack monitoring instrumentation)
S. R. Holdsworth (&) Z. Chen

Swiss Federal Laboratories for Materials Science and Technology (EMPA),
Uberlandstrasse 129, 8600 Dubendorf, Zurich, Switzerland
e-mail: stuart.holdsworth@empa.ch
© Springer Nature Singapore Pte Ltd. 2018 1

R. Prakash et al. (eds.), Advances in Structural Integrity,
https://doi.org/10.1007/978-981-10-7197-3_1
2 S. R. Holdsworth and Z. Chen
f Frequency
FIB Focussed ion beam
HCFCG High-cycle fatigue crack growth (typically for 80 < f < 100 Hz)
TDFAD Time-dependent failure assessment diagram
kp Oxidation parabolic growth constant
k′ Inelastic strain constant in e(r) relationship
K, DK Stress intensity factor, range of stress intensity factor
Kc Critical stress intensity factor responsible for unstable fracture
C Material creep toughness (for a given temperature and time)
Kmat
Kmax Maximum stress intensity factor (in cycle)
Kmin Minimum stress intensity factor (in cycle)
Kr K Ratio representing proximity to fracture
DKth Fatigue crack growth threshold stress intensity factor
dDKthox Enhancement to DKth due to oxide-induced crack closure
Lr Stress ratio representing proximity to plastic collapse or creep rupture
m Exponent in Paris mid-K regime power law
N Number of cycles
R Load ratio (Kmin/Kmax)
Rp0.2 0.2% proof strength
Rm Ultimate tensile strength
RR Creep-rupture strength
RC0:2 0.2% creep strength (stress responsible for 0.2% inelastic strain for a
given temperature and time)
RT Room temperature
SEM Scanning electron microscope
t Time
W Specimen width
x Oxide thickness
b Inelastic strain exponent in e(r) relationship
e, eref Strain, Reference strain
r, rref Stress, Reference stress
rmax
ref Maximum reference stress (in cycle)
t Poisson’s ratio
1 Background and Introduction
High-cycle fatigue crack growth (HCFCG) behaviour may conveniently be con-

sidered in terms of three regimes, Fig. 1a. These are a low-DK regime close to the
fatigue crack growth threshold DKth, a mid-K regime in which fatigue crack growth
rates are represented by a power law [1], and a high-DK regime in which da/dN is
high and Kmax approaches the critical stress intensity factor responsible for unstable
High-R Fatigue Crack Growth Threshold Stress Intensity Factors … 3
Fig. 1 a Schematic representation of high-cycle fatigue crack growth rate characteristics in low-,
mid- and high-DK regimes; and b the influence of frequency, environment and R ratio on fatigue
crack growth threshold values for 1CrMoV at 550 °C [2, 5]
fracture, Kc [2]. The focus of the following paper is behaviour in the low-DK regime,
in particular for high-R (Kmin/Kmax) load ratios at high temperatures.
Conceptually, DKth is the maximum DK associated with no fatigue crack
extension. Experimentally, it is assumed to be the DK consistent with da/
dN 1 10−10 m/c [3] or da/dN 1 10−11 m/c [4]. At low temperatures, the DKth
criteria difference may not be important, but this is not the case at high temperatures
as will become apparent in the following paper.
In the low-DK regime, low da/dN and DKth are sensitive to load ratio (R = Kmin/
Kmax), with the magnitude of DKth decreasing with increasing R (Fig. 1a). At
ambient temperature (RT), higher DKth values at low-R are primarily the conse-
quence of plasticity-induced crack closure, but may also be due to contributions
from fracture surface roughness-induced closure and fretting oxide-induced closure.
At high temperatures, DKth(R) profiles become increasingly elevated with
reducing frequency [2, 5], with the evidence for 1CrMoV steel at 550 °C (which
also includes data collected at 0.01 Hz in vacuum), clearly indicating the
enhancement of DKth to be mainly the consequence of oxide-induced crack closure,
Fig. 1b, although plasticity-induced closure is influential at lower R due to lower
Rp0.2 at higher temperatures.
New high-temperature DKth results for three steels with different oxidation
characteristics (respectively containing 1%Cr, 9%Cr and 17%Cr) have been
assessed in terms of fracture surface oxide thickness measurements and a K(CTOD)
analysis.
During the course of the study, it became apparent that creep cracking could
occur at very low da/dN(DK) close to DKth under high-R conditions at high tem-
peratures, even for high frequencies not normally expected to be affected by
time-dependent cracking processes. A time-dependent failure assessment diagram
(TDFAD) approach is adopted to predict the incidence of creep cracking at the tips
of very slowly propagating high-R high-cycle fatigue cracks.
2 High-Temperature Fatigue Crack Growth Thresholds
High temperature DKth values were determined in accordance with [3] for three
steels with different chromium levels, namely 1CrMoV, 9CrMoCo and 17Cr12Ni.
As a generality, tests were conducted using compact tension (CT) specimens
instrumented with electrical DCPD crack monitoring instrumentation at a frequency
(f) of *80 Hz. Typically, the tests were performed using proportional W(2B) CT
specimens with thicknesses (B) of 12.5 or 25 mm. The results determined for a
temperature of 550 °C are summarised as a function of R and Kmax in Fig. 2. The
high-temperature values are compared with RT values when available and/or the
lower bound RT DKth trend line defined in [6].
Fig. 2 Variation of DKth with R and Kmax, respectively, at 550 °C for a, b 1CrMoV, c,
d 9CrMoCo and e, f 17Cr12Ni (with filled data points representing those tests involving creep
crack development when high-cycle fatigue cracking was almost stationary)
As will be considered later, evidence of creep cracking in certain circumstances

could be observed at high-R high-cycle fatigue DKth test crack tips (when Kmax and/
or rmax
ref were high). The filled DKth data points in Fig. 2 acknowledge the incidence
of creep cracking associated with these tests.
For each steel, there is indication of a DKth enhancement, which at high-R can
almost exclusively be attributed to oxide-induced crack closure. The extent of high-
R DKth enhancement (dDKthox ) is not the same for each steel (Table 1) and appears to
be related to the chromium content. This is examined further in the following
section.
3 Oxide-Induced Crack Closure
The growth of oxide scales in air at elevated temperatures on the surfaces of many
steels may be predicted with some certainty up to a thickness of *100 lm by
adopting parabolic growth kinetics, i.e.
x2 ¼ kp :t ð1Þ
where upper bound kp values for 1%Cr, 9%Cr and 17%Cr steels for a temperature
of 550 °C are summarised in Table 1 [7]. For oxide thicknesses above *100 lm,
spallation becomes increasingly probable, and adoption of the listed kpUB values
should be with caution. Typically, oxidation kinetics are relatively insensitive to
chromium content up to *10%, above which there is a significant increase in
oxidation resistance to *12%Cr, before a relative stabilization, Fig. 3. The vari-
ation in oxidation kinetics is accompanied by the changes in scale formation
mechanism shown as insets in Fig. 3 which are likely to be influential in their
contribution to dDKthox .
Fracture surface oxide thickness measurements for the three steels were made
immediately adjacent to the final crack tips established during DKth measurement
Table 1 Summary of oxide-induced crack closure data (550 °C)

rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi xðFIBÞ
Steel dDKthox kpUB [7] UB t R550 C
E550 C K ðCTODÞ [9]
k p0:2
MPa√m 2
mm /h p lm MPa MPa MPa√m
lm
1% 2.7 1 10−6 2.5 3.51 445 160,800 2.5
Cr
9% (1.8) 3 10−7 2.7 0.51 483 159,078 1.0
Cr
17% (1.2) 3 10−9 0.5 0.46–1.47 197 165,405 0.4–0.7
Cr
dDKthox values given in parentheses are estimates with reference to the BS7910 RT DKth lower
bound [6] and are likely to be higher than in practice. NB DKth defined in accordance with [3]
Fig. 3 Variation of oxide

growth kinetics with
chromium content in steels at
550 °C [8]
campaigns conducted in accordance with [3]. With fracture surface oxide thicknesses
being in the range 0.5–3.5 lm as a consequence of being exposed for *24 h during
the final very low da/dN(DK) phase of high-R DKth determination, an SEM-FIB
drilling technique had to be adopted for maximum measurement accuracy. This
involved FIB drilling a small trench, typically 1–2 mm deeper than the anticipated
oxide scale thickness, with one vertical face and one stepped face. Prior to drilling, a
platinum coating was locally sputter-deposited on the fracture surface to provide
edge protection and a marker at the position of excavation. Following ‘rough’ dril-
ling, the ion beam was used to polish the vertical surface of the trench. The sample
could then be tilted for oxide thickness measurement directly in the SEM.
Observed FIB-determined oxide thickness measurements compare reasonably
well with those calculated using the kpUB values established in [7], Table 1.
The apparent dDKthox enhancements experienced by the three steels at 550 °C
(Fig. 2) are consistent with the oxide-induced crack closure Ks which may be
predicted using the plane strain Rice relationship [9], i.e.
2
CTOD ¼ 0:225 K Rp0:2 ð2Þ
The degree of consistency is evident in Table 1 and Fig. 4. An alternative and

widely adopted K(CTOD) formulation is that proposed by Stewart [10] (Eq. 3),
although dDKthox predictions appear to be excessive using this relationship (Fig. 4).
K 2 ð 1 t2 Þ
CTOD ¼ ð3Þ
2ERp0:2
4 Creep Cracking
An additional mechanism interaction which may be encountered in

high-temperature high-R DKth tests is creep cracking (Fig. 5). Experience has
shown that the incidence of creep cracking ahead of very slowly growing
Fig. 4 Comparison of observed dDKthox for 1%Cr, 9%Cr and 17%Cr steels at 550 °C with
predictions based on Rice and Stewart K(CTOD) formulations (respectively, Eqs. 2 and 3)
Fig. 5 Evidence of creep cracking (dark bands) at the ends of three high-R DKth fatigue cracking
campaigns in a W(2B)12.5 CT specimen (with inset showing W(2B)12.5 CT specimen with DCPD
instrumentation attached)
high-cycle fatigue cracks may go undetected by examination only of da/dN(DK)

test records. Three additional examples are shown in Fig. 6. Perhaps surprisingly, in
these circumstances, fracture surface post-test examination may be the only way to
reveal the presence of creep cracking with certainty (e.g. Fig. 6c). However, it is
shown that when fracture surface examination is not possible, a TDFAD approach
may be adopted to predict the existence of creep cracking when Kmax approaches
Kcmat and/or rref approaches RC0:2 .
While there are two types of high-temperature failure assessment diagram [11],
the one favoured for this application is the R5 TDFAD construction [12]. The
failure envelope for this diagram is defined by:
Fig. 6 Examples of high-R high-cycle da/dN(DK) records (with accompanying post-test fracture
surfaces) for DKth tests during which creep cracking has occurred with; a expected da/dN(DK)
response, b a discernible but unexpected da/dN(DK) response and c a hardly discernible da/dN
(DK) response (W(2B)12.5 CT specimens are 12.5 mm thick)
Table 2 Summary of Steel k′ b C

Kmat MPa√m RC0:2 MPa
TDFAD material parameter
data (550 °C) 1%Cr 730 9 48 366
9%Cr 1200 5 42 345
17%Cr 331.9 11.4 130 123

Eeref ðLr Þ3 RC0:2
Kr ¼ C
þ for Lr Lmax
r ð4Þ
Lr R0:2 2Eeref
Kr ¼ 0 for Lr [ Lmax
r
C
where Kr ¼ K Kmat , Lr ¼ rref RC0:2 and Lmax
r ¼ RR RC0:2 . The parameters used for
TDFAD construction for the three steels considered in this study at 550 °C are
summarised in Table 2.
Important information for construction of the TDFAD envelope (Eq. 4) is iso-
chronous e(r, t) data for the time period of interest. Typically, for DKth tests
involving an ASTM da/dN(DK) criteria [3], high-cycle fatigue cracks can be almost
Fig. 7 TDFAD constructions for 1CrMoV, 9CrMoC and 17Cr12Ni steels at 550 °C (with filled
data points representing those tests involving creep crack development when high-cycle fatigue
cracking was almost stationary)
stationary for *24 h, and e(r) for this period are modelled from data from various
sources (e.g. [11, 13]), Table 2, using:
r r b
e¼ þ 0 ð5Þ
E k
TDFADs for the three steels at 550 °C are given in Fig. 7. As in Fig. 2, filled
co-ordinates are used to represent those high-R DKth tests involving creep crack
development when high-cycle fatigue cracking is almost stationary. The TDFAD
envelope acceptably predicts creep crack development in high-R DKth tests for the
1CrMoV and 9CrMoCo steels at 550 °C. There is not yet the same evidence for
17Cr12Ni in Fig. 7, primarily because 550 °C is low in the creep range for this steel.
5 Practical Implications
5.1 da/dN(DKth) Criterion Sensitivity: Oxidation
It has already been acknowledged that there are two international standards cov-
ering the determination of DKth with very different da/dN criteria defining fatigue
crack growth threshold stress intensity factor. The DKth da/dN criterion in the
ASTM standard is 1 10−10 m/c [3], whereas that in the ISO standard is
1 10−11 m/c [4].
While this anomaly is unlikely to have a big influence at low temperatures for
materials tested in what for them are relatively inert conditions, the evidence
indicates that the situation is very different at high temperatures.
Depending on the oxidation resistance of a material at a given temperature, DKth
can be significantly enhanced as a consequence of oxide-induced crack closure. For
example, for a DKth da/dN criterion of 1 10−10 m/c, high-cycle fatigue cracking
Fig. 8 Influence of time

(oxidation) on da/dN(dDKthox )
(diagonal broken lines for
three steels) and ultimately on
material independent high-R
da/dN(DK) (connected black
crosses), and the consequence
of different DKth da/
dN criteria [3, 4] (for crack
size resolution of 0.25 mm
[3])
can be close to stationary for *24 h, and dDKthox is *2.7 MPa√m for 1CrMoV at
550 °C (Table 1). For a DKth da/dN criterion of 1 10−11 m/c, high-cycle fatigue
cracking may be close to stationary for >100 h, and, in these circumstances, dDKthox
could be >4 MPa√m for 1CrMoV at 550 °C (i.e. with DKth > 6 MPa√m for
R = 0.9).
The situation is illustrated in Fig. 8. This shows a relatively material independent
reducing high-R da/dN(DK) record representative of time insensitive conditions
(connected black crosses). It also shows three diagonal iso-da/f da/dN(dDKthox ) lines
(one for each of the Cr steels) representing crack size resolution (*0.25 mm [3])
and minimum test times (with the iso-da/f da/dN(dDKthox ) reference lines being
determined by substituting dt = da(dN/da)/f into Eq. 2).
With longer test times at temperature, the diagonal da/dN(dDKthox ) lines move
towards the right (Fig. 8). The consequence of this is enhanced DKth values.
Importantly, if the adopted DKth test DK reduction rates are slower than the oxi-
dation rates, da/dN(DK) can be prevented from reducing to the standard defined
DKth da/dN criteria.
5.2 da/dN(DKth) Criterion Sensitivity: Creep Cracking
The evidence in Fig. 7 indicates that, while it could have been possible to predict
the incidence of creep cracking during high-R DKth determinations for the 1CrMoV
C
steel with only a knowledge of Kmat , it would not for the 9%Cr and in 17%Cr steels,
for which a TDFAD analysis is required.
6 Concluding Remarks
High-R high-cycle fatigue crack growth thresholds for power plant steels are
enhanced at high temperatures due to oxide-induced crack closure.
The extent of any enhancement of DKth due to oxide-induced crack closure
depends on time at low da/dN, temperature and material/oxidation mechanism.
Consequently, the adopted standard defined da/dN(DKth) criterion becomes
increasingly important with increasing temperature.
High-R DKth values are increasingly influenced by the development of creep
cracking at very low da/dN with increasing temperature. While this may only be
possible to detect with certainty by fracture surface examination, evidence is pre-
sented to show that creep cracking in these circumstances may be predicted using a
TDFAD analysis.
References
1. P.C. Paris, F. Erdogan, A critical examination of crack propagation laws. J. Basic Eng. 85(4),
528–533 (1963)
2. S.R. Holdsworth, in High Temperature Fatigue Crack Growth, ed. by J.B. Marriott. High
Temperature Crack Growth in Steam Turbine Materials, (Commission European
Communities, COST Monograph EUR 14678EN, 1994), pp. 129–176
3. E 647, in Standard Test Method for Measurement of Fatigue Crack Growth Rates, (ASTM
Standard, ASTM International, West Conshohocken, PA, US)
4. ISO 12108, in Metallic Materials—Fatigue Testing—Fatigue Crack Growth Method,
International Standard
5. R.P. Skelton, J.R. Haigh, Fatigue crack growth rates and thresholds in steels under oxidising
conditions. Mat. Sci. Eng. 36, 17–25 (1978)
6. BS 7910, in Guide to Methods for Assessing the Acceptability of Flaws in Metallic Structures,
(British Standards Institution, 2013)
7. S.R. Holdsworth, in Review of Air Oxidation Kinetics for a Range of Low and High Alloy
Steels, unpublished (2000)
8. P.J. Ennis, W.J. Quadakkers, Mechanisms of steam oxidation in high strength martensitic
steels. Int. J. Pres. Ves. Pip. 84, 75–81 (2007)
9. J.R. Rice, in The Mechanics of Crack Tip Deformation and Extension by Fatigue. Fatigue
Crack Propagation, vol. 415 (ASTM STP 1967), pp. 247–311
10. A.T. Stewart, The influence of environment and stress ratio on fatigue crack growth at near
threshold stress intensities in low alloy steels. Eng. Fract. Mech. 13(3), 463–478 (1980)
11. D.W. Dean, R.D. Patel, A. Klenk, F. Mueller, Comparison of procedures for the assessment
of creep crack initiation. OMMI 3(3), (2004)
12. R5, An Assessment Procedure for the High Temperature Response of Structures, EDF
Energy, 3 (2003)
13. ASME, in Boiler and Pressure Vessel Code III, Rules for construction of nuclear facility
components, Class 1 components in elevated temperature service, Division 1—Subsection
NH, ASME (2004)
Part I
Aerospace
Fracture Mechanics Analysis of Fuselage
Structural Joints with Cracks
H. Ramesha, N. C. Mahendra Babu and H. V. Lakshminarayana
Abstract Fracture mechanics provides a methodology for prediction, prevention

and control of fracture in materials, components, and structures. Fracture mechanics
analysis is the backbone of damage tolerant design. Objectives of fracture
mechanics analysis are: (1) stress analysis of cracks to derive crack tip stress field
equations and define crack tip stress intensity factors, (2) determination of SIF
solutions as function of crack length, orientation and applied loads for a given
geometry, (3) prediction of mixed mode fracture under static, dynamic, and sus-
tained loads, (4) prediction of residual strength as a function of crack length. The
focus of this paper is on fracture mechanics analysis of longitudinal and circum-
ferential joints of aircraft fuselage structure with cracks. The loading is by internal
pressure. Commercial FEA software ANSYS and a special purpose post-processing
subprogram called 3MBSIF are used to determine mixed mode membrane and
bending SIF solutions. Residual strength prediction is based on the use of strain
energy density theory of fracture. Significant results of this study are graphically
presented and discussed in this paper.

Keywords Fuselage Structural integrity Longitudinal joint Circumferential

joint Crack Residual strength Stress intensity factor
Nomenclature
t Thickness
E Modulus of elasticity
t Poisson’s ratio
a0 Initial half crack length
H. Ramesha (&) H. V. Lakshminarayana

Department of Mechanical Engineering, Dayananda Sagar College of Engineering,
Bangalore 560078, India
e-mail: bharatavarsh@gmail.com
N. C. Mahendra Babu
Department of Mechanical and Manufacturing Engineering, MSRUAS,

https://doi.org/10.1007/978-981-10-7197-3_2
16 H. Ramesha et al.
acr Critical half crack length

K1 Mode 1 stress intensity factor
ΔK1 Mode 1 stress intensity factor range
ΔK2 Mode 2 stress intensity factor range
N Number of cycles
1 Background
Fuselage of commercial transport aircraft is a complicated structure designed to

withstand aircraft loads. Parts and assembly need to be designed to give more life to
the structure there by reducing high replacement costs. Nowadays, more research is
going on in extending the life of aircraft beyond its designed life. Fracture
mechanics analysis helps in this aspect by determining the structural integrity,
remaining life, and residual strength of the structure.
Longitudinal and circumferential joints formed on fuselage structure are more
susceptible to fatigue damage [1]. As aircraft undergoes pressurization cycles, the
parts are subjected to cyclic load which leads to initiation and growth of cracks at
joint regions. After number of flight cycles, these cracks coalesce and will lead to
failure of the structure [2]. Inspection of structure using NDT tests and damage
tolerance analysis helps in analyzing the structure ensuring the safety of passengers
and aircraft.
Experimental investigation of fuselage structure is a costly, tedious, and
time-consuming process. However, in the absence of suitable reliable analytical
means, many such investigations are carried out throughout the world, and a few of
these results are also available in open literature [3, 4].
In this work, effort is made for fracture mechanics analysis of curved stiffened
panels with longitudinal and circumferential joint using commercial FEA software
ANSYS.
2 Problem Statement
Longitudinal and circumferential joints of fuselage structure are considered for

fracture mechanics analysis. The specific problem was chosen for investigation,
because complete geometry, dimensions, material property, applied loads, static test
results, fracture test results, and fatigue crack growth test results are readily
available in a report [4]. This is essential for validation and verification of pre-
dictions made in this study.
Fracture Mechanics Analysis of Fuselage Structural Joints … 17
Analysis is carried out in following steps

• Stress analysis
• Evaluation of crack tip stress intensity factors
• Prediction of residual strength
2.1 Longitudinal Joint
A segment of fuselage structure shown in figure is considered for the analysis.

Figure 1a shows details of the joint, consisting of outer skin, inner skin, outer
doubler plate, inner doubler plate, and stringer bonded together. Evidently, the
structure is quite complicated demanding finite element modeling for fracture
mechanics analysis.
Figure 1b shows views of curved panel with lead crack. The assembly detail of
parts is also shown in the same Fig. 1. A lead crack in the skin panel at joint row A
is considered. Fracture mechanics analysis of the panel for various crack lengths is
the main objective of this work.
2.2 Circumferential Joint
Figure 2a shows the details of a circumferential joint configuration in a fuselage

structure consisting of skin, finger doubler plate, and tapered doubler plate.
Location of starter crack can be observed between frame assemblies as shown in
Fig. 2b..
Fig. 1 a Longitudinal joint configuration; b curved panel with a starter crack [4]
Fig. 2 a Circumferential joint configuration; b Circumferential joint in a curved panel with a

starter crack [4]
A segment of Circumferential joint with seven stringers, six frame assemblies,

two skin layers, and doubler plates shown in Fig. 2b is considered for the analysis
[4]. A lead crack in the skin panel at joint row A is considered. Fracture mechanics
analysis of the panel for various crack lengths is the main objective of this work.
Analysis of the curved stiffened panel is carried out in following steps
• FEA of the curved panels (without crack)
• Validation of results with available experimental data
• Fracture mechanics analysis of the curved panels
3 FE Modeling for Stress Analysis and Validation
3.1 Introduction
Finite element modeling is defined here as the analyst’s choice of material models
(constitutive relations and failure criteria), finite elements (types, shapes, orders),
meshes, constraint equations, analysis procedures, governing matrix equations and
their solution methods, specific pre- and post-processing options available in a
chosen commercial FEA software (ANSYS) for the intended analysis (LEFT
analysis) of the candidate component. The finite element model is developed as per
dimensions and material property data extracted from reference [4]. Analysis is
carried out for internal pressure of 0.06937 N/mm2 with symmetric boundary
conditions on all four sides. Thickness and material property data of curved panel
components are shown in Table 1.
Table 1 Material property and thickness of parts

Part name t (mm) Material E (MPa) t
Skin 1.6 Al 2024-T3 72,395 0.33
Stringer–middle 1.8 Al 7075-T3 71,016 0.33
Stringer 1.6 Al 7075-T3 71,016 0.33
Shear clip 1.6 Al 7075-T5 71,016 0.33
Frame 1.8 Al 7075-T4 71,016 0.33
Finger doubler 3.1 Al 2024-T3 72,395 0.33
Tapered doubler 0.6–1.6 Al 2024-T3 72,395 0.33
3.2 Longitudinal Joint
The geometry used to compute stress distribution and deformation is presented in

Figs. 3a, b, and 4. Evidently, it is a very tough task even with the best preprocessor
capability available in ANSYS.
For the specific panel under investigation, measured strains using strain gages
for various applied pressure loads are reported in reference [4]. The measured and
computed strain on frame assembly near the joint region is presented in Table 2.
The agreement is good enough to accept the proposed FE model for further
investigations.
Figure 5a displays the resulting deformations at the maximum pressure. Figure 5b
presents contour plots of von Mises equivalent stress for the whole panel and
identifies a maximum value of 485.77 MPa for internal pressure of 0.06937 N/mm2.
The graphical post-processing capability in ANSYS enables us to locate the region
where it occurs as a reentrant corner in a frame close to the maximum deformation
region in the panel.
Fig. 3 a Curved panel considered for analysis; b assembly showing frame, stringer, and skin
panel
Fig. 4 Assembly showing stringer, doubler plates, outer and inner skin panels
Table 2 FE model validation Strain (µЄ) experiment [4] Strain (µЄ) analysis % Error
using measured strain
605.4 586.02 3.201
Fig. 5 a Contour plots of resultant displacement; b von Mises stress contour
3.3 Circumferential Joint
The geometry used to compute stress distribution and deformation is presented in

Figs. 6a, b, and 7 highlights circumferential joint region of the panel. The computed
and measured strain value on frame assembly near the joint region is presented in
Table 3. The proposed FE model is used for further investigations.
Figure 8a displays the resulting deformations for internal pressure of
0.06937 N/mm2. Figure 8b presents contour plots of von Mises equivalent stress
for the whole panel and identifies a maximum value of 488.434 MPa in stringer
near frame assemblies.
Fig. 6 a Curved panel considered for analysis; b assembly showing frame, stringer, and skin
panel
Fig. 7 Circumferential joint
Table 3 FE model validation Strain (µЄ) experiment [4] Strain (µЄ) analysis % Error
using measured strain
357.03 303.4 15.02
4 Fracture Mechanics Analysis
4.1 Sub-Modeling and Computation of SIF’s
4.1.1 Longitudinal Joint
ANSYS sub-modeling capability is used for evaluation of stress distribution and

stress intensity factors at crack tips. Figure 9a shows the sub-model region con-
sidered for fracture mechanics analysis. Location of longitudinal lead crack in the
curved panel can be observed from Fig. 9b.
It is to be noted that a single element type namely eight noded quadrilateral
element quadratic in order (SHELL 281) is exclusively used. Around each crack tip,
a refined mesh of singular elements (S-TRIA 6) with NS = 20, Δa = a/50, and ‘a’
Fig. 8 a Contour plots of resultant displacement; b von Mises stress contour
(a) (b)
Crack tip - A Crack tip - A
Crack tip - B
Crack tip - B
Fig. 9 a Sub-model for detailed analysis; b modeling of crack in outer skin panel (rivet row A)
is half crack length, and a compatible mesh of regular elements (QUAD 8 and
TRIA 6) is used. The pre-processing command KSCON is exclusively used here.
Parametric design language capability is used to model cracks of different lengths.
Macros are written to generate FE models for given crack lengths.
Figure 10a, b displays von Mises stress contours around crack tips A and B. It is
possible to identify the mode of fracture and estimate crack tip plastic zone shape
and size from these figures. These data are essential for the choice of fracture
criteria and prediction of fracture pressure and crack growth directions.
ANSYS software provides a post-processing command KCALC to compute
mixed mode Stress Intensity Factors (KI, KII, KIII). The computed SIFs are pre-
sented in Fig. 11a, at the maximum pressure equal to 0.06937 N/mm2. These
results are essential for the prediction of residual strength. Figure 11b provides
ΔK values predicted for load range (Minimum/Maximum) equal to 0.1.
Fig. 10 a von Mises stress contour at crack tip—A; b von Mises stress contour at crack tip—B
(a) (b)
6000 5000
DELTA SIF ( MPa √mm)
MODE 1 SIF DELTA MODE 1 SIF

5000
4000
SIF ( MPa √mm)

4000
3000
3000
2000
2000
1000
1000
0 0
181
210
238
272
313
343
395
439
506
545
594
653
181
210
238
272
313
343
395
439
506
545
594
653
Crack Length -2a (mm) Crack Length -2a (mm)
Fig. 11 a Stress intensity factor as a function of crack length; b stress intensity factor range as a
function of crack length
4.1.2 Circumferential Joint
ANSYS sub-modeling capability is used for evaluation of stress distribution and

stress intensity factors at crack tips. Figure 12a shows the sub-model region con-
sidered for fracture mechanics analysis. Location of circumferential lead crack in
the curved panel can also be observed in the same figure.
Figure 13a, b displays von Mises stress contours around crack tips A and B. It is
possible to identify the mode of fracture and estimate crack tip plastic zone shape
and size from these figures. These data are essential for the choice of fracture
criteria and prediction of fracture pressure and crack growth directions.
ANSYS software provides a post-processing command KCALC to compute
mixed mode Stress Intensity Factors (K1, K2, K3). The computed SIFs are presented
in Fig. 14a, at the maximum pressure equal to 0.06937 N/mm2. These results are
essential for the prediction of residual strength. Figure 14b provides ΔK values
predicted for load range (Minimum/Maximum) equal to 0.1.
Fig. 12 a Sub-model for detailed analysis; b von Mises stress contour in the skin panel near crack
location
Fig. 13 a von Mises stress contour at crack tip—A; b von Mises stress contour at crack tip—B
(a) (b)
2500 2000
DELTA K (MPa √mm)

2000
SIF (MPa √mm)
1500
1500
1000
1000
500
500
0 0
178 195 210 238 288 312 360 430 445 178 195 210 238 288 312 360 430 445
2a (mm) 2a (mm)
Fig. 14 a Stress intensity factor as a function of crack length; b stress intensity factor range as a
function of crack length
4.2 Residual Strength Prediction and Correlation with Test

Data
This section presents prediction of residual strength for longitudinal and circum-
ferential joint curved panels using strain energy density theory of fracture.
According to this crack extension will occur in the direction of minimum strain
energy density (Smin), and the extension will occur when it reaches a critical value
(Scr) which is a material dependent parameter [5].
4.2.1 Longitudinal Joint
Figure 15a shows variation of strain energy density factor with respect to crack
length. Critical SED factor for skin panel material is 30.072. From the figure, it is
evident that Minimum SED factor reaches critical value for crack length equal to
585 mm, which can be termed as critical crack length. From this analysis, it is
found that catastrophic failure of the structure occurs when lead crack at the lon-
gitudinal joint region grows to its critical length of 585 mm. Figure 15b shows
variation of residual strength for different crack lengths.
4.2.2 Circumferential Joint
Figure 16a shows variation of strain energy density factor with respect to crack
length. Its value gradually increases to 17.12 for crack length equal to 550 mm on
further increase in crack length SED factor deviates from its path and remained
constant as shown. This is due to stringer underneath the crack which resists its
growth. So crack will not extend unless this stringer fails. Similar observation is
(a) 45 (b)
40 170
Residual Strenght (MPa)
35 150
30
130
SED Factor
25
S min 110
20
Scr
15
90
10
70
5
0 50
150 250 350 450 550 650 750 150 250 350 450 550 650 750
Crack Length - 2a (mm) Crack Length - 2a (mm)
Fig. 15 a Variation of SED factor for different crack lengths; b variation of residual strength for
different crack lengths
(a) (b)
35 700
Strain Energy Density Factor
Residual Strength (MPa)

30 600
S min
25 500
Scr
20 400
15 300
10 200
5 100
0 0
150 250 350 450 550 650 0 200 400 600 800
Crack Length - 2a (mm) Crack Length - 2a (mm)
Fig. 16 a Variation of SED factor for different crack lengths; b variation of Residual strength for
different crack lengths
found in experimental investigation, crack grows steadily till 560 mm and on

further increase of load, the stringer underneath the crack breaks. Figure 16b shows
variation of residual strength for different crack lengths. At crack length equal to
210 mm, a slight increase in residual strength can be observed, which is due to
resistance by stringers located on either side of the crack.
References
1. S. Pitt, R. Jones, Multiple-site and widespread fatigue damage in ageing aircraft. Eng. Fail.
Anal. 4(4), 237–257 (1997) (Elsevier)
2. G. Samavedam, D. Hoadley, Fracture and fatigue strength evaluation of multiple site damaged
aircraft fuselages—curved panel testing and analysis. Final Report, DOT/FAA/CT-94/10,
DOT-VNTSC-FAA-93-8 (1994)
3. G. Samavedam, D. Hoadley, J. Davin, in Structural Integrity of Ageing Airplanes, Test Facility
for Evaluation of Structural Integrity of Stiffened and Jointed Aircraft Curved Panels,
(Springer, 1991), pp. 321–337
4. J. Bakuckas, Full-scale testing and analysis of fuselage structure containing multiple cracks.
Report DOT/FAA/AR-01/46, (2002)
5. D.Y. Jeong, O. Orringer, G. C. Sih, Strain energy density approach to stable crack extension
under net section yielding of aircraft fuselage. Theor. Appl. Fract. Mech. 127–137 (1995)
(Elsevier)
Effect of Silica Nanoparticles
on the Fatigue Life of a Glass Fiber
Reinforced Epoxy Composite Under
an Aircraft Spectrum Load Sequence
N. Jagannathan, K. Sakthivel, Ramesh Bojja and C. M. Manjunatha
Abstract Two types of glass fiber reinforced plastic (GFRP) composites viz.,
(i) GFRP employing unmodified LY556 epoxy matrix (GFRP-neat), and (ii) GFRP
incorporated with 10 wt% of well-dispersed silica nanoparticles in the LY556
epoxy matrix (GFRP-nano), were tested to determine their fatigue life under
mini-FALSTAFF, a standard fighter aircraft spectrum load sequence. Spectrum
fatigue tests were conducted on standard test specimens in a 50 kN servo-hydraulic
test machine with sinusoidal waveform at an average frequency of 3 Hz. Tests were
conducted on both types of GFRP composites with various reference stresses to
determine the fatigue life expressed as number of blocks required for failure. The
fatigue life of GFRP-nano composite was observed to be about four times higher
than that of GFRP-neat composite over the entire range of reference stresses
investigated. For a given number of applied load cycles, both the matrix crack
density and stiffness reduction rates were observed to be lower in GFRP-nano
composite when compared to that of GFRP-neat composite. Presence of silica
nanoparticles in the epoxy matrix of GFRP appear to reduce matrix cracking and
also retard crack growth rate in the composite leading to enhanced fatigue life.
Further, using constant fatigue life diagrams of these materials, the spectrum fatigue
life under mini-FALSTAFF load sequence was predicted. Good correlation was
observed between the predicted and experimental fatigue life for both types of
composites.
Keywords Glass fiber Polymer composite Spectrum fatigue

Silica nanoparticle
N. Jagannathan (&) K. Sakthivel R. Bojja C. M. Manjunatha

Structural Technologies Division, Fatigue and Structural Integrity Group,
CSIR-National Aerospace Laboratories, Bengaluru 560017, India
e-mail: njagan@nal.res.in

https://doi.org/10.1007/978-981-10-7197-3_3
28 N. Jagannathan et al.
1 Introduction
Continuous fiber reinforced polymer (FRP) composites are replacing conventional

metallic alloys in many structural applications such as airframe, wind turbine, ship
hull. Since engineering structures experience cyclic fatigue loads in service, the
fatigue performance of such composites assumes significant importance in design,
safety, damage tolerance, and durability issues of these structures. Although FRP
composites exhibit high specific strength and stiffness with good fatigue resistance,
the continuous demand for high-performance structural materials has resulted in use
of nanotechnology to further improve mechanical properties of composites.
FRP composites containing nano fillers in the epoxy matrix, called as nanocom-
posites, have gained tremendous importance in recent times [1, 2]. Significant
improvements in static mechanical properties such as tensile and compressive
strength, stiffness, interlaminar shear strength, flexural strength, and fracture tough-
ness have been obtained by addition of various types of nanofillers such as SiO2, SiC,
Al2O3, and TiO2 particles, carbon nanotubes, and carbon nanofiber [3–10]. Epoxies
modified with layered fillers such as clay, graphite platelets, and fullerene have also
been shown to improve the mechanical properties of epoxies and FRPs based upon
such modified epoxies [11–15].
The fatigue behavior of bulk epoxies modified with nanofillers has been inves-
tigated. Presence of silica nanoparticles, carbon nanotubes, and carbon nanofibers in
the epoxy improves constant amplitude fatigue life [16–19]. Further, the use of
nano-modified epoxies as the matrices in FRPs has been observed to improve the
fatigue properties of the FRPs. The addition of small amounts of carbon nanotubes
[20], carbon nanofiber [21], silica nanoparticles [16, 22], and nanoclay [12] has been
shown to enhance the fatigue properties of FRP composites.
Hybrid-modified epoxy composites wherein two or more different types or sized
fillers added have been developed recently which dramatically improve fracture
toughness and fatigue properties of polymer composites. Various hybrid combi-
nations such as nano- and micro-sized silica particles [23], nano-silica and
nano-rubber [17], nano-silica and micron rubber [24–28], nano-silica and MWCNT
[29], carbon nanotubes and graphite nanoplatelets [30] have been employed to
significantly improve fracture toughness and fatigue properties of composites.
Most of the fatigue studies on nanocomposites have been limited to constant
amplitude fatigue but spectrum fatigue studies which simulate real service loads
have not been studied in detail. Recently, we have observed that a polymer
nanocomposite containing silica nanoparticles exhibit improved constant amplitude
fatigue life [16] as well as enhanced fatigue life under a wind turbine spectrum load
sequence [31]. The main aim of this investigation was to measure and predict the
fatigue behavior of a GFRP-nano composite, under a standard aircraft spectrum
load sequence. Fatigue life was determined experimentally and compared with
predictions through empirical equations.
Effect of Silica Nanoparticles on the Fatigue Life … 29
2 Experimental
2.1 Material
Two types of glass fiber reinforced polymer (GFRP) composites were considered in
this study viz., (i) GFRP with unmodified LY556 epoxy matrix (termed as
GFRP-neat), and (ii) GFRP with modified epoxy matrix containing 10 wt% silica
nanoparticles (termed as GFRP-nano). The complete details of the materials used
and the processing employed to fabricate the GFRP composites can be found in
Manjunatha et al. [16]. However, for the sake of completeness, they are briefly
explained here.
The epoxy resin used was a diglycidyl ether of bisphenol A (DGEBA) resin. The
silica (SiO2) nanoparticles were obtained as a colloidal silica sol. with a concen-
tration of 40 wt% in a DGEBA resin. The curing agent was an accelerated
methylhexahydrophthalic acid anhydride. The E-glass fiber cloth used was a
non-crimp-fabric. The required quantity of the neat epoxy resin and the calculated
quantities of silica nanoparticle-epoxy resin to give 10 wt% of nano-silica in the
final resin were all individually weighed, degassed, and mixed together, and a
stoichiometric amount of curing agent was added to produce a resin mix. This resin
mixture was used to prepare the GFRP composite laminate by resin transfer
molding technique.
E-glass fiber fabric pieces were cut and laid up in a quasi-isotropic sequence
[(+45/−45/0/90)s]2. The resin mixture was infused into the glass-cloth layup and
cured at 100 °C for 2 h and post-cured at 150 °C for 10 h. In this way, two types of
GFRP composites were fabricated, i.e., GFRP-neat and GFRP-nano. The laminates
produced were about 2.5–2.7 mm thick and had a fiber volume fraction of about
57%.
The silica particles of about 20 nm in diameter were evenly distributed in the
epoxy as shown in atomic force microscopic image [16] in Fig. 1. The tensile and
compressive properties of the composite laminates [31] are shown in Table 1. Both
the strength and stiffness of GFRP-nano composite are slightly higher in both
tension and compression due to the presence of hard silica nanoparticles in the
epoxy matrix.
2.2 Fatigue Testing
Fatigue tests on composites were conducted under a standard spectrum load

sequence. The load sequence considered in this investigation was a fighter aircraft
loading standard for fatigue evaluation, mini-FALSTAFF [32, 33] shown in Fig. 2.
It is a short version of standard FALSTAFF load spectrum, which is a standardized
variable-amplitude test load sequence developed for the fatigue analysis of mate-
rials used for fighter aircraft. In Fig. 2, the normalized stress is plotted against peak/
50 nm
Fig. 1 Atomic force microscopy image of a 10 wt% silica nanoparticle-modified bulk epoxy
polymer [31]
Table 1 Mechanical properties of the GFRP composites [31]

Type of test Mechanical property Material % change
GFRP-neat GFRP-nano
Tension rUTS (MPa) 365 ± 13 382 ± 12 +4.65
ET (GPa) 17.5 ± 0.1 18.8 ± 1.7 +7.42
Compression rUCS (MPa) 355 ± 47 361 ± 28 +1.69
EC (GPa) 21.3 ± 0.4 22.6 ± 0.4 +6.10
One block
Fig. 2 Standard mini-FALSTAFF spectrum load sequence [32]

12.5 mm
50 mm 35 mm 50 mm
6 mm
~2.7 mm thick laminate
Fig. 3 Schematic diagram showing the dimensions of the fatigue test specimens
trough points of load sequence. One block of this load sequence consists of 18,012
reversals at 32 different stress levels and represents loading equivalent of 200
flights. The stress sequence for our experiments was obtained by multiplying with a
constant reference stress value, rref for all the peak/trough points in the entire block.
Spectrum fatigue tests with various reference stress levels were conducted on
both GFRP composites. The geometry and dimensions of the test specimens
employed for spectrum fatigue tests are shown in Fig. 3. Tests were conducted in a
computer controlled 50 kN servo-hydraulic test machine. Sinusoidal waveform
with an average frequency of 3 Hz (six reversals per second) was employed in the
tests. For any given reference stress, the number of load blocks required to fail the
test specimen, Nb, was determined. Whenever a specimen failed in-between a full
block, it was rounded-off to the nearest complete block number.
The stiffness variation of the specimen subjected to spectrum fatigue loads was
determined during the test as a function of the number of applied load blocks.
Whenever stiffness measurement data were required, the fatigue test was inter-
mittently stopped, a load cycle with rmax = 0.5 rref and stress ratio R = rmin/
rmax = 0 was applied, the load versus displacement data were obtained and then
analyzed. Considering the large number of load cycles in one block, insertion of
this one cycle was assumed not to alter the fatigue damage in the material signif-
icantly. For the purpose of comparison, the normalized stiffness of the specimen
was defined as the ratio of measured stiffness at any given time to the initial stiffness
(obtained before application of the first spectrum load block). For one particular test
with rref = 250 MPa, the specimens were dismounted at the end of the application
of three complete load block, and photographs showing matrix cracks were
obtained, as explained in [16].
2.3 Fatigue Life Prediction
The fatigue life of the GFRP composites under the mini-FALSTAFF load sequence
was predicted and compared with the experimental results. The general method-
ology followed in prediction of fatigue life under spectrum loads in composites may
be found elsewhere [31, 34]. A schematic of flow chart for life prediction is shown
in Fig. 4 which involves [31]: (i) the rainflow counting of the fatigue cycles in the
spectrum load sequence, (ii) the determination of the cycles to failure, Nf, for each
of the counted load cycles using a constant life diagram (CLD) of the material,
(iii) the calculation of the damage fraction for each of the counted load cycles as the
ratio of cycle count to Nf obtained from the CLD, and finally (iv) the determination
of the total fatigue damage per load block by summation of the damage fraction.
The material is assumed to fail when the total damage fraction reaches 1.0, and,
hence, the fatigue life under the spectrum load sequence is equal to the reciprocal of
the total damage estimated per load block. This procedure was followed in the
present study for estimation of fatigue life.
Fig. 4 Schematic diagram

showing the procedure for Spectrum load S-N curves for
prediction of spectrum fatigue sequence the material
life [31]
Rainflow cycle Constant life

count diagram (CLD)
Determine Nf for each

counted cycle from CLD
(use interpolation technique)
Calculate damage fraction

for each counted cycle
Di = ni/Nf,i
Calculate total fatigue

damage per block
D = ∑ (ni / Ni)
Determine Spectrum
fatigue life
N = 1/D
3 Results and Discussion
3.1 Experimental Spectrum Fatigue Life
The experimentally determined spectrum fatigue life of GFRP composites under

mini-FALSTAFF load sequence at various reference stresses is shown in Fig. 5.
The fatigue life was observed to increase with a reduction in the reference stress in
both GFRP composites, as observed in many other investigations [31, 35, 36]. It
may also be seen that for any given reference stress level, the GFRP-nano com-
posite shows higher fatigue life compared to that of GFRP-neat composite by about
three times. The spectrum fatigue life enhancement is observed over the entire
range of reference stress levels investigated.
The normalized stiffness of the specimen as a function of spectrum load blocks
determined for the fatigue test with rref = 250 MPa, for both GFRP composites is
shown in Fig. 6. The stiffness reduction trend observed in these materials is similar
to generally observed behavior in FRP composites [37–42]. It may be seen that the
stiffness reduction rate is quite high in the GFRP-neat composite than in
GFRP-nano composite.
Photographs of the matrix cracks observed on the surface of the composite
subjected to thee complete load block of the mini-FALSTAFF spectrum load
sequence with rref = 250 MPa are shown in Fig. 7. Initiation and growth of such
matrix cracks under cyclic fatigue loads in GFRP composite have been reported by
others [16, 42]. The GFRP-neat composite exhibits more severe cracking than
GFRP-nano composite suggesting suppression of matrix cracks by nano modifi-
cation of epoxy matrix.
The fatigue failure mechanisms under cyclic loads in polymer composites have
been investigated extensively. The generally accepted mechanisms include [37, 40]
(i) initiation and growth of matrix cracks, (ii) initiation and growth of disbonds and
delaminations, and (iii) individual fiber breaks throughout the fatigue life. Studies
have clearly shown that the fatigue crack growth rate of the bulk epoxy containing
Fig. 5 Experimental and 325

predicted fatigue lives of the Lines : PredicƟon
Reference Stress, σref (MPa)
GFRP composites under the 300 Points : Experimental

mini-FALSTAFF spectrum
load sequence 275
250
GFRP - nano
GFRP - neat
225
200
175
0.1 1 10 100 1000
Number of Blocks, Nb
Stage I
Stage II
GFRP-nano
Stage III
GFRP-neat
Fig. 6 Variation of the normalized stiffness for the GFRP composites determined under the
mini-FALSTAFF spectrum load sequence, with rref = 250 MPa
3 mm
(a) GFRP-neat composite (b) GFRP-nano composite
Fig. 7 Photographs showing matrix cracks (indicated by the arrows) in the GFRP composites
subjected to three complete mini-FALSTAFF spectrum load blocks with rref = 250 MPa
nano-silica particles is over an order of magnitude lower than that of the neat epoxy
[43, 44]. Further, it has been shown that the use of such particles to formulate a
modified epoxy matrix in a GFRP composite material enhances the constant
amplitude fatigue life due to suppressed matrix cracking, delayed initiation of
delamination, and reduced crack/delamination growth rate [16].
The stiffness loss in stage I and stage II results primarily from matrix cracking.
Once the matrix crack density saturates and attains the characteristic damage state
(CDS), the disbonds and delaminations created due to the coalescence of primary
and secondary matrix cracks grow, and this leads to a further loss in stiffness. The
present results show that when both composites are subjected to the same number of
spectrum load blocks, the crack density is lower in the GFRP-nano composite
compared to the GFRP-neat composite. Thus, the stiffness loss curves shown in
Fig. 6 indicate the underlying mechanisms, i.e., suppressed matrix cracking,
delayed initiation of delamination, and a reduced crack/delamination growth rate.
All these mechanisms lead to an improvement in the spectrum fatigue life of the
GFRP-nano composite.
3.2 Spectrum Fatigue Life Prediction
The fatigue life of GFRP-neat and GFRP-nano composites was predicted following
the procedure shown in Fig. 4. The constant life diagrams (CLDs) for the GFRP
composites determined in an earlier investigation [31] are shown in Fig. 8.
Rainflow counting of the fatigue cycles in the spectrum load sequence was per-
formed following the ASTM standard specified procedure [45]. Since each of the
rainflow-counted load cycles will be of different load amplitude and mean stress, it
is necessary to interpolate and determine Nf for all these load cycles using the CLD.
A piecewise linear interpolation technique [46] was used in the present investiga-
tion. For damage accumulation model, the Miners’ linear damage accumulation rule
[47] was used:
X
D¼ ðni =Ni Þ ð1Þ
where D is the damage fraction, ni is the cycle count, and Ni is the cycles to failure
for a given load cycle amplitude. Thus, the total damage per load block was
estimated following the above procedure, and the spectrum fatigue life was pre-
dicted for the GFRP composites.
The fatigue life predicted as a function of the reference stress following the
above procedure is shown in Fig. 5, along with the experimental results. It may be
observed from these results that very good agreement exists between the experi-
mental and the predicted fatigue lives for both GFRP composites. Thus, the pre-
dictions also suggest an improvement in the fatigue life of GFRP-nano composite
by about four times over that of GFRP-neat composite.
200
R =-1 R =0.1
R =-4
R =10
Stress Amplitude (σ ), MPa
R =0.3
150
103 cycles
a
R =0.5
100
105
R =0.7
50
107
UCS UTS
0
-400 -300 -200 -100 0 100 200 300 400
Mean stress (σ m ), MPa
(a) GFRP-neat composite
250
R =-1 R =0.1
R =-4
R =10
Stress Amplitude (σ ), MPa
200 R =0.3
103 cycles
a
150
R =0.5
7 5
100 10 10
R =0.7
50
UTS
UCS
0
-400 -300 -200 -100 0 100 200 300 400
Mean stress (σ ), MPa
m
(b) GFRP-nano composite
Fig. 8 Constant life diagrams (CLDs) for the GFRP composites [31]
4 Conclusions
Based on the results obtained in this investigation, the following conclusions may
be drawn:
1. The addition 10 wt% of silica nanoparticles to the epoxy matrix of a GFRP
composite improves the fatigue life under mini-FALSTAFF spectrum load
sequence by about four times.
2. The suppressed matrix cracking and retarded crack and delamination growth
rate in the silica nanoparticle-modified epoxy matrix of the GFRP composite
appear to enhance the spectrum fatigue life.
3. The predicted fatigue lives under the mini-FALSTAFF spectrum load sequence
are in very good agreement with the experimental observations for both the
GFRP-neat and GFRP-nano composites.
Acknowledgements The authors from CSIR-NAL wish to thank Mr. Shyam Chetty, Director and
Dr. Satish Chandra, Head, Structural Technologies Division, CSIR-National Aerospace
Laboratories, Bangalore, India, for their constant support and encouragement during this work.
The laminates were fabricated in the department of Mechanical Engineering, Imperial College,
London, UK. Thanks to Prof. A.J. Kinloch and Dr. A.C. Taylor for their assistance and encour-
agement during this work. The authors also wish to thank the technical support staff members of
the Department of Mechanical Engineering and the Composites Centre of the Aeronautics
Department, Imperial College London, and the Materials Evaluation Lab, STTD, NAL, Bangalore,
for their assistance in the experimental work.
References
1. E.T. Thostenson, C. Li, T.W. Chou, Comp. Sci. Tech. 65, 491 (2005)
2. F. Hussain, M. Hojjati, M. Okamoto, R.E. Gorga, J. Comp. Mater. 40, 1511 (2006)
3. J. Ma, M.S. Mo, X.S. Du, P. Rosso, K. Friedrich, H.C. Kuan, Polymer 49, 3510 (2008)
4. C.M. Manjunatha, A.C. Taylor, A.J. Kinloch, S. Sprenger, J. Mater. Sci. 44(16), 4487 (2009)
5. N. Chisholm, H. Mahfuz, V.K. Rangari, A. Ashfaq, S. Jeelani, Compos. Struct. 67, 115
(2005)
6. S. Zhao, L.S. Schadler, H. Hillborg, T. Auletta, Compos. Sci. Tech. 68, 2976 (2008)
7. B. Wetzel, P. Rosso, F. Haupert, K. Friedrich, Eng. Fract. Mech. 73, 2375 (2006)
8. F.H. Gojny, M.H.G. Wichmann, B. Fiedler, W. Bauhofer, K. Schulte, Compos. A: Appl. Sci.
Manufac. 36, 1525 (2005)
9. Z. Spitalsky, D. Tasis, K. Papagelis, C. Galiotis, Prog. Polym. Sci. 35, 357 (2010)
10. C.M. Manjunatha, A.C. Taylor, A.J. Kinloch, S. Sprenger, J. Mater. Sci. 44(1), 342 (2009)
11. Y. Zhou, V. Rangari, H. Mahfuz, S. Jeelani, P.K. Mallick, Mater. Sci. Eng. A 402, 109 (2005)
12. S.U. Khan, A. Munir, R. Hussain, J.K. Kim, Compos. Sci. Tech. 70, 2077 (2010)
13. Y. Zhou, M. Hosur, S. Jeelani, P. Mallick, J. Mater. Sci. 47, 5002 (2012)
14. J. Cho, J.Y. Chen, I.M. Daniel, Script. Mater. 56, 685 (2007)
15. A.M. Rafiee, F. Yavari, J. Rafiee, N. Koratkar, J. Nano. Part Res. 13, 733 (2011)
16. C.M. Manjunatha, A.C. Taylor, A.J. Kinloch, S. Sprenger, Compos. Sci. Tech. 70, 193 (2010)
17. G.T. Wang, H.Y. Liu, N. Saintier, Y.W. Mai, Eng. Fail. Analy. 16, 2635 (2009)
18. N. Yu, Z.H. Zhang, S.Y. He, Mater. Sci. Eng. A 494, 380 (2008)
19. D.R. Bortz, C. Merino, I. Martin-Gullon, Compos. Sci. Tech. 71, 31 (2011)
20. C. Grimmer, C. Dharan, J. Mater. Sci. 43, 4487 (2008)
21. D.R. Bortz, C. Merino, I. Martin-Gullon, Compos. Sci. Tech. 72, 446 (2012)
22. L. Böger, J. Sumfleth, H. Hedemann, K. Schulte, Compos. Part A: Appl. Sci. Manufact. 41,
1419 (2010)
23. T. Adachi, M. Osaki, W. Araki, S.C. Kwon, Acta Mater. 56, 2101 (2008)
24. Y.L. Liang, R.A. Pearson, Polymer 51, 4880 (2010)
25. A.J. Kinloch, R.D. Mohammed, A.C. Taylor, C. Eger, S. Sprenger, D. Egan, J. Mater. Sci. 40,
5083 (2005)
26. A.J. Kinloch, R.D. Mohammed, A.C. Taylor, S. Sprenger, D.J. Egan, J. Mater. Sci. 41, 5043
(2006)
27. A.J. Kinloch, K. Masania, A.C. Taylor, S. Sprenger, D. Egan, J. Mater. Sci. 43, 1151 (2008)
28. C.M. Manjunatha, S. Sprenger, A.C. Taylor, A.J. Kinloch, J. Compos. Mater. 44, 2095 (2010)
29. L. Böger, J. Sumfleth, H. Hedemann K. Schulte, Compos. Part A: Appl. Sci. Manufact. 41,
1419 (2010)
30. J. Li, P.S. Wong, J.K. Kim, Mater. Sci. Eng. A 483, 660 (2008)
31. C.M. Manjunatha, R. Bojja, N. Jagannathan , Mater. Perf. Char. 3(1), 327 (2014)
32. G.M. Van Dijk, J.B. de Jonge, National Aerospace Laboratory NLR, Netherlands, NLR MP
75017 U (1975)
33. P. Heuler, H. Klatschke, Int. J. Fat. 27, 974 (2005)
34. N.L. Post, S.W. Case, J.J. Lesko, Int. J. Fat. 30, 2064 (2008)
35. T.P. Philippidis, A.P. Vassilopoulos, Compos. Part A: Appl. Sci. Manufact. 35, 657 (2004)
36. C.M. Manjunatha, Ramesh Bojja, N. Jagannathan, A.J. Kinloch, A.C. Taylor, Int. J. Fat. 54,
25 (2013)
37. R. Talreja, Proc. Roy. Soc. London. A. Math. Phys. Sci. 378, 461 (1981)
38. J.S. Tate, A.D. Kelkar, Compos. B 39, 548 (2008)
39. K.L. Reifsnider, R. Jamison, Int. J Fat. 4, 187 (1982)
40. S.W. Case, K.L. Reifsnider, in Fatigue of Composite Materials, eds. I. Milne, R.O. Ritchie, B.
Karihaloo. Comprehensive Structural Integrity, vol. 4 Cyclic loading and fatigue, (Elsevier
Science, Amsterdam, 2003)
41. A. Gagel, B. Fiedler, K. Schulte, Compos. Sci. Tech 66, 657 (2006)
42. A. Gagel, D. Lange, K. Schulte, Compos. A 37, 222 (2006)
43. H.Y. Liu, G. Wang, Y.W. Mai, Compos. Sci. Tech. 72(13), 1530 (2012)
44. C.M. Manjunatha, N. Jagannathan, K. Padmalatha, A.C. Taylor, A.J. Kinloch, Int. J. Nanosci.
10, 1095 (2011)
45. Standard Practices for Cycle Counting in Fatigue Analysis, ASTM E1049, Annual book of
ASTM Standards, American Society for Testing and Materials, PA, vol. 15.03, (2003)
46. A.P. Vassilopoulos, B.D. Manshadi, T. Keller, Int. J. Fat. 32, 659 (2010)
47. M.A. Miner, J. Appl. Mech. 12A, 159 (1945)
Fatigue Life Estimation of Typical Fighter
Aircraft Main Landing Gear Using Finite
Element Analysis
T. Sivaranjani, D. V. T. G. Pavan Kumar, C. M. Manjunatha

and M. Manjuprasad
Abstract Fatigue analysis of main landing gear of typical fighter aircraft was
carried out using MSC Fatigue software considering the strain spectra measured at
critical location during the flight starting from taxiing, take-off to landing. In the
absence of load information on the landing gear, a methodology was proposed to
calculate the equivalent load system from measured strains and verified. Fatigue life
of landing gear was obtained using various stress spectra based on stress-life
approach and was compared with the fatigue life calculated using constant life
diagram (CLD) approach. Fatigue life estimation using Gerber mean stress cor-
rection was found to be in good agreement with the fatigue life estimation using
CLD approach. Fatigue analysis using maximum absolute principal stress criterion
and linear damage rule give the conservative fatigue life, and hence, these criteria
may be used as fatigue life estimation criteria for design.

Keywords Main landing gear Fatigue life Stress spectra Goodman and

Gerber mean stress corrections Maximum absolute principal stress
1 Introduction
Landing gear system is one of the critical systems of an aircraft and is configured
along with the aircraft structure because of its substantial influence on the aircraft
structural configuration itself. Landing gear and its attachments are one of the
principal structural elements and are useful for aircraft during taxiing, take-off and
landing. Principal structural element’s failure results in catastrophic failure. The
main failure mechanisms in landing gear are fatigue, corrosion-related problems and
overload failures. Fatigue failure is brittle-like (relatively little plastic deformation)
—even in normally ductile materials and hence, it is sudden and catastrophic.
T. Sivaranjani D. V. T. G. Pavan Kumar (&) C. M. Manjunatha M. Manjuprasad

Structural Technologies Division, CSIR-National Aerospace Laboratories, HAL Airport
Road, Bangalore 560017, India
e-mail: pavan@nal.res.in

https://doi.org/10.1007/978-981-10-7197-3_4
40 T. Sivaranjani et al.
Catastrophic metal fatigue can lead to aircraft tearing apart in mid-flight. Estimation
of fatigue life for the landing gear system is mandatory in order to certify the landing
gear and frame the schedule for maintenance.
Beaudet and Roth [1] assessed the trends in failure mechanisms of landing gear
components and their causes. The different failure causes identified were fatigue,
corrosion and overload failures. Design deficiencies and manufacturing defects led
mainly to fatigue failures, while poor material selection and improper field main-
tenance are the principal origins of corrosion-related failures. They emphasized that
fatigue can be best addressed by improving the manufacturing quality and better
characterizing the service and manufacturing stresses. Khondker et al. [2] showed
that the composite material structure is having high efficiency and reduced weight.
Imran et al. [3] showed that composite material landing gear is suitable for struc-
tural safety for both static and spectrum loads. Divakaran et al. [4] presented a work
dealing with the various stages of landing gear design and development. They
showed the current technology landscape and how they are going to meet the
challenges in the development of landing gear and demonstrated design tools to
reduce the time and cost.
The fatigue failures can be figured out mostly through non-destructive testing in
the landing gear. These kinds of fatigue failures will result in the catastrophic failure
of aircraft whether it is transport or military aircraft. Hence, there is a need to study
the fatigue phenomena of landing gear, and it is mandatory to consider the fatigue
effects on the landing gear during the design phase itself. Among the various parts
present in the landing gear, oleo strut is one of the critical components under fatigue
condition. In the present fatigue analysis study, the main landing gear outer cylinder
nothing but the shock absorbing strut cylinder was taken into consideration.
2 Static Analysis of Main Landing Gear
As a preliminary step towards modelling and analysis of landing gear, a simple case
of landing gear reported in the literature [2] was studied using finite element
(FE) method. The landing gear model was made up of standard steel having
Young’s modulus of 200 GPa, Poisson’s ratio of 0.266 and yield strength of
250 MPa.
During landing, the piston in the landing gear retracts inside by compressing the
fluid in the chamber. The pressure caused by the compression is the input load for
this analysis, and its magnitude is 20.68 MPa. The piston was clamped at the
bottom end in this analysis. The landing gear was modelled using four-noded
tetrahedral elements using Altair HyperMesh [5], the static analysis was performed
using MSC Nastran [6], and the results are shown in Fig. 1. The hoop stress
obtained from the finite element static analysis is 1030 MPa.
Since the pressure load was applied on the cylinder, the hoop stress can be
calculated analytically as PD/2t = 1034 MPa, where ‘P’ is the applied pressure, ‘D’
is the mean diameter, and ‘t’ is the thickness of the cylinder. The FE hoop stress
Fatigue Life Estimation of Typical Fighter Aircraft … 41
Fig. 1 Landing gear model with boundary conditions and stress contour
value is in good agreement with the analytical hoop stress value. Since, using
four-noded tetrahedral elements, the finite element results are agreeing with the
analytical results, the same elements were used further for the static and fatigue
analysis of a fighter aircraft main landing gear.
2.1 Modelling of Main Landing Gear
A typical fighter aircraft main landing gear was considered. The landing gears are
constructed using low alloy steel, and its general composition in weight percentage
is C: 0.27–0.33, Si: 0.9–1.2, Mn: 1.0–1.3, Cr: 0.9–1.2, Ni: 1.4–1.8, Cu: 0.25 max,
S: 0.025 max and P: 0.025 max. The mechanical properties such as ultimate tensile
strength, yield strength, Young’s modulus and Poisson’s ratio are 1141 MPa,
1010 MPa, 202 GPa and 0.33, respectively. The CAD model was generated using
the geometric dimensions measured from the actual landing gear. The geometry and
the CAD model of the landing gear outer cylinder are shown in Fig. 2. The upper
Fig. 2 Geometric dimensions, CAD model and strain gauge locations of main landing gear
cylinder
portion of the landing gear was assumed as solid, and the lower portion of the
landing gear was modelled as a hollow cylinder (Fig. 2). During landing or take-off
or taxiing, the piston in the landing gear compresses or decompresses the fluid
present inside the strut and hence the internal pressure gets generated. The pressure
magnitude is not known for this main landing gear cylinder. But there are strain
values available at the critical points measured from the experiment. Using those
strain values, the pressure load applied in the main landing gear cylinder was
calculated.
Among the five gauges shown in Fig. 2, four gauges (R1, R2, R3 and R4) were
bonded at the upper portion of the landing gear. Since there was an assumption in
the modelling of upper portion, the strain gauge at the lower portion only was
considered. R5 is the location present in the lower hollow portion of the cylinder,
and the strain values at that point for 0°, 45° and 90° directions are −120 le,
215 le and 570 µe, respectively [7].
2.2 Load Calculation from Strain Measurements
The procedure proposed/used to calculate the load from the experimentally mea-
sured strain values is given here. Principal strains were calculated from 0°, 45° and
90° directions strain measurements, and then, the principal stresses were calculated.
After that, the reference coordinate system stresses were calculated from the prin-
cipal stresses, and they are longitudinal stress, hoop stress and shear stress having
the magnitude of 15 MPa, 120 MPa and 1 MPa, respectively. Since the strain
values are from the main landing gear experiment subjected to static compressive
load on the piston, there should be only internal pressure generated and hence
mainly the hoop stress alone should occur. The longitudinal stress on the main
landing gear cylinder may be due to the bending effect caused by the eccentricities
in load application. In the absence of these eccentricity details, for simplicity, an
equivalent axial load instead of bending load was assumed in order to get back the
measured stress and strain values.
Finally, the loads were calculated from the reference coordinate system stresses
using the hoop stress, axial stress and shear stress formulae, and their values are
60 MPa pressure, 112 kN axial load and 4 kN shear force. The shear force cal-
culated here is just 3.6% of the axial load, and when compared to the hoop stress of
120.26 MPa, the shear stress of 1.588 MPa is 1.3% only. Hence, this shear stress
was considered as negligible in the analysis carried out here.
2.3 Finite Element Analysis
The full landing gear finite element model (Fig. 3) was modelled using four-noded
tetrahedral elements as per the dimensions shown in Fig. 2. The constrained
boundary condition was applied at the top portion of the cylinder based on the
laboratory experiment carried out earlier. The calculated internal pressure load of
60 MPa was applied to the inner wall of the cylinder. From the stress analysis, the
hoop stress at the inner wall is 171 MPa and at the outer wall is 119 MPa. The
stress at the outer wall of the cylinder is in good agreement with the calculated
stress value from the measured strains. The longitudinal stress obtained is 2 MPa
which is not in agreement with the experimental value. So, there should be some
load other than internal pressure acting in the cylinder. The static pressure of
60 MPa was applied internally at the cylinder portion, an axial tensile load of
112 kN was applied at the bottom portion of the cylinder, the static stress analysis
was carried out, and the results are shown in Fig. 3. The FE stresses obtained are
hoop stress and longitudinal stress with the magnitude of 119 MPa and 13 MPa,
respectively, and are in good agreement with the calculated stresses from the
experimentally measured strains. The comparison between the finite element and
experimental strain values is shown in Table 1.
From the above results, the load boundary conditions are clearly understood for
the main landing gear cylinder, i.e. by using the equivalent load system comprising
of internal pressure load and the axial load, so that the original landing gear
behaviour can be achieved. In this way, from the static analysis of main landing
gear, the boundary and loading conditions to be applied on the landing gear were
obtained and verified. With the above-obtained information, fatigue analysis was
carried out using the measured strain spectrum data.
Fig. 3 Static analysis of main landing gear cylinder
Table 1 Comparison of finite element analysis (pressure and axial load) results with experimental
results for main landing gear cylinder
Direction FEA Experimental FEA Experimental
stress (MPa) stress (MPa) strain (µe) strain (µe)
Hoop 119 120 576 570
Longitudinal 13 15 −122 −120
(axial)
3 Fatigue Analysis of Main Landing Gear
The fatigue analysis of main landing gear was carried out, considering the spectrum
loads obtained from the in-flight data measured during taxiing, take-off and landing
using MSC Fatigue tool. For the low alloy steel landing gear, the fatigue material
properties such as Fatigue Strength Coefficient and Fatigue Strength Exponent
values are Sf = 6839 MPa and b = −0.264, respectively.
The strain gauges were bonded at the critical locations (Fig. 2), and the corre-
sponding strain values were measured during the flight using the data acquisition
system. Among the five gauge locations, only the R5 location’s strain spectra
(Fig. 4) at the starboard side of landing gear were taken into consideration for the
fatigue analysis study.
Since the strain spectra were measured starting from taxiing, take-off and landing
stages, the loads acting on the landing gear during those three stages have to be
considered. In general, landing gear is subjected to vertical load, drag load and side
load as shown in Fig. 5. The landing gear types, landing load cases and the cor-
responding combination of landing and ground loads are available in detail in MIL
Fig. 4 Strain spectra measured at R5 location along 0°, 45° and 90° directions on the main
landing gear
Fig. 5 Typical loads on the landing gear
standards [8] and FAR 25 regulations [9]. Based on the type of landing cases such
as two-point landing, tail-down landing and single-point landing, the loads acting
on the landing gear will differ. For example, in the case of drift landing [8], the
vertical ground reaction shall be one half of the maximum vertical load resulting
from all symmetrical landing conditions. The side load on the main gear in drift
landing is 60–80% of the vertical load. Similarly, for each landing case, the load
combinations among the vertical, drag and side loads will differ.
The landing load case information of the above-measured strain spectra was not
available. The longitudinal stress at the R5 location may be due to the bending
loads acting on the cylinder. In the absence of distance of the R5 location from the
ground based on the piston position and wheel diameter, the bending stress at R5
location cannot be calculated. Hence, in the present work, for simplicity, an
equivalent axial load instead of bending load was assumed to be acting on the outer
cylinder of the landing gear to reproduce the measured longitudinal stress. Hence,
the axial load on the outer cylinder acting along the axis of the shock absorbing
strut and the internal pressure load were considered for the landing gear static and
fatigue analysis.
From the strain spectra, stress spectra were generated using the procedure dis-
cussed in the Sect. 2.2. From the stress spectrum, initial and peak point values were
extracted and the equivalent loads were calculated. Using the calculated equivalent
loads at various time instances, static stress analyses were carried out on the landing
gear by applying both the pressure load and the longitudinal load using MSC
Nastran. The respective stress and strain results are presented in Table 2 and can be
observed that FE stress/strain values obtained from the analysis of landing gear are
in good agreement with the calculated stress/strain values. Hence, with the appli-
cation of equivalent load system, the original behaviour of the landing gear was
Table 2 Comparison of measured/calculated stress and strain values and FEA stress and strain values
Time (s) 0 1133 1605 1642
Calculated equivalent loads (internal −14 MPa and −1245 kN 22 MPa and 2814 kN −39 MPa and −3617 kN −67 MPa and −2948 kN
pressure and axial load)
Stress/strain Measured/ FEA Measured/ FEA Measured/ FEA Measured/ FEA
calculated calculated calculated calculated
Longitudinal strain (0°) (µe) −798 −797 1835 1833 −2324 −2323 −1778 −1775
Hoop strain (90°) (µe) 140 139 −412 −410 427 424 −6 −12
Maximum principal strain (µe) 141 138 1900 1833 433 421 4 −16
Fatigue Life Estimation of Typical Fighter Aircraft …
Minimum principal strain (µe) −799 −797 −442 −408 −2300 −2323 −1788 −1775
Maximum principal stress (MPa) −28 −28 390 384 −76 −77 −133 −134
Minimum principal stress (MPa) −171 −170 39 44 −496 −493 −405 −401
von-Mises stress (MPa) 159 159 371 365 462 462 358 358
Hoop stress (MPa) −28 −28 44 44 −77 −76 −134 −133
Longitudinal stress (MPa) −170 −170 385 385 −495 −493 −404 −401
47
achieved. This verified finite element model was used to carry out the fatigue
analysis using the calculated spectra.
3.1 Fatigue Life Estimation—Hoop and Longitudinal Stress

Spectra
The landing gear is one of the critical components in the aircraft, and hence, it
should be designed under the category of safe life components. If the design life of
the landing gear component is expired, then that component should get replaced
even though there is no damage. There is no repair technology used in the case of
safe life components. The component should not lose its strength or stiffness before
it gets replaced. Because of this, stress-life (S-N) approach was considered for the
fatigue analysis of main landing gear cylinder using MSC Fatigue software [10]. In
the stress-life approach, the mean stress corrections available are Goodman, Gerber
and Soderberg. Since the stress spectra calculated here are having different R ratios
and the fatigue material properties are available for R ratio ‘−1’, mean stress cor-
rections should be considered to account for the different mean stresses present in
the spectrum. Among the three mean stress corrections, Goodman mean stress
correction and Gerber mean stress correction were used here. The stresses obtained
from the finite element static analyses were considered as inputs for the fatigue
analyses. Furthermore, in the fatigue life calculations using MSC Fatigue, nor-
malization of stress spectrum approach was used. This means the given stress
spectrum was normalized using the maximum stress value of the spectrum (scale
factor to normalize stress spectrum) so that maximum normalized stress of unity is
available in the normalized spectrum. During the fatigue analysis, the stress
obtained at the point of interest in FEA gets multiplied with the unit value in the
spectrum at the maximum point. The normalized spectrum should contain the
values ranging from −1 to +1. So, we have to use the maximum stress available in
the spectrum to normalize and hence the load case corresponding to maximum
stress (−67 MPa internal pressure load for hoop stress spectrum and 3616 kN
longitudinal compressive load for longitudinal stress spectrum) was used in this
approach. Using the hoop stress spectrum and longitudinal stress spectrum, the
fatigue lives were calculated, by applying the pressure load and axial load in the
static analysis, and are given in Table 3.
In the actual landing gear, both the loads are acting simultaneously. Hence, it is
required to calculate the fatigue life by applying both the stress spectra simulta-
neously. In order to incorporate the effect of both the stress spectra in the fatigue life
calculation, linear damage rule was used here.
Table 3 Fatigue life of main landing gear based on normalized hoop and longitudinal stress
spectra
Static analysis Fatigue analysis
Applied load Stress spectrum Scale Fatigue life (landings)
factor Goodman Gerber
−67 MPa (internal Hoop stress −134 9.328 10 6
8.278 106
pressure) spectrum
3616 kN (axial Longitudinal stress −496 35,914 28,946
compressive load) spectrum
3.2 Fatigue Life Estimation—Linear Damage Rule
According to Miner’s linear damage rule, the damage occurred in the landing gear
due to the application of the stress spectra independently was calculated using the
following formula:
ni
Di ¼ ð1Þ
Ni;f
where ni denotes the number of load cycles applied, and Ni,f denotes the number of
cycles required for the component to fail under the applied load cycle. The fatigue
lives estimated using normalized hoop and longitudinal stress spectra with the
corresponding loads applied independently in static analysis are given in Table 3.
1
Damage due to hoop stress spectrum ¼ ¼ 1:072 107 ð2Þ
9:328 106
1
Damage due to longitudinal stress spectrum ¼ ¼ 2:7844 105 ð3Þ
35; 914
Damage due to the longitudinal stress spectrum is very high compared to the
hoop stress spectrum.
Accumulated Damage ¼ Damagehoop þ Damagelongitudinal ð4Þ
1 1
Fatigue Life ¼ ¼ ¼ 35776 Landings ð5Þ
Accumulated Damage 2:79512 105
The longitudinal stress is the critical stress when compared to the hoop stress in
this analysis, and hence, the accumulated damage is closer to the damage due to the
longitudinal stress spectrum. Therefore, the equivalent axial load acting on the
landing gear is more critical when compared to the internal pressure load in the
equivalent load system considered. It may be noted here that above fatigue life
estimation is based on Goodman mean stress correction.
3.3 Fatigue Life Estimation—Principal Stress Spectra
The fatigue analysis was carried out in MSC Fatigue using the calculated principal
stress spectra. The load corresponding to the maximum/minimum principal stress
value in the spectrum was applied in the static analysis, and the corresponding
spectra were used in the fatigue analysis. In the MSC Fatigue software, appropriate
stress combination and the corresponding spectra were considered for the fatigue
analysis. The fatigue life estimation using normalized spectrum was carried out, and
the results were compared with the fatigue life estimated using constant life diagram
(CLD) approach. In CLD approach [7], the S-N curves for different R ratios were
generated experimentally and they were used to calculate the life for each and every
cycle in the spectrum. The R ratio for which S-N curve is not available could be
calculated using the interpolation or extrapolation techniques. The comparison
between the fatigue lives obtained from the MSC Fatigue and CLD approach is
given in Table 4.
From Table 4, it is clear that MSC Fatigue tool with Gerber mean stress cor-
rection can be used to perform fatigue analysis and predict the fatigue life of
metallic components as it gives comparable/similar fatigue life that can be obtained
from CLD approach.
3.4 Summary of Fatigue Analyses of Main Landing Gear
The fatigue analysis of main landing gear cylinder of typical fighter aircraft was
carried out using the normalized longitudinal, hoop, maximum principal, minimum
principal, von-Mises and maximum absolute principal stress spectra (maximum
value of absolute principal stress at each and every point of time in the stress
spectrum) by considering both the Goodman and Gerber mean stress correction
methods in MSC Fatigue. The fatigue life results obtained using all the spectra are
presented in Table 5.
Table 4 Comparison of fatigue life obtained from MSC fatigue and CLD approach
Stress spectrum Fatigue life based on CLD Fatigue life based on normalized
(landings) spectrum (landings)
MSC fatigue MSC fatigue
(Goodman) (Gerber)
Load applied (22 MPa internal pressure and 2813 kN axial tensile load)
Maximum 2.28 105 1.579 105 2.357 105
principal stress
Load applied (−38 MPa internal pressure and 3616 kN axial compressive load)
Minimum 1.28 105 2.088 105 1.235 105
principal stress
Table 5 Fatigue analysis/life of main landing gear using MSC fatigue

Normalized spectrum Fatigue life (landings)
Goodman Gerber
Maximum principal stress 1.579 105 2.357 105
Minimum principal stress 2.088 105 1.235 105
von-Mises stress 6.7731 104 1.427 105
Maximum absolute principal stress 3.9397 104 3.2033 104
Linear damage rule (hoop and longitudinal stress) 3.5776 104 2.8845 104
From the fatigue analysis of main landing gear results presented in Table 5, it
may be said that the fatigue life estimated using maximum absolute principal stress
spectrum is in good agreement with the fatigue life estimated using linear damage
rule (combination of longitudinal and hoop stress spectra). Hence, it is advisable to
use either the maximum absolute principal stress spectrum or linear damage rule as
the fatigue life estimation criterion for design, since both of them yield the con-
servative fatigue life when compared to the fatigue lives from other criteria.
4 Conclusions
Fatigue life estimation of typical fighter aircraft main landing gear was explored
using MSC Fatigue for various stress spectra/criteria. In the absence of load
information on the landing gear, a methodology was proposed to calculate the
equivalent load system from measured strains and verified using static analysis of
landing gear. The approach using the normalized stress spectrum (only the signature
of the spectrum was used) along with the finite element stress distribution was
explored to calculate the fatigue life. Linear damage accumulation rule was also
used to incorporate the effect of two independent stress spectra in the fatigue life.
Fatigue life of landing gear was obtained using various stress spectra based on
stress-life approach and was compared with the fatigue life calculated using con-
stant life diagram (CLD) approach. Among the Goodman and Gerber mean stress
corrections, fatigue life estimation using Gerber mean stress correction was found to
be in good agreement with the fatigue life estimation using CLD approach.
Furthermore, maximum absolute principal stress spectrum/criterion and linear
damage rule give the conservative fatigue life and hence these criteria may be used
as fatigue life estimation criteria for design.
Acknowledgements The authors would like to thank Head, Structural Technologies Division,
CSIR-NAL and the Director, CSIR-NAL for their support and encouragement to carry out the
work reported in this paper. The authors would also like to express their thanks to the people who
contributed to this work directly or indirectly.
References
1. P. Beaudet, M. Roth, in Failure analysis case histories of canadian forces aircraft landing
gear components. Landing Gear Loads, AGARD Conference Proceedings, 484 (1990)
2. M. Khondker, A.K.M. Lutful Kabir, A. Younes, Md Shelimuzzaman, S. Islam, in Landing
Gear Shock Absorber Design, Concordia University, Summer 2009
3. M. Imran, R.M. Shabbir Ahmed, M. Haneef, Static and dynamic response analysis for landing
gear of test air crafts. Int. J. Innovative Res. Sci. Eng. Technol. 3(5) (2014)
4. V.N. Divakaran, G.V.V. Ravi Kumar, P. Srinivasa Rao, Aircraft landing gear design and
development, Infosys, 2015
5. Altair HyperMesh software, version 12.0, Altair HyperWorks
6. MSC Nastran solver, version 2012, MSC Software
7. C.M. Manjunatha, Inflight strain data acquisition and fatigue life assessment of aircraft
landing gear (Project Report, CSIR-NAL, 2013)
8. Military Specification—Airplane Strength and Rigidity Ground Loads for Navy Acquired
Airplanes, MIL-A-8863B (AS), 6 May 1987
9. Part 25: Airworthiness Standards: Transport Category Airplanes Special Federal Aviation
Regulations, SFAR No. 13, SFAR No. 109
10. Fatigue module in MSC Patran, version 2001, MSC Software
Designing for Strength: ASME’s DBA
Guidelines and Beyond
Kotur S. Raghavan
Abstract Finite element stress analysis is carried out with the aim of assessing the
strength of structural and other load-bearing components. It is an essential feature of
structural mechanics that the stress values cannot be taken at their face value.
Damage which can be caused by stress of a given magnitude is strongly dependent
on the nature of stress distribution and the loads causing them. These issues are
addressed in ASME’s Boiler and Pressure Codes, Section VIII, Div. 2 which is
referred to as Design by Analysis (DBA) guidelines. In this paper, the guidelines
are assessed by carrying out elastic-plastic analysis.
Keywords Design Strength ASME DBA Stress linearization
1 Introduction
In a great majority of cases, we use finite element analysis for structural or machine
design whether it is the case of new design development or verification of an
existing design. Further, in a significantly high percentage of the studies that are
carried out focus is on strength. In relatively smaller percentage of studies, the
focus is on other design considerations such as free vibration, buckling, transient
dynamics, creep.
Historically speaking, structures and machines have been built from time
immemorial. Many have withstood the test of time. In comparison, finite element
method can be considered to be a baby. The method came into existence exactly
sixty years ago with the development of constant strain triangle (CST), which is
considered to be the mother of all finite elements. Widespread application of the
method, facilitated by commercially available analysis packages, is still more recent
being not more than four decades old.
K. S. Raghavan (&)
Cyient Limited, Gachibowli, Hyderabad 500032, India
e-mail: kotur.raghavan@cyient.com

https://doi.org/10.1007/978-981-10-7197-3_5
54 K. S. Raghavan
Having said this, it needs to be stated here that finite element method has not
made great inroads into design practices. Design approaches vary depending on the
industry practices which are product dependent. Design by rules based of ele-
mentary mechanics and strength of materials are still in vogue in many industries.
Boiler and pressure vessel (B and PV) codes of American Society of Mechanical
Engineers (ASME) are a case in question. The present paper is concerned with finite
element analysis and design for strength. The paper centres on ASME’s B and PV
codes. However, the contents and concepts are of general nature.
2 ASME’s Boiler and Pressure Vessel Codes
American Society of Mechanical Engineers is world’s premier engineering insti-

tution. It was founded about 130 years ago. The society’s B and PV codes are about
a century old. ASME came out with mandatory design guidelines for all pressure
vessels used in process and power industry. The first edition contained a little over
one hundred pages. The codes have gone continuous evolution ever since, and as of
today, the codes span more than twenty thousand pages in nearly thirty volumes.
The codes cover all aspects related to design, manufacturing and material selection
and testing. Of particular, relevance here is the Section VIII of the codes. This
section has three divisions. Division 1 provides guidelines for design by rules.
Division 2 and Division 3 provide guidelines for design by analysis (DBA).
Division 2 was released in mid-sixties, and this was developed as it was realized
that Div. 1 rules were too conservative and the components were becoming too
bulky.
There are two routes in Div. 2. They are the linear or elastic route and the
nonlinear or elastic-plastic route. ASME recommends the use of latter to overcome
any ambiguities or conflicts. ASME further recommends that decisions based on the
elastic-plastic approach override those based on the elastic route. Thus, on the
whole, there are three approaches available to the designer: one of Div. 1 and two of
Div. 2. They are design by rules (DBR), design by analysis-elastic (DBA-E) and
design by analysis-elastic-plastic (DBA-EP). Conceptually, DBR is based on peak
stress values. The guidelines are akin to what we normally find in text books on
machine design and structural design. DBA-E is based on the principle of “damage
potential” of stresses. The damage caused by a stress of given magnitude is a
function of the nature of stress distribution and the loads causing that stress. The
guidelines consider section on the whole. On the other hand, DBA-EP takes into
account stress redistribution effects.
Designing for Strength: ASME’s DBA Guidelines and Beyond 55
3 Earlier Work and Scope of the Present Work
In an earlier work [1], the present author addressed the issues concerned with DBA
codes. The paper focused on a few ambiguities, inadequacies and limitations of the
elastic analysis guidelines. On the basis of a number of case study problems, a
strong case was made for making the nonlinear route (DBA-EP) mandatory instead
of giving option to the designer. One of the reasons for arriving at this recom-
mendation was that nonlinear structural analysis is not a challenge in the present
day scenario.
The main recommendation of the cited paper still holds. In the current work,
more studies are carried out which provide more insights into the guidelines of
DBA. These findings further reinforce the earlier recommendations.
4 Design by Analysis
The essence of DBA-E guidelines is classification of stress into membrane–bend-

ing–peak and primary–secondary categories. Different categories of stress will have
different allowable stress values. Membrane stress caused by primary (mechanical)
loads has the highest damage potential, and hence, the allowable stress for this
category is the lowest (say S). The allowable bending stress is 1.5 * S, and
allowable secondary stress is 3 * S. Secondary stresses are self-limiting stresses
generally caused by thermal loads and structural discontinuities.
There is no issue as far as membrane stress is concerned. It is the uniform
component of the stress across a section (thickness) also equal to the average stress
across the section. There are, however, three different definitions for the so-called
bending stress to be found in the literature. The three definitions are as follows:
1. Bending stress is the non-uniform component of the stress across a section. The
variation may or may not be linear.
2. Bending stress is the linearly varying component of stress across a section. Also,
bending stress is that component of stress which is proportional to the distance
from the centroid of the section.
3. Bending stress is the linearly varying component of stress if peak stress (stress
raiser) is present. Otherwise, it is equal to the difference between maximum
stress and the membrane stress.
ASME’s Section VIII codes of 2007 and later [2] contain definitions 1 and 2. It
is notable that only the first definition is to be found even in the latest edition of
Section III (Nuclear, Class 1). In relevant documents of ANSYS [3], the second
definition is to be found. This definition is not precise as one faces uncertainties in
transition regions. In documents related to ABAQUS [4] program, we find the third
definition. This definition is essentially same as the first one but without vagueness
as it makes distinction between two different scenarios.
56 K. S. Raghavan
The foregoing definitions are easily realized in two-dimensional structures (plane

stress or axisymmetric). Both ANSYS and ABAQUS programs contain capabilities
for computing membrane and bending components along specified paths using
procedure known as “linearization”. The calculations are just algebraic, and the
programs compute for any path whether meaningful or not. In fact, one can easily
carry out linearization using Microsoft EXCEL program. One has to use engi-
neering judgment in order to extract numbers which make sense. Thus, one of the
limitations of the elastic route guidelines is that it is difficult to apply for
three-dimensional domains.
5 “Bending” Stress and Strength
In this section, the effect of bending, actually the maximum value of membrane plus
bending stress on the component stress is assessed under different geometry con-
ditions. Only two-dimensional geometries are considered, and use is made of lin-
earization capability available in ANSYS. Strength is assessed by carrying out
elastic-plastic analysis. Materials are assumed to follow elastic-perfectly plastic
rule. The strength is evaluated by taking into account the response trend, divergence
behaviour and nature of stress distribution at critical regions.
Strength is assessed with DBA guideline as reference. As already stated, If S is
the permissible (von Mises) stress for primary membrane (M) component, then the
allowable membrane plus bending component (MB) is 1.5 * S. This is based on the
fact that a section yields only when the entire section and not just the extreme fibres
becomes plastic.
5.1 Parabolically Tapering Cantilever Beam
As a first case study, a parabolically tapering cantilever beam with tip load is
considered. Geometry and elastic analysis results for unit load are in Fig. 1. The
figure also contains the stress contours and linearization plot. This is case of pure
bending. Maximum unit stress under unit load is 0.3. Elastic-plastic analysis is next
carried out. The material yield stress is 200 MPa. The section is expected to yield
when the extreme fibre stress reaches 300 MPa. The tip load at section yield is
1000 N. Actually, the load applied is twice this value.
Nonlinear response is shown in Fig. 2. Here, the X-axis denotes tip load
non-dimensionalized with respect to elastic limit load which is 1000 N. It is seen
that the structure has not reached instability even at twice the elastic limit load
although it is obvious that plastic flow has set in at non-dimensional load of unity.
We also see that the beam has undergone very large deformation at full load.
A study of two-colour stress contours shown in Fig. 3 is informative. At full
load, a very small part of the span has become fully plastic. Because of the
Fig. 1 Tapering cantilever beam, elastic analysis results
Fig. 2 Nonlinear response and deformation at full load
Fig. 3 Two colour von Mises stress contours
geometry, there is lot of stress redistribution taking place along the length and that
is why the beam has not reached instability in spite of very large deflection. From
strength point of view, however, load corresponding to non-dimensional load of
unity is of design significance.
5.2 Perforated Ring
Half sector of a perforated ring subject to internal pressure is the next case study
problem. Geometry, elastic stresses and linearized curves are shown in Fig. 4.
These are for unit pressure application. There are two load paths here, and the inner
path is more highly stressed. MB component is a little more than 1.5 times the
58 K. S. Raghavan
Fig. 4 Perforated ring, diameter ratio = 0.5. Geometry and stress description for unit pressure
membrane component. Going by DBA guidelines, the strength is governed by MB

component. Pressure for nonlinear analysis is appropriately chosen.
Results of nonlinear analysis are presented in Fig. 5. Most importantly, insta-
bility is seen to set in at a non-dimensional pressure of 1.2 and not unity. A part of
the reason for this is the outer ligament, which is also a load path, supporting the
load almost equally at failure or instability. The other reason is that MB components
tend to be overestimated along sections involving peak stress. Another reason is
that the classification line is across a partial load path, and bending stress about the
midpoint may not be physically meaningful.
For this configuration, the pressure limit (for YS = 200 MPa) based on nonlinear
analysis is (300/5.9 * 1.2) which works out to about 61 MPa. Based on DBA-E, the
pressure limit is about 51 MPa. A designer or an analyst without recourse to (or
knowledge of) linearization or nonlinear analysis would perhaps estimate on the
basis of highest stress magnitude which is 8 MPa (Fig. 4) and the corresponding
pressure limit would be 25 MPa.
In order to get a little more insight, parametric results are presented in Table 1.
The parameter is the ratio of whole diameter to the width. The second column
numbers signify whether membrane or MB component governs the strength
depending on whether the number is less than or greater than 1.5. The last three
columns denote the strength in terms of pressure. The fourth column values are
based on DBA elastic route. The last column values refer to membrane basis
assessment. The message of this table is that membrane stress has dominant
influence on strength. The pressure limits computed by nonlinear analysis are
higher by 10–20% because of the load sharing between the inner and outer ligament
sections.
Fig. 5 Nonlinear response. Contours at theoretical and actual failure loads
Table 1 Perforated plate D/W MB-U/ Pressure limits

parametric analysis results ratio M-U DBA-EP DBA-E DBA-E-MEMB
0.3 1.24 82.40 74.91 74.91
0.4 1.41 71.38 64.31 64.31
0.5 1.57 60.61 50.51 52.91
0.6 1.71 49.39 36.32 41.41
5.3 Analysis of Fillet Welds
A common fillet-welded configuration is shown in Fig. 6. The structure is obliquely

loaded as shown, with angle with the horizontal being an analysis parameter. The
bottom welded will be more highly stressed for angles greater than zero. Linearized
stresses are typically as shown on the right-hand side of the figure for the path
chosen along the path shown. The peak stress is spurious and is due to stress
singularity.
Fig. 6 Structural member with fillet welds

60 K. S. Raghavan
Fig. 7 Nonlinear analysis results for three load orientations
Elastic-plastic analysis results for three load orientations are presented in Fig. 7.
For horizontally applied load, membrane plus bending component gives conser-
vative results. For other two cases, it is seen to be slightly non-conservative. It is
notable that failure planes are closer to the horizontal line. They are away from the
throat (45° plane) which is considered in design (also in this problem for
linearization).
5.4 Tension Members with Groove
The last problem in this section pertains to a tension member with a groove. The
geometry and loading are as shown on the left-hand side of Fig. 8. Two cases are
considered. The first case pertains to plane stress condition (strip). The second case
is that of axisymmetry with XX as the axis of symmetry (grooved shaft).
Non-dimensional depth (=D/W) is a parameter.
Stress distribution along the critical path is typically as shown on the right-hand
side of Fig. 8. In all the cases, (M + B) component was more than 1.5 times the
membrane component. Hence, strength is governed by (M + B) component.
Analysis was carried out for four values of non-dimensional depth of groove, and
consolidated results for plane stress and axisymmetric cases are presented in Fig. 9.
Here, the ordinates correspond to non-dimensional limit loads from nonlinear
analysis. A value of unity corresponds to DBA-E limits. For the plane stress
condition, the bending stresses are genuine ones. The bottom curve is thus genuine.
The strength is within ±10% of the DBA-E limit. The same is not true, however,
Fig. 8 Tension member with a groove
Fig. 9 Non-dimensional load

limits for grooved bar
for the axisymmetric condition. By definition, bending stress is the component of

stress which is proportional to the distance from the centre of the path. In
axisymmetry condition, there is no such deformation taking place. The bending
stress which is computed in axisymmetric cases is due to non-uniformity of stress
across the section resulting from stress concentration and thus is spurious. Any
design evaluation involving bending stress in axisymmetry conditions will result in
undue conservatism. This is a very important point which is often overlooked by
analysts and designers. ASME codes are also silent about this aspect.
62 K. S. Raghavan
6 Other Concerns
One of the provisions of DBA-E guidelines is related to “secondary” stresses. They

are caused by thermal loads and geometric discontinuities. Secondary stresses are
classified as “self-limiting”, and allowable value is more as compared to those for
mechanical loads (three times the value allowable for primary membrane). The
main concern is that the allowable limit for secondary stress has no theoretical
basis. This has been addressed by the present author in Ref. [1] where it was
established that stresses due to thermal loads cause no plastic flow. In a structure
subject to pure thermal load, response will be linear well beyond the design limit as
shown at the left of Fig. 10.
Another concern is related to limited applicability for problems involving 3D
elements. In general, it can be stated that the only way of assessing the strength of
3D solid structures is elastic-plastic analysis. Using nonlinear analysis, it is possible
to develop interaction curves for multi-axial load conditions like those shown on the
right side of Fig. 10. This figure pertains to biaxial bending of a beam of square
cross-section. The figure contains three interaction lines. The outermost is generated
by way of nonlinear analysis. The middle one is DBA-E based and considers plastic
moment into consideration but not redistribution. The innermost line is based on
strength of materials using only the yield moment. This is the most conservative.
This figure graphically presents relative impacts of the three approaches available to
the analyst according to Section VIII, Divisions 1 and 2. It is worth stating here that
“design by rules” and machine design textbook procedures are akin to one another.
Fig. 10 Left: response to thermal load, right: biaxial bending, interaction curves
7 Summary
The purpose of the paper was to assess and evaluate DBA-E guidelines. Main focus
has been on the meaning, application and interpretation of bending stress. A few
simple case studies have established that the bending stress is not meaningful in all
situations. Caution is needed while taking design decisions where in “bending”
stresses happen to be critical. In situations where it is meaningful, the design may
become slightly non-conservative. In situations where it is not meaningful, the
design may become over-conservative.
Recourse to nonlinear route is highly recommended for all design problems.
8 Further Work
In this work, the strength is evaluated by joint interpretation of response pattern,

nature of stress distribution at critical sections and divergence trends. ASME’s
codes, on the other hand, suggest consideration of only the collapse load. There
may be some ambiguities in this regard. This aspect remains to be addressed.
References
1. Raghavan, K.S, in An Evaluation of ASME’s Design by Analysis Guidelines. Proceedings

ASME’s GTINDIA Conference, (Bangalore, 2013)
2. American Society of Mechanical Engineers, Boiler and Pressure Vessel Code, Section VIII,
Division 2, 2010
3. ANSYS Mechanical APDL User’s Manual, ANSYS Inc., www.Ansys.com
4. ABAQUS/CAE User’s Manual, http://www.3ds.com/products-services/simulia/products/
abaqus/
Premature Fatigue Failures in the Hot
Zone of Low Pressure Turbine Rotor
(LPTR) Blades of Aero-Engine
M. Madan, R. Bharathanatha Reddy, K. Raghavendra, M. Sujata

and S. K. Bhaumik
Abstract Generally, the fatigue failures in gas turbine blades occur at 1/3rd height
of the airfoil, where the bending stresses on the blades are maximum. In contrast,
the creep and the stress-rupture damages are observed at around 2/3rd height of the
blade, which is the maximum temperature zone on the leading edge. However,
irrespective of the mechanism of crack generation, the major part of the crack
propagation in the turbine blades takes place by fatigue mechanism because of the
cyclic nature of the loads. Therefore, identification of the primary mechanism of
crack initiation while analyzing these failures is of prime importance. Unless this is
established unambiguously, identification of the primary cause of failure is not
possible. In this paper, analysis of a prematurely failed low pressure turbine rotor
(LPTR) blade is presented. Although the fracture in the blade occurred by fatigue
mechanism, the initial crack initiation in the blade was found to be by stress-rupture
mechanism. Detailed metallurgical investigation showed that the stress-rupture in
the blade was promoted due to overheating of the blades for short durations. The
reason for overheating of the blades was due to failure elsewhere in the system
which resulted in faulty feedback on the actual engine temperature during
operation.
Keywords Superalloy Gas turbine blades Overheating Stress-rupture

Fatigue
1 Introduction
Premature failures in gas turbine blades of aero-engine are of serious concern for
the reliability and safety of the aircraft. In most of the cases, the specific mechanism
of failure is dependent on either the defects in the material that have been present in
the blade from the manufacturing stage or those generated during the service life.
M. Madan R. Bharathanatha Reddy K. Raghavendra M. Sujata (&) S. K. Bhaumik

Materials Science Division, CSIR-National Aerospace Laboratories, Bangalore 560017, India
e-mail: msujata@nal.res.in

https://doi.org/10.1007/978-981-10-7197-3_6
66 M. Madan et al.
Most common failure mechanisms in the gas turbine blades are fatigue and creep
rupture [1–4]. There are numerous studies on the effect of service exposure on the
microstructure and mechanical properties of the blade materials [5, 6]. Some of the
microstructural changes that influence the creep and fatigue strength of the blades
include excessive precipitation of carbides and/or brittle topological close-packed
(TCP) phases, rafting of c′ precipitate phase, and void formation at the triple or
grain boundaries. The extent of microstructural changes that occurs in the blade
material depends on time, temperature, and stress during service. In this paper,
failure of low pressure turbine rotor (LPTR) blades of an aero-engine has been
reported. The blades were manufactured from Nimonic 115 alloy by closed die
forging.
2 The Failure
There was an incident of engine surge in a fighter aircraft. On strip examination of

the engine, one of the LPTR blades was found fractured. The LPTR blades were in
service for 1123:50 h against a total technical life of 1200 h. Examination revealed
that the fracture in the airfoil occurred at a distance of about 20 mm from blade
shroud. Only a small part of the fractured and dislodged part of the blade was
retrieved from the engine during strip examination. Following the fracture, the
dislodged part of the blade resulted in severe damages to the engine. Examination
of the engine revealed cracks in a few more LPTR blades. The fractured and
cracked LPTR blades along with a HPTR blade were subjected to detailed metal-
lurgical failure analysis (Fig. 1) for establishing the primary cause of failure.
(a) (b)
1 2 1 2
20 mm
Fig. 1 Failed LPTR blades (marked 1 and 2) and the HPTR blade of the aero-engine: a concave
surface, and b convex surface
Premature Fatigue Failures in the Hot Zone … 67
3 Laboratory Investigation
3.1 Macroscopic Features of Fractured LPTR Blade
Figure 2a shows the fractured LPTR blade. The blade had fractured in the airfoil
section at a height of about 50 mm from the blade root platform. Examination did
not reveal the presence of any incipient cracks in the airfoil or leading edge or
trailing edge. Apart from the fracture, the blade did not have any other damages.
Figure 2b shows the fracture surface of the blade. Examination of the fracture
surface at low magnification showed that the crack had initiated at the leading edge
and there were three distinctive fracture zones. Over a length of about 4 mm from
the leading edge, the fracture surface was irregular in appearance. Following this,
the fracture surface showed a smooth appearance over a length of 2.5 mm with a
semi-circular crack front, typical of progressive crack propagation by fatigue
mechanism. The rest of the fracture surface showed intergranular mode of crack
propagation, typical of overload failure in superalloy components at elevated
temperatures. In the first two zones of crack propagation, that is, over 6.5 mm
length from the leading edge, the fracture surface was found severely oxidized
compared to the rest of the fracture surface.
3.2 Mode of Crack Propagation and Fracture Mechanism
Typical appearance of the fracture surface in the zone I is shown in Fig. 3a, b. At
the leading edge, over about a length of 250 lm, the crack propagation was purely
by cleavage mode. Following this, the crack propagation was by mixed mode of
transgranular and intergranular fracture, and this region extended up to a length of
about 600 lm. The remaining part of the zone I revealed intergranular fracture
features (Fig. 3b). The fracture features in zone II showed predominantly trans-
granular mode of crack propagation. Vaguely delineated crack arrest marks with
semi-circular crack fronts were found present in this zone. Due to severe oxidation,
the micro-fractographic features were not clearly discernible. However, the gross
fracture features confirmed that the crack propagation in this zone was by
(a) (b)
Zone II
Zone I (2.5 mm)
(4 mm)
Zone III
50 mm (12.6 mm)
Fig. 2 a Fractured LPTR blade, b fracture surface showing three zones of crack propagation
68 M. Madan et al.
(a) (b)
Zone II
Zone I
(c) (d)
Fig. 3 Scanning electron fractographs: a zone I and II, b magnified view showing transgranular
cleavage fracture up to a depth of 250 lm, c crack arrest marks in zone II, and d intergranular
fracture in zone III
progressive fatigue mechanism. A transition in the fracture mode from zone II to

zone III involving transgranular mode of crack propagation to intergranular mode
of crack propagation was clearly identified. In the zone III, the crack propagation
was purely by intergranular mode. Fine dimples on the grain facets were also
observed in this region (Fig. 3d). Such features are typical of overload fracture in
superalloys at elevated temperatures.
3.3 Microstructural Examination and Phase Analysis
3.3.1 Optical Microscopy
A transverse section was cut from the fractured LPTR blade at approximately 2 mm
below the fracture and metallographically prepared for microstructural study.
Figure 4 shows the optical microstructures of the blade at the leading and trailing
edges. The microstructures consisted of c′ precipitate phase (Ni3Al) in a Ni-rich c
matrix. The blade surface showed the presence of an oxide layer, and this was
followed sequentially by a diffused zone, a thin c′ depleted zone, and the base
material of the blade.
(a) (b)
γ′ depleted zone
Oxide layer and diffused

zone
400 μm
40 μm
Fig. 4 a, b Optical microstructures of the leading edge of the fractured LPTR blade at different
magnifications
3.3.2 Scanning Electron Microscopy
Figure 5 shows the secondary electron microstructures at the leading and trailing
edges of the blade. The c′ depleted zone between the diffused layer and the base
material of the blade can be seen. In the base material of the blade, no abnormal
microstructural changes such as complete dissolution or excessive coarsening of
either the c′ precipitate phase or the carbide phases were observed. A similar
observation was also made on the concave and convex surfaces of the blade. The
dimensions of the different zones of microstructure, that is, oxide and diffused
layers and c′ depleted zone at different locations of the blade cross section were
measured. The study revealed that the thickness of the oxide layer and the diffused
layer together was significantly less at the leading edge compared to that a little
away and on the concave and convex surface sides of the leading edge. The
presence of oxide layers at these locations suggested that there was continuous
erosion and formation of the oxide layer on the surface of the leading edge during
the service. The microstructures of the broken blade were compared with those of
the standard microstructures provided by the engine manufacturer. A comparative
Fig. 5 Secondary electron micrographs of LPTR blades: a leading edge, and b trailing edge
70 M. Madan et al.
study showed that the microstructural changes observed in the hotter section of the
blade after 1123:50 h of service were within the acceptable limit. The
microstructures did not suggest running of the engine persistently well above the
normal operating temperatures.
3.3.3 Analysis of Phases Present at the Blade Surface
Analysis showed that the oxide layer on the convex and concave surfaces of the
LPTR blades consisted of a mixture of oxides, namely, Al2O3, Cr2O3, and/or TiO2
(Fig. 6a). The metallic phases within the diffused zone were found to be rich in Ni
and Co and depleted in Al and Cr compared to the base material (Fig. 6b). In
contrast, the leading edge of the blade showed the presence of a very thin diffused
Fig. 6 Energy dispersive X-ray (EDX) spectrums: a oxide layer on convex surface, and b c′-
denuded layer
layer, and the outer surface was found to contain predominantly a Si–Ca–O rich
layer.
3.4 Cracked LPTR Blade
Figure 7a shows the cracked LPTR blade received for investigation. Examination
revealed that the blade had developed two cracks at the leading edge located at 52
and 48 mm distance from the blade root platform. The length of the cracks on the
convex surface was measured to be 1.5 and 2.0 mm, respectively. The cracks were
pulled open for further study, and a typical fractograph of the resulting fracture
surfaces is shown in Fig. 7b. Examination revealed that the crack initiation and the
propagation were by mixed mode of transgranular and intergranular fracture. There
was no region on the crack surface which could be indicative of fatigue mode of
crack propagation. The crack surfaces were found covered with thick oxide layers
suggesting presence of the cracks over a prolonged period of time. In the freshly
opened region, the fracture surface showed faceted, intergranular features with fine
dimples on the grain facets.
3.5 HPTR Blade
Visual observation did not show any damages on the HPTR blade (Fig. 1).
However, examination of the blade at low magnification showed severe surface
oxidation at the hotter region on the leading edge. The blade was subjected to
detailed metallurgical investigation to examine whether or not there were any
abnormal microstructural changes in the material of construction.
(a) (b)
Fig. 7 a Cracks on the leading edge of the HPTR blade, and b fracture surface showing
transgranular cleavage
72 M. Madan et al.
3.5.1 Microstructure, Thermal Barrier Coating, and Composition
The microstructures of the blade at the root region and the hot zone are shown in
Fig. 8. Observations revealed coarsening of the c′ precipitate phase at the hot zone
of the blade. Figure 8 also shows the thermal barrier aluminide coating on the
leading edge of the blade at the root region as well as in the hot zone. Examination
showed rafted microstructure in the hot zone of the blade on the leading edge
(Fig. 8b). Precipitation of TCP phases was also observed at a few places adjacent to
the aluminide coating (Fig. 8c, d). The coating thickness including the diffusion
zone was measured to be about 50 and 90 lm at the root region and hot zone of the
blade, respectively.
Fig. 8 Secondary electron microstructures at the leading edge of HPTR blade showing a, b c′ in a
c-matrix, and c, d thermal barrier coating; a, c root region, b, d hot zone
4 Analysis of Failure
4.1 Failure Mechanism
Fracture surface of the broken LPTR blade showed three stages of crack propa-
gation, namely (i) crack initiation by mixed mode of transgranular and intergranular
fracture, (ii) progressive fatigue crack propagation by transgranular mode, followed
by (iii) final fracture by intergranular mode. These fractographic features confirmed
that the failure in the blade was progressive in nature. Laboratory studies showed
that there were two distinct mechanisms for the crack initiation and the subsequent
crack propagation. From the fracture features, it could be clearly established that
following the initiation, further crack propagation was by fatigue mechanism. The
fatigue crack propagated to a critical length before giving way to final overload
fracture.
In the cracked blade, there were multiple crack initiations at the leading edge at a
height of 48–52 mm from the blade root platform. This region is the hot zone of the
blade, and the mode of crack initiation in this blade was similar to that observed in
the fractured blade. In this blade, the crack surface displayed mixed mode of
fracture up to a depth of 2 mm.
4.2 Micromechanism(s) of Crack Initiation
In both the LPTR blades, the crack initiation occurred at the leading edge at a height
between 48 and 52 mm from the blade root platform, which is essentially the hotter
region of the blade. This clearly indicates that the crack initiation in the blades was
by a thermally activated process. Examination of the remaining LPTR blades of the
engine revealed the presence of cracks in seven more LPTR blades. In all the seven
blades, cracking occurred at the hotter region of the leading edge, located in and
around a height of 50 mm from the blade root platform. These observations sug-
gested that the failure was not blade specific. Rather, it was because of reason(s)
which was global in nature and related to the engine operating conditions.
The predominant mechanisms for temperature-assisted failures in superalloy
turbine blades are (a) creep, (b) stress-rupture, (c) thermal fatigue, and (d) thermal
shock. However, each of these mechanisms of failure has characteristic features
which can be distinctively identified through fractography and microstructure
studies. In the context of the present failure, these are discussed below.
• Creep: Generally, the creep failure in turbine blades is associated with detectable
deformation surrounding the crack initiation region. Also, during creep defor-
mation, grain-boundary sliding results in the generation of wedge-shaped cracks
at the triple points, which can be readily identified through metallographic
examination. Examination of the LPTR blades in the present case did not show
74 M. Madan et al.
any such characteristic features, and hence, the possibility of failure by creep
mechanism is ruled out.
• Stress-rupture: The stress-rupture fracture may be macroscopically either brittle
or ductile in nature. Brittle fracture is intergranular and occurs with no elon-
gation or necking. Ductile fracture is transgranular and is accompanied by
discernible elongation or necking. But superalloy gas turbine blades are gen-
erally designed based on about 1% creep, and hence, they show low
stress-rupture ductility. As a result, the fracture surface in this case often shows a
mixture of transgranular and intergranular fracture features. Also, stress-rupture
results in fracture surfaces which are generally irregular in nature, and they show
uniform discolouration or oxidation. The fracture features seen at the leading
edge in the present LPTR blades resemble these characteristics to a large extent.
• Thermal shock: Crack generation by thermal shock occurs in the hot zone of the
blade and may occur on either the trailing edge or the leading edge. Generally,
such cracking takes place in multiple blades in a similar fashion. The crack
surface in this case is generally irregular in appearance with a mixed mode of
transgranular and intergranular fracture. Thermal shock may result in generation
of closely spaced multiple cracks in a single blade.
4.3 Primary Cause for Crack Initiation
Fracture features and microstructural examination clearly established that the crack
initiation in the LPTR blades was by a thermally activated process, involving either
thermal shock or stress-rupture. The crack initiation in the blades by these mech-
anisms would have been possible only if there were variations in the engine
operating temperatures in excess of the normal. In Ni-base superalloys, the frac-
tographic features on the crack surfaces generated by thermal shock and
stress-rupture are very similar and not clearly distinguishable [4]. However, at
times, it may be possible to distinguish between these mechanisms of failure
through careful analysis of the fractographic features and microstructural changes in
the material of construction together.
4.3.1 Oxide Layer on the Blades Surface
In the present case, uncoated blades made of Nimonic 115 alloy were used in the
low pressure turbine section of the aero-engine. The surface oxidation of the LPTR
blades during engine operation is, therefore, normal. At high temperatures and in
oxidizing atmosphere, a surface oxide layer is formed on the LPTR blades. The
primary constituents in the surface oxide layer are Al2O3, Cr2O3, and TiO2. The
elements Al and Cr diffuse out of the blade material at the surface to form these
oxides, resulting in a diffused zone in the blade comprising base material depleted
in Al and Cr interspersed with oxide phases Al2O3 and Cr2O3. Hence, it is often
difficult to assess the possibility of over-temperature in the blades unambiguously
based on the characteristics of the oxide layer on the blades’ surface.
4.3.2 Microstructural Changes in the Material of Construction
Microstructural changes do take place in the turbine blades, especially in the hotter
regions of the blade airfoil, during engine running. Although, these changes are
unavoidable even under the normal operating conditions, they do not reduce the
material properties beyond acceptable limits within the technical life of the blades.
Overheating of the blades beyond acceptable limit, if any, results in abnormal
microstructural changes in the material which are identifiable through microstruc-
tural examination. In the present case, considering that the LPTR blades were in
service for 1123:50 h, the microstructural changes observed were within the
acceptable limit as per the engine overhaul manual. Also, there were no abnormal
changes in the microstructure of the blades which could be suggestive of persistent
running of the engine well above the normal operating temperatures. But, the
presence of rafting and TCP phase in the HPTR blade material adjacent to the
coating clearly indicated possible over-temperature in the engine.
4.4 Follow-Up of the Laboratory Investigation
As a follow-up of the laboratory investigation, the manufacturer subjected the

aero-engine to detailed inspection. During this time, it was found that two out of the
six thermocouples used for measuring the exhaust gas temperature were broken. It
is important to note that the engine works on a feedback system wherein the fuel
supply to the engine is controlled based on the exhaust gas temperature. In the
present engine, the average value of the temperatures measured by the six indi-
vidual thermocouples was the input. Because of the failure of the two thermo-
couples, the average temperature of the exhaust gas fed to the feedback system was
less than the actual temperature. Hence, more fuel was supplied to the engine, and
as a result, the engine temperature was naturally higher than that of the normal
operating temperature for a substantial period of time. This was further confirmed
through study of the flight data recorder (FDR). Following this study, it was pos-
sible to conclude that the failure in the LPTR blades was by stress-rupture mech-
anism as a consequence of engine over-temperature.
76 M. Madan et al.
5 Conclusions
(a) The fracture/cracking of the LPTR blades occurred by stress-rupture. In the

broken blade, after initiation, the crack had propagated by fatigue mechanism
before giving way to final overload fracture.
(b) The crack initiation in the LPTR blades by stress-rupture was due to persistent
over-temperature in the engine and not due to any metallurgical deficiencies in
the material of construction.
(c) The primary reason for the over-temperature in the engine was due to failure in
two of the six thermocouples that were used for measuring the engine exhaust
temperature. The faulty feedback to the engine operating system due to failure
of the thermocouples led to more fuel supply to the engine resulting in
over-temperature.
References
1. J. Hou, B.J. Wicks, R.A. Antoniou, An investigation of fatigue failures of turbine blades in gas
turbine engine by mechanical analysis. Eng. Fail. Anal. 9, 201–211 (2002)
2. J. Blachnio, W.I. Pawlak, Damageability of gas turbine blades-evaluation of exhaust gas
temperature in front of the turbine using a non-linear observer, in E. Benini (ed.) Advances in
Gas Turbine Technology. InTech. ISBN: 978-953-307-611-9 (2011)
3. D.P. Walls, R.E. deLaneuville, S.E. Cunningham, Damage tolerance based life prediction in
gas turbine engine blades under vibratory high cycle fatigue. J Eng. Gas Turbines Power 11,
143–146 (1997)
4. T.M. Maccagno, A.K. Koul, J.P. Immarigeon, L. Cutler, R. Allem, G. L’Esperance, Met
Trans A 21A, 3115–3125 (1990)
5. Z. Shi, S. Liu, X. Wang, X. Yue, J. Li, Effect of melting temperature on the microstructure
stability of a Ni-based single crystal superalloy. Proc Eng. 99, 1415–1420 (2015)
6. X.W. Jiang, W. Wang, G. Xie, H. Li, L.H. Lou, J. Zhang, The effect of long term thermal
exposure on the microstructure and stress rupture property of a directionally solidified Ni-based
superalloy. Met. Trans. A 45A, 6016–6026 (2014)
Fatigue Life Prediction of Composite
Airframe Panel
P. K. Sahoo, Shriram Gujar and M. Manjuprasad
Abstract Fatigue failure analysis of pristine composite laminates and composite

laminates with circular cut-outs made of CFRP T300/5208 with stacking sequence
{0}5s, {90}5s, {0/90/0/90/0}s and {+45/−45/0/90/0}s are carried out. Fatigue
damage model using physics-based approach based on multicontinuum theory and
kinetic theory of fracture considering matrix cracking failure criterion is developed
to calculate the fatigue life of above-mentioned composite laminates. Following the
above model, S-N curves are derived for different stress ratios for the pristine
composite laminates and laminates with circular cut-outs. The proposed fatigue
model has been validated by comparing the predicted results with experimental
results available in the literature. The results show that the fatigue life to failure of
pristine composite laminates is more as compared to the laminates with circular
cut-outs. For different stress ratio values, the stress-life (S-N) curves move upward
indicating that fatigue life is more as the stress ratio decreases for all the laminates.
The physics-based approach can be used as an alternative approach for predicting
fatigue life of composites. This analytical approach minimizes expensive testing
activities significantly. The studies are important and essential to evaluate the
structural integrity of composite airframe structures.
Keywords CFRP composite laminates Fatigue life Matrix cracking

Multicontinuum theory Kinetic theory of fracture
1 Introduction
In new generation civilian/military aircrafts, polymer matrix composite

(PMC) materials such as CFRP/GFRP are widely used in aircraft structures such as
fuselage, wing and control surfaces due to their several benefits and abilities to be
P. K. Sahoo (&) S. Gujar M. Manjuprasad

Structural Technologies Division, CSIR-National Aerospace Laboratories,
HAL Airport Road, Bangalore 560017, India
e-mail: pks@nal.res.in; sahoopradeepk1@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_7
78 P. K. Sahoo et al.
tailored for specific stiffness and strength properties requirements. They offer
superior performances over the traditional materials such as Aluminium alloy (Al
2024-T3) in terms of lightweight, high specific strength, high stiffness and good
fatigue and corrosion resistance.
The aircraft structural failure in service is mainly due to fatigue loadings.
Therefore, the main consideration, while designing aircraft structures, is greatly
concerned with fatigue life evaluation of structural components. Failure mechanism
in CFRP composites is a very complex phenomenon. Various failure modes such as
matrix yielding, matrix cracking, fibre/matrix interface debonding, fibre pull-out,
fibre fracture and inter-ply delamination could occur either separately or simulta-
neously. Interface delamination is of particular interest in view of the relatively
weak ply-to-ply interface strength, and the large interface shear stresses developed
during loading. Interface delamination failure leads to significant loss of the
load-carrying capacity and occurs even in the absence of any visible damage of the
CFRP composite.
The composites in aircraft structures were, earlier (1970), used marginally in
tertiary and secondary aircraft structural elements. Now, these are most preferred
materials for large primary structures in modern frontline advanced aircraft in both
combat and transport category.
Fertig and Kenik [1] presented a physics-based methodology for predicting
fatigue life of composite structures using minimal coupon level data for charac-
terization. Methodology is based on multicontinuum theory for extracting con-
stituent stresses from a composite stress and applies the kinetic theory of fracture for
predicting fatigue life of the matrix constituent. Kinetic theory of fracture in
combination with a damage variable accurately predicts the fatigue life of a com-
posite. Jen and Lee [2] investigated the static and fatigue behaviour of AS4/PEEK
APC-2 laminates using extended Tsi-Hill fatigue failure criterion. Naderi and
Maligno [3] developed a three-dimensional finite-element model to predict the
progressive fatigue damage and the life of a carbon/epoxy flat laminates (9AS4/
3501-6) based on the longitudinal, transverse and in-plane shear fatigue charac-
teristic. The model uses fatigue failure analysis, stress analysis, random distribution
and material property degradation. Fertig and Kenik [4] presented a multiscale
physics based fatigue life prediction methodology for composite structures, which
requires minimal fatigue characterization and computationally efficient, also taking
into account arbitrary loads and load histories. Liu and Mahadevan [5] proposed a
new damage accumulation model for multidirectional composite laminates to pre-
dict the fatigue life. The model is built on the ply level, and a new multiaxial
damage index has been used to consider the damage caused by different stress
components. With the help of proposed model and making use of the experimental
results of the unidirectional laminates, the fatigue life has been predicted. Wu and
Yao [6] presented a phenomenological fatigue damage model by making the use of
the stiffness degradation rule of composites having two material parameters. The
fatigue damage model is capable of describing the nonlinear damage development
in the whole fatigue life period of the materials. Attia et al. [7] proposed a
methodology and studied growth of damage in and fatigue life of I-beams made of
Fatigue Life Prediction of Composite Airframe Panel 79
CFRP composite using finite-element analysis in conjunction with experiments (to

obtain fracture mechanics data). Eason and Ochoa [8] modelled composite struc-
tures using the plate/shell elements, which capture a three-dimensional state of
stress for damage progression incorporating a shear deformable composite element
with built-in progressive damage capability into a commercial finite-element pro-
gramme, ABAQUS. Varvani-Farahani et al. [9] developed an energy-based crite-
rion to evaluate the fatigue damage of unidirectional CFRP/GFRP composites. The
suggested fatigue damage model effectively linked fatigue lives of unidirectional
composites at various off-axis angles h and stress ratios Lian and Yao [10] per-
formed three-dimensional finite-element simulations for plain E-glass/epoxy to
simulate the fatigue life. Papanikos et al. [11] developed a progressive fatigue
damage model for predicting damage accumulation and life of CFRP laminates
with arbitrary geometry and stacking sequences subjected to constant amplitude
loading. The model consists of components of stress analysis, fatigue failure
analysis and fatigue material property degradation. Ratwani and Kan [12] studied
the effect of stacking sequence on failure modes and damage growth in composite
laminates of AS4/3501-6 material. Experiments were conducted for coupons with
four different stacking sequences and were investigated under X-ray radiography.
The existing fatigue models can be classified into three different categories:
fatigue life model (S-N curves), residual stiffness or residual strength model and
progressive damage model [6, 7]. The first two models used in fatigue analysis of
composites are at the macroscopic level, while third one, the progressive failure
model takes into account local damage mechanisms, such as fibre breakage, local
ply buckling and delamination. Micromechanics model based on Mori-Tanaka
method was attempted by Reifsnider and Gao to predict fatigue failure of composite
materials [13].
The literature studies show that considerable amount of work has been carried
out on static strength and fatigue life evaluation of metallic airframe at both
component and full-scale levels using either experimental or computational
approach. As for fatigue evaluation studies of composite airframes, there exit few
such studies using experimental/NDT techniques. However, there is no significant
or visible amount of work carried out on fatigue life estimation of composite
airframe using the computational technique. Therefore, it is necessary to develop a
fatigue damage model for composites to predict fatigue life.
2 Problem Statement
Fatigue failure analyses are carried out on both pristine composite laminates and
laminates with circular cut-outs using the physics-based fatigue model as detailed in
Eq. (13). The MATLAB code of this model considering different stress ratios
(R = 0.1 and R = −0.1) is developed. The laminates are made of CFRP (T300/
5208) material.
Fig. 1 Schematic draw and dimension of the laminate (all dimensions are in mm)
Table 1 Mechanical and thermal properties for CFRP T300/5208 lamina

E1 E2 G12 G23 t12 t23 a1(10−6/°C) a2 (10−6/°C)
(GPa) (GPa) (GPa) (GPa)
Composite 136 8.729 6 3.1175 0.36 0.4 −0.1134 25.236
(T300/5208) lamina
T300 Carbon fibre 207 11.382 24.226 40.706 0.33 0.21 −0.51 10.08
5208 Epoxy 4.556 4.556 1.631 1.631 0.4 0.4 41.4 41.4
The dimensions and stacking sequences are as follows, dimensions: length

50 mm, width 10 mm, ply thickness 0.15 mm; stacking sequences: {0}5s, {90}5s,
{0/90/0/90/0}s and {+45/−45/0/90/0}s with 10 plies for each sequence.
The schematic draw of the laminate is shown in Fig. 1. The mechanical and
thermal properties of T300/5208 CFRP composite material, T300 carbon fibre and
5208 epoxy are given in Table 1 [1, 14]. The lamina properties shown in the table
indicate that lamina is transversely isotropic because elastic properties in second
and third direction are same, i.e. E2 = E3, G12 = G13, t12 = t13. Elastic properties
are different in all three directions indicate laminate is orthotropic.
3 Problem Formulation and Procedures
The initiation of damage in FRP composites occurs mainly due to the kinetic
process of microcrack accumulation and coalescence. After a critical density of
microcracks is attained, a macroscopic crack forms and leads into ultimate failure.
This kind of fatigue failure can be modelled by making use of the kinetic theory of
fracture (KTF). In FRP composite materials, the stress values in matrix are not same
as those in the composite. For applying KTF to the polymer composites, deter-
mination of matrix stresses from composite stresses is essential. For extracting
matrix stress from composite stress, a methodology called multicontinuum theory
(MCT) has been used [1].
• Multicontinuum theory
The most numerically efficient theory to be implemented for a failure analysis of
composites is the MCT approach. In this approach, the macroscopic composite stress
state is efficiently decomposed to the constituent level. This multiscale decompo-
sition allows for easy identification between fibre and matrix failure states [15].
According to continuum mechanics, any physical quantity of interest can be
evaluated at a material point by averaging the quantity over a representative volume
element that surrounds the point of interest. The size of this representative volume
element (RVE) must be very small compared to the overall physical dimensions of
the material body, yet large enough to provide an accurate statistical representation
of the quantity within the body’s material microstructure. A continuum point for a
continuous fibre composite at lamina scale is shown in Fig. 3a. The macroscopic
value used for characterizing the stress tensor at this point in a single continuum is
derived by taking a volume average of all stresses as
Z
1
r¼ rðxÞdV ð1Þ
V
D
where the volume averaging occurs over both matrix and fibre constituents.
Let us consider a continuous fibre composite material where the fibres (f) and
matrix (m) are permitted for retaining their identity in the continuum, as shown in
Fig. 2. Using Eq. (1) for each constituent, we get
Z
1
rf ¼ rðxÞdV ð2Þ
Vf
Df
Fig. 2 Multicontinuum
decomposition
Z
1
rm ¼ rðxÞdV ð3Þ
Vm
Dm
Assuming the continuum forms of rm and rf therefore taking them as the

original variables. Combining Eqs. (1)–(3) gives
r ¼ /f rf þ /m rm ð4Þ
where /m and /f are the volume fractions of matrix and fibre, respectively.
Similarly for strains,
¼ /f f þ /m m ð5Þ
The capability of decomposing lamina level continuum fields to the constituent

level involves a relationship between constituent elastic constants and composite.
The relation between material properties is obtained by making use of a
finite-element micromechanics model of a representative composite microstructure.
The multicontinuum analysis described takes into account the internal stress states
of each constituent. The main advantage of the multicontinuum approach is that it
allows the use of physics-based theories that have been recognized for polymers
directly to the matrix material of the composite [16].
• Linking composite stress with matrix
For applying kinetic theory to composite structures, the stress in the matrix and
fibre constituents need to be determined. In this study, the main focus is on matrix
dominated fatigue failure, so rm exact value of average stress in the matrix con-
stituent can be written as
rm ¼ Qm rc km DT ð6Þ
where,
1
Qm ¼ Cm Cc /m I þ /f A
1 1
km ¼ Cm /f Cc Cf /m I þ /f A a þ gm /m I þ /f A gc
/ 1 ð7Þ
A ¼ m Cc Cf ðCc Cm Þ
/f
a ¼ Cc gc /f Cf gf /m Cm gm
In Eqs. (6) and (7), rc is the six-component composite stress vector; DT is the
temperature change from the stress free state; Ci (i = m, f, c) is the reduced stiffness
matrices for matrix, fibre and composite, respectively; /m and /f are the matrix and
fibre volume fractions, respectively; and gi (i = m, f, c) is the thermal expansion
coefficients of the matrix, fibre and composite, respectively, written as

six-component strain vectors where the shear components are zero.
• Kinetic theory of fracture (KTF)
KTF gives a description of bond breaking via thermally activated processes. It
predicts fatigue failure in the matrix constituent, which translates to composite
failure. After obtaining composite stress state or stress history, MCT can be used to
determine the average matrix stress. All atoms and molecules that are having
temperatures greater than absolute zero oscillate with a frequency proportional to m0
given by Planck’s Law [17].
v0 ¼ kT=h ð8Þ
where,
v0 Oscillation frequency of the atom
h Planck’s constant (h = 6.626068 10−34 J s/K atom)
kT Thermal energy, described by the product of the Boltzmann constant
k (k = 1.3806504 10−23 J/K atom) and the absolute temperature T
The thermal energy linked with each oscillation is not a single number but rather
a distribution; there may be a possibility that any given oscillation will have suf-
ficient energy for overcoming the energy barrier U and move from one equilibrium
state to another. The chances of this occurrence for any given oscillation is given by
the known exponential form exp (−U/kT). The bond-breaking rate Kb can be written
as

U cr
Kb ¼ m0 exp ð9Þ
kT
The stress is acting to ease the bond-breaking process; therefore, it must reduce
the energy barrier to bond breaking. The amount of this reduction is determined by
the volume of material c the activation volume, over which the process occurs so
that the reduction in the activation energy is only the product of activation volume
and the stress. Adjusting Eq. (9) to reflect the effect of an applied stress results in

kT U cr
Kb ¼ exp ð10Þ
h kT
Equation (10) is the baseline equation for the KTF that is used to predict
composite fatigue. The above equation with multicontinuum theory and kinetic
theory of fracture is put together to get an algorithm for performing fatigue failure
analysis of composites.
Kinetic theory of fracture in conjunction with MCT gives a bond-breaking rate in
the matrix, provided that U and c are appropriately calibrated. The only remaining
effort is to link the bond-breaking rate with a macroscopic measurement of damage.
A differential equation to evaluate the damage variable n with time t, where the
evolution of the damage variable is directly related to the bond rupture rate as:
dn
¼ ðn0 nÞKb ð11Þ
dt
where
e
n0 ¼
e1
The damage variable, representing the fraction of microcrack density essential for
fracture, is zero initially (n(t = 0) = 0) and unity at failure (n(t = tf) = 1). Merging
Eqs. (10) and (11) gives the initial equation for determining the fatigue life of a
polymer.
dn KT U crðtÞ
¼ ðn0 nÞKb exp ; nð 0Þ ¼ 0 ð12Þ
dt h kT
Assuming a sawtooth-shaped load history having frequency f, maximum stress

rmax and minimum stress rmin, and as a first-order approximation, assuming that the
stiffness properties are not degrading with increasing n, as saw in some experi-
mental work on composites [16]. Using these assumptions and solving Eq. (12)
gives the number of cycles to fatigue failure Nf.

fhcðrmax rmin Þ U h crmax crmin i1
Nf ¼ 2
exp exp exp ð13Þ
ðkT Þ kT kT kT
The above formula (Eq. 13) is used to calculate fatigue life of polymers and has
been implemented in this study to calculate the fatigue life of composite laminates.
3.1 Fatigue Analysis Algorithm
The flow chart for stress calculations and fatigue analyses of both pristine and
laminate with circular cut-outs is described in Fig. 3. For pristine laminates, stresses
at each ply were calculated by inputting load, lamina properties and stacking
sequences to MATLAB programming; whereas, for laminate with circular cut-outs,
peak stresses at stress critical locations (around cut-out region) were obtained from
ABAQUS FEA code output results. Then, the maximum stress values at respective
ply for both types of laminates (pristine and laminate with cut-outs) were used in
fatigue analysis model to calculate fatigue life.
Fig. 3 Process flow chart for fatigue analysis
Fatigue analyses have been carried out for pristine composite laminates and lami-
nates with circular cut-outs made of CFRP T300/5208 composite materials with
different stacking sequences for different stress ratios (R = 0.1 and −0.1). The
fatigue model using a physics-based approach is developed, and the MATLAB
code of this model is developed. Stress-life (S-N) data are generated by increasing
or decreasing the load values and determining the corresponding number of cycles
to failure. Thus, S-N curve is plotted by considering the maximum stress values
occurring in a lamina to the corresponding number of cycles to failure for pristine
composite laminates and laminates with circular cut-outs (stress concentrations at
cut-out sections).
4.1 Model Validation
In order to validate the proposed theoretical model, fatigue life results of (0, ±15, 0)
T300/5208 laminate for stress ratio, R = 0.1 have been compared with experimental
ones from literature [1] as shown in Fig. 4. The comparison shows that both the-
oretical and experimental results are in good agreement.
Fig. 4 Comparison of fatigue

life prediction results with
experimental
4.2 Stress-Life (S-N) Plots for Pristine T300/5208

Composite Laminates with Different Stacking Sequence
Figures 5, 6, 7 and 8 show the S-N curve for pristine T300/5208 composite lami-
nates for stacking sequences {0}5s, {90}5s, {0/90/0/90/0}s and {+45/−45/0/90/0}s,
respectively, with different stress ratios (R = 0.1 and −0.1).
Fig. 5 S-N curve for {0}5s

pristine composite laminate

pristine composite laminate
Fig. 7 S-N curve for {0/90/0/

90/0}s pristine composite
laminate
Fig. 8 S-N curve for {+45/

−45/0/90/0}s pristine
composite laminate
It is observed, in {0}5s pristine composite laminates (Fig. 5) that fatigue life is

more for the laminates having stress ratio −0.1 compared to laminate at stress ratio
0.1. This trend is quite acceptable by observing the fact that, the trend is similar to
that established for metals. However, the difference in fatigue life is not that sig-
nificant. For {90}5s, {0/90/0/90/0}s and {+45/−45/0/90/0}s pristine composite
laminates (Figs. 6, 7 and 8), it is observed that both the fatigue curves almost
coincide at both stress ratios (R = 0.1 and −0.1). The possible reasons for such
trend might be due to the fact that {0}5s pristine laminates fail due to failure of 0°
ply, whereas other pristine composite laminates {90}5s, {0/90/0/90/0}s and {+45/
−45/0/90/0}s fail due to the failure of 90° lamina.
4.3 Stress-Life (S-N) Plots for T300/5208 Composite

Laminate with Circular Cut-Outs with Different
Stacking Sequences
Figures 9, 10, 11 and 12 show the S-N curves for T300/5208 composite laminate
with circular cut-outs for stacking sequences {0}5s, {90}5s, {0/90/0/90/0}s and
{+45/−45/0/90/0}s, respectively, with different stress ratios (R = 0.1 and −0.1).
It may be seen from Figs. 9, 10, 11 and 12 that similar trend of fatigue life for
different stress ratios for T300/5208 composite laminate with circular cut-outs as

composite laminate with
circular cutout

composite laminate with
circular cutout
Fig. 11 S-N curve for {0/90/

0/90/0}s composite laminate
with circular cutout
Fig. 12 S-N curve for {+45/

−45/0/90/0}s composite
laminate with circular cutout
discussed earlier for pristine composite laminates. It is observed that the fatigue life
of laminates has decreased due to the presence of stress concentration around
cut-out region, which is more prone to the damage initiation.
Fig. 13 Fatigue life

comparison for pristine
composite laminates and
laminates with circular
cut-outs
4.4 Comparison of Fatigue Lives
Figure 13 shows the comparison of fatigue life of pristine composite laminates and
laminates with circular cut-outs under same load (50 N/mm) and at the same stress
ratio (R = 0.1) for all the stacking sequences considered.
It is observed that fatigue life of pristine composite laminates is much higher
compared to the fatigue life of composite laminates with circular cut-outs. This is
due to the presence of stress concentrations around cut-out regions, and damage
initiates first for laminates with cut-outs in comparison with pristine laminate.
Finite-element stress analysis results are post-processed using multicontinuum

theory and kinetic theory of fracture to obtain the fatigue lives of composite pristine
laminates and laminates with cut-outs. Multicontinuum theory was used for
extracting matrix stresses from composite stresses. Effective matrix stress values are
processed using KTF to predict fatigue life of composite material for two different
stress ratios 0.1 and −0.1.
Fatigue lives are estimated for CFRP T300/5208 pristine composite laminates
and laminate with circular cut-outs subjected to specified loading with different
stress ratios (R = 0.1 and −0.1) and frequency (f = 10 Hz). The fatigue failure
cycles are obtained at different stress levels, and S-N curves are derived for all the
laminates considered. It is observed that pristine laminates have higher cycles to
failure compared to laminates with circular cut-out due to the cut-out region
experiencing higher stress concentrations and the damage initiation occurs first at
those locations.
Fatigue failure analysis results show that {0}5s pristine laminates fail due to the
failure of 0° ply, whereas {90}5s, {0/90/0/90/0}s and {+45/−45/0/90/0}s pristine
composite laminates fail due to failure of 90° lamina. Fatigue life is more for the
laminate having stress ratio −0.1 as compared to laminate at stress ratio 0.1. It is
observed that pristine laminates have higher cycles to failure compared to laminates
with circular cut-outs. This is due to the cut-out region experiencing higher stress
concentrations, and the damage initiation occurs first at those locations. The fatigue
life of laminates decreases due to the presence of stress concentration around
cut-out region. This method can be used as an alternative approach to obtain
stress-life (S-N) curves for composite laminates for predicting fatigue life of
aerospace composites, thus minimizing expensive testing activities.
Acknowledgements The authors kindly acknowledge the financial support of council of

Scientific and Industrial Research (CSIR) under ASTA grant ESC-02-12-03. They would like to
thank the Director, CSIR-National Aerospace Laboratories, Bangalore and Dr. Satish Chandra,
Head, STTD, CSIR-NAL, Bangalore for encouragement and permitting them to present and
publish the work.
References
1. R.S. Fertig III, D.J. Kenik, in Predicting Composite Fatigue Life Using Constituent Level
Physics, Presented at the 52nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural
Dynamic Material, Denver (Colorado, 2011)
2. M.H.R. Jen, C.H. Lee, Strength and life in thermoplastic composite laminates under static and
fatigue loads. Part I: experimental. Int. J. Fatigue 20, 605–615 (1998)
3. M. Naderi, A.R. Maligno, Fatigue life prediction of carbon/epoxy laminates by stochastic
numerical simulation. Compos. Struct. 94, 1052–1059 (2012)
4. D.R.S. Fertig III, D.J. Kenik, Physics Based Fatigue Life Prediction of Composite Structures
5. Y. Liu, S. Mahadevan, Probabilistic fatigue life prediction of multidirectional composite
laminates. Compos. Struct. 69, 11–19 (2005)
6. F. Wu, W. Yao, A fatigue damage model of composite materials. Int. J. Fatigue 32, 134–138
(2010)
7. O. Attia, A.J. Kinloch, F.L. Matthews, Modelling the fatigue life of polymer–matrix
fibre-composite components. Compos. Sci. Technol. 61, 2273–2283 (2001)
8. T.G. Eason, O.O. Ochoa, Modeling progressive damage in composites: a shear deformable
element for ABAQUS® Compos. Struct. 34, 119–128 (1996)
9. A. Varvani-Farahani, H. Haftchenari, M. Panbechi, An energy-based fatigue damage
parameter for off-axis unidirectional FRP composites. Compos. Struct. 79, 381–389 (2007)
10. W. Lian, W. Yao, Fatigue life prediction of composite laminates by FEA simulation method.
Int. J. Fatigue 32, 123–133 (2010)
11. P. Papanikos, K.I. Tserpes, S.P. Pantelakis, Modelling of fatigue damage progression and life
of CFRP laminates. Fatigue Fract. Eng. Mater. Struct. 26, 37–47 (2003)
12. M.M. Ratwani, H.P. Kan, in Effect of Stacking Sequence on Damage Propagation and
Failure Modes in Composite Laminates, ed. by K. Reifsnider, Damage in Composite
Materials (ASTM STP 775, ASTM, 1982), pp. 211–228
13. K.L. Reifsnider, Z. Gao, A micromechanics model for composites under fatigue loading. Int.
J. Fatigue 13, 149–156 (1991)
14. D.E. Bowles, in Micromechanics Analysis of Space Simulated Thermal Deformations and
Stresses in Continuous Fiber Reinforced Composites. ed. by M.Y. University C. Langley
Research. NASA Technical Memorandum, National Aeronautics and Space Administration,
Langley Research Center
15. A.C. Hansen, E.E. Nelson, D.J. Kenik, A comparison of experimental data with
multicontinuum failure simulations of composite laminates subjected to tri-axial stresses.
J. Compos. Mater. (2013)
16. E.E. Nelson, J.A. Gies, R.S. Fertig III, The Virtues of Multicontinuum Mechanics for
Composites Analysis, in Presented at the Structural Dynamics and Materials Conference,
Palm Springs, California (2009)
17. H. El Kadi, F. Ellyin, Effect of stress ratio on the fatigue of unidirectional glass fibre/epoxy
composite laminae. Composites, 25, 917–924 (1994)
Integrity Evaluation of Feature Level Test
Specimen of an Aircraft Primary
Composite Structure
S. Venkatesh, S. C. Lakshminarayana and Byji Varughese
Abstract It is well known to the aircraft designers that the design of structure using
only metals or only composite materials can be realized with higher confidence
level based on the database or tools available present in their aircraft design bureau.
However, data availability for design of primary structural element/joint using
metal, composites, and mechanical fasteners is minimal, and many of them are
proprietary; moreover, they are depending on type of design. There is a need to
design and test a feature level test specimen with appropriate boundary and loading
conditions, which simulates behavior of full-scale component. Primary structural
joint for actuator mounting on the composite fin has been considered for feature
level testing. The purpose of the tests was to demonstrate the validity of the
structural joint by applying static limit and ultimate loads. The feature level test
specimen consists of composite and metal joints was successfully loaded to the
ultimate design load without evidence of failure. The global behavior predicted by
the finite element (FE) model of the test arrangement was in close agreement with
the experimental results as evidenced by the strain and displacement results. The
success of the testing program demonstrated the importance of feature level test in
the design stage of primary composite structural joints of a civil transport aircraft.
Keywords Feature level test Composite FE analysis Actuator assembly
S. Venkatesh (&) B. Varughese

Advanced Composites Division, CSIR-NAL, Bangalore 560017, India
e-mail: venkateshs@nal.res.in
B. Varughese
e-mail: byji@nal.res.in
S. C. Lakshminarayana
Centre for Civil Aircraft Design and Development Division,
CSIR-NAL, Bangalore 560017, India
e-mail: sclnara@nal.res.in

https://doi.org/10.1007/978-981-10-7197-3_8
94 S. Venkatesh et al.
1 Introduction
Over the last two decades, CSIR-National Aerospace Laboratories (CSIR-NAL) has
developed primary components for both civil and military aircraft programmes
using composite materials. The application of composite materials to primary air-
craft structures requires proven design concepts and certification procedures to
demonstrate their structural integrity during service life. Development phase of
structural design for civil and military aircraft involves an enormous amount of
design and qualification tests before building a full-scale component for testing. In
the preliminary design phase of an aircraft, structural designer would like to know
the adequacy of the design parameters for all primary components/features/joints in
aircraft. The design parameters include strength, stiffness, and number/type of
fasteners used [1, 2]. Composites have a more linear load–strain response, a greater
sensitivity to stress concentrations and environments, higher data scatter, and a
multiplicity of potential failure modes. Application of the metallic certification
database to composite structures is limited by the inherent differences between
composites and metals. In order to have confidence in the compliance of composite
structural design, it is imperative to generate the database of primary structural
joints of composite and metal. This is achieved by interpretation of feature level
tests at the development phases.
The design of feature level test specimen and testing requires a lot of effort and
brings challenges in terms of technical study/understanding, design and detailing of
joints, application of proper software tools, and simulation using testing equipments
to capture the strain, displacement, and local damage/structural collapse. In this
study, one of the primary composite structural joint on fin is considered for feature
level testing [3]. This paper presents the importance of feature level test, test design,
and test-FE correlation of a primary composite structural joint of a fin component of
civil transport aircraft.
2 Geometry and Structural Features of Test Component
In many of the Light Transport Aircrafts, a T-tail configuration of the empennage in

which horizontal tail (HT) is mounted on fin. For variable incidence horizontal
stabilizer, the actuation of the horizontal stabilizer is done by either an elec-
tromechanical or a hydraulic actuator with its fixed end mounted in the fin structure
and the actuation/moving end connected to the horizontal stabilizer. These joints are
classified under primary joints; any failure of such a joint would cause loss of
aircraft. For this design, horizontal stabilizer trim actuator (HSTA), which is an
electromechanical linear actuator, is used. HSTA consists of two load paths, the
primary load path works under normal conditions, and the secondary load path
works after the failure of the primary load path, and therefore, HSTA has built-in
fail-safe design feature. Therefore, installation of this HSTA on an aircraft is very
Integrity Evaluation of Feature Level Test Specimen … 95
(a) (b)
Fig. 1 a Actuator installation on HT and fin; b detail view of the attachment
important and considered to be the primary joint and hence requires proven design.
Fixed or actuator end of the HSTA is connected to the composite fin structure
between the spar #2 and spar #3. Hinge or movable end of the actuator connected to
the horizontal stabilizer front spar, top, and bottom skins as shown in Fig. 1a.
Details of the actuator attachment on HT and fin are shown in Fig. 1b.
Fin actuator fitting assembly is a critical joint, and an experimental testing of the
same has to be performed before installation. The complete fin cannot be considered
for experimental testing as the complete setup is huge, and the materials used are
very expensive, and this joint should be validated before fabrication of the full-scale
component; thereby, any changes found after the testing can be implemented in the
design before fabrication of full-scale component. Hence, in this study, a segment of
the fin close to the actuator fitting assembly has been considered instead of the
complete fin. A feature level test model of fin actuator fitting assembly was
developed to demonstrate the composite and metal joints for design ultimate loads.
Feature level test specimen was designed using finite element tools. Basic under-
standings with the global behavior of the fin for the critical load cases were studied.
Stress parameters were considered in deriving the boundary and loading conditions.
Finally, required components fabricated and assembled as per the airworthiness
requirements.
3 Finite Element Modeling and Stress Analysis
Finite element (FE) preprocessing and post-processing of the structure was done
using hypermesh. The analysis was carried out using MSC/NASTRAN [4]. In this
section, FE modeling of the global structure and the test specimen is described. The
global FE model of the fin structure with actuator assembly is shown in Fig. 2a. The
structural members such as spar, skin, stiffeners, and metallic brackets were mod-
eled using the CQUAD4 and CTRIA3 elements in MSC/NASTRAN. All com-
posite parts were treated as 2D orthotropic layered shell structural elements
modeled with PCOMP property in NASTRAN. Test configuration derived from the
global FE model is shown in Fig. 2b. Parts of spar #2 and spar #3 at the actuator
bracket fitting assembly, top and bottom skins, and bracket fitting assembly were
considered for the test model analysis. The FE models of the components of fin and
actuator fitting assembly were generated by idealizing the skin, spars, ribs, and
surfaces of actuator fitting assembly by shell elements as the thickness of these
components is small as compared to the other dimensions. The rivets and flanges
were idealized by bar and rod elements, respectively.
The configuration of the test specimen was derived based on the configuration of
spar, rib, and skins of the composite fin. The loading and boundary conditions of
the test specimen should result in similar strain conditions as computed by FE
analysis for the composite fin. To achieve this, a combination of actuator forces has
to be determined, which gave an optimum correspondence between strains in the fin
structure and in the test specimen. Boundary conditions place a very important role
in achieving the same. The end fittings of the test specimen attached to the
mounting brackets of the test rig were designed to take the test loads and transfer in
the intended direction. To understand the structural behavior of the global model
and feature level test article, displacement, von-Mises stress patterns and failure
indices were studied. The failure index was calculated based on the Yamada-Sun
failure criteria. Finally, the Feature level test article was designed to capture the
global behavior of the fin structure.
Fig. 2 a Global FE model, b Feature level test specimen

3.1 Loads and Boundary Conditions
The FE modeling was carried out to cater for the application of loads as per the test
loads. The design limit load (DLL) on this actuator mounting point along the
actuator axis in the neutral axis is 2000 kg, and design ultimate load (DUL) is
3000 kg [5]. The load reacted at this point due to aerodynamic load on the hori-
zontal stabilizer is lower than the DLL of 2000 kg which is used to design the
actuator. Therefore, actuator supporting structure should be able to withstand the
DLL of 2000 kg and DUL of 3000 kg. The displacement boundary conditions for
the feature level test model with respect to the global axes were constrained as
UX = 0, UY = 0 and UZ = 0 in the FE model. The applied load and boundary
conditions on the feature level test model are shown in Fig. 2b.
3.2 Material Properties
The materials used for fin actuator fitting assembly and bracket fitting assembly
were aluminum alloy steel, bushes, and pins were made out of corrosion resistant
steel. The spars, ribs, and skins of fin were made of UD prepreg (carbon-epoxy)
composites with ‘quasi-isotropic’ layup sequence.
4 Test Configuration
The feature level test was conducted in the specially built test rig. It is a steel frame
supported in all direction to provide the rigidity for the test loads. The test article
was mounted on the test rig at the mounting brackets. The mounting brackets were
designed to take the test loads and capture the global behavior of the fin structure.
Figure 3 shows the schematic view of the test setup and loading arrangement of the
test article.
A hydraulic jack was connected to the actuator mounting bracket through which
load was applied. Design limit load of ±2000 kg (compression and tensile) was
applied on the actuator mounting point using hydraulic actuator. Test carried out for
tension and compression load of 2000 kg was applied in steps of 10% DLL.
Subsequently, the DUL test was carried out for tension load of 3000 kg
(2000 1.5). This case is more critical for composite and metal joints. The load
was applied in steps of 10% DLL up to 160% (i.e., 10% above DUL) and unloaded
to zero in steps of 10% DLL. Figure 4 shows the photography of the test setup.
Strain and dial gauge locations were selected on composite skin and metal fitting
assembly by understanding the finite element stress analysis. Strains and deflection
were recorded at every load step during loading and unloading. Rosettes were used
to measure strains, and dial gauges were used to measure the deflections.
Fig. 3 Test setup with

loading arrangement
Test component
mounted on test rig
Load cell
Actuator
Compression
Tension
Fig. 4 Photography of the

test setup
5 Test Results Comparison and Discussion
Displacement and strains were recorded during the test at various locations are
compared with the finite elemental analyses results. Displacements are small in
magnitude, and it is more appropriate to compare the relative displacement or twist
of the test article. This eliminates the influence of test rig deflections. Maximum
deflection measures under compression and tensile load cases at DLL was
2.83 mm, and from FE, it was 1.59 mm. Displacement measured from ultimate
load test (160% of DLL) for the tensile case was 2.63 mm and FE estimate was
2.54 mm as shown in Fig. 5. It has observed that the comparison between test and
FEM deflections is quite good. Both deflections are showing similar profile. This
shows that the FE model is capturing global elastic behavior very well. It has
observed that deflection from DLL test for tension shows little deviations as
compared to DUL test. Furthermore, the variation observed in DLL test can be
attributed to fixture and component fittings stabilization during DUL test.
The strains recorded during the test at various locations were compared with FE
strains determined from the ply strain output from MSC Nastran. The FE strains
were extracted, and it is compared with test strain gauge data. The ply shear strains
were derived from the strains measured in all three elements. Comparisons between
experimental and FE strains are presented for the gauges at the maximum strain at
two locations for the compression case (DLL) as shown in Fig. 6 and for the tensile
case (DUL) as shown in Fig. 7. The strains measured from the test were within the
allowable limit of 3300 µe. In general, there is a good correlation between the FE
and test results. The trend observed in test and FE is similar. As the strains are low
at many locations, there is no real concern for the integrity of the structure. The
Fig. 5 Comparison of test and FE displacement for tensile case (DUL) for dial gauge 1 and dial
gauge 2
Fig. 6 Comparison of test and FE strains for the compression case (DLL)
Fig. 7 Comparison of test and FE strains for the tensile case (DUL)
actuator assembly was subjected to ultrasonic inspection after the static test and
results showed no debonds or delaminations on the tested article.
6 Conclusion
The structural static testing of critical joints is an essential activity to prove the
integrity of composite aerospace parts. In this paper, requirements, importance in
design of feature level test component, and testing were discussed. Static testing of
fin actuator fitting assembly was carried out to prove the structural integrity of the
joint and its attachment to the surrounding composite structure. The component was
tested up to design limit load both in tension and compression loads and up to
design ultimate load (up to 160% DLL) for tensile load. The deflection and strains
are monitored during the DLL tests, and DUL test was compared with those from
FE analysis. A good correlation between test and analysis results was observed at
most locations. This proved the integrity of the structure for the load which it is
designed. It may be noted that the selected geometrical size of the feature level test
component and the boundary conditions chosen proved to be proper throughout the
test. Hence, it may be concluded that the design of test specimen and its validation
is acceptable.
Acknowledgements Authors would like to express their gratitude to Director, CSIR-National

Aerospace Laboratories (NAL) and Heads of Advanced Composites Division (ACD) and Centre
for Civil Aircraft Design and Development Division (C-CADD), NAL for their encouragement.
Authors also thank all staff members of ACD, NAL for their unstinted support in various stages of
this work.
References
1. H.G.S.J. Thuis, C. Biemans, Design, fabrication and testing of a composite bracket for
aerospace applications. Compos. Struct. 38(1–4), 91–98 (1997)
2. K.I. Tserpes, R. Ruzek, R. Mezihorak, G.N. Labeas, Sp.G. Pantelakis, The structural integrity
of a novel composite adhesively bonded flap-track beam. Compos. Struct. 93, 2049–2059
(2011)
3. B. Ramanaiah, H.V. Ramachandra, S. Sanjeev Kumar, M.G. Kotresh, B. Varughese, R.
Sundaram, Challenges in static testing of co-cured co-bonded composite aircraft structure, in
National Conference on Scientific Achievements of SC & ST Scientists & Technologists,
National Aerospace Laboratories, Bangalore-17, 14–16 Apr 2009
4. R.S. Lahey, M.P. Miller, M. Reymond, MSC/NASTRAN Version 70 Reference Manual (The
MacNeal-Schwendler Corporation, 2003)
5. M.C.-Y. Niu, Airframe Stress Analysis & Sizing, 2nd edn. (Hong Kong Conmilit Press Ltd, Jan
1999)
A Study on the Effect of Crack Stopper
for Enhanced Damage Tolerance Behavior
of a Fuselage Stiffened Panel
G. Akshitha, A. Shailesh Rao, N. Srinivasan and M. Mohan Kumar
Abstract During the design and development phase of a large transport aircraft, a
considerable amount of parametric analysis and testing is carried out to bring out an
optimum structure. Two-bay longitudinal crack arrest feature is the main aspect of
design for damage tolerance of the pressurized fuselage cabin. Under fuselage
pressurization load cycles, fatigue cracks develop at location of maximum tensile
stress. There are locations on the airframe which are favorable for the initiation of
longitudinal cracks. This investigation identifies one such location in a fuselage
panel from where a longitudinal crack can initiate and studies the fast fracture and
crack arrest features under the action of uniaxial hoop stress. The main crack
arresting features are the bulkheads and crack stopper straps. A finite element
modeling and analysis approach will be used for simulating stiffened panel with and
without the presence of tear strap and their role in the two-bay crack arrest capa-
bility of the aircraft fuselage is assessed. Stress intensity factor for progressive crack
lengths and the fatigue crack growth rate of the cracked stiffened panel was esti-
mated. A tear strap crack stopper was introduced for evaluating the damage tol-
erance capability of the stiffened panel again fatigue crack growth rate was
estimated for a realistic representation of two-bay cracking scenario, it will be
examined under what condition a two-bay crack can be arrested. By this new
design, the stiffened panel was found to be more damage tolerant compared to the
earlier one.

Keywords Damage tolerance Crack Fracture Bulkhead Tear strap

Finite element analysis Residual strength
G. Akshitha A. Shailesh Rao

Department of Mechanical Engineering, K.S. School of Engineering and Management,
N. Srinivasan M. Mohan Kumar (&)
Fatigue and Structural Integrity Group, Structural Technological Division, CSIR-National
Aerospace Laboratories, Bangalore 560017, India
e-mail: mmk@nal.res.in

https://doi.org/10.1007/978-981-10-7197-3_9
104 G. Akshitha et al.
1 Introduction
Aircraft structure is the most obvious example where structural efficiency results in
light weight and high operating stresses. An efficient structure must have three
primary attributes, namely the ability to perform its intended function, adequate
service life, and the capability of being produced at a reasonable cost. The major
part of the aircraft structure consists of built-up panels of sheets and stringers, e.g.,
wing and fuselage skin panels, spar webs, and stiffeners. Despite all precautions,
cracks have arisen in many of these structural elements. These cracks reduce the
stiffness and the total load-carrying capacity of the structure. The fuselage is the
main structure in the aircraft with the basic functions is to transmit and resist the
applied loads; to provide an aerodynamic shape and to protect passengers, payload
systems, etc., from the environmental conditions encountered in flight.
These requirements, in most aircraft, result in thin shell structures where the
outer surface or skin of the shell is usually supported by longitudinal stiffening
members and transverse frames to enable it to resist bending, compressive, and
torsion loads without buckling. Such structures are known as semi-monocoque,
while thin shells which rely entirely on their skins for their capacity to resist loads
are referred to as monocoque. In most modern aircrafts, the skin plays an important
role in carrying loads. Sheet metals can usually only support tension. But if the
sheet is folded, it suddenly does have the ability to carry compressive loads,
stiffeners are used for that. A section of skin, combined with stiffeners, called
stringers, is termed a thin-walled structure. An aircraft fuselage structure must be
capable of withstanding many types of loads and stresses. The principal source of
the stresses in this structure is the internal pressure in high altitude caused by
difference of cabin pressurization and reduction of the outside pressure with
increase in altitude, but the structure is subjected to other loads, as bending, torsion,
and thermal loads.
Fatigue loads in a pressurized fuselage are mostly due to pressure cycles that
occur with each takeoff or landing cycle during flight. The most common fatigue
crack orientation in a pressurized fuselage is a longitudinal crack along the direction
of maximum hoop stress. Damage-tolerant designs use fracture mechanics data and
analysis to predict crack growth rates and critical crack lengths [1]. Cabin pressure
results in radial growth of the skin and this radial growth is resisted by frames and
stringers giving local bending along the fastener lines. Fuselage skin panels are
curved, and these panels are under biaxial tension loading due to cabin pressure.
Cabin pressurization is the main source of loading causing longitudinal skin cracks.
Two types of damage most frequently associated with the structural integrity of the
fuselage are longitudinal cracks under high hoop stresses induced by cabin pres-
surization and circumferential cracks under stresses from vertical bending of the
fuselage. The objective of paper was to present a systematic investigation of the
damage tolerance design capability of typical aircraft fuselage structure for longi-
tudinal cracks using linear elastic fracture mechanics [2]. Damage-tolerant fuselage
is supposed to sustain cracks safely until it is repaired or its economic service life
A Study on the Effect of Crack Stopper for Enhanced Damage … 105
has expired. Strength assessment of the structures is necessary for their in-service
inspection and repair. Damage tolerance analysis should provide information about
the effect of cracks on the strength of the structure. Damage tolerance evaluation
must include a determination of the probable locations and modes of the damage
due to fatigue or accidental damage.
The crack propagation stage is studied by using stress intensity factor. There are
different methods used in the numerical fracture mechanics to calculate stress
intensity factors (SIF). The crack opening displacement (COD) method and the
force method were popular in early applications of FE to fracture analysis. The
virtual crack extension (VCE) methods proposed by Hellen [3] and Parks [4] lead to
increased accuracy of stress intensity factor results. The virtual crack extension
method requires only one complete analysis of a given structure to calculate SIF.
Both the COD and VCE methods can be used to calculate SIF for all three fracture
modes. However, additional complex numerical procedures have to be applied to
get results. The equivalent domain integral method which can be applied to both
linear and nonlinear problems renders mode separation possible [5]. The VCCT,
originally proposed in 1977 by Rybicki and Kanninen [6], is a very attractive SIF
extraction technique because of its good accuracy, a relatively easy algorithm of
application capability to calculate SIF for all three fracture modes. Leski [7], the
implementation of the virtual crack closure technique in engineering FE calcula-
tions. SIF was fundamental quantity that governs the stress field near the crack
tip. It depends on the geometrical configuration, crack size, and the loading con-
ditions of the body.
2 Problem Description
Stiffened panels are the most generic structural elements in an airframe. Currently,
large transport airplanes are being developed with large damage tolerance capability
as a design goal. An important concept in the design of the pressurized fuselage of
large transport aircraft is the provision of crack stopper straps to arrest the fast
fracturing of a crack. In this project, the role of the crack stopper strap in the
fail-safe design of the fuselage is investigated. As a first approximation, a stiffened
flat panel with a center longitudinal crack is considered. The strength of this cracked
panel is investigated as a function of crack length in the absence of crack stopper
straps. Crack stopper straps is then introduced at the locations of stiffeners per-
pendicular to the crack line, and strength of the cracked flat panel is investigated as
a function of crack length in the presence of crack stopper straps. The failure criteria
that are used in this study that the skin crack will have a fast fracture when the
maximum stress intensity factor becomes equal to the fracture toughness of the skin
material. It is assumed that there is no rivet failure, and there is no failure of the
stiffener normal to the crack line.
3 Geometrical Configuration
The stiffened panel represents a most generic in fuselage structure. The stiffened
panel dimensions are 1750 mm in the longitudinal direction and 990 mm in
transverse direction. The thickness of the stiffened panel skin is 1.5 mm. The
stiffened panel has five frames (four bays) with 400 mm spacing and seven stringers
(six bays) with 140 mm spacing. The frame has Z and L cross section with
1099 mm2 of cross-sectional area and stringer has Z cross section with 419 mm2 of
cross-sectional area. The frames and stringers are attached to the skin by row of
3 mm diameter rivet, placed at a pitch of 21 mm. Geometric modeling is carried out
using MSC PATRAN software shown in Fig. 1.
Fig. 1 a Flat stiffened panel, b details of bulkhead and stiffener

4 FE Analysis
In the present study, a segment of fuselage structure of a transport aircraft is modeled

and analyzed in MSC NASTRAN/PATRAN for pressurization load case to iden-
tify the critical location on skin for the initiation of crack, and evaluating the SIF
values for progressive crack lengths of a stiffened panel without and with tear strap.
Fatigue life was estimated for the progressive crack lengths in the stiffened panel.
4.1 Flat Stiffened Panel
It involves modeling the different components of the flat stiffened panel such as
bulkheads, stringers skin, tear strap, and rivets as shown in Fig. 2. The geometry of
all the stiffened panel components in the form of models was used for the finite
element modeling using MSC Patran. The modeled fuselage is a 2D integral FE
model, and there are totally around 320,743 elements comprising of QUAD4,
TRIA3, 1D beam elements. Various material properties for the different compo-
nents that are considered in the study are shown in Table 1.
See Table 1.
Fig. 2 a FE modeling of the stiffened panel, b parts of the stiffened panel
Table 1 Details of Material properties

Component Material Properties
Skin Al-2024-T3 Young’s modulus, E = 72 GPa
Bulkhead Poisson’s ratio, m = 0.3
tear strap Ultimate tensile strength, rult = 427 GPa
p
Stiffener rivets Fracture toughness, KIC = 31 MPa m
Material constants: C = 3.2; m = 3E-8
4.3 Loads and Boundary Conditions
A differential internal pressure of 0.062 MPa is considered for the current case. The
hoop stress corresponding to the internal pressure of 0.062 MPa was considered for
the present stiffened panel analysis. Uniformly distributed load is applied on edges
of skin and frame in the transverse direction. But, the stringers are passing in
longitudinal direction in the stiffened panel. Since it is uniaxial tensile loading, the
stringers are not subjected to loading. At other end, all the edge nodes of stiffened
panel are constrained in all six degree of freedom (three translations and three
rotations) shown in Fig. 3.
4.4 SIF Calculation
Modified virtual crack closure integral (MVCCI) method is used to determine stress
intensity factor for different crack lengths in the stiffened panel. MVCCI method is
based on the energy balance. Stresses at the crack tip and the nodal displacements
are noted for calculating the strain energy release rate, G, and hence SIF by MVCCI
method [7–9]. In this technique, SIF is obtained for fracture mode from Eqs. (1)
and (2).
4.5 Calculation of Stress Intensity Factor by MVCCI

Method
See Fig. 4.
p
Stress intensity factor; K I ¼ GE ð1Þ
UDL
UDL
6 DOF
Constrained
6 DOF
Constrained
Fig. 3 Loads and boundary conditions of the stiffened panel

Fig. 4 MVCCI method
ðF DvÞ
Energy release rate; G ¼ : ð2Þ
ð2 Da tÞ
4.6 Fatigue Life Calculation
Fatigue life calculation for an initial crack length af to ai.

From Paris’ law we have,
da
¼ CðDK m Þ ð3Þ
dn
Integration of above equation will give the following simple expression for the
number of cycles, Nf.
1m=2 1m=2
af ai
Nf ¼ pffiffiffiffim : ð4Þ
C FDr P 1 m2
The linear static stress analysis of the flat stiffened panel was carried out. Pressure
of 0.062 MPa was considered as a load case for the present problem. Effect of tear
strap on the stress distribution was studied; effect of tear strap on SIF values under
progressive crack lengths was analyzed. Residual strength was evaluated for the
considered configuration. Finally, fatigue life was estimated for the flat stiffened
panel from an initial crack length of 5 mm to a final crack length of 450 mm.
5.1 Stress Analysis of a Flat Stiffened Panel
A linear static analysis of a flat stiffened panel was conducted and an overall
structural response was analyzed to determine the maximum tensile stress location
which was on the skin at the rivet location where the rivets are used to fasten the
frames and stringer on the skin. The maximum stress distribution in the skin was 86
and 55 MPa in a flat stiffened panel without and with tear strap, respectively
(Fig. 5).
5.2 Design of Tear Strap
Tear strap is principally designed for arresting longitudinal skin cracks. Tear strap is
designed with thickness 1.2 mm and width 50 mm and inserted in between bottom
flange of frame and skin. Selection of tear strap was based on the optimization
criteria.
5.2.1 Optimization of Tear Strap Thickness
In this case, the thickness of the tear strap was varied to 1.5, 2.5, 2.7, 3, and 5 mm.
Iterations are carried for a crack length of 5–500 mm, and the corresponding stress
intensity factor was calculated for the corresponding crack lengths. The values of
SIF for the progressive crack lengths obtained from the stiffened panel analysis by
MVCCI method are plotted to obtain the characteristic curve shown in Fig. 6. From
this, it is clear that for a tear strap thickness of 2.5, 2.7, and 3 mm, there is a close
matching in SIF values between them. For a thickness of 1.5 and 5 mm, the SIF
values are correspondingly high compared with that of 2.7 mm. Hence, tear strap of
2.7 mm thickness is considered for the present work.
(a) Without tear strap (b) With tear strap
Fig. 5 Stress plots of flat stiffened panel skin

Fig. 6 Tear strap thickness optimization
5.2.2 Optimization of Tear Strap Width
In order to optimize the width of a tear strap, various iterations were carried out for
a crack lengths ranging from 50 to 500 mm, and graph of SIF versus crack length
was plotted shown in Fig. 7, From the graph, it was observed that the width had no
effect on the SIF variation for increasing crack length.
5.3 Effect of Tear Strap on SIF
The values of SIF for the progressive crack lengths are obtained from the flat
stiffened panel analysis by MVCCI method without and with tear strap and are
plotted to obtain the characteristic curve shown in Fig. 8. It can be observed from
the curve that the stress intensity factor (SIF) of the panel varies linearly till it
reaches the stiffener position. As the crack reaches the stiffener position, the load is
transferred to the stiffener through rivets, thus reducing the SIF near the stiffener,
which ensures in arresting the crack from propagating further. When the crack is
propagated beyond the stiffener position, there will be an increase in SIF because
the load shared by the skin increases gradually. The increasing trend in the curve is
observed as the crack moves away from the stiffener position.
SIF values for the progressive crack lengths were plotted for without and with
tear strap were plotted. For a same crack length stiffened panel without tear strap
was having higher SIF values compared to that of stiffened panel with tear
Fig. 7 Tear strap width optimization
SIF variation of flat stiffened panel without and with tear strap
35
30
25
SIF, MPa√m
20
Without
15 tear strap
With tear
10 strap
0
0 50 100 150 200 250 300 350 400 450 500 550
a, mm
Bulkhead Tear strap
Fig. 8 SIF variation of flat stiffened panel without and with tear strap
strap. The SIF values decreased by 20% for the initial crack lengths and once the
crack reaches the tear strap position the SIF values got decreased by 50% compared
to the panel without tear strap. Once it crosses the tear strap, the SIF value incre-
mental for the further crack lengths was also less compared to that of initial
incremental SIF values.
0.5
0.45
0.4
without tear strap
0.35 with tear strap
0.3
a, m
0.25
0.2
0.15
0.1
0.05
Nf
Fig. 9 Fatigue failure plot for a flat stiffened panel without and with tear strap
5.4 Fatigue Life Calculation for Progressive Crack Lengths
Fatigue life was estimated for crack lengths ranging from 5 to 450 mm. Number of
life cycles required to reach the corresponding crack length were calculated for a
stiffened panel without and with tear strap. A plot of number of cycles required to
attain various crack lengths was plotted as shown in Fig. 9 for a flat stiffened panel
without and with tear strap. In case of a flat stiffened panel without tear strap
curve is varying linearly, once it crosses the bulkhead there is a small increment in
the number of cycles required to attain the further crack lengths. In case of a
flat stiffened panel with tear strap, once the crack reaches the tear strap position,
there is an increment in the life of a panel which is indicated from the graph. The
life of a panel with tear strap is almost 50% higher than that of a panel without
tear strap.
6 Conclusion
In the present work, a detailed study of a flat stiffened panel was carried out in order
to make the structure to be damage tolerant. Initially, a flat stiffened panel was
subjected to tensile load corresponding to pressure of 0.062 MPa to identify the
maximum stress location region in the panel. Effect of tear strap was studied, and
SIF values are estimated for the flat stiffened panel without and with tear strap for
progressive crack lengths. In the later stage, fatigue life estimation was carried out
with both the configurations of the panel to demonstrate the effect of tear strap in
slowing down the crack growth of the panel.
• Stress analysis of the stiffened panel was carried out and maximum tensile stress
was identified at the rivet hole of skin. Center longitudinal crack was initiated
from rivet location of skin.
• Fatigue crack propagation was estimated by using stress intensity factor
approach. Stress intensity factor calculations were carried out for various
incremental crackspfrom 5 to 500 mm. The maximum value of stress intensity
factor 28.67 MPa m was observed at crack length p of 500 mm. After intro-
ducing tear strap, the value
p of SIF was 2.3 MPa m at crack length of 50 mm
and increased
p to 18 MPa m at crack length of 500 mm and finally decreases to
13 MPa m at frame location. This SIF value is much less than the fracture
toughness of material. Using tear strap, the crack propagation in the stiffened
panel was found to be reduced by 50%.
• The fatigue life without the tear strap showed 11,440 cycles at the crack length
of 475 mm, whereas the life for the same crack length with the tear strap showed
19,024 cycles. An increase in the fatigue life of 7584 cycles was observed by
providing the tear strap in the bulkhead region. From this, it is conclude that the
tear strap is used to slow down the fast fracturing longitudinal crack much
effectively.
Acknowledgements The authors thank Director, NAL, and the Head, STTD, NAL, Bangalore
for their support and encouragement during this work. The Principal and Head, Department of
Mechanical Engineering, K.S. School of Engineering and Management, is thanked for their
support and guidance during the course of this work.
References
1. G. Bibel, Fuselage metal fatigue in large commercial aircraft. Int. J. Forensic Eng. 1(1), 47–57
(2012)
2. P.M. Toor, On damage tolerance design of fuselage structure (longitudinal cracks). Eng. Fract.
Mech. 24(6), 915–927 (1986)
3. T.K. Hellen, The finite element calculations of stress intensity factors using energy
techniques, in 2nd International Conference on Structural Mechanics in Reactor Technology,
Paper G5/3 (Berlin, 1973)
4. D.M. Parks, A stiffness derivative finite element technique for determination of crack tip
stress intensity factors. Int. J. Fract. 10, 487–501 (1974)
5. K.N. Shivakumar, I.S. Raju, An equivalent domain integral method for three-dimensional
mixed-mode fracture problems. Eng. Fract. Mech. 42(6), 935–959 (1992)
6. E.F. Rybicki, M.F. Kanninen, A finite element calculation of stress intensity factors by a
modified crack closure integral. Eng. Fract. Mech. 9, 931–938 (1977)
7. A. Leski, Implementation of the virtual crack closure technique in engineering FE
calculations. Finite Elem. Anal. Des. (Polish Air Force Institute of Technology, Poland) 43
(6), 261–268 (2006)
8. N. Muthu, B.G. Falzon, S.K. Maiti, S. Khoddam, Modified crack closure integral technique
for extraction of SIFs in mesh free methods. Finite Elem. Anal. Des. 78, 25–39 (2014)
9. D. Xie, S.B. Biggersjr, Progressive crack growth analysis using inter face element based on
the virtual crack closure technique. Finite Elem. Anal. Des. 42, 977–984 (2006)
Failure Analysis of a Squirrel Cage
Bearing of a Gas Turbine Engine
Swati Biswas, Jivan Kumar, V. N. Satishkumar

and S. N. Narendra Babu
Abstract Premature failure of a squirrel cage bearing of a developmental

aero-engine was encountered during its structural test. Four ribs (out of 12) in the
cage were fractured from both ends, and the rear flange was completely separated
from the cage. One of the support bearings was also found fractured, and the other
was damaged to a lesser extent. There were secondary damages to the other
components in the test assembly. Fractographic analysis established fatigue failure
of the ribs. Two crack fronts were originated from the diagonally opposite corners
of the ribs indicating failure of the bearing cage under reverse bending fatigue
loading condition. No metallurgical anomalies were present at the crack origin. The
rear flange was separated from the bearing cage in shear mode of failure, whereas
the support bearing outer race failure was by instantaneous overloads. Design
modifications were introduced after thorough stress analysis of the component.
Subsequently, no such failure incidents were encountered, and the modified design
has been put into service.
Keywords Gas turbine engine Bearings Fatigue Beach marks

Striations Dimples
1 Introduction
Bearings are important components of an aircraft propulsion system that allows free
rotation of shafts, discs and aerofoils [1]. Key characteristics required for the
bearings [2] of jet engine main shaft are (i) suitability for high-speed rotation,
(ii) operability in high temperature and (iii) high reliability. The bearing types for jet
engine main shafts are mainly three-point contact ball bearings and cylindrical
roller bearings. The three-point contact ball bearings support load in the axial and
S. Biswas (&) J. Kumar V. N. Satishkumar S. N. Narendra Babu

Gas Turbine Research Establishment, Defence Research & Development Organization,
Bengaluru 560093, India
e-mail: swati@gtre.drdo.in

https://doi.org/10.1007/978-981-10-7197-3_10
118 S. Biswas et al.
Fig. 1 Photograph showing a failed squirrel cage bearing; b two support bearings
radial directions. Different bearing manufacturers and agencies have documented

the varied nature of bearing damage and failure of its various assemblies [3, 4].
Wear, fretting, pitting and corrosion are most commonly encountered problem for
the inner and outer races and the rolling elements [5].
The present article describes analysis carried out on a failed roller bearing on a
developmental engine. The squirrel cage bearing that supports the high pressure
(HP) rotor in the aero-gas turbine propulsion system encountered failure during the
structural integrity test (Fig. 1a). The component experiences high cyclic loads
during operation, which comes from the unbalance loads on the structure. In order
to validate the design of the component with respect to the structural integrity, the
squirrel cage bearing was subjected to high-cycle fatigue test. However, the rear
flange of the cage was separated from the ribs during the test. There were secondary
damages to the other components of the test assembly. During dismantling, it was
found that the one of the support bearings in the testing assembly has also fractured
into three pieces along the outer race region, while the other support bearing was
found to be slightly damaged at the inner circumference of the outer race (Fig. 1b).
The material of the bearings was M50 NiL (AMS-6278A).
2 Observations
2.1 Squirrel Cage Bearing
2.1.1 Visual Observation
The bearing consisted of the inner race (IR), outer race (OR), front flange (FF), balls
and rear flange (RF). The rear flange is connected to the main bearing by means of
ribs. From the visual observation, it was found that out of total twelve ribs, all have
failed from rear flange side. The mating fractured surfaces of both the ribs and
corresponding region of the rear flange were severely damaged. It appeared most
probable that the dislodgement of the rear flange from the squirrel cage was
associated with the relative shear movement with respect to the connecting ribs
Failure Analysis of a Squirrel Cage Bearing of a Gas Turbine … 119
Fig. 2 Photograph showing a rib, inner race, outer race along with balls intact of the failed
bearing cage; b the separated ribs
which would have caused the damage on mating fractured surfaces of the ribs and
the flange. Four of the ribs failed from both the flange side and bearing side
(Fig. 2b). However, three of them could be traced from the test assembly after the
failure. The fractured surface of the ribs which failed from the bearing side was well
preserved. Half of the inner race showed indication of overheating (Fig. 3).
2.1.2 XRF Analysis
Semi-quantitative chemical analysis was carried out on the bearing by using

Portable XRF analyzer. The nominal composition details of the component are
Fig. 3 Photograph showing

presence of overheated region
(arrows) on the inner race
surface
Table 1 Chemical composition of the component as per XRF

S. No. Element Wt (%) of #3 bearing M50 NiL (ASTM standard)
1 Molybdenum 4.15 4
2 Vanadium 1.17 1
3 Nickel 3.42 3.5
4 Chromium 4.44 4
5 Carbon – 0.15
6 Iron Balance Balance
presented in Table 1. The chemical composition of M50 NiL as per ASM

Handbook [6] is listed in Table 1. Based on analysis, it was found that the material
of the squirrel cage bearing showed a match with M50 NiL.
2.1.3 Stereo Zoom Microscopy
The fractured surface of one of the ribs (Fig. 4) of #3 bearing was observed under
Stereo Zoom Microscope. The fractured surface revealed the presence of beach
marks [7] (Fig. 4) indicating that the failure of the rib was in progressive mode.
Tracing the orientation of the beach marks, two crack origins were identified; which
were located at the two diagonally opposite corners of the rib (Fig. 4). Similar
features were also observed on other two ribs fractured surfaces. It may be noted
that failure of all the ribs was from same position, i.e. rib joint region to the bearing
where chamfer is provided to avoid stress concentration. Macroscopic examination
at the failure locations of the ribs was carried out on the bearing cage. Two other
ribs indicated presence of crack near the rib-bearing joint (Fig. 5).
Fig. 4 Macrograph showing

location of crack
origins (arrows) and beach
marks
Fig. 5 Macrographs showing cracks (encircled) at two other ribs
2.1.4 Scanning Electron Microscopy for #3 Bearing Cage
A suitable piece was cut from the separated rib and was subjected to ultrasonic
cleaning with acetone before the scanning electron microscope (SEM) examination
on the fractured surface. Striations [7] (Fig. 6) were observed at higher magnifi-
cation confirming that the failure of the rib was by fatigue crack initiation and
propagation. Two independent fatigue crack fronts from two diagonally opposite
corners were found to propagate towards each other. These two crack fronts were
separated by a thin region (approximately 0.2 mm width) where dimples were
observed (Fig. 7). This observation indicated that as the two fatigue crack fronts
propagated from two diagonally opposites corners, the remaining cross section in
between the two crack fronts kept on reducing and finally when the remaining area
was unable to carry the load, it failed by overload. Distinct striations could be
observed on the fractured surface covering almost 90–95% fractured surface. No
metallurgical abnormalities (i.e. inclusions) were observed at the fatigue crack
origin.
2.2 Support Bearings
2.2.1 Visual Observation
The outer race inner circumference surface of the failed support bearing showed
presence of deformed and overheated region, as if the rollers have rubbed against
Fig. 6 Striations at higher

magnification on the fractured
surface of the rib
Fig. 7 Dimples (at the top of

the frame) observed in
between the two opposite
fatigue crack fronts
the mating surface (Fig. 8a). Similar features were observed on the inner race as
well (Fig. 8b). The cage supporting the ball bearings was crushed and separated
into two pieces (Fig. 8c). A small deformed region was also observed on the inner
circumference of the outer race of the other support bearing (Fig. 8d). Further,
discolorations of the balls were observed in the failed support bearings (Fig. 9).
2.2.2 Scanning Electron Microscopy for Support Bearing
The fractured surface of the failed support bearing, as observed under scanning
electron microscope, showed presence of dimples [8], indicating overload failure
(Fig. 10).
Fig. 8 a Photograph showing the deformed region in the inner circumferential region of outer
race of failed support bearing; b overheated and deformed region on the outer circumference of
inner race of failed support bearing; c fractured surface of the failed cage (inset) of support bearing;
d deformed overheated region in the other support bearing
Fig. 9 Photograph showing discoloration in the balls of failed support bearing

Fig. 10 Fractured surface of

the failed support bearing
outer race showing presence
of dimples
3 Analysis
Fractographic studies confirmed that three of the twelve ribs of the bearing cage
have failed in progressive mode, i.e. by fatigue crack initiation and propagation
from the bearing side. Cracks have originated from diagonally opposite corners of
the rib and have propagated inward towards the centre of the surface. 90–95% of
the fractured surface of the ribs showed a smooth region of fatigue crack propa-
gation followed by final fracture due to overload. The overload region was confined
to thin strip region of approximately 0.2 mm width. Cracks were also identified at
the similar location of two more ribs connecting the flange with the bearing. The
locations had sharp corners which are associated with stress concentration regions
and are normally prone to develop fatigue crack initiation sites. Existence of fatigue
crack origins at two opposite corners indicated that the bearing cage was subjected
to reverse bending fatigue loading during testing [7–9]. Such operating condition
may arise in two situations, (i) misalignment in the assembly and (ii) design defi-
ciency. The roller bearing in the present case experiences high axial load during
operation, while the centrifugal loads due to high rotational speed contribute to the
radial loads. Under such operating condition, a slight misalignment is expected to
introduce reverse bending fatigue loading [10] on the structural element of the
bearing, which is the cage in the present case. However, health monitoring
parameter like vibration did not indicate any abnormality in the signal. Therefore,
possibility of the misalignment leading to failure was ruled out. Stress analysis on
the component at a later stage revealed higher than expected imbalance load at the
failure location. Overload failure of the outer race of one of the support bearings
and the crushed cage indicates possible application of sudden load leading to the
failure of this part.
4 Conclusion and Recommendations
Based on the analysis, it is concluded that the few of the ribs in the squirrel cage
bearing failed by fatigue crack initiation and propagation. The failure of the ribs
occurred under the reverse bending fatigue loads. The failure of the ribs resulted in
instantaneous redistribution of the applied load causing the flange to dislodge from
the bearing as well as failure of the outer race of the support bearing.
Design modifications were introduced after thorough stress analysis of the
component. Subsequently, no such failure incidents were encountered, and the
modified design has been put into service.
Acknowledgements The authors express their gratitude to DRDO and Director, GTRE, for
supporting the work and granting permission to publish the data. The support and guidance
provided by the Technical and Associate Directors and the help extended by the Structural
Mechanics Group and NDT-Quality Assurance Group in completing the analysis also gratefully
acknowledged.
References
1. R. Rzadkowski et al., Analysis of middle bearing failure in rotor jet engine using tip-timing
and tip-clearance techniques. Mech. Syst. Signal Process. 76–77, 213–227 (2016)
2. T. Nishikawa, N. Hayashi, A. Hayakawa, Technical trend of aircraft bearings. NTN Tech.
Rev. 82, 83–87 (2014)
3. Bearing Failure: Causes and Cures, Technical Brochure, Barden Precision Bearings (2007)
pp. 1–15
4. Ball & Roller Bearings: Failures, Causes and Countermeasures, CAT. NO. B300 1 E (2009)
pp. 1–20
5. ASM Handbook, Properties and Selection: Irons, Steels, and High Performance Alloys, vol.
1 (2005)
6. Aircraft Turbine Engine Reliability and Inspection Investigations, AD-A274 860, DOT/FAA/
CT-92/29 (U.S. Department of Transportation, Federal Aviation Administration, 1993)
7. ASM Handbook, Failure Analysis and Prevention, vol. 11 (2002)
8. ASM Handbook, Fractography, vol. 12 (1987)
9. ASM Handbook, Fatigue and Fracture, vol. 19 (1996)
10. P.P. Milella, Fatigue and Corrosion in Metals (Springer-Verlag Italia, 2013). doi:https://doi.
org/10.1007/978-88-470-2336-9_2
Part II
Civil Structures
On-line Deformation Monitoring
of Building Estate Above Mining
R. Tsvetkov, Igor Shardakov, A. Shestakov and V. Yepin
Abstract Mineral extraction and mine flooding cause rock deformation and sub-
sidence of ground surface. Also it causes deformation and damage to structures and
buildings located on this surface. The paper presents deformation monitoring of
estate above mining. This system monitors vertical displacement (settlement) of
building foundations by the method of hydrostatic leveling. The monitoring data
allow to estimate the deformation of buildings and to evaluate changes in the tilt of
the ground surface at the location of the buildings. The article presents the results of
settlements of the buildings and the tilt of the ground surface over the 2 years of
observations. The obtained data are the basis for the analysis of the evolution of
deformation parameters, forecast implementation and correction of rock mass
behavior models.
Keywords Deformation monitoring Ground surface subsidence

Hydrostatic level
Nomenclature
S Settlement or vertical displacement
dSi;j Differential settlement between points i and j
b Angular distortion
x Tilt
k Curvature
R. Tsvetkov (&) I. Shardakov A. Shestakov V. Yepin

Institute of Continuous Media Mechanics UB RAS, 1, Korolev Street, 614013 Perm, Russia
e-mail: flower@icmm.ru

https://doi.org/10.1007/978-981-10-7197-3_11
130 R. Tsvetkov et al.
1 Introduction
The world economy constantly requires new natural resources which have led to
increases in mineral extraction, such as oil, coal or potash. Sometimes, mines are
located under building estate. Mineral extraction causes rock deformation and
ground surface subsidence. Also it causes deformation and damage to structures
and buildings located on this surface. Mine flooding increases these deformation
processes, and sinkholes may appear on the ground surface. For example, this
problem is described for the Verkhnekamsk salt deposit in [1]. To control
geomechanical processes of rock massif, different monitoring methods are required:
geomechanical [2], microseismic [3], and geochemical [4]. Monitoring of ground
surface subsidence is very important. Monitoring of ground surface can be done by
instrumental observations with extensometers [5]; by GPS (GNSS) [6, 7], by
satellite radar interferometry [8]. GPS detect quasi-static displacements of several
numbers of points with an accuracy of 3–4-mm GPS and instrumental observations
provide low-spatial resolution. Satellite radar interferometry (SAR) allows con-
trolling a large area, but it is less often (one time per two weeks). The accuracy of
the observations depends on the trajectory of the satellite in orbit and atmospheric
precipitation (snow). Therefore, SAR is not applicable for monitoring in the winter
time. Some buildings and structures that are located on land surface are equipped
with continuous deformation monitoring systems. This monitoring systems work at
any time of the year.
The aim of this work is the assessment of ground surface and rock massif based
on deformation monitoring data of building groups. This method of land surface
monitoring can be carried out continuously at any time of the year. This information
allows to verify monitoring data of rock massif monitoring, obtained by other
methods and to forecast a ground surface tilt.
2 Deformation Monitoring of Buildings
2.1 Measurement System
The foundation movement is the main reason of building damage. The measure-
ment system has to control the foundation displacement. Vertical displacement of
the foundation (or settlement) can be measured by the hydrostatic leveling method
[9]. This method controls the difference in vertical displacements of points where
sensors are installed. To measure the field of vertical displacements, it is possible to
use a multipoint hydrostatic leveling system with a necessary number of sensors.
The paper [10] describes the main aspects of using measurement systems for
controlling the differential settlements and estimates of the accuracy of this mea-
surement technique as (±0.5 mm).
On-line Deformation Monitoring of Building Estate Above Mining 131
Considering the design features of each building, positions of the sensors were
chosen. Installation of monitoring system components is carried out in the interior
basements that reduced the influence of environmental conditions. These systems
have been installed in more than 30 buildings that were located at several parts of
the massif.
2.2 The Approach of Estimation of a Deformation State

of the Building
The assessment of the safe operation of the building can be carried out based on
various criteria: estimating differential settlement dSi,j or angular distortion bi,j.
Differential (non-uniform) settlement is defined as the difference between vertical
displacements of two points where sensors are located. Angular distortion between
i and j sensors, having distance Li,j, is determined by the formula:
dSi; j
bi; j ¼ ð1Þ
Li;j
According to reference [11], the greatest tolerable differential settlement is

45 mm for a clayey soil and maximum angular distortion for buildings is 1/300.
The deformation assessment is also possible by calculating the change of cur-
vature of foundation section between three foundation points where sensors are
installed. The curvature of the building foundation between sensors i−1, i, and i + 1
is:
2 dSi1 2 dSi 2 dSi þ 1

k¼ þ ð2Þ
Li1;i Li þ 1;i1 Li1;i Li;i þ 1 Li þ 1;i1 Li;i þ 1
where dS—vertical displacements (settlements) of the sensors; L—distance

between the sensors.
In addition, based on the finite element modeling, it is possible to estimate the
change in the stress–strain state of building elements, using the measured values of
vertical displacements at the sensor location points.
2.3 Results of Deformation Monitoring of the Building
Several buildings have deformation monitoring systems. For example, consider a

typical 3-story brick building (length—30 m, width—12 m). The position of this
building on the terrain is shown in Fig. 1a marked as number 1. The location of the
sensors in the building foundation is shown in Fig. 1b. The sensors are placed on
Fig. 1 The position of the buildings on map (a), the location of the sensors (b)
the building perimeter, including basement corners where ground movements may
potentially represent danger.
The continuous monitoring of building started in 2014 and continues today.
Evolutions of vertical displacements (or differential settlements) at the sensors are
shown in Fig. 2a. The maximal settlement difference is about 14 mm for 2 years.
The diagrams show that the settlements occur at constant rates. Maximal differential
settlement rate (between sensor No. 5 and No. 2) is 0.5 mm per month.
Figure 2b shows changes in angular distortion of the foundation part between
the sensors, and Fig. 3 shows change of the line curvature of the foundation,
between the 3 sensors (1–3–5 and 2–4–6).
The submitted data demonstrate that differential settlements and angular dis-
tortions of the building foundation are increasing. However, the curvature of the
building foundation changes slightly. A certain number of buildings observed
exhibit similar behavior.
Fig. 2 Differential settlements at the sensor location points (a), changes in angular distortion of
the foundation (b)
Fig. 3 Change of the curvature of the building foundation
3 Rock Massif Monitoring
3.1 Estimation of Change in the Tilt of the Ground Surface
Local hydrostatic level systems are unable to give absolute values of vertical dis-
placements of buildings and give only the differential settlement of foundations,
which are localized in certain parts on the surface of the rock massif. Since the size
of buildings is less than a massif and magnitudes of curvature change slightly, let us
consider each building as a rigid element. According to this approximation, we
estimate the tilt of the building foundation, which is equal to the tilt of the part of
the ground surface under the foundation.
On the basis of the sensor positions in the building foundation ðxi ; yi Þ and
changes in vertical displacements for the appropriate time interval, calculation of
coefficients of the plane equation characterizing a building tilt was made by the
ordinary least squares method.
a xþb xþc xþd ¼ 0 ð2Þ
Using the calculated coefficients (a, b, and c), tilt angle x between the calculated
plane and the reference plane (0, x, and y) was estimated:
1
cos x ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ
ða=cÞ þ ðb=cÞ2 þ 1
2
Also, the angle a characterizing the tilt direction in the coordinates system
associated with the building was calculated:
b
tg a ¼ ð4Þ
a
Accuracy of the surface tilt measurement depends on accuracy of the hydrostatic

leveling system, number of sensors and building size. For this case, accuracy is
about 0.00002 rad.
3.2 Results
The calculation of the tilt change was made for each building with a monitoring
system. Figure 4a presents the map of building estate with monitored buildings
marked with numbers 1–7. The arrows associated with buildings represent the
projection of the normal to the ground surface under the building onto the hori-
zontal plane. The arrow direction corresponds to the direction of a tilt of a normal,
and the arrow length is proportional to the tilt value. The change in the tilt for
1-year observation is shown in Fig. 4a.
On this part of the massif, the tilt of the buildings and ground in the direction of
existing sinkholes which are in the left part of the figure is observed. The closer the
building to a sinkhole is located on the surface tilt is observed more intensively. The
obtained results show that it is also possible to estimate tilts of the constructions
located close to the buildings equipped with monitoring systems.
Figure 4b presents the evolution of the change in the ground surface tilt at
locations corresponding to the location of the buildings. The diagram shows that the
building tilt rate varies slightly with time. It is possible to notice some increase rates
of a tilt of the buildings located closer to sinkholes. Observation of a stable tilt rate
and direction angle for several years allows us to make a several month forecast.
Fig. 4 The change in the tilt of the buildings for 1 year (a) and evolution of the building tilts (b)
4 Conclusion
The presented results show that it is possible to monitor the ground surface above
mining using deformation monitoring data of buildings. This procedure is done all
year round, as well as deformation monitoring of buildings. The estimate of the
accuracy of slope measuring is about 0.00002 rad. This experimental results can
supplement the measurements obtained by other geodetic measuring systems and
can also be used to verify models that describe ground subsidence above mining [1].
Acknowledgements This research was performed under the grant of the Russian Science
Foundation (grant № 14-29-00172).
References
1. A.A. Baryakh, SYu. Devyatkov, N.A. Samodelkina, Theoretical explanation of conditions for
sinkholes after emergency flooding of potash mines. J. Min. Sci. 52, 36–45 (2016)
2. Z. Kalab, Lednicka long-term geomechanical observation in the Jeronym Mine. M. Acta
Geophys. 64, 1513–1524 (2016)
3. D. Amitrano et al., Microseismic activity analysis for the study of the rupture mechanism in
unstable rock masses. Nat. Hazard Earth. Syst. Sci. 10, 831–841 (2010)
4. B.A. Bachurin, A.A. Borisov, Gas-geochemistry approach to monitoring mining-induced
genesis processes in the geological environment of the Upper Kama Potassium Salt Deposit.
J. Min. Sci. 47, 861–866 (2011)
5. J.M. Pardo et al., Instrumental monitoring of the subsidence due to groundwater withdrawal in
the city of Murcia (Spain). Environ. Earth Sci. 70, 1957–1963 (2013)
6. F. Mancini et al., Monitoring ground subsidence induced by salt mining in the city of Tuzla
(Bosnia and Herzegovina). Environ. Geol. 58, 381–389 (2009)
7. H. Zhao, F. Ma, Y. Zhang, J. Guo, Monitoring and mechanisms of ground deformation and
ground fissures induced by cut-and-fill mining in the Jinchuan Mine 2. Environ. Earth Sci. 68,
1903–1911 (2013)
8. D. Walter et al., in Radar interferometric monitoring of sinkhole formations above a flooded
potash mine in Berezniki Russia. Proceedings XV International ISM Congress,
(Clausthal-Zellerfeld 2013), pp. 136–153
9. E. Meier et al., Hydrostatic levelling system: measuring at the system limits. J. App. Geodesy.
4, 91–102 (2010)
10. I.N. Shardakov, A.P. Shestakov, R.V. Tsvetkov, V.V. Yepin, The hydrostatic level method
for continuous monitoring of building foundations. SSP 243, 105–111 (2016)
11. J.E. Bowles, Foundation analysis and design (McGraw-Hill Companies, Singapore, 1997)
Mechanical Properties of Nano Concrete
Ajaysinh R. Vaghela and Gaurang R. Vesmawala
Abstract Nano materials have recently been introduced as an advanced pozzolana

to improve the microstructure and stability of the cement-based system. This paper
presents the effect of the addition of nano materials on the mechanical properties of
concrete. To compare the different nano materials total three nano materials were
taken along with polypropylene fiber. Nano silica and nano alumina were taken as
1, 2, 3, 4, and 5%; carbon nanotube was taken as 0.1, 0.2, 0.3, 0.4, and 0.5% of
cementitious material. Further, nano silica and nano alumina were mixed with
polypropylene fiber to check the combination of powder and fiber material.
Mechanical tests such as compression test, indirect tensile test, and flexure test were
performed to find mechanical properties. Results show an increase in compressive,
tensile, and flexural strength with the addition of nano materials.

Keywords Carbon nanotubes Cement composite Concrete Fiber reinforce-

ment Mechanical properties Fracture mechanics
1 Introduction
Nanotechnology in the construction industry is a new area where the majority of

research interest is focused. Nanotechnology is the most modern technology which
has its roots of origin and growth in last decades.
Due to increasing demand of concrete (about 2 billion tons per year), total
production of cement is also increasing which has very high amount of carbon
dioxide emission. It is expected that this value would go even higher and will go up
A. R. Vaghela (&) G. R. Vesmawala

Department of Applied Mechanics, Sardar Vallabhbhai National Institute of Technology,
Ichhanath, Surat 395007, Gujarat, India
e-mail: vaghelaajay11111@gmail.com
A. R. Vaghela
G H Patel College of Engineering and Technology, Vallabh Vidya Nagar 388120,
Gujarat, India

https://doi.org/10.1007/978-981-10-7197-3_12
138 A. R. Vaghela and G. R. Vesmawala
to 4 times than what was in 1990 [1]. One possibility of reducing the emission of
CO2 is manufacturing of strong concrete which allows to use less amount of
concrete for construction. To increase, the strength researchers have tried different
supplementary materials to cement. It is considered that nanotechnology is one of
the main research areas which can improve the concrete properties, i.e., strength and
durability.
Nanoparticles quicken C–S–H gel creation as an aftereffect for expanded crys-
talline Ca(OH)2 amount particularly at the initial period of hydration and increase
the strength of the concrete [2]. Researchers have shown that among largely
available materials, SiO2 and Al2O3 exhibit more effect on mechanical properties of
concrete [3]. Nano silica and nano Alumina increase compressive strength and
decrease water absorption with different % of nanoparticles [4]. Nano Silica (5% by
weight of cementitious materials) improves durability parameters of concrete more
as compared to nano alumina (3% by weight of cementitious materials), i.e.,
compressive strength and frost resistance increased up to 30 and 83%, respectively,
for nano Silica as compared to 8 and 81%, respectively, for Nano Alumina [5].
Nano Silica, nano alumina, and their combination increase surface hardness as well
as abrasion and fracture resistance [6] which prove their applicability for improving
the fracture toughness of concrete. Nano Silica also increases electrical resistivity of
concrete about 100–150% [7].
Apart from widely famous nano materials carbon nanotubes (CNT) also gained
popularity due to their unique properties. CNTs have very high Young’s modulus,
high strength, and high electrical and thermal conductivity. CNTs allow the
development of materials with enhanced performance characteristics. Researchers
have found that addition of nanotubes affected the structure of cement hydration
products, both the initial C3A and C3S hydration products. They also observed that
CNTs accelerate the rate of hydration process by acting as a matrix for the
development of C–S–H and Ca(OH)2 produced during the hydration [8]. Presence
of oxygen-containing groups on the surface of CNTs makes possible chemical
reaction between CNTs and the cement matrix [9]. CNTs go about as nucleating
agent in cement hydration which increases the possibilities of reaction to occur and
increases the formation of reaction products [10]. It is important to increase the
flexure strength of concrete since the same would result in increased fracture
toughness of the material.
In this paper, the compressive strength, tensile strength, and flexure strength of
concrete containing various proportions of nano Silica (NS), nano Alumina
(NA) and carbon nanotubes (CNT) are presented. Moreover, combination of NS
and NA with polypropylene fiber is also checked.
Mechanical Properties of Nano Concrete 139
2 Experimental Program
2.1 Material Used
Ordinary Portland Cement of 53 grade was used in the study. Sand from river
having fineness modulus of 2.82 was used as fine aggregate. Coarse aggregate of
maximum size 20 mm was used. Multiwall carbon nanotubes (MWCNTs) pur-
chased from Cheap Tubes (USA), nano Alumina purchased from Advanced
Technology Materials (Mumbai, India), and nano Silica purchased from Adinath
Industries (Ajmer, Rajasthan, India) having properties as shown in Table 1 were
used in the work. The transmission electron microscope image of the MWCNT and
nano Alumina is also presented in Fig. 1.
Table 1 Properties of MWCNTs, nano silica, and nano alumina

Property MWCNTs Nano silica Nano alumina
Appearance form Powder Powder Powder
Purity 90% 98% >99.8%
Outer diameter 20–40 nm 15–30 nm 40–75 nm
Inner diameter 5–10 nm – –
Length 10–30 lm – –
Specific surface area >110 m2/g – 20–30 m2/g
Bulk density 0.07 g/cm3 – –
True density *2.1 g/cm3 – –
Fig. 1 a TEM image of nano alumina and b TEM image of CNT (by the manufacturer data sheet)
Table 2 Mix proportions Material Quantity

Cement 450
Water 180
Coarse aggregate 1121.208
Fine aggregate 659.604
Admixture 0
2.2 Mix Proportions
Mix proportion of ordinary concrete was done by IS: 10262-2009, and the quan-
tities of different materials were obtained as shown in Table 2. Nano materials were
taken as % by weight of cement. Totally 17 different mixes were taken into
consideration. Nano Silica and nano Alumina were taken in 5 different proportions,
i.e., 1, 2, 3, 4, and 5% by weight of total cementitious materials. CNTs were taken
as 0.1, 0.2, 0.3, 0.4, and 0.5% by weight of total cementitious materials. The higher
percentage of CNTs were taken to check the behavior of concrete with increased
pozzolana. Moreover, combinations of nano silica and nano alumina with
polypropylene fiber were also taken. Nano silica and nano alumina were taken as 3
and 2% of total cementitious material with 0.2% of polypropylene fiber.
2.3 Testing Procedure
For compression test of concrete, cubes were prepared for each mix with dimension
15 cm 15 cm 15 cm according to IS:516-1959. Three cubes were prepared
for each test day and for each mix. Cubes were tested at 7, 14, and 28 days. Indirect
tensile test specimens were prepared for each mix having a cylindrical shape with
dimension 15 cm diameter 30 cm length according to IS:5816-1999. Cylinders
were tested at 28 days for tensile strength. For flexure, beams were prepared for all
mix having dimensions 10 cm 10 cm cross section with a 50 cm length
according to IS:516-1959.
Calculation based on peak load for indirect tensile strength and flexure strength
was done with following Eqs. (1) and (2), respectively.
2P
fct ¼ ð1Þ
pld
where,
P Maximum load applied in N,
d Cross-sectional dimension of the specimen in mm, and
l Length of the specimen in mm.
pl
fb ¼ ð2Þ
b d2
where,
P Maximum load applied in kg,
b Measured width of the specimen in cm,
d Measured depth of the specimen at failure point in cm, and
l Length or span on which the specimen is supported in cm.
3 Result and Discussion
3.1 Compressive Strength
Compressive strength of nano concrete at 7, 14, and 28 days is shown in Table 3.

The values are the average of 3 cube strength at each testing day. It can be found
that the compressive strength of nano concrete increases in almost all cases as
compared to ordinary concrete.
Table 3 Compressive strength of specimen (fc) in MPa

Mix Compressive strength at Compressive strength at Compressive strength at
7 days in MPa 14 days in MPa 28 days in MPa
fc Enhanced fc Enhanced fc Enhanced
extent (%) extent (%) extent (%)
PC 36.7 0.00 43.5 0.00 49.2 0.00
NS1 37.2 1.36 41.8 1.61 51.2 4.07
NS2 39.2 6.81 44 1.15 53.8 9.35
NS3 40.3 9.81 49.2 13.10 58.7 19.30
NS4 36.4 0.27 44.2 1.61 52.6 6.91
NS5 38 3.54 46.6 7.13 53.8 9.35
NA1 35.2 0.54 44.8 2.99 50.6 2.85
NA2 38.8 5.72 49.6 14.02 56.5 14.84
NA3 36.6 0.27 48.6 11.72 53.8 9.35
NA4 37.3 1.63 46.2 6.21 52.4 6.50
NA5 36.8 0.27 49.3 13.33 54.7 11.18
CNT0.1 36.6 0.54 45.3 4.14 52.7 7.11
CNT0.2 36.9 0.54 48.2 10.80 54.3 10.37
CNT0.3 38.2 4.09 47.5 9.20 56.3 14.43
CNT0.4 37.8 3.00 49.2 13.10 57.2 16.26
CNT0.5 38.3 4.36 48.2 10.80 58.5 18.90
NS3PF0.2 37.8 3.00 46.9 7.82 53.2 8.13
NA2PF0.2 36.7 0.00 45.8 5.29 52.6 6.91
Concrete-containing nano silica increases the strength up to 9.81 and 19.30% for
7 and 28 days, respectively. The reason for increased strength is pozzolanic reac-
tion of nano silica. Nano materials react with calcium hydroxide Ca(OH)2 crystals
which form C–S–H gel. Nano materials also act as a filler of voids in concrete
which makes concrete denser. It can be seen from the Table 3 that the optimum
dose of nano silica in concrete is 3% by weight of all cementitious materials as
results beyond 3% decrease with further increase in nano silica content. Decrease in
strength may be due to increase in week zones in concrete with increased nano
silica content as it imparts negative effects on hydration process. Nano alumina also
shows same behavior as nano silica with maximum dosage of nano Alumina found
as 2%.
Addition of MWCNTs increases the formation of C–S–H gel resulting in
stronger material. The reason for increased strength could be possession of voids by
MWCNTs particles which resist loading. Moreover, it makes concrete denser which
increases its crack resistivity. For MWCNTs, increases in the strength were reported
as 18.90% as compared to ordinary concrete.
3.2 Tensile Strength
Table 4 shows values of tensile strength for nano concrete with all mixes, and result
increases in all cases. MWCNTs addition is found to be suitable as far as tensile
strength is concerned. The increase in tensile strength as compared to ordinary
concrete is up to 69.57%. This is mainly because of the large modulus of elasticity
of MWCNTs which increases stress carrying capacity of concrete. Similar results
have been reported by other researcher [3]. For other nano materials, the increase in
tensile strength is not as higher as MWCNTS but an increase in tensile strength can
be seen up to an extent. For combination of nano silica and nano alumina with
polypropylene fibers stand beneficial as compression as well tensile strength also
increase with combination.
3.3 Flexural Strength
Flexural test results of all specimens at 28 days are shown in Table 5. Results show
same behavior of nano material in flexure as in compression. Nano silica and nano
alumina show same optimum dosage of 3 and 2%, respectively, for flexure as well.
Table 4 Tensile strength of Mix Tensile strength in MPa

specimen (fct) in Mpa
fct Enhanced extent (%)
PC 2.3 0.00
NS1 2.5 8.70
NS2 2.6 13.04
NS3 3.1 34.78
NS4 2.7 17.39
NS5 2.2 −4.35
NA1 2.4 4.35
NA2 2.8 21.74
NA3 3.0 30.43
NA4 2.9 26.09
NA5 2.5 8.70
CNT0.1 2.9 26.09
CNT0.2 3.2 39.13
CNT0.3 3.5 52.17
CNT0.4 3.9 69.57
CNT0.5 4.2 82.61
NS3PF0.2 3.4 47.83
NA2PF0.2 3.2 39.13
Table 5 Flexural strength of Mix Flexure strength in MPa

concrete (fb) in MPa
fb Enhanced extent (%)
PC 5.8 0.00
NS1 6.1 5.17
NS2 6.5 12.07
NS3 7.2 24.14
NS4 7.0 20.69
NS5 6.9 18.97
NA1 5.9 1.72
NA2 6.6 13.79
NA3 6.4 10.34
NA4 6.2 6.90
NA5 6.3 8.62
CNT0.1 7.8 34.48
CNT0.2 8.6 48.28
CNT0.3 9 55.17
CNT0.4 9.3 60.34
CNT0.5 9.4 62.07
NS3PF0.2 8.2 41.38
NA2PF0.2 7.9 36.21
Fig. 2 a and b SEM images of MWCNTs
4 Conclusions
1. Content of nano silica increases compressive and flexural strength up to 3% by

weight of cementitious material, and after further addition, strength decreases
which give optimum dosage of nano silica as 3%.
2. Nano Alumina exhibits better compression, tensile, and flexural behavior as
compared to ordinary concrete, but takes a back step in comparison with other
nano material studied. The optimum dosage is found out to be 2% in com-
pression only for nano alumina.
3. MWCNTs due to their high modulus of elasticity can withstand stresses more as
compared to other nano materials studied. For improvement of tensile strength
and Flexural strength, MWCNTs are the best available nano material.
4. Combination of nano silica and nano alumina with polypropylene fibers holds
good as nano materials increase compressive strength and fibers increase tensile
strength of concrete.
5. Increased flexural strength opens up a new scope for increasing fracture
toughness of nano concrete using Nano silica, Nano Alumina, and MWCNTS.
References
1. The Concrete Conundrum, Chemistry World, March (2016). www.chemistryworld.org

2. A. Nazari, S. Riahi, The effects of SiO2 nanoparticles on physical and mechanical properties
of high strength compacting concrete. Compos. B 42, 570–578 (2011)
3. K. Behfarnia, N. Salemi, The effects of nano-silica and nano alumina on frost resistance of
normal concrete. Constr. Build. Mater. 48, 580–584 (2013)
4. M. Oltulu, R. Sahin, Effect of nano-SiO2 nano-Al2O3 and nano-Fe2O3 powders on
compressive strengths and capillary water absorption of cement mortar containing fly ash.
Energy Build. 58, 292–301 (2013)
5. N. Salemi, K. Behfarnia, Effect of nano-particles on durability of fiber-reinforced concrete

pavement. Constr. Build. Mater. 48, 934–941 (2013)
6. N. Leon, Effect of nano-SiO2 and nano-Al2O3 on cement mortars for use in agriculture and
livestock production. Biosystem engineering 123, 1–11 (2014)
7. H. Madami, A. Bagheri, T. Parhizkar, A. Raisghasemi, Chloride penetration and electrical
resistivity of concretes containing nanosilica hydrosols with different specific surface areas.
Cem. Conc. Compos 53, 18–24 (2014)
8. T. Manzur, N. Yazdani, M.A.B. Emon, Effect of carbon nanotube size on compressive
strengths of nanotube reinforced cementitious composites. J. Mat. 2014 (2014)
9. N. Zohadi, Functionalized graphitic nanoreinforcement for cement composites (University of
South Carolina 2014)
10. S.G. Shah, S.G. Shah, Q.I. Barodawala, Nano modification of self compacting concrete.
CTMC (2016)
11. Concrete Mix Design Revised, Ambuja technical literature series no. 79
12. Design of High strength Concrete Mixes, Proceeding of a Symposium on Mix Design and
Quality control of concrete. (Cement and Concrete Association, London)
13. Design of Normal Concrete Mixes. (Department of the Environment, Building research
Establishment, Walford, U.K. 1988)
14. IS: 10262, Concrete mix proportioning—guidelines (2009)
15. IS: 383, Specification for coarse and fine aggregates from natural sources for concrete (second
revision) (1970), (reaffirmation, 1997)
16. IS: 516, Indian standard methods of tests for strength of concrete (1959), (reaffirmation, 1999)
17. IS: 5816, Splitting tensile strength of concrete—method of test (1999), (reaffirmation 2004)
18. M.S. Shetty, Concrete Technology Theory and Practice. (S. Chand publication 2009)
Estimation of Critical Microcrack Length
in Concrete by Considering Interfacial
Properties
Keerthy M. Simon and J. M. Chandra Kishen
Abstract Concrete is considered as a homogeneous material from a design per-

spective. However, on a close examination, it is observed that concrete is hetero-
geneous, which consist of coarse aggregate and fine aggregate embedded in cement
paste. Further, there exists an interfacial region, which bonds the aggregate with the
cement paste. The strength of this interfacial transition zone (ITZ) depends on its
microstructural characteristics. This interfacial region neither possesses the prop-
erties of aggregate nor of the cement paste. Interface being the weakest zone, the
microcracks are likely to initiate here when the local major principal stress exceeds
the initial tensile strength of the interface. When these microcracks reach certain
critical length, it propagates and coalesces with the existing macrocrack to form a
major crack resulting in the failure of the bond. The microstructural character of the
interfacial zone governs the mode I crack propagation in conventional concrete. The
material behavior of concrete is influenced by the geometry, the spatial distribution,
and the material property of the individual constituents and their interactions. This
study aims at estimating the critical microcrack length using the principles of linear
elastic fracture mechanics (LEFM) by analyzing the crack opening displacement at
different scales. Also, a procedure to determine the material properties such as the
elastic modulus and fracture toughness at the interface by knowing the concrete mix
proportion is explained.
Keywords Concrete Critical microcrack Interfacial transition zone
1 Introduction
During mixing of concrete, the shearing stress exerted on the cement paste by the
aggregates tends to separate water from cement particles and results in a small
region around the aggregate particles with fewer cement particles termed as the ITZ
K. M. Simon J. M. Chandra Kishen (&)

Indian Institute of Science, Bangalore 560012, India
e-mail: chandrak@iisc.ac.in

https://doi.org/10.1007/978-981-10-7197-3_13
148 K. M. Simon and J. M. Chandra Kishen
[1]. The composition of this zone was found to be different from that of cement
paste as the density of cement particles was lesser than that of its surroundings.
Although several techniques such as X-ray diffraction, X-ray photoelectron spec-
troscopy, secondary ion mass spectroscopy, mercury intrusion porosimetry, optical
microscope, and electron microscope have been employed to study the ITZ, the
complete characterization of the distinctive features of this zone has still not been
achieved. The interface acts as a bridge between aggregate and the cement paste,
which neither possesses the properties of aggregate nor the cement paste.
Upon loading, microcracking initiates in the ITZ when the local major principal
stress exceeds the initial tensile strength of the interface [2]. Ansari [3] reported that
the macrocrack would occur only when the microcrack reaches a critical length.
This critical microcrack propagates and coalesces with existing macrocrack thereby
resulting in the failure of the bond. The stiffness of the interface decreases although
the individual components on either side of the interface possess high stiffness. This
is due to the presence of voids and microcracks in this region, which do not allow
the transfer of stress.
The packing and density of the cement particle around the aggregates define the
strength of the interfacial zone. The strength of an interface decides whether a crack
should grow around an aggregate particle or through the aggregate. Also, the
fracture energy of the interface is found to be less than that of the cement paste and
aggregate. The microstructural character of the interfacial zone governs the mode I
crack propagation in concrete [4]. Thus, the material behavior of concrete is
influenced by the geometry, the spatial distribution, and the material property of the
individual material constituents and their mutual interaction. Hence, the failure of
concrete structure can be viewed as a multiscale phenomenon, wherein the infor-
mation of the material properties at a microlevel can be used to determine the
system behavior at the macrolevel.
In this paper, the critical length of the microcrack present in the ITZ is estimated
using the principles of linear elastic fracture mechanics. This is achieved by relating
the crack opening displacements at two different scales, namely micro and macro.
A procedure to determine the material properties at the ITZ including the elastic
modulus and the fracture toughness by knowing the mix proportions of the
ingredients in concrete is explained. The critical microcrack length thus determined
can be treated as a material property and can be used further in predicting the
macroscopic behavior of concrete.
2 Critical Microcrack Length
The initiation and propagation of microcrack at the interface between the cement
paste and the aggregates are attributed to the toughness of the interface. In this
study, the critical microcrack length is a parameter, which is used to characterize the
interface. As the critical microcrack length increases, the toughness of the interface
increases. Thus, it influences the overall strength of concrete. The existence of
Estimation of Critical Microcrack Length in Concrete … 149
Fig. 1 Representation of microcrack at interface
microcrack and restraining stress at the macrocrack tip is a well-known fact. Upon
loading, the microcrack will initiate at the interface and later coalesce with the
existing macrocrack resulting in the crack increment. In this study, the critical
length of the microcrack is determined by analyzing a small element near the
macrocrack tip and along the interface between the cement paste and aggregate, as
depicted in Fig. 1.
The following assumptions are made in the derivation of the critical microcrack
length: (1) The microcrack grows in a direction perpendicular to the maximum
principal stress. (2) The initial microcrack length is assumed to be much smaller
than the size of the element considered. (3) The microcrack tip is sharp for linear
elastic fracture mechanics to be applied.
The stresses and displacement along the crack tip for this two-dimensional crack
problem are determined through an inverse method by making use of an Airy stress
function (Ф), which satisfies the biharmonic equation (∇2∇2 Ф = 0). The stresses
and strains in the polar coordinate system are further determined. Using these
strains, the displacements in the polar coordinate system are obtained as
X
1
1 4
Vr ¼ r kn ðkn þ 1ÞBn cosðkn þ 1Þh þ Dn cosðkn 1Þh ðkn þ 1Þ
2lmicro n¼0
1 þ vmicro
ð1Þ
X
1
1 4
Vh ¼ r kn ðkn þ 1ÞBn sinðkn þ 1Þh þ Dn sinðkn 1Þh þ ðkn 1Þ
2lmicro n¼0
1 þ vmicro
ð2Þ
In the above equations, lmicro and mmicro are the shear modulus and the Poisson’s
ratio of the interface, respectively, where the microcrack is likely to occur. The
microcrack present in the interface is assumed to be sharp in order to initiate the
crack propagation. The crack surface of the microcrack is considered as stress free,
and the corresponding boundary conditions along the upper surface denoted by (+)
and lower surface denoted by (−) are given by
rhh ¼ rrh ¼ 0; h ¼ p ð3Þ
Assuming the microcrack to be very sharp, the microcrack angle attains a value
of p, and the characteristic equation reduces to sin(2knp) = 0. The roots of this
characteristic equation give the eigenvalues (kn = n/2; n = 0, ±1, ±2, …). The
crack-tip singularity is observed when the eigenvalue becomes 0.5. The displace-
ment field near the microcrack tip reduces to the following form

r 1=2 1 1 1 1
Vr ¼ þ 1 B1 cos þ1 h þ 1 D1 cos 1 h
2lmicro 2 2 2 2
ð4Þ
4 1
þ D1 cos 1 h
1 þ vmicro 2

r 1=2 1 1 1 1
Vr ¼ þ 1 B1 sin þ1 hþ 1 D1 sin 1 h
2lmicro 2 2 2 2
ð5Þ
4 1
þ D1 sin 1 h
1 þ vmicro 2
A relationship between B1 and D1 is obtained by substituting the eigenvalue into

the boundary condition (B1 = D1/3). This reduces the displacement field in terms of
one unknown parameter D1. Further, by defining D1 = K1/√(2p), the displacement
field becomes a function of stress intensity factor (K). The crack opening dis-
placement corresponding to the microcrack is determined by transforming the
displacement component perpendicular to the loading direction (V) into rectangular
coordinates by making use of the following transformation
V ¼ V r sinðhÞ þ V h cos h ð6Þ
By substituting h = + b* and h = −b*, the displacement field along the upper

(V ) and the lower (V−) surface of the crack is obtained. The corresponding dis-
+
placement fields are given by
V þ ¼ Vr sinðhÞ þ Vh cosðhÞjh¼ þ b ð7Þ
V ¼ Vr sinðhÞ þ Vh cosðhÞjh¼b ð8Þ
The following equation provides the corresponding microcrack opening dis-

placement (dmicro):
dmicro ¼ V þ V ð9Þ
rffiffiffiffiffiffi
r K
d micro
¼8 ð10Þ
2p E
Fig. 2 Representation of
microcrack an macrocrack at
peak load
As the microcrack initiates in the interfacial region, the stress intensity factor
(K), elasticity modulus (E), and r are replaced with KInterface, EInterface, and l/2 (as
the total length of microcrack is assumed to be l as in Fig. 2). When the microcrack
reaches a critical length (i.e., l = lc), the corresponding fracture toughness and crack
opening displacement reach their critical value (KInterface
IC , dc) and are given by,
rffiffiffiffi
Interface
lc KIC
d micro
¼4 Interface
ð11Þ
pE
Microcracks are nucleated in concrete structure when the applied load reaches
80% of peak load. When the applied load reaches the peak load, the microcrack
becomes critical, and it coalesces with the macrocrack. Figure 2 represents the
existence of the microcrack at the macrocrack tip considered in this study.
The macrocrack length and the stress corresponding to the peak load (Ppeak) are
represented as the critical macrocrack length (ac) and peak stress (rp). The crack
length corresponding to the peak load is determined by knowing the crack mouth
opening displacement (dp) at the peak load. The following equation provides the
crack opening displacement at any point (x) along the macrocrack:

4rp ac ac ac x
dmacro ¼ g2 g3 ; ð12Þ
E D D ac
where E is the elastic modulus of concrete, D is the depth of the specimen, and the
geometric factors g2(ac/D) and g3(x/ac, ac/D) for beam specimen are taken from
standard reference [5]. From Fig. 2, it is clear that the crack opening displacement
due to the macrocrack at (x = ac − lc/2) will be equal to the crack opening dis-
placement due to the microcrack at r = lc/2. The critical microcrack length is
determined by equating both the crack opening displacements that are given by
Eqs. (11) and (12). The solution of the following equation gives the critical
microcrack length.
dmacro x¼ðac l2c Þ ¼ dmicro ð13Þ

r¼l2c
rffiffiffiffi
Interface
4rp ac ac ac x lc KIC
g2 g3 ; ¼4 ð14Þ
E D D ac p EInterface
As seen from the above equation, the critical microcrack length can be obtained
by knowing the fracture toughness and elastic modulus of the interfacial zone. The
procedure for estimating the interfacial properties is explained in the following
section.
3 Estimation of Interfacial Properties
3.1 Interfacial Fracture Toughness
The lower density of cement particles around the aggregates makes the interface
weaker. The initiation and propagation of a microcrack depend on the toughness of
the interface. Hillemier and Hilsdorf [6] conducted experiments and analytical
investigations to determine the fracture properties of hardened cement paste,
aggregate, and cement-paste interface. Huang and Li [7] considered the nucleation of
a crack along the interface of aggregate and mortar and derived a relation between
the effective toughness of the material and the mortar in terms of volume fractions by
considering the crack deflection and interfacial cracking effects, which is
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffisffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
KIC 1
m ¼ 1 þ 0:87 Vf ðcaÞ ð15Þ
KIC 1 16 Vf ðcaÞ 1 m2eff
p2
where KIC is the fracture toughness of concrete, Km IC is the fracture toughness of

mortar, Vf(ca) is the volume fraction of coarse aggregate, and meff is the Poisson’s
ratio of concrete. In this work, a relation between the toughness of matrix and
cement paste is derived by suitably modifying and replacing the fracture toughness
of concrete and mortar with those of mortar and cement paste. Also, the volume
fraction of coarse aggregate and Poisson’s ratio of concrete is replaced with that of
fine aggregate and Poisson’s ratio of mortar, which takes the form
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
KICm qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
cp ¼ 1 þ 0:87 Vf ðfaÞ ð16Þ
KIC 1 16 Vf ðfaÞ 1 m2m
p2
The fracture toughness of cement paste–aggregate interface is much lower than

that of cement paste. Hillemier and Hilsdorf [6] conducted experiments to deter-
mine the fracture toughness of the interface and the cement paste and observed the
toughness of the interface to be 0.4 times of the cement paste (i.e., KInterface
IC = 0.4
Kcp
IC ), which is used in the present study.
3.2 Interfacial Elastic Modulus
The elastic modulus of the interface is another major parameter used to define the
critical microcrack length (lc) and is reported to be 0.7 times the elastic modulus of
cement paste (Ecp) [8] and is used in this study. Hashin [9] obtained a relation
between the elastic modulus of homogeneous material with and without inclusion,
based on volume fraction and modulus of the inclusion by considering the change
in the strain energy of a loaded homogeneous body due to the insertion of inho-
mogeneities using the variational theorems in the theory of elasticity. Accordingly,
the relation between the elastic modulus of concrete and mortar is expressed as,
E Vf ðmÞEm þ 1 þ Vf ðcaÞ Eca

¼ ð17Þ
Em 1 þ Vf ðcaÞ Em þ Vf ðmÞEca
where Vf(m) and Vf(ca) are the volume fraction of mortar and coarse aggregate, and
E, Em, and Eca are the elastic modulus of concrete, mortar, and coarse aggregate,
respectively. The above equation can be simplified to the form,

Vf ðmÞEm2 þ 1 þ Vf ðcaÞ Em ½Eca E Vf ðmÞEca E ¼ 0 ð18Þ
The elastic modulus of the mortar is obtained by solving the above quadratic
equation. The relationship between the modulus of elasticity of mortar and cement
paste is obtained in a similar manner by replacing the material properties of con-
crete and mortar with the properties of mortar and cement paste. Also, the volume
fraction of coarse aggregate and mortar is replaced with those of fine aggregate and
cement paste. The modulus of elasticity of cement paste is determined by solving
the following quadratic equation.

Vf ðcpÞEcp
2
þ 1 þ Vf ðfaÞ Ecp Efa Em Vf ðcpÞEfa Em ¼ 0 ð19Þ
where Vf(cp) and Vf(fa) are the volume fraction of cement paste and fine aggregate,
and Em, Ecp, and Efa are the elastic modulus of mortar, cement paste, and fine
aggregate, respectively.
4 Analysis of Experimental Data
The interfacial properties, such as the modulus of elasticity and fracture toughness
and the critical microcrack length as discussed in the previous section, are evaluated
for normal strength concrete used by researchers in their experimental program. The
following experimental data are considered in this analysis: Bazant and Xu [10] and
Shah and Chandra Kishen [11]. In both research works mentioned above, tests have
been carried out on beams of three different sizes (designated as small, medium, and
large), which are geometrically similar under three-point bending. Table 1 shows
the dimensions of the beams, the peak stress, and the crack mouth opening dis-
placement (CMOD) at peak load. The material properties, including the fracture
toughness and elastic modulus as reported in these experimental works, are given in
Table 2. The mix proportion used in the experiments conducted by Bazant and Xu
[10] and Shah and Chandra Kishen [11] are 1:1.86:2.61:0.54 and 1:2:2:0.6,
respectively.
As discussed in the previous sections, the interfacial properties of concrete
depend on the volume fraction of each of its constituents. The interfacial fracture
toughness, the interfacial elastic modulus, and the critical microcrack length, cal-
culated using the above-mentioned procedure, are tabulated in Tables 2 and 3. It is
seen that the interface is the weaker region and is therefore more prone to cracking
and that justifies the analysis of the interfacial region. Furthermore, as reported in
the literature [5], if the interface is weaker than the coarse aggregates, the crack
propagates around the aggregate. In Table 2, it is also observed that for a particular
concrete mix, the computed interfacial properties remain constant.
As reported in Table 3, the critical microcrack length is found to be dependent
on the properties of the interface as well as the geometry of the specimen. Also, it is
found to be increasing with the size of the specimen, even though the interfacial
properties remain the same. Also, Mobasher et al. [12] have reported that 80% of
microcracks are smaller than 1.5 mm in length. The critical length of microcrack is
found to be comparable with the size of the fine aggregate and cement particles.
Further, the critical microcrack length is normalized with the depth of beam as
shown in Table 3. It is seen that this normalized value remains almost constant for a
given mix of concrete and could be treated as a material property.
Table 1 Geometry and material properties of specimens

Specimen Depth Span Thickness Stress at peak CMOD at peak
designation load load
D (mm) S (mm) B (mm) ru (MPa) dp (mm)
Small [11] 76 190 50 4.476 0.0567
Medium [11] 152 380 50 3.589 0.0743
Large [11] 304 760 50 3.338 0.0483
Small [10] 38.1 95 38.1 4.750 0.0270
Medium [10] 76.2 191 38.1 3.833 0.0354
Large [10] 152.4 381 38.1 3.364 0.0436
Table 2 Material properties of concrete and interfaces

Fracture toughness Elastic modulus (N/mm2) References
(MPa√mm)
Concrete Interface Concrete Interface
KIC KInterface
IC E EInterface
44.6 16.76 30,000 5413.5 [11]
32.0 12.45 27,120 5488.7 [10]
Table 3 Critical microcrack length

Specimen designation Critical microcrack length (lc) (mm) lc/D
10−2
Small [11] 0.3659 0.39
Medium [11] 0.5717 0.38
Large [11] 0.9238 0.37
Small [10] 0.2370 0.62
Medium [10] 0.4540 0.60
Large [10] 0.9167 0.60
5 Conclusions
In this study, a critical microcrack length parameter is defined, and an expression is

derived by analyzing the crack opening displacement at micro- and macroscales.
The critical microcrack length is found to be dependent on the specimen size. Also,
it is observed that, for a particular mix of concrete, the ratio of critical microcrack
length to the specimen depth is found to be constant and can be used as a material
property. The critical microcrack thus determined can be further used in predicting
the macroscopic behavior of concrete.
References
1. J.C. Maso, Interfacial transition zone in concrete, RILEM report, vol. 11 (CRC Press, UK,
2004)
2. Iulia Carmen Mihai and Anthony Duncan Jefferson, A material model for cementitious
composite materials with an exterior point eshelby microcrack initiation criterion. Int.
J. Solids Struct. 48, 3312–3325 (2011)
3. Farhad Ansari, Mechanism of microcrack formation in concrete. ACI Mat. J. 86, 459–464 (1989)
4. J.G.M. Van Mier, A. Vervuurt, Numerical analysis of interface fracture in concrete using a
lattice-type fracture model. Int. J. Damage Mech 6, 408–432 (1997)
5. S.P. Shah, S.E. Swartz, C. Ouyang, Fracture mechanics of concrete: applications of fracture
mechanics to concrete, rock and other quasibrittle materials. (Wiley, NY, 1995)
6. B. Hillemeier, H.K. Hilsdorf, Fracture mechanics studies on concrete compounds. Cem.
Concr. Res. 7, 523–535 (1977)
7. J. Huang, V.C. Li, A meso-mechanical model of the tensile behavior of concrete. Part II:
modelling of post-peak tension softening behavior. Composites 20, 370–378 (1989)
8. C.C. Yang, Effect of the transition zone on the elastic moduli of mortar. Cem. Concr. Res. 28,
727–736 (1998)
9. 1Zvi Hashin, The elastic moduli of heterogeneous materials. J. Appl. Mech. 29, 143–150
(1962)
10. Z.P. Bazant, K. Xu, Size effect in fatigue fracture of concrete. ACI Mat. J. 88, 390–399 (1991)
11. S.G. Shah, J.M. Chandra Kishen, Fracture properties of concrete–concrete interfaces using
digital image correlation. Exp. Mech. 51, 303–313 (2011)
12. B. Mobasher, H. Stang, S.P. Shah, Microcracking in fiber reinforced concrete. Cem. Concr.
Res. 20, 665–676 (1990)
Fracture Analysis and Remaining Life
Assessment of Ultra High Strength
Concrete Beams
A. Ramachandra Murthy, N. R. Iyer and B. K. Raghu Prasad
Abstract This paper presents the details of crack growth study and remaining life
assessment of concrete specimens made up of ultra-high-strength concrete (UHSC).
Flexural fatigue tests have been conducted on UHSC beams under constant
amplitude loading with a stress ratio of 0.2. A “Net K” model has been proposed by
using nonlinear fracture mechanics principles for crack growth analysis and re-
maining life prediction. SIF (K) has been computed by using the principle of
superposition. Bilinear tension softening obtained through inverse analysis corre-
sponding to size-independent fracture energy. Remaining life values have been
predicted and compared with the corresponding experimental values and observed
that they are in good agreement with each other.

Keywords Ultra-high-strength concrete Characterization Fatigue loading

Tension softening Fracture energy Stress intensity factor Crack growth
Remaining life
1 Introduction
Concrete has been one of the most commonly used construction materials in the
world. One of the major problems civil engineers face today is concerned with
preservation, maintenance, and retrofitting of structures. The historical development
A. Ramachandra Murthy (&)

CSIR–Structural Engineering Research Centre, Taramani, Chennai 600113, India
e-mail: murthyarc@serc.res.in
N. R. Iyer
Academy of Scientific & Innovative Research (AcSIR), New Delhi, India
e-mail: nageshiyer@acsir.res.in
B. K. Raghu Prasad
Civil Engineering Department, Indian Institute of Science, Bengaluru, India
e-mail: bkr@civil.iisc.ernet

https://doi.org/10.1007/978-981-10-7197-3_14
158 A. Ramachandra Murthy et al.
of concrete material may be marked and divided into several stages. The first is the
traditional normal strength concrete (NSC) where only four kinds of ingredients,
namely cement, water, fine aggregates, and coarse aggregates are used. With the
increasing development in physical infrastructure, such as high-rise buildings,
long-span bridges, flyovers, power plant structures, and higher compressive
strength, concrete is preferred in most cases. When the compressive strength of
concrete is generally higher than 50 MPa, it is usually defined as high-strength
concrete (HSC). The easiest way to reach high compressive strength is to reduce the
water–cement ratio. However, sometimes the compressive strength is not as
important and necessary as some other properties, such as low penetrability, high
durability, and excellent workability. Thus, high-performance concrete (HPC) was
proposed and widely studied at the end of the last century. Currently, an
ultra-high-strength concrete (UHSC) with axial compressive strength more than
100 MPa and also with a high tensile strength (more than 10% of the compressive
strength) was successfully developed [1–4]. Since UHSC is a relatively new
material, its characterization and fracture behavior are yet to be understood.
Concrete contains numerous flaws, such as holes or air pockets, precracked
aggregates, lack of bond between aggregate and matrix, etc. from which cracks may
originate. Cracks generally propagate in a direction, which is perpendicular to the
maximum tensile stress. The fracture behavior of concrete is greatly influenced by
the fracture process zone (FPZ). The variation of FPZ along the structure thickness
or width is usually neglected. The inelastic fracture response due to the presence of
FPZ may then be taken into account by cohesive pressure acting on the crack faces.
A few experimental investigations on fatigue crack propagation in concrete were
reported [3, 5–8].
In the present study, fracture mechanics-based remaining life has been predicted
for high-strength concrete (HSC) and ultra-high-strength concrete (UHSC) and
compared with the corresponding experimental observations. A “Net K” model has
been proposed by using principle of superposition. Bilinear tension softening
relationship corresponding to size-independent fracture energy has been used to
consider the cohesive stresses acting ahead of the crack tip.
2 Experimental Investigations
Two mixes designated as high-strength concrete (HSC) and Ultra-high-strength

concrete (UHSC) are characterized, and their mix proportions have been arrived at
using appropriate method and several trials. For HSC, the materials are Portland
cement, silica fume, quartz sand, high-range water reducer, water, and steel fibers.
for UHSC, the materials are Portland cement, silica fume, quartz sand, quartz
powder, high range water reducer, water, and steel fibers. The main difference
between HSC and UHSC is the absence of quartz powder in the case of HSC mix.
Several trials had to be tried before a final mix design. The final mix proportions
and ratio obtained are given in Table 1.
Fracture Analysis and Remaining Life Assessment … 159
Table 1 Mix ratio of HSC and UHSC

Mix Cement Fine Coarse Silica Quartz Quartz Steel Water SP
ratio aggregate aggregate fume sand powder fiber (%)
(%)
HSC 1 – – 0.25 1.5 – 2 0.33 2.5
UHSC 1 – – 0.25 1.1 0.4 2 0.23 3.5
2.1 Mechanical Properties
Various mechanical properties such as compressive strength, spilt tensile strength of

HSC and UHSC mix at 28 days are shown in Table 2. Table 2 also shows the
size-independent fracture energy [9].
2.2 Casting of Beams
Details of the specimens prepared for experimental study and notch depth details
are given in Table 3. The experimental setup consists of a MTS 2500 kN capacity
servo-hydraulic UTM with online data acquisition system. Flexural fatigue tests
have been conducted on HSC and UHSC specimens. Three specimens for each
notch depth have been tested under load control. A stress ratio of 0.2 is maintained
for all the specimens. Frequency of loading is varied for HSC and UHSC speci-
mens. At regular intervals, crack depth versus number of cycles has been monitored
for HSC and UHSC specimens. Table 3 shows the maximum load (Pmax), mini-
mum load (Pmin), frequency of loading, and no. of cycles to failure. Pmax has been
arrived at based on the failure load obtained from the static test. Paris crack growth
constants have been evaluated from the experimental data of da=dN and DK: The
values of C and m for various cases are shown in Table 3.
Table 2 Mechnanical properties of HSC and UHSC

Mix Compressive Split tensile Modulus of Fracture
ID strength (MPa) strength (MPa) elasticity (MPa) energy (N/m)
HSC 87.71 15.38 37,890 6393
UHSC 122.52 20.65 42,987 14,184
Table 3 Remaining life predicted by net K model

Beam Notch Max. Min Paris crack Remaining life,
dimensions depth load load growth cycles
(mm) (kN) (kN) constants
log C m Exptl Analytical
250 * 50 * 50 5 8.0 1.6 −17.24 5.31 2112 1876
(UHSC) 10 6.5 1.3 −16.99 5.23 1464 1280
15 4.9 0.9 −16.54 5.13 1172 1035
20 3.3 0.7 −16.35 5.01 820 745
400 * 50 * 80 8 11.3 2.3 −18.78 6.17 5621 5223
(UHSC) 16 8 1.6 −18.52 6.02 4159 3845
24 6.1 1.2 −18.23 5.86 3215 2876
32 4.4 0.8 −17.45 5.32 2432 2230
600 * 50 * 130 13 17.5 3.5 −19.99 6.76 7654 6801
(UHSC) 26 11.4 2.3 −19.34 6.33 4302 3910
39 8.1 1.6 −18.98 5.96 3214 2810
52 5.84 1.2 −18.54 5.69 2483 2231
250 * 50 * 50 5 3.36 0.673 −15.46 5.12 2012 1812
(HSC) 10 2.69 0.54 −14.76 5.34 1617 1503
15 2.27 0.45 −15.21 5.01 1234 1134
20 1.682 0.34 −14.79 5.11 1032 890
500 * 50 * 100 10 6.67 1.335 −17.23 6.01 4356 4089
(HSC) 20 4.081 0.816 −16.89 6.11 3219 2820
30 3.193 0.64 −16.78 5.99 2489 2178
40 2.55 0.51 −16.56 5.67 1876 1694
3 Net K Model Considering Tension Softening
In the Net K model, nonlinear fracture mechanics principles have been used for
crack growth analysis and remaining life prediction. The merits of the model are
two-fold: (i) it is possible to get SIF variation from crack initiation to unstable crack
propagation, and (ii) it is possible to model the fatigue behavior. SIF (K) has been
computed by using the principle of superposition.
To incorporate the tension softening behavior, based on the principle of super-
position, SIF has to be modified as (Fig. 1),
KI ¼ KIP þ KIq ð1Þ
where KI is called net K, and KIq is negative where KIP is SIF for the concentrated
load P on a three-point bend specimen, and KIq is SIF due to the closing force
applied on the effective crack face inside the process zone, which can be obtained
through Green’s function approach by knowing the appropriate softening relation.
(a) (b)
(c) (d) +
Fig. 1 Illustration of superposition principle
3.1 Computation of KIP
SIF due to the concentrated load P can be calculated by using LEFM principles.
A three-point bending beam is shown in Fig. 1. The SIF for the beam can be
expressed as
pffiffiffiffiffiffi a 3PS

KIp ¼ r pa g1 where; r ¼ ð2Þ
b 2b2 t
where P = applied load, a = crack length, b = depth of the beam, t = thickness and
g1(a/b) = geometry factor, which depends on the ratio of span to depth of the beam
and is given below for S/b = 2.5 [10]
a 1:0 2:5a=b þ 4:49ða=bÞ2 3:98ða=bÞ3 þ 1:33ða=bÞ4

g1 ¼ ð3Þ
b ð1 a=bÞ3=2
3.2 Computation of KIq
The incremental SIF due to the closing force dq can be written as [11].
2 a x
dKIq ¼ pffiffiffiffiffiffiffiffiffi dq g ; ð4Þ
pDa D a
where dq can be expressed as a function of softening stress distribution over the

crack length Da, the function ‘g’ represents the geometry factor.
3.3 Calculation of ‘dq’
By using the above concept (Fig. 1d), cohesive crack can be modeled in the fol-
lowing manner (Fig. 2)
The crack opening displacement w at any point x is assumed to follow linear
relationship (Fig. 2) and can be expressed as,
a x
0
w¼d þ1 a0 x aeff ð5Þ
Da
where d is the crack tip opening displacement, and a0 is initial crack length.
As an example, let us consider linear softening law
r ¼ ft ð1 w=wc Þ ð6Þ
where ft = tensile strength of concrete and wc = critical crack opening

displacement.
Substituting for w from Eq. (5) in the linear softening law given by Eq. (6), one
can obtain,

d a0 x
dq ¼ r ¼ ft 1 þ1 ð7Þ
wc Da
The crack opening displacement at any point d(x) can be calculated using the
following equation
a x
dðxÞ ¼ CMOD g3 ; where
b a
a x x 2

a x x 2 1=2
ð8Þ
g3 ; ¼ 1 þ 1:081 1:149
b a a b a a
Fig. 2 Cohesive crack CTOD

modeling-schematic diagram
of crack opening
displacement
Δa
x
ao
CMOD
where CMOD is crack mouth opening displacement and is calculated using the
following formula.
4ra a
CMOD ¼ g2 ð9Þ
E b
where g2(a/b) is geometric factor, depends on the ratio of span to depth of the beam
and is given below for S = 2.5b
1:73 8:56a=b þ 31:2ða=bÞ2 46:3ða=bÞ3 þ 25:1ða=bÞ4

g2 ða=bÞ ¼ ð10Þ
ð1 a=bÞ3=2
Hence, replacing dq in Eq. (4) and integrating over length Da, KIq can be
obtained as,
Zaeff a x
2ft d a0 x
KIq ¼ pffiffiffiffiffiffiffiffiffi 1 þ 1 g ; dx ð11Þ
pDa wc Da b a
a0
where
a x 4:35 5:28x=a
3:52ð1 x=aÞ
g ; ¼
ð1 a=bÞ1=2
3=2
b a ð1 a=bÞ
" # ð12Þ
1:30 0:30ðx=aÞ3=2 x h x ai
þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2 þ 0:83 1:76 1 1
1 ðx=aÞ a a b
Similar expressions can be obtained for other models such as bilinear, trilinear,
exponential, and power law.
3.4 Remaining Life Prediction
Analysis of fatigue crack growth and remaining life prediction requires data
regarding loading conditions, type of material and crack geometry, among others.
Then, a suitable crack growth law must be selected. One general expression for
such a law is
da
¼ f ðDK; R; . . .Þ ð13Þ
dN
The number of loading cycles required to extend the crack from an initial length
a0 to the final critical crack length af is given by
Zaf
da
N¼ ð14Þ
f ðDK; R; . . .Þ
a0
Using the Paris law, the number of cycles to failure can be obtained
da
¼ CðDKÞm ð15Þ
dN
Here, DK can be computed by using the following expression
DK ¼ Kmax Kmin ; where Kmax ¼ K p K q :
4 Numerical Studies and Discussion of Results
Crack growth studies and remaining life prediction have been carried out for
three-point bending UHSC and HSC beams by using the above methodologies.
Bilinear tension softening model developed by using inverse analysis has been used
to account for tension softening. Bilinear tension softening model obtained from
inverse analysis corresponding to size-independent fracture energy has been
employed for crack growth analysis and remaining life prediction (Fig. 3) [12].
Paris crack growth constants (C, m), maximum load, minimum load, and predicted
remaining life are shown in Table 3. From Table, it can be observed that the
predicted remaining life is in good agreement with the corresponding experimental
observations. A significant advantage of net K model is that it is possible to obtain
DK as an increment in the value of K at every step.
(a) HSC (b) UHSC
Fig. 3 Bilinear stress-crack opening relationship for HSC and UHSC corresponding to their GF
5 Summary
A “Net K” model has been proposed by using nonlinear fracture mechanics prin-
ciples for crack growth analysis and remaining life prediction. SIF (K) has been
computed by using the principle of superposition. Bilinear tension softening rela-
tionship obtained from the inverse analysis has been used to consider the cohesive
the stresses acting ahead of the crack tip. Numerical studies have been conducted on
three-point bending concrete structural component under constant amplitude
loading. The predicted remaining life values have been compared with the corre-
sponding experimental observations and it is noted that they are in good agreement
with each other.
Acknowledgements We acknowledge with thanks the valuable technical suggestions and support
provided by our colleagues of Computational Structural Mechanics Group, CSIR-SERC.
References
1. P. Richard, M.H. Cheyrezy, Reactive powder concretes with high ductility and 200–800 MPa
compressive strength, ACI SP144. 24 (1994) 507–518
2. P. Richard, M. Cheyrezy, Composition of reactive powder concretes. Cem. Concr. Res. 25,
1501–1511 (1995)
3. A.N. Mingzhe, Y.U. Ziruo, M. Sun, S. Zheng, L. Liang, Fatigue properties of RPC under
cyclic loads of single-stage and multi-level amplitude. J. Wuhan Univ. Technol. Mat. Sci. Ed.
25, 167–173 (2010)
4. P. Goltermann, V. Johansen, L. Palbol, Packing of aggregates: an alternate tool to determine
the optimal aggregate mix. ACI Mat. J. 94, 435–443 (1997)
5. A.R. Ingraffea, Discrete fracture propagation in rock: laboratory tests and finite element
analysis, Ph.D. Dissertation, (University of Colorado, Boulder, 1977)
6. M.V.K.V. Prasad, C.S. Krishnamoorthy, Computational model for discrete crack growth in
plain and reinforced concrete. Comput. Methods Appl. Mech. Engg. 191, 2699–2725 (2002)
7. Z. Wu, S. Yang, X. Hu, J. Zheng, An analytical model to predict the effective fracture
toughness of concrete for three-point bending notched beams. Engg. Fract. Mech. 73,
2166–2191 (2006)
8. V. Slowik, B. Beate Villmann, N. Bretschneider, T. Villmann, Computational aspects of
inverse analyses for determining softening curves of concrete. Comput. Methods Appl. Mech.
Engg. 195, 7223–7236 (2006)
9. A.R. Murthy, B.L. Karihaloo, N.R. Iyer, B.K. Raghu Prasad, Determination of size—
independent specific fracture energy of concrete mixes by two methods. Cem. Concr. Res. 50,
19–25 (2013)
10. H. Tada, P.C. Paris, G.R. Irwin, The Stress Analysis of Cracks Handbook, 2nd edn. (St. Louis.
MO, Paris Productions, 1985)
11. S.P. Shah, S.E. Swartz, Fracture Mechanics of Concrete: Applications of Fracture Mechanics
to Concrete, Rock and other Quasi-brittle Material (Wiley, New York, 1995)
12. A.R. Murthy, B.L. Karihaloo, N.R. Iyer, B.K. Raghu Prasad, B.K. Raghu Prasad, Bilinear
tension diagrams of concrete mixes corresponding to their size-independent specific fracture
energy. Constr. Build. Mater. 47, 1160–1166 (2013)
Part III
Composites
Rupture in Abdominal Aortic Aneurysm
Qusay Alfaori, Ashok Saxena, Hanna Jensen and Morten Jensen
Abstract Ability to predict the growth of abdominal aortic aneurysm and the
rupture event is important for prognostics and treatment for this life-threatening
cardiovascular disease. Uniaxial stress–strain testing of healthy and collagen-
degraded aortic specimens from pigs was performed in order to study the biome-
chanical properties of the aortic tissue. A mathematical model for the stress–strain
relationship suitable for describing the behaviour of abdominal aortic tissue was
used to derive tissue properties as a function of specimen orientation and collagen
degradation. The resultant properties varied significantly between healthy tissue and
collagen-degraded tissue and were found to be orientation dependent. Maximum
stress, maximum strain and critical strain values were significantly higher for healthy
tissue group than for collagen-degraded tissue group. Finite element modelling of
healthy and collagen-degraded abdominal aortas was performed using the measured
properties. It is shown that this approach can potentially lead to the development of
an analytical tool for assisting in prognostics and treatment of the disease.
Keywords Aneurysm Aorta Collagen degradation Rupture

Nomenclature
r Stress
e Strain
A = ec Critical strain for rupture
K Inverse of the elastic modulus in Region 3 of the stress–strain behaviour
B Regression constant characteristic of the stress–strain behaviour in Region
1
emax Maximum strain sustained prior to rupture
ru Ultimate strength or maximum stress at rupture
Q. Alfaori A. Saxena (&) H. Jensen M. Jensen

Department of Biomedical Engineering, University of Arkansas, Fayetteville,
AR 72701, USA
e-mail: asaxena@uark.edu

https://doi.org/10.1007/978-981-10-7197-3_15
170 Q. Alfaori et al.
1 Introduction
In spite of the advances in medical sciences, cardiovascular diseases remain the

leading cause of death [1, 2]. Abdominal aortic aneurysm (AAA) is an irregular
expansion of the abdominal aorta due to the weakening of abdominal aortic wall
and is considered a form of cardiovascular disease. A patient is said to have AAA
when the abdominal aortic diameter increases by 50% of the healthy original
diameter [3]. Every year, 500,000 new patients are diagnosed with AAA throughout
the world, 150,000 of those in the USA alone. AAA is present in about 2% of the
elderly (60 years and older), a population group that is rapidly increasing in the
developed countries [4, 5]. Abdominal aortic aneurysms are one of the most lethal
aneurysms that can develop in the human body, with a mortality rate approaching
90% in the case of rupture [1]. Every year in the USA, a country for which accurate
numbers are available, approximately 15,000–20,000 people die from AAA, and it
is listed as the 13th most common cause of death and the 10th among men that are
55 years of age and older [4].
Aneurysm development is a complex process related to variability of molecular
mediators and matrix-degrading proteases often related to ageing [6]. The process
causes the connective tissue of the aorta to degrade, caused by destruction of
important structural proteins such as elastin and collagen [6]. AAA is diagnosed
using expensive medical imaging techniques such as ultrasonography, computed
tomography (CT) scanning or MRI scanning that are not performed routinely
during physical examinations. AAA is often being asymptomatic, making it difficult
for the patients to seek critical, timely help and treatment.
Once AAA is detected, assessment of the rupture risk is important. The most
common method for rupture risk assessment is measurement of the diameter of the
dilated aorta using CT scan. If the aneurysm exceeds 1.5 times the healthy aorta
diameter (20–30 mm), the risk of rupture is considered high at a mortality rate of
95% [3]. While not all aneurysms with diameters greater than 1.5 times the healthy
diameter rupture, some aneurysms do rupture at diameters less than 1.5 times the
original diameter, and thus, regular monitoring is highly recommended in order to
assess growth of the aneurysm and to treat it in time to avoid catastrophic rupture.
Thus, a prognostic analytical tool to assist in determining the frequency of such
examinations will be very useful.
Over the past 15 years, other factors that have been identified as relevant to
rupture risk assessment are wall stress distribution, wall strength and formation of
intraluminal thrombus (ILT) on AAA wall [1, 7, 8]. Engineering tools such as finite
element analysis (FEA) are now frequently used to model hypothetical and
patient-specific aneurysms. Based on work done using FEA, factors such as
expansion rate, degree of symmetry and geometrical parameters of AAA can be
used as additional parameters for assessing potential risk of rupture [1].
This study was designed to address the following questions: (a) How does
collagen degradation contribute to the development of AAA and (b) what is the
effect of collagen degradation on the tissue’s elastic modulus and other tensile
Rupture in Abdominal Aortic Aneurysm 171
properties? It is likely that abnormal collagen degradation overtime accelerates the

development of the aneurysm. Thus, quantification of the growth kinetics and
rupture criteria involving collagen degradation is important.
2 Experimental Methods
An Instron 5944 (Instron, Norwood, MA, USA) universal testing machine was used
for the tensile tests conducted in this study. The set-up included a Bioplus bath in
which testing could be performed under environmental conditions that resemble the
in vivo environment of soft tissue, as shown in Fig. 1 along with a picture of a
ruptured specimen.
A total of 50 abdominal aortic specimens were obtained from pigs (5 months old
weighing 80 kg). Certain sections of the abdominal aorta were cut along the sagittal
plane to obtain a longitudinal orientation, and some sections were cut along the
transverse plane to obtain a circumferential orientation. Each specimen had a length of
5–7 cm, a width of 2–2.5 cm and a thickness of approximately 1.5–2 mm. Specimens
were frozen at a temperature of −80 °C to protect them from degradation, and prior to
mechanical testing and to eliminate most artefacts in measured mechanical properties,
they were removed from the freezer to thaw at room temperature for 24 h. Next, the
specimens were mounted on the stress–strain apparatus and were immersed in the
Bioplus bath, containing a phosphate-buffered saline solution (1 PBS), at 37 °C
for 30 min before performing the mechanical tests. The abdominal aortic specimens
marked as Type-I collagen degraded were immersed in Type-I collagenase solution
for 24 h prior to mechanical testing. The collagenase solution had a concentration of
0.30 g/mL and was made via mixing of Type-I collagenase powder with 1 HBSS
(Hank’s Balanced Salt Solution). Uniaxial stress–strain tests on healthy abdominal
aorta specimens from longitudinal and circumferential orientations and Type-I
Fig. 1 Instron testing system

equipped with a Bioplus bath
and a picture of a tested
specimen
collagen-degraded specimens from the same orientations were then performed. All
specimens from both groups were preconditioned by applying five partial loading and
unloading cycles in the elastic regime.
3 Experimental Results and Discussion
A three-phase mathematical model from Raghavan et al. was [9] adapted to rep-
resent the stress–strain relationship for the aortic material, Eqs. (1) and (2),
respectively, below.
1
rjRegion 1 ¼ e ð1Þ
K þ AB

1 A
rjRegion 3 ¼ e ð2Þ
K K
where K, A and B are model constants derived from the experimental stress–strain
data. Figure 2a and b show a schematic representation of the stress–strain
Fig. 2 Schematic representation of a changing stiffness the aortic material with strain due to
activation of collagen fibres and b the three regions of stress–strain behaviour of the aortic material
and parameters representing various properties [9]
relationships as described by Eqs. (1) and (2) and the physical reasoning behind the
three regions. In Region 1, the stress is entirely borne by the matrix material
consisting of elastin. Region 2 represents the onset of collagen fibre participation in
bearing the stress and occurs sequentially as shown in Fig. 2a for three fibres, as an
example. Region 2 continues until all fibres participate in bearing the stress leading
to Region 3 in which complete recruitment of all collagen fibres occurs leading to
greatly increased stiffness. The constant A in the above equations is the intercept of
the extrapolated stress–strain trend line in Region 3 with the strain axis at zero
stress. This also represents the residual strain in a ruptured specimen that can be
directly measured on the tested sample and is therefore called ec . Equation (2) can
be rewritten as Eq. (3). Similarly, the constant K can be derived from the slope of
the Region 3 stress–strain behaviour. The constant K represents inverse of the
stiffness of the sample. The constant B can be derived from regression of the Region
1 data using Eq. (1) and does not have a physical meaning beyond a regression
constant.

1 ec
rregion 3 ¼ e ð3Þ
K K
While Eqs. (1)–(3) are important for understanding the physical deformation
processes in the aortic material, they do not provide a single mathematical rela-
tionship between stress and strain for use in analyses. For example, no equation is
available to describe the behaviour in Region 2. An exponential relationship of the
type in Eq. (4) can accurately represent the relationships in all three regions and can
be used to reliably extract the constants A, B and K for each of the material
conditions.
r ¼ aebe ð4Þ
where a and b are regression constants whose values were determined from the
results of each test and then used to extract the value of K, A and B as listed in
Table 1 along with other properties such as the maximum stress and maximum
strain sustained by the aorta samples. The typical engineering stress–strain diagrams
are shown in Fig. 3 for both healthy and collagen-degraded aorta samples in the
longitudinal and circumferential directions. The shape of the stress–strain curves for
healthy aorta is very similar to the schematic shown in Fig. 2, but they very dif-
ferent for the collagen-degraded samples. The properties in the longitudinal and
circumferential directions in the healthy aorta are significantly different from each
other. Collagen degradation results in severe degradation of strength and stiffness,
and it appears that collagen fibres no longer contribute to the mechanical strength of
the aorta wall in either longitudinal or circumferential direction.
Figure 4 shows some sample results from the various stress–strain diagrams and
the corresponding fits using Eq. (4). The exponential form represents the entire data
very well.
174
Table 1 Material parameters of the aortic material for healthy and collagen-degraded conditions in longitudinal and circumferential orientation
Specimen group No of duplicate B (MPa) A = ec emax K (MPa)−1 Ultimate strength, ru
samples (MPa)
Healthy, longitudinal 12 0.46 ± 0.01 0.45 ± 0.06 0.94 ± 0.11 0.15 ± 0.01 2.34 ± 0.09
Healthy, circumferential 12 0.35 ± 0.02 0.73 ± 0.08 1.19 ± 0.12 0.22 ± 0.01 2.19 ± 0.07
Degraded longitudinal 6 22.2 ± 0.4 0.02 ± 0.0 0.68 ± 0.15 9.09 ± 0.14 0.09 ± 0.00
Degraded 6 10.5 ± 0.5 0.03 ± 0.0 0.83 ± 0.10 7.14 ± 0.11 0.076 ± 0.00
circumferential
Raghavan et al. values B (N/cm2) A K (10(cm2/
N)
AAA (longitudinal) 45 3.90 ± 0.6 0.091 ± 0.009 39 ± 5
AAA (circumferential) 16 4.47 ± 0.9 0.103 ± 0.02 25 ± 4
Q. Alfaori et al.
Fig. 3 Typical plot of stress and strain for healthy and collagen-degraded aorta samples in the
longitudinal and circumferential orientations. The samples’ values of ultimate strength stress and
strain at rupture are shown in Table 1
Fig. 4 Exponential fit of stress–strain data of longitudinal (a) and circumferential (b) healthy
specimens and linear fit of stress–strain data of longitudinal (c) and circumferential (d) colla-
gen-degraded specimens
The data in Table 1 and Fig. 3 strongly suggest that one of the causes of AAA
could be the degradation of collagen in regions of the aorta wall that causes weakening
of the aorta in those regions. The values of A in Table 1 are somewhat higher than for
the values reported for humans between age 31 and 58 by Raghavan [9, 10] but
comparable. Thus, samples from pigs that are much more readily available from
slaughter houses than human tissue are suitable model materials for human aortas.
If the collagen degradation occurs axisymmetrically over a length of the aorta,
the bulge will also be symmetric about the axis. On the other hand, if the collagen
degradation occurs only on one side of the aorta wall, the bulge will be in that
region only. Both types of aneurysms are lethal and whether rupture occurs or not
will depend on the strain in the aneurysmal region and how close it is to the critical
residual strain of ec , although other factors may also influence when rupture is
expected to occur. This will be discussed further later in the paper. It is important to
remember that collagen is not the only protein responsible for all macroscopic
properties of the tissue. In fact, the cohesiveness among elastin, smooth muscle and
collagen provides the tissue’s ultimate structural integrity. Thus, AAA development
is complex and can be related to elastin degradation, collagen fibre cross-linking
and smooth muscle deterioration.
Another aspect of the deformation behaviour of aorta materials investigated was
the sensitivity of the deformation properties to applied strain rate. The data in Fig. 3
were obtained from tests conducted at a strain rate of 0.1 min−1. Tests were con-
ducted at strain rates of 1, 0.01 and 0.001 min−1, and it was found that the stress–
strain behaviour was sensitive to strain rates indicating viscoelastic and/or vis-
coelastic tendencies. Thus, the long-term goal of this project includes building those
capabilities in the stress–strain relationships. In this paper, only the basic rela-
tionships are used to model the aneurysm in aortas.
4 Finite Element Modelling
Finite element modelling was carried out using the commercially available
ABAQUS code to explore how degradation in the wall material properties affects
the development of the aneurysm. The aneurysm model included the deformation of
healthy and aneurysmal aortas using the deformation constants obtained from the
tested specimens. Specifically, the elastic modulus value of the healthy (in Region
3) and collagen-degraded specimens was used. Figure 5 shows the model used for
the healthy and degraded aortas. In the latter, a region over which the degraded
properties were used is shown in a different colour. As mentioned previously, the
healthy aortic tissue properties typically display nonlinear, anisotropic, hyperelastic
or viscoelastic behaviour. In this study, the analyses were conducted with simplified
assumptions of isotropic, nonlinear behaviour including the characteristics of
hyperelasticity using the measured value of elastic modulus (1/K) of 5.4 MPa for
the healthy tissue (Region 3) and a value of 0.13 MPa for the collagen-degraded
Fig. 5 Finite element model of a healthy and b aneurysmal aorta
tissue. The Poisson’s ratio of 0.45 and arterial wall density of 1.2 g/cm3 taken from
the literature [11].
The structural geometry of a healthy abdominal aorta resembles that of a pipe
with a circular cross section of outer radius R and inner radius r, wall thickness
w = R − r and Length L. The geometry of a healthy abdominal aorta was con-
structed with dimensions that are similar to the actual dimensions of the abdominal
aorta. The outer radius of the abdominal aorta is approximately 1.5 cm, the wall
thickness is approximately 2 mm, and the length ranges from 10 to 15 cm.
A refined mesh was applied to the geometry which created at least 20,000 solid
elements.
The boundary conditions at the edge or end of the pipe may affect the results and
may cause irregular deformation behaviour in the model near the ends. Thus, the
actual length of the healthy and aneurysmal abdominal aortic models was extended
by 2 cm on each end so the results were not affected by the end effects. Further, the
elastic modulus value of collagen-degraded aortic specimen was applied to an area
to simulate the aneurysmal region in the aneurysmal aortic model. A cyclically
fluctuating uniform internal pressure mimicking physiological conditions was
applied to the interior wall along the entire length of the aortic models. The applied
cyclic pressure is a sinusoidal wave function with a maximum amplitude of
approximately 16 kPa (120 mm Hg), minimum amplitude of approximately 11 kPa
(80 mm Hg) and a frequency of 60/72 s−1 (heart’s 72 beats per minute).
A cyclically varying pressure of constant amplitude with a sinusoidal waveform
was applied to the internal surface of the healthy aortic model, as mentioned earlier.
This was applied over a time of 10 s, and the resulting von Mises stress distribu-
tions for healthy and collagen-degraded models were calculated by applying the
load and are shown in Fig. 6.
A typical element in the finite element model was chosen from the healthy aorta
model and also from the collagen-degraded region of the aneurysmal aorta model to
illustrate the von Mises stress and the circumferential strain magnitudes as a
function of time. These results are shown in Fig. 7a and b for healthy aorta@@ and
c, d for the aneurysmal aorta. In the healthy aorta, Fig. 7a, b, the peak stress is
approximately 150 kPa, and the peak strain is approximately 5.5 10−5 which is
much below the levels required for rupture. Obviously, this can change with age as
the properties of the aortic wall degrade from collagen degradation and also the wall
Fig. 6 Deformation of a a healthy aorta during systole and b a collagen-degraded aorta during a
diastole showing the bulge in the aneurysmal region
thickness itself may decrease. By comparison, the peak stress in the

collagen-degraded region is 62 kPa and the strain 0.01 that is dangerously high!
With repeated cycling, the strains will accumulate each cycle making the aneurysm
worse and increasing the risk of rupture. This supports the notion put forward by
others that as collagen degrades, elastin is the only tissue component available to
support the stress [9, 10].
Using the properties from mechanical testing in the finite element models in the
modelling software clearly demonstrates that the risk of rupture will be much higher
for collagen-degraded tissue. From a biological point of view, collagen degradation
occurs due to the action of collagenases, which are class of proteases enzymes.
Excessive collagenase activity causes weakening and destruction of the aortic wall.
In fact, biomechanical studies have shown that collagen presents in media and
adventitia layers of the aortic wall is responsible for maintaining tissue
microstructural stability [12, 13].
The kinetics of the aneurysm growth have not been modelled in this study
because a different constitutive relationship that incorporates the effects of degra-
dation with time will be needed to model that trend. This will be the focus of future
studies. Relying completely on uniaxial test data is also not optimal, and studying
the stress–strain behaviour of aortic tissue under biaxial stress is needed to more
accurately model the state of stress in the aorta. Conducting biaxial testing on aortic
specimens will help validate the accuracy of critical strain values that were cal-
culated in this study and to establish rupture criterion that can improve rupture
prognostics for AAA patients. In addition, due to the fact that the elastic modulus in
the longitudinal orientation differs from that in the circumferential orientation,
orthotropic material property assumption is more appropriate in our finite element
model than the isotropic material properties assumed in the calculations thus far.
Also, biaxial testing will allow us to derive more tissue property constants that can
be utilized in applying FEA on a viscoelastic model under the conditions of
dynamic loading.
Fig. 7 a Circumferential strain b von Mises stress for a healthy aorta and c, d are the same,
respectively, for collagen-degraded aortic material properties
5 Conclusions and Future Work
It has been shown that significant changes in the biomechanical properties of the
aortic tissue result in the development of aneurysms. The results indicate that
changes in the collagen microstructure of the tissue due to degradation greatly affect
the critical strain and the maximum stress that the tissue can sustain. The critical
strain to rupture and the changes that occur in its value from collagen degradation
are likely will be good indicators of impending rupture in AAA patients. Also, the
changes in critical strain to rupture under biaxial loading conditions are likely to be
relevant but have not been investigated in this study. To our knowledge, this study
is the first to examine the biomechanical behaviour of the aortic tissue after collagen
degradation but lot more work must be performed to fully understand the degra-
dation kinetics and the variables that determine its progression. More accurate data
through testing of abdominal aortic tissue can improve the accuracy of the models.
Acknowledgements The animal aortic specimens were obtained from Cockrums Meat
Processing and Taxidermy, Rudy, Arkansas, USA. Dr. Kartik Balachandran is thanked for his help
in providing valuable insight and allowing us to use some of his laboratory space and materials.
References
1. B.J. Doyle, 3D Imaging of Abdominal Aortic Aneurysms Techniques and Applications (Nova
Science, Hauppauge, 2010)
2. The Top 10 Causes of Death, Retrieved 26 Oct 2015, from http://www.who.int/mediacentre/
factsheets/fs310/en/ (2012)
3. J.D. Humphrey, C.A. Taylor, Intracranial and abdominal aortic aneurysms: similarities,
differences, and need for a new class of computational models. Annu. Rev. Biomed. Eng. 10
(1), 221–246 (2008)
4. D.A. Vorp, J.P. Vande Geest, Biomechanical determinants of abdominal aortic aneurysm
rupture. Arterioscler. Thromb. Vasc. Biol. 25(8), 1558–1566 (2005)
5. Projections of the Population by Selected Age Groups and Sex for the United States: Table 2.
2015 to 2060 (NP2012-T2), December 2012. Source: U.S. Census Bureau, Population
Division at http://www.census.gov/population/projections/files/summary/NP2012-T2.xls
6. R.W. Thompson, J.A. Curci, T.L. Ennis, D. Mao, M.B. Pagano, C.N. Pham, Pathophysiology
of abdominal aortic aneurysms: insights from the elastase-induced model in mice with
different genetic backgrounds. Ann. NY. Acad. Sci. 1085, 59–73 (2006)
7. J. Biasetti, P.G. Spazzini, J. Swedenborg, G.T. Christian Gasser, An integrated fluid-chemical
model toward modelling the formation of intra-luminal thrombus in abdominal aortic
aneurysms. Front. Physiol. 3, 1–16 (2012)
8. W.R. Mower, W.J. Quiñones, S.S. Gambhir, Effect of intraluminal thrombus on abdominal
aortic aneurysm wall stress. J. Vasc. Surg. 26(4), 602–608 (1997)
9. M.L. Raghavan, M.W. Webster, D.A. Vorp, Ex vivo biomechanical behaviour of abdominal
aortic aneurysm: assessment using a new mathematical model. Ann. Biomed. Eng. 24(5),
573–582 (1996)
10. M.L. Raghavan, Cardiovascular bio-solid mechanics section, in Biomechanics 51:050,
pp. 3–25
11. D.A. Vorp, Biomechanics of abdominal aortic aneurysm. J. Biomech. 40(9), 1887–1902
(2007)
12. H. Abdul-Hussien, R.V. Soekhoe, E. Weber, J.H. von der Thüsen, A. Mulder, J.N. Lindeman,
Collagen degradation in the abdominal aneurysm: a conspiracy of matrix metalloproteinase
and cysteine collagenases. Am. J. Pathol. 170(3), 809–817 (2007)
13. E. Choke, G. Cockerill, W.W. Wilson, S. Sayed, J. Dawson, I. Loftus, M.M. Thompson, A
review of biological factors implicated in abdominal aortic aneurysm rupture. Eur. J. Vasc.
Endovasc. Surg. Off. J. Eur. Soc. Vasc. Surg. 30(3), 227–244 (2005)
Damage Tolerance Assessment
of Laminated Composite Plates
Subjected to Ballistic Impact
H. L. Vinayaka and Shivashankar R. Srivatsa
Abstract Polymer matrix composite materials in laminated plate configurations are

extensively used in a wide range of fields like defense, navy, aerospace, and
automotive industries due to their low weight, high-specific strength, and tai-
lorability. However, in these applications, impact damage tolerance of laminated
composite structures is a constant source of concern over the years. Ballistic impact
damage tolerance assessment is a concern in many fields, besides obvious appli-
cation of protective armor where the primary design objective is that of preventing
projectile penetration. Soft body armors made of composites are subjected to bal-
listic impact from projectiles where the structure has no time to respond and could
result in localized damage. Analysis of composite structures under ballistic impact,
which is a transient nonlinear dynamic phenomenon, requires nonlinear finite ele-
ment analysis (NLFEA). LS-DYNA, a general purpose code based on finite element
method with a wide range of material models, which uses explicit time integration
technique for analyzing large deformation dynamic response of solids and struc-
tures, has proven to be relevant in this respect. This paper focuses on finite element
modeling of impact on composite plates subjected to ballistic impact by a fragment
simulating projectile (FSP) using LS-DYNA software. Finite element modeling is
defined here as analyst’s choice of material models (constitutive equations and
failure criteria), nonlinear finite elements, meshes, constraints, boundary conditions,
analysis procedures, governing matrix equations, and their solution methods,
specific pre- and post-processing options available in a chosen commercial FEA
software for the intended analysis of candidate components and structures. The
focus of this work is on the use of LS-DYNA software and prediction by analysis
of: (1) Impact damage resistance (contact force-time history); (2) Impact response
(displacement-time history, ply-by-ply strain history, ply-by-ply stresses in the
H. L. Vinayaka (&)
Department of Mechanical Engineering, Dayananda Sagar College of Engineering,
e-mail: yoganarasimha123@gmail.com
S. R. Srivatsa
Department of Mechanical Engineering, BMS College of Engineering,

https://doi.org/10.1007/978-981-10-7197-3_16
184 H. L. Vinayaka and S. R. Srivatsa
material coordinates, and contour plots of failure index); (3) Impact damage
propagation (ply-by-ply failure from first ply failure to last ply failure and
delamination between plies); and (4) Impact damage tolerance (stiffness, strength,
life without and with damage). Significant results like energy balance, contact
force-time history, displacement-time plots, and damage zone shape and size are
presented. Also, a study of the failure modes and effect of failure theories on
ballistic impact response of a composite structure is presented.
Keywords FEM Nonlinear finite element analysis Composite structure

Ballistic impact
1 Introduction
Composite materials offer many advantages compared to conventional materials,

especially where high strength and stiffness to weight ratio is a concern. These
properties make the composites the candidate materials in many applications such
as aerospace, sports equipment, pressure vessels, and automotive parts [1].
However, structures made of composite material are susceptible to damage from
impact loading as a result of their low toughness. Impact damage must be con-
sidered in the design of composite structures considering the potential impacts
during fabrication, operation, and repair.
Typical damage occurring in impacted laminates consists of a combination of
intralaminar damage mechanisms (such as matrix cracking or plasticity, fiber/matrix
debonding and fiber fracture) and interlaminar failure, which develops at the
interface between adjacent plies in the form of debonding between layers (delam-
ination). Interface delaminations are probably the most critical and insidious failure
mechanisms, since they may severely degrade the strength and the integrity of the
structure, and may propagate undetected during service, leading to the unexpected
collapse of the component [2].
Ballistic impact is generally a low-mass, high-velocity impact caused by a
propelling source or a projectile. The velocity of the projectile varies from 300 to
2500 m/s and energy up to 20 kJ. The structure will have no time to react and may
involve complete penetration of the target.
Research on the impact response of various materials including composites is
available in the literature with varying ply orientations, projectile shapes, projectile
velocities, etc. Literature available in the field of impact varies from the use of
conventional metals and their combinations with composite materials for different
applications [3, 4]. Impact experiments on composites need to be conducted with
varying velocities, and associated significant failure modes can be identified [3, 5].
Analytical formulations are available for characterization of impact responses for
different materials [6].
Of the methods available for evaluating the damage induced during ballistic
impact (analytical, experimental, and numerical methods), numerical method is
Damage Tolerance Assessment of Laminated Composite Plates … 185
cost-effective as well as time-saving as the other two methods are tedious and
expensive in nature. Recent research includes numerical simulation as a tool for
validation of experimental results and prediction of impact damage for varying
conditions [1–4, 7–10]. A large number of commercial software are available for
the numerical simulation of impact phenomenon.
The main objective of this work is to develop a finite element model for the
analysis of impact of an impactor on a target for evaluating the damage induced
during ballistic impact, using two classical lamina failure theories, namely
Chang-Chang and Tsai-Wu theory using a chosen commercially available finite
element analysis (FEA) software LS-DYNA. The results include prediction of
ballistic limit, capture of different response plots of the structure such as force-time
plots, energy balance plots, kinetic energy plots, and also damage progression
contour plots for individual plies as well as for the laminate using the failure
theories available in the software. However, the interlaminar failure mode (de-
lamination) is not discussed here.
2 Finite Element Model Development
A laminated composite plate with dimensions 100 mm 100 mm 5.25 mm is

considered. The laminate is E-glass fiber reinforced polymer matrix composite with
plain weave architecture having 8 laminae with (0/90)4 configuration with ply
thickness of 0.656 mm each. The impactor is a 0.50 caliber (12.7 mm diameter and
13.65 mm length) flat-nosed cylinder made of alloyed tool steel with a mass of
13.4 g ± 0.03 g consistent with the mass of a NATO 0.50 caliber fragment sim-
ulating projectile (FSP), whose velocity is varied from 172 to 187.1 m/s. The
material properties of the composite plate and the projectile are given in Tables 1
and 2, respectively.
Plate geometry is discretized using 4-nodded quadrilateral shell elements with a
mesh of 25 25 based on convergence of impact damage tolerance (force-time
history). Use of a graded mesh is avoided because of the dispersion of stress waves.
Use of solid elements to discretize a multilayered composite laminates subjected to
ballistic impact is never used in practice. The projectile is discretized using
8-nodded hexahedral solid elements with a total of 18,400 elements. The finite
element model of the plate and projectile is shown in Fig. 1.
Material type *MAT_054/055—ENHANCED_COMPOSITE_DAMAGE [11, 12]
is used to model the composite plate, and *MAT_03—PLASTIC_KINEMATIC
[11, 12] is used to model the projectile. All the four edges of the plate are fixed with
automatic surface-to-surface contact between the impactor and the plate. The analysis is
carried out for a time period of 0.3 ms with a time step of 1.44 10−6 ms.
Table 1 Material properties Material property Value

of composite plate
(polypropylene E-glass Density, q 1850 kg/m3
composite) [4] Young’s modulus, E11, E22 14 GPa
Young’s modulus, E33 5.3 GPa
Poisson’s ratio, t31 0.14
Poisson’s ratio, t32 0.15
Tensile strength, XT 0.45 GPa
Tensile strength, YT 0.45 GPa
Tensile strength, ZT 0.15 GPa
Compressive strength, XC 0.25 GPa
Compressive strength, YC 0.25 GPa
Matrix mode shear strength, S12 0.032 GPa
Shear modulus, G21 1.8 GPa
Table 2 Properties of Material property Value

projectile [4]
Density, q 7800 kg/m3
Poisson’s ratio 0.28
Yield strength 1.08 GPa
Young’s modulus 210 GPa
Fig. 1 Finite element model of plate and projectile

3 Results Presentation and Discussion
3.1 Ballistic Limit Prediction
For a given target-projectile combination, the ballistic limit is defined as the lowest
initial velocity of the projectile that will result in complete penetration. At that
impact velocity, the residual velocity of the projectile is zero. Recognizing that a
certain amount of variability is always present, the ballistic limit is often defined as
the velocity that will result in penetration of 50% of the samples when a large
number of tests are performed. It is necessary to determine the ballistic limit and to
predict the residual velocity of the projectile when the initial velocity exceeds the
ballistic limit.
Here, the velocity of the projectile was initially considered at 172 m/s and
increased gradually to arrive at the ballistic limit. Table 3 provides numerically
predicted residual velocities for various given initial velocities, based on the two
failure criteria. (Chang-Chang failure criterion identifies modes of ply-by-ply failure
in addition to the failure of the ply as provided by Tsai-Wu criterion, and the two
criteria is distinct and hence their predictions.)
From Table 3, it is observed that the ballistic limit velocity for an 8-ply laminate
based on Chang-Chang failure criteria is 175 m/s with the projectile residual
velocity of 0.856 m/s.
Significant results of impact damage resistance, impact response, and impact
damage propagation are captured in the form of various response plots like energy
balance plots, displacement-time plots, and contact force-time plots for the con-
sidered failure criteria along with a study of failure modes, and the effect of failure
theories are presented as follows. (It is to be noted that the predictions of
Chang-Chang and Tsai-Wu failure criteria is distinct and cannot be combined.
Chang-Chang criterion was introduced to identify the modes of failure which is not
provided by Tsai-Wu criteria.)
Table 3 Simulation results Initial velocity (m/s) Residual velocity (m/s)

for 8-layer PP/E-glass
Chang-Chang Tsai-Wu
composite material
172.0 11.727 11.574
173.0 8.974 8.369
174.0 2.607 9.333
174.5 4.213 7.329
174.8 4.674 9.018
175.0 0.856 10.341
176.0 21.049 12.003
178.0 31.168 13.231
179.5 7.867 21.292
186.8 28.454 18.944
187.1 18.778 36.351
Contact Force-Time history

See Figs. 2 and 3.
Energy Balance Plot
See Figs. 4 and 5.
Displacement-Time history
See Fig. 6.
Fig. 2 Contact force (kN)—time (ms) plot Chang-Chang
Fig. 3 Contact force (kN)—time (ms) plot Tsai-Wu

Fig. 4 Energy (J)—time (ms) plot Chang-Chang
Fig. 5 Energy (J)—time (ms) plot Tsai-Wu
Fig. 6 Displacement (mm)—time (ms) plot, a Chang-Chang criterion; b Tsai-Wu criterion

Failure Index Contour Plots

(Failure Index Contour Plots for Tsai-Wu criterion are distinct compared to
Chang-Chang criterion. They cannot be superposed.)
See Figs. 7 and 8.
Damage Propagation Failure Index Contour Plots
See Figs. 9 and 10.
Fig. 7 Failure index line contour plots for different plies for Chang-Chang criterion at time
t = 0.3 ms
0° (ply 1) 90° (ply 5) 0° (ply 6) 90° (ply 8)
Fig. 8 Failure index line contour plots for different plies for Tsai-Wu criterion at time t = 0.3 ms
t = 0.02 ms t = 0.06 ms t = 0.025 ms t = 0.3 ms
Fig. 9 Failure index line contour plots showing damage propagation at different time instants for
the ensemble for Chang-Chang criterion
t = 0.02 ms t = 0.06 ms t = 0.025 ms t = 0.3 ms
Fig. 10 Failure index line contour plots showing damage propagation at different time instants for
the ensemble for Tsai-Wu criterion
Failure Index Line Contour Plots for Damage Propagation for Various Failure
Mechanisms
See Figs. 11, 12, 13, 14, 15 and 16.
In this study, only a fraction of the results obtained in the modeling, analysis,
and simulation of ballistic impact on a laminated composite square plate using the
LS-DYNA software is given above. There is a clear need to validate the finite
element model developed for the intended analysis using a benchmark with known
target solutions. This work is in progress.
t = 0.03 ms t = 0.04 ms t = 0.19 ms t = 0.3ms
Fig. 11 Failure index line contour plots showing tensile fiber mode damage propagation at
different time instants for the ensemble for Chang-Chang criterion
t = 0.03 ms t = 0.04 ms t = 0.19 ms t = 0.3ms
Fig. 12 Failure index line contour plots showing tensile fiber mode damage propagation at
different time instants for the ensemble for Tsai-Wu criterion
t = 0.06 ms t = 0.2 ms t = 0.25 ms t = 0.3 ms
Fig. 13 Failure index line contour plots showing compressive fiber mode damage propagation at
different time instants for the ensemble for Chang-Chang criterion
t = 0.06 ms t = 0.2 ms t = 0.25 ms t = 0.3 ms
Fig. 14 Failure index line contour plots showing compressive fiber mode damage propagation at
different time instants for the ensemble for Tsai-Wu criterion
t = 0.03 ms t = 0.15 ms t = 0.25 ms t = 0.3 ms
Fig. 15 Failure index line contour plots showing compressive matrix mode damage propagation
at different time instants for the ensemble for Chang-Chang criterion
t = 0.03 ms t = 0.15 ms t = 0.25 ms t = 0.3 ms
Fig. 16 Failure index line contour plots showing compressive matrix mode damage propagation
at different time instants for the ensemble for Tsai-Wu criterion
The graphical post-processing capability in LS-DYNA is demonstrated in the

results presented above. However, there is a need to verify these predictions
through correlation studies where the numerical results are verified using experi-
mental data. This demands the generation of experimental results which is quite
tough in terms of research and test facilities. However, this work is also in progress
but in an indirect route. We are on the search for experimental results in the
published literature for which we will repeat the modeling analysis and simulation
using LS-DYNA software.
4 Conclusions
Nonlinear finite elements for continua and structures, in general, and a commercial
FEA software LS-DYNA, in particular, are used in this study for ballistic impact
damage tolerance assessment of laminated composite plate-type structures.
However, there is a clear need to validate finite element modeling using LS-DYNA
software for simulation of ballistic impact on laminated composites components
and structures. This demands benchmarks with known target solutions. Ballistic
impact damage consists of ply-by-ply failure and delamination between all the
plies. The ensemble of the results over all the plies and ply interfaces quantifies
impact damage. However, measurement of this ensemble demands computerized
ultrasonic C-scan of impact damage components, and there is also a need to exploit
software capabilities to graphically display the ensemble of damage. These are
topics of current research.
References
1. Y. Shi, T. Swait, C. Soutis, Modelling damage evolution in composite laminates subjected to

low velocity impact. Compos. Struct. 94, 2902–2913 (2012)
2. D. Feng, F. Aymerich, Finite element modelling of damage induced by low-velocity impact
on composite laminates. Compos. Struct. 108, 161–171 (2014)
3. M.A. Iqbal, A. Chakrabarti, S. Beniwal, N.K. Gupta, 3D numerical simulations of sharp
nosed projectile impact on ductile targets. Int. J. Impact Eng 37, 185–195 (2010)
4. L. J. Deka, S. D. Bartus, U. K. Vaidya, Damage evolution and energy absorption of FRP
plates subjected to ballistic impact using a numerical model, in 9th International LS-DYNA
Users Conference, 2006, Dearborn, Michigan, USA
5. J. Pernas-Sánchez, J.A. Artero-Guerrero, J. Zahr Viñuela, D. Varas, J. López-Puente,
Numerical analysis of high velocity impacts on unidirectional laminates. Compos. Struct. 107,
629–634 (2014)
6. N.K. Naik, P. Shrirao, Composite structures under ballistic impact. Compos. Struct. 66,
579–590 (2004)
7. M. Loikkanen, G. Praveen, D. Powell, Simulation of ballistic impact on composite panels, in
10th International LS-DYNA Users Conference, 2008, Dearborn, Michigan, USA
8. Sebastian Heimbs, Sven Heller, Peter Middendorf, Simulation of low velocity impact on
composite plates with compressive preload (LS-DYNA Anwender Forum, Bamberg, 2008)
9. S.K. Chelluru, Finite element simulations of ballistic impact on metal and composite plates,
M.Sc. thesis, Wichita State University, 2007
10. Costantino Menna, Domenico Asprone, Giancarlo Caprino, Valentina Lopresto, Andrea
Prota, Numerical simulation of impact tests on GFRP composite laminates. Int. J. Impact Eng
38, 677–685 (2011)
11. LS-DYNA Theory Manual, Livermore Software Technology Corporation (Livermore, CA,
2006)
12. LS-DYNA Keywords Manual, Volume I, Version 971 Livermore Software Technology
Corporation, Livermore, CA, May 2007
Prediction of Mode II Delamination Onset
Life Under Spectrum Fatigue Loads Using
Equivalent Strain Energy Release Rate
Concept
A. R. Anilchandra, M. Seshagirachari, Ramesh Bojja,

N. Jagannathan and C. M. Manjunatha
Abstract End notched flexure (ENF) test specimens of unidirectional IMA/M21

carbon fiber composite (CFC) were fabricated using standard autoclave process.
A Teflon insert was used to simulate a delamination at the midplane. Three-point
bend setup tests were conducted at an average frequency of 2 Hz using a 25 kN
servo-hydraulic test machine in room temperature conditions. Constant amplitude
fatigue tests were done at three different stress ratios, viz. R = 0.0, 0.5, and −1.0 to
construct the standard G-Nonset diagram, similar to S–N curve in its usefulness.
Nonset was identified as 5% change in initial compliance value. Using an equivalent
energy release rate parameter, Geq, all the curves were collapsed into a single curve
in the form of Basquin’s equation. The equation was subsequently used in pre-
dicting the delamination onset-of-growth life under a standard mini-FALSTAFF
spectrum load sequence. A fairly good correlation was found between the predicted
and experimental mode II onset-of-growth behavior.
Keywords Composite Delamination Spectrum load Onset-of-growth

Compliance
Nomenclature
R Stress ratio
Nonset Number of constant amplitude (CA) fatigue cycles for onset-of-growth
Gmax Maximum strain energy release rate (SERR)
Gmin Minimum SERR
Geq Equivalent SERR
GIIC Critical mode II SERR
G0eq Basquin's coeffecient
DG Range SERR
A. R. Anilchandra (&) M. Seshagirachari R. Bojja N. Jagannathan C. M. Manjunatha

Structural Technologies Division, Fatigue and Structural Integrity Group, CSIR-National
Aerospace Laboratories, Bengaluru 560017, India
e-mail: anilchandraar.mech@bmsce.ac.in

https://doi.org/10.1007/978-981-10-7197-3_17
196 A. R. Anilchandra et al.
c Best fit parameter
Nb Number of spectrum fatigue load blocks
D Damage fraction
1 Introduction
Aircraft structures are increasingly being replaced with fiber reinforced plastic
composites (FRPs) due to their high specific mechanical properties [1].
Delamination in such laminated FRPs possesses a serious threat to the structural
integrity since these structures are highly susceptible to in-service-related events
such as barely visible impact damages (BVID), assembly and service related events
(over tightening, collisions), and so on [2]. While identifying such defects are quite
challenging, deciding the fate of such “damaged” structure is also equally chal-
lenging [3]. The total life of a delamination under fatigue loading is the sum of
number of cycles spent in onset-of-growth (Nonset) and subsequently the number of
cycles needed for it to propagate to a critical length (NP). The fatigue loads could be
mode I (opening mode), mode II (shearing), mode III (tearing), or a mixed-mode
condition. The “no-growth” philosophy of damage tolerant design presumes exis-
tence of a delamination owing to the aforementioned reasons and the energy needed
in terms of load cycles for it to begin-to-grow, that is, onset-of-growth is regarded
as failure of the structure/component [4]. Owing to cost and safety considerations,
the aircraft industry would prefer life prediction so that catastrophic failures could
be prevented and various prediction models have been proposed in the literature [5].
Life prediction under spectrum fatigue loads can be classified into three major
categories, viz., empirical, phenomenological, and physics-based damage models.
Empirical models are simple and rely on experimental data (stress levels, stress
ratio, or frequency) without considering the inherent damage mechanisms. The
Palmgren-Miner (PM) cumulative damage accumulation model is one such life
prediction model and was used in our earlier work to predict the onset-of-growth
under mode II spectrum loading in a carbon fiber reinforced plastic composite
(CFRP) [6]. A novel constant onset life diagram (COLD), similar to constant life
diagram (CLD), was proposed and used in the prediction. However, the prediction
using CLD involves tedious interpolation technique and this is overcome in the
present work by using the equivalent strain energy release rate (Geq) concept.
Attempts have been made in the past to normalize stress (S)-number of cycles to
failure (N) plot in order to eliminate the effect of specimen geometry and test setup
for a particular stress ratio (R) [7]. Petermann and Plumtree [8] used an
energy-based approach to merge different R-ratio curves into a single curve and
used it in life prediction with a good correlation between the experimental and
predicted values. Geq concept, similar to Keq proposed by Hojo et al. [9], was shown
to merge various R-ratio curves into a single curve in the G-Nonset plot [10].
Prediction of Mode II Delamination Onset Life Under Spectrum … 197
The aim of the present work is,

• to make use of the Geq concept to merge all the R-ratio curves into a single R-
ratio curve in the G-Nonset and subsequently use it in predicting the
onset-of-growth under mini-FALSTAFF spectrum load sequence and
• validate with experimental results.
2 Experiment
2.1 Material and Specimen
The unidirectional carbon fiber/epoxy IMA/M21 prepreg was obtained from M/s
Hexcel in the form of 300 mm wide roll. The nominal thickness of prepreg (lamina)
was about 0.18 mm. The prepreg was cut into 300 mm 450 mm size and laid up
on a tool. Teflon sheet of 30 lm thickness was introduced at the mid-thickness
(neutral axis) to create an artificial delamination in the test specimens. The com-
posite was then cured in an autoclave maintained at a pressure of 7 bar at 180 °C
for 2 h [11]. The vacuum at 1 bar was maintained during curing process. The fiber
volume fraction of the fabricated laminate was 59%. The prepared laminates of
about 3.8 mm thick were ultrasonically C-scanned. End notched flexure (ENF) test
specimen measuring 130 mm 25 mm was cut from these laminates.
2.2 Spectrum Fatigue Tests
Spectrum fatigue tests were carried out on the ENF specimen in servo-hydraulic test
machine under load control mode at an average frequency of 2 Hz (four reversals
per second). The photograph of the test setup is shown in Fig. 1a. The schematic, in
Fig. 1b, shows the specimen dimensions and the loading arrangement. The fixture
is so designed that the specimen can undergo bending either way about the neutral
axis and further detailed description about the fixture could be found elsewhere
[12]. A standard mini-FALSTAFF load sequence was employed for the spectrum
tests. Mini-FALSTAFF is a short version of the FALSTAFF load spectrum [13],
which is a standardized variable-amplitude test load sequence developed for the
fatigue analysis of materials used for fighter aircraft and is shown in Fig. 2a. One
block of this load sequence (Nb = 1) consists of 18,012 reversals at 32 different
stress levels and represents loading equivalent of 200 flights. The actual load
sequence for experiments and fatigue life prediction was obtained by multiplying
all the peak/trough points in the entire block with a reference load. The compliance
of the specimen was determined on completion of every load block.
Onset-of-growth was assumed, similar to an accepted practice [14, 15] of whenever
Fig. 1 a Photograph of the test setup highlighting the specimen and support rollers; b schematic
of the test setup
Fig. 2 a Mini-FALSTAFF spectrum load sequence; b truncated mini-FALSTAFF
5% increase in compliance from the initial value was observed. In our earlier work,
a modified form of mini-FALSTAFF, called truncated mini-FALSTAFF, was used
wherein all the reversals below zero level were truncated as shown in Fig. 2b [6].
Although the modified mini-FALSTAFF was not used for testing in the present
work, the proposed life prediction methodology was tried on the truncated
mini-FALSTAFF as well and the results were compared with the results of our
earlier work.
2.3 Constant Amplitude Fatigue Tests
As a part of life prediction requirement, constant amplitude (CA) fatigue tests were
conducted at three different stress ratios, R = 0.0, 0.5, and −1.0 using a computer
controlled 25 kN servo-hydraulic test machine under load control mode, at 2 Hz.
The maximum strain energy release rate (maximum SERR or Gmax) for each of the
CA fatigue test was fixed on certain percentage of critical mode II SERR (GIIC) of
the material, determined in our earlier work [6]. Measuring the crack tip advance
through techniques such as the microscope is extremely difficult and might lead to
erroneous results as the crack remains closed under mode II condition and hence the
researchers have suggested the compliance calibration (CC) technique to monitor
the crack length [16]. A 5% increase in compliance from the initial value was
regarded as Nonset.
3 Onset-of-Growth Prediction Methodology
The Nonset under the mini-FALSTAFF load sequence was predicted using
Palmgren-Miner (PM) empirical method and compared with the experimental
results. The methodology employed was similar to the general procedure used for
fatigue life prediction for composites under spectrum loads [4]. But the uniqueness
in this prediction method lies in the fact that it makes use of only the Gmax-Nonset
data (or G-Nonset), similar to the S-N data. The flow chart for onset-of-growth
prediction is schematically shown in Fig. 3. The procedure of prediction contains:
• Rainflow counting of the fatigue cycles [17] in the spectrum load sequence to
obtain individual single cycles with specific Gmax and Gmin.
Fig. 3 Flowchart depicting the procedure used for prediction of onset-of-growth under spectrum
loading
• Construction of CA G-Nonset curves for various stress ratios. Further, the various
stress ratio curves are merged using the Geq concept [18] to get the Geq-Nonset
plot where
Geq ¼ DGð1cÞ Gcmax ¼ Gmax ð1 RÞ2ð1cÞ ð2Þ
and
DG ¼ Gmax ð1 RÞ2 ð3Þ
where c is the best fit for the Geq curve and will be discussed in subsequent
section.
• Determination Nonset for each of the counted load cycles using the Geq-Nonset
plot.
• Calculation of the damage fraction (Di) for each of the counted load cycles, i, as
the ratio of cycle count to Nonset,i obtained from the Geq-Nonset plot using the
Basquin’s equation, and, finally,
• Determination of the total fatigue damage per load block (D) by summation of
the damage fraction. The material is assumed to fail when the total damage
fraction reaches 1.0.
4.1 Experimental Onset-of-Growth Under Spectrum

Fatigue Tests
The onset-of-growth behavior of mode II delamination under mini-FALSTAFF

load sequence determined experimentally at various reference SERR, GRef is shown
in Fig. 4. Figure 4 also contains the predicted onset-of-growth life which will be
Fig. 4 Experimental and predicted GRef – (Nb)onset curves for mode II delamination of IMA/M21
CFC determined under, a mini-FALSTAFF spectrum load and b truncated mini-FALSTAFF
discussed in the subsequent sections. As expected, the onset-of-growth of mode II

delamination increased with decrease in reference SERR, GRef. These results are in
similar trend of reference stress-life curves for composites under spectrum loads
[19, 20].
4.2 Prediction of Onset-of-Growth Behavior
For the purpose of prediction of onset-of-growth under spectrum loads, the fracture
toughness GIIC value was determined (average value is 2.2 kJ/m2) and the proce-
dure is elaborated in [6]. Following the procedure explained earlier (Fig. 3), the
onset-of-growth life was predicted. The predicted number of blocks for
onset-of-growth (Nb)onset at various reference SERR (GRef) under spectrum load
sequence is shown along with experimental results in Fig. 4a. Prediction appears to
converge with the experimental results at higher GRef (1.25 kJ/m2) and drift away
as the GRef decreases. In our earlier work, COLD was used for prediction under
modified mini-FALSTAFF load sequence and the prediction was found to be
non-conservative (Fig. 4b) [6]. However, it is seen that by making use of proposed
methodology (Geq-Nonset) the accuracy of prediction may be improved.
4.3 Constant Amplitude Fatigue Tests and Merger of Stress

Ratios
The G-Nonset curves for IMA/M21 CFC material under mode II condition at three
different stress ratios are shown in Fig. 5a. It may be clearly seen that increasing the
stress ratio increases the Nonset. This is in agreement with similar observations made
earlier in polymer composites [21, 22]. The endurance limit also increased with
stress ratio, as has been observed by Bak et al. [21]. The experimental G-Nonset
curves of the IMA/M21 composite were merged into a single curve using the Geq
concept, explained in Sect. 3, and fit to an equation similar to Basquin’s law as
shown in Fig. 5b. A c value of 0.34 was found to best fit the data points and
indicates the relative contribution of DG and Gmax. The Basquin’s coefficient
(G0eq = 5.37) and exponent (b = −0.24) values were used to estimate Nonset and
subsequently to calculate the damage fraction per rain flow counted cycle, as
explained in Sect. 3. The advantage of having a single R-ratio curve in the G-Nonset
data is that the subsequent exercise of constructing COLD, used in life prediction
models, is overcome and the life prediction is shown to be even better compared to
the empirical model tried earlier using COLD data, shown in Fig. 4b.
Delamination studies under static and fatigue loads have been shown to exhibit
some significant scatter in test results [23]. The constant amplitude data shown in
Fig. 5 also shows a wide scatter. Since this data is fit to an empirical equation and
Fig. 5 a The constant amplitude G–Nonset curves determined for mode II delamination in IMA/
M21 CFC material, b the three R-ratio curves were merged into a single curve and fit to Basquin’s
equation
used further in prediction, the scatter may induce errors in predictions. The anomaly
in the experimental and the predicted results may be attributed to the scatter in the
G-Nonset data. Hence, further work would be necessary in understanding the effect
of large scatter in the G-Nonset data and its effects on prediction.
5 Conclusions
Based on the results obtained from the present investigation, the following con-
clusions may be drawn:
• The three stress ratio curves of R = 0.5, 0.0, and −1.0 in the G-Nonset plot were
successfully merged into a single curve using the equivalent SERR (Geq) con-
cept and fit in the form of Basquin’s equation. This master plot of Geq-Nonset was
subsequently used in life prediction methodology
• The predicted results are non-conservative. Nevertheless, further work is
required in understanding the effect of large scatter in the G-Nonset data on
prediction, and possible the use of nonlinear damage accumulation models for
onset life prediction.
Acknowledgements Authors wish to thank the AR&DB for financially supporting the project.
The support and encouragement provided by Mr. Shyam Chetty, Director, Dr. Satish Chandra,
Head, STTD, Dr. Ramesh Sundaram, ACD, CSIR-NAL are acknowledged. Thanks are also due to
scientists and technical support staff members of FSIG-STTD and ACD, CSIR-NAL for their
assistance in experimental work.
References
1. Hexcel ready to fly on the A350 XWB, Reinf. Plast. 57, 25–26 (2013)
2. A.C. Garg, Delamination—a damage mode in composite structures. Eng. Frac. Mech. 29,
557–584 (1988)
3. R.P. Wei, Fracture Mechanics: Integration of Mechanics, Materials Science, and Chemistry
(Cambridge University Press, 2010)
4. FAA, Composite Aircraft Structure: Advisory Circular (AC) 20-107B, Change 1, FAA (2010)
5. N.L. Post, S.W. Case, J.J. Lesko, Modeling the variable amplitude fatigue of composite
materials: a review and evaluation of the state of the art for spectrum loading. Inter. J. Fatigue
30, 2064–2086 (2008)
6. N. Jagannathan, A.R. Anilchandra, C.M. Manjunatha, Onset-of-growth behavior of mode II
delamination in a carbon fiber composite under spectrum fatigue load. Compos. Struct. 132,
477–483 (2015)
7. M. Wisnom, M. Jones, Through thickness fatigue failure of fibre-reinforced composites.
Aeronaut. J. 102, 83–88 (1998)
8. J. Petermann, A. Plumtree, A unified fatigue failure criterion for unidirectional laminates.
Compos. Part A: Appl. Sci. Manuf. 32, 107–118 (2001)
9. M. Hojo, K. Tanaka, C.G. Gustafson, R. Hayashi, Effect of stress ratio on near-threshold
propagation of delamination fatigue cracks in unidirectional CFRP. Compos. Sci. Technol.
29, 273–292 (1987)
10. I. Maillet, L. Michel, F. Souric, Y. Gourinat, Mode II fatigue delamination growth
characterization of a carbon/epoxy laminate at high frequency under vibration loading. Eng.
Frac. Mech. 149, 298–312 (2015)
11. http://www.hexcel.com/Resources/DataSheets/Prepreg-Data-Sheets/M21_global.pdf
12. A.R. Anilchandra, R. Bojja, N. Jagannathan, C.M. Manjunatha, Variable amplitude fatigue
testing to characterize mode II delamination in a polymer composite. Trans. Indian Inst. Met.
69, 421–424 (2016)
13. P. Heuler, H. Klätschke, Generation and use of standardised load spectra and load–time
histories. Int. J. Fatigue 27, 974–990 (2005)
14. W.X. Wang, M. Nakata, Y. Takao, T. Matsubara, Experimental investigation of test methods
for mode II interlaminar fracture testing of carbon fiber reinforced composites. Compos.
Part A Appl. Sci. Manuf. 40, 1447–1455 (2009)
15. ASTM D6115, Standard test method for mode I fatigue delamination growth onset of
unidirectional fiber-reinforced polymer matrix composites, vol. 15.03. Annual Book of
ASTM Standards, ASTM International, West Conshohocken, PA, 2003
16. A.J. Vinciquerra, B.D. Davidson, J.R. Schaff, A.L. Smith, Determination of the mode II
fatigue delamination toughness of laminated composites. J. Reinf. Plast. Compos. 21, 663–
677 (2002)
17. ASTM E1049, Standard practices for cycle counting in fatigue analysis. Annual Book of
ASTM Standards, ASTM International, West Conshohocken, PA, 2003
18. I. Maillet, L. Michel, F. Souric, Y. Gourinat, Mode II fatigue delamination growth
characterization of a carbon/epoxy laminate at high frequency under vibration loading. Eng.
Fract. Mech. 149, 298–312 (2015)
19. C.M. Manjunatha, R. Bojja, N. Jagannathan, Enhanced fatigue performance of a polymer
nanocomposite under spectrum loads. Mater. Perform. Charact. 3, 327–341 (2014)
20. C.M. Manjunatha, R. Bojja, N. Jagannathan, A.J. Kinloch, A.C. Taylor, Enhanced fatigue
behavior of a glass fiber reinforced hybrid particles modified epoxy nanocomposite under
WISPERX spectrum load sequence. Int. J. Fatigue 54, 25–31 (2013)
21. B.L.V. Bak, C. Sarrado, A. Turon, J. Costa, Delamination under fatigue loads in composite
laminates: A review on the observed phenomenology and computational methods. Appl.
Mech. Rev. 66, 1–24 (2014)
22. A. Argüelles, J. Viña, A.F. Canteli, M.A. Castrillo, J. Bonhomme, Interlaminar crack
initiation and growth rate in a carbon-fibre epoxy composite under mode-I fatigue loading.
Compos. Sci. Technol. 68, 2325–2331 (2008)
23. T.K. O’Brien, W.M. Johnston, G.J. Toland, Mode II interlaminar fracture toughness and
fatigue characterization of a graphite Epoxy Compos Mater, NASA/TM–2010-216838
Dependence of Crack Velocity on Stress
Intensity Factor in PMMA Using
Single-Edge-Notched Clamped Beams
G. R. Rahul, V. Jayaram and S. Bose
Abstract The Single-Edge-Notched Clamped Beam (SENCB) geometry, which

has been used for studying fracture in graded coatings, is adapted to the study of
fracture in macroscopic specimens of PMMA, a brittle, transparent polymer. Test
materials were fabricated by compression molding, pre-cracked using a safety
razor, and loaded at constant displacement rate at room temperature, while crack
propagation was monitored with a video camera. Crack growth with time was
tracked and velocity computed. The load measured from the experiment was used
in FEM in conjunction with the crack length to determine the applied stress
intensity factor at the different instance of crack propagation and was correlated
with crack velocity. Crack grew from the interior, and a jump in crack velocity was
observed as the crack spanned the width of the sample. Pre-cracking is done by
physically restraining the legs and unloading as the crack propagates across the
width. The average velocity of crack front was correlated with stress intensity factor
(K) and was found to broadly agree with what is reported for PMMA at velocities
more than 0.1 mm s−1. At low velocity, the curve appears to become insensitive to
K at a plateau of *0.7–0.9 MPa m1/2.
Keywords PMMA
Single-Edge-Notched Clamped Beams Stress intensity

factor v(K) curve Slow crack growth
Nomenclature
SENCB Single-Edge-Notched Clamped Beam
PMMA Polymethyl meta acrylate
SENB Single edge notched beam
DT Double torsion
PC Parallel cleavage
TC Tapered cleavage
DSLR Digital single lens reflex
G. R. Rahul (&) V. Jayaram S. Bose

Materials Engineering Department, Indian Institute of Science, Bangalore 560012, India
e-mail: to.grrahul@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_18
206 G. R. Rahul et al.
FEM Finite element method

K Stress intensity factor
R Resistance to crack extension
a/w Crack length to width ratio
µ Coefficient of friction
1 Introduction
The Single-Edge-Notched Clamped Beam (SENCB) has been shown to be suitable

for fracture testing in microscale [1]. The test geometry is stable even under con-
stant load because the stress gets re-distributed to the corners of the geometry as the
crack propagates, whereas in most of the conventional geometries, the stress
intensity factor is either constant or rises with crack extension. The microscale test
specimen was made by machining out a cavity in bulk material and cutting a notch
on the bottom using focused ion beam machining as shown in Fig. 1a. This con-
figuration can be seen as a beam with connected legs to support it. In principle, the
advantages observed in micron-scale should be capable of being exploited in
macroscale testing. The purpose of the present work is to explore the application of
this technique to a well-characterized material, i.e., polymethyl methacrylate
(PMMA). Because access to the lower part of the beam for making a pre-crack
would be difficult, the present work uses a modified geometry wherein two finite
legs are present, shown in Fig. 1b. If the legs were physically constrained, the
effective stiffness would approach that of the clamped beam; however, since the
legs are free to slide sideways, the constraint is limited by the friction between the
leg and the supporting substrate. The crack stability is a function of geometry,
initial a/w, machine stiffness [2], and also of the initial crack front and the material’s
crack resistance behavior. In SENCB geometry, the driving force for crack prop-
agation keeps decreasing with crack advance, such a geometry allows the study of
slow crack growth. Thus, in comparison with the double torsion geometry [3]
which employs a constant stress intensity factor, the SENCB offers the combined
effect of material resistance and stress redistribution during the crack advance to
produce a range of velocity and hence a range of v(K) data points.
The effect of temperature on the velocity–toughness relationship in PMMA has
been studied extensively [3–5]. Toughness decreases with temperature; however, at
a constant temperature, the toughness increases with crack velocity in region I as
shown in Fig. 2. Nonlinear behavior of energy release rate has been reported at low
velocities [3, 4]. However, much of the data come from extrapolations of velocity
from load relaxation. A schematic of stress intensity factor variation with velocity is
given in Fig. 2 [3]. The stress intensity factor rises up to a certain velocity, and then,
it drops and rises again. It has been reported that the drop is due to thermal effects
(softening under conditions of adiabatic heating) at the crack tip [6]. The current
Dependence of Crack Velocity on Stress Intensity Factor … 207
Fig. 1 a SENCB
microbeam; b SENCB beam
in macroscale under loading
Fig. 2 Schematic of stress

intensity factor versus crack
velocity [3]
study is confined to the velocity region I, where stress intensity factor rises with
velocity.
Applied stress intensity factor at the different instances of crack propagation was
computed by Finite Element Modeling. The contour integral approach in com-
mercial FEM package ABAQUS® was employed to evaluate stress intensity factor
[7]. A modulus-independent method is required for computation of SIF, as crack
growth is suspected to result in an effective change of modulus ahead of the crack
tip in PMMA. With load as input, stress intensity factor becomes independent of the
modulus. Care is taken to simulate realistic boundary conditions with respect to
frictional sliding of the legs.
The experiments were done at room temperature in displacement control mode
as they are inherently more stable than load control, and crack growth was recorded
using a camera. Using the video of crack propagation, the crack length variation
with time was tracked and correlated with load. The instantaneous crack length and
the load of the geometry define the condition of the specimen. The crack is initiated
in the interior of the beam and propagated across its width as it propagated. An
acceleration of crack was observed when the crack front spanned the width of the
beam. The stability of the geometry resulted in the drop of driving force and
therefore in the reduction in crack velocity as the crack moved. A wide range of
initial crack velocities was produced by varying the displacement rate and due to
random variations in the initial crack profile.
2 Experimental Procedure
2.1 Sample Preparation
Polymethyl methacrylate, commonly called PMMA, is a quasi-brittle polymer.

PMMA (Guzpol-P, 876G) with melt flow index of 6 g/10 min was procured from
Gujarat State Fertilizers and Chemicals, India. The sample was prepared by pres-
sure casting PMMA pellets in a Teflon mold in a compression molding machine.
The machine can apply constant pressure while casting. The casting was done at
220 °C with plates preheated to 220 °C. The pellets were added in steps into the
preheated mold, giving enough time for the material to flow, thus ensuring adequate
supply of material in the mold cavity when the pressure is applied. The excess
material is flushed out in the process. The sample is then allowed to slowly cool to
room temperature under pressure inside the mold by switching off the plate heating.
The surface of the specimen was polished to remove surface cracks and to obtain
transparency.
2.2 Notching and Pre-cracking
A blunt notch is made on the bottom center of SENCB using a high-speed linear
precision saw. The machine has a cutting disk which can transverse linearly. The
cutting disk has a thickness of 0.8 mm which leads to a notch width of
0.8 ± 0.1 mm. The cutting is done on the center of the beam at 4000 rpm and at a
constant feed rate of 1.2 mm/min. Care was taken to ensure that the cutting face is
normal to the diameter of the cutting disk. Since PMMA is a soft material, an initial
crack could be made by pressing a double-edged safety razor on the notch.
However, such a crack may not be sharp and uniform across the thickness. In order
to produce an ideal pre-crack, the beam’s legs are clamped (to ensure maximum
stability), loaded until the crack pops in and propagates across through the width,
and then unloaded.
2.3 Testing
The specimen was tested in Instron make Universal Testing Machine (UTM).
A load cell with a capacity of 5 KN was used for load measurement. The load cell
can measure load above 0.2% of maximum capacity (i.e., 10 N). The specimen was
loaded using a cylindrical indenter of 3 mm diameter. A flat plate was mounted as
base. Sample was kept on the base plate, and indenter aligned above the pre-crack
by visual inspection. The test was captured using a Digital Single Lens Reflex
(DSLR) camera. The camera is focused on to the crack plane. Optimum illumi-
nation was provided to get a contrast between the sample and the crack. The test
and the video recording are synchronized with a tolerance of ± 0.03 s. Tests were
conducted in displacement control at constant crosshead velocity ranging from
5 10−5 to 1 10−2 mm/s. The test was stopped when the sample fractured
completely.
3 Results and Analysis
3.1 Friction Coefficient Measurement
The friction between the bottom of the specimen leg and the base plate affects the
stiffness of the beam and contributes to the redistribution of stress in the body.
Physically, the role of friction may be seen as follows: when the beam is centrally
loaded, the legs will tend to splay outward, thus opening the crack further.
Prevention of such lateral sliding will, therefore, reduce the stress intensity factor.
Therefore, friction has to be accounted for in the FEM model. The coefficient of
friction was evaluated in a two-step experiment. In the first step, a beam without a
Fig. 3 Load displacement

simulated with coefficient of
friction 0.1, 0.14, 0.2
overlapped with experimental
load displacement data in the
case of beam with
unconstrained legs
notch was clamped and loaded in the elastic regime. The clamping constrained the
leg from sliding when the load was applied. The second step was to load the same
beam without any constraint. The beam was then modeled in FEM, and stiffness of
the clamped beam was matched by varying Young’s modulus. By fitting in the
measured modulus, the stiffness was again matched for the second experiment by
varying the friction coefficient as shown in Fig. 3. By this method, a friction
coefficient value of 0.14 was obtained and was used throughout the study for all
beams.
3.2 Load and Crack Length Correlation
In displacement control, as the crack propagates, the load drops in response to the
drop in stiffness. Figure 4a shows the load versus time during crack growth and
crack length corresponding time superimposed. The average crack length was
measured by splitting the recorded video into frames and measuring the area of
cracked region at various instances. Dividing the cracked area with the width of the
specimen yields average crack length. And the derivative of crack length with time
was taken to obtain crack velocity. Crack length and crack velocity are synchro-
nized with experimental load values. Figure 4b shows velocity versus time for the
same experiment shown in Fig. 4a. The velocity decreases steadily as crack
advances. A range of velocity data is obtained from each experiment.
The factors that limit the range of analysis are as follows:
• Uncertainty in synchronizing the load displacement and crack length-time data.
• The ability to resolve crack increments much smaller than 0.1 mm.
Fig. 4 a Load and crack

length with time. As the crack
propagates, the combination
of crack resistance and falling
driving force retards the
crack; b crack velocity with
time for the same experiment
The load displacement and the test video can be synchronized to within
± 0.03 s. The uncertainty in load value at each point of crack increment would then
vary with crack velocity. At low crack speeds, the load drops, and therefore, the
uncertainty in measurement would also be low. However, at high crack velocity, a
small uncertainty results in large variation of load values and hence high uncer-
tainty in stress intensity factor. Therefore, better accuracy in experimental video
synchronization is required for exploring behavior in the high-velocity regime. In
the setup used a crack increment of 0.1 mm can be resolved. Therefore, in the
low-velocity regime, uncertainty due to crack length measurement dominates. The
uncertainty in measurement in crack length also leads to a larger error in K for long
cracks, which becomes a factor even for slow cracks.
3.3 Stress Intensity Factor Using FEM
Average crack length measured at discrete instances of crack propagation from the
video recorded and the load applied at corresponding instances obtained from the
experimental data describes the state of the specimen during the crack advance.
Applied stress intensity factor at these discrete instances is evaluated by solving the
FE model of the test specimen under corresponding conditions. The friction
between the specimen and the base plate, measured as shown in Sect. 3.1, is
incorporated in the FE model.
The validity of the FE formulation which was tested by comparing the numerical
solution of stress intensity factor to the analytical one for the ASTM three-point
bend SENB geometry was modeled, and the numerical approximations closely
matched with the analytical results. Contour integral method in FEM commercial
was used for evaluating stress intensity factor. The contour integral method
calculates J integral from displacement field and then calculates K values in each
mode [7].
3.4 Crack Velocity and Profile in PMMA
The pre-crack made by the double-sided razor blade would never be uniformly
through thickness. This results in lower stress intensity factor near crack tip than if
it were a through crack. Hence, the crack starts moving at a higher stress intensity
factor, and therefore, higher load than that needed to move a straight crack. It is
observed that the initial crack develops at the interior as shown in Fig. 5a and grows
simultaneously toward the indenter and the free edges, as shown in Fig. 5b. Once
the crack becomes straight across the width, a sudden increase in stress intensity
factor is experienced. The excess in K above equilibrium would manifest as driving
force and accelerate the crack. Hence, this excess in K manifests as an initial high
crack velocity even if the tests are done at quasi-static displacement rates.
Subsequently, the crack stabilizes and decelerates, as shown in Fig. 5c. To achieve
Fig. 5 a Crack initiated; b crack grown across the width; c crack face stabilized
slow crack growth, the crack was allowed to grow under loading and unloaded as
the crack spreads fully across the beam width.
3.5 Crack Velocity Toughness Relation in PMMA
Average velocity of the crack front is plotted against stress intensity factor in a log–
log plot, Fig. 6a. The data from earlier work on SENB, parallel cleavage, tapered
cleavage, and double torsion tests are also plotted for comparison [3, 8, 9] in
Fig. 6b. Unlike the double torsion and tapered cleavage beams which yield one data
point per experiment, a single SENCB beam yields stress intensity factor for more
than a decade of velocity values. Results from all the geometries suggest increasing
velocity with stress intensity factor. Considering beams individually as given in
Fig. 6a, a particular stress intensity factor yields a range of crack velocity. Tests
Fig. 6 a Stress intensity

factor (K) versus crack
velocity (v) from different
SENCB beam; b Stress
intensity factor (K) versus
crack velocity (v) obtained
from SENCB beams
compared with the data from
SENB, PC, TC, and DT
geometries [3, 8, 9] in log
linear plot
done on single edge notched beams and parallel cleavage geometries also yield
considerable scatter. It must also be recognized that the previous double torsion data
[3, 9] are not a product of direct measurements but come from models that relate
load relaxation data to velocity.
Broadly speaking, the data from the present work fit in well with published
results. Two bands of data points can be identified: above a velocity of around
0.1 mm/s where the stress intensity factor rises steeply and a plateau below
0.1 mm/sec where one can suggest the existence of a threshold stress intensity
factor of 0.75–0.9 MPa m1/2 below which crack growth ceases. However, confir-
mation of such a threshold requires sample sizes and microscopic measurements
that allow smaller crack increments to be measured than are realizable in the present
optical system.
4 Conclusions
The single edged clamped beam geometry has been shown to work satisfactorily as
a stable macroscopic test system to study crack propagation. Direct crack visual-
ization coupled with load measurements allows the velocity dependence of crack
growth with stress intensity factor to be mapped over four decades in crack velocity
over which the toughness of PMMA varies from 0.8 to 1.6 MPa m1/2.
Acknowledgements Financial support from Defense Research and Development Organization is

gratefully acknowledged.
References
1. N.B. Jaya, S. Bhowmick, O.L. Warren, V. Jayaram, Optimization of clamped beam geometry
for fracture toughness testing of micron scale samples. Phil. Mag. 95, 1945–1966 (2015)
2. F.I. Baratta, W.A. Dunlay, Crack stability in simply supported four point and three point loaded
beams of brittle material. Mech. Mater. 10, 149–159 (1990)
3. G.P. Marshall, L.H. Coutts, J.G. Williams, Temperature effects in fracture of PMMA. J. Mater.
Sci. 9, 1409–1419 (1974)
4. A.G. Atkins, C.S. Lee, R. Caddell, Time-temperature dependent fracture toughness of PMMA.
J. Mater. Sci. 10, 1381–1393 (1975)
5. J.C. Pollet, S.J. Burns, An analysis of slow crack propagation data in PMMA and brittle
polymers. Int. J. Fract. Mech. 13, 775–786 (1977)
6. J.G. Williams, Visco-Elastic and thermal effects on crack growth in PMMA. Int. J. Fract.
Mech. 8, 393–401 (1972)
7. D.S. Simulia, Abaqus/CAE Users Manual version 6.7. Retrieved from egr.msu.edu: http://
www.egr.msu.edu/software/abaqus/Documentation/docs/v6.7 (2015)
8. G.P. Marshall, J.G. Williams, The correlation of fracture data for PMMA. J. Mater. Sci. 8, 138–
140 (1973)
9. J.Y. Robert, W.R. Beaumont, Time-dependant failure of poly(methyl methacrylate). Polymer
17, 717–722 (1976)
Effect of Wrap Thickness and Ply
Orientation on Cracking and Failure
of FRP-Wrapped Columns
Pradeep B. Kodag and Gaurang R. Vesmawala
1 Introduction
Nowadays, retrofitting and rehabilitation of existing reinforced concrete structures

are a major concern all over the world. FRP sheets are lightweight and
non-corrosive, show high strength, and can be modified to satisfy performance
requirements. This makes the fiber-reinforced composites a popular alternative
material over the conventional materials available in the market. FRP jackets sig-
nificantly increase the load carrying capacity and ductility of RC columns when
used as an external reinforcement [1–3].
Most of studies showed that load carrying capacity of confined RC columns is
mainly affected by parameters such as aspect ratio of specimens, FRP thickness,
concrete compressive strength, modulus, and Poisson’s ratio of wrapping sheet fiber
orientation and type of loading [4, 5].
Some experimental studies were conducted to assess the effect of the type of
loading on load carrying capacity and stiffness of confined RC columns. The results
of tests conducted showed that the effectiveness of FRP-wrapped RC columns was
influenced by eccentric loading. As the eccentricity increases, the effect of confining
action of FRP decreases for eccentrically loaded RC columns as compared to
concentrically loaded RC columns [6].
Punurai et al. studied experimental and analytical results of RC columns con-
fined with carbon fiber-reinforced polymer (CFRP) subjected to biaxial bending.
Total five RC column specimens wrapped with different layers of CFRP jackets
were experimentally tested up to failure. Effect of orientation of CFRP sheets in
longitudinal and transverse directions of column was studied. For all specimens,
average load carrying capacity was achieved with CFRP sheets applied on all sides.
P. B. Kodag (&) G. R. Vesmawala

Department of Applied Mechanics, Sardar Vallabhbhai National Institute of Technology,
Ichhanath, Surat 395007, Gujarat, India
e-mail: pkodag@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_19
216 P. B. Kodag and G. R. Vesmawala
Result showed that the lateral deformation was reduced when the orientation of
CFRP wrapping is along the longitudinal direction. Computational analysis done to
calculate stress–strain equation for CFRP-confined concrete confirmed the behavior
of tested columns. The results of experimental work and analytical program for the
load versus deflections and the moment versus curvature curves were compared [7].
Rahai and Hamed carried out the experimental study on eight RC columns
wrapped with carbon fiber-reinforced polymer (CFRP) under axial load and biaxial
loading up to failure. This study includes variables such as CFRP sheet thickness,
fiber orientations, and bi-eccentric loading. To assess the effect of these variables on
the moment–curvature relationship and load–longitudinal displacement relation-
ship, this study was carried out. Increasing longitudinal layers slightly higher than
transverse layers results in a greater loading and displacement capacity of the
specimens due to the overall behavior of RC wall-like columns. There was an
improvement in the strength and ductility of confined RC columns [8].
Li and Hadi conducted the parametric study on FRP-wrapped columns using
finite element-based software. They studied the effect of parameters like FRP
thickness, stiffness, eccentric loading, and fiber orientation, comparative study of
CRFP, E-glass, and steel fibers on concrete specimen wrapped with FRP. After
reaching the unconfined compressive strength of columns, the mechanical properties
of FRP jacket play a major role on the performance of FRP-confined columns [9].
Youcef et al. carried out an experimental work on CFRP-wrapped square RC
columns under biaxial loading. For this experimentation, one to three FRP layers
were wrapped to square RC columns. The load carrying capacity and deformation
capacity were improved as the number of layers of FRP wrapping was increased [10].
2 Experimental Program
In this project, the experimental work was carried out for eccentrically loaded
FRP-wrapped column to study the effect of change in wrap thickness and ply
orientation. The experimental work was carried out by casting of 45 circular RC
columns. Concrete of M20 grade was used. Experimental work was divided into
two parts such as casting of unwrapped specimens and casting of GFRP-wrapped
specimens. To check the consistency of the concreting, test on three cubes per batch
of concreting was carried out. This experimental study was carried out to investi-
gate the effect of following parameters on the strength of the columns reinforced
with fiber composites subjected to concentric and eccentric loading.
1. Behavior of RC column due to change of loading pattern.
2. To study the effect of GFRP strengthening on the ultimate load carrying capacity
of FRP-confined columns.
3. To understand the associated cracking and failure mechanism.
Effect of Wrap Thickness and Ply Orientation on Cracking … 217
For the study, forty-five non-slender circular RC columns were cast; each col-
umn is having 170 mm diameter and 700 mm height. To check consistency of mix
design, three cubes having size 150 mm 150 mm 150 mm were cast and
tested, per batch of concrete. Parameters such as the GFRP thickness, fiber orien-
tation, and the eccentricities were considered as variables. The different GFRP
thickness as 0.43 and 0.86 mm (1–2 layers, respectively) and the fiber orientations
as 0°, 45° and 90° were investigated. 0° referred to the orientation of fibers per-
pendicular to the column axis, and 90° referred to the orientation of fibers parallel to
the column axis. The experimental program is summarized in Table 1.
All columns were cast from different batch of concrete. A concrete with a com-
pressive strength of 20 MPa and a slump of 80–100 mm was maintained for this
study. The longitudinal steel reinforcements were 4–10 mm Ø. The transverse
reinforcement was provided with circular ties of 6 mm Ø stirrups 200 mm c/c. The
reinforcement details for the columns are shown in Fig. 1.
2.2 Specimen Preparation and Test Setup
Column molds had to be suitably made for all the specimens with semicircular
metal as shown in Fig. 2. RC cylinders were cast as per M-20 mix proportion in six
batches as per dimension and after 28-day curing; the surface of cylinder was
cleaned and prepared for strengthening on the next day. The GFRP fiber sheets
were applied to the surface of column cylinder, saturated with epoxy. To maintain
bond between GFRP sheet and concrete surface, resin was used. In transverse layer
of fiber direction, 70 mm overlap was provided. The load was applied on cylinder
by means of a 2000-kN load capacity universal testing machine up to failure. At top
of the cylinder, eccentric load was applied with the help of special arrangement of
plates of 250 mm in diameter and 25 mm thick. A steel ball of 25 mm in diameter
was then placed in the relevant groove, and then, the assembly was placed under the
UTM for loading.
218
Table 1 Experimental program

Designation Ply Number of layers Internal Number of Eccentricity
orientation (°) of GFRP reinforcement specimens (mm)
ex ey
UE0 – – Yes 3 00 00
UE30 – – Yes 3 30 30
UE40 – – Yes 3 40 40
SWO0E0 0 Single Yes 3 00 00
DWO0E0 0 Double Yes 3 00 00
Note D—double wrap; E—eccentricity; ex—eccentricity about X-axis; ey—eccentricity about Y-axis; O—ply orientation; S—single wrap; U—unwrapped; W
—wrapped
P. B. Kodag and G. R. Vesmawala
(a) (b)
Fig. 1 Reinforcement details of circular specimen a elevation, b cross section
Fig. 2 Molds for circular columns
3 Result and Discussion
The columns were tested under load-controlled universal testing machine. The axial
load was applied to the column specimen at a load rate of 1 kN/s. Figure 3 shows
an experimental setup for axially and eccentrically loaded column specimens.
Fig. 3 Laboratory test setup for RC columns a unwrapped specimen, b wrapped specimen
3.1 Ultimate Load Carrying Capacity
Circular columns with 0° orientation exhibited higher axial load carrying as com-
pared to that of other columns. As eccentricity increases, load carrying capacity of
specimen decreases. 90° orientation of fiber exhibited lowest load carrying capacity.
Results clearly demonstrated that variation of 0° orientations under eccentric
loading plays significant role to increase load carrying capacity for RC columns.
45° orientation gives average load carrying capacity under loading. The increase in
the load carrying capacity of GFRP-wrapped circular columns was 150% as
compared to that of unwrapped columns. Circular shape gives full confinement and
therefore has resulted in attainment of the highest compressive strength.
3.2 Load Versus Axial Deformation
It could be observed from Fig. 4 that GFRP-wrapped circular RC columns with ply
orientation of 45° had high load carrying capacity than 90° ply orientations. Load
carrying capacity decreases with increase in the eccentricity and ply angle. So as a
result, 45° gives better confinement than 90° orientations.
Axial strain for columns was calculated from vertical deformation measured
from LVDT. Axial stress for columns was calculated as load divided by
cross-sectional area. Average axial strain was calculated from average axial
deformation for the columns. Average axial stress and axial strain for unwrapped
and GFRP-wrapped columns had exhibited higher axial strain as compared to that
of unwrapped columns. The stress–strain specimen confinement can be considered
1000 UE0
UE30
800 UE40
SWO0E0
600 SWO0E30
Load, kN
SWO0E40
400 SWO45E0
SWO45E30
200 SWO45E40
SWO90E0
0 SWO90E30
0 2 4 6 8 10 12 14
SWO90E40
Axial deformation, mm
Fig. 4 Axial load versus Axial deformation
as two zones. In the first zone, stress–strain behavior of GFRP-wrapped specimen

was linear, i.e., obeys Hooke’s law. In the second zone, deformation in vertical
direction increases with a little increase in load and GFRP wrapping got activated,
transferring the stress from dilated concrete to FRP.
3.3 Failure Modes and Crack Pattern
After reaching the ultimate strength, the failure of unwrapped RC columns was
because of excessive spalling of concrete at surface and splitting of concrete between
the stirrups. The governing mode of failure was shearing and spalling of the concrete
in case of the unwrapped columns. Under eccentric loading, columns specimens
failed mainly because of shearing. Concrete came out of the stirrups at the ultimate
compressive load because of buckling of reinforcement as shown in Fig. 5.
Fig. 5 Failure pattern of unwrapped column specimen

Fig. 6 Failure pattern of wrapped specimen a 0° wrap, b 45° wrap, and c 90° wrap
The failure of the GFRP-wrapped RC columns has been divided into three
modes: (i) tensile rupture of the GFRP layer, (ii) delamination of the GFRP layer,
and (iii) a combination of delamination and tensile rupture of the GFRP layer. The
failure modes for the GFRP-wrapped columns are shown in Fig. 6. Most of the
confined columns were failed by the rupture of the GFRP at top or bottom location.
During the application of loading for the columns, typical sound was heard sig-
nifying the straining of the GFRP laminate and the cracking of the epoxy resin.
Final failure for the column occurred suddenly with a significant explosive noise.
Once the column reaches to the failure load, sudden rupture of GFRP wrapping was
observed.
The tensile rupture of FRP composite indicates the FRP confinement has started
contributing in load carrying capacity of column. Failure at top or bottom was
observed for specimen wrapped with GFRP 0° and 45° orientation. The specimens
were wrapped with GFRP at 90° orientation, the failure pattern observed for these
specimens was vertical tearing of the GFRP sheet. Debonding failure was not
observed for 0° and 90° orientation, which revealed that the overlap provided was
sufficient to transfer the stress from concrete to GFRP confinement. However, for
few specimens wrapped with GFRP at 45° orientation, debonding was observed.
As eccentricity increases, abrupt behavior could be seen as the result of inef-
fectiveness of the GFRP confinement. Thus, single layer of GFRP confinement
proved to be insufficient as compared to double wrapping. In case of doubly
wrapped specimens, large deformations were observed without much rupture of
FRP jacket.
4 Conclusions
In order to study the behavior of FRP-confined column, total 45 reinforced concrete

columns confined with GFRP sheets were cast and tested under concentric and
eccentric load. The experimental work was carried out for FRP-wrapped column to
access the effect of change in wrap thickness and ply orientation. From comparison
and investigation between the different eccentricities, a number of conclusions were
drawn:
• GFRP wrapping increased the axial load carrying capacity by providing addition
confinement to the concrete without much increase in the original column size
and weight of the structure.
• Effective confinement with GFRP wrapping resulted in enhancement of the
compressive strength of GFRP-wrapped circular columns by 150% compared
with unwrapped specimens for 0° orientation.
• All unwrapped columns failed under brittle mode with spalling effect.
GFRP-wrapped circular column failed without any sign of debonding.
• Magnitude of eccentricity affects the maximum load carrying capacity of a
confined column under eccentric load. As the eccentricity increases, the load
carrying capacity decreases.
• GFRP-wrapped circular RC columns with 0° ply orientation showed highest
load carrying capacity than 45° and 90° orientations. Load carrying capacity
decreases with increase in the eccentricity. So as a result, 0° gives better con-
finement than 45° and 90° orientation.
References
1. S. Rocca, N. Galati, A. Nanni, Interaction diagram methodology for design of FRP confined
reinforced concrete columns. Constr. Build. Mater. 23(4), 1508–1520 (2009)
2. M.N.S. Hadi, Comparative study of eccentrically loaded FRP wrapped columns. Compos.
Struct. 74(2), 127–135 (2006)
3. M.N.S. Hadi, Behavior of FRP strengthened concrete columns under eccentric compression
loading. Compos. Struct. 77, 92–96 (2007)
4. M.H. Harajli, Axial stress—strain relationship for FRP confined circular and rectangular
concrete columns. Cement Concr. Compos. 28(10), 938–948 (2006)
5. H.J. Lin, Liao CL compressive strength of reinforced concrete column confined by composite
material. Compos. Struct. 65(2), 239–250 (2004)
6. El Maaddawy, Strengthening of eccentrically loaded reinforced concrete columns with
fiber-reinforced polymer wrapping system: experimental investigation and analytical mod-
eling. J. Compos. Constr. 13(1), 13–24 (2009)
7. W. Punurai, C.T.T. Hsu, J. Chen, Biaxially loaded RC slender columns strengthened by
CFRP composite fabrics. Eng. Struct. 46, 311–321 (2012)
8. Alireza Rahai, Hamed Akbarpour, Experimental investigation on rectangular RC columns
strengthened with CFRP composites underaxial load and biaxial bending. Compos. Struct.
108, 538–546 (2014)
9. J. Li, M.N.S. Hadi, Behavior of externally confined high strength concrete columns under
eccentric loading. Compos. Struct. 62(2), 145–153 (2003)
10. Youcef Y, Amziane S, Chemrouk M, The influence of CFRP on the behavior of reinforced
concrete subjected to buckling, in Proceedings of the 4th inter conf on FRP compos eng
(CICE2008) (Zurich, Switzerland, 2008)
Studies on Fracture Features and Fracture
Toughness of CFRP Composites
Fabricated by Resin Infusion Technique
K. Panbarasu, M. Suresh Kumar, V. R. Ranganath and R. V. Prakash
Abstract Fiber reinforced polymer matrix composite laminates are extensively

employed in aerospace and other advanced applications owing to their desirable
properties like high specific stiffness and specific strength. However, full weight
saving potential of these composites has not been completely utilized because of
their susceptibility to delamination failure. The delamination failures are usually
considered as propagation of a crack between two adjacent layers, individually or in
combination of Mode I and II loading conditions. Mode I fracture toughness studies
were conducted on carbon fiber reinforced plastic (CFRP) composites fabricated by
resin infusion technique. A small amount of fiber or other material runs in a
direction perpendicular to the carbon fibers, with the main intention to hold the
primary fibers in position. The unidirectional (UD) fabric employed in the present
study contains 3–4 wt% of glass fibers as warp fibers. This configuration could
influence the fracture toughness of the laminate. Hence, in the present study,
interlaminar fracture studies were conducted under Mode I loading condition on
resin infused UD fabric composites. The fracture surface characteristics were
studied using scanning electron microscope (SEM). The studies reveal that Mode I
fracture toughness initiation value is lower than the toughness values reported for
neat resin and is ascribed to weak interface between fiber and matrix. The Mode I
fracture propagation values are influenced by the use of warp glass fibers employed
for holding the UD fibers intact in resin infused composites. This favors additional
energy absorption mechanisms, such as crack detour, fiber displacement, and
fracture in addition to fiber bridging. These and other aspects of the studies are
discussed in the paper.
Keywords CFRP composites Interlaminar fracture toughness

Resin infusion Fractography
K. Panbarasu M. Suresh Kumar (&) V. R. Ranganath

CSIR-National Aerospace Laboratories, Bangalore, India
e-mail: mskumar@nal.res.in
R. V. Prakash
Department of Mechanical Engineering, Indian Institute Technology, Chennai, India

https://doi.org/10.1007/978-981-10-7197-3_20
226 K. Panbarasu et al.
1 Introduction
Fiber reinforced polymer matrix composite laminates are extensively employed in

aerospace and other advanced applications owing to their desirable properties like
high specific stiffness and specific strength. However, full weight saving potential
of these composites has not been completely utilized because of their susceptibility
to delamination failure. Delaminations can significantly alter the structural prop-
erties of the composites resulting in loss of mechanical strength and stiffness,
especially the compressive strength as well as fatigue life [1]. Further, delamina-
tions can lead to exposure of the fibers to environmental conditions such as
moisture and lead to failure before it can be detected or monitored.
Delaminations can be generated due to tool drop, hailstorm, and runway debris
impacting on the panel surfaces and/or due to the free edge stresses arising around
functional cutouts, rivet/bolt holes, etc. [2]. In general, excessive out-of-plane or
interlaminar stresses could result in local delaminations [3]. Efforts are continuing
to improve the delamination resistance of laminated composite materials [4–9].
The delamination failures are usually considered as propagation of a crack between
two adjacent layers, individually or in combination of Mode I and or Mode II
loading conditions. Some authors have studied interlaminar fracture toughness of
different composite materials with emphasis on various data reduction methods for
calculation of strain energy of different kinds of failure modes [10, 11]. Only a few
authors have addressed this issue through fractography to understand the fracture
process. Thus, Bonhomme [12] studied fracture surfaces of UD AS4/8552 carbon/
epoxy prepreg composite laminate subjected to Mode I and Mode II loading
conditions.
An effort was also made to correlate the fracture surface appearance with the
state of stress. Recently, Greenhalgh presented an extensive study of fractography
on delaminated composites [13]. However, fractographic studies on delamination
failure in UD fabric carbon fiber reinforced plastic (CFRP) composites fabricated
by resin infusion technique are rare in the available literature. The resin infusion
technique employs UD fabric. A small amount of fiber or other material runs in a
direction perpendicular to the carbon fibers, with the main intention of holding the
primary fibers in position. The UD fabric employed in the present study contains
3–4 wt% of glass fibers as warp fibers [14]. This configuration could influence the
fracture toughness of the laminate. Hence, in the present study, interlaminar fracture
studies were conducted under Mode I loading condition on resin infused UD fabric
composites. Standard ASTM recommended procedures were employed for data
reduction and calculating the fracture toughness values [15]. The fracture surfaces
were studied in detail to understand the mechanisms that influence the fracture
toughness values.
Studies on Fracture Features and Fracture Toughness of CFRP … 227
Table 1 Resin properties Tensile strength Elastic modulus Fracture toughness

(MPa) (GPa) (J/m2)
57 3.1 300a
a
Mean of the value arrived at from the results in Ref. [17, 18]
Table 2 Fiber properties Tensile strength (MPa) Elastic modulus (GPa)

4000 236
2 Materials and Methods
The laminates of UD CFRP are fabricated through resin infusion technique with
epoxy resin system. The reinforcement is UD carbon fabric manufactured by M/s
Hexcel composites. The resin system is Epolam 2063 supplied by M/s. Axson,
France. The reinforcement/fiber constitutes 68% by weight. The properties of resin
and fiber provided by the supplier [14, 16] are presented in Tables 1 and 2,
respectively. A total of 26 plies, each of *0.15 mm thick, were used to obtain a
nominal laminate thickness of 4 mm. The dry fabric (reinforcement) is infused with
resin-hardener mixture at 40 °C under vacuum (0.1 MPa). The infused laminates
are heated to 65 °C and dwelled for 3 h. Later, it is heated to 80 °C and dwelled
for 7 h. This is followed by a free-standing post-cure at 160 °C for 4 h. The UD
dry fabric contains bundles of glass fibers (3–4 wt%) in a direction perpendicular to
that of UD carbon fibers in order, as stated earlier on, to hold the primary fibers in
position. A 12.7 µm thick Teflon foil was inserted at the midplane of the panels
during layup process to define a starter delamination/crack for all the test
specimens.
3 Specimen Preparation and Test Conditions
The sample geometries and test conditions used for Mode I double cantilever beam
(DCB) are shown in Fig. 1. The test specimens were prepared from the fabricated
laminates. The nominal dimensions of length *152 mm, width *25 mm, and
thickness *4 mm were employed for each case. The crack length *32 mm was
uniformly maintained in each specimen. Prior to bonding the load-hinges, the
specimens were slightly abraded using 240-grade emery papers. The loading blocks
Fig. 1 DCB sample

geometry and test conditions.
All dimensions are in
millimeter
were bonded with AralditeTM, a two-component room temperature cure adhesive.

All the tests were performed on a computer-controlled test machine having a load
cell maximum capacity of 5 kN. The specimens were mounted on the test fixture
and tests were performed under displacement control at a cross-head displacement
rate of 1 mm/min. At least five specimens were tested. Load (P) versus displace-
ment (d) plots were recorded during the tests. From these plots, PNL [19] was
obtained to calculate Mode I fracture toughness for delamination initiation (GIC ini).
PNL is the load at point of deviation from linearity of the initial P versus d plot. The
plateau values in the R-curve (GIC versus delamination length) were taken as
fracture toughness for delamination propagation (GIC prop) values. The fracture
toughness values were obtained as per ASTM standards D5528-13 via modified
beam theory [15] as per equation.
GI ¼ ð3PdÞ = ½2b ða þ DÞ ð1Þ
where P is the applied load, d is the load point displacement, b is the specimen
width, a is the delamination length, and D is a correction term for the crack length
and determined by generating a least square plot of the cube root of compliance, C1/
3
as a function of the crack length.
Specimens of suitable size for observation under LEO 440i scanning electron
microscope (SEM) were prepared by sectioning the delaminated specimens with a
jeweler’s saw. The debris, formed on the fractured surface during sectioning, was
minimized by cutting the samples before the delamination surfaces were separated.
The generated debris was blown away using compressed air. The delaminated
surfaces were examined in detail under SEM for fractographic characterization. The
fracture surfaces were vacuum sputter coated with gold before examining under
SEM.
Figure 2 shows typical P versus d plot of the composite material tested for Mode I
condition. Significant nonlinear behavior is observed before the maximum load is
reached. The fracture toughness values obtained for delamination initiation, GIC,ini and
propagation, GIC,prop are *190 and 450 J/m2, respectively (Table 3). For the case of
neat resin, a band of 210 to 390 J/m2 has been reported in the literature [17, 18] from
which if an average of 300 J/m2 is considered for discussion, the value of GIC,ini for the
composite considered is low. It may be noted that epoxy resins are brittle and usually
exhibit relatively smaller deformation zone. Therefore, GIC, ini values of the composite
laminate are not expected to be lower than the ones displayed by the neat resin [20].
Moreover, GIC, prop is higher than that of GIC, ini.. The higher value of GIC, prop points
to greater resistance to delamination propagation. The GIC variation against crack
length, a, is shown in Fig. 3. GIC increases monotonically in the first few millimeters
Fig. 2 P versus d plot for

resin infusion composite
Table 3 Mode I fracture Fracture toughness GC ± standard deviation

toughness values (J/m2)
Initiation Propagation
190 ± 11 450 ± 2
Fig. 3 GIC against crack

length, a, for resin infusion
composite
of crack propagation, and then stabilizes with further crack growth. The reasons for
differing aspects of recording lower GIC, ini and higher GIC, prop values are discussed in
detail through fractographic analysis on both crack initiation and propagation regions.
Typical fracture surface of resin infusion composite tested in Mode I loading is
shown in Fig. 4. Detailed examination on the crack origin region revealed matrix
Fig. 4 Fracture surface of delaminated composite samples
cracks near fiber/matrix interface and lesser tendency for fibers to display smearing
of matrix on them indicating a lesser adherence at the interface regions (Fig. 5a).
This lesser adherence factor favors crack initiation and propagation along the fiber/
matrix interface rather than through the matrix. Due to this, the inherent toughness
of matrix could not be fully realized and leads to lower GIC,ini. values of composite
as compared to neat resin toughness.
The fracture surface in crack propagation region exhibited features of fiber
breakages and lifting (Fig. 5b) which are the characteristic marks of fiber bridging
phenomenon [21]. According to a report [22], fiber bridging is one of the factors
responsible for increase in fracture toughness value above GIC,ini. Therefore, fiber
bridging mechanism is a factor paving way for increased GIC,prop.
The weak interface also highlights extensive fiber bridging during crack prop-
agation, resulting in higher GIC,prop as seen in the present case. Further, fracture
surface also reveals an undulated surface (Fig. 6a) in the crack propagation region.
Undulations are due to presence of warp glass fibers used in the fabrication of UD
fabric. These warp glass fibers could favor a detour in the crack path. It has also
been observed that significant amount of resin-rich regions (Fig. 6b) develop in the
resin infused zones. The local resin-rich areas could act as crack detour zones owing
to these regions containing less of the fibers and the associated lower adherence
tendencies discussed earlier on. This situation could also lead to an increase in the
propagation toughness [23]. Additionally, warp fiber pullout together with
Fig. 5 a Presence of cleaner fibers with sparsely adhered matrix; b a photograph illustrating
fracture and lifting of the fibers
displacement (Fig. 6c) and fracturing at a few locations (Fig. 6d) during the test
could also contribute to higher energy absorption thereby resulting in increased GIC,
prop values. Therefore, fiber bridging, crack deflection, encountering resin-rich
region, fiber pullout, fiber fragmentation could all be included as the reasons for
increased delamination propagation toughness in resin infused composites.
Fig. 6 Fracture surface features observed in crack propagation regions: a undulations; b typical
features noticed in some resin-rich region; c pullout with marginal displacement of some warp
fibers; and d fracturing of warp fibers
5 Conclusions
UD CFRP composites fabricated by resin infusion process exhibit lower GIC,ini and
higher GIC,prop as compared to neat resin GIC values cited in literature. Fractographic
study reveals that poor interface bonding between matrix and fiber lead to such
differences. Additionally, the woven glass fiber structure in resin infused composite
triggers various other local energy absorption mechanisms viz., crack deflection, fiber
pullout, fiber fracture, etc. that in turn lead to higher GIC,prop values.
Acknowledgements Authors are deeply acknowledging the staff and Head of Departments of
Advanced Composite Division, Material Sciences Division and Structural Technologies Divisions
of CSIR-NAL for their direct and indirect support rendered toward this work.
References
1. K. Singh, B. Dattaguru, T. Ramamurthy, P. Mangalgiri, Delamination tolerance studies in

laminated composite panels. Sadhana 25, 409–422N (2000)
2. N.S. Choi, A. Kinloch, J. Williams, Delamination fracture of multidirectional carbon-fiber/
epoxy composites under mode I, mode II and mixed-mode I/II loading. J. Compos. Mater. 33,
73–100 (1999)
3. T. Nyman, Fatigue and residual strength of composite aircraft structures. Ph.D. Thesis,
Institutionen för flygteknik (1999)
4. N. Sela, O. Ishai, Interlaminar fracture toughness and toughening of laminated composite
materials: a review. Composites 20, 423–435 (1989)
5. S. Sankaran, M. Chanda, Chemical toughening of epoxies-II. Mechanical, thermal, and
microscopic studies of epoxies toughened with hydroxyl-terminated poly
(butadiene-co-acrylonitrile). J. Appl. Poly. Sci. 39, 1635–1647 (1990)
6. Kishore, S.M. Kulkarni, D. Sunil, S. Sharathchandra, Effect of surface treatment on the impact
behaviour of fly-ash filled polymer composites. Poly. Inter. 51, 1378–1384 (2002)
7. Y. Hirai, H. Hamada, J.K. Kim, Impact response of woven glass-fabric composites—I.: Effect
of fiber surface treatment. Compos. Sci. Tech. 58, 91–104 (1998)
8. A. Peijs, R. Venderbosch, P. Lemstra, Hybrid composites based on polyethylene and carbon
fibers Part 3: Impact resistant structural composites through damage management. Composites
21, 522–530 (1990)
9. C. Marshall, International encyclopedia of composite materials (VCH Publishers, NewYork,
1990)
10. S.P. Blake, K.A. Berube, R.A. Lopez-Anido, Interlaminar fracture toughness of woven
E-glass fabric composites. J. Compos. Mater. 46(13), 1583–1592 (2012)
11. M.S. Prasad, C. Venkatesha, T. Jayaraju, Experimental methods of determining fracture
toughness of fiber reinforced polymer composites under various loading conditions. J. Miner.
Mater. Charact. Eng. 10(13), 1263–1275 (2011)
12. J. Bonhomme, A. Argüelles, J. Vina, I. Vina, Fractography and failure mechanisms in static
mode I and mode II delamination testing of unidirectional carbon reinforced composites.
Polym Test. 28, 612–617 (2009)
13. E.S. Greenhalgh, Failure analysis and fractography of polymer composites (CRC Presss,
Washington, DC, 2009)
14. Hexcel composites, Aerospace selector guide, USA (2010)
15. ASTM D5528–13A, Standard test method for mode I interlaminar fracture toughness of
unidirectional fiber-reinforced polymer matrix composites
16. Epolam 2063 data sheet, Axson, France
17. G. Quino, J. El Yagoubi, and G. Lubineau, Characterizing the toughness of an epoxy resin
after wet aging using compact tension specimens with non-uniform moisture content. Polym.
Degra. and Stab. 109, 319–326 (2014)
18. Q.Q. Gustavo, Characterization of fracture toughness of epoxy resin after hygrothermal aging.
M.Sc. Thesis, King Abdullah University of Science and Technology (2013)
19. K.Y. Kim, L. Ye, K.M. Phoa, Interlaminar fracture toughness of CF/PEI and GF/PEI
composites at elevated temperatures. Appl. Compos. Mater. 1, 173–190 (2004)
20. W.L. Bradley, R.N. Cohen, Matrix deformation and fracture in graphite-reinforced epoxies.
ASTM STP 876, 389–410 (1985)
21. I. Chou, I. Kimpara, K. Kageyama, I. Ohsawa, Mode I and mode II fracture toughness
measured between differently oriented plies in graphite/epoxy composites. ASTM STP 1230,
132–151 (1995)
22. P. Compston, P.Y. Jar, P.J. Burchill, K. Takahashi, The transfer of matrix toughness to
composite mode I interlaminar fracture toughness in glass-fibre/vinyl ester composites. Appl.
Compos. Mater. 9(5), 291–314 (2002)
23. A.J. Brunner, B.R.K. Blackman, P. Davies, Mode I delamination. In: D.R. Moore, A. Pavan,
J.G. Williams (Eds.), Fracture Mechanics Testing Methods for Polymers, Adhesives, and
Composites (Elsevier Science Ltd, 2001), pp 277–305
Numerical and Experimental Analysis
of Double-Sided Stepped Lap-Repaired
CFRP Laminates Under Tensile Loading
Matta Seshadri and M. Ramji
Abstract Adhesive layer plays a critical role in the strength restoration of the scarf
repaired carbon fibre-reinforced plastic (CFRP) laminates. In this work, Araldite
2015 is used. Hence, it is crucial to model the behaviour of adhesive layer accu-
rately in case of numerical model. Modelling of adhesive layer by cohesive zone
law characterises the fracture behaviour of the bonded joint accurately. In this
paper, cohesive zone law parameters for mode I and mode II are determined by
comparing numerical predictions to experimental observations of a double can-
tilever beam (DCB) for mode I and end notched flexure (ENF) for mode II fracture
test. In this work, Araldite 2015 (supplied by Huntsman) is used for repair work.
Strain energy release rate for both mode I and mode II is determined by performing
DCB and ENF test, respectively. Traction–separation law for mode I is generated
by direct method which involves differentiation of the relation between the strain
energy release rate and crack tip opening displacement which is measured using
digital image correlation (DIC) technique. Traction–separation law for mode II is
generated by inverse method which involves fitting the numerical and experimental
load–displacement curves. The obtained cohesive law is used to model the adhesive
layer in numerical analysis of double-sided stepped lap joint repair of CFRP lam-
inate subjected to tensile loading. The numerical predictions are validated by
comparing the load–displacement curve obtained from the experimental study.
A good agreement exists between numerical and experimental results confirming
that the proposed cohesive law for mode I and mode II can be applied to model
adhesive layer with CFRP as adherend.

Keywords Cohesive zone law Strain energy release rate Digital image

correlation Double-sided stepped lap joint
M. Seshadri M. Ramji (&)

Engineering Optics Lab, Department of Mechanical and Aerospace Engineering,
IIT Hyderabad, Hyderabad 502205, India
e-mail: ramji_mano@iith.ac.in

https://doi.org/10.1007/978-981-10-7197-3_21
236 M. Seshadri and M. Ramji
List of symbols
a Crack length
ae Equivalent crack length
a0 Initial crack length
B Width of the specimen
C Compliance of the specimen
C0 Initial compliance
Ef Flexural modulus
GI Mode I strain energy release rate
GIC Mode I fracture toughness
GII Mode II strain energy release rate
GIIC Mode II fracture toughness
h Height of single adherend
L Total specimen length
P Applied load on the specimen
w Crack tip opening displacement (CTOD)
d Displacement of the specimen
d0 Damage initiation relative displacement
dc Ultimate relative displacement
D Correction factor for root rotation effect
ta Adhesive thickness
r Normal traction
1 Introduction
In adhesive bonding, the transfer of load between panel and patch is through the
adhesive layer and it constitutes as the weakest link in repaired configuration and
failure initiates in the adhesive layer. Hence, adhesive layer plays a precarious role
in the strength restoration of the scarf repaired carbon fibre-reinforced plastic
(CFRP) laminates. It is crucial to model the behaviour of adhesive layer accurately
in case of numerical model. The three approaches based on fundamental mechanics
applied in finite element modelling of the adhesively layer of bonded joints are
fracture mechanics, continuum mechanics, and damage mechanics. Continuum
mechanics approach is based on stress and strain analysis. It is challenging to use
strain- and stress-based criteria as singularity arises in the bonded joints.
Energy-based parameter (joint toughness) is used in linear elastic fracture
mechanics (LEFM) approach. Toughness values in opening, sliding, and mixed
mode loading conditions are estimated to predict the strength of the adhesive joint.
Linear elastic fracture mechanics is based on linear elastic behaviour of adhesive
and the presence of initial flaws. In many cases, it can be challenging to locate flaw
a prior in the composite structural component and also most of the recent adhesives
Numerical and Experimental Analysis of Double-Sided … 237
have ductile behaviour, which is the reason why LEFM approach is not suitable for
practical applications. A thorough review of various analytical models on behaviour
of adhesively bonded joints related to both single and double-lap joint configuration
is presented in [1, 2]. In order to overcome the above difficulties, both the
stress-based and the energy-based criteria are combined to model the adhesively
bonded joints. Modelling of adhesive layer by cohesive zone law characterises the
fracture behaviour of the bonded joint accurately. It combines both stress-based
criteria and energy-based fracture mechanics criteria to simulate damage initiation
and its propagation, respectively. A comprehensive review of finite element
methods applied to analyse the adhesively bonded joints is presented in [3].
Advanced techniques in finite element modelling of adhesive layer are described in
[4].
To obtain cohesive law of CFRP composite bonded joints, initially, strain energy
release rate is measured using experimental techniques. The estimation of strain
energy release rate involves measurement of crack tip progression throughout the
test. It is difficult to identify the crack tip during the test which results in
non-negligible errors in the estimation of strain energy release rate. A data reduction
technique based on specimen compliance and beam theory is proposed in [5]. In
this method, the measurement of strain energy release rate is not dependent on crack
tip progression during the test, but depends wholly on the specimen compliance,
which results in accurate measurement. Later, an inverse approach based on curve
fitting involving curve fitting of experimental load displacement data with
numerical prediction is carried to obtain the cohesive law [6]. Alternatively, in
direct methods, strain energy release rate (G) is defined in terms of crack tip
opening displacement (CTOD) based on the experimental behaviour [7]. The
cohesive damage law is further obtained by differentiating the relation
G = f (CTOD).
In the present work, the cohesive law for the adhesive bonding of the CFRP
composite joints under Mode I and Mode II loading conditions is determined, and
the obtained cohesive law is used to describe the adhesive layer behaviour in
numerical model of the double-sided stepped lap-repaired CFRP laminate subjected
to tensile loading. The numerical predictions are validated by comparing the load–
displacement curve obtained from the experimental study.
2 Cohesive Damage Model
2.1 Pure Mode I Loading Condition
2.1.1 Estimation by Direct Method
The strain energy release rate of CFRP composite bonded joint under mode I (GI)
loading condition is determined using equivalent crack method [7]. In this method,
specimen compliance is utilised. Based on Timoshenko beam theory, the specimen

compliance is given by
8a3 12a
C¼ þ ð1Þ
E1 Bh3 5BhG13
An equivalent flexural modulus is obtained from Eq. (1), which includes the
effect of adhesive presence, stress concentrations, and material variability [8]. Initial
crack length and compliance are considered to estimate Ef, which is given in the
Eq. (2).

12ða0 þ hDÞ 1 8ða0 þ hDÞ3
Ef ¼ C0 ð2Þ
5BhG13 Bh3
From Irwin–Kies equation, we have
P2 dC
GI ¼ ð3Þ
2B da
By substituting the value of compliance (C) from (1) in the above equation, we
get

6P2 2a2 1
GI ¼ 2 þ ð4Þ
B h Ef h2 5G13
From the above equation, to evaluate GI experimentally crack length monitoring

is required which is not easy because it is not visible clearly to identify the crack
tip. Instead, equivalent crack length can be evaluated by the current compliance
using the relation ae = f(C) [6]. Following this procedure is advantageous as
equivalent crack length, and GI are obtained directly from load–displacement curve.
To determine the cohesive damage law for adhesively bonded CFRP composite
joint using direct method, the following relation between GI and (CTOD) is very
essential.
Zw
GI ¼ ðrÞdw ð5Þ
o
The differentiation of strain energy release rate GI with respect to the CTOD
leads to cohesive law
r ¼ f ðwÞ ð6Þ
dGI
rðwÞ ¼ ð7Þ
dw
Fig. 1 Schematic of the CTOD measurement location in the DCB specimen
To obtain the CTOD in this work, DIC technique is utilised. The sensitivity
analysis of digital image correlation technique is described in [9]. The CTOD is
defined as the relative displacement between cracked face, measured in the normal
to the crack plane, at the initial crack tip. The CTOD is estimated from the dis-
placement field obtained from DIC, by considering displacements at the crack tip
which are perpendicular to the crack surface. The schematic of the CTOD mea-
surement location in the DCB specimen is shown in Fig. 1.
2.1.2 Experimental Test
The adherends of the DCB specimen are fabricated using carbon fibre supplied by
Hindustan Technical Fabrics Ltd. India, weighing 230 gsm. The layup sequence is
[00]8. The matrix is made from epoxy resin (CY 230) and hardener (HY 951)
supplied by Huntsman in 10:1 weight ratio. The weight ratio of fibre and matrix is
1:1. The CFRP composite adherends are fabricated in-house by vacuum bagging
technique under room temperature with a curing time of 24 h at room temperature.
The specimen dimensions are Lc = 143 mm, L = 130 mm, B = 25 mm,
a0 = 50 mm, tp = 2 mm, ta = 0.2 mm. An epoxy adhesive Araldite 2015 from
Huntsman is used to bond the CFRP composite adherends. Initially, the bonding
surfaces are roughened using emery paper and cleaned with acetone to enhance the
adhesion and avoid the adhesive failure. The adhesive bond is cured at room
temperature for a period of twenty-four hours, and later, piano hinges are bonded as
shown in Fig. 2. The initial crack is introduced by means of Teflon tape of
thickness 40 lm. Both the CFRP adherend surfaces are applied with adhesive.
A Teflon tape is placed on the adhesive layer of one of the adherend, and both the
CFRP panels are joined and held in a fixture which applies equal pressure on the
joint so as to maintain a constant thickness of adhesive. Teflon tape is placed in
between the two adherends after applying the adhesive. Two specimens are fabri-
cated and tested at room temperature.
The experimental set-up shown in Fig. 3 consists of computer controlled,
electric motor operated Instron mechanical testing machine of 2 kN capacity.
2 mm/min displacement rate is applied, and the load–displacement (P–d) curve is
captured from the test. DIC system supplied by Correlated Solution Inc. is utilised
to grab images during the test procedure. On the front edge surface of the DCB
Fig. 2 Geometry of the DCB test specimen
Fig. 3 Experimental set-up of DCB specimen test
specimen, a random speckle pattern is generated to perform DIC post-processing.

Initially, the surface is sprayed with an acrylic white spray paint, and later, random
speckle pattern is generated with carbon black colour using air gun.
To capture the speckle images, a CCD Camera manufactured by Grasshopper®
(POINTGREY-GRAS-50S5M-C) with a spatial resolution of 2448 2048 pixels
is used. Tamron lens (with SP AF 180 mm f/3.5 Di) is mounted on the CCD
camera to capture the entire length of the specimen. Two white LED lamps of 30 W
capacity are used to illuminate the surface of the specimen. Five images per second
are captured and saved on a computer with Vic-Snap software from Correlated
Solution Inc.
The P–d curve generated from the Instron machine test data and GI–d curve
evaluated using equivalent crack length method are shown in Fig. 4a, b,
respectively. The obtained GIC value is closed to 0.45 N/mm which compares very
well with the average GIC mentioned in Ref. [5].
In order to get the cohesive law, a sixth order polynomial cure was fitted
between GI and CTOD which is shown in Fig. 5a. The polynomial is differentiated
with respect to CTOD to obtain the relation r = f(w), which in turn gives the
cohesive law of adhesively bonded CFRP composite joints under mode I loading
condition. A bilinear cohesive law is adjusted over the traction–separation curve as
shown in Fig. 5b. This bilinear law is then applied to model the adhesive layer in
the finite element method (FEM) to validate the procedure. And the same procedure
is repeated for mode II characterisation of adhesive layer in subsequent section.
2.1.3 Numerical Simulations
To validate the above method, numerical simulations are performed.

A two-dimensional eight-nodded plane strain elements are selected to represent the
adherends. The adhesive layer is modelled as 2D contact elements at the
mid-thickness of the specimen. The lamina mechanical properties of the CFRP
Fig. 4 Mechanical behaviour of DCB specimen. a P–d curve; b strain energy release rate (GI)–d
curve
Fig. 5 Traction-separation behaviour. a GI versus w curve; b experimental and adjusted cohesive

law
Table 1 CFRP laminates Properties Value

and patch ply mechanical
properties Longitudinal modulus, E11 (GPa) 82.64
Transverse modulus, E22 = E33 (GPa) 7.12
In-plane Shear modulus, G12 = G13 (GPa) 3.3
Out-of-plane Shear modulus, G23 (GPa) 2.47
In-plane Poisson’s ratio, m12 = m13 0.31
Out-of-plane Poisson’s ratio, m23 0.43
Fig. 6 P–d curve for

comparison of experimental
and numerical solution
composite are presented in Table 1. Damage initiation and progression are pre-
dicted using a pure mode I bilinear cohesive law. The P–d curve is generated, and
they are compared with the experimental one as shown in Fig. 6. There is a close
match between experiment and FEM result thereby confirming the accuracy of the
traction–separation law which has been arrived upon.
2.2 Pure Mode II Loading Condition
2.2.1 Estimation by Compliance-Based Beam Method
To estimate GII value of CFRP composite bonded joint under mode II loading
condition, compliance-based beam method (CBBM) is followed. Three different
data reduction techniques to estimate the strain energy release rate under mode II
loading condition are mentioned in [10]. The drawback in these methods is that in
accurate measurement of crack length during propagation. Also, modern adhesives
have ductile behaviour which results in the development of fracture process zone
(FPZ) ahead of the crack tip. This leads to a non-negligible amount of energy
dissipation, which cannot be accounted for in the standard data reduction tech-
niques. In order to overcome these difficulties, CBBM utilises the crack equivalent
concept which is experimentally determined based on the specimen compliance.
According to the beam theory, ENF specimen compliance can be written as
3a3 þ 2L3 3L
C¼ þ ð8Þ
8E1 Bh3 10G13 Bh
Using initial specimen compliance as well as the initial crack length, the flexural
modulus of the specimen is defined as
3a30 þ 2L3
Ef ¼ ð9Þ
8Bh3 C0corr
where C0corr is given by
3L
C0corr ¼ C0 ð10Þ
10G13 Bh
For evaluating the equivalent crack length, a correction of real crack length is
measured to account for the FPZ influence. The equivalent crack length is given by
1=3
Ccorr 3 2 Ccorr
ae ¼ a þ DaFPZ ¼ a þ 1 L3 ð11Þ
C0corr 0 3 C0corr
where Ccorr is given by (10), using C instead of C0. Using Irwin–Kies relation,
strain energy release rate under mode II (GII) is given by
9P2 a2e
GIIC ¼ ð12Þ
16B2 Ef h3
2.2.2 Experimental Analysis
The materials and fabrication process for the adherends of the ENF specimen are
similar to Sect. 2.1.2. The measurements of the specimen are L = 100 mm,
Ls = 50 mm, B = 25 mm, ta = 0.2 mm, a0 = 30 mm, and h = 2 mm as shown in
Fig. 7. The experimental set-up shown in Fig. 8 consists of a computer-controlled
MTS Landmark® servo-hydraulic cyclic test machine of 100 kN capacity.
A three-point bend fixture with radius of the support rollers equal to 25 mm is used.
The load–displacement curve generated from the MTS machine test data for two
specimens is shown in Fig. 9. The strain energy release rate under mode II, evaluated
by CBBM is given in Table 2, and its average value is found to be 3.14 N/mm.
Fig. 7 Schematic presentation of the ENF test
Fig. 8 Experimental set-up of ENF specimen test
Fig. 9 Load–displacement curve of ENF test

Table 2 Fracture energies in Specimen GIIC (N/mm)

pure mode II by CBBM
1 3.21
2 3.54
3 2.69
Average 3.14
2.2.3 Numerical Simulations
The cohesive law for pure mode II loading condition is determined by inverse
method, which involves fitting the experimental and numerical load–displacement
curves. A two-dimensional eight-nodded plane strain elements are selected to
model the adherends. The adhesive layer is modelled as 2D contact elements at the
mid-thickness of the specimen. The lamina elastic properties of the CFRP com-
posite are presented in Table 1. Damage initiation and progression are predicted
using a pure mode II bilinear cohesive law. The parameters required to define the
cohesive law for pure mode II are maximum tangential traction, tangential dis-
placement jump at the completion of debonding, and critical fracture energy density
for tangential slip. The cohesive parameter selection criteria are explained in [11].
The critical fracture energy density value is acquired experimentally. The traction–
separation law before damage initiation is given by
r ¼ Kdr ð13Þ
where K is the stiffness defined as the ratio of elastic modulus of the material in
tension or shear to the adhesive thickness. Assuming maximum tangential traction
(rIIU) equal to the shear strength of the adhesive material, the value of tangential
displacement jump at the completion of debonding (dU) can be estimated from the
following relationship
rIIU dU
GIIC ¼ ð14Þ
2
Figure 10 shows the numerical and experimental load–deformation curve for

single tested specimen after the fitting procedure. From the plot, one could ascertain
there is a very good coherence between experiment and FEA.
3 Double-Sided Stepped Lap Joint Repair
3.1 Numerical Analysis
A numerical analysis of double-sided stepped lap joint repair of adhesively bonded

CFRP adherends subjected to tensile loading is performed. The cohesive law is
Fig. 10 P–d curve for

comparison of experimental
and numerical solution
defined for modelling the adhesive layer, which was obtained from the
above-prescribed methods. The CFRP laminate is modelled with a layup sequence
of [00]12. The geometry of the model is shown in Fig. 11. The dimensions are as
follows: length of the laminate a = 225 mm, width b = 45 mm, thickness of the
laminate t = 2.4 mm, scarf angle a = 2°, and step length = 13 mm. The elastic
properties of the laminate are mentioned in Table 1. A two-dimensional
eight-nodded plane strain elements are assigned to the laminates, and
two-dimensional interface elements are assigned to the adhesive layer. A total of
three hundred interface elements are meshed along the bond length in order to have
a refined mesh for the adhesive layer. The application of interface elements in
modelling the adhesive layer, for progressive analysis of adhesive bond, is pre-
scribed in [12]. The meshed model of the double-sided stepped lap joint repair
including the boundary conditions is represented in Fig. 12. The geometry of the
model is modelled according to the Ref. [13].
Fig. 11 Geometry of double-sided stepped lap joint repair

Fig. 12 FE model of double stepped lap repair
3.2 Experimental Work
To validate the numerical solution of double-sided stepped lap joint repair of CFRP
laminate subjected to tensile loading, experimental work is carried out. The
dimensions of the specimen are specified in Sect. 3.1. The CFRP laminates are
fabricated using carbon fibre manufactured by Hindustan Technical Fabrics Ltd.
India, weighing 200 gsm. The layup sequence is [0°]12. The matrix is made up of
epoxy resin (CY 230) mixed in the weight ratio of 10:1 with hardener (HY 951)
supplied by Huntsman. The weight ratio of fibre and matrix is maintained as 1:1.
The CFRP composite laminates are fabricated in-house by vacuum bagging tech-
nique under room temperature with curing time set for 24 h.
The laminate and the patch are cut from the same casting. It is presumed that the
laminate suffered damage at an intermediate section, and repair is carried out.
Following, the damaged region is removed by diamond-coated milling cutter with a
stepped shape as shown in the Fig. 10. Patch with the corresponding shape of the
removed material is adhesively bonded over the damaged area of panel using an
Araldite 2015 (Huntsman) adhesive of thickness 0.2 mm, and then, it is cured at
room temperature under vacuum. Bevelled aluminium tabs of dimension
50 mm 45 mm 2 mm are adhesively bonded at each end of the specimen for
gripping purpose. The testing of the three specimens is performed on a
computer-controlled MTS Landmark® servo-hydraulic cyclic test machine of
100 kN capacity. The experimental set-up for the testing of double stepped lap
repaired CFRP panel is shown in Fig. 13. The P–d curve generated from MTS data
is shown in Fig. 14a. From the graph, we can infer that initially, the patch gets
detached from the panel with failure of adhesive layer. Hence, there is drop in the
load at point 1. This can be depicted from the Fig. 15a where the patch gets
detached from the panel as a result of adhesive layer failure. But the load starts
increasing because the undamaged part of the panel (unmachined part–bottom
layers) starts to take the load. Finally, there is a drop in the load at point 2, as the
panel fails which is the ultimate failure of the specimen. The final damaged
specimen is shown in Fig. 15b, which characterise the existence of several inter-
acting failure modes like matrix cracking, longitudinal splitting of fibres, fibre pull
out, and delamination. Figure 14b displays the comparison between the load–dis-
placement curves of experimental test and numerical simulation up to the failure
point of the adhesive layer. The post-adhesive layer failure behaviour is not sim-
ulated numerically as progressive damage model is not implemented at this
moment.
Fig. 13 Experimental set-up of DSSLJ repaired CFRP panel test
Fig. 14 a Experimental load–displacement curve, b comparison of experimental and numerical

P–d curve
Fig. 15 Failure surface of double-sided stepped lap repaired CFRP panel. a Surface with the joint,
b surface without the joint
4 Conclusion
The objective of this work is to estimate the cohesive damage law of adhesive layer
(Araldite 2015) used in CFRP laminate repair under mode I and mode II loading
conditions. For mode I and mode II strain energy release rate, experimental test of
DCB and ENF, respectively, is performed. To obtain the cohesive law for mode I and
mode II, direct method and inverse method are employed, respectively. Numerical
simulation is carried out to validate the obtained cohesive law for adhesive layer by
modelling it as interface elements. For mode I cohesive law, the maximum normal
traction value is 3.94 MPa and critical fracture energy is 0.45 N/mm. For mode II
cohesive law, the maximum tangential traction value is 25 MPa and critical fracture
energy is 3 N/mm.
The obtained cohesive law for mode I and mode II is used to model the adhesive
layer in numerical analysis of double-sided stepped lap joint repair of CFRP lam-
inate subjected to tensile loading. The validation of the above-described procedure
is evaluated by comparing numerical and experiments results. The initial stiffness,
the maximum load until patch debonding, and the respective displacement are
the parameters for comparison. Looking at the load displacement curve, a good
coherence exists between experimental and numerical values corresponding to the
above-mentioned parameters until adhesive layer fail. It is concluded that the
procedure presented in this work can be effectively applied to predict the cohesive
law for mode I and mode II loading conditions and can be used to model adhesive
layer bonded with CFRP panel as adherend.
Acknowledgements The first author thank Dr. Viswanath R Chintapenta, Assistant Professor,
Indian Institute of Technology Hyderabad for his guidance, lab mates in engineering optics lab and
central workshop staff at Indian Institute of Technology Hyderabad.
References
1. M.D. Banea, L.F.M. da Silva, Adhesively bonded joints in composite materials: an overview.
Proc. Inst. Mech. Eng. Part L: J. Mater. Design Appl. 223 (2009)
2. L.F.M. da Silva, P.J.C. das Neves, R.D. Adams, J.K. Spelt, Analytical models of adhesively
bonded joints—Part I: Literature survey. Int. J. Adhes. Adhes. 29, 319–330 (2009)
3. X. He, A review of finite element analysis of adhesively bonded joints. Int. J. Adhes. Adhes.
31, 248–264 (2011)
4. L.F.M. da Silva, R.D.S.G. Campilho, Advances in Numerical Modeling of Adhesive Joints
(Springer, Berlin, 2012)
5. M.F.S.F. de Moura, J.P.M. Gonçalves, A.G. Magalhães, A straightforward method to obtain
the cohesive laws of bonded joints under mode I loading. Int. J. Adhes. Adhes. 39, 54–59
(2012)
6. M.F.S.F. de Moura, J.J.L. Morais, N. Dourado, A new data reduction scheme for mode I
wood fracture characterization using the double cantilever beam test. Eng. Fract. Mech. 75,
3852–3865 (2008)
7. M.F.S.F. de Moura, R.D.S.G. Campilho, J.P.M. Gonçalves, Crack equivalent concept applied
to the fracture characterization of bonded joints under pure mode I loading. Compos. Sci.
Technol. 68, 2224–2230 (2008)
8. G.F. Dias, M.F.S.F. de Moura, J.A.G. Chousal, J. Xavier, Cohesive laws of composite bonded
joints under mode I loading. J. Compos. Struct. 106, 646–652 (2013)
9. M. Kashfuddoja, R.G.R. Prasath, M. Ramji, Study on experimental characterization of carbon
fibre reinforced polymer panel using digital image correlation: a sensitivity analysis. J. Opt.
Lasers Eng. 62, 17–30 (2014)
10. M.F.S.F. de Moura, R.D.S.G. Campilho, J.P.M. Gonçalves, Pure mode II fracture
characterization of composite bonded joints. Int. J. Solids Struct. 46, 1589–1595 (2009)
11. K. Song, C.G. Dávila, C.A. Rose, Guidelines and parameter selection for the simulation of
progressive delamination, in 2008 Abaqus Users’ Conference
12. J.P.M. Gonçalves, M.F.S.F. de Moura, A.G. Magalhães, P.M.S.T. De Castro, Application of
interface finite elements to the three-dimensional progressive failure analysis of adhesive
joints. Fatigue Fract. Eng. Mater. Struct. 26, 479–486
13. D. Li, G. Qing, Y. Liu, A three-dimensional semi-analytical model for the composite
laminated plates with a stepped lap repair. Compos. Struct. 93, 1673–1682 (2011)
Part IV
Computational Mechanics
Lower and Upper Bound Estimates
of Material Properties of Pristine
Graphene: Using Quantum Espresso
T. Chaitanya Sagar and Viswanath Chinthapenta
Abstract In this paper, the basic groundstate (T = 0) properties of pristine gra-

phene are calculated using first principles with the aid of Quantum Espresso (QE).
QE software suite is a tool based on ab initio quantum chemistry methods to obtain
the electronic structure for materials modeling. It is an open source package built
basing on the formalism of density functional theory (DFT). Using QE, the band
structure, cohesive energy, and second-order elastic constants are estimated for a
pristine graphene. Upper bound estimates based on the generalized gradient
approximation (GGA) and lower bound estimates based on local density approxi-
mation (LDA) are obtained. The cohesive energy is found to be −7.917 eV/atom
using LDA and −5.673 eV/atom using GGA. Further, the elastic properties are
determined using a post-processing tool ElaStic. The second-order elastic stiffness
C11 is found to be 491.5, and 506.7 GPa using LDA and GGA approaches,
respectively.

Keywords Graphene DFT Quantum Espresso Band structure
Cohesive energy and second-order elastic constants
Nomenclature
DFT Density Functional Theory
GGA Generalized Gradient Approximation
ICME Integrated Computational Materials Engineering
LDA Local Density Approximation
MD Molecular Dynamics
PP Pseudo Potentials
QE Quantum Espresso
SOEC Second-Order Elastic Constants
T. Chaitanya Sagar V. Chinthapenta (&)

Micro-Mechanics Lab, Indian Institute of Technology Hyderabad, Kandi, Sangareddy
502285, Telangana, India
e-mail: viswanath@iith.ac.in

https://doi.org/10.1007/978-981-10-7197-3_22
254 T. Chaitanya Sagar and V. Chinthapenta
1 Introduction
Conventional design of materials at the component level requires experiments to be

conducted at the components scale without taking information from the entities at
the underneath length and timescales. In this methodology, the choice of design is
limited by the approximation that the material is homogeneous. This approximation
is manifested by an arbitrarily large choice of a factor of safety. The design of
materials at multiple scales accounts for the uncertainties by physically modeling
the inhomogeneities at appropriate length scales and scale bridging their effect to
the higher scales. The concept of design of materials at multiple scales had gained a
lot of importance in the recent times [1]. The multiscaling approach implemented is
not limited to computations; it is also widely used in experiments, modeling, and
design.
Integrated Computational Materials Engineering (ICME) [2] and predictive
modeling of materials is based on multiscale modeling [3]. Wherein, the material
properties are calculated with limited information from the experimental inputs [4].
Although experiments play a crucial role, the contributions of simulations per-
formed at a lower length scales provide fundamental insights such as existence/
feasibility of certain class of materials. And with the availability of high-speed
computers and simulating tools, exploring new materials has become attainable.
Moreover, validation of experiments with simulations has become a mandate.
Hence, multiscale modeling opened a new window for computational materials
science.
ICME has gained a lot of importance in the current times, and several
researchers are working in this domain [5, 6]. For example, Horstemeyer [7] pro-
vides a detailed overview of a case study on the control arm of an automobile
(Cadillac). In this study, the mechanical properties of the higher length scale were
obtained from mesoscale simulations, mesoscale properties are obtained from the
atomic-scale simulations, and atomic properties are obtained from electronic scale
simulations. Different mathematical models are used at various length scales, such
as FEM at continuum scale, crystal plasticity or discrete dislocations at mesoscale,
molecular dynamics at atomic-scale, and DFT at electronic scale [7].
Materials can be perceived at four different levels depending on the length scales
and timescales involved. These length scales range from a few angstroms to a
micron to macroscopic lengths, and the timescale ranges from femtoseconds to a
few seconds to years. In the top-down approach, as shown in Fig. 1, they can be
classified into:
1. Continuum/Bulk scale
2. Mesoscale
3. Molecular/Atomistic scale
4. Quantum/Electronic/ab initio scale
The smallest length scale of about a few angstroms corresponds to the electronic
scale. At electronic scale, the interaction of the electrons among themselves as well
Lower and Upper Bound Estimates of Material Properties … 255
Fig. 1 Multiple scales of length and time in materials modeling
as with the nuclei is vital. Here, the total energy of the system is derived by solving
computationally intensive Schrödinger equation of quantum mechanics. Hence,
with the current computational architecture, systems with only a few tens of elec-
trons can be simulated. While, with the usage of density functional theory (DFT),
which can describe the properties at the ground state of a quantum mechanical
system with the knowledge of the density of electrons in real space, it is possible to
study systems with at least few thousands of electrons. The results from DFT
calculations can provide accurate information about properties of crystalline
materials. The objectives of the current study are to estimate upper and lower
bounds to the electrical and also mechanical properties of pristine graphene such as
electronic band structure, cohesive energy, and SOEC.
As shown in Fig. 2a, b, graphene is an allotrope of carbon having a hexagonal
honeycomb structure which is one atom thick with a 0.142 nm length for the
carbon–carbon bond [9]. As carbon is its source, it is abundantly available (over
95% all chemical compounds are carbon compounds). Graphene existing in three
Fig. 2 a Allotropic forms of carbon depicting graphene in hexagonal form [8]; b graphene
monolayer arm chair along the vertical direction
forms namely 3D-fullerenes, 2D-nanotubes, and 0D-buckyball [10] is known to

show magnificent electronic [11] thermal [12–16], surface optical, and mechanical
properties [17–19]. It is widely used in applications like nanoelectronics, catalysis,
gas sorption, supercapacitors, and photovoltaics [20].
The manufacturability of pure graphene at an industrial scale is still a challenge
compared to that of laboratory production [21]. There is tremendous variability in
the values of electrical and mechanical values reported in the literature [22–27]. The
factors mentioned above motivated to carry out the upper and lower bound esti-
mates of the electrical and mechanical properties for the pristine graphene.
The paper is outlined as follows: In Sect. 2, a brief methodology of DFT-based
QE is presented, followed by a procedure for generating SOEC using the ElaStic
tool. In Sect. 3, the structure of graphene analyzed in the current study is presented.
Using methodology listed in Sect. 2, the material properties of pristine graphene are
estimated and are shown in Sect. 4. Section 4.1 illustrates the electronic properties
such as Band structure, and the cohesive energies obtained using LDA and GGA
approach. Similarly, Sect. 4.2 presents an upper and lower bound estimation of
SOEC using LDA and GGA, determined using the ElaStic tool.
2 Methodology
2.1 Quantum Espresso Suite
Quantum Espresso suite is used to perform ground state calculations of total energy
using and to find out the material properties of various systems. With the use of QE,
a variety of models is studied to compute the behavior of materials from nanoscale
and upwards. The atomic number and the pseudopotentials (PPs) for each con-
stituent element are taken as inputs to a system. Further, QE also requires the
structural input to compute the materials properties of the system. The structural
input consists of details of the unit cell. The unit cell on repeating it in all direction
would yield the macroscopic system to be simulated. In the lower dimension
systems, the unit cell is filled with the vacuum spaces such as in a film or wire or a
quantum dot. QE also supports the periodic boundary conditions which scale in all
the three directions to model the infinitely large crystalline systems.
QE uses plane wave (PW) basis set to express the Kohn–Sham orbitals, while
the pseudopotentials represent the interactions of the ions with only the valence
electrons. These PPs are classified into ultrasoft (US) PPs, separable
norm-conserving (NC) PPs, or by projector augmented wave (PAW) based on the
description of atomic cores. QE is based on DFT, wherein the intermolecular
interactions are taken care using the exchange and correlation functional.
Generalized gradient approximation (GGA) [28] and local density approximation
(LDA) [29, 30] to the exchange-correlation functional can be applied using the
pseudopotentials.
Fig. 3 DFT algorithm for implemented [31] by QE
The PWscf (the plane wave self-consistent fields based DFT) tool in QE is
utilized in the current study. The algorithm of PWscf code is as shown in Fig. 3. In
the first step, the pseudopotential for each element type is generated. Further, the
energy cutoff for the basis set of a plane wave is determined using energy opti-
mization. Next, the trial density for the system n(r) is supplied.
Furthermore, the local potential is taken as the sum of potential contributions of
ionic term (Vion), Hartree term (VH), and exchange-correlation (Vxc) term. This local
potential is supplied to the Kohn–sham equations, and these equations are solved by
diagonalization of the Hamiltonian, and a new charge density n(r) is determined.
This solution for the charge density is accepted to be the solution only when it
satisfies the self-consistency. If the solution is not self-consistent, a new local
potential is generated by giving a mixing parameter, and this solution is obtained by
the iterative process.
2.2 SOEC Methodology
Elastic properties of a material quantify its resistance to various types of deforma-

tion, and hence they play a vital role in materials design. Elastic tensors can be
defined as the Taylors expansion of stress or the elastic energy in terms of the applied
strain. Elastic constants of single crystals can be obtained from first principle cal-
culations. There are two methods based on fitting total energies or stresses of
appropriately deformed crystals called energy method or stress method, respectively.
Different strain patterns are designed for crystals of different symmetry, and except
for crystals of high symmetry, these calculations are computationally expensive.
Alternative methods based on linear response theory and interatomic force constants
[32] are also used in computing the elastic properties. However, these alternative
methods are more computationally demanding. Moreover, due to high accuracy
calculations of DFT, it is widely used in the determination of elastic constants of
complex crystal structures, where experimental determination is less feasible.
In this study, ElaStic software tool is used in generating the stiffness matrix of
pristine graphene. The Elastic tool functions in the following manner:
Firstly, the space group is given as input, which is determined using sgroup
package [33].
• The space-group number determines the number of independent elastic con-
stants. It then generates sets of independent deformation matrices. Then, the
corresponding set of input structure files is created.
• The total energy is calculated for every set of structure data created using the
DFT code QE.
• The total energies computed are plotted against the applied strain. Fitting pro-
cedure is used to determine the derivatives at zero strain.
• These derivatives help in identifying all the independent components of an
elastic tensor.
• The polynomial fit and the order are decided to achieve the required accuracy of
the results.
ElaStic package can be used to determine the second- and third-order elastic
constants. Further, they also determine, the Voigt and Reuss averaged Young’s
modulus, shear modulus, bulk modulus, and Poisson ratio. The flowchart of the
ElaStic tool is as shown in Fig. 4. The crystal structure of the material to be studied
is provided as inputs. This crystal structure is optimized for atomic positions and
cell parameters to the required accuracy to attain equilibrium configuration. Thus,
the optimized structure is used as the reference system. The crystal structure is
deformed to the chosen strain level, and structure is further optimized to attain the
equilibrium in a deformed state. For a detailed description of evaluating SOEC, see
the Refs. [34, 35].
Fig. 4 Flowchart of the algorithm used for calculating SOEC using ElaStic tool [34]
3 Modeling of Graphene
The structure of graphite obtained from the experiments [36] is used in the current
study, in building the graphene structure. Graphite consists of planar sheets of
carbon in hexagonal form. Its intra-planar bonding is adamant, and the forces
holding these sheets are weak. Hence, the structure of graphene is generated from
graphite by making the inter-planar distance of parallel planes very high. Due to
this, the two parallel planes of graphene remain non-interacting. Since Graphene is
two dimensional in nature, only the planar properties are calculated.
Two carbon atoms are sufficient to define the primitive cell for graphene
structure. As shown in Fig. 5a, the two carbon atoms are placed at 1/3, 2/3 positions
of the lattice vector a and b. Conventional cells of the generated structure, indicated
by 2 2 repetitions of the unit cell are shown in Fig. 5b, and 5 5 3 repeti-
tions are shown in Fig. 5c, denoting the macroscopic system is graphene structure.
However, in the present analysis, the unit cell shown in Fig. 5a is used in the rest of
the analysis.
Variable cell relaxation calculations are performed on the primitive cell, and
equilibrated lattice parameters are determined and tabulated in Table 1. Figure 6a
shows the energy variation with change in lattice parameter ‘a’, while Fig. 6b
shows the energy variation with change in lattice parameter c. An equilibrated
lattice parameter of 2.4685 Å for ‘a’ is obtained in the relaxation simulations for
this study. Similarly, the minimum value of ‘c’ is found to be *2.6 Å. However, a
sufficiently large value of ‘c’ *6.928 Å is taken to ensure graphene structure.
Fig. 5 a Primitive cell containing two carbon atoms, b conventional cell containing 2 2
repetitions of primitive cell in the planar direction, c conventional cell containing 5 5 3 unit
repetitions of the primitive cell
Table 1 Structural parameters of graphite and pristine graphene

Structure Space A (Å) b (Å) c (Å) a b c Atomic Details
(hexagonal) group positions
—2 atoms
(fractional)
Graphite P63mc 2.470 2.470 6.790 90 90 120 C1 (1/3 2/ From [36]
3 0)
C2 (2/3 1/
3 0)
Graphene P6/ 2.468 2.468 6.928 90 90 120 C1 (1/3 2/ Parameters after
mmm 3 0) relaxing the
(191) C2 (2/3 1/ structure in QE
3 0)
Fig. 6 a Convergence of lattice parameter (a = b) with total energy for pristine graphene,
b convergence of lattice parameter ‘c’ with total energy (QE-GGA)
4 Results
4.1 Electronic Properties
Band Structure
Band structure calculations provide the quantum mechanical states of electrons in
the crystal that are characterized by a Bloch wave vector denoted by k and a band
index denoted by n. The Bloch wave vector considered as an element of the
reciprocal space, which is limited to the first Brillouin zone. Band structure helps in
explaining many magnetic, optical, and electrical properties of crystals. The loca-
tion of Fermi level, the energy of the highest energy state filled at absolute zero of
temperature, is of interest. If the Fermi level falls in a band gap, then the material is
electrically insulating (in some cases semiconducting when the band gap is small)
while in other cases it is metallic [37]. The first electronic band structure of gra-
phene using tight binding technique was calculated more than six decades ago [38],
and the study shows that it is a conductor. In our case, the band structure is
Fig. 7 Band structure of

pristine graphene calculated
using QE
Table 2 Cohesive energies of pristine graphene

Method Cohesive energy (eV/atom) Reference
DFT −7.906 [40]
Experimental −7.600 [42]
DFT, LDA pseudopotential −9.717 Present study
DFT, GGA pseudopotential −5.673 Present study
Average −7.695 Present study
calculated by choosing the K grid along selected points within the Brillouin zone
which are in the reciprocal space, and the result shows no band gap along the
position K of the Brillouin zone. Figure 7 indicates that it is conductive at ground
state which agrees well with the result from Ref. [39].
Cohesive Energies
Cohesive energy also called as formation energy or binding energy is the gain in
energy for a structure when placed in a crystalline form.
Ecohesive ¼ Estructure Efreeatoms
By the common trend in total energy calculations using DFT, it is seen that LDA
under-estimates the lattice parameter due to over binding, and GGA over-estimates
them due to under binding. The reported experimental values from literature fall
midway between the LDA and GGA results as shown in Table 2.
4.2 Elastic Properties
In this section, the SOEC obtained by DFT calculations using ElaStic code are
presented. As a first step, the structure of graphene is relaxed using ionic relaxation,
and lattice parameters are updated. This updated structure is deformed using different
strain patterns (deformation modes) based on the crystal symmetry. Graphene pos-
sesses hexagonal structure and therefore relates to transverse isotropy. Consequently,
five independent elastic constants are sufficient to define the SOEC of graphene as
shown below completely. To determine the five independent elastic constants,
energy variation with strain for five different deformation modes is sufficient.
0 1
C11 C12 C13 0 0 0
B C12 C11 C13 0 0 0 C
B C
B C13 C13 C33 0 0 0 C
B C
B 0 0 0 C44 0 0 C
B C
@ 0 0 0 0 C44 0 A
ðC11 C12 Þ
0 0 0 0 0 2
Table 3 Input parameters for Conditions Ground state (T = 0)

SOEC calculation
System Hexagonal
Material model Transversely isotropic
No. of SOEC 5
Number of deformed structures 41
Order of the polynomial fit 6
Maximum Lagrangian strain 0.05
Table 4 Elastic properties Stiffness components in LDA GGA Hills

calculated for pristine (GPa) average
graphene
C11 491.50 506.70 499.10
C33 23.30 2.90 13.10
C44 1.60 0.20 0.90
C12 80.70 88.00 84.35
C13 0.10 0.40 0.25
BV 132.31 128.69 130.55
EH 103.72 102.14 102.95
mH (Poisson’s ratio) 0.20 0.23 0.22
In the current calculations, the energy variation for each deformation mode is
calculated for 41 strain inputs. Using fitting procedure, the optimized maximum
Lagrangian strain is determined to be 0.05. Hence, the strain is varied from 0 to
0.05 by 41 equal partitions. A sixth-degree polynomial is fitted to each deformation
mode. The input parameters supplied in the present study are listed in Table 3.
QE procedure requires pseudopotentials (PP) to be supplied to run the calcula-
tions. The total energy obtained through LDA and GGA is the lower and upper
bounds because of the formalism of exchange term in the energy functional. Hence,
an averaged estimate is also obtained and presented in Table 4.
Apart from the elastic constants, the Voigt Bulk modulus (BV), Hills shear
modulus (EH), and Hills Poisson ratio (mH) values are also calculated as shown in
Table 4. Hills values are an average of the Voigt and Reuss values [41].
5 Summary
Using the first principle calculation through QE and ElaStic tool, the electronic and
mechanical properties of graphene are obtained using two different approaches
(LDA, GGA). LDA and GGA assume two distinct pseudopotentials, and hence,
cohesive energy for LDA is −9.717 and GGA is −5.673. However, the average of
the LDA and GGA is in reasonable agreement with experimental results [42].
Second-order elastic stiffness is found to be 491.5, and 506.7 GPa using LDA and
GGA. The Poisson ratio of 0.215 is obtained for graphene in these calculations
compared to 0.165 observed in the experiment [17, 43].
References
1. J.J. De Pablo, W.A. Curtin, J.J. de Pablo, Multiscale modeling in advanced materials research:
challenges, novel methods, and emerging applications. MRS Bull. 32(11), 905–911 (2011)
2. M.F. Horstemeyer, Multiscale modeling: a review, in Practical Aspects of Computational
Chemistry (Springer Netherlands, Dordrecht, 2009), pp. 87–135
3. R.E. Miller, E.B. Tadmor, A unified framework and performance benchmark of fourteen
multiscale atomistic/continuum coupling methods. Modell. Simul. Mater. Sci. Eng. 17(5),
53001 (2009)
4. E.A. Carter, Challenges in modeling materials properties without experimental input. Science
(80-.) 321(5890), 800–803 (2008)
5. R.W. Sullivan, S.M. Arnold, An annotative review of multiscale modeling and its application
to scales inherent in the field of ICME, in Models, Databases, and Simulation Tools Needed
for the Realization of Integrated Computational Material Engineering (2011), pp. 6–23
6. D.G.B. Pollock, M. Tresa, J.E. Allison, Integrated Computational Materials Engineering: A
Transformational Discipline for Improved Competitiveness and National Security
(Washington, DC: The National Academies Press, 2008)
7. M.F. Horstemeyer, From Atoms to Autos—A new Design Paradigm Using
Microstructure-Property Modeling Part 1: Monotonic Loading Conditions (Albuquerque,
NM, and Livermore, CA, 2001)
8. E.H. Falcao, F. Wudl, Carbon allotropes: beyond graphite and diamond. J. Chem. Technol.
Biotechnol. 82(6), 524–531 (2007)
9. J. Slonczewski, P. Weiss, Band structure of graphite. Phys. Rev. 109(2), 272–279 (1958)
10. A.K. Geim, K.S. Novoselov, The rise of graphene. Nat. Mater. 6(3), 183–191 (2007)
11. K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, M.I. Katsnelson, I.V. Grigorieva, S.V.
Dubonos, A.A. Firsov, Two-dimensional gas of massless Dirac fermions in graphene. Nature
438(7065), 197–200 (2005)
12. A.A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao, C.N. Lau, Superior
thermal conductivity of single-layer graphene. Nano Lett. 8(3), 902–907 (2008)
13. C. Berger, Z. Song, T. Li, X. Li, A.Y. Ogbazghi, R. Feng, Z. Dai, A.N. Marchenkov, E.H.
Conrad, P.N. First, W.A. de Heer, Ultrathin epitaxial graphite: 2D electron gas properties and
a route toward graphene-based nanoelectronics. J. Phys. Chem. B 108(52), 19912–19916
(2004)
14. S. Ghosh, I. Calizo, D. Teweldebrhan, E.P. Pokatilov, D.L. Nika, A.A. Balandin, W. Bao, F.
Miao, C.N. Lau, Extremely high thermal conductivity of graphene: prospects for thermal
management applications in nanoelectronic circuits. Appl. Phys. Lett. 92(15), 151911 (2008)
15. J.H. Seol, I. Jo, A.L. Moore, L. Lindsay, Z.H. Aitken, M.T. Pettes, X. Li, Z. Yao, R. Huang,
D. Broido, N. Mingo, R.S. Ruoff, L. Shi, Two-dimensional phonon transport in supported
graphene. Science (80-.) 328(5975), 213–216 (2010)
16. T. Schwamb, B.R. Burg, N.C. Schirmer, D. Poulikakos, An electrical method for the
measurement of the thermal and electrical conductivity of reduced graphene oxide
nanostructures. Nanotechnology 20(40), 405704 (2009)
17. C. Lee, X. Wei, J.W. Kysar, J. Hone, Measurement of the elastic properties and intrinsic
strength of monolayer graphene. Science (80-.) 321(5887), 385–388 (2008)
18. G. Tsoukleri, J. Parthenios, K. Papagelis, R. Jalil, A.C. Ferrari, A.K. Geim, K.S. Novoselov,
C. Galiotis, Subjecting a graphene monolayer to tension and compression. Small 5(21), 2397–
2402 (2009)
19. C. Lee, X.D. Wei, Q.Y. Li, R. Carpick, J.W. Kysar, J. Hone, Elastic and frictional properties
of graphene. Phys. Status Solidi B-Basic Solid State Phys. 246(11–12), 2562–2567 (2009)
20. V. Singh, D. Joung, L. Zhai, S. Das, S.I. Khondaker, S. Seal, Graphene based materials: the
past, present, and future. Prog. Mater Sci. 56(8), 1178–1271 (2011)
21. K.S. Novoselov, V.I. Fal’ko, L. Colombo, P.R. Gellert, M.G. Schwab, K. Kim, A roadmap
for graphene. Nature 490(7419), 192–200 (2012)
22. X. Wei, B. Fragneaud, C.A. Marianetti, J.W. Kysar, Nonlinear elastic behavior of graphene:
Ab initio calculations to continuum description. Phys. Rev. B Condens. Matter Mater. Phys.
80(20), 1–8 (2009)
23. F. Liu, P. Ming, J. Li, Ab initio calculation of ideal strength and phonon instability of
graphene under tension. Phys. Rev. B Condens. Matter Mater. Phys. 76(6), 1–7 (2007)
24. K.N. Kudin, G.E. Scuseria, B.I. Yakobson, C2F, BN, and C nanoshell elasticity from ab initio
computations. Phys. Rev. B 64(23), 235406 (2001)
25. G. Van Lier, C. Van Alsenoy, V. Van Doren, P. Geerlings, Ab initio study of the elastic
properties of single-walled carbon nanotubes and graphene. Chem. Phys. Lett. 326(1–2), 181–
185 (2000)
26. D. Sánchez-Portal, E. Artacho, J.M. Soler, A. Rubio, P. Ordejón, Ab initio structural, elastic,
and vibrational properties of carbon nanotubes. Phys. Rev. B 59(19), 12678–12688 (1999)
27. E. Konstantinova, S.O. Dantas, P.M.V.B. Barone, Electronic and elastic properties of
two-dimensional carbon planes. Phys. Rev. B Condens. Matter Mater. Phys. 74(3), 1–6
(2006)
28. J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple. Phys.
Rev. Lett. 77(18), 3865–3868 (1996)
29. W. Kohn, L.J. Sham, Self-Consistent Equations Including Exchange and Correlation Effects,
vol. 385, no. 1951 (1965)
30. J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C.
Fiolhais, Atoms, molecules, solids, and surfaces: applications of the generalized gradient
approximation for exchange and correlation. Phys. Rev. B 46(11), 6671–6687 (1992)
31. Giannozzi, P., Baroni, S., Bonini, N., Calandra, M., Car, R., Cavazzoni, C., … Wentzcovitch,
R. M. . QUANTUM ESPRESSO: a modular and open-source software project for quantum
simulations of materials.J. Phys. Condens. Matter 21(39), 395502 (2009)
32. M. Born, Dynamical theory of crystal lattices. Am. J. Phys. 23(7), 474 (1955)
33. B.Z. Yanchitsky, A.N. Timoshevskii, Determination of the space group and unit cell for a
periodic solid. Comput. Phys. Commun. 139(2), 235–242 (2001)
34. R. Golesorkhtabar, P. Pavone, J. Spitaler, P. Puschnig, C. Draxl, ElaStic: a tool for calculating
second-order elastic constants from first principles. Comput. Phys. Commun. 184(8), 1861–
1873 (2013)
35. R. Yu, J. Zhu, H.Q. Ye, Calculations of single-crystal elastic constants made simple. Comput.
Phys. Commun. 181(3), 671–675 (2010)
36. O. Hassel, H. Mark, Uber die Kristallstruktur des Graphits. Zeitschrift fur Phys. 25(1), 317–
337 (1924)
37. P. Hofmann, Solid State Physics: An Introduction (Wiley, 2015)
38. P.R. Wallace, The band theory of graphite. Phys. Rev. 71(9), 622–634 (1947)
39. L.-F. Huang, Y.-L. Li, M.-Y. Ni et al., Lattice dynamics of hydrogen-substituted graphene
systems. Acta Phys. Sin. 58(13), S306–S312 (2009)
40. H. Shin, S. Kang, J. Koo, H. Lee, J. Kim, Y. Kwon, Cohesion energetics of carbon allotropes:
quantum Monte Carlo study. J. Chem. Phys. 140(11), 114702 (2014)
41. R. Hill, The elastic behaviour of a crystalline aggregate. Proc. Phys. Soc. London, Sect. A 65
(389), 349–354 (1952)
42. B. Aufray, A. Kara, Ś. Vizzini, H. Oughaddou, C. Ĺandri, B. Ealet, G. Le Lay, Graphene-like
silicon nanoribbons on Ag(110): a possible formation of silicene. Appl. Phys. Lett. 96(18),
10–12 (2010)
43. O.L. Blakslee, Elastic constants of compression-annealed pyrolytic graphite. J. Appl. Phys. 41
(8), 3373 (1970)
Studies on Behaviour of Steel Tubular
Compression Members Subjected
to Accelerated Corrosion
A. Cinitha, P. K. Umesha and G. S. Palani
Abstract The performance of steel structural components is strongly influenced by

the damage due to atmospheric corrosion, whose control is a key aspect for design
and maintenance of both new and existing structures. In extreme situations, it can
lead to catastrophic failure of structural components. This paper presents experi-
mental and numerical studies on corroded compression members under stressed and
unstressed conditions. Experimental studies are carried out on corroded members
with different percentage in thickness and weight loss. A methodology has been
developed numerically to simulate three types of corrosion cases by using finite
element method, namely (1) uniform corrosion, (2) pitting corrosion and (3) uni-
form and pitting corrosion. The failure modes and the ultimate load carrying
capacity of the specimen are determined numerically and validated with experi-
mental results. A significant reduction in load carrying capacity is observed for all
corroded specimens compared to uncorroded control specimen. The failure modes
and load carrying capacity of corroded members for uniform, pitting and combined
effect of uniform and pitting corrosion are discussed. In the light of experimental
results, it is inferred that the failure of the members is due to localised axisymmetric
imperfections induced in the tubular members due to corrosion.

Keywords Corrosion Compression member Uniform corrosion

Pitting corrosion Imperfections introduction
1 Introduction
Corrosion is a natural phenomenon and major cause of deterioration of steel struc-

tures which exists as part of our everyday life. In extreme situations, catastrophic
failure such as collapse occurs due to reduction in the load bearing capability of a
structure. Corrosion damage can also results in life threatening situations; hence, it
A. Cinitha (&) P. K. Umesha G. S. Palani

CSIR-Structural Engineering Research Centre, Chennai 6000113, India
e-mail: cinitha@serc.res.in

https://doi.org/10.1007/978-981-10-7197-3_23
268 A. Cinitha et al.
has to be addressed for safety, environment and economic reasons. Steel structures
exposed to the extreme atmosphere, especially marine and highly polluted industrial
environment are subjected to corrosion. The conventional approach to evaluate
residual capacity is to perform visual inspection of the corroded members and
classify the members according to their level of damage. Rahgozar [1] reviewed
various forms of corrosion and the effects of uniform corrosion on steel structures.
They developed corrosion decay models based on the information on the locations
where corrosion occurs. Beaulieu et al. [2] studied corrosion of steel structures
exposed to various environmental conditions. They estimated the residual capacity
of corroded members in order to decide whether to change the member, repair it or
just remove corrosion and re-protect the member. Landolfo et al. [3] presented the
modelling approaches of atmospheric corrosion damage of metal structures. Based
on the studies conducted by Damgaard et al. [4], it was concluded that corrosion can
significantly reduce the service lives/life of weathering steel girders. This paper
presents the experimental and numerical studies conducted on tubular compression
members subjected to accelerated corrosion of various amount. The numerical
method developed to simulate uniform, pitting and combined form of uniform and
pitting corrosion is discussed.
2 Experimental Set-up
2.1 Unstressed Corrosion Set-up
Among the accelerated corrosion techniques, Galvanostatic method is adopted to

corrode the desired regions of the tubular compression members. In Galvanostatic
method, constant current from a DC (direct current) source is applied, till the
required amount of corrosion is achieved. Figure 1 shows the experimental set-up
to induce corrosion in the tubular specimen under unstressed condition.
Fig. 1 Experimental set-up to induce corrosion in the tubular compression members under
unstressed condition
Studies on Behaviour of Steel Tubular … 269
(a) Tubular section (b) Set-up to corrode (c) After corrosion

under service load
Fig. 2 Experimental set-up to induce corrosion under stressed condition
2.2 Stressed Corrosion Set-up
In order to achieve the stressed corrosion condition, a load under service condition
is applied to the joint component with the help of a reaction frame. After applying
the required compressive load, the load is retained by tightening the check nuts of
the upper arm of the specially fabricated load retaining frame. The corrosion
chamber is fixed to the region scheduled to corrode, and galvanic corrosion is
carried out as per Faraday’s law to achieve target % of weight loss due to corrosion,
i.e. is 20%. Figure 2 shows the experimental set-up to induce corrosion under
stressed condition.
3 Compression Test on Tubular Sections
The details of tubular members considered for experimental study are given in
Table 1. The uncorroded specimen is named as STJC-1, corroded and unstressed
specimens as STJC-2 and STJC-5, and stressed and corroded specimen as STJC-4.
The specimens were subjected to static axial compression under displacement
control by a hydraulically operated UTM of 250T capacity. The specimens were
placed between the two heads in such a way that the centre of the flange plates of
the specimen coincides with the centre of bottom platform and the top head of the
compression testing machine. The experimental set-up is shown in Fig. 3. In order
to measure the lateral deflection, two LVDTs are placed at mid-height of the
corroded region of the specimen. Four numbers of strain gauges are glued in
affected regions, i.e. two gauges along the direction of load in perpendicular to each
other and other two in horizontal direction. At each load stage, deflection and strain
measurements were recorded automatically using a data logger and analysed.
Table 1 Details of tubular specimens

Specimen Type of Initial Final Percentage reduction
identification pipes used thickness thickness in thickness
(mm) (mm)
STJC-1 80NB 3.40 3.40 0
65NB 3.80 3.80 0
STJC-2 80NB 3.40 2.10 38.24
65NB 3.80 2.30 39.47
STJC-4 80NB 3.40 3.14 7.65
65NB 3.80 3.50 7.89
STJC-5 80NB 3.40 2.40 29.41
65NB 3.80 2.70 28.94
REACTION FRAME
LOADING SYSTEM
LOAD CELL
STIFFENED PLATE
LOAD RETAINING
SYSTEM LVDT
CORROSION CHAMBER
CORRODED REGION CPU COMPUTER

DATA LOGGER
SUPPORTING SYSTEM
Fig. 3 Schematic sketch of experimental set-up
4 Numerical Investigation of Corroded Steel Tubular

Sections
In the present study, numerical investigation is carried out on the corroded steel
tubular sections by using the ABAQUS general-purpose finite element analysis
software. A methodology is developed to simulate different forms of corrosion,
namely uniform, pitting and combined, i.e. uniform and pitting, by varying the
percentage of corrosion from 10 to 50%. The observed strength reduction of cor-
roded tubular leg members for varying rate of corrosion is discussed.
ABAQUS software is used for the numerical simulation of behaviour of cor-
roded steel tubular compression member (80NB and 65NB, heavy) using STATIC,
RIKS procedure to account for the nonlinear effects. Young’s modulus and
Poisson’s ratio of the material were taken as 2.1 105 N/mm2 and 0.3, respec-
tively. In linear elastic bodies, the displacements are proportional to the applied
loads. But in the case of buckling, even if the stress–strain relation remains linear,
large deformations and the changing geometric configuration cause the structure to
behave nonlinearly. The actual behaviour can be predicted by the nonlinear anal-
ysis. The incremental plasticity model is considered and the stress–strain values
corresponding to the tensile coupon test are considered. Hence, the material is
modelled as elastic-perfectly plastic material. The model is discretised using the
element C3D8R, i.e. 8-noded linear brick element with reduced integration. It is
found that the solution converges to a unique value by using the global seed size of
50. For the ABAQUS analyses conducted for the present study, load increment size
was chosen on a trial and error basis in an attempt to provide an accurate solution
with reasonable computation time.
4.1 Simulation of Boundary Conditions
To simulate the experimental set-up in finite element analysis (FEA), the cross
section centroid of the top end of the column was defined as a reference point for
creating constraint, where all three translational and rotational degrees of freedom
are specified. A kinematic coupling constraint was defined to constrain the motion
of the top surface to that of the single reference point, where all three translational
and rotational degrees of freedom were specified as shown in Fig. 4. In this case, all
Fig. 4 Coupling constraints

Fig. 5 Boundary conditions
coupling nodes on the top surface follow the rigid body motion of the reference
point. All translational degrees of freedom of the reference point and also the nodes
in the top surface were fixed, except the vertical displacement. The bottom of the
member is fixed by restraining all degrees of freedom. The loading is applied at the
reference point to simulate the axial loading condition. The bolts are tied to the
flanges using tie constraints, and hence it will act as a rigid connection. Figure 5
shows the FE model with boundary condition.
4.2 Simulation of Corrosion Effect
4.2.1 Uniform Corrosion
The uniform corrosion effect is simulated by equal removal of thickness from the
region of interest of the member. Since in tubular sections, the outer surface is most
susceptible area of corrosion. A uniform corrosion effect is modelled by reducing
the thickness by 10–50% of original thickness, for a region of 235 mm from either
side of the mid-region of the member excluding the flanges. Figure 6 shows the
modelling of uniform corrosion effect.
Fig. 6 FE model for simulating uniform corrosion effect
4.2.2 Pitting Corrosion
In pitting corrosion, surface of the corroded member is characterised by the for-

mation of smaller pits resulting in the damage of the material property. Hence in the
present study instead of modelling the pit holes in the geometry, the pitting cor-
rosion is numerically modelled by reducing thickness as 1.9 mm in the pitted region
of interest, as shown in Fig. 7. The thickness is removed from the randomly
selected mesh elements in the centre portion of the member. The percentage of pit
corrosion is assigned by the number of pitted elements to the total number of
elements which is varied from 0 to 50%. The degree of pitting is defined as given in
Eq. (1).
Number of elements deleted

Degree of pitting ¼ ð1Þ
Total number of elements on the surface
4.2.3 Uniform and Pitting Corrosion
Simulation of combined form of corrosion is carried out for a region of 235 mm

from either side of the mid-region of the member. Initially, uniform corrosion is
Fig. 7 FE model for simulating pitting corrosion effect

Fig. 8 FE model for simulating combined effect of uniform and pitting corrosion effect
simulated by incorporating the thickness reduction of 20% followed by the pitting

corrosion effect is induced for 10–50% of the pit depth up to 30% of the original
thickness. Figure 8 shows simulation of combined effect of uniform and pitting
corrosion.
5.1 Load Bearing Capacity of Tubular Compression

Members
From the results, it is observed that the load carrying capacity of corroded tubular
members decreases significantly for both uniform and pitting corrosion. The mode
of failure is buckling which occurs at the corroded region due to the reduction in
thickness, the geometric properties such as area, moment of inertia, radius of
gyration and section modulus. Table 2 gives the load carrying capacity of the
tubular members studied for various types of corrosion effect. Figure 9 shows the
comparison of load vs deflection behaviour.
Table 2 Load carrying capacity of the tubular sections studied numerically for various types of
corrosion effect
Percentage reduction in thickness Load (kN)
Uniform Pitting Combined
0 240.2685 240.2685 211.0608
10 231.9262 226.4492 195.9038
20 211.0608 216.6892 188.1662
30 181.7877 204.8408 181.5685
40 158.3254 201.8231 173.9538
50 132.0331 187.7304 166.8362
UNIFORM PITTING COMBINED

300
250
200
Load (kN)
150
100
50
0
0 10 20 30 40 50 60
Deflection (mm)
Fig. 9 Load versus deflection
5.2 Validation of Experimental Results with FEM
The experimental results are compared with FEA results and found that both the
results agree well as given in Table 3. The experimentally observed deformed
modes are shown in Fig. 10.
From the comparison, it is evident that the results obtained using numerical
method agree well with the experimental results and the failure modes also
matching well. Figure 11 shows the comparison of load vs deflection behaviour. It
has been observed that the load carrying capacity decreases with the percentage
increase in type and level of corrosion. And among the various foams and per-
centage of corrosion studied, 40% of uniformly corroded tubular members has
significant reduction in strength.
It is clear that the load bearing capacity goes on reducing as the percentage of
corrosion increases. Buckling is observed at the region of corrosion. This section is
considered as the critical region of the tubular compression members studied. The
stress concentration is more in the regions where the pits are available. Hence, there
is the possibility of sudden failure. From the experimental and numerical studies,
combined corrosion is more dangerous than the other forms of corrosion studied.
The drastic reduction in strength may be further justified with localised axisym-
metric imperfections due to corrosion along with material degradation.
Table 3 Comparison of experimental results with FEA for columns with joints
Specimen Reduction in Experiment FEM
identification thickness Putest % Remaining PuFEM % Remaining
(kN) capacity (kN) capacity
STJC-1 0% 236.45 100.00 240.27 100.00
STJC-2 40% uniform 147.89 62.54 158.33 65.89
STJC-3 10% uniform 219.34 92.76 231.93 98.40
STJC-4 30% combined 117.65 49.75 121.24 50.46
Fig. 10 Failure modes of specimens
Fig. 11 Comparison of numerical and experimental results for the tubular sections studied
6 Conclusions
This paper presents experimental and numerical studies of corrosion effect on steel
tubular compression members, generally used in transmission towers. The study is
conducted by varying the level and type of corrosion. The level of corrosion has
been varied from 10 to 50%. The various types of corrosion effect studied are
uniform, pitting and combined effect of uniform and pitting. The remaining strength
capacity of corroded tubular compression members for varying rate of uniform and
pitting corrosion was presented and validated with experimental results.
Acknowledgements This paper is published with the permission of the Director, CSIR-Structural
Engineering Research Centre, Chennai, India.
References
1. R. Rahgozar, Remaining capacity assessment of corrosion damaged beams using minimum

curves. J. Constr. Steel Res. 65, 299–307 (2009)
2. L.V. Beaulieu, F. Legeron, S. Langlois, Compression strength of corroded steel angle
members. J. Constr. Steel Res. 66, 1366–1373 (2010)
3. R. Landolfo, L. Cascini, F. Portioli, Modelling of metal structure corrosion damage: A state of
the art report. Sustainability 2, 2163–2175 (2010)
4. N. Damgaard, S. Walbridge, C. Hansson, J. Yeung, Corrosion protection and assessment of
weathering steel highway structures. J. Constr. Steel Res. 66, 1174–1185 (2010)
Effect of Delamination Defects on Buckling
Behavior of Wind Turbine Blades
Vinodkumar Boniface
Abstract Blades in wind turbines are used to capture energy from wind. For
MW-class wind turbines, these blades are usually composite structures. They are
generally made of glass-fiber composites, while carbon-fiber composites are used
selectively in large blades. Stability of these structures is one of the key design
drivers. In spite of increasing focus on stringent manufacturing processes, some
defects are occasionally seen in the products which lead to reduced margins and
lowered life spans. These defects could be delaminations, waves, debond, etc.,
which can impact various design parameters such as strength, stability, and fatigue
life. The specific problem of the effect of delaminations on buckling is addressed in
this paper. A typical laminate stack-up is used along with different defect locations
and sizes, to arrive at a comprehensive understanding and propose a new delami-
nation model for finite element analysis. A review of recent literature is also
presented.
Keywords Wind turbine blade Composite laminate Delamination

Buckling
1 Introduction
Blades on a wind turbine are critical for energy capture. They are typically
fiber-reinforced composite structures made of monolithic laminates and sandwich
panels which are either infused or bonded together. They are designed for specific
environmental conditions (wind, air density, etc.) and are certified to appropriate
standards. One such is the Germanischer Lloyd (GL) guideline [1], which details
various aspects of wind turbine design and certification.
V. Boniface (&)
GE Renewable Engineering, John F. Welch Technology Center,
Plot #122, EPIP Phase 2, Whitefield Road, Bangalore 560066, India
e-mail: vinodkumar.boniface@ge.com

https://doi.org/10.1007/978-981-10-7197-3_24
280 V. Boniface
Defects in these composite structures could occur during manufacturing or

service. These defects could be delaminations, waves, debond, etc. and tend to
reduce margins of safety and reduce service life of the blade. In this paper, the effect
of one type of defect, i.e., delamination, on buckling margin is studied.
1.1 Wind Turbine Blade
Kim et al. [2] provided excellent information on various assessments required for a
blade structure design. Though the paper deals with designing a blade having
flat-back airfoils, the methodology is quite similar for other blades as well.
Typically, 3D finite element analysis (FEA) is used to ensure adequate margins in
fiber failure(FF), inter-fiber failure(IFF), bond stresses, and buckling stability with
extreme loads. Further, fatigue loads are used for damage assessment.
Lee and Park [3] presented results from their investigation on the effect of fatigue
damage on residual strength of a wind turbine blade. A 48.3 m-long-blade that had
experienced initial static tests and fatigue tests in accordance with the technical
specification IEC61400-23 [4] was considered. The blade had failed in negative
flapwise direction during post-fatigue static test. Delaminations were one of the
many failure mechanisms observed during post-failure examination.
1.2 Buckling Due to Delamination
Nilsson et al. [5] presented a numerical and computational study of delamination

buckling and growth for slender composite panels, focusing on global buckling
rather than thin-film buckling. Hosseini-Toudeshky et al. [6, 7] investigated de-
lamination propagation using the softening behavior of interface elements. They
showed that buckling load, delamination growth, and buckling of composite plates
depended on delamination size and laminate stack-up. Batra and Xiao [8] used a
layer-wise third-order shear and normal deformable plate/shell theory incorporating
cohesive zone model to study initiation and growth of delamination in double
cantilever beams. Butler et al. [9] developed an analytical model using thin-film
assumption and cohesive zone at interface to predict the threshold value of com-
pressive strain for buckle-driven propagation of delaminations. Lee and Park [10]
studied buckling behavior of laminated composites with delamination using FEA
with enhanced assumed strain (EAS) solid element.
Ovesy et al. [11] showed a considerable reduction in load-carrying capacity of
the laminate in the presence of delamination using higher order shear deformation
theory. Gong et al. [12, 13] investigated the effect of delamination size and shape on
buckling behavior using circular and elliptical delaminations in thin beams under
four-point bending. Results from stereoscopic digital image correlation and scan-
ning electron microscopy used in experiments were compared with FEA
Effect of Delamination Defects on Buckling … 281
predictions. Wang et al. [14] studied post-local buckling-driven delamination in

bilayer composite beams, by partitioning energy release rate using various partition
theories. Cappello and Tumino [15] presented results from FEA of buckling and
post-buckling behavior of unidirectional and cross-ply composite laminated plates
with multiple delaminations. Wang et al. [16] used a cohesive element in FEA to
predict delamination buckling and growth in a slender laminated composite for
various parameters such as model aspect ratio and delamination parameters such as
size, shape, and depth. Liu and Zheng [17] studied multiple delaminations and their
interaction toward buckling and post-buckling behavior under compressive loads
for symmetric and unsymmetric laminates.
Riks and Rankin [18] described implementation of a new sandwich FE model in
the STAGS code that was better than models based on classical sandwich theory,
with the ability to study core-to-face delaminations. Roberts et al. [19] presented
experimental work on fiber-reinforced plastic sandwich panels tested in buckling in
uniaxial compression for PVC foam and balsa cores. While balsa core had higher
buckling loads, core bonding to face sheet was better with PVC foam core. Mahfuz
et al. [20] reported experimental work, FEA, and analytical calculations showing
that core density had direct influence on global buckling of a sandwich panel, while
embedded delamination had minimal effect on both global as well as local buckling.
Gaudenzi [21] investigated the effect of a delamination on buckling load of com-
posite laminates under compression through two- and three-dimensional solid FEA,
which were compared with existing analytical solutions. Rajendran and Song [22]
used a sub-laminate approach and FEA to study delamination buckling and pre-
sented buckling loads and strain energy release rates.
2 Delaminations in Wind Turbine Blades
As seen earlier in [2], delaminations are an important mechanism in the potential

failure of a wind turbine blade. Delaminations may be caused by inadequate control
in manufacturing and also due to transport loads (impact) and in-service loads
(progressive damage). Delaminations have been investigated from various per-
spectives such as buckling loads, post-buckling, delamination propagation, etc.
Also, different approaches, analytical, experimental and computational, have been
used to study the effect of delaminations.
This study is focused on the effect of delaminations on buckling, with particular
reference to wind turbine blades. Typically, wind turbine blades are modeled using
shell elements for FEA [1]. These shell elements have details of the laminae (and
core in the case of sandwich panels) in the stack-up.
Here, a representative 300 300 mm laminate panel made of 11 layers of
1 mm thick carbon-fiber UD is considered for buckling analysis. Properties are
shown in Table 1. A 100 100 mm delamination is considered at the center of the
panel. For FEA, the model has an in-plane mesh density of 10 10 mm; for solid
elements, mesh density in the thickness direction is 1 mm.
282 V. Boniface
Table 1 Material properties Property Units Value

of CFRP [13]
E11 GPa 132.0
E22 GPa 10.3
E33 GPa 10.3
m12 0.25
m13 0.25
m23 0.38
G12 GPa 6.5
G13 GPa 6.5
G23 GPa 3.91
All edges of the panel are simply supported, i.e., uy displacements are con-
strained. For solid elements, only the mid-layer of edge nodes is constrained.
A compressive load of Fy ¼ 1e5N is applied on one edge, while the opposite edge
is constrained in uy (see Fig. 1). In Fig. 1b, delamination is shown at the center in
dark blue.
For FEA, the commercial software ANSYS R15.0 is used. The panel described
above is modeled using various options in FEA, resulting in the following
configurations:
1. SHELL181 (one element through thickness, with 11 layers)
2. SHELL181 (11 elements through thickness, with 1 layer each)
3. SHELL181 (11 elements through thickness, with 1 layer each) + CONTA173 at
delamination
4. SOLSH190 (11 elements through thickness, with 1 layer each)
(a) (b)
1 1
Y Y
Z X Z X
SHELL181, 11 layers, delam (0x0) between 6-7 SHELL181, 11 layers, delam (100x100) between 6-7
Fig. 1 Finite element model a without delamination, b with delamination at the center, shown in
dark blue
5. SOLSH190 (11 elements through thickness, with 1 layer each) + CONTA173 at

delamination
6. SOLID185 (11 elements through thickness, with 1 layer each)
7. SOLID185 (11 elements through thickness, with 1 layer each) + CONTA173 at
delamination.
Note that these element types are:
• CONTA173: 3D 4-node Surface-to-Surface Contact
• SHELL181: 4-node Structural Shell
• SOLID185: 3D 8-node Structural Solid
• SOLSH190: 3D 8-node Structural Solid Shell.
2.1 Models with SHELL181 Elements
For SHELL181 elements with 11 elements through thickness (i.e., configurations

2–4), the CERIG command in ANSYS is used (which defines a rigid region using
constraint equations), see Fig. 2. Each row of 11 nodes is connected using CERIG.
Delaminations are modeled through a break in CERIG, i.e., using two CERIG
definitions on either side of the delamination. In Fig. 2, this is shown for a de-
lamination between layers 6 and 7.
1 1
No Delamination:
CERIG across all layers
Delamination between
layers 6 & 7:
1) CERIG across layers
1 to 6
2) CERIG across layers
7 to 11
11 elements (through
thickness) at this No Delamination:
Y delamination corner Y CERIG across all layers
Z X elaborated alongside X
Z
SHELL181, 11 layers, delam (100x100) between 6-7 SHELL181, 11 layers, delam (100x100) between 6-7
Fig. 2 Finite element model: Delamination in SHELL181 model with 11 layers

284 V. Boniface
2.2 Models with Solid Elements (SOLID185

and SOLSH190)
For solid elements with 11 elements through thickness (i.e., configurations 5–7),
elements are attached by merging nodes at the interface between layers. To model
delamination, nodes on either side of the delamination are not merged.
3 Buckling Analysis
In this paper, the effect of delaminations on panel buckling is studied through linear
eigenvalue buckling analysis. While modeling with solid elements is relatively
straightforward, the CERIG approach used for shell elements needs to be validated.
This is done by comparing FEA results with available analytical results.
3.1 Benchmark Problem
To establish the CERIG modeling approach mentioned in Sect. 2.1, a panel made of
steel without delamination is analyzed first. This corresponds to configuration #2
with 11 elements through the thickness having 1 layer each, attached with CERIG.
Properties of steel are provided in Table 2.
The first five buckling modes are shown in Fig. 3 for the steel panel of con-
figuration #2 without delamination.
From Table 15.2 in [24], critical buckling load for a panel simply supported on
all edges and subject to a compressive load is
t 2
E
Pcr ¼ K bt ð1Þ
ð 1 m2 Þ b
here, E is Young’s Modulus, m is Poisson’s Ratio, t is thickness, b is width, a is

length and K is a function of ða=bÞ ratio and is equal to 3.29 for a=b ¼ 1.
Substituting values for this steel panel, Pcr ¼ 3:368e6N. On comparing with the
applied load of Fy ¼ 1e5N, it is seen that buckling load factor is 33.68. From FEA,
buckling load factor is 32.34 (see Fig. 3), which matches analytical prediction
within 4%. This shows that the CERIG approach used for the 11-layer shell models
is acceptable. Composite panels are analyzed in the next section.
Table 2 Material properties Property Units Value

of steel [23]
E GPa 210.0
m 0.3
Fig. 3 First five buckling modes, steel panel without delamination, configuration 2
3.2 Composite Panels
The first five buckling modes are shown in Fig. 4 for the composite panel of
configuration #2 without delamination. Note that some of the modes are similar to
those in Fig. 3 for steel, though buckling load factors are quite different as seen in
the table in Fig. 5.
A summary of buckling load factors from the different configurations is shown
in Fig. 5. As expected, the SOLID185 model shows anomalous behavior, while
SHELL181 and SOLSH190 are in agreement. Two locations with delaminations
are studied:
(a) Delamination between layers 6 and 7: In this case, the trend is similar to the
panel with no delaminations, i.e., only SOLID185 model shows a different
behavior. Models with and without contact elements exhibit similar behavior.
This is because the delamination is at 54.5% thickness and the two parts of the
delaminated laminate have approximately equal stiffness and their buckling
behavior is similar.
(b) Delamination between layers 9 and 10: In this case, however, the effect of
contacts is seen. Both SHELL181 and SOLSH190 with contact elements
(CONTA173) show similar results. This delamination is at 81.8% thickness,
and the thinner part of the delaminated laminate has a stiffness value signifi-
cantly lower than the thicker part and so approaches thin-film buckling. Unless
contact elements are used, the delaminated surfaces can penetrate each other as
286 V. Boniface
Fig. 4 First five buckling modes, composite panel without delamination, configuration 2
Fig. 5 Buckling Comparison a no delamination, b delamination between layers 6 and 7,

c delamination between layers 9 and 10
shown next. As before, SHELL185 models show anomalous behavior, while

the effect of CONTA173 in them is seen.
Fig. 6 Effect of contact elements on buckling for mode #1, delamination between layers 9 and 10
a–c without contacts, d–f with CONTA173
Effect of contact elements for the case of a delamination introduced between

layers 9 and 10 is shown in Fig. 6. Buckling mode #1 is considered. It is seen that
without contacts, Fig. 6a–c, the two surfaces of the delamination penetrate into each
other. Note, however, that deformations shown here are scaled by a factor of 20. On
introducing contacts, Fig. 6d–f, the surfaces do not interfere and the buckling mode
is quite different.
3.3 Alternative Delamination Model
Another approach to model delamination as a soft laminate layer is investigated. In

configuration #1, i.e., SHELL181 elements with one element through thickness
made of 11 layers, a 12th layer is introduced at the location of delamination. This
layer is modeled with negligible stiffness, Ex ¼ 0:001 GPa, mxy ¼ 0:3. As before,
two locations with delaminations are studied, i.e., (a) between layers 6 and 7 and
(b) between layers 9 and 10. Thickness of this soft layer is varied and buckling
results are compared with those from configuration #2, i.e., SHELL181 elements
with 11 elements through thickness made of one layer each. Figure 7 shows this
comparison for mode #1. Also shown are ±5% bounds from which it is seen that
for a delamination thickness of around 0.02 mm, similar buckling load factors are
obtained.
288 V. Boniface
Fig. 7 Effect of introducing a thin layer of delamination in SHELL181 model with 1 element
through thickness made of 11 layers a delamination between layers 6 and 7, b delamination
between layers 9 and 10
This approach would be convenient for use in FEA of a Wind Turbine Blade
including the effect of delaminations. However, robustness of this approach and
domain of application needs to be established. A design-of-experiments
(DoE) study including variables like mesh density, materials, delamination geom-
etry, stack-up, and load is required to get a better understanding. Additionally,
interaction of delaminations also needs to be studied.
Effect of delaminations on buckling with specific reference to wind turbine blades is

addressed. Different element types are used in this study using FEA. It is seen that
contacts have to be modeled in the delamination to obtain realistic buckling modes.
A novel approach of using a soft layer to model delamination is investigated.
Results from this approach depend on thickness of this soft layer. For the specific
problem looked at here, a thickness of around 0.02 mm seems to match results from
the shell model with contact elements where the 11 layers are modeled explicitly.
However, the effect of various parameters like mesh sensitivity, materials, de-
lamination geometry (size, shape, location, etc.), stack-up, and loads need to be
investigated before this approach can be used reliably. Further, experimental vali-
dation of FEA results using sub-article tests is recommended.
References
1. Guideline for Certification of Wind Turbines, Edition 2010, Germanischer Lloyd Industrial
Services GmBH, Hamburg, Germany (2010)
2. S.H. Kim, H.J. Bang, H.K. Shin, M.S. Jang, Composite structural analysis of flat-back shaped
blade for multi-MW class wind turbine. Appl. Compos. Mater. 21, 525–539 (2014)
3. H.G. Lee, J. Park, Static test until structural collapse after fatigue testing of a full-scale wind
turbine blade. Compos. Struct. 136, 251–257 (2016)
4. International Electrotechnical Commission International Standard IEC 61400-23, Wind

turbines—part 23: full-scale structural testing of rotor blades (2014)
5. K.-F. Nilsson, L.E. Asp, J.E. Alpman, L. Nystedt, Delamination buckling and growth for
delaminations at different depths in a slender composite panel. Int. J. Solids Struct. 38, 3039–
3071 (2001)
6. H. Hosseini-Toudeshky, S. Hosseini, B. Mohammadi, Delamination buckling growth in
laminated composites using layerwise-interface element. Compos. Struct. 92, 1846–1856
(2010)
7. H. Hosseini-Toudeshky, S. Hosseini, B. Mohammadi, Buckling and delamination growth
analysis of composite laminates containing embedded delaminations. Appl. Compos. Mater.
17, 96–109 (2010)
8. R.C. Batra, J. Xiao, Analysis of post-buckling and delamination in laminated composite St.
Venant-Kirchhoff beams using CZM and layer-wise TSNDT. Compos. Struct. 105, 369–384
(2013)
9. R. Butler, A.T. Rhead, W. Liu, N. Kontis, Compressive strength of delaminated aerospace
composites. Philos. Trans. R. Soc. A 370, 1759–1779 (2012)
10. S.Y. Lee, D.Y. Park, Buckling analysis of laminated composite plates containing delami-
nations using the enhanced assumed strain solid element. Int. J. Solids Struct. 44, 8006–8027
(2007)
11. H.R. Ovesy, M. Taghizadeh, M. Kharazi, Post-buckling analysis of composite plates
containing embedded delaminations with arbitrary shape by using higher order shear
deformation theory. Compos. Struct. 94, 1243–1249 (2012)
12. W. Gong, J. Chen, E.A. Patterson, An experimental study of the behaviour of delaminations
in composite panels subjected to bending. Compos. Struct. 123, 9–18 (2015)
13. W. Gong, J. Chen, E.A. Patterson, Buckling and delamination growth behaviour of
delaminated composite panels subject to four-point bending. Compos. Struct. 138, 122–133
(2016)
14. S. Wang, C.M. Harvey, B. Wang, A. Watson, Post-local buckling-driven delamination in
bilayer composite beams. Compos. Struct. 133, 1058–1066 (2015)
15. F. Cappello, D. Tumino, Numerical analysis of composite plates with multiple delaminations
subjected to uniaxial buckling load. Compos. Sci. Technol. 66, 264–272 (2006)
16. R.G. Wang, L. Zhang, J. Zhang, W.B. Liu, X.D. He, Numerical analysis of delamination
buckling and growth in slender laminated composite using cohesive element method.
Comput. Mater. Sci. 50, 20–31 (2010)
17. P.F. Liu, J.Y. Zheng, On the through-the-width multiple delamination, and buckling and
postbuckling behaviors of symmetric and unsymmetric composite laminates. Appl. Compos.
Mater. 20, 1147–1160 (2013)
18. E. Riks, C.C. Rankin, Sandwich modeling with an application to the residual strength analysis
of a damaged compression panel. Int. J. Non-Linear Mech. 37, 897–908 (2002)
19. J.C. Roberts, M.P. Boyle, P.D. Wienhold, G.J. White, Buckling, collapse and failure analysis
of FRP sandwich panels. Compos. B 33, 315–324 (2002)
20. H. Mahfuz, S. Islam, M. Saha, L. Carlsson, S. Jeelani, Buckling of sandwich composites:
effects of core-skin debonding and core density. Appl. Compos. Mater. 12, 73–91 (2005)
21. P. Gaudenzi, On delamination buckling of composite laminates under compressive loading.
Compos. Struct. 39(1–2), 21–30 (1997)
22. S. Rajendran, D.Q. Song, Finite element modelling of delamination buckling of composite
panel using ANSYS. In Proceedings of 2nd Asian ANSYS User Conference, Singapore, 11–13
Nov 1998
23. H.A. Kuhn, Overview of Mechanical Properties and Testing for Design, Mechanical Testing
and Evaluation, vol. 8 (ASM International, ASM Handbook, 2000), pp. 49–69
24. Roark’s Formulas for Stress and Strain, Table 15, 7th edn (McGraw Hill, New York, 2001)
Analysis of the Performance
of the Cantilever-Type Piezoelectric
Generator Based on Finite Element
Modeling
A. N. Soloviev, I. A. Parinov, A. V. Cherpakov, V. A. Chebanenko,

E. V. Rozhkov and L. V. Duong
Abstract The paper analyzes the influence of the characteristics of proof mass on
the output characteristics of cantilever-type piezoelectric generators (PEGs) with
bimorph elements by using finite element modeling (FEM) in the ANSYS software.
This investigation studies the effects of the geometric dimensions of proof mass and
the place of its attachment. Symmetrical and asymmetrical cases of mass placement
are also studied. The problems of the evaluation of influence of PEG’s electrome-
chanical characteristics on output parameters of the generators are considered. The
results of the computations of output voltage and power of the PEGs are present.

Keywords Finite element method (FEM) Piezoelectric generator (PEG)

Bimorph Cantilever Proof mass Damped oscillations Harmonics

Frequency Voltage Power
1 Introduction
The problem of the development of piezoelectric generators (PEGs) for different

purposes has not been completely solved yet due to the low-energy efficiency of
PEGs and low output power of the existing experimental models. Among the
various problems arising during the development of PEGs, one can highlight the
A. N. Soloviev I. A. Parinov (&) A. V. Cherpakov V. A. Chebanenko

E. V. Rozhkov
Southern Federal University, Rostov-on-Don, Russia
e-mail: parinov_ia@mail.ru
A. N. Soloviev A. V. Cherpakov
Don State Technical University, Rostov-on-Don, Russia
A. N. Soloviev V. A. Chebanenko
Southern Scientific Center of RAS, Rostov-on-Don, Russia
L. V. Duong
Department of Mechanical Engineering, Le Quy Don Technical University,
Hanoi, Vietnam

https://doi.org/10.1007/978-981-10-7197-3_25
292 A. N. Soloviev et al.
most significant of them, including a choice of energy-efficient piezoceramic

materials (PCMs), the development of electrical circuits of accumulation of elec-
trical energy with minimal leakage of charge, the search of geometrical configu-
rations, and assembly technologies of PEG’s sensitive element, which induce
maximum output power. Approaches used to solve these problems differ substan-
tially, depending on the application of PEG. Previously, problems of energy har-
vesting systems, including PEGs, were widely discussed in the reviews [1–6].
Later, researches were focused toward the study of PEGs with low-power char-
acteristics, where already have achieved quite tangible results [7, 8].
The finite element method (FEM) is widely used together with the analytical
methods of studying PEGs (the main types of these models are present, e.g., in [9–
11]). The main advantage of FEM is the capability to conduct numerical simula-
tions of specific models of the devices. This allows one to optimize the model
parameters, meet the specified performance, and achieve the maximum PEG’s
output characteristics. We discuss the examples of such studies below. In [12], it is
investigated the material effects of various piezoelectric ceramics on energy effi-
ciency and some resonance characteristics of a cantilever bimorph harvester. The
authors concluded that PMN-32 is of greatest interest to the energy storage prob-
lems among all investigated compounds. In [13], the possibility of creating a power
supply for non-destructive testing devices was investigated. Together with this, the
researchers proposed a model of multimodal PEG operating from oscillations
arising in the structures, where the sensors are installed usually. Previously [14, 15],
we have conducted, by using the finite element modeling, the analysis of the effect
of proof mass and thickness of the piezoelectric elements on the output charac-
teristics of axial-type PEG and studied the electrical energy harvesting schemes.
2 Finite Element Modeling of PEG
2.1 Continuous Models of Composite Elastic, Electroelastic,

and Electroacoustic Medium
Piezoelectric energy harvesting device presents itself a composite elastic and

electroelastic body, oscillating with small amplitude in the moving coordinate
system. The vertical displacement of the system is defined by the law yðtÞ or, in the
case of external force excitation, by FðtÞ (or pressure rðtÞ), due to the device’s base
moves. The initial-boundary value problem of linear electrodynamics coincides
with the mathematical model, describing the operation of such a device [16].
In this paper, we use the linear theory of elasticity and electrodynamics, taking
into account the dissipation of energy and realizing in the ANSYS software [17], as
well as the equations of motion in the acoustic approximation [18].
Analysis of the Performance of the Cantilever … 293
The piezoelectric medium could be described by the following equations:
q€ui þ aqu_ i rij;j ¼ fi ; Di;i ¼ 0; ð1Þ
rij ¼ cijkl ðekl þ b_ekl Þ eijk Ek ; Di þ 1d D_ i ¼ eikl ðekl þ 1d e_ kl Þ þ 3ik Ek ; ð2Þ
ekl ¼ ðuk;l þ ul;k Þ=2; Ek ¼ u;k ; ð3Þ
where q is the material density; ui are the components of the displacement vector;
rij are the components of the stress tensor; fi are the components of the vector of the
density of mass forces; Di are the components of the electric induction vector; cijkl
are the components of the fourth-rank tensor of the elastic moduli; eijk are the
components of the third-rank tensor of piezoelectric coefficients; eij are the com-
ponents of strain tensor; Ei are the components of the electric field vector; u is the
electric potential; 3ij are the components of the dielectric constants tensor; and
a; b; 1d are non-negative damping coefficients (the value of 1d ¼ 0 is used in
ANSYS software).
2.2 Modeling
The full-scale finite element model of PEG presents itself a cantilever structure with
a bimorph of thin symmetric piezoelectric layers, polarized in thickness, and glued
to an elastic plate. The cantilever PEG’s geometrical dimensions are shown in
Fig. 1a: The substrate has dimensions l b h = 150 9.8 1 mm3, piezo-
electric elements consist of two the same piezoelectric plates, polarized in thickness
with dimensions lp bp hp = 54 6 0.5 mm3. The center of the proof mass
is fixed at a distance lm from the cantilever clamp. We varied the size of lm in
different numerical calculations into range from 65 to 150 mm. Electric scheme of
compound PEG under active load is shown in Fig. 2. Parametrization of the proof
mass’s geometrical dimensions is shown in Fig. 3a. The length Lm and width bm of
the inertial element are constant and equal to 8 mm and 22 mm, respectively.
Fig. 1 Structural scheme a and its finite element model b for studied cantilever PEG:
1 piezoelectric element, 2 substrate, 3 proof mass, 4 clamp
Fig. 2 Electric scheme of compound PEG under active load
Fig. 3 Two cases of the attachment of proof mass: a symmetrical arrangement, b asymmetrical
arrangement
Thickness parameters hm3, hm1 can vary from 0.1 to 5 mm; hm2 is the base thickness
equal to 1 mm.
The value of the proof mass was varied from 5 to 25 g, and material of
piezoelectric plates was hot-pressed piezoelectric ceramic PCR-7 M, whose prop-
erties given in Table 1 [11].
As the substrate materials for PEG, we consider duralumin, steel, and fiberglass.
The density values of duralumin are above density of the fiberglass, but lower than
the density of steel. At the same time, the material of proof mass is a fiberglass. The
properties of abovementioned materials are present in Table 2.
Figure 1b demonstrates three-dimensional finite element model of PEG. The
electroelastic FE SOLID5 with 4 degrees of freedom models piezoelectric medium 1
Table 1 Elastic moduli CEpq (1010 Pa), piezoelectric coefficients ekl (in C/m2) and relative
permittivity enkk/e0 (at room temperature)
CE11 CE12 CE13 CE33 CE44 e31 e33 e15 en11/e0 en33/e0
12.5 8.4 8.1 12.1 2.36 −9.0 28.3 17.9 1,430 1,350
Table 2 Mechanical Material q (kg/m3) E 1010 (Pa) m

properties of the used
materials Fiberglass 2,500 7.0 0.25
Steel 7,800 21.0 0.30
Duralumin 2,700 7.4 0.34
as follows: VOLT is the electric potential, and UX, UY, and UZ are the components of
the displacement vector; the elastic finite element SOLID186 models media 2 and 3.
The developed FE models based on using ANSYS software and exact formu-
lation of the problem (1–3) allowed us to conduct modal and harmonic analysis.
2.3 Results of Calculations
The natural frequencies and their eigenforms of PEG vibrations were calculated.
Table 3 shows corresponding values, and Fig. 4 presents the eigenforms of
vibrations. Modal analysis shows that the first, second, fifth, and sixth oscillation
modes are transverse modes of vibration in the vertical plane OXY. The third
oscillation mode is a transverse oscillation mode in a horizontal plane OXZ. The
fourth mode is a torsion oscillation mode with respect to the axis OX. The differ-
ences of the natural vibration frequencies for first to fourth and sixth modes of
vibrations are less than 1.7%, and for fifth oscillation mode, it is 6.4%.
At the first stage, it was solved the problem on the comparison of the PEG’s
output parameters in symmetric and asymmetric attachments of proof mass at
various values of electric resistive load on each of the piezoplates for the proof
mass, located on distance lm = 65 mm from the cantilever clamp. The output
Table 3 Natural frequencies Mode Proof mass geometric parameters

of the model PEG with
Symmetric case (Hz) Asymmetric case (Hz)
different symmetry of proof
mass location at lm = 65 mm 2.55 mm 2.55 mm 0.1 mm 5 mm
1 49.440 49.368
2 170.08 167.90
3 244.43 240.62
4 389.31 382.81
5 627.14 587.17
6 1184.5 1168.8
1 mode 2 mode
3 mode 4 mode
5 mode 6 mode
Fig. 4 First six modes of oscillation eigenforms for PEG with asymmetrical location of proof
mass
Table 4 Output voltage and power for each of piezoplates at proof mass location lm = 65 mm;
results are shown for first oscillation mode
Rn, X Asymmetric case Symmetric case
U1 (V) U2 (V) P1 P2 U1 = U2 (V) P1 = P2
(10−6 W) (10−6 W) (10−6 W)
1106 15.17 15.19 230.00 230.70 15.11 228.30
5105 14.19 14.17 402.80 401.50 14.12 398.70
2.5105 12.28 12.25 602.70 599.80 12.21 596.30
1105 8.18 8.17 669.60 667.30 8.15 664.20
5104 5.02 5.01 503.80 502.00 5.00 500.00
1104 1.17 1.17 137.10 136.70 1.17 136.00
voltage of the piezoelectric plates 1 and 2 and output power of every plate for the
symmetric and asymmetric cases of proof mass attachments are present in Table 4.
Analysis of the results from Table 4 shows that the values of output voltage
differ weakly for the piezoplates in the asymmetric and symmetric cases at various
resistive loads. The relative voltage deviation does not exceed 0.6%. As a result,
asymmetrical arrangement of the proof mass had low effect on the output param-
eters of PEG’s voltage in this study.
At the next stage, we analyzed output voltage on each electrode for first six
vibration modes in symmetric and asymmetric attachment of the proof mass for
value of distance lm = 65 mm and electric load resistance equal to R = 1105 X.
Table 5 shows the deviation of maximum voltage (DU), as well as maximum power
(DP) on one of the electrodes at symmetric and asymmetric location of proof mass.
Analysis of the results presented in Table 5 shows that the maximum output
power at the excitation of oscillations at the corresponding resonance frequency by
a certain force, applied at the point on the free end of the PEG in the direction OY
(Fig. 4), is achieved only at the first resonance frequency. Third and fourth oscil-
lation modes have the least values of output power at a given vibration excitation. It
is caused that the third oscillation mode excites oscillations in OXZ plane, which is
perpendicular to the exciting force. The fourth oscillation mode corresponds to
torsion oscillation that also affects the efficiency of the output power. The values of
the output power for second, fifth, and sixth vibration modes are equal to
25.9 10−6 W, 33.7 10−6 W, and 24.9 10−6 W, respectively. However, they
are more than 25 times less than the output power at the first vibration mode. The
deviation of the output voltage on each electrode at symmetric and asymmetric
location of proof mass and electric resistance load R = 1 105 X is less than 1%
for first, second, fifth, and sixth modes of vibrations. The difference of output power
does not exceed 1.7%. For third and fourth oscillation modes, output voltage and
output power on each electrode do not exceed values of 0.42 V and 1.7 10−6 W,
respectively, which is quite low compared to other modes of vibration.
At the next step of the simulation, two PEG models with different conditions of
the proof mass attachment and mechanical excitation were considered (the schemes
for both models are shown in Fig. 5).
Table 5 Output voltage and power for each of piezoplates at the proof mass, located on distance
lm = 65 mm from the cantilever clamp; results are shown for first oscillation mode at load
resistance R = 1105 X
Mode Asymmetric case Symmetric case DU (%) DP (%)
U1 (V) U2 (V) P1 P2 U1 = U2 P1 = P2
(10−6 W) (10−6 W) (V) (10−6 W)
1 8.17 8.18 667.3 669.6 8.15 664.2 −0.4 −0.8
2 1.53 1.61 23.4 25.8 1.61 25.9 0.3 0.6
3 0.39 0.42 1.5 1.7 0.4 1.6 −4.0 −8.2
4 0.31 0.34 1.0 1.2 0.31 0.9 −10.4 −21.9
5 1.58 1.82 25.0 33.1 1.84 33.7 0.9 1.7
6 1.51 1.58 22.8 25.0 1.58 24.9 −0.1 −0.3
(a) l
(b) l
lm l2 lm l2
l1 lp l1 lp
h
h
hm
y(t) F(t)
hp
hp
ux = 0 ux = 0
hm
uy = 0 1 2 3 uy = 0
uz 1 2 3
uz = 0
Fig. 5 Schemes for two models of cantilever PEG with proof mass (explanations are into text):
1 piezoelements; 2 substrate; 3 inertial element (proof mass)
Fig. 6 Dependencies of voltage amplitude: a and c for the first model; b and d for the second
model: 1 fiberglass; 2 duralumin; 3 steel
The first model performs small oscillations in the moving coordinate system,
associated with the surface, which locates at the left side of the plate. The vertical
vibrations of the moving system are determined by the relation:
yðtÞ ¼ y0 eið2pf Þt ; ð4Þ
where y0 = 1 mm is the oscillation amplitude, and f is the frequency of forced

oscillations.
In the second model of PEG, its base can move freely only in the vertical
direction, but horizontal direction is fixed. The oscillations are excited by the
external force:
FðtÞ ¼ F0 eið2pf Þt ; ð5Þ

Fig. 7 Dependencies of voltage amplitude: a for the first model; b for the second model;
1 fiberglass; 2 duralumin; 3 steel
where F0 = 1 N is the amplitude of exciting force, and f is the frequency of forced

oscillations.
The dependencies of voltage amplitude, calculated on the free electrode, on the
material properties, and on the length l of substrate, at the substrate thickness of
1.2 mm and vibrations, excited at a non-resonance frequency of 10 Hz, are
demonstrated in Fig. 6. The calculation results in the case of the proof mass absence
are shown in Fig. 6a, b, and the results for the inertial mass equal to 5 g are present
in Fig. 6c, d.
The dependencies of the amplitude values of voltage on the upper electrode, at
the non-resonance frequency of 10 Hz, on the value of the proof mass, which varies
in the range of 0–15 g, are present in Fig. 7. The substrate thickness h and length
l were 1.2 and 110 mm, respectively.
3 Conclusions
Based on the finite element modeling performed by using ANSYS software, the
problem of influence asymmetric attachment of proof mass on the output voltage at
the first resonance frequency, taking into account electric resistive load, was con-
sidered. Analysis of the parameters of PEG output voltage shows that the output
voltage in the case of asymmetric location of proof mass varies no more than 0.6%
compared to the value in the case of symmetric proof mass attachment.
Analysis of the results of the simulation of PEG oscillations at non-resonance
frequency of 10 Hz, taking into account substrates of different materials and
location of attached proof mass, shows that the elastic characteristics of the sub-
strate directly define the output voltage of PEG. Thus, by comparing the results with
various proof mass locations, we showed that the ratio of the values of output
voltage V (at l = 65 mm) and V (at l = 150 mm) can differ more than four times.
The results, presented in Figs. 6 and 7, show that the output voltage V at the
non-resonance frequency 10 Hz is significantly less than the corresponding quan-
tities in the case of resonance; however, these results allow us to choose optimal
materials and geometrical dimensions of the substrate and proof mass.
Analysis of output voltage on different electrodes of PEG at asymmetric
attachment of proof mass shows that a deviation of the output voltage on each
electrode at symmetrical and asymmetrical fixing the attached mass and electric
resistive load R = 1105 X is for first, second, fifth, and sixth modes of vibrations
less than 1%. This difference of output powers does not exceed 1.7%. For third and
fourth oscillation modes, the output voltage and output power on each electrode
does not exceed the values of 0.42 V and 1.710−6 W, respectively, which is quite
low compared to other modes of vibration.
Acknowledgements This study has been performed at partial support of Russian Foundation for
Basic Research (grants Nos. 16-08-00740, 14-08-00546) and proposals Nos. 213.01-11/2014-25,
0110-11/2017-20 of the Russian Ministry of Education and Sciences (basic part). I.A. Parinov
acknowledges financial support of the Russian Ministry of Education and Sciences in the
framework of “Organization of Scientific Research” Government Assignment.
References
1. S.R. Anton, H. Sodano, A review of power harvesting using piezoelectric materials Smart
Mater. Struct. 16, К1–К21 (2007)
2. S. Priya, Advances in power harvesting using low profile piezoelectric transducers.
J. Electroceramics. 19, 167–184 (2007)
3. K.A. Cook-Cherault, N. Thamb, A.M. Sastry, Powering MEMS portable devices—a review
of non-regenerative and regenerative power supply systems with emphasis on piezoelectric
energy harvesting systems. Smart Mater. Struct. 17, 043001 (2008)
4. H.S. Kim, J.H. Kim, J. Kim, A review of piezoelectric energy harvesting based on vibration.
Int. J. Precis. Eng. Manuf. 12(6), 1129–1141 (2011)
5. R. Caliò, U.B. Rongala, D. Camboni, M. Milazzo, C. Stefanini, G. de Petris, C.M. Oddo,
Piezoelectric energy harvesting solutions. Sensors 14(3), 4755–4790 (2014)
6. V.A. Chebanenko, V.A. Akopyan, I.A. Parinov, Chapter 10. Piezoelectric generators and
energy harvesters: modern state of the art, pp. 243–277. in Piezoelectrics and Nanomaterials:
Fundamentals, Developments and Applications, ed. by I.A. Parinov (Nova Science
Publishers, New York, 2015), 283 pp.
7. V.A. Akopyan, Y.N. Zakharov, I.A. Parinov, E.V. Rozhkov, S.N. Shevtsov, V.A.
Chebanenko, Chapter 4. Optimization of output characteristics of the bimorph power
harvesters, pp. 111–131. in Nano- and Piezoelectric Technologies, Materials and Devices, ed.
by I.A. Parinov (Nova Science Publishers, New York, 2013), 261 pp.
8. V.A. Akopyan, I.A. Parinov, Y.N. Zakharov, V.A. Chebanenko, E.V. Rozhkov, Chapter 24.
Advanced investigations of energy efficiency of piezoelectric generators, pp. 417–436. in
Advanced Materials—Studies and Applications, eds. by I.A. Parinov, S.H. Chang, S.
Theerakulpisut (Nova Science Publishers, New York, 2015), 527 pp.
9. A. Erturk, P.A. Tarazaga, J.R. Farmer, D.J. Inman, Effect of strain nodes and electrode
configuration on piezoelectric energy harvesting from cantilevered beams. J. Vib. Acoust. 131
(1), 011010 (2009)
10. N.G. Elvin, A.A. Elvin, A general equivalent circuit model for piezoelectric generators.
J. Intell. Mater. Syst. Struct. 20(1), 3–9 (2009)
11. Y. Liao, H.A. Sodano, Optimal parameters and power characteristics of piezoelectric energy
harvesters with an RC circuit. Smart Mater. Struct. 18(4), 045011 (2009)
12. A. Nechibvute, A. Chawanda, P. Luhanga, Finite element modeling of a piezoelectric
composite beam and comparative performance study of piezoelectric materials for voltage
generation. ISRN Materials Science (2012)
13. Y. Zhang, W. He, Multi-mode piezoelectric energy harvesters for wireless sensor network
based structural health monitoring. in The 15th International Symposium on Smart Structures
and Materials & Nondestructive Evaluation and Health Monitoring 2008, pp. 69340Z–
69340Z
14. A.N. Soloviev, I.A. Parinov, L.V. Duong, C.C. Yang, S.H. Chang, J.C.Y. Lee, Chapter 24.
Analysis of finite element models for piezoelectric devices of energy harvesting, pp. 335–352.
in Physics and Mechanics of New Materials and their Applications, eds. by I.A. Parinov, S.H.
Chang (Nova Science Publishers, New York, 2013), p. 444
15. S. Shevtsov, V. Akopyan, E. Rozhkov, V. Chebanenko, C.C. Yang, C.Y. Lee, C.X. Jenny
Kuo, Chapter 24. Optimization of the electric power harvesting system based on the
piezoelectric stack transducer. Advanced materials—manufacturing, pp. 639–650. in Physics,
Mechanics and Applications, eds. by I.A. Parinov, S.H. Chang, V.Y. Topolov (Springer,
Heidelberg, 2016), pp. 175, 707
16. A.V. Belokon, A.V. Nasedkin, A.N. Soloviev, New schemes of the finite element dynamic
analysis of piezoelectric devices. Appl. Math. Mech. 66(3), 491–501 (2002)
17. A.V. Nasedkin, The wave field in anisotropic elastic media with complicated properties and
methods of finite element dynamic analysis. DrSc thesis, Rostov State University Press,
Rostov-on-Don. 2001, 271 p. (In Russian)
18. V.A. Krasilnikov, V.V. Krylov, Introduction to Physical Acoustics (Nauka, Moscow,1984),
403 pp. (In Russian)
Energy Absorption Characteristics
of AA7075-T6 Tube Filled with Aluminum
Foam
S. Vignesh, C. Lakshmana Rao and Simhachalam Bade
Abstract In this article, energy absorption and dynamic compressive stress–strain

behavior of aluminum foam-filled and empty tube of aluminum alloy AA7075-T6
were investigated using a split Hopkinson pressure bar (SHPB) technique. Both
quasi-static and dynamic compression tests were conducted on closed-cell alu-
minum foam specimens having a relative density of 0.16. The experimental results
showed that the plateau stress and energy absorption of foam-filled aluminum alloy
AA7075-T6 tube specimen increase with increase in strain rate. The
load-deformation characteristics, failure modes, and energy absorption capacity of
different structures under dynamic loading were investigated. At higher strain rates,
the energy absorption capacity of foam-filled aluminum alloy tubes was found to be
increased, which is useful for crashworthy applications.

Keywords SHPB Aluminum alloy Energy absorption Crashworthiness

ALPORAS Crushable foam
Nomenclature
q* Density of closed-cell aluminum foam
rpl Plateau stress
eD Densification strain
W Energy absorption
Ai Cross-sectional area of incident bar
At Cross-sectional area of transmitted bar
C0 Longitudinal wave speed
S. Vignesh (&) C. Lakshmana Rao S. Bade

Department of Applied Mechanics, Indian Institute of Technology Madras, Chennai, India
e-mail: am14m016@smail.iitm.ac.in
S. Bade
Corporate Technology Centre, Tube Investment of India Ltd., Avadi, Chennai, India

https://doi.org/10.1007/978-981-10-7197-3_26
304 S. Vignesh et al.
1 Introduction
Aluminum alloys are mostly used in the automotive and aerospace industries
because of their improved strength-to-weight ratios, which are a key factor for fuel
economy. Aluminum alloys are also an ideal material for the energy absorption and
crashworthiness application. Mostly metallic circular cylindrical aluminum alloy
tubes are used as energy absorbers because of low weight and ease of manufac-
turing process. In recent days, aluminum foams have shown interesting properties
for shock absorption and crashworthiness applications in aviation and automotive
sectors. In particular, many works related to the energy absorption of foam-filled
aluminum tubes have been published in the recent years [1–3]. Raj et al. [4] studied
the dynamic crushing behavior of closed-cell aluminum foam and identified the
energy absorbed by the aluminum foam found to be higher in dynamic compres-
sion. Mukai et al. [5] studied compression behavior of closed-cell aluminum foam
at a strain rate of 2.5 103 s−1. He found that the energy absorption of aluminum
foam at the dynamic strain rate of 2.5 103 s−1 is 50% greater than that at the
quasi-static strain rate (1 10−3 s−1). Hanssen et al. [6] investigated the energy
absorption of aluminum foam-filled square aluminum extrusions subjected to a
quasi-static load. It was reported that there was an increase in specific energy
absorption by using foam-filled square aluminum extrusions. Hall et al. [7] com-
pared the effects of aluminum foam in brass, aluminum, and titanium tubes and
found that the energy absorption was highest in the aluminum tube. In this paper,
the closed-cell aluminum foam produced by liquid melt route using titanium
hydride (TiH2) was studied for strain rate sensitivity using SHPBtechnique. The
energy absorption of empty and foam-filled AA7075-T6 tubes was studied under
dynamic compression at high strain rates using SHPB technique.
2 Materials and Measurements
Aluminum foams were prepared by melting aluminum and then adding 1–2% of
TiH2 in the form of 5–20 µm diameter particles. The viscosity of the melt was
raised by adding 1.5% of calcium. As soon as TiH2 was added in the melt, the
stirring process was withdrawn, and foam was allowed to form. The melt was then
cooled to solidify the foam.
The compression test specimens were cut carefully by electro-discharge
machining from prepared bulk foam. The regular square prisms of size 20 mm
20 mm 40 mm were used for the quasi-static test. The quasi-static compression
tests were done at a constant strain rate of 1 10−3 s−1 in a universal testing
machine as shown in Fig. 1a. The foam specimens were compressed up to
approximately 70% of their nominal strain. The six specimens were weighed
accurately to determine its relative density (q*/qs), and the average was estimated
around 0.16 (16%). The tomographic images were used to characterize the cell
Energy Absorption Characteristics … 305
Fig. 1 a Quasi-static compression setup with the specimen. b Tomographic image of aluminum
foam
Fig. 2 Empty and aluminum foam-filled AA7075-T6 tubes
shape and size. The typical tomographic image is shown in Fig. 1b. The mean cell
size was found to be 3.5 mm. The regular square prisms of size 13 mm 13
mm 10 mm were used for dynamic compression of foam specimen.
Al alloy AA 7075-T6 having an inner diameter and the outer diameter of 19 and
24 mm was cut to a length of 30 mm. The inner surfaces of the tubes were wire
brushed. The prepared foam was cut into cylindrical specimens by wire cut EDM
process. Each tube was filled by a cylindrical foam specimen of diameter 24 mm
and length 30 mm. Foam core and tubes were pasted using epoxy. Empty and
foam-filled AA7075-T6 tube for dynamic compression loading is shown Fig. 2.
3 Experimental Methods
3.1 Dynamic Compression of Aluminum Foam
Kolsky [8] developed the split Hopkinson pressure bar (SHPB) technique to
characterize the dynamic mechanical behavior of solid materials at high strain rates.
Mostly solid bars are used as incident and transmitted bars in SHPB apparatus, and
the cross-sectional areas of the incident bar (Ai) and transmitted bar (At) are same.
When low-strength materials like aluminum foams are tested with a conventional
split Hopkinson pressure bar, the transmitted pulse will have very low amplitude
and is difficult to record. Chen et al. [9] reported the use of a hollow transmitted bar
when testing low-strength materials like aluminum foam to amplify the strength of
the transmitted pulse. In the present study, the aluminum incident bar used was
1.25 m long and 19.5 mm in diameter. The striker had a length of 350 mm and a
diameter of 19.5 mm. A hollow aluminum transmitter bar of length 1.5 m with an
area ratio of At/Ai = 0.3 was used. At the end of the hollow transmitted bar, an end
cap was press fitted and its effect on the transmitted signal is negligible. The
displacement of the incident bar–specimen interface is obtained from the incident
(ei) and reflected (er) pulse. The axial displacement u1 is given by
Zt
u1 ¼ C0 ðei er Þdt ð1Þ
0
Similarly, the displacement of the specimen-transmitted bar interface is calcu-

lated from the transmitted (et) pulse. The axial displacement u2 is given by
Zt
u2 ¼ C0 et dt ð2Þ
0
where C0 is the longitudinal wave speed in the bar and is given by

sffiffiffiffi
E
C0 ¼
q
The compressive strain in the specimen esp is given by

u1 u2
esp ¼ ð3Þ
Ls
Substituting Eqs. (1) and (2) into (3) provides

Zt
C0
2sp ¼ ðei er et Þdt ð4Þ
Ls
0
where Ls is the length of the specimen. The forces P1 and P2 acting at the incident
bar-specimen interface and specimen-transmitted bar interface are given by
P1 ¼ EAi ðei þ er Þ
and
P2 ¼ EAt et
where Ai and At are the cross-sectional area of the incident and transmitted bars. The
specimen is in equilibrium under the action of the above forces ðP1 P2 Þ, we can
write
Ai
et ¼ ðei þ er Þ ð5Þ
At
Substituting Eq. (5) into (4) provides the expression for the compressive strain in
the specimen when the hollow transmitter bar is used in SHPB apparatus as
2 3
Zt Zt
C0 4 Ai Ai
esp ¼ 1 ei dt 1 þ et dt5 ð6Þ
Ls At At
0 0
The stress acting on the specimen (rsp ) is given by
P2 At
rsp ¼ ¼ Eet
Asp Asp
where Asp is the cross-sectional area of the specimen. The specimen strain rate was
obtained by evaluating the slope of the linear portion of the stain versus time plot.
Modified SHPB bar set up is shown in Fig. 3.
3.2 Dynamic Compression of Foam-Filled Tubes
Dynamic compression test on empty and foam-filled AA7075 tubes were carried
out using conventional split Hopkinson pressure bar as shown in Fig. 4. Maraging
steels were used for striker, incident, and transmitted bar. Properties of material
used in the striker and the incident and the transmitted bar are listed in Table 1.
Fig. 3 Modified split

Hopkinson pressure bar set up
with foam specimen
Fig. 4 Conventional Split Hopkinson pressure bar arrangement for testing foam-filled tubes
Table 1 Properties of striker, incident, and transmitter bar

Material Bar diameter Young’s modulus Poisson’s Wave speed
(mm) (GPa) ratio (m/s)
Maraging 50 200 0.3 5000
steel
Quasi-static stress–strain curve of aluminum foam showed three different modes of

deformation; namely the initial elastic deformation, after which the foam cells start
to buckle and finally the densification or failure mode. Densification takes place
Fig. 5 Compressive stress versus nominal strain curve
when the foam was fully compressed. The compressive stress–nominal strain curve
obtained for the foam is shown in Fig. 5.
The most important mechanical properties obtained from quasi-static compres-
sion tests are summarized in Table 2.
Dynamic compression experiments of aluminum foam showed that there is strain
rate sensitive behavior in closed-cell aluminum foam. Factors, which cause strain
rate sensitivity of closed-cell aluminum foam, are as follows,
• Cell wall material
• Micro-inertia
• Air-compressed pressure
• Localized deformation.
In aluminum foam, there is no strain rate sensitivity of cell wall material. Hence,
it just needs to consider the effect of inertia, air-compressed pressure, and localized
deformation. To estimate the strain rate sensitivity, the plateau stress was plotted
against the strain rate in a log–log plot (see Fig. 6b). The strain rate sensitivity of
closed-cell aluminum foam was found to be 0.3 from the slope of the log–log plot.
SEM photograph of a deformed cell wall is showed that the foam cell wall
fractured due to tension and shear stress induced by stretching cell faces. The tensile
Table 2 Mechanical properties of aluminum foam under quasi-static compression

Foam density, Relative Young’s Plateau stress, Densification
q* (g/cm3) density, q*/qs modulus, E rpl (MPa) strain, eD
(GPa)
0.4191 0.16 0.7 2.5 0.57
Fig. 6 a Stress–strain curve of an aluminum foam tested under dynamic and quasi-static
conditions. b Strain rate sensitivity of aluminum foam
Fig. 7 Tensile and shear failure modes in aluminum foam cell wall
stress r1 developed is perpendicular to the compression axis. Sketch of the tension

and shear failure of closed-cell aluminum foam cell wall is illustrated in Fig. 7. The
fractured cell wall of deformed closed-cell aluminum foam at 20% of strain in
quasi-static uniaxial compression is shown Fig. 8.
Filling foam inside AA7075 tubes significantly increases the force necessary to
deform the tubes. Compressive stress–strain curve of empty and foam-filled tube at
different strain rates is shown in Fig. 9. Energy absorption of foam-filled tube is
higher than the empty tube, and filling foam inside the tube also increases the strain
rate sensitivity of the tube. At higher strain rate, foam-filled tube showed an
increase in stress, which is useful for the crashworthy application.
When the specimen is subjected to compression, the energy absorption capacity
is defined as the energy necessary to deform a given specimen to a specific strain.
So the absorption energy per unit volume of a sample, up to any particular strain eo,
can be evaluated by integrating the area under the stress–strain curve as given by
the expression:
Fig. 8 SEM photograph of undeformed and deformed closed-cell aluminum foam
Fig. 9 Stress–strain response

of empty and foam-filled tube
at different strain rates
Ze0
w¼ rðeÞde
0
Energy absorption capacity empty and foam-filled tubes subjected to high strain
rate compression up to 5% strain as shown in Figs. 10 and 11.
Fig. 10 Energy absorption of empty and foam-filled tube at different strain rates
Fig. 11 Deformed shape of empty and aluminum foam-filled AA7075-T6 tubes after dynamic
compression
5 Conclusion
Closed-cell aluminum foams with a relative density of 0.16 were prepared by the
decomposition of TiH2 in molten metal. The behavior of foam subjected to
quasi-static and dynamic compression showed that the foam with a relative density
of 0.16 was sensitive to strain rate because of localized deformation and
micro-inertial effect. Energy absorption of foam-filled tube is considerably higher
than empty tube, and energy absorption capacity of foam-filled tubes increases with
strain rates. The higher energy absorption capacity at higher strain rate is a useful
property for using this material for lightweight impact energy absorption
applications.
Acknowledgments The authors acknowledge Professor P. Venkitanarayanan from Indian

Institute of Technology, Kanpur for valuable discussions and collaboration.
References
1. M.F. Ashby, A.G. Evans, N.A. Fleck, L.J. Gibson, J.W. Hutchinson, H.N.G. Wadley, Making
Metal Foams, Metal Foams (2000)
2. J. Banhart, J. Baumeister, M. Weber, Damping properties of aluminium foams. Mater. Sci.
Eng., A 205(1–2), 221–228 (1996)
3. Aluminum Foam Technology Applied to Automotive Design, Cymat Corporation (2004)
4. R.E. Raj, V. Parameswaran, B.S.S. Daniel, Comparison of quasi-static and dynamic
compression behavior of closed-cell aluminum foam. Mater. Sci. Eng., A 526, 11–15 (2009)
5. T. Mukai, H. Kanahashi, T. Miyoshi, M. Mabuchi, T.G. Nieh, K. Higashi, Experimental study
of energy absorption in a close-celled aluminum foam under dynamic loading.
ScriptaMaterialia 40(8), 921–927 (1999)
6. A.G. Hanssen, O.S. Hopperstand, M. Langseth, Validation of constitutive models applicable to
aluminium foams. Int. J. Mech. Sci. 44(2), 359–406 (2002)
7. I.W. Hall, M. Guden, C.J. Yu, Crushing of aluminum closed cell foams: density and strain rate
effects. ScriptaMaterialia 43(6), 515–521 (2000)
8. H. Kolsky, An Investigation of the Mechanical Properties of Materials at very High Rates of
Loading. Proceedings of the Physical Society. Section B 62(11), 676–700 (1949)
9. W. Chen, B. Zhang, M.J. Forrestal, A split Hopkinson bar technique for low-impedance
materials. Experimental Mechanics 39(2), 81–85 (1999)
Strain Intensity Factor and Interaction
of Parallel Rigid Line Inclusion in Elastic
Matrix Using FEA
Prataprao Patil, S. N. Khaderi and M. Ramji
Abstract When a rigid line inclusion embedded in an elastic matrix is subjected to

an external load, stress singularity is generated at the tips. The magnitude of this
singularity can be quantified in terms of a strain intensity factor rather than a stress
intensity factor. The principal reason is that the strain intensity factor is independent
of the material properties of the matrix. The strain intensity factor can be analyti-
cally calculated for the case of single inclusion. However, for two parallel rigid line
inclusions, the strain intensity factor analytical expression is not readily available.
Numerical calculation of the strain intensity factor for two parallel rigid line
inclusions embedded in an infinite elastic matrix and the effect of distance between
rigid line inclusions on the strain intensity factor is the objective of this paper. To
this end, first the stress and displacement fields near the inclusions are calculated
using the finite element method (FEM). Then, we use a numerical method based on
the reciprocal theorem to calculate the strain intensity factor. It is found that the
strain intensity factor is equal to that of the single rigid line inclusion case when the
distance between the two parallel rigid inclusions is more than twice their length.
Keywords Strain intensity factor Rigid line inclusion FEM

Singularity
Nomenclature
E Young’s modulus of the matrix material
2l Length of rigid line inclusion
2w Width of the matrix containing rigid line inclusions
2h Height of the matrix containing rigid line inclusions
P. Patil S. N. Khaderi M. Ramji (&)

Engineering Optics Lab, Department of Mechanical and Aerospace Engineering, IIT
Hyderabad, Kandi 502205, Telangana, India

https://doi.org/10.1007/978-981-10-7197-3_27
316 P. Patil et al.
nj Unit vector normal to contour

xi Coordinates in the ith direction, i = 1, 2
r, h Variables defining cylindrical coordinate reference system at an inclusion
tip
k Order of singularity
m Poisson’s ratio of the matrix material
rij, ui Components of the stress tensor and displacement vector
rij ; ui Components of the auxiliary stress tensor and displacement vector
Lcr Critical buckling load
1 Introduction
Nowadays, composite materials find wide application in aerospace, marine, auto-

mobile, and packing sectors. Both continuous and short fiber composites are used
for structural applications. Short fiber composites usually use discontinuous or short
fibers as reinforcing elements. They are used instead of plain polymeric components
in electrical, packaging, and automobile applications. These composites are
developed to fill the property gap between continuous fiber laminate and unrein-
forced polymer. The short fiber composites are used in secondary load-bearing
structures. The load transfer mechanism is attributed to the interfacial shear stresses
at the fiber/matrix interface. The fibers in short fiber composites have random
orientation and spacing. The spacing in between fibers influences the structure
performance in terms of damage and load-bearing capacity. As a step toward
understanding the interaction between the fibers, the problem of two parallel rigid
line inclusions embedded in an elastic matrix is studied. The fiber is modeled as a
rigid line inclusion. This is a reasonable assumption since the elastic modulus of
fiber is much larger than that of the matrix material.
The elastic field near a rigid line inclusion, subjected to a remote load, is known
for isotropic and anisotropic matrix materials [1–5]. Few studies exist on charac-
terizing fracture parameter for rigid line inclusion using analytical and experimental
approach [6]. The objective of this paper is numerical estimation of the strain
intensity factor near the tip of an inclusion (inclusion 1), embedded in an isotropic
elastic matrix, when another similar inclusion (inclusion 2) is placed parallel to it,
see Fig. 1. Now, the strain intensity factor near the tip of the inclusion 1 is influ-
enced by the remote load as well as the distance between the inclusions.
The article is organized as follows. In Sect. 2, we detail the boundary value
problem to be solved. In Sect. 3, we describe the numerical method used to cal-
culate the strain intensity factor. In Sects. 4 and 5, we present the results and
concluding remarks, respectively.
Strain Intensity Factor and Interaction of Parallel Rigid … 317
Fig. 1 General configuration of two parallel rigid line inclusions in an elastic matrix subjected to
traction
2 Problem Definition
Two parallel rigid line inclusions embedded in an elastic isotropic matrix are
considered, see Fig. 1. The length and width of specimen are 2w and 2h, respec-
tively. The rigid line inclusions have length 2l and are separated by a distance
2b. The isotropic matrix material has Young’s modulus (E) and Poisson’s ratio (m).
The boundary conditions applied are as follows:
tx ðw; x2 Þ ¼ P;
u1 ðw; x2 Þ ¼ 0; ð1Þ
u2 ðw; hÞ ¼ 0;
where tx is the traction along the x1 direction, and ui ði ¼ 1; 2Þ is the displacement

along the i direction.
An elastic analysis of the rigid line inclusion embedded in an elastic matrix
reveals that the stress and strain fields are singular at the tip of inclusion. In fracture
mechanics, such a singularity is quantified in terms of stress intensity factor. In our
previous work [7], we have shown that it is appropriate to quantify the singularity in
terms of a strain intensity factor defined as
pffiffi
KIe ¼ lim err ðh ¼ 00 Þ r ; ð2Þ
r!0
where r and h are the coordinate system shown in Fig. 1, KIe is the strain intensity
factor, and err is the radial strain. It was noted that the strain intensity factor is
independent of matrix material property [8].
318 P. Patil et al.
3 Numerical Method for Estimating the Strain Intensity

Factor
In this section, the method for strain intensity factor estimation is discussed. The
strain intensity factor is calculated by using the elastic full-field solution and an
auxiliary asymptotic solution in the reciprocal theorem. The elastic full-field
solution is obtained from finite element method. To proceed further, we discuss the
asymptotic solution below.
The asymptotic field near the tip is of the form
rij ¼ KIe r k1 fij ðk; hÞ;

ð3Þ
ui ¼ KIe r k gi ðk; hÞ;
where (i, j) = (r, h), fij and gi are functions of k and h, r is the distance from tip of
inclusion, and KIe is the intensity of singularity, i.e., the strain intensity factor, and k
is the order of singularity.
The strain intensity factor for arbitrary geometry is estimated using reciprocal
theorem. The reciprocal theorem can be stated as
I
rij ui rij ui nj dS ¼ 0; ð4Þ
C
where nj is the unit normal

to the contour C, (rij, ui) are the actual stress and

displacement fields, and rij ; ui are suitably chosen auxiliary asymptotic fields
(with a unit strain intensity factor) that satisfy boundary conditions. The contour
C can be subdivided into four parts. Equation 4 vanishes over C1 and C3 because of
displacement boundary conditions (See Fig. 2). Equation 4 can be rewritten as,
Fig. 2 Contour surrounding the tip of inclusion to evaluate strain intensity factor
Z Z
rij ui rij ui nj dS ¼ rij ui rij ui nj dS ð5Þ
C2 C4
Substituting the actual elastic fields (in which KIe is unknown) and auxiliary
fields in left-hand side of Eq. 5, we get
Z
1
KIe ¼ rij ui rij ui nj dS; ð6Þ
c
C4
where
Zþ p

c¼ fij ðkÞgi ðk Þ fij ðk Þgi ðkÞ nj dh: ð7Þ
p
The value of c can be calculated by numerical integration. To calculate KIe from

Eq. 6, we need to know the actual stress fields. These are obtained by performing a
finite element analysis. For more details on the implementation of numerical
method, the reader is referred to [7, 9].
4 Numerical Analysis
As mentioned above, the finite element model is used to obtain stress and dis-
placement field that will be used to estimate the strain intensity factor. The finite
element (FE) model consists of two parallel rigid line inclusions in an elastic matrix
(see Fig. 3). The FE simulations are performed using ABAQUS (version-6.9) in a
plane strain setting. The analysis is done with 8-nodded plane strain elements. The
dimensions for model are a = 10 mm, w = 100 mm, and h = 100 mm. The mesh
pattern around inclusion tip is fine enough to capture the high stress gradients near
the inclusion tip. Quarter point elements are used to capture square root singularity
surrounding the tip of inclusion. A mesh convergence study has been performed to
arrive at the number of element surrounding the tip of inclusion. In the simulations
reported, we use 75 elements in circumferential and 60 in radial direction as shown
in Fig. 3. The rigid inclusion is modeled as line using constraint equations. The
boundary conditions applied are shown in Fig. 1. The magnitude of the applied
traction is 0.1 MPa. The elastic properties of the matrix are taken to be
E = 7.95 MPa and m = 0.45.
320 P. Patil et al.
Fig. 3 Finite element mesh used for simulations
To qualitatively understand the interaction of two parallel line inclusions, we

compare the contours of maximum shear stress for two inclusion spacing:
2b = 20 mm and 2b = 5 mm. The results are shown in Fig. 4. Comparison of the
contour plots reveals the following aspects. The inclusion spacing has an effect on
the stress distribution due to interaction between the elastic of the individual
inclusions. However, the maximum stress always occurs at inclusions tip despite
the spacing between the successive inclusions. This is obvious because of the stress
singularity at inclusion tip.
Photoelasticity experiments are performed to qualitatively verify the stress dis-
tribution obtained from FE simulations. For experiment, a specimen of epoxy
matrix with two parallel rigid line inclusions of length 20 mm and spaced 5 mm
apart is fabricated. The conventional photoelastic experiment is carried out.
Fig. 4 Maximum shear stress distribution surrounding the rigid line inclusion from FEM for
different spacing in line inclusions a 2b = 5 mm, b 2b = 20 mm
Fig. 5 Comparison of maximum shear stress contour for rigid line inclusion of spacing
2b = 5 mm between a FEM and b photoelastic fringes
Fig. 6 Ratio of strain

intensity factor considering
interaction effect for two
parallel rigid line inclusions
Figure 5b shows the isochromatic fringe contours surrounding rigid line inclusion.
The isochromatic contours correspond to the principle stress difference, i.e., max-
imum shear stress. From Fig. 5, we see that the qualitative features of the maximum
shear stress distribution are nicely captured by FE simulations.
The strain intensity factors for different values of inclusion spacing are esti-
mated. The strain intensity factor of two parallel inclusions is normalized with
respect to single rigid line inclusion. The normalized strain intensity factor with
respect to b/l ratio for parallel rigid line inclusion in an elastic matrix is shown in
Fig. 6.
The magnitude of the strain intensity factor increases with decreasing distance
between the line inclusions. Strain intensity factor seems to asymptotically
approach that of the single rigid line inclusion when distance between the two
parallel rigid inclusions is more than their length.
322 P. Patil et al.
6 Conclusion
We have numerically estimated the strain intensity factor when two parallel rigid
line inclusions are subjected to external loading. It is found that the strain intensity
factor is equal to that of the single rigid line inclusion case when the distance
between the two parallel rigid inclusions is more than their length. For small values
of inclusion spacing (b = 0.25l), the strain intensity factor is amplified by 25%.
References
1. C. Atkinson, Some ribbon-like inclusion problems. Int. J. Eng. Sci. 11(2), 243–266 (1973)
2. L. Ni, N.S. Nasser, A general duality principle in elasticity. Mech. Mater. 24, 87–123 (1996)
3. Z.Y. Wang, H.T. Zhang, Y.T. Chou, Characteristics of the elastic field of a rigid line
inhomogeneity. J. Appl. Mech. 52, 818–822 (1985)
4. D. Radaj, S. Zhang, Loading modes and stress intensity factors at rigid layer edges. Eng. Fract.
Mech. 46(2), 199–208 (1993)
5. P. Pingle, J. Sherwood, L. Gorbatikh, Properties of rigid-line inclusions as building blocks of
naturally occurring composites. Compos. Sci. Technol. 68, 2267–2272 (2008)
6. G. Noselli, F.D. Corso, D. Biboni, The stress intensity near a stiffener disclosed by
photoelasticity. Int. J. Fract. 166, 91–103 (2010)
7. P. Patil, S.N. Khaderi, M. Ramji, Numerical estimation of strain intensity factors at the tip of a
rigid line inclusion embedded in a finite matrix. Eng. Fract. Mech. 172, 215–230 (2017)
(Online Available)
8. A. Akisanya, N. Fleck, Interfacial cracking from the free edge of a long bi-material strip. Int.
J. Solids Struct. 34(13), 1645–1665 (1997)
9. P. Patil, S.N. Khaderi, M. Ramji, Finite geometry correction factor for stress singularities of
rigid line inclusion, in International Conference on Computer Aided Engineering (CAE-2015),
pp. 269–272 (2015)
Part V
Creep-Fatigue
Creep-Fatigue Damage Assessment
of Reactor Hot Pool Components
During Crash Cooling
Rosy Sarkar, S. Jaladeen and K. Velusamy
Abstract In the prototype fast breeder reactor (PFBR), when there is a reactor
SCRAM, the shutdown takes place in two phases. First, there is hot shutdown, i.e.
the hot pool sodium temperature is rapidly brought down from 820 to 623 K in
25 min, and the II phase is to proceed to the cold shutdown (453 K isothermal)
condition gradually by deploying operating grade decay heat removal system
(OGDHR). The controlled cooling involving high and varying operating mode of
suction pressure of the pump makes the pump design complex. In view of this, the
requirement of controlled cooling has been relooked in the commercial breeder
reactor (CBR), and it is proposed to deploy the OGDHR only after reaching cold
shutdown. Thermo-mechanical analysis of the reactor components, i.e. control plug
(CP), inner vessel (IV) and intermediate heat exchanger (IHX) has been carried out
using a finite element (FE) code. The approach followed for writing the program
(imposing the fall in sodium level (*16 mm/min) at the free surface along with the
drop in temperature (*7 K/min) and the creep-fatigue damage assessment for each
case has been discussed in this paper.

Keywords OGDHR Crash cooling CBR Control plug Inner vessel
Intermediate heat exchanger
Nomenclature
TNa Sodium temperature
TAr Argon cover gas temperature
lti Inner surface top length
lbi Inner surface bottom length
lto Outer surface top length
lbo Outer surface bottom length
ΔL Level change
L Sodium level
R. Sarkar (&) S. Jaladeen K. Velusamy

Indira Gandhi Center for Atomic Research, Kalpakkam 603102, Tamil Nadu, India
e-mail: rosysarkar@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_28
326 R. Sarkar et al.
1 Introduction
In the prototype fast breeder reactor (PFBR), the hot pool sodium is maintained at
820 K during normal operating condition. When there is a reactor SCRAM, the
shutdown takes place in two phases. In the I phase, the hot pool sodium is rapidly
cooled from 820 to 623 K within 25 min by operating the main condensers. The II
phase is to cool down the sodium systems from 623 to 453 K at a controlled rate of
20 K/h by deploying the operating grade decay heat removal system (OGDHR).
When there is controlled cooling, the pumps incorporated in the OGDHR system
face the high suction pressure of the order of 170 bars and vary up to about 15 bars.
The high and varying operating mode of suction pressure of the pump makes the
pump design complex, mainly from the point of view of sealing arrangement. In
view of this, the requirement of controlled cooling rate from 623 to 453 K has been
relooked in the commercial breeder reactor (CBR) and it is proposed to deploy the
OGDHR only after reaching cold shutdown condition [1].
In CBR, the need for an alternate system having a lower capacity for decay heat
removal (DHR) during shutdown state is envisaged. The simplified scheme for
DHR through steam generators has been proposed [1]. During the phase when the
reactor systems are at high temperature (820 K), the cool down takes place at a high
rate (*400 K/h), and subsequently, when the reactor systems are at low temper-
ature (623 K), the cool down is performed at lower rate (100 K/h), when main
condenser is used for cool down. The thermo-mechanical analysis of the reactor
components in primary sodium has been carried out using a finite element
(FE) code. The approach followed for writing the program [imposing the fall in
sodium level (*16 mm/min) at the free surface along with the drop in temperature
(*7 K/min)] and the creep-fatigue damage assessment for each case has been
discussed in the subsequent sections.
2 Reactor Components in the Hot Pool
During crash cooling, the sudden change in the hot and cold pool sodium tem-
peratures induce stresses in the reactor components like control plug (CP), inner
vessel (IV) and intermediate heat exchanger (IHX), which are immersed in the
primary sodium. This results in creep and fatigue damage of the components.
Hence, detailed investigation of the hot pool components has been carried out to
estimate the damage. The critical region in CP, IV and IHX is the sodium-free level
because of the continuous change in the sodium level and sodium temperature. The
geometrical details of the CP, IV and IHX are given in Fig. 1 [2], Fig. 2 [3] and
Fig. 3 [4], respectively.
Creep-Fatigue Damage Assessment of Reactor Hot Pool … 327
Fig. 1 Vertical section of CP
USP
Junction
Fig. 2 3D details of IV
3 Loading Conditions
Following SCRAM, the sodium temperature is brought down by operating the main
condensers. With a decrease in sodium temperature, the density of sodium increases
and thereby the volume decreases, resulting in continuous fall in sodium level with
fall in sodium temperature.
3.1 Change in Sodium Level
Following SCRAM, the initial sodium level (at 820 K) is at 27,400 mm. During the
I phase, when the sodium is cooled to 623 K in 25 min, the sodium level reaches to
27,000 mm. Thus, the average rate of change in the sodium level is 16 mm/min.
During the II phase when the sodium at hot shutdown condition (623 K) is
cooled to the cold shutdown condition (453 K), the sodium level falls from 27,000
Fig. 3 Vertical section of

IHX
to 26,700 mm at 100 K/h (1.67 K/min). Thus, the average rate of change in sodium
level is 3 mm/min.
The sodium level change is same for all the hot pool components.
3.2 Change in Sodium Temperature
The evolution of hot and cold pool temperature, following SCRAM, in the case
without the deployment of OGDHRS is shown in Fig. 4 [5]. The hot pool
Fig. 4 Evolution of hot and cold pool temperature following SCRAM, without deployment of
OGDHR
temperatures given in Fig. 4 are the input for both inside and outside the CP
(12 mm thick). The hot pool and cold pool sodium temperatures, shown in Fig. 4,
are given as input for the inside and outside surfaces of the inner vessel (15 mm
thick), respectively.
The IHX shell (16 mm thick) is filled in with secondary sodium, and the sleeve
valve (5 mm thick) is surrounded by primary sodium inside (annulus—50 mm) and
outside. There is argon cover gas above the primary sodium in the annulus and
outside the sleeve. Following SCRAM, the temperatures at each surface during hot
shutdown (I phase) are as shown in Fig. 5. The temperature drop during cold
shutdown (II phase) is 100 K/h.
580
TSLo, °C Sleeve out
555
TSLi, °C Sleeve in
530
TSHo, °C Shell Out
Temperature (C)
505
TSHi, °C Shell in
480
455
430
405
380
355
330
0 500 1000 1500 2000
Time (s)
Fig. 5 Temperatures at shell and sleeve in IHX

4 Flow Chart for Writing FE Program
The generalized approach followed to obtain the stresses and temperature distri-
bution at every minute, at the transient with varying sodium level and sodium
temperature is shown in the flow chart in Fig. 6.
Start
Axisymmetric Model
Material Properties (with change in temperature)
Define lti, lto, lbi, lbo, l, li.i, lo.i

lti-top-in;lbi-bottom-in;lto-top-out;lbo-bottom-out,∆L-level change
L-sodium level; li.i-mesh size in; lo.i-∆L mesh out; .i-iteration
Time j=0; i=0
No
j < 25
Yes
No Yes
(lbi - li.i) >
L- (j*∆L)
lti=lti lti=lti+li.i; lto=lto+lo.i

lto=lto lbi=lbi-li.i; lbo=lbo-lo.i
lbi=lbi
lbo=lbo
i=i+1
TAr-Argon Temperature, TNa-Sodium Temp
Extract Temperature and Stresses
TNa =TNa-∆TNa.i, j = j+1

TAr = (((393 + TNa) / 2) + TNa) / 2
End
Fig. 6 Flow chart for writing a finite element program to simulate the change in length and
temperature of primary sodium
An axisymmetric solid element model for each component has been modelled,
and the radial components have been arrested for simulating the boundary condi-
tions. The material for the construction of CP, IV and IHX is SS 316 LN. Density
has been conservatively taken at 423 K, and all the other material properties (except
conductivity) have been taken at 823 K. The change in the conductivity values with
a change in temperature has been incorporated.
The length of the component that is immersed in sodium is categorized as
bottom (lbi and lbo), and the rest is the top region (lti and lto), where ‘i’ and ‘o’ are
inner and outer surfaces. ‘L’ is the initial sodium level, and ΔL is the change in
sodium level at each iteration.
Convection heat transfer is considered, with h = 10 W/m2 K at the top region of
the component that is surrounded by argon cover gas. Conduction heat transfer is
considered for the bottom portion which is surrounded by sodium, which has a very
high heat transfer coefficient.
5 Control Plug (CP)
5.1 Methodology
During the I phase (i.e. to reach the hot shutdown condition), there is *400 mm
fall in sodium level and the maximum stresses are expected at this region (shown in
blue in Fig. 7a).
During the II phase, there is *300 mm fall in sodium level and this is the region
of interest as shown in Fig. 7b (in pink). The zones A, B, C in I phase and A′, B′, C′
in II phase are shown in Fig. 7a, b, respectively. Both the phases are modelled
separately, and the stresses are evaluated separately (because of different loading
conditions), and the creep-fatigue damages at each location are added to get the
final creep-fatigue damages.
5.2 I Phase (To Reach Hot Shutdown Condition)
Transient analysis of the CP with the above-mentioned loading and boundary

conditions has been carried out. The length considered for this analysis is from
elevation 27,500 to 26,900 mm. The whole length is divided into 10 sub-zones
namely 1S, 2S, 3S, 4S, 5S, 6S, 7S, 8S, 9S, 10S, where 1S is the bottom-most
sub-zone and 10S is the topmost sub-zone. The evolution of temperature and stress
is shown in Fig. 8a, b, respectively. The stress and temperature distribution curves
for each sub-zone are extracted. The temperature distribution curve and the stress
distribution curve at the maximum damaged sub-zone (in zone-B) are shown in
Fig. 9a, b, respectively.
(a) (b)
Maximum
Damage A’
Location 400 mm
B’
C
300 mm
C’
Fig. 7 a Categorization of zones in I phase, b categorization of zones in II phase
(a) (b)
At time (t) = 0 s t = 1500 s At time (t) = 0 s t= 1500 s
Fig. 8 a Temperature evolution in CP during I phase shutdown, b stress evolution in CP during I

phase shutdown
(a) 840 (b)

180
820
Temperature (K)
160
Temperature (K)
800 Stress (Mpa)
140
Temperature (K) 120
780
100
760 80
740 60
40
720
20
700 0
0 250 500 750 1000 1250 1500 1750 2000 0 250 500 750 1000 1250 1500 1750 2000
Time (s) Time (s)
Fig. 9 a Temperature versus time (CP, I phase), b stress versus time (CP, I phase)
Table 1 Creep-fatigue damage for CP during I phase shutdown

Zone (sub-zone) Creep effect Fatigue damage Creep damage
A (10S) Significant 6.32E−19 2.88E−06
B (7S) Significant 1.71E−07 6.0E−04
C (1S) Negligible 1.26E−08 –
The zones which are under significant creep and negligible creep are identified,
and the creep and fatigue damages are calculated accordingly. The creep and fatigue
damages for significant creep zones are calculated by taking the maximum stress at
each grid (Fig. 9a) at the corresponding temperature at the corresponding grid
(Fig. 9b), and the cumulative damage is determined. For the negligible creep zones,
the stress range at the maximum temperature is considered and the fatigue damage
for the cycle is calculated directly. The creep and fatigue damage calculations are
done as per RCC MR-RB [6]. Table 1 gives the creep and fatigue damages at the
maximum damaged sub-zones 10S, 7S, 1S at the zones A, B, C, respectively.
5.3 II Phase (From Hot Shutdown Condition to Cold

Shutdown Condition)
Transient analysis has been carried out for the length considered from elevation
27,100 to 26,500 mm. The whole length is divided into 10 sub-zones namely 1S′,
2S′, 3S′, 4S′, 5S′, 6S′, 7S′, 8S′, 9S′, 10S′, where 1S′ is the bottom-most sub-zone
and 10S′ is the topmost sub-zone. The stress distribution curve and the temperature
distribution curve at every location have been extracted. The evolution of tem-
perature and stress in II phase is similar to that of I phase.
As the maximum temperature 623 K is below the creep regime, only fatigue
damage is calculated, taking the maximum stress range at the maximum
Table 2 Creep-fatigue damage for CP during II phase shutdown

Zone (sub-zone) Creep effect Fatigue damage Creep damage
A′ (10S′) Negligible 1.14E−33 1.14E−33
B′ (4S′) Negligible 1.7E−11 1.7E−11
C′ (1S′) Negligible 1.76E−28 1.76E−28
Table 3 Combined Zone (sub-zone) Fatigue damage Creep damage

creep-fatigue damage for CP
during shutdown A + A′ 6.32E−19 2.88E−06
B + A′ 1.17E−07 6E−04
C + B′ 1.262E−08 –
C + C′ 1.26E−08 –
temperature. Table 2 gives the creep and fatigue damages at the maximum damaged
sub-zones 10S′, 4S′, 1S′ at the zones A′, B′, C′, respectively.
5.4 Combined Damage for the I and II Phases
From Fig. 7a, b, it can be observed that, on combining the zones of the two phases,
we get four zones, i.e. A + A′, B + A′, C + B′ and C + C′. The creep and fatigue
damages for these zones are added as shown in Table 3.
Thus, the maximum fatigue and creep damages are found to be at B + A′, with
the values, 1.17E−07 and 6E−04, respectively, for 861 cycles. The maximum
damage location is shown in Fig. 7a.
6 Inner Vessel (IV)
The methodology followed is same as that in CP. The zones categorized in I phase
and II phase are shown in Fig. 10a, b, respectively. The IV, which separates the hot
pool sodium from the cold pool, has 2.25 m height difference between hot and cold
pool sodium during normal operating condition (in CBR). Following SCRAM, the
height difference between hot (lbi) and cold pool (lbo) sodium reduces to 75 mm.
This difference in sodium height has been modelled in the analysis. The temperature
evolution and the difference in hot and cold pool sodium level can be observed in
Fig. 10c.
From Fig. 10a, b, it can be observed that, on combining the zones of the two
phases, we get four zones, i.e. D + D′, E + D′, F + E′ and F + F′. The creep and
fatigue damages for these zones are added and are given in Table 4.
(a) (b)
D
Maximum
Damage D’
Location 400 mm
F E’
300 mm
F’
(c)
lti lto
lbi lbo
At time (t) = 0 s t= 1500 s
Fig. 10 a Zones in I phase in IV, b zones in II phase in IV, c temperature evolution in IV during I
phase shutdown
Table 4 Combined creep-fatigue damage for IV during shutdown

Zone Fatigue damage Creep damage
D + D′ 6.16E−04 2.21E−03
E + D′ 0.0057 0.0019
C + B′ 2.61E−07 –
C + C′ 2.4E−07 –
7 Intermediate Heat Exchanger (IHX)
The intermediate heat exchanger (IHX) transfers thermal power from the primary
sodium which is radioactive to the non-radioactive secondary sodium circuit. There
are four IHX per secondary loop. The methodology followed is same as that in CP.
The zones categorized in I phase and II phase are shown in Fig. 11a, b, respec-
tively. During the I phase of the shutdown, there is *400 mm fall in primary
sodium level, and during the II phase, there is *300 mm fall in primary sodium,
but the secondary sodium level in IHX shell is always constant. There is no level
change inside the shell. There is only temperature change in secondary sodium. For
primary sodium, both level change and temperature changes are considered.
(a) (b)
G J
K G’ J’
400
H
Maximum
Damaged
Location
300
I H’
K’
L
I’ L’
Fig. 11 a Zones in I phase in IHX, b zones in II phase in IHX

Table 5 Combined creep-fatigue damage for IV during shutdown

Zone Fatigue damage Creep damage
G + G′ 7.04E−08 –
H + G′ 1.1E−07 –
I + H′ 8.22E−14 –
I + I′ 8.22E−14 –
J + J′ 9.13E−16 1.94E−05
K + J′ 7.58E−04 1.01E−03
L + K′ 6.78E−07 –
L + L′ 6.78E−07 –
lpot lsit
817 K
220 MPa -4 MPa
lsot 697 K
lpi
lsib
lpob
lsob 237 MPa
636 K 41 MPa
Where, lpot-Shell out top, lpi-Shell in, lpob- Shell out bottom,
lsit- Sleeve in top, lsot- Sleeve out top, lsib- Sleeve in bottom, lsob- Sleeve out bottom
Fig. 12 Temperature evolution in IHX during I phase shutdown
From Fig. 11a, b, it can be observed that, on combining the zones of the two
phases, we get eight zones, i.e. G + G′, H + G′, I + H′, I + I′, J + J′, K + J′, L + K′
and L + L′. The creep and fatigue damages for these zones are added and are given
in Table 5.
The evolution of temperature for IHX is shown in Fig. 12.
8 Conclusion
In the future fast breeder reactor, the requirement of controlled cooling rate from
350 to 180 °C has been relooked to avoid various complexities and it is proposed to
deploy the OGDHR only after reaching cold shutdown condition. Hence, the main
condenser is used to cool down the hot pool sodium from 550 to 350 °C (I phase) at
an average rate of 400 K/h and from 350 to 180 °C (II phase) at the rate of 100 K/h.
The thermo-mechanical design of the shutdown systems, i.e. control plug (CP),
inner vessel (IV) and intermediate heat exchanger (IHX), has been carried out for
the shutdown involving two phases, and the combined creep-fatigue damage at each
location has been estimated. It is found that the maximum fatigue damage values
are 1.17E−07, 0.0057, 7.58E−04 and the maximum creep damage values are 6.0E
−04, 0.0019 and 0.001 for CP, IV and IHX, respectively. As the values are neg-
ligible, the simplified scheme for DHR through steam generators is recommended.
Acknowledgements The authors thank Mr. K. Natesan, Mr. Juby Abraham and Mr.
U. Parthasarathy of Thermal Hydraulics Section, IGCAR, for providing the temperature input for
the reactor components.
References
1. K. Natesan et al., Simplified scheme for DHR through steam generators, Internal Report (2015)
2. R. Sarkar et al., Dependence of control plug life on primary sodium heating rate during power
raising, Internal Report (2015)
3. R. Sarkar et al., Design and life assessment of inner vessel for FBR-1&2. Trans. Indian Inst.
Met. 69(2), 543–547 (2016)
4. R. Srinivasan et al., Conceptual design of intermediate heat exchanger, Internal Report
5. K. Natesan, Integrated plant operations, Internal Report
6. RCC-MR Section I, Subsection B, design and construction rules for class-1 components of
FBR nuclear islands (2010)
The Influence of Prior Plasticity
on the Creep Resistance of Two
Elevated Temperature Power Plant Steels
Valliappa Kalyanasundaram and Stuart R. Holdsworth
Abstract Analysis of the risk of creep–fatigue (C-F) cracking is an integral part of

the condition assessment of elevated temperature structural components operating in
modern power plants. The advanced martensitic steels used in such applications tend
to exhibit a high degree of C-F deformation interaction relative to low-alloy bainitic
steels. Of particular interest to this work is characterising the influence of plasticity
on the creep resistance of two power plant steels from both a qualitative and
quantitative perspective. Uniaxial strain-controlled cyclic/hold C-F tests, with a hold
period at peak tensile strain, have been performed in this context on 1%Cr and 10%
Cr steels at their respective peak operating temperatures of 550 and 600 °C. In
addition to C-F tests, monotonic stress relaxation tests have also been conducted on
both the steels at different elastic/inelastic strains (strain amplitudes). With this
information the influence of magnitude of the loading transients, both monotonic
and cyclic, on the subsequent creep behaviour is systematically studied.
Experimental results indicate that the creep deformation resistance of these steels is
clearly modified as a consequence of prior, but not exclusively cyclic, plasticity. In
particular, creep deformation kinetics are found to be influenced by the magnitude of
the first inelastic loading transient for both the monotonic and cyclic loading his-
tories under consideration. This effect is therefore not limited to the stress relaxation
response during the midlife cycle hold, but also to the creep strain accumulated
during the first cycle hold when preceded by an inelastic loading transient. This work
also examines the determination of forward creep properties from the obtained stress
relaxation data (and vice versa) without and with prior monotonic or cyclic inelastic
loading.

Keywords Creep–fatigue interaction Prior fatigue Creep strength degradation

Creep curve prediction Stress relaxation Cyclic softening
V. Kalyanasundaram (&) S. R. Holdsworth

ICMI, Inspire AG c/o. EMPA, Überlandstrasse 129, 8600 Dübendorf, Switzerland
e-mail: kalyanasundaram@inspire.ethz.ch
S. R. Holdsworth
Swiss Federal Laboratories for Materials Science and Technology (EMPA),
Überlandstrasse 129, 8600 Dübendorf, Switzerland

https://doi.org/10.1007/978-981-10-7197-3_29
342 V. Kalyanasundaram and S. R. Holdsworth
1 Introduction
The thermomechanical loading profile of metallic structural components operating

at elevated temperature typically includes cyclic transients (during start-up and
shut-down) and steady-state operating periods during service. Under such operating
conditions, fatigue, creep and their synergistic interaction become the predominant
modes of material deformation and damage accumulation [1, 2]. As operating
conditions become increasingly flexible, frequent power transients induce enhanced
cyclic plasticity in the material especially in critical locations involving multiaxial
stress concentrations. There is mounting evidence to indicate the significant effects
of prior plasticity on the subsequent creep deformation resistance of several ele-
vated temperature materials [2]. As a consequence, the influence of plasticity and its
magnitude on creep needs to be systematically characterised for more effective
remnant life assessments and for improved design protocols in the future. Since the
temperature and stress handling capability of fossil-fired turbomachinery, for
example, is expected to gradually increase over time so as to minimise environ-
mental pollution and to maximise operating efficiency and flexibility, the require-
ment for such fundamental knowledge becomes more important [3, 4]. The major
objective of the current study is therefore to characterise the effect of plasticity, both
monotonic and cyclic, on creep in two commonly adopted power plant steels that
exhibit cyclic softening at/near their respective peak operating temperatures. Equal
focus is given to obtaining qualitative and quantitative information in this context,
with the aid of experimental data from cyclic/hold, monotonic stress relaxation and
forward creep tests. In particular, the importance of the first loading transient and its
magnitude is analysed with reference to recent observations that implicitly suggest
accelerated creep kinetics as a consequence of the initial few cycles of plastic
deformation in sequential [5, 6] or simultaneous test schemes [7, 8].
The detrimental influence of synergistic creep–fatigue (C-F) interactions from a
damage perspective has been well documented in the literature for various steels
within the broad class of cyclic softening materials [7–13]. However, relatively few
literature exists focussed specifically on obtaining a systematic understanding of the
effect of cyclic plasticity on creep resistance at elevated temperatures [5, 6, 14–17].
These studies clearly show the creep resistance to be significantly reduced as a
consequence of plasticity as evidenced by higher minimum creep-rates in com-
parison with those under pure creep. Other researchers suggest the degree of creep
softening due to plasticity to be much less significant if cycling is responsible for
predominantly elastic deformation [6, 8, 9]. All these experimental observations
thus highlight the need for a better understanding of the influence of prior plasticity
and its magnitude on subsequent creep from a deformation viewpoint for cyclic
softening steels. The experiments conducted in this study explicitly offer clear
qualitative insights into this regard for two materials (1%Cr and 10%Cr steels) that
have been preferred for critical applications in power plants in recent years.
The Influence of Prior Plasticity on the Creep Resistance … 343
Besides qualitative analyses, quantitative comparisons further facilitate a com-

plete characterisation of the experimentally observed material response under dif-
ferent deformation histories. Differences in the nature of stress relaxation during a
tensile hold after elastic/inelastic loading transients either under monotonic or cyclic
plasticity therefore need to be systematically evaluated. Since degradation in
observed creep resistance needs to be compared with that under pure creep, the
amount of stress relaxation evaluated under the different deformation histories
should be converted to equivalent creep curves. In summary, r(t) stress relaxation
curves obtained during the strain-controlled tensile holds under varying stress need
to be converted to equivalent e(t) creep curves under a nominally constant stress.
Conversely, it should also be possible to convert conventional forward creep curves
to equivalent stress relaxation curves to address the same issue. The applicability of
both these approaches will be explored in this work for experimental results
obtained with the 10%Cr steel alone.
Among the existing analytical methodologies [18–22], commonly employed
approaches are those based on the Norton-Bailey (NB) equation either assuming
time- or strain-hardening as appropriate for the application under consideration.
More often than not, the strain-hardening assumption is judged to more closely
simulate material response in practical applications than time-hardening (e.g. [23]),
although concerns exist regarding its adequacy under variable loading conditions
[24].
A brief discussion of the experimental details is provided next followed by a
detailed description of the NB model assuming strain-hardening chosen for study in
this work. The obtained experimental and analytical results are comprehensively
discussed in the section thereafter followed by a summary of the major conclusions.
2 Experimental Details
2.1 Test Materials
A low-alloy 1%Cr steel and an advanced martensitic 10%Cr steel were the test
materials chosen for this study, both exhibiting cyclic softening at/near their
respective peak operating temperatures of 550 and 600 °C. The former had a
tempered bainitic microstructure while the latter a tempered martensitic structure
after quenching and tempering heat treatments. The 10%Cr steels generally exhibit
a 50 °C temperature advantage relative to 1%Cr steels [15], primarily due to
enhanced strengthening against elevated temperature deformation imparted by
chromium-rich carbide and carbonitride precipitates.
2.2 Test Matrix
Standard 8 mm diameter specimens with parallel gauge lengths of 50 mm (creep)

and 20 mm (monotonic and cyclic/hold) were used for the conducted isothermal
tests. Conventional creep tests were performed only on the 10%Cr steel for three
different stress levels (75, 130 and 170 MPa) at 600 °C in accordance with [25]
until the accumulation of 1% inelastic strain.
Uniaxial strain-controlled cyclic/hold tests and monotonic stress relaxation tests
were conducted in ambient environment in an induction heating based
servo-hydraulic test system of 100 kN capacity. The middle (control) thermocouple
was kept to within ± 1 °C of the target temperature during the course of the entire
test. A Class 0.5 side-entry extensometer, with a datum leg spacing of 15 mm, was
used for the measurement and control of axial strain. Loading started
compression-going for the cyclic/hold tests following a trapezoidal waveform under
fully reversed conditions (strain ratio, Re = −1) with a hold at the peak tensile
strain. The cyclic/hold tests were stopped after macroscopic crack initiation within
the gauge length as per the 2% load-drop criterion. In the stress relaxation tests, the
specimens were strained monotonically in tension to a prescribed strain, held
therein for a few hours and unloaded thereafter to zero strain. A detailed summary
of the monotonic and cyclic/hold C-F tests conducted on both the steels is provided
in Table 1. When multiple stress relaxation routines were performed on a single
specimen, only the last loading transient was responsible for macroscopic yielding
while the earlier one(s) were only responsible for elastic deformation.
Table 1 Summary of the experimental test matrix for both the test materials considered in this
work
Loading type Test T (°C) e_ (%/s) Test Strain Hold
material ID amplitude (%) time, th
Cyclic/hold 10%Cr steel 600 0.1 1 ±0.39 15 min
ε 2 ±0.51 15 min
3 ±0.50 30 min
4 ±0.50 60 min
1%Cr steel 550 0.1 5 ±0.50 30 min
6 ±0.50 60 min
Monotonic stress 10%Cr steel 600 0.01 7 +0.075 3h
relaxation +0.15 3h
+0.40 4h
8 +0.075 7h
+0.50 7h
1%Cr steel 550 0.1 9 +0.075 2h
+0.15 2h
+0.40 3h
3 Analytical Method
3.1 NB Strain-Hardening Formulation
One of the most commonly used analytical formulations to represent material

deformation during primary creep is the Norton-Bailey (NB) equation, i.e.
ecr ¼ ANB rn tp ð1Þ
where ecr is the creep strain (in %), ANB is a temperature-dependent constant, n is
the exponent for stress, r and p is the exponent for time, t. To predict forward creep
behaviour from stress relaxation data (and vice versa), a formulation derived from
Eq. (1) assuming strain-hardening is employed in this study. Following this
approach, the instantaneous creep-rate in the material (_ecr ) is assumed to depend on
the current stress and strain irrespective of the current time [26]. A closed-form
expression for the NB strain-hardening formulation is possible only for specific
values of n and p. Hence, n is left as a variable for optimisation with experimental
data while p is kept at 1/3 in this study. Accordingly, the predicted time (tpred: ) for
stress to relax from ri to that at rt can be expressed as shown in Eq. (2):
2 np3 3
rt
6 p
3pn 1 ri 7
1p 6 7
1 1 1 6 6 n
2 7 1
tpred: ¼ n1 6 þ
2p
1 rrti
p
7 when p ¼ ð2Þ
p ANB E r p 6 2pn 7
t 4 np1 7 3
5
pn p
1 rrti
where ri is the stress at the onset of hold (t = 0), rt is the stress at time t and E is the
elastic modulus. To ensure robust fits, the assessment involved both the prediction
of time required for stress to relax from its peak value to that at t and the stress
expected at a given time t. The optimised NB model parameters (ANB , n and p)
assessed from stress relaxation data were then substituted in Eq. (1) for forward
creep curve construction. Similarly, NB parameters assessed from experimental
primary creep data as per Eq. (1) were used in Eq. (2) to obtain stress relaxation
curves at 0.075 and 0.50%.
3.2 Model Parameter Optimisation
The root-mean-square (RMS) error calculated between experimental and correspond-

ing predicted values was minimised using Z, the scatter factor, formulated as [27]:
Z ¼ 10ð2:5RMSÞ ð3Þ
While a Z value of unity implies a perfect fit, values slightly above this are not
unusual as a consequence of data measurement errors and/or to testing uncertain-
ties, especially at elevated temperatures. Parameter optimisation was carried out in
Microsoft Excel® using the SOLVER routine, and a systematic protocol ensured
that the best-fit values were not arbitrarily influenced by the chosen starting values.
4.1 Creep Relaxation Data Comparisons
Figure 1 shows creep relaxation curves obtained for both test materials that have
been stress-normalised to account for the different peak stress values at the onset of
tensile hold periods in monotonic and cyclic/hold tests. The creep response for the
different deformation histories can be qualitatively appreciated from these profiles
by realising that increasing amounts of stress relaxation indicate the propensity of
the material to accumulate more creep strain (i.e. to creep soften) during the hold
times of such tests. The figures clearly indicate the detrimental influence of
increasing strain (strain amplitude), and initial peak stress, on subsequent creep
deformation resistance for both the steels after monotonic and cyclic (cycle-1)
loading. The larger the strain (strain amplitude), and initial peak stress preceding the
hold period, the more the steels creep soften at their respective test temperatures,
with the effect tending to saturate progressively after macroscopic yielding.
Comparison of the stress relaxation profiles obtained for the 10%Cr steel corre-
sponding to +0.50% under monotonic and cyclic (cycle-1) straining to ±0.50% also
clearly shows further creep softening due to reduced flow stresses after strain
reversal. Although not shown here, the existence of such material response due to
the Bauschinger effect is also observed for the 1%Cr steel.
Figure 2 shows stress relaxation profiles obtained during cycle-1 and midlife
cycle tensile holds (60 min) for specimens strained to ±0.50% strain. For both
steels, the component figures also compare observed midlife cycle stress relaxation
responses with those predicted using the appropriate cycle-1 NB parameters,
assuming strain-hardening. For both the steels, the plastic strain accumulated
between cycle-1 and the midlife cycle is apparently responsible for further creep
softening. Hence, the consideration of plasticity effects on creep in both the steels is
critical and detailed microscopy studies are required to systematically understand
the microstructural reasons for the observed material response.
Strain rates observed as a function of instantaneous stress during the cycle-1 and
midlife cycle hold periods of the cyclic/hold creep–fatigue tests are compared with
forward creep test minimum creep-rates for the two steels in Fig. 3. These indicate
the effect of plastic strain accumulated between cycle-1 and the midlife cycle on
15min. T hold (± 0.39%) 30min. T hold (± 0.50%)

1.2 10%Cr 15min. T hold (± 0.51%) 1.2 1%Cr 60min. T hold (± 0.50%)
30min. T hold (± 0.50%)
60min. T hold (± 0.50%) 2hr. T hold (+0.075%)
3hr. T hold (+0.075%)
1 3hr. T hold (+0.15%) 1 2hr. T hold (+0.15%)
4hr. T hold (+0.40%) 3hr. T hold (+0.40%)
7hr. T hold (+0.075%)
7hr. T hold (+0.50%)
0.8 0.8
σ/σi
σ/σi
0.6 0.6
0.4 0.4
0.2 0.2
600˚C 550˚C
0 0
0 2 4 6 8 0 1 2 3 4
Hold Time (hours) Hold Time (hours)
Fig. 1 Normalised stress relaxation response during tensile hold for 10%Cr steel at 600 °C and
1%Cr steel at 550 °C after monotonic and cyclic (cycle-1) straining to different strain amplitudes
(‘+’ represents a tensile-going ramp directly to the indicated strain value at tensile hold, whereas
‘±’ represents compressive-going and tensile-going ramps to the indicated strain value prior to the
tensile hold)
450 500
60min. T hold (± 0.50%) - First Cycle 60min. T hold (± 0.50%) - First Cycle
400 450
60min. T hold (± 0.50%) - Midlife Cycle 60min. T hold (± 0.50%) - Midlife Cycle
350 400
Midlife Cycle Prediction (Cycle 1 NB Parameters) Midlife Cycle Prediction (Cycle 1 NB Parameters)
350
300
Stress (MPa)
Stress (MPa)
300
250
250
200
200
150
150
100 100
50 50
10%Cr 600˚C 1%Cr 550˚C
0 0
0 0.25 0.5 0.75 1 1.25 0 0.25 0.5 0.75 1 1.25
Fig. 2 Experimentally observed stress relaxation responses for 10%Cr steel at 600 °C and 1%Cr
steel at 550 °C during first- and midlife cycle tensile hold times for a cyclic strain range of 1% (the
predicted midlife cycle relaxation profiles were determined using cycle-1 NB parameters to
illustrate the influence of plasticity on creep in both the cyclic softening steels)
primary creep rates, and how the influence is greater for the 10%Cr steel (Fig. 3a)
than for the 1%Cr steel (Fig. 3b).
Defining a short-form terminology for the different deformation histories under
consideration is necessary for brevity in the following text concerned with the
quantitative comparison of the discussed results. The isothermal assessment of stress
relaxation profiles obtained from 10%Cr specimens that were strained monotoni-
cally to either +0.075% or +0.15% preceding the tensile hold is termed NPP (no
prior plasticity), while those strained to either +0.40% or +0.50% are termed 1PT-M
(1 plastic transient—monotonic). Similarly, the isothermal assessment of stress
relaxation profiles obtained from 10%Cr specimens that were strained cyclically first
in compression and then to either +0.39%, +0.50% or +0.51% immediately
(a) 1E+1 (b) 1E+1

15min. T hold (± 0.39%) -
1E+0 First Cycle 1E+0
15min. T hold (± 0.39%) -
30min. T hold (± 0.50%) -
Midlife Cycle
First Cycle
1E-1
Strain Rate (/h)
Strain Rate (/h)

15min. T hold (± 0.51%) - 1E-1
First Cycle 30min. T hold (± 0.50%) -
15min. T hold (± 0.51%) - Midlife Cycle
1E-2 Midlife Cycle 1E-2
60min. T hold (± 0.50%) -
30min. T hold (± 0.50%) -
First Cycle
First Cycle
1E-3 30min. T hold (± 0.50%) - 1E-3 60min. T hold (± 0.50%) -
Midlife Cycle Midlife Cycle
60min. T hold (± 0.50%) -
1E-4 First Cycle 1E-4
60min. T hold (± 0.50%) -
Midlife Cycle
1E-5 1E-5
1E-6 1E-6
10 100 1000 10 100 1000
Stress (MPa) Stress (MPa)
Fig. 3 Comparison of cycle-1 and midlife cycle hold time strain rates (mainly due to primary
creep) with minimum creep-rates from forward creep tests of a 10%Cr steel at 600 °C and b 1%Cr
steel at 550 °C
preceding the tensile hold is termed 2PTC (2 plastic transients—cyclic), with this
terminology being adopted with reference to both cycle-1 and midlife cycle
response. Figure 4 shows creep curves constructed from stress relaxation data using
the NB model, assuming strain-hardening, for the 10%Cr steel at (a) 75 and
(b) 170 MPa. The figures for the two stress levels, at least qualitatively, illustrate a
systematic influence of the plastic strain accumulated by the different cycle types on
creep response. It is acknowledged that any uncertainty in these creep response
predictions would be minimised by the informed consideration of anelastic recovery
(e.g. [23]). In particular, the amount of creep accumulation is observed to be higher
at midlife (2PTC—Midlife curves) than that at cycle-1 (2PTC—Cycle-1 curves),
possibly as a consequence of the steel’s pronounced cyclic softening over time.
The NB model fit parameters for some of these profiles are given elsewhere [28]. In
summary, increasing the strain (strain amplitude) above the limit of proportionality
reduces the creep deformation resistance of cyclic softening steels.
0.2 1
NPP (a) 0.9
NPP (b)
1PT - M 1PT - M
0.16 0.8 2PTC - Cycle 1
Creep Strain, ε cr (%)
Creep Strain, εcr (%)
2PTC - Cycle 1
2PTC - Midlife 0.7 2PTC - Midlife
0.12 0.6
0.5
0.08 0.4
0.3
0.04 0.2
0.1
600˚C 600˚C
0 0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 800
Time (hours) Time (hours)
Fig. 4 Primary creep curves constructed from stress relaxation data for 10%Cr steel at 600 °C
using the NB model, assuming strain-hardening, for stress levels of a 75 MPa and b 170 MPa
4.2 Forward Creep Data Comparisons
Primary creep data generated using the NB model, assuming strain-hardening, for
the 75 and 170 MPa stress levels can also be presented in the form of creep-rate
versus time plots to further illustrate relative differences between the various prior
deformation histories (Fig. 5). Predictably, NB modelled creep-rates drop rapidly at
the start of the hold period, with the rate of drop being higher for cases where the
material had yielded macroscopically before the start of the tensile hold and lower
when the peak strain was below the limit of proportionality, i.e. when the defor-
mation preceding the hold time was elastic. The enhancement of creep-rates due to
prior plastic deformation, as reported in [6, 14–17], can also be observed at both the
creep stress levels. Moreover, the increase in creep-rates, at any point in the time
range considered, as a consequence of macroscopic yielding can be clearly
observed for both the stress levels.
The experimentally observed forward creep lines in the log–log representations
in Fig. 5 are essentially linear in the primary creep regime. Since they originate
from tests conducted on the same heat of 10%Cr steel, it is not surprising that the
agreement is good with the NPP-based NB model predicted behaviour at the lower
stress level (Fig. 5a). At the higher stress (Fig. 5b), the NPP model predictions
underestimate the experimentally observed forward creep-rates largely due to the
greater influence of anelastic recovery under these conditions.
Results obtained using the NB model, assuming strain-hardening, to predict
stress relaxation curves from forward creep data for the 10%Cr steel at 600 °C are
presented in Fig. 6. While application of the NB model assuming strain-hardening
approximates the experimentally observed stress relaxation behaviour well at lower
strains (below the limit of proportionality, Fig. 6a), this is not the case for strains
(strain amplitudes) and initial stresses in excess of yield (Fig. 6b). Once again, the
1E+1 1E+1
Forward Creep Forward Creep
(a) NPP (b) NPP
1E+0 1PT - M 1E+0 1PT - M
2PTC - Cycle 1 2PTC - Cycle 1
2PTC - Midlife 1E-1 2PTC - Midlife
1E-1
Creep-rate (1/h)
Creep-rate (1/h)
1E-2 1E-2
1E-3 1E-3
1E-4 1E-4
1E-5 1E-5
75 MPa 170 MPa
1E-6 1E-6
0.01 0.1 1 10 100 1000 10000 0.01 0.1 1 10 100 1000 10000
Time (hours) Time (hours)
Fig. 5 Creep-rate versus time plots for 10%Cr steel at 600˚C obtained using data predicted by the
NB model assuming strain-hardening for the different deformation histories at a 75 MPa and
b 170 MPa
150 500
3hr. T hold 30min. T hold (± 0.50%) - First Cycle
450
60min. T hold (± 0.50%) - First Cycle
125 7hr. T hold
400 7hr. T hold (+ 0.50%)
Strain-Hardening
350 Strain-Hardening
100
Stress (MPa)
Stress (MPa)
300
75 250
200
50
150
100
25
(a) 0.075% 50
(b) 0.50%
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Fig. 6 Stress relaxation curves predicted using the NB model determined from forward creep data
for 10%Cr steel at 600 °C for strain levels of a 0.075% and b 0.50% (assuming strain-hardening)
discrepancy in creep relaxation responses is attributed to the greater influence of

anelastic recovery in the high stress regime, which is currently not formally con-
sidered in the present model.
The major focus of this study has been a systematic evaluation of the influences of
monotonic and cyclic plasticity and their extent on the creep deformation response
of a 10%Cr steel at 600 °C and a 1%Cr steel at 550 °C. The obtained results
indicate the degree of creep softening to increase with the accumulated plastic strain
associated with increasing prior strain (strain amplitude) and initial stress at the start
of the relaxation period.
The evidence indicates that for hold times of up to 60 min in strain-controlled
creep–fatigue tests on the two power plant steels at their respective maximum
application temperatures, stress relaxation is due to primary creep deformation.
The effectiveness of the NB model fitting approach applied to the stress relax-
ation data to examine the influence of prior plasticity on forward creep on 10%Cr
steel at 600 °C is examined by using forward creep determined NB parameters to
predict stress relaxation behaviour for the same heat of 10%Cr steel. It is concluded
that adoption of the NB model, assuming strain-hardening, can effectively trans-
form forward creep to stress relaxation and vice versa in the low stress regime, but
not at higher stresses when the consequences of anelastic recovery are more
influential.
Acknowledgements The authors would like to acknowledge the assistance offered by Mr. Freddy
Bürki during creep testing and by other members of the High Temperature Integrity Group at
EMPA during the course of this work.
References
1. D.A. Miller, R.H. Priest, Materials response to thermal-mechanical strain cycling, in High
Temperature Fatigue: Properties and Prediction, ed. by R.P. Skelton (Elsevier Applied
Science Publication, Dordrecht, 1987), pp. 113–175
2. S.R. Holdsworth, E. Mazza, L. Binda, L. Ripamonti, Development of thermal fatigue damage
in 1CrMoV rotor steel. Nucl. Eng. Des. 237, 2292–2301 (2007)
3. R.L. Klueh, Elevated temperature ferritic and martensitic steels and their application to future
nuclear reactors. Int. Mater. Rev. 50, 287–310 (2005)
4. F. Abe, T.U. Kern, R. Viswanathan, Creep-resistant Steels (CRC Press, Cambridge, 2008)
5. C. Vorpahl, A. Möslang, M. Rieth, Creep-fatigue interaction and related structure property
correlations of EUROFER97 steel at 550 °C by decoupling creep and fatigue load. J. Nucl.
Mater. 417, 16–19 (2011)
6. A. Sarkar, V.D. Vijayanand, P. Parameswaran, V. Shankar, R. Sandhya, K. Laha, M.D.
Mathew, T. Jayakumar, E. Rajendra Kumar, Influence of prior fatigue cycling on creep
behavior of reduced activation ferritic-martensitic steel. Metall. Mater. Trans. A 45, 3023–
3035 (2014)
7. J.S. Dubey, H. Chilukuru, J.K. Chakravartty, M. Schwienheer, A. Scholz, W. Blum, Effects of
cyclic deformation on subgrain evolution and creep in 9–12% Cr-steels. Mater. Sci. Eng.
A 406, 152–159 (2005)
8. B. Fournier, M. Sauzay, C. Caës, M. Noblecourt, M. Mottot, L. Allais, I. Tournie, A. Pineau,
Creep-fatigue interactions in a 9 pct Cr-1 pct Mo martensitic steel: Part I. Mechanical test
results. Metall. Mater. Trans. A 40, 321–329 (2009)
9. E.G. Ellison, A.J.F. Paterson, Creep fatigue interactions in a 1 CrMoV steel. Proc. Inst. Mech.
Eng. 190, 319–330 (1976)
10. D. Sidey, Creep-fatigue interactions in a low alloy steel. Part III—Fatigue: Micromechanisms,
in Fracture 1977, ICF4, Waterloo, vol. 2 (1977), pp. 813–819
11. S.R. Holdsworth, Prediction of creep-fatigue behaviour at stress concentrations in 1CrMoV
rotor steel, in Proceedings of Conference on Life Assessment and Life Extension of
Engineering Plant, Structures and Components, Cambridge (1996), pp. 137–146
12. S.R. Holdsworth, Component assessment data requirements from creep-fatigue tests, ASTM
STP1539. J. ASTM Int. 8, 1–14 (2011)
13. V. Kalyanasundaram, Creep, fatigue and creep-fatigue interactions in modified 9%
Chromium-1%Molybdenum (P91) steels, Ph.D. Dissertation, University of Arkansas,
Fayetteville, 2013
14. L. Binda, S.R. Holdsworth, E. Mazza, Influence of prior cyclic deformation on creep
properties of 1CrMoV. Mater. High Temp. 27, 21–27 (2010)
15. S.R. Holdsworth, Creep-fatigue in steam turbine materials, in Proceedings of the 6th
International Conference on Advances in Materials Technology for Fossil Power Plants
(EPRI, Sante Fe, USA, 2010), pp. 487-503
16. S.R. Holdsworth, Influence of prior fatigue on creep properties, in Proceedings of the 12th
International Conference on Creep and Fracture of Engineering Materials and Structures
(CREEP 2012), Kyoto (2012)
17. A. Sarkar, V.D. Vijayanand, V. Shankar, P. Parameswaran, R. Sandhya, K. Laha, M.D.
Mathew, T. Jayakumar, E. Rajendra Kumar, Cyclic softening as a parameter for prediction of
remnant creep rupture life of a Indian reduced activation ferritic-martensitic (IN-RAFM) steel
subjected to fatigue exposures. Fus. Eng. Des. 89, 3159–3163 (2014)
18. J.H. Laflen, C.E. Jaske, Cyclic relaxation response under creep-fatigue conditions, in Stress
Relaxation Testing ed. by A. Fox (ASTM STP676, 1979), pp. 182–206
19. N.K. Sinha, Limitations of stress relaxation tests for determining stress dependence of strain
rate at high temperatures. Scr. Mater. 48, 731–736 (2003)
20. H. Altenbach, K. Naumenko, Y. Gorash, Creep analysis for a wide stress range based on
stress relaxation experiments. Int. J. Mod. Phys. B 22, 5413–5418 (2008)
21. G.G. Grzywinski, D.A. Woodford, Design data for polycarbonate from stress-relaxation tests.
Mater. Des. 14, 279–284 (1993)
22. D.A. Woodford, Accelerated Testing for High-Temperature Materials Performance and
Remaining Life Assessment, Report TR-114045 (EPRI, Palo Alto, USA, 1999)
23. J. Bolton, A “characteristic-strain” model for creep. Mater. High Temp. 25, 197–204 (2008)
24. E. Krempl, Cyclic Creep: An lnterpretive Literature Survey, Bulletin 195 (Welding Research
Council, New York, 1974), pp. 63–123
25. BS EN ISO 204, Metallic materials—Uniaxial creep testing in tension—Method of test
(2009)
26. C.C. Davenport, Correlation of creep and relaxation properties of copper. Trans.
ASME J. Appl. Mech. 60 A55–A60 (1938)
27. S.R. Holdsworth, G. Merckling, ECCC developments in the assessment of creep-rupture
properties, in Proceedings of the 6th International Charles Parsons Turbine Conference
(IOM, Dublin, 2003), pp. 411–426
28. V. Kalyanasundaram, S.R. Holdsworth, Prediction of forward creep behaviour from stress
relaxation data for a 10% Cr steel at 600 °C. Trans. Indian Inst. Met. 69, 573–578 (2016)
Structural Integrity Mechanics
and Creep Life Prediction of 304HCu
Austenitic Stainless Steel Under Multiaxial
State of Stress
Kanhu Charan Sahoo, Sunil Goyal, P. Parameswaran, S. Ravi

and K. Laha
Abstract For increasing the efficiency of fossil power plants, the boiler tube
material has to withstand higher temperature and pressure which calls for extensive
research for identifying materials having high-temperature strength, good corrosion
resistance and adequate mechanical properties. The 304HCu stainless steel
9304HCu SS is one of the candidate materials for boiler tubes employed in
advanced ultra-super critical power plants. However, under above circumstances,
material will subject to multiaxial state of stress that arises from internal pressure,
weld joint, inhomogeneous structure, sudden change in dimension and change in
cross section of the tube. Present study aims to introduce multiaxial state of stress
through notches of different root radius on creep samples. Notch of different root
radii, e.g. 0.25, 0.5, 2.5 was creep tested by keeping notch throat diameter 5 mm.
Both plain and notch specimen had creep tested at same stress level at a particular
temperature. Notch specimen possesses higher rupture life as compared to plain
specimen. Based on these observations, the material is found to be ‘notch
strengthening’. Then both SEM and optical micrograph were carried out on the
unfailed notch which revealed that cavity density was decreased from notch root
towards centre for relatively sharper notch, while random distribution of cavity for
shallow notch. FE analysis has been carried out to understand the contribution of
different components of stresses, i.e. von-Mises, maximum principal and hydro-
static stresses. Different models given by Cane, Hayhurst and Nix were examined
for creep life prediction under multiaxial state of stress. The average hardness near
to notch root was more in comparison with regions away from the notch root and
shallow notch was found exhibit more hardness in comparison with sharper notch.

Keywords AUSC Multiaxial creep Finite element analysis

Life prediction Hardness measurement
K. C. Sahoo (&) S. Goyal P. Parameswaran S. Ravi K. Laha

Metallurgy and Materials Group, Indira Gandhi Centre for Atomic Research,
Kalpakkam 603102, Tamil Nadu, India
e-mail: kanhu2011.sahoo@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_30
354 K. C. Sahoo et al.
1 Introduction
Fossil-fired power plant contributes nearly half of the total amount of electrical
energy production made all over the world. This is true for both developing and
developed countries as far as producing electricity is concerned. However, the main
problem associated with the coal-fired power plants is the poor efficiency and
generation of green house gas. Increasing efficiency of electricity generation is
related to steam line temperature and pressure. For increasing efficiency, the steam
line temperature of the plant should be above 60–70 K in comparison with con-
ventional power plant. This essentially calls for materials that withstand higher
temperature and pressure. Therefore, the structural components operating under this
harsh condition causes creep damage which results from the nucleation, growth and
coalescence of cavities and enhanced microstructural degradation resulting from
coarsening of precipitates and production of dislocation sub-structural network [1].
Advanced ultra-super critical power plant (AUSC) is conceived specifically for
increasing efficiency and decreasing carbon emission. In this respect, 304HCu SS is
a candidate material for advanced ultra-super critical power plant [2]. While
designing components, generally the designers use the uniaxial data with factor of
safety. However, the structural components experience multiaxial state of stress due
to internal pressure, change in geometry, inhomogeneous microstructure as in a
weld joint and possible changes in the mode of loading during service [3, 4]. Thus,
the failure of the material will be decided based on the multiaxial conditions that
prevail upon the components. The life and damage prediction of such components
could be assessed by using notched specimen testing to simulate the multiaxial state
of stress. The notched specimens are the most convenient and widely used ones to
generate multiaxial state of stress in laboratory scale [5]. Presence of notch intro-
duces some constraints which depend on the notch geometry, material chosen and
stress level. Notched specimens are designed to simulate the behaviour of various
stress-raising features in components, and, generally they fall into two categories—
(i) those which exhibit a high degree of constraints where deformation of the locally
high-stressed region is limited by the low value of the second stress invariant and
corresponds to a region of high-stress triaxiality and (ii) situations of low constraint
where the stress peak is usually less pronounced and a lower degree of triaxiality is
observed [6]. Different types of notch root radius, specimens are used to investigate
the range of behaviour associated from high degree to low degree of constraints [7].
The contribution of different components of stresses, i.e. von-Mises, maximum
principal and hydrostatic stresses which can be used to study the effect of multiaxial
state of stress on creep rupture life. A concept of representative stress which
accounts the relative contribution of von-Misses and maximum principal stresses
have been introduced to estimate the rupture life under multiaxial state [8–11].
According to Hayhurst et al. [9], the representative stress is the algebraic sum of
maximum principal stress and von-Mises stress. Cane [10] proposed that the
Structural Integrity Mechanics and Creep Life Prediction … 355
representative stress is the multiplication of the maximum principal stress and

von-Mises stress by incorporating material-dependent parameter which influences
the relative contribution of each stress.
In the present study, three notches were machined in the steel to establish a
relationship between the creep rupture life of uniaxial and multiaxial data. While
the specimen would fail from one of the notches, the other two notches will serve
for post-creep microstructural studies. Among the different notches employed,
variation of root radius of notch from 0.25, 0.5 and 2.5 mm was employed, keeping
a fixed depth of notch. In addition, smooth specimen was also tested under the same
conditions of stress and temperature (973 K). Further, for predicating the rupture
life by simulation, finite element analysis was carried out. Subsequently, scanning
electron microscopy (SEM) studies were carried out on unfailed notch to know the
propagation of crack and distribution of cavities and wedge cracks from the root of
the notch to the centre of the specimen. The hardness distribution as a function of
distance from the notch root was measured along the notch plane.
2 Experiment
Effect of multiaxial state of stress has been studied by carrying out creep test on
both notch and smooth specimen in the stress range 140–220 MPa at 973 K. The
effect of notch sharpness has been studied by carrying out test on notch root radii of
0.25–5 mm (r) by keeping notch throat diameter as 5.00 mm (d). These notches
resulted notch acuity ratio (d/r) of 1–20. The geometry of notched specimen is
illustrated in Fig. 1. Creep tests were systematically carried out in smooth specimen
to know the coefficients in Norton’s creep law (es = Arn).
Fig. 1 Schematic representation of notch specimen of different root radius

3 FE Analysis of Stress Distribution
Finite element analysis of stress distribution has been carried out in notch plane to
know the distribution of different components of stresses, viz. von-misses, maxi-
mum principal and hydrostatic components of stresses during creep exposure [12].
The 2D axisymmetric analysis was carried out using 4 noded quadrilateral elements
with ABAQUS 6.10 finite element solver [13]. In the analysis, the elastic modulus
was assumed to be 189 GPa. The value of the parameters A and n obtained from
smooth specimen data was 1.74 10−24 and 8.84, respectively. The stress con-
centration factor of the notched specimen is 1.4, 2.46 and 3.4 for notch acuity 2, 10
and 20, respectively. Initially, the deformation of the material was assumed to
behave elastically followed by creep deformation [14]. Then element size was
reduced closer to the notch root by structured mesh configuration and elastic
analysis was used to assure that the mesh configuration was sufficiently refined near
the notch root for the theoretical stress concentration factor. The stress redistribu-
tion was generally considered to attain stationary state condition when the total
strain at the notch throat plane reached the elastic strain in the material as stated by
Calladine [15].
Figure 2 shows the variation of (a) rupture life with net applied stress, (b) rupture
life with notch acuity ratio and (c) creep ductility (%RA) with notch acuity ratio.
Creep rupture life of 304HCu SS increases in the presence of notch, exhibiting the
ability of the material to ‘notch strengthen’. The ‘notch strengthening’ in the
presence of notch with notch acuity ratio of 2, 10 and 20 when creep tested at
973 K is found to exhibit a trend in the strengthening effect as it increases with
notch acuity ratio and, it tends to saturate for higher notch acuity ratio. This is
similar to what was observed in P91 steel and 2.25Cr–1Mo steel earlier similar
studies [12, 13, 16]. On the other hand, creep rupture ductility decreased in the
Fig. 2 Variation of (a) creep rupture life with net applied stress (b) creep rupture life with notch
acuity ratio (c) creep rupture ductility (reduction in area %) of both smooth and notch specimens of
the steel at 973 K
presence of notch, and the decrement was more profound for higher notch acuity.
The fracture behaviour of the steel for shallow and sharper notches [2, 12] was also
found to be varied in nature.
4.1 Creep Cavitation and Deformation Study Through SEM
Generally, it is known that the austenitic stainless steels are more prone to cavi-
tation and deformation. Shallow notches are more prone to cavitation while sharper
notches are more prone to crack. Sharper notches form crack easily than shallow
notches because of higher stress concentration factor. Therefore, scanning electron
microscopy (SEM) studies were carried out on the unfailed notch region of the
creep tested specimen. The specimens were grounded and polished up to half of
their diameter and then systematically SEM studies were carried out. Figure 3
shows the progressive increase in wedge crack and voids (r-type) from notch centre
to the root region.
Generally, the cavities nucleate, on locations specifically where three-
dimensional defects, namely precipitates occupy on the grain boundaries perpen-
dicular to the tensile axis. It has been observed that the creep fracture can occur by
‘w’ or wedge-type cracking at grain boundary triple points. The wedge cracking
develops as a consequence of grain boundary sliding, i.e. grain boundary sliding is
not accommodated which arises at higher stresses (low temperature) and larger
Fig. 3 SEM micrographs from notch root to centre of unfailed notch creep specimen of notch root
radius 2.5 mm creep tested at 200MPa, 973K
Fig. 4 SEM micrographs from notch root to centre of unfailed notch creep specimen of notch root
radius 0.25 mm, creep tested at 200 MPa and 973 K
grain sizes. These wedge cracks are propagated by the formation of r-type voids as
well. The wedge cracks are simply accumulation of r-type voids and small cavities.
The presence of voids, microcracks, wedge cracks and their accumulation and
propagation is shown in Fig. 4. It suggests that because of high-stress concentration
factor, crack has nucleated at the notch root and propagated towards centre. The
generation of lots of cavities and wedge crack during creep exposure is attributed to
the high strength and limited ductility of the material. Figure 4a, b distinctly shows
the migration and coalescence of voids into wedge-type cracks and then into cracks
which lead to final failure in sharper notch specimen.
4.2 Effect of Notch Sharpness on Creep Rupture Behaviour
To understand the effect of multiaxial stress on creep strengthening, FE analysis

was carried out on notch throat plane. Effect of plastic behaviour of the material in
the stress distribution across the notch throat plane was also assessed considering
the material’s elasto-plastic deformation behaviour. Shallow notch has less plastic
deformation at notch root than sharper notch because of less stress concentration at
the notch plane. The plastic deformation and plastic zone increase with notch acuity
ratio and tend to saturate at higher notch acuity ratio. FE analysis of stress distri-
bution across the notch throat plane in 304HCu SS steel during creep deformation
has been carried out by incorporating material’s time-independent (plastic) and
time-dependent (creep) behaviour in the model [17].
In the FE analysis, the material was considered to undergo elastic/elasto-plastic
deformation followed by creep deformation. Under creep condition, the distribution
of stress was found to change with creep exposure and attain a stationary state. The
variations of von-Mises, maximum principal and hydrostatic state of stress across
the notch in the steel for notch acuity ratio 20 and 2 are shown in Fig. 5. von-Mises
Fig. 5 Distributions of (a) von-Mises, (b) Maximum principal and (c) Hydrostatic stresses across
the notch throat plane for different notch acuity ratio (creep tested at 200 MPa, 973 K)
stress increases with increase in distance from centre to root which is more sig-
nificant in case of sharper notch. For shallow notch, the Von-misses stress was
uniform at notch root while it is more concentrated for sharper notch. Maximum
principal stress decreases at the notch root in case of shallow notch while it is
increasing in its trend as observed in case of sharper notch. Hydrostatic stress
exhibited similar behaviour as that of the maximum principal stress. The contour
plots of von-Mises and maximum principal stress are shown in Figs. 6 and 7 for
notch acuity ratio (a) 2 and (b) 20 at 200 MPa, 973 K, respectively. For shallow
notch, the Von-misses stress was uniform at notch root (Fig. 6a) while it is more
concentrated for sharper notch (Fig. 6b). In case of maximum principal stress, it
was not uniform across the notch and found to be concentrated at the central region
of notch for relatively shallow notches (Fig. 7a) and close to notch root for sharper
notches (Fig. 7b).
4.3 Creep Life Prediction Based on the Representative

Stress
The creep rupture life under uniaxial loading is expressed in terms of the applied
stress as
Fig. 6 The variation of von-Mises stress for notch acuity ratio (a) 2 and (b) 20, for creep tested at
200 MPa, 973 K
Fig. 7 The variation of Maximum principal stress for notch acuity ratio (a) 2 and (b) 20, for creep
tested at 200 MPa, 973 K
tr ¼ Mrm ð1Þ
where ‘m’ is the slope of uniaxial creep plot.

Creep rupture life under multiaxial state of stress can also be described by
similar equation in which rrep is the representative stress calculated through notched
specimen data by incorporating suitable factor from different models, viz. Hayhurst,
Cane and Nix model.
tr ¼ Mrm
rep ð2Þ
The representative stress (rrep) is applied to the uniaxial specimen which will
result in the same rupture life as that of notched specimen. If at a particular tem-
perature rrep > rnet (net stress applied to the notched specimen), then the presence
of notch weakens the material, i.e. ability of the material to ‘notch weaken’. If
rrep < rnet, then the presence of notch strengthens the material, i.e. material exhibits
‘notch strengthening’. As already discussed, creep deformation and cavitation
depends upon three components of stress, viz. maximum principal stress, hydro-
static stress and von-Mises stress in governing the creep deformation and cavitation
[2, 12]. The von-Mises stress controls the deformation and creep cavity nucleation
processes, the maximum principal and hydrostatic stress control the continuum
cavity growth whereas the maximum principal stress controls the stress directed
diffusion controlled intergranular cavity growth [12, 13].
In this paper, creep life prediction has been done by the relationships provided
by Hayhurst et al. [9], Cane [10] and Nix et al. [11]. The prediction obtained from
these three methods compared and based on the close fitting by least square method,
their applicability has decided.
Hayhurst’s model
rrep ¼ ar1 þ ð1 aÞrvm ð3Þ
Cane’s model
c=m
ðmcÞ=m
rrep ¼ r1 rvm ð4Þ
Nix’s model
rrep ¼ 2:24 r1 0:62ðr2 þ r3 Þ ð5Þ
where m, c and a are material constants and ‘r1’ is the maximum principal stress
and ‘rvm’ is the von-misses stress, r2 is the intermediate principal stress and ‘r3’ is
the minimum principal stress, respectively.
For a = 0 and c = 0, rrep = rvm, i.e. representative stress fully depends on the
von-misses stress.
For a = 1 and c = m, rrep = r1, i.e. representative stress fully depends on the
maximum principal stress.
During creep exposure the stresses vary across the notch throat plane but at
skeletal point is attained on reaching a stationary state. But it is difficult to identify
the location in notch plane at which the stress can be considered as representative
stress. Hayhurst et al. [9], Goyal et al. [12] based on FE analysis have introduced
the skeletal point concept in determining the representative stress.
4.4 Skeletal Point Concept
The skeletal point is the location in notch throat plane for a given notch geometry,
where the variation of stress across the notch throat plane for different stress
exponent ‘n’ in Norton’s law intersects. FE analysis has been carried out for several
values of ‘n’ ranging from 1 to 10 to obtain the skeletal point at a particular stress
200 MPa. The value of coefficient ‘A’ was obtained based on the creep strain of
10−5 h−1. Similar works were reported by Webster [8, 13] and Goyal et al. [12] The
details regarding the variation of skeletal point with notch acuity ratio have been
discussed in detail by Goyal [12] where the authors show that the maximum
principal and hydrostatic stresses at skeletal point were found to increase with notch
acuity ratio and tend to saturate at higher notch acuity ratio; whereas, von-Mises
stress decreased with increase in notch acuity ratio.
From Hayhurst Eq. (1), the value of ‘a’ was decided by carrying out the
regression analysis for the representative stress based on the skeletal point stress for
each notch geometry. The best fit value of ‘a’ is 0.32 with high correlation coef-
ficient (0.96107) for this steel. So creep rupture life is mainly depended by
von-Mises stress to an extent of contribution of 68% while the rest is from maxi-
mum principal stress.
Cane’s model also used for prediction of rupture life which incorporates ‘c’ as a
material parameter. By carrying out systematic regression analysis, the best value of
‘c’ was found to be 3.8 with high correlation coefficient (0.968). Figure 8 shows the
presentation of multiaxial creep rupture data based on the calculation of repre-
sentative stress based on the models proposed by (a) Hayhurst, (b) Cane.
4.5 Effect of Notch in Change in Hardness Due to Creep
The creep tested samples failed in one of the three notches. Other two notches of
steel were used to generate the information regarding change in microstructural and
mechanical property like hardness. In order to judge the change in hardness at notch
plane due to creep exposure, the steel was chopped from both sides of notch and
ground and polished up to half of their length. Systematically Vickers hardness was
measured with distance for sharper (0.5 mm) notch case and shallow notch
Fig. 8 Multiaxial creep life

prediction based on
representative stress
calculated based on the
models proposed by
(a) Hayhurst, (b) Cane at
973K
(2.5 mm) case across the notch plane. Figure 9 shows the hardness profile of creep
tested and failed specimen of two cases, viz. (a) one with a notch of 0.5 mm root
radius and (b) another with a notch of 2.5 mm root radius, traversing from notch
root to another side. The above figure suggests that in both the cases, the average
Vickers hardness at notch root region was more than at the central region. Since the
specimen exhibits more stress concentration at the notch root, it is more strained
during creep testing. The change in hardness could be attributed to grain boundary
strengthening caused by grain size changes, dislocation density produced by creep
deformation and sub-structure development, the precipitation hardening derived
from fine precipitation of nano-precipitates and other phases or creep damage
accumulation. So any of the above phenomena can cause significant amount of
change in hardness [18]. But as compared to sharper notch shallow, notch showed
significant increase in hardness in both notch root and centre. In sharper notch, the
decrease in hardness is due to long rupture time, softening phenomena is more due
Fig. 9 3-D Hardness variation across the notch throat plane (un-failed notch) (a) shallow notch
(notch root radius 0.5 mm) (b) sharp notch (notch root radius 2.5 mm), creep tested at 200 MPa,
973 K
to predominance in tertiary creep while in shallow notch, increase in hardness is due

to the phenomena of strain hardening and lack of recovery due to creep exposure
[19].
5 Conclusions
1. Presence of circumferential-U notch increases the creep rupture life of 304HCu

SS material in comparison with material taken in the form of smooth specimen.
2. SEM investigations reveal that the random distribution of wedge crack and
cavities was distinct in the case of shallow notch material compared to uniform
distribution of void, wedge crack and sub-crack in the case of sharper notch one.
3. Effect of notch sharpness shows von-misses stress was uniform at notch root in
shallow notch while it is more concentrated for sharper notch. The maximum
principal stresses were found to concentrate at the central region of notch for
relatively shallow notches and close to notch root for sharper notches.
4. The creep life prediction by means of representative stress through skeletal point
method was done by both Hayhurst and Cane’s model. Cane’s model repre-
sented well for the multiaxial creep data with uniaxial creep data.
5. The average Vickers hardness at notch root region was more than the central
region for sharper notch, and the average hardness of shallow notch both at
centre and root was more due to the absence of recovery phenomena in tertiary
creep.
Acknowledgements The authors gratefully acknowledge the encouragement and continuous

support received from Dr. A. K. Bhaduri, Director, IGCAR, Kalpakkam. The authors also wish to
record their sincere thanks to V. D. Vijayanand, SO/D, IGCAR, Kalpakkam for his help and
support during analysis.
References
1. W.D. Nix, Mechanisms and controlling factors in creep fracture. Mater. Sci. Eng. A 103,
103–110 (1988)
2. K.C. Sahoo, S. Ravi, S. Goyal, P. Parameswaran, K. Laha, Creep ruptures behavior and
microscopy analysis of 304 HCu austenitic stainless steel under multiaxial state of stresses at
973 K. Trans. Indian Inst. Met. 69(2), 451–455 (2016)
3. S. Goyal, K. Laha, K.S. Chandravathi, P. Parameswaran, M.D. Mathew, Finite element
analysis of Type IV cracking in 2.25Cr-1Mo steel weld joint based on micro-mechanistic
approach. Phil. Mag. 91, 3128–3154 (2011)
4. T.H. Hyde, W. Sun, A.A. Becker, Failure prediction for multi-material creep test specimens
using a steady-state creep rupture stress. Int. J. Mech. Sci. 42, 401–423 (2000)
5. A. Forghani, M. Shahbazi, N. Zobeiry, A. Poursartip, R. Vaziri, An overview of continuum
damage models used to simulate intralaminar failure mechanisms in advanced composite
materials. in Composites Science and Engineering (A volume in Woodhead Publishing
Series) (2015), pp. 151–173
6. P.S. Westera, C. Pickard Rolls-Royce plc, Derby Rolls-Royce plc, Derby the prediction of
stress rupture life of notched specimens in theti5331s beta-processed titanium alloy
7. W. Li, F. Liao, T. Zhou, H. Askes, Ductile fracture of Q460 steel: effects of stress triaxiality
and lode angle. J. Construct. Steel Res. 123, 1–17 (2016)
8. G.A. Webster, S.R. Holdsworth, M.S. Loveday, K. Nikbin, I.J. Perrin, H. Purper, R.
P. Skelton, M.W. Spindler, Fatigue Fract. Eng. Mater. Struct. 27, 319–342 (2004)
9. D.R. Hayhurst, J. Mech. Phys. Solids 20, 381–390 (1972)
10. B.J. Cane, in Proceedings of the International Conference on Mechanical behavior of
Materials, ed. by K.J. Miller, R.F. Smith (Pergamon Press, Oxford, 1979), pp. 173–182
11. W.D. Nix, J.C. Earthman, G. Eggeler, B. Ilschner, Acta Metall. 37, 1067–1077 (1989)
12. S. Goyal, K. Laha, Creep life prediction of 9Cr-1Mo steel under multiaxial state of stress.
Mater. Sci. Eng. A 615, 348–360 (2014)
13. S. Goyal, K. Laha, M.D. Mathew, Creep life prediction of modified 9Cr-1Mo steel under
multiaxial state of stress. Procedia Eng. 86, 150–157 (2014)
14. S. Goyal, K. Laha, C.R. Das, S. Panneerselvi, M.D. Mathew, Effect of constraint on creep
behavior of 9Cr-1Mo steel. Metall. Mater. Trans. A 45(2), 619–632 (2014)
15. C.R. Calladine, Proc. R. Soc. Lond. A 309, 363 (1969)
16. S. Goyal, K. Laha, C.R. Das, S. Panneerselvi, M.D. Mathew, Finite element analysis of effect
of triaxial state of stress on creep cavitation and rupture behaviour of 2.25Cr–1Mo steel. Int.
J. Mech. Sci. 75, 233–243 (2013)
17. S. Goyal, K. Laha, A.K. Bhaduri, Response of triaxial state of stress to creep rupture life and
ductility of 316 LN austenitic stainless steel. J. Mater. Eng. Perform. 26(2), 752–763 (2017)
18. Z. Yan, Z. Jie, L.I. Xiaona, Microstructural evolution and change in hardness during creep of
NF709 austenitic stainless steel. Acta Metall. Sin. (Engl. Lett.) 24(3), 220–224 (2011)
19. H. Tanaka, M. Murata, F. Abe, H. Irie, Microstructural evolution and change in hardness in
type 304H stainless steel during long-term creep. Mater. Sci. Eng. A 319, 788–791 (2001)
Stress Analysis for Integrity Assessment
of High-Energy Hot Reheat Pipe Bends
of 210 MW Coal-Fired Unit
Rajesh Daga and Mahendra Kumar Samal
Abstract The critical high-energy piping components of thermal power plant are
predominantly subjected to damage mechanisms of creep, fatigue, and their inter-
actions during their service. These pipings at high temperature and transient loadings
under sustained loadings and thermal movements are designed as per the piping
code. The creep damage mechanism causes irreversible thermal expansion in the
piping inducing thermal stresses. The pipings under altered state of stresses due to
thermal expansion are supported by suitable hangers and supports for ensuring
stresses to be within the permissible design limits. The present paper discusses the
layout of hot reheat piping of a coal-fired 210 MWe unit. The piping material as per
specification ASTM A335 P22 with hot reheat steam temperature 540 °C and hot
reheat steam pressure of 4.7 MPa pressure operates under creep domain. The piping
under these pressure and temperature loadings is subjected to load ramps during the
start-ups, shutdowns, and load fluctuations and the steady loading during the sus-
tained load operations. The scope of the study includes hot reheat pressure and
low-pressure bypass pipelines. The piping layout with respect to the designed
stresses, stress ratios, support loads, element forces, and displacements at each node
has been considered for the present ongoing assessment. The design of the piping
layout is accomplished to ensure that the structural integrity of the piping doesn’t
exert excessive load to the nozzles of the connecting equipment. The stress analysis
of the piping and support system should be carried out to ensure that the stresses are
within the allowable values as per the applicable design code during the service. The
R. Daga (&)
NTPC Energy Technology Research Alliance, NTPC Ltd,
Greater Noida 201306, India
e-mail: daga.raj@gmail.com
M. K. Samal
Division of Engineering Sciences, Homi Bhabha National Institute,
Mumbai 400094, India
e-mail: mksamal@yahoo.com; mksamal@barc.gov.in
M. K. Samal
Reactor Safety Division, Bhabha Atomic Research Centre, Trombay,

https://doi.org/10.1007/978-981-10-7197-3_31
370 R. Daga and M. K. Samal
stress analysis data were utilized in modeling the pipe bend before control valve of
the intermediate pressure turbine for online creep fatigue damage monitoring.

Keywords Stress analysis High-energy piping Hanger and support

Piping code Creep Thermal movement Creep fatigue damage assessment
1 Introduction
Main stream, hot and cold reheat, and high-pressure and low-pressure bypass piping
are the critical high-energy pipings of thermal power plants at high pressure and
temperature subjected to a number of damage mechanisms like creep, fatigue,
thermal fatigue, microstructural instability, and flow-accelerated corrosion fatigue
during their service [1]. These high-energy pipings under the effect of service-
induced stresses experience damage which lead to cracking at base metal locations
and weldments. Such damages under high stresses during the service cause cracking,
steam leak, or failure. Thus, it is imperative to predict and prevent failures under the
effect of in-service damage mechanisms in such high-energy critical pipings. With an
understanding of the damage mechanism, extent of damage is evaluated on applying
rules of damage accumulation and remaining life of the component is assessed.
Globally, considerable work is being done for integrity of such critical structures in
service at elevated temperatures [2]. The life assessment of a hot reheat pipe bend of a
200 MW subcritical unit on real-time basis is discussed elsewhere [3]. The stress
analysis of the hot reheat piping loop including low-pressure bypass piping was
performed to evaluate the piping load. The stresses were computed induced due to
piping loads in hot reheat pipe bend for the fluid temperature and pressure.
To prevent stresses beyond the code permissible values in piping systems, major
aspect of a thermal power plant is to have an effective inspection program [4]. For a
safe and reliable operation of the power plant, condition assessments of the critical
components are essential. For maintaining the system stresses during the online and
offline conditions, the pipe support systems are to be monitored during their service.
The present paper discusses on the guidelines and its importance for ensuring the
integrity of the piping structure.
2 Overview: Stress Analysis of Design Layout of the Hot

Reheat Piping
High-energy piping generally refers to main steam, hot and cold reheat, feedwater,
and extraction steam piping. High-energy piping has no expected design life. When
in service, the piping undergoes continuous damage. The fabrication codes have built
up design margins to ensure that the anticipated damage rate would not result in
failure of major components during the economic life of a unit. The failure occurs
Stress Analysis for Integrity Assessment of High-Energy Hot … 371
normally due to inadequate design, overestimation of material strength, an incom-

plete understanding of the metallurgical impact of fabrication variables, and signif-
icant change in operating practices. The stress analysis of the design layout of
210 MW unit of high-energy piping of hot reheat and low-pressure bypass piping
was carried out as per design code ASME B31.1. The stresses within the piping
system are limited within the code allowable limit under the defined load cases
expected to occur in service conditions. The forces and moments expected to pass on
to the equipment nozzles through the connected piping under the defined loading
conditions are limited to the equipment manufacturer’s allowable limit. The objective
of the stress computations carried out for flexibility analysis was to ensure structural
integrity of the piping system.
3 Piping System Surveys
The high-energy piping system designed and fabricated as per Power Piping Code
B31.1 experiences different loading types, namely sustained, thermal, and occa-
sional loads. The system stresses of the piping are to be maintained, while the
hangers are in the online (hot) and offline (cold) positions. The pipe support system
which primarily accomplishes the task of maintaining the stresses in the piping
within the code allowable limits is also the vital component. The hanger and
support system supports the dead weight of the piping, the insulation, the contained
fluid, and the weight of other attached components, which allows for thermal
RH OUTLET
HEADER
IP TURBINE
HRH pipe bend
INLET PIPE-BEND
BEFORE IP
CONDENSER STOP
CONTROL
VALVE#2
Zoomed region shown in Fig. 3
Fig. 1 Piping layout of the reheat pipe line where the bend is located
Fig. 2 Piping loads on HRH pipe bend used in the online creep–fatigue monitoring
expansion of the piping while minimizing the load transfer at the terminal points of
boiler header and turbine connections. Additionally, the support systems are
designed to withstand dynamic loadings, namely vibration, water, and steam
hammer. For an improperly designed support system, considerable stresses are
imposed in the piping if the thermal expansion differences are not accommodated
and hanger support malfunction. The elevated stresses under improper design cause
distortion of the steam piping or premature failure of girth or attachment welds.
The walkdown inspections of the piping system during the cold and hot con-
ditions are carried out to ensure the proper operation of the hanger in accordance
with its designed capacity. Figure 1 shows the piping layout of the hot reheat
pipeline where the bend is located. The piping system supported by the hangers
should not get stressed beyond the allowable stresses as per the applicable piping
design codes. Survey of condition of piping and support system is performed to
identify broken or bottomed out hangers and trend the hanger readings. Such
monitoring of the support system helps in identifying the potential problems with
the critical piping and their support systems which may become significant. The
timely corrective measures for the repair are initiated when hanger failure is
identified. The effect of hanger failure on the piping system under service loadings
is to be assessed by stress analysis of the piping system. The cause of the hanger
failure is to be ascertained. A reason for the hanger failure may be attributed to the
hanger load exceeding its designed load capacity. The consequential damage to the
piping system in the event of hanger failure causes system stresses to exceed
beyond allowable code stresses leading to premature creep failure and fatigue
cracks. The stress analysis of the piping system is carried out to determine the
location of the damage in the piping. The piping loads acting on the pipe bend are
shown in Fig. 2.
4 Effect of Damage Mechanisms on the Piping and Life

Assessment
Primarily, a comprehensive understanding of the potential damage mechanisms

affecting the piping component is essential in predicting a component’s effective
life span and preventing failures. The damage depends on factors, namely design,
operating modes, water chemistry, and fuel characteristics which are unit-specific.
The piping components must be considered individually to obtain realistic life
assessments. The life span of any high-energy piping component is primarily a
function of factors such as operating conditions, namely stresses and temperatures;
geometry, namely piping layout, support placement, and wall thickness; material;
and type of damage. The presence of uncertainties like material properties, fabri-
cation tolerances, and stress states makes the life assessment of the piping com-
ponent difficult. Thus, life assessment is focused on setting inspection intervals to
obtain inputs for taking decisions to whether repair or replace the component.
Several researchers have stressed the need to develop a life prediction methodology
to address various aspects of failure mechanisms [5–12].
5 Analysis of Hot Reheat Pipe Bend as a Part of Piping

Loop
In order to carry out thermo-mechanical analysis of pipe bend of hot reheat loop of
the coal-fired thermal power plant, the geometrical and boundary condition details
of the piping loop are required. As shown in Fig. 1, the entire loop considering
different anchor points, support, and hangers has been modeled using finite element
code. 3D piping elements have been used for the analysis. Both straight and bend
pipe elements have been included in the loop in order to realistically calculate the
stresses and loading coming in the loop due to dead weight as well as thermal loads.
As the hot reheat pipings have been designed to operate at 540 °C and 4.63 MPa,
the same has been considered in the analysis. The segment of pipe bend for which
detailed thermo-mechanical analysis has been carried out is shown in Fig. 2. The
actual geometry has been modeled using 3D 20-nodded solid brick elements. The
loads and moments coming from the connected pipings are applied as boundary
loads in the finite element model of the 3D pipe bend. The online damage moni-
toring system incorporates this piping bend as a module in the system, and entire
information regarding geometry, loading, and boundary conditions is fed into the
system through necessary input files. The input contains information regarding the
discretization of the component, material data, and associated loading and boundary
conditions. Piping forces in three Cartesian coordinate systems and moments (two
bending and one torsional) about three axes are also considered. The above pro-
cedure ensures that the thermo-mechanical behavior of the pipe bend is simulated as
accurately as possible. In the next section, the details of thermal analysis, stress
analysis, and damage evaluation (due to creep and fatigue) of the hot reheat pipe
bend as carried out by the online damage monitoring system are described in detail.
6 A 3-D FE Model Hot Reheat Pipe Bend and Damage

Calculation
A detailed 3-D finite element (FE) model of the pipe bend was made using
20-noded brick elements. The element information for the FE mesh is stored as
element–node relationship and nodal coordinates in a static file specific to the
component. Similarly, all the loadings and boundary conditions at specific nodes
and element surfaces are stored in the same file. During run-time, the software
computes the element stiffness matrices and load vectors, assembles, and then
solves for metal temperature distribution (in thermal module) and metal stress
information (in the structural module). The element loading information is for unit
internal pressure rise, dead weight, and associated piping loadings due to temper-
ature rise (from room temperature to design temperature condition). Actual loading
is calculated at run-time once the fluid pressure and temperature information are
known from the data acquisition system. The bending moments, end pull, and shear
forces at the ends of the pipe bends are converted into equivalent loading infor-
mation in the element surfaces covering the two ends. The 3-D FE model along
with material parameters and loading (due to internal pressure, dead weight, thermal
gradient, and piping expansion loads) is used by the system BARC Online
Structural Safety Evaluation System (BOSSES) for online monitoring of damage
due to creep, fatigue, and creep–fatigue crack growth. The nodal temperature data
of the FE mesh (obtained by thermal or transient temperature module of BOSSES)

are used to calculate the load due to temperature transients seen by the component.
Thermal load due to temperature gradient, internal pressure, associated piping
loads, etc., is then used in a stress analysis module to calculate stresses at different
gauss points of the pipe bend. In our analysis, the pipe bend is in equilibrium under
the action of all the forces and bending moments. For prevention of rigid body
motion, some soft springs are provided at the two ends of the bend.
The metal temperature information is stored in file (as restart files) to be used for
heat conduction analysis in the next time step. All the stress information at the
Gauss points is stored for a specific time interval (one hour) so that rain-flow cycle
counting algorithm can be used to know the number of stress cycles. The stress
cycles are used further to calculate fatigue damage at the Gauss points using
Minor’s cycle summation rule. The stress and temperature information at each time
step is used to calculate creep damage using Robinson’s life fraction rule. For the
above purpose, the appropriate creep and fatigue properties of the material are used.
The temperatures and stresses at different time are also used to compute crack
growth of different postulated cracks due to creep and fatigue at various locations of
the pipe bend in the fracture mechanics module. The life consumed by the com-
ponent is predicted by extrapolating the damage and crack growth information to
the previous operating hours. The contour information of stress intensities, metal
temperature, creep, and fatigue damage, etc., at the last time step is written into
specific dynamic plot files. The history of stress intensity, temperature, creep, and
fatigue damage at different critical locations is also stored in plot files for visual-
ization. All the information is upgraded, and restart files are saved for successive
computation. The system also keeps automatic backup of contour and history
information of different parameters of the component at user-specified intervals.
The real-time process data of the pipe bend are made available to the researcher’s
desk through client-server network.
For the components operating at high temperature and subjected to plant tran-
sients and cyclic loads during plant start-up and shutdown, creep–fatigue interaction
is an important damage mechanism. Due to cyclic loads, fatigue micro-cracks
initiate at the metal surface which can join with the grain boundary cavitation sites
(due to creep damage and grain boundary sliding, there is progressive cavitation
and micro-crack formation), and the micro-cracks can grow along the grain
boundaries. Hence, computation of crack growth due to creep–fatigue interaction is
an important aspect of life assessment which has been carried out here using
fracture mechanics methods. For fatigue crack growth calculation, modified Paris
law has been used in this analysis which uses effective stress intensity factor range
for the given loading cycle in the pre-cracked geometry. For creep crack growth
calculation, the C* parameter is used as a crack tip loading parameter and the
universal creep crack growth law is used. The evaluation of C* integral requires
calculation of reference stress and stress intensity factor for the given crack con-
figuration in the component.
For selection of components for online monitoring, one must make an assess-
ment of the potential risk associated in case of failure of the component in terms of
plant unavailability, economic loss, consequences to environment and human life,

etc. The risk of damage depends upon the local stress concentration due to abrupt
change in geometry (such as shell–nozzle junctions and tee joints) as creep damage
accumulation depends upon the effective stress and temperature. Definitely, the
headers (especially the shell–nozzle junctions) are potential creep damage-prone
areas. Again, the shell–nozzle junctions contain welded regions where there can be
further complications due to in-homogeneities in the mechanical and creep prop-
erties, microstructures of the material, residual stresses, etc. The pipe bends are also
subjected to stress in-homogeneity due to their geometry (which depends upon the
bend radius, bend radius to thickness ratio, diameter to thickness ratio, etc.).
However, the particular pipe bend under consideration is important for online
monitoring because of the prior bulge observed (which may be because of some
prior imperfection). The consideration of existing local imperfection such as ‘bul-
ges’ as mentioned above is very important from the point of view of stress analysis
and damage prediction. As we discretize the actual 3D geometry using 3D con-
tinuum finite elements, it will be possible to incorporate the bend ovality easily if
we know the details of it (i.e., the location, profile, and amount of ovality). The
information has to come from actual measurement in the plant. In this work, the
ovality details are not measured, and hence, the same is not considered here.
However, incorporation of this information is crucial to life prediction, and it will
be carried out in future.
As the damage monitoring system was implemented recently and has accumu-
lated data for roughly 25,937 h, it is necessary to extrapolate the results of creep
and fatigue damage for the actual service life (e.g., 78,251 + 25,937 h.) of the HRH
pipe bend by the system during online calculations. The remaining life assessment
is performed on the basis of cumulated material damage and crack growth. The
consumed life of the pipe bend is very small (i.e., of the order of 5%) when the
creep and fatigue usage factor (the limiting factor being unity for crack initiation at
a material point) is considered. However, when crack growth is considered as the
lime-limiting mechanism, the predicted life consumed is still less (smaller than 1%),
when the crack of 1.5 mm depth was considered and the most severe crack was
found to be the longitudinal semi-elliptical surface crack at the outer surface of the
intrados of the pipe bend. Hence, the pipe bend is found to be very safe, and the
usual annual inspection during shutdown of the plant is not necessary for this
component. However, the above prediction can be fine-tuned once more, and more
data are available from the previous operation and inspection history, which can be
used to suitable schedule the future inspection interval. The fluctuation of steam
pressure, temperature, and boiler load index (tones/hour) for the hot reheat pipe
bend is shown in Fig. 3, and the computed metal temperature, stress intensity, and
damage index in the pipe bend are shown in Fig. 4. The contour for maximum
stress intensity and accumulated creep and fatigue damage in the pipe bend is
shown in Figs. 5 and 6, respectively.
Fig. 3 Recorded thermal hydraulic parameters (pressure, temperature, and flow of steam through
the bend)
Fig. 4 Computed temperature, stresses, and damage factor (due to creep and fatigue) at a material
point in the HRH pipe bend
Fig. 5 Contour of maximum stress intensity in MPa (due to mechanical, thermal, and piping
loads) over HRH pipe bend at 25,937 h of monitoring
Fig. 6 Contour of accumulated damage (due to creep and fatigue) over HRH pipe bend at the end
of 25,937 h of monitoring
7 Discussion and Recommendation
A properly operating high-energy piping system essentially requires the weight,

thermal, and pressure stresses to be kept at acceptable levels as per the piping codes.
The stress analysis of the piping involves calculation of piping code stresses, loads,
and deflections under static and dynamic loading conditions. The piping should be
well supported and should have controlled deflections under its weight, thermal,
and other loads. Thus, stress analysis of the piping system ensures the piping layout
conforms to allowable code values of weight, thermal, and pressure stresses during
its service loads.
The hanger and support system are the vital and critical components which
properly maintain high-energy piping system. They maintain the stresses of the
piping system in online (hot) and offline (cold) conditions. This primary function is
achieved by supporting the dead weights (piping, insulation, fluid, and other
components), allowing thermal expansion of piping and supporting dynamic loads
(vibration, water, and steam hammer). Summarily, the piping system with hangers
and supports should be code compliant for both sustained and thermal load cases.
To properly maintain the high-energy piping system, performance of hangers
and supports has to be evaluated. A thorough hot and cold walkdown of the piping
and support system is required to ensure its healthiness. A hanger inspection should
be carried out for maintaining the hanger operation properly. The failure of hanger
causes damage to the piping. The increased piping load due to hanger failure causes
deformation in the piping, accelerated creep and fatigue damage, and thus even-
tually reduces the piping life.
Piping code ASME B31.1 recommends that the hangers should be periodically
examined to look for changes both in the piping and in the hangers. Hangers can
become inoperable as a result of creep deformation of the piping, a dynamic event
in the line such as a water or steam hammer, or deterioration of the spring and other
components. For constant-support and variable-load hangers, the code recommends
that hanger readings should be obtained both in the fully hot position and when the
unit is offline in the cold position. The hanger readings should be recorded and
permanently stored. The temperature of the pipe at the time of observations should
also be noted.
The issue of validity of damage predictions versus actual observed plant damage
is very crucial and needs to be discussed here. One needs to measure the actual
damage in specific critical locations of the plant components in order to validate the
assessment approach and have a better understanding of the models used. The
measurements are usually in terms of change in microstructure, hardness, carbide
precipitation, grain boundary cavitation, grain boundary micro-cracks, etc. There
exits quantification techniques in literature, like measurement of cavity density and
A-parameter. The authors are actually in the process of quantifying the damage due
to creep and fatigue in the above pipe bend. Replicas at some critical locations of
the pipe bend have been taken and are currently being analyzed. However, this
aspect will be addressed in the future work. Apart from creep and fatigue damage in
high-temperature components, flow-assisted or erosion corrosion, corrosion pitting,
corrosion fatigue, stress corrosion cracking, etc., are of major concern in
low-temperature pipework systems. However, laboratory experiments are now
carried out by researchers worldwide to address specific cases, to have an in-depth
understanding and to generate experimental data from which some empirical or
analytical models can be derived. For flow-assisted corrosion in nuclear pipings,
some popular models exist. Once we generate such models for the thermal power
plant feedwater line conditions, it may be possible to deal with it analytically.
Otherwise, one has to resort to periodic inspections based on plant-specific and
component-specific experience for dealing with life assessment of components
subjected such dominant mechanisms.
The accuracy of predictions of the above online system is based on the level of
modeling used in the analysis and use of proper material properties. However, if
there is some prior imperfection which is not known and not considered explicitly
in the modeling, the results may not be very accurate. For these reasons, site
investigations are required. However, the extent of site investigation and the
associated maintenance and inspection costs can be minimized if such a system is
used a guiding tool as it provides online information on status of the damage in
critical components which is based on available data and actual loading conditions
seen by them. The above system actually guides the plant maintenance engineer to
decide inspection and maintenance intervals of critical plant components and hence
fits into a comprehensive risk assessment program. The correlations between pre-
dictions and actual damage can be improved by conducting metallurgical investi-
gations at certain intervals for the critical locations to provide better material
properties so that the system gets more accurate information for prediction. Though
component inspections cannot be ruled out, these can be minimized by adopting a
risk-based in-service inspection program in which online monitoring is an integral
part.
The prevailing approach/practice of the utilities to estimate the need for
inspection is on the basis of offline inspection and the past operation and mainte-
nance (O&M) experience. These result in frequent inspections of problem-free
equipment by the utilities and also often cause unexpected failures. Presently, the
system uses only one deterministic value, i.e., mean value of the material properties
for calculation of damage and crack growth due to creep and fatigue. However, a
probabilistic analysis module is currently being developed where the effect of type
of distribution, coefficient of variation, etc., of the material properties can be
accommodated to calculate the band of variation of the predicted damage param-
eters (i.e., 5 and 95% percentile values) instead of a single value. The system is
capable of monitoring welds as well. One needs to specify the mechanical, creep,
and fatigue properties of the weld in the weld region. In the finite element model,
the material properties are associated with a material number which is in turn
allocated to the corresponding elements falling in the respective regions of the
materials. In the present analysis, the bend does not have weld regions and hence
weld material properties are not used.
The issue of prior system loading, i.e., the pull-up stresses that will be applied
during original plant construction, is also an important consideration in the pre-
diction of damage due to creep and fatigue in high-temperature components. If the
system pull-up stresses are known, it can be easily incorporated in the finite element
model as an initial stress problem. However, when the above information is not
known (especially for older plants), the assessment may be done with applications
of some initial pull-up stresses in the FE model according to certain codal guide-
lines. The assessment of damage due to creep–fatigue interaction is an important
aspect of life assessment and remaining life prediction exercise. In ASME, the rules
for damage accumulation due to creep–fatigue interaction are different for different
materials (e.g., low alloy steel and stainless steel). The effective damage depends
upon the relative damage of creep and fatigue damages (the locus of effective
damage curve is not a straight line, but sometimes a bilinear or nonlinear curve).
Again, because of many sources of uncertainties such as system initial stresses and
local in-homogeneities, usually the predictions are done based on codal guidelines
which are conservative in nature. However, given sufficient laboratory database for
a given material under wide range of loading conditions for the above effects, it
may be better to use own database (specific to the material of the component under
monitoring) which can remove the conservatism to some extent.
Use of non-destructive and metallurgical investigations for the critical bulged
location of the pipe bend under monitoring will provide a method of validation of
the predicted results of the current online monitoring system. The damage pre-
diction due to creep depends upon the constitutive equations used. However, the
selection of creep constitutive equation depends upon the particular material under
consideration. It has been shown in literature that no single creep constitutive law
can fit well for all the materials for all temperature and stress ranges. Of course,
creep laws based on continuum damage mechanics theories (e.g., Kachanov’s and
Hayhurst’s) are more promising as they consider the coupled effect of stress and
damage in the material. In the current analysis, Only Robinson’s life fraction rule
has been used for creep damage prediction. More advanced theories such as con-
tinuum damage mechanics will be considered in future work of the authors.
The authors envisage covering more number of critical components in an unit as
well as covering many further coal- and gas-based thermal power plants of NTPC
Ltd in near future so that the integrity of these components can be assessed in the
real time and this shall help in decision making by the concerned plant engineers as
well as maintenance staff regarding minimization of cost and overall downtime of
the plants. The presence of online damage monitoring system shall also generate the
necessary information which shall help in future design and material improvements.
Acknowledgements Dr. M.K. Samal acknowledges the support from Dr. D.N. Badodkar,
Director RD&DG, Mr. V. Bhasin, Associate Director, NFG, BARC for carrying out this R&D
activity regarding online damage monitoring of plant components. Mr. R. Daga acknowledges the
support from Dr. P. Jain, General Manager, NETRA, NTPC and engineers from NTPC thermal
power plants at different locations.
References
1. EPRI Report 1012201, Fossil Plant High-Energy Piping Damage, vol. 1. (June 2007)
2. R. Viswanathan, Damage Mechanisms and Life Assessment of High Temperature
Components (Ohio, ASM International, 1989), pp. 182–263
3. R. Daga, G. Bandyopadhyay, M.K. Samal, B.K. Dutta, J. of power plant operation.
Maintenance Mater. Issues 5(1), 1–19 (2008)
4. C. Basavaraju, W.S. Sun, Piping handbook, “Stress analysis of piping systems” (The
McGraw-Hill Companies, NY, 2004)
5. R. Daga, M.K. Samal, Real-Time Monitoring of High Temperature Components, 6th
International Conference on Creep, Fatigue and Creep-Fatigue Interaction (CF-6), (23–25
January 2012, Chennai, India)
6. M.K. Samal, B.K. Dutta, H.S. Kushwaha, R. Daga, G. Bandyopadhyay, Creep damage
evaluation of a power plant header using combined FEM analysis and quantitative
metallography. Trans. Indian Inst. Met. 63(2–3), 411–416 (2010)
7. Assessment of components operating in the creep regime, Section 10 of API-579 document
(June 2001)
8. R. Daga, G. Bandyopadhyay, M.K. Samal, B.K. Dutta, A.K. Mohindru, Combined creep life
fraction assessment of critical locations of an in-service superheater outlet header under
surveillance programme. Trans. Indian Inst. Met. 63(2–3), 423–429 (2010)
9. R. Daga, M.K. Samal, Structural integrity assessment of superheater header through FE
analysis and in-situ metallography (Global Energy Technology Summit (GETS), 2014, India,
7–9 November 2014)
10. R. Daga, M.K. Samal, Creep Fatigue Damage Assessment of an In-service Superheater Outlet
Header. International Conference on Fatigue Durability and Fracture Mechanics, (Bangalore,
India, 28–30 May 20150
11. Power Piping Code, ASME B31.1
12. R. Daga, M.K. Samal, Remote monitoring of health of critical high temperature components
by real-time finite element technique, (Global Energy Technology Summit (GETS) 2015,
India, 7–9 November 2015)
Part VI
Energy and Transportation
Structural Integrity Study for Additional
Piles on an Existing Jacket Structure
in Western Indian Offshore
Praveen Bhat and Bakul Master
Abstract Oil and Natural Gas Corporation (ONGC) operates more than 265 steel
jacket-supported platforms installed in water depths ranging from 25 to 90 mts. and
secured to the seabed using steel piles. Quite large number of platforms have either
exceeded their design lives and/or undergone modifications/mitigation/strengthening
due to change in design premises and/or revamping projects executed for enhanced oil
recovery. This calls for requalification studies for their extended “fit for use purpose.”
This paper describes the structural integrity assessment of an existing platform found
to have highly overstressed piles and failing members and joints based on design-level
analysis. A pushover analysis has been carried out to assess the reserve strength ratio
(RSR) for checking the structural adequacy of the jacket structure, and suitable
mitigation measures have been suggested for the particular platform as a combination
of additional retrofit members/piles, removal of redundant facilities, and strength-
ening of members/joints, etc.
Keywords Jacket structure Design in-place analysis Ultimate strength analysis

RSR
1 Introduction
The hunt for Black Gold started in the Indian offshore region when Oil and Natural
Gas Corporation Limited (ONGC) started its operations in the western offshore
region of India in the year 1976. Since then, more than 265 well, process, and living
quarter platforms have been installed and operated by ONGC. By now, quite a large
number of platforms have either exceeded their design lives or undergone/
undergoing modification/mitigation measures due to change in design premises
and/or revamping projects executed for enhanced oil recovery. Re-assessment of
P. Bhat (&) B. Master

Institute of Engineering and Ocean Technology, ONGC-Panvel, Panvel 410221,
Maharashtra, India
e-mail: bhat_praveen@ongc.co.in

https://doi.org/10.1007/978-981-10-7197-3_32
386 P. Bhat and B. Master
these offshore platforms involves structural integrity check after taking due con-
sideration of change in design premises along with new loads and structural
damages, if any.
This paper describes the summary of the work carried out at the Institute of
Engineering and Ocean Technology (IEOT), ONGC, on the project of global static
in-place structural integrity check of an existing platform and consequential miti-
gation measures. The mitigation measures suggested for the particular platform are
in combination with the removal of redundant equipment and appurtenances,
restriction on marine growth, strengthening of members/joints, and additional ret-
rofit members/piles. The paper highlights the main findings of the study with
special emphasis on the jacket and pile structure.
2 Structural Analysis
2.1 Design-Level Analysis
The study for this typical jacket structure, located in western offshore, for the
structural integrity check on specific requirements from the asset/platform operator
in the context of its design life has been outlived. The salient features of the
platform have been listed in Table 1 and Fig. 1.
An initial global static in-place design level analysis of the platform has been
carried out for 100-year extreme storm condition with 85% environmental loading
considering all lateral loads, i.e., Wave, Current, and Wind, with a load factor 0.85
(reduced environmental criteria1) in combination with other design loads including
gravity loads due to self-weight and production facilities installed on the platform
topside. A marine growth thickness of 100 mm from EL (+) 6 m to EL (−) 30 m
and 50 mm from EL (−) 30 m to mud line has been considered in the analysis.
The primary structural members have been checked for yield, stability, and
nominal joint strength assessment for 100-year extreme storm condition [1]. In the
Table 1 Platform details

Water depth 77.455 m
No. of main piles 4
No. of skirt piles 2
Diameter of main piles 1.372 m
Diameter of skirt piles 1.372 m
Piles vertical penetration for main piles 95.555 m
Piles vertical penetration for skirt piles 77.724 m
Production details
No. of conductors 12 (9 inside + 3 clamp on)
No. of risers 6
(1 * 4″ + 2 * 6″ + 1 * 10″ + 1 * 12″ + 1 * 14″)
Structural Integrity Study for Additional Piles … 387
Skirt Pile, SKB
Main Pile, B1
Main Pile, B2
Main Pile, A1
Main Pile, A2
Skirt Pile, SKA
Fig. 1 Key plan and 3-D view of the jacket structure
design-level analysis, all the piles have been checked for axial load carrying
capacity and the pile head stresses.
The results of the design-level analysis revealed that for 100-year extreme storm
condition with 85% environmental loading and associated design loads, the factor
of safety (FOS) against axial capacity for all the piles is more than 1.50 under
various load cases in both compression and pull out. In pile head stress utility ratio
check, all the piles were found to have material utilization higher than the maximum
permissible limit of 1.0. Hence, re-analysis was performed after incorporating load
reduction measures. The load reduction measures comprised of removal of
redundant sump and pump caisson, non-consideration of future riser protector, and
restriction of marine growth thickness to 50 mm throughout. The results of the
re-analysis showed that still some of the piles were highly overstressed with a UC
value higher than the permissible limit of 1.00 (refer Table 2).
Member strength assessment revealed that six structural members were having
capacity utilization more than the API (RP-2A-WSD) specified limit value of 1.0
with a maximum capacity overutilization by 24%. In joint strength check, three
structural joints were showing high capacity utilization with UC ratios exceeding the
acceptable limit of UC ratio of 1.0 with a maximum capacity overutilization by 28%.
2.2 Simplified (Linear) Ultimate Strength Analysis
Subsequently, simplified (linear) ultimate strength in-place analysis with 100%

environmental loading was performed, and few primary structural members and
joints were found to be having utility ratio higher than the acceptable value of
1.2781. Hence, few structural members and joints were not passing the assessment
Table 2 Pile FOS and pile UC values for original condition (with 85% environmental loading)
with suggested load reduction measures
Pile Location Axial force Pile Min. FOS in Axial Pile Min. Max.
No. in capacity in compression force capacity FOS pile
comp. (MN) comp. (MN) in in in UC
pull pullout pull
out (MN) out
(MN)
1 A1 14.334 38.781 2.71 10.225 30.020 2.94 0.948
2 B1 17.057 38.781 2.27 8.984 30.020 3.34 1.006
3 A2 19.475 38.781 1.99 13.539 30.029 2.22 1.300
4 B2 20.452 38.781 1.90 12.521 30.029 2.40 1.382
5 SKA 10.695 23.152 2.16 4.268 23.759 5.57 0.945
6 SKB 13.189 23.152 1.76 3.544 23.759 6.70 1.063
requirement even as per the simplified ultimate strength analysis check (Linear
global analyses check as per API-RP-2A (WSD)).
Member Check Results:
S. No. Member UC ratioa UC ratiob
1 203L-0743 1.034 1.278
2 201L-0004 1.099 1.352
3 202L-0005 1.043 1.255
4 203L-0006 1.243 1.530
5 204L-0007 1.119 1.347
6 0001-404L 1.008 1.181
Joint Check Results:

S. No. Brace Location UC UC
member ratioa ratiob
1 0007 Row-A, X-brace joint, b/w EL (−) 52.76 m and 1.280 1.538
EL (−) 77.455 m
2 0253 Horizontal level, EL (−) 52.760 m 1.232 1.426
3 0005 Row-B, X-brace joint, b/w EL (−) 52.76 m and 1.225 1.466
EL (−) 77.455 m
Note aDepicts design-level analysis with 85% environmental loading
b
Depicts simplified ultimate strength analysis with 100% environmental loading
2.3 Nonlinear Ultimate Strength Analysis
As the results of the design-level analysis and simplified ultimate strength analysis
revealed that some of the primary structural members, joints, and piles do not meet
the assessment requirement, a higher level nonlinear ultimate strength analysis was
carried out for eight directions of environmental loading. In this paper, results of
nonlinear plastic collapse analysis [2] using USFOS [3] software have been dis-
cussed for assessing the ultimate strength of the jacket platform, and reserve
strength ratio (RSR) values for the structure have been presented. The RSR is
defined as
Ultimate lateral load carrying capacity

RSR =
100 year environmental condition lateral loading
The results of the ultimate strength analysis revealed that the structure is not able
to withstand the environmental forces up to the target RSR [4] level of 1.323 for all
the considered directions, primarily due to failure of Row-2 piles (refer Fig. 3) and
X-brace joints present on Row-A and Row-B of the jacket structure, around a load
level of 1.05 (refer Fig. 2).
Thereafter, a re-analysis for nonlinear ultimate strength check was carried out
after considering conductors as piles for providing the lateral support [5]. The
re-analysis results revealed that the structural adequacy of the jacket structure of
platform still could not be documented for all the wave approach directions even
after considering the conductors as piles.
The analysis results (both design-level and ultimate strength) revealed that the
structure was having overstressed piles (predominantly Row-2 piles) and X-brace
joints. The main reason for that is the change in the design regime (design hydro-
dynamic coefficients, hydrodynamic marine growth thickness, etc.) and installation
of additional facilities on the platform over a period of time. All these factors have
contributed to an increase in the hydrodynamic loading by more than 45%.
Unloadding observed at
Collapse
Fig. 2 Jacket structural collapse around a load level of 1.05

Fig. 3 Failure of row-2 piles,

depicting plastic utilization
3 Additional Piles/Strengthening Requirement
3.1 Load Transfer and Load Distribution Mechanism

for Additional Pile Arrangement
In view of the insufficient capacity of the jacket structure to withstand the design
loading, an analysis study for installation of additional piles along with grouting of
overstressed joints and members is contemplated.
The existing piles on Row-2 of the platform are found to be highly overstressed,
and hence, additional piles have been contemplated to be installed near them. For
ensuring proper load transfer, adequate strength pile connection needs to be
designed, fabricated, and installed. The load distribution pattern needs to consider
the fact that the vertical loads due to self-weight of the structure and topside (deck)
loading would have already mobilized the pile-soil resistances for the existing piles,
and the additional piles will only be contributing toward sharing the incident
environmental loading. To assess the structural adequacy of the additional piles for
sharing the environmental loading on the jacket structure, two separate analysis
studies (with 100% loading) have been carried out:
1. In-place design-level analysis with only existing piles considering only gravity
loads (100%)
2. In-place design-level analysis with all piles (incl. additional piles) considering
only environmental loads (100%)
The effects on the support system (pile-soil system) were then combined to
achieve the pile head loads for ascertaining the pile material utilization and soil
capacity utilization (refer Table 3). It is important to note that pile material uti-
lization is of utmost concern as the jacket structure was having highly overstressed
piles in its original condition. An iterative procedure was adopted for the selection
of adequate pile size for providing sufficient support capacity, and therefore, pile
sizes ranging from 60 in. up to 84 in. were checked for. Following this procedure, it
was finally proposed to install four 84 in. (2.134 m) additional piles on Row-2 of
the platform (two additional piles each at pile location A2 and B2); thereby all the
piles been found to meet the requisite strength requirements.
Additionally, the exact new pile intra-spacing and distance from existing piles on
Row-2 (refer Fig. 4) will have to be ascertained for assessing the load distribution
and pile group effects. The proximity to which the installation barge could approach
the platform and the obstruction posed due to the structure deck needs to be duly
considered.
Table 3 Factor of safety for extreme storm condition for both compression and pull out for well
platform
Pile Location Min. FOS in Min. FOS in pull Max. pile
No. compressiona outa UCa
1 A1 2.46 3.51 0.408
2 B1 2.08 4.08 0.683
3 A2 2.88 7.51 0.647
4 B2 2.63 20.58 0.925
5 SKA 1.93 5.05 0.246
6 SKB 1.55 5.83 0.532
7 A2 2.56 1.95 0.302
(NP1)
8 A2 3.24 2.29 0.262
(NP2)
9 B2 3.18 2.25 0.276
(NP1)
10 B2 2.23 2.23 0.303
(NP2)
a
Sump-pump caisson and future riser protector considered to be removed
Fig. 4 Structural model with additional piles
3.2 Proposed Additional Pile Details
Number: 4
Size: 2.134 m (84 in. outer diameter) 0.065 m (Thickness)
Vertical Penetration below mud line: 45.5 m
It is to be noted that after carrying out the strengthening of the jacket structure
with additional four piles on Row-2 of the structure, the load redistribution has
effectively resulted in the reduction of stresses in some of the adjacent members and
joints , e.g., the Row-A and Row-B primary X-brace joints (b/w EL (−) 52.76 m
and EL (−) 77.455 m) were now found to be stressed within permissible limits.
3.3 Grouting Scheme
One of the primary structural members and two primary structural joints were found
to be overstressed after carrying out the analysis with additional piles; so,
strengthening of these components has been proposed. The analysis has been
carried out after considering the strengthening of the member 0001-404L with
adequate strength grout [6] (Fig. 5), the results of which reveal that the member is
stressed within the permissible limits. The results of the re-analysis performed after
0252-0253
0001-404L
L
0257
0253
0248-02499
Fig. 5 Structural components proposed for grouting (as highlighted)
considering the strengthening of the respective chords for joints 0253 and 0257
with adequate strength grout (Fig. 5) reveal that the structural joints 0253 and 0257
are stressed within the permissible limits.
4 Conclusion and Recommendations
The results of the structural analysis show that in the original condition, even after
incorporating the suggested mitigation measures, the jacket structure of the platform
did not meet the structural adequacy requirements primarily due to overstressed
Row-2 piles and X-brace joints.
In view of the same, it has been contemplated to install four additional piles near
Row-2 of the platform. A design-level analysis has been carried out with additional
piles proposed to be installed on the platform. The structural adequacy of the
platform could be documented for all the considered environmental directions
subject to the following mitigation/strengthening measures:
1. Removal of sump and pump casing,
2. Non-consideration of future riser protector on Row-2 of the structure,
3. Installation of additional four 84″ (2.134 m) piles on row-2 of the platform (two
each at A2 and B2), and
4. Grouting of primary structural member 0001-404L and primary joints 0253
(Joint Chord member 0248-0249 to be grouted) and 0257 (Joint Chord member
0252-0253 to be grouted)
The additional pile-jacket connection details along with final additional pile
penetration below mud line, and members grout properties will be required to be
designed while carrying out the detailed engineering. From above, it can be
concluded that old-age platforms can be re-qualified with the implementation of

appropriate mitigation measures adopted after carrying out detailed nonlinear
assessment of the structure. These structures can continue production of hydro-
carbons without pressing the need of platform abandonment process or installation
of new platforms.
Acknowledgements We acknowledge the support and resources provided by ONGC required for
carrying out this study. The study has been immensely beneficial in understanding the pertinent
issues relevant to the structural behavior of jacket structures and is of vital importance in purview
of ONGC’s operational requirements to carry out life extension studies of existing platforms. We
extend our sincere thanks to Shri C. Tandi, ED-HoI, IEOT-ONGC for his generous support and
encouragement. We are also immensely grateful for the continuous motivation received from Shri
Dinesh Kumar, GGM (Civil)-Head of Structures Section, IEOT-ONGC.
References
1. IEOT—ONGC: re-qualification methodology for fixed offshore platforms in west coast of

India, IEOT-LIFE-001, rev. 4, 2012-06-05
2. J. Chakrabarty, Theory of Plasticity (McGraw-Hill International, New York, 1987), p. 1987
3. ULTIGUIDE—guidelines for ultimate strength analyses of fixed offshore structures, DNV—
SINTEF—BOMEL (1999)
4. DNV-OS-C101, Design of offshore steel structures, general (LRFD method) (April 2011)
5. API RP-2A WSD, Recommended practice for planning, design and constructing fixed offshore
platforms—working stress design. API Recommended practice 2A-WSD, 21st ed. Supplement
3 (2007)
6. Assessment of repair techniques for ageing or damaged structures, doc. ref. C357R001 Rev 1,
MSL Engineering Limited
Structural Integrity Assessment
of a Reactor Pressure Vessel Using
State-of-the-Art Methodologies—A
Case Study
V. Chaudhry, S. M. Ingole, A. K. Balasubrahmanian

and U. C. Muktibodh
Abstract Structural integrity assessment of reactor pressure vessel (RPV) has been
carried out for design life of the reactor using state-of-the-art methodologies.
Irradiation damage is the governing material degradation for core belt region of
RPV, and it is quantified based on the surveillance program. Structural integrity
assessment has been carried out for the postulated flaw at the core belt region to
demonstrate the protection of the vessel against catastrophic failure, plastic col-
lapse, and tearing instability. Protection against catastrophic failure has been
demonstrated based on the linear elastic fracture mechanics (LEFM) approach.
Stable crack extension and flaw stability due to ductile tearing have been demon-
strated based on J-integral and failure assessment diagram (FAD). Limiting values
of flaw depth and flaw length have been evaluated to demonstrate the protection
against all failure modes. Further, leak-before-break (LBB) criterion has been
adopted to demonstrate the defense in depth philosophy. Limiting flaw length to
depth ratio has been quantified in order to demonstrate leak-before-break criterion.
The assessment has demonstrated the healthiness of the vessel for safe operation.

Keywords Irradiation damage Catastrophic failure Plastic collapse

Tearing instability Leak-before-break
1 Introduction
Reactor pressure vessels (RPVs) are designed and fabricated in accordance with
consensus codes and are considered as life-limiting component for light water
reactors (LWRs). Selection of the material for RPVs is generally based on the
special considerations pertaining to high strength, good hardenability, and excellent
fracture properties. Commonly used materials in the construction of vessels for
LWRs are ASTM SA 302Gr B (plates)/SA 508 class 2 in the USA, 20MnMoNi55
V. Chaudhry (&) S. M. Ingole A. K. Balasubrahmanian U. C. Muktibodh

Nuclear Power Corporation of India Limited, Mumbai 400094, India
e-mail: vchaudhry@npcil.co.in

https://doi.org/10.1007/978-981-10-7197-3_33
396 V. Chaudhry et al.
in Germany, 16MnD5 in France, and 15X2HMUA in Russia. Although many other

materials are acceptable for reactor vessels, the choice is limited due to consider-
ation of fracture toughness and radiation-induced material degradation.
A typical RPV of boiling water reactor has been considered to demonstrate its
structural integrity for its design life, say 40 years. The life-limiting region of RPV
is the core belt region where the irradiation damage is the governing material
degradation. Large efforts have been made in many countries to characterize the
irradiation-induced damage in RPV steels as well as to understand and model the
mechanism governing the damage through experimental studies in research reactor
and through surveillance programs. The results obtained can be used to forecast the
behavior of RPV material under irradiation environment. Now, it is well understood
that the susceptibility of RPV steel to radiation damage is strongly affected by the
presence of copper, nickel, phosphorus, and manganese [1]. The correlations for
predicting the irradiation damage to the material are available in US Nuclear
Regulatory Commission (USNRC) Guide RG 1.99 Revision 2 [2], ASTM E 900-02
[3], and ASME Section XI Appendix G [4].
Structural integrity assessment of RPV has been carried out to rule out the
typical failure modes of RPV for the design life of the reactor, and also to
demonstrate the protection against the instability for the postulated defect in the
core belt region. Typical failure modes are non-ductile failure, plastic collapse, and
tearing instability. ASME codes provide the guidelines for integrity assessment
considering these failure modes [4]. The assessment is made based on postulating a
defect in core belt region having a depth of quarter thickness and length of
one-and-half the thickness. Limiting flaw length to depth ratio has been quantified
in order to demonstrate leak-before-break criterion.
2 Design Details of RPV
RPV considered in the present study is a cylindrical shell of ASTM SA 302Gr B

(plates) with an inside diameter of around 4000 mm, length of 20,000 mm, and
thickness of 124 mm having hemispherical top and bottom heads with thickness of
102 mm. The inside surface of the vessel is cladded with stainless steel of 5 mm
thickness. Figure 1 shows the schematic of a typical BWR pressure vessel [5]. The
loading conditions considered for the assessment of the vessel are:
• Design condition: Design pressure of 8.75 MPa and design temperature of
300 °C.
• Normal operating condition (NOC): Normal operating pressure of 7.0 MPa,
operating temperature of 286 °C, and heating and cooling rate of 55 °C/h.
• Upset condition: Normal operating pressure of 7.0 MPa, operating temperature
of 286 °C, and cooling rate of 150 °C/h.
• Test condition: Hydro test at 7.7 MPa with heating rate of 55 °C/h (the heating
rate are much less than 55 °C/h in test condition).
Structural Integrity Assessment of a Reactor Pressure Vessel … 397
Fig. 1 Typical boiling water reactor vessel [5]
The most severe transient for light water reactors comes from the combination of
thermal stresses (resulting from rapid cooling) and stresses due to internal pressure.
BWR vessels operate with a large inventory of water at saturation condition so that
any cooling transient would result in steam condensation, and results in system
depressurization, preventing the coupling of high thermal and pressure stresses.
Structural integrity assessment for different failure modes has been carried out
based on postulating an elliptical defect in vessel wall core belt region having a
depth of quarter wall thickness and length of one-and-half thickness.
3 Irradiation Damage Quantification
As discussed earlier, the correlations for prediction of shift in ductile-to-brittle

transition temperature are available in US Nuclear Regulatory Commission
(USNRC) Guide RG 1.99 Revision 2, ASTM E 900-02, and ASME Section XI
Appendix G. These correlations can be used to evaluate the shift in ductile-to-brittle
transition temperature due to irradiation damage in a material under operating
condition and neutron fluence. These correlations were developed from large
number of surveillance data from pressurized water reactors (PWRs) and boiling
water reactors (BWRs). Over a period of time, understanding of the damage
mechanisms was developed such that the correlations are mechanistically driven
rather than being purely statistical in nature. This aspect is of great importance
when the correlations are extrapolated beyond the maximum value of the database,
for example, deciding the design life of a reactor.
In the USA, the radiation embrittlement of RPV beltline material is assessed by
RG1.99 Revision 2 and 10CFR 50.61. The 41 J temperature is used as the index for
transition temperature shift based upon ASTM specified Charpy V-notch
(CVN) impact test. The guide RG1.99 gives a method for estimating the shift in
the ductile-to-brittle transition temperature (DBTT) and the drop in the Charpy
upper-shelf energy (USE). The correlations specified in RG1.99 were derived sta-
tistically from a database containing fewer than 170 surveillance data points. Later,
significant data has been generated and published from material test, a program
which has enhanced the mechanistic understanding of the microstructural changes
in the material leading to embrittlement. ASTM E 900-87, a standard guide for
predicting neutron radiation damage to reactor vessel materials, was issued based
on 216 CVN surveillance data points. Later, Electric Power Research Institute
(EPRI) sponsored to evaluate independently the mechanistic and statistical bases
from the data consisting of 752 measurements (80–90 were from BWR surveil-
lance) and new correlation was developed, details of which are provided in EPRI
technical report [6]. The EPRI correlation was accepted by ASTM E 900-02.
Further, additional 62 data points from BWR surveillance program were produced.
The assessments showed that the BWR data points were not predicted well by the
correlations based on PWR dominated data. In particular, the embrittlement shift for
BWR material tends to underpredicted by ASTM E 900-02. As a result, additional
work was carried out to generic improved correlations for both PWR and BWR.
One of the correlations proposed by Oak Ridge National Laboratory (ORNL) [7]
with additional 24 BWR data values is presently adopted by ASME Section XI

Appendix G [4].
4 Assessment Details
4.1 Protection Against Catastrophic Failure
To demonstrate the protection against catastrophic failure of vessel, structural

integrity assessment has been carried out as per ASME Section XI Appendix G [4].
The criteria for the protection against catastrophic failure is given by,
2K1m þ K1t \K1c ð1Þ
where K1m and K1t are the stress intensity factors (SIF) under pressure and thermal
loadings and are evaluated using the methodology given in ASME Section XI
Appendix G [4]. The defect is postulated in axial and circumferential direction at
inside and outside surface of the vessel. Equation (1) needs to be satisfied for
startup/shutdown condition for the postulated flaw in axial and circumferential
direction for different operating conditions. Table 1 provides the SIFs for different
operating conditions and its comparison with the critical stress intensity (K1c). The
K1c value has been evaluated according to ASME Section XI Appendix G
depending on the operating temperature and the shift p in DBTT (discussed in
Sect. 3). The K1c value is considered here as 240 MPa m (upper bound value).
It is evident from Table 1 that the maximum SIF is observed to be for an axial
flaw on inside surface under shutdown condition. This is from the fact that under
shutdown condition, the thermal gradient produces significant tensile stress at inside
surface of the vessel.
Table 1 SIF for postulated flaw under different operating conditions

p
S. Operating Postulated flaw 2K1m + K1t (MPa m) K1c p
No. conditions location Hydro Design NOC Upset (MPa m)
test condition condition
1 Startup Axial flaw on inside 74.15 84.26 67.40 67.4 240
surface
2 Startup Axial flaw on outside 75.66 85.39 69.17 69.17 240
surface
3 Startup Circumferential flaw on 39.73 44.56 36.5 36.5 240
outside surface
4 Shutdown Axial flaw on inside 79.58 89.69 72.8 82.23 240
surface
5 Shutdown Circumferential flaw on 40.87 45.69 37.64 47.03 240
inside surface
4.2 Protection Against Plastic Collapse
The limiting values of flaw depth and flaw length have been evaluated as per ASME
Section XI Appendix C [4] presuming that plastic collapse is the dominant mode of
failure.
The allowable flaw length (Lf) for end-of-evaluation period is given by,
" #1=2
0:5 r2f
Lf ¼ 1:58 ðRm tÞ 1 ð2Þ
r2h
where rf is the flow stress, rh is the hoop stress, Rm is the mean radius of vessel, and
t is the vessel thickness.
The allowable flaw lengths for hydro test pressure (7.7 MPa) and design pres-
sure (8.75 MPa) are 2874.8 and 2501.57 mm, respectively. The allowable flaw
depth at the end-of-evaluation period flaw length Lf is given by,
M2ð1 SÞ
a=t ¼ ð3Þ
M2 S
h i1=2
where M2 ¼ 1 þ L2f 4tR
1:61
m
and S = SFmrh/rf, SFm is the structural factor taken
as 2.7 and 2.4 for normal and upset operating conditions, respectively.
The allowable flaw depths of cylindrical shell for hydro test condition and design
pressure are 119.61 and 119.75 mm, respectively. The applicability of Eq. (3) is
limited to a flaw depth of 93 mm (0.75t). Thus, protection against plastic collapse
for a flaw of depth 93 mm and length 2501.57 mm has been demonstrated.
4.3 Protection Against Tearing Instability
The evaluation procedure and acceptance criteria used to demonstrate structural

integrity of RPV are specified in ASME Section XI Appendix K [4]. The approach
is used to check the effect of low upper-shelf energy on the plastic zone at the crack
tip which dictates the tearing stability. The original purpose of appendix K was to
demonstrate the structural integrity of RPV with low upper-shelf energy (USE). The
technical basis for Appendix K is described in WRC bulletin 413 [8]. The approach
is equally applicable to any region of the reactor vessel where the fracture tough-
ness can be described with elastic plastic parameters.
In order to demonstrate the flaw stability, there are two acceptable methods as
per ASME Appendix K, for a vessel with wall thickness lesser than 150 mm.
Method-I
Criteria-1: This criteria state that a crack starts growing when the applied JI equals
the material resistance to crack initiation JIC (J0.1).
JI J0:1 ð4Þ
where JI is the applied J-integral considering a structural factor of 1.15 on maxi-

mum pressure (overpressure) and 1.0 on thermal loading, and J0.1 is J-integral
resistance of a ductile flaw extension of 2.5 mm.
The material properties used in the assessment are taken from EPRI report [9].
To be on a conservative side, the sulfur content of the vessel is assumed to be more
than 0.018%. The estimated mean minus two standard deviation bilinear J-R curve
of the material (200–287 °C) is shown in Fig. 2, having J = JIC at da = 0 to
J = 1.31JIC at da = 2.54 mm, and J = 1.31JIC for all crack extensions greater than
2.54 mm.
Applied J-integral for axial flaw, Japp = 18.16 kJ/m2
Applied J-integral for circumferential flaw, Japp = 5.0145 kJ/m2
J0.1 = 92.7 kJ/m2 (Fig. 2)
Thus, JI J0.1
Criteria-2: This criterion demonstrate flaw stability due to ductile tearing,
@J @JR
at Japp ¼ Jmat ð5Þ
@a @a
The structural factor considered are 1.25 on pressure and 1.0 on thermal loading.
Figure 3 shows the comparison of the slopes of the applied J-integral curve and
the material J-R curve for axial and circumferential flaws, respectively. The slopes
of applied J-integral curve are lower than the J-R curve for the material, and thus,
Fig. 2 Bi-linear J-R curve for low-alloy steel material [9]

Fig. 3 Comparison of the slopes of the applied J-integral curve and the J-R curve for a axial flaws
b circumferential flaws
demonstrate the stability against the ductile tearing for a flaw depth equal to half the
vessel wall thickness.
Thus, the Method-1 demonstrated the stability against crack growth for a pos-
tulated flaw of quarter thickness depth and stability against ductile tearing for a flaw
depth equal to half the vessel wall thickness.
Method-II
This method uses failure assessment diagram (FAD) curve for checking the stability
of postulated axial and circumferential cracks. Using this method, crack initiation or
tearing instability can be determined by plo4tting the calculated assessment points
on FAD. For crack initiation, the assessment point must fall inside the curve. The
curves provided in ASME Appendix K [] are based on material properties which are
characteristics of reactor pressure vessel steels, which includes SA 302Gr A and B,
SA 533 Gr B class 1, and SA 508 classes 2 and 3 steels. The FAD curves for axial
and circumferential flaws are reproduced in Fig. 4. Figure 4 shows that all the
assessment points for a flaw having depth ranging from 0.25t to 0.35t lie inside the
failure assessment curve. Thus, the acceptance criteria based on J-integral approach
(method-1) are validated using alternative FAD approach, which accounts the
interaction effect of fracture and plastic instability.
Fig. 4 FAD along with failure assessment points for the postulated a axial flaws b circumferential
flaws for depth varying from 0.25t to 0.35t
4.4 Leak-Before-Break (LBB)—A Conceptual Approach
A part through crack in pressure vessel may grow by fatigue or stress corrosion
until it reaches the outer surface. This possibility exists if there is a crack at vessel
inside surface and it starts growing. If this crack once become through wall, the
vessel starts leaking, and there is sufficient time (time available between the
detection and the crack reaches a critical size or become unstable) available for
corrective action, then the criterion of leak-before-break is applicable. In order to
demonstrate the LBB criterion, it is essential to evaluate the limiting flaw length to
depth ratio which provides leak indication prior to break.
A simplified leak-before-break criterion was purposed by Irwin [10]. The criteria
were based on the fact that a surface flaw will be semicircular when it pops through
the vessel wall, implying that it develops a through-wall crack of length equal to
twice the vessel thickness. This approach is oversimplified approach and does not
give a solution for surface flaws longer than twice the thickness. A more general
LBB criterion can be derived on the basis of fracture mechanics principles in order
to evaluate the limiting flaw length to depth ratio.
The fracture condition of a surface flaw, after accounting the correction for back
free surface, plastic zone, and the proximity of the free surface in front of the crack
(Fig. 5), is given by [11]
p
K1ct ¼ 1:12Mk ðp1 =UÞð1 þ R=tÞ pa ð6Þ
where Mk is the Kobayashi stress intensity magnification factor, R is the mean

radius of vessel, p1 is the internal pressure for unstable propagation of surface flaw,
a is the flaw depth, and U can be approximated as,

U ¼ ðp=8Þ 3 þ a2 =c2 ð7Þ
where c is the half crack length.
Fig. 5 Schematic
representation of the crack
Consider a surface flaw which will grow to become a through-wall crack of size
2c. The fracture condition is given by [11],
p
K1c ¼ MF p2 ðR=tÞ pc ð8Þ
where MF is the Folias correction factor and is equal to 1 for large R/t ratio, p2 is the
unstable propagation of the through-wall crack, and c is the half crack length.
Crack will be arrested when the pressure to propagate a through-wall crack of
length ‘2c’ is greater than the pressure for instability of a flaw having a depth ‘a.’
Hence, the LBB criterion follows from p2 > p1 and is given by,
K1c UMF ðR=tÞ

[ pffiffiffi ð9Þ
K1ct 1:12MK 1 þ Rt a=c
Assuming that the vessel is made of rolled plates and possess anisotropy in
mechanical properties, specifically, in CVN or fracture toughness. The CVN or
fracture toughness in the transverse direction is around 65% of the longitudinal
values [12]. Further, it has been assumed that fracture toughness in transverse and
normal direction is same. This implies that the ratio K1c/K1ct for pressure vessel will
be 1.54.
Figure 6 shows the variation of the ratio K1c/K1ct versus a/2c for different
a/t ratio. The variation of limiting flaw length with flaw depth is shown in Fig. 7. It
is evident that for demonstration of LBB, the limiting length of flaw is around 4–6
times the flaw depth.
Fig. 6 Variation of the ratio (K1c/K1ct) versus a/2c for different a/t ratio
Fig. 7 Limiting flaw length

and flaw depth based on LBB
criterion
5 Conclusion
Structural integrity assessment of reactor pressure vessel of a typical BWR has been
carried out to demonstrate the protection of vessel against catastrophic failure,
plastic collapse, and tearing instability using state-of-the-art methodologies. As a
defense in depth, the vessel has been qualified for LBB criterion and limiting flaw
length to depth ratio has been quantified. The assessment has demonstrated the
healthiness of the vessel for its design life.
References
1. IAEA Nuclear Energy Series, No NT-T-3.11 (2009)

2. USNRC Regulatory Guide 1.99 Rev 2 (1988)
3. ASTM E 900-02, Standard guide for predicting radiation-induced transition temperature shift
in reactor pressure materials, E706, Reapproved-2007
4. ASME Section XI, 2015
5. S.C. Katiyar, S.S. Bajaj, Tarapur atomic power station units-1 and 2: design features,
operating experience, and license renewal. Nucl Energy Des 236(7–8), 881–893 (2006)
6. EPRI technical report, BWR Pressure Vessel Embrittlement correlation studies, 1019056
(2009)
7. E. Eason, G.R. Odette, R.K. Naustad, T. Yamamoto, A physically based correlation of
irradiation-induced transition temperature shifts for RPV steels, ORNL report, ORNL/
TM-2006/530 (2007)
8. WRC bulletin 413, Development of criteria for assessment of reactor vessels with low upper
shelf fracture toughness (Welding Research Council, New York, 1996)
9. EPRI report, Evaluation of the toughness properties and ASME service A/B upper shelf
toughness criteria for A 302B vessel steel, report no. TR-103434 (1994)
10. G.R. Irwin, Fracture of Pressure Vessels, McGraw-Hill (1963)
11. D. Broek, Elementary Engineering Fracture Mechanics (1978)
12. USNRC RG 1.161, Evaluation of Reactor Pressure vessel with Charpy upper–shelf energy
less than 50ft-lb (1995)
Sodium-Free Level Fluctuations: Concern
on Structural Integrity for the Pool-Type
Sodium-Cooled Fast Breeder Reactor
V. R. Chandan Reddy, R. Suresh Kumar, Anil Kumar Sharma,

K. Velusamy and P. Selvaraj
Abstract In the case of sodium-cooled pool-type fast breeder reactor, the weight of
the entire reactor assembly along with sodium coolant is carried by a large-sized
cylindrical vessel called the main vessel. Since main vessel is one of the primary
boundaries from the radiation shielding point of view, the structural integrity of the
main vessel is an essential safety feature to be ensured under all operating condi-
tions. The inner surface of the main vessel is partly filled with high-temperature
sodium coolant, and the remaining part is in contact with relatively low-temperature
argon cover gas. Due to large surface area of the main vessel pool, the free surface
level of the sodium oscillates. This environment creates drastic temperature cycling
in the main vessel wall at the sodium-free level interface. This transient temperature
cycling imposed on the main vessel with negligible attenuation can lead to
high-cycle thermal fatigue damage in the vicinity of sodium-free level. This
high-cycle thermal fatigue can be detrimental in ensuring the structural integrity of
the main vessel considering the number of cycles (approximately 9.5 108 cycles)
applied in the plant’s lifetime. This paper presents the numerical studies carried out
towards assessing the structural integrity of the main vessel by considering the
effect of sodium-free level fluctuations. The magnitudes of level fluctuations of 100
and 30 mm with a frequency range of 0.1–10 Hz with different increment of fre-
quencies are considered as a parametric study. In this study, the critical portion due
to level fluctuation has been identified from the thermal stress cycling point of view,
and the effect of level fluctuation frequency on the structural integrity of the main
vessel has been quantified.
Keywords Level fluctuation Fast breeder reactor High-cycle fatigue

Structural integrity
V. R. Chandan Reddy (&)

Homi Bhabha National Institute, Anushaktinagar, Mumbai 400094, India
e-mail: chandanreddy.vr@gmail.com
V. R. Chandan Reddy R. Suresh Kumar A. K. Sharma K. Velusamy P. Selvaraj
Reactor Design Group, Indira Gandhi Centre for Atomic Research, Kalpakkam 603102,
Tamil Nadu, India

https://doi.org/10.1007/978-981-10-7197-3_34
408 V. R. Chandan Reddy et al.
1 Introduction
Prototype fast breeder reactor (PFBR) is a 500-MWe-pool-type reactor, with

sodium as a coolant, which is contained within the main vessel of diameter 12,900
and 12,700 mm height. The total surface area of the main vessel is around 700 m2
and the volume is 1500 m3. Liquid sodium coolant with its excellent heat transfer
properties provides better natural heat removal capability in the pool-type reactor.
However, from the structural mechanics point of view, there are certain critical
problems, associated with sodium particularly during thermal transients that occur
under various operating conditions [1]. The structural material 316LN austenitic
stainless steel (SS 316LN) has low thermal conductivity and high thermal expan-
sion coefficient. The critical combination of SS 316LN and sodium induces high
thermal stress even for a small change in temperature. The main vessel is filled with
hot liquid sodium, and it is covered with inert argon cover gas to prevent it from
interacting with atmospheric gases. The heat transfer properties of argon are weak.
The thin walled vessel in the vicinity of sodium free level experience high tem-
perature fluctuations just below the free level, and the portion above the free level
experience low-temperature argon cover gas. This induces a large temperature
gradient in the vicinity of sodium-free level as shown in Fig. 1. Further, due to large
sodium pool surface area, the sodium-free level oscillates. So the temperature
gradient at the sodium-free level also oscillates inducing transient temperature
cycling in the main vessel-free level portion. These thermal fluctuations can lead to
high-cycle thermal fatigue of the immersed components.
Fig. 1 Temperature profile

oscillation due to sodium
level fluctuation in a reactor
vessel
Sodium-Free Level Fluctuations: Concern on Structural Integrity … 409
Due to good heat transfer properties of sodium, the wetted surface of the
structure attains the temperature of sodium immediately. But the depth of pene-
tration of temperature depends on the amplitude and frequency of surface tem-
perature oscillations. In the present study, detailed transient thermal stress analysis
has been performed using the idealised model of the reactor vessel as depicted in
Fig. 2a to study the characteristics of thermal stress fluctuation in the main vessel
structural wall.
Fig. 2 a Mathematical model of reactor vessel, b finite element mesh

2 Frequency and Amplitude of Level Fluctuations
The typical main vessel geometry considered for studying the level fluctuation is
12,900-mm-diameter and 25-mm-thick vessel. The experiments conducted by
Laxman et al. [2] on the ¼-scaled model of reactor assembly described the
amplitude and frequency of level fluctuations in the reactor pool. The amplitude of
level fluctuation varies from 38 to 92 mm after applying the proper scaling laws.
The frequency of oscillations during the normal operating condition is in the range
of 0.1–2.0 Hz. This paper also presents the study carried out for the reactor vessel
for the amplitude of level fluctuations of 100 and 30 mm for a frequency range of
0.1–10 Hz using finite element simulation. The loading cycle for the level fluctu-
ation is shown in Fig. 2a.
3 Transient Heat Transfer Analysis
The reactor vessel is modelled as an axisymmetric shell structure. Since the effect of
variation is confined within 1000 mm above and 1000 mm below the vicinity of
sodium-free level the main vessel is analysed for 2100 and 2030 mm for 100 and
30 mm of level fluctuation, respectively. It is analysed for a temperature difference
of 100 K between sodium (675 K) and argon (575 K). An additional subroutine is
formulated in the numerical tool ABAQUS [3] to create cyclic temperature varia-
tion at the curved surface of the cylinder by imposing moving temperature profile
along the axis of the cylinder. The temperature profile was generated by varying the
film coefficient along the axis of the cylinder [4]. Film coefficients considered for
sodium is 25,000 W/m2 K while that of argon is 10 W/m2 K. The rate of change of
the film coefficient with respect to the surface temperature is kept equal to zero.
Temperature-dependent properties of SS 316LN used for the analysis are shown in
Table 1. The reactor vessel is analysed for 1000 s for a frequency range of 0.1–
10 Hz for both 100 and 30 mm level fluctuations.
The finite element mesh for the analysis is shown in Fig. 2b. The outcome of the
analysis is the transient temperature variation at each node of the finite element
model. The variation of temperature for the node 17 (node corresponding to the
maximum amplitude of temperature fluctuation on the surface of the main vessel
wall) and other nodes along the thickness of the reactor vessel wall at 1 Hz fre-
quency for 100 s is shown in Fig. 3a. Initially, the amplitude of temperature
variation is about 60 K on the wall surface for a temperature difference of 100 K
between argon and sodium. This temperature amplitude decays with time, and
steady temperature amplitude of about 20 K is achieved after about 200 cycles of
level variation, and it remains same further. The temperature amplitude decay trend
remains similar for all other frequencies.
Table 1 Temperature-dependent properties of SS 316 LN

Temperature Density Thermal Specific Coefficient of Young’s
(K) (kg/m3) conductivity heat thermal modulus
(W/m K) capacity expansion (GPa)
(J/kg K) (10−6/K)
293 7966 13.94 470 15.9 192
323 7949 14.37 476 16.1 190
373 7932 15.08 486 16.4 186
423 7910 15.80 497 16.7 182
473 7889 16.52 508 17.0 178
523 7867 17.24 518 17.2 174
573 7846 17.95 529 17.5 170
623 7824 18.67 539 17.7 166
673 7803 19.39 550 17.9 161
723 7781 20.10 560 18.1 157
3.1 Temperature Variation Along the Length

of the Reactor Vessel
The amplitude of temperature fluctuation on the surface of the reactor vessel wall is
a function of the frequency of level oscillation and the plots for 100 and 30 mm
level fluctuations at 1000 s are shown in Fig. 4a, b respectively. The maximum
temperature amplitude is for 0.1 Hz frequency in both the cases, as the reactor
vessel wall will be in contact with hot sodium for more time at 0.1 Hz frequency
compared to other higher frequencies. The amplitude of temperature fluctuation on
the surface varies along the length of the cylinder and is highest at approximately 90
and 27 mm from the bottom most point of sodium-free level for 100 and 30 mm
level variations, respectively. These are the points which will be in contact with
low-temperature argon cover gas for major part of the cycle. Figure 4a, b depict the
amplitude of temperature fluctuations with height of reactor vessel in the vicinity of
sodium-free level. From these figures, it is clear that the amplitude of temperature
variation on the surface of reactor vessel wall depends only on the frequency of
level variation and not a function of its amplitude.
3.2 Temperature Variation Along the Thickness

of the Reactor Vessel Wall
The amplitude of temperature fluctuation in the wall decays with the increasing
depth from the surface of reactor vessel wall. The temperature amplitude along the
thickness of the cylinder at node 17 for 100 mm level fluctuation at 1000 s is
shown in Fig. 6a. The temperature amplitude reduces to a very low value with in
Fig. 3 a Temperature variation at node 17 and other nodes across wall thickness, b temperature
contour plot at node 17 at 100 s
half the thickness of the reactor wall. And hence, the point on the inner surface of
the reactor vessel wall is the location with maximum temperature amplitude at any
point along the length of the cylinder, and the temperature amplitude is negligible
beyond half the thickness of reactor vessel wall.
Fig. 4 a Amplitude of temperature fluctuation along the length of the reactor vessel for 100 mm
level fluctuation, b amplitude of temperature fluctuation along the length of the reactor vessel for
30 mm level fluctuation
4 Transient Thermal Stress Analysis
The results of the transient heat transfer analysis are used as input for the transient
linear elastic stress analysis. The same mesh of heat transfer analysis as shown in
Fig. 2b is used for the stress analysis. The thermal expansion coefficient for the SS
316LN is defined as a temperature dependent property and the nodal stress com-
ponents field outputs are requested. The stress variation with time for node 17 and
other nodes along the thickness of the reactor vessel wall are shown in Fig. 5a. It
shows that initially the amplitude of stress is high and decays with time. Steady
stress amplitude is achieved after about 200 cycles of level variation. This initial
transient stress region appears whenever the reactor is restarted. Since 100 reactor
scrams are assumed in the design during the plant life of 40 years, this transient
Fig. 5 a Stress across wall thickness at node 17, b stress fluctuation amplitude along the length of
the reactor vessel
region is getting repeated accordingly. The damage fraction for this initial transient
region is higher that the damage caused during steady-state stress amplitude. Thus
depending on the number of scrams in a reactor life time, this transient stress needs
to be converted to equivalent steady-state stress amplitude for accurate damage
prediction.
Stress amplitude along the length of the reactor vessel increases with the
amplitude of level fluctuation similar to temperature amplitude and is shown in
Fig. 5b. It shows that the stress amplitude is a function of frequency of the level
fluctuation and it increases with the decrease in frequency thus maximum stress
amplitude is for a frequency of 0.1 Hz. The location with the maximum amplitude
of stress variation is at about 90 and 27 mm from the bottom most point of
sodium-free level for 100 and 30-mm-level variation, respectively. This is equiv-
alent to 0.9 times the amplitude of level variation. Hence, the location of maximum
stress amplitude is at about 0.9 times the amplitude of level variation from the
bottom most point of sodium-free level irrespective of the amplitude of level
variation.
The stress amplitude variation along the thickness of the reactor vessel is a
function of the frequency of level fluctuation and is shown in Fig. 6b. The stress
amplitude decays exponentially along the thickness and the stress variation beyond
half the thickness of reactor vessel wall can be neglected. This indicates that the
stress in the reactor vessel wall due to sodium level fluctuation acts like skin
stresses with a large temperature gradient along the thickness of the wall. These
skin stresses are critical in structural integrity point of view as the cracks are mostly
initiated at the surface of the metal walls.
It can be seen from Fig. 7 that the maximum stress amplitude induced in the
reactor vessel wall subjected to sodium-free level fluctuation is a function of the
frequency of level fluctuation and independent of amplitude of level variation. Even
though the difference between the two amplitudes of level variations is about
70 mm, there is no significant change in the maximum stress amplitude induced in
the reactor vessel wall.
5 Conclusion
The analysis carried out in the present study deals with the numerical studies carried
out to access the affect on the structural integrity of the main vessel subjected to
sodium-free level fluctuations in a typical 500-MWe-sodium-cooled fast breeder
reactor. The magnitude of level fluctuation of 100 and 30 mm for different fre-
quencies in the range of 0.1–10 Hz are considered in the numerical analysis. It is
found that the critical location of maximum stress amplitude is independent of
amplitude of sodium-free level fluctuation and is predominant particularly at lower
frequency range of 0.1–2 Hz which is also the frequency range of interest of present
study.
Fig. 6 a Amplitude of temperature fluctuation along the thickness of the reactor vessel at node 17,
b amplitude of stress fluctuation along the thickness of the reactor vessel at node 17
Fig. 7 Maximum amplitude

of stress variation induced in
the reactor vessel wall due to
sodium-free level fluctuation
Apart from the aforementioned high-cycle thermal fatigue (HCF), the

sodium-free level zone is further subjected to low-cycle fatigue (LCF) due to
reactor scram and restart cycles. As the zone is subjected to both LCF and HCF
simultaneously the cumulative damage will be higher than the individual damages.
At present, matured design rules are not available to calculate damage due to this
complex LCF–HCF interaction phenomenon. Considering the importance of the
main vessel in maintaining the structural integrity of the plant, this paper is the first
step in accurate prediction of damage due to this complex phenomenon.
Acknowledgements The authors are thankful for the support and motivation by Dr. S.A.V. Satya
Murthy, Director, IGCAR and Shri P. Puthiyavinayagam, Director, RDG. The Authors gratefully
acknowledge the grant of the research fellowship from Indira Gandhi Centre for Atomic Research,
Department of Atomic Energy, India.
References
1. K. Velusamy, P. Chellapandi, S.C. Chetal, B. Raj, Overview of pool hydraulic design of Indian
prototype fast breeder reactor. Sadhana-Acad. Proc. Eng. Sci. 35(2), 97–128 (2010)
2. D. Laxman, P.K. Gupta, C. Asokane, J. VenkataRamanan, C.A. Gopal, K. Rajesh, G.
Padmakumar, V. Prakash, G. Vaidyanathan, Experimental study of free level fluctuations in 1/4
scale reactor assembly model of PFBR, NUTHOS-6 (Nara, Japan, 2004)
3. ABAQUS 6.10, Dassault Systèmes Simulia Corp (Providence, RI, USA, 2010)
4. A. Mishra, R.S. Kumar, P. Chellapandi, Progressive deformation behaviour of thin cylindrical
shell under cyclic temperature variation using Combined Hardening Chaboche Model. Lat.
Am. J. Solids Struct. 11(6), 980–992 (2014)
Monotonic Fracture Studies on Bi-metallic
Pipe Weld Joints Having Circumferential
Through-Wall Crack
G. Raghava, S. Vishnuvardhan, M. Saravanan, P. Gandhi,

Suranjit Kumar, P. K. Singh, I. A. Khan and V. Bhasin
Abstract Bi-metallic welded joints are necessary in pressurized and boiling water
reactors where heavy section low alloy carbon steel components are connected to
stainless steel primary piping systems. The behaviour of such bi-metallic weld
joints is very difficult to understand, and very limited experimental data are
available in the literature on the fracture behaviour of such joints. It is essential to
develop fracture assessment methodologies for these joints and validate them
experimentally to ensure safe operation of power plants. In this background, frac-
ture studies were carried out under monotonic bending on seven numbers of
bi-metallic pipe weld joints of 324 mm nominal outer diameter having circumfer-
ential through-wall crack. The specimens were made of SA 508 Gr.3 Cl-1 low alloy
steel (ferritic) pipe on one half and SA312 Type 304LN stainless steel (austenitic)
pipe on the other half. The initial through thickness notch was located in the
different regions of the weld joints such as weld centre, buttering, heat-affected
zones and base metals (low alloys steel and stainless steel). Initial notch angle was
either 60° or 90°. Prior to the fracture tests, fatigue pre-cracking was carried out to
create sharp crack front. Subsequently, fracture tests were conducted under
four-point bending and displacement control. The collapse load of the weld joint
having crack in the buttering layer is 12% less compared to that having crack in the
heat-affected zone. Increase in the crack angle present in the heat-affected zone
from 60° to 90° reduced the collapse load of the weld joint by 19%.

Keywords Bi-metallic pipe weld joints Monotonic fracture Heat-affected zone

Buttering Through-wall notch Four-point bending
G. Raghava (&) S. Vishnuvardhan M. Saravanan P. Gandhi

e-mail: raghavag@serc.res.in
S. Kumar P. K. Singh I. A. Khan V. Bhasin
Bhabha Atomic Research Centre, Mumbai 400085, India

https://doi.org/10.1007/978-981-10-7197-3_35
420 G. Raghava et al.
Nomenclature
2C Initial notch length
D Diameter of the pipe
l Length of the pipe
L Inner span
Rm Mean radius of the pipe
t Thickness of the pipe
W Width of the notch
Z Outer span
h Half notch angle
ry Yield strength
ru Ultimate tensile strength
rf Flow stress of the material
1 Introduction
Bi-metallic weld joints are used in the primary heat transport system of light water
cooled reactors of nuclear power plants. These weld joints are used to join pressure
vessel nozzle (low alloy steel) to stainless steel piping. Large difference in physical,
chemical and mechanical properties of low alloy steel and stainless steel calls for a
deposit of material (buttering) on ferritic steel with intermediate properties. This
helps in minimizing the residual stresses and distortion during welding. The but-
tering material is generally nickel-rich Inconel alloy. The bi-metallic weld joints
have varying metallurgical, mechanical and fracture toughness properties across the
joint, which affects the integrity of the joints. The behaviour of such bi-metallic
weld joints is very difficult to understand, and very limited experimental data are
available in the literature on the fracture behaviour of such joints. Therefore, it is
essential to develop fracture assessment methodologies for these joints and validate
them experimentally to ensure safe operation of power plants.
Brocks et al. [1] conducted two large-scale four-point bend tests on a piping
assembly containing a SA 508 ferritic steel to AISI 304 L stainless steel weld with a
notch at the interface between the ferritic steel and the buttering layer. Crack growth
analyses were also carried out at material level, and the results were compared with
the results of finite element analysis. A crack tip opening displacement (CTOD)
criterion was applied to predict crack initiation and growth. CTOD appeared to be
better suited for welded structures than J as CTOD is a near-field quantity which
Monotonic Fracture Studies on Bi-metallic Pipe Weld Joints … 421
does not suffer from problems like path dependence, material gradients, etc., and
CTOD-based resistance curves can be easily determined experimentally even for
welded structures. Faidy et al. [2, 3] studied performance of two benchmark
four-point bend pipe tests, conducted on a nominal 6-inch piping assembly, con-
taining a ferritic to stainless steel (A508-308/309SS-304SS) bi-metallic weld
(BMW). The weld was notched at the ferritic steel/buttering layer interface, to
simulate plant experiences of the cracking behaviour of such components. A range
of analysis methodologies including conventional flaw assessment methods,
J approaches and local approach methods were applied to predict the critical load
for initiation of the defect, the extent of crack growth and the path followed by the
crack through the weld up to the maximum load.
Guozhen et al. [4] conducted fracture tests and microscopic observations on an
Alloy 52 M dissimilar metal welded joint (DMWJ) between A508 ferritic steel and
316L stainless steel in nuclear power systems. The results showed that the crack
growth always deviates to the materials with lower strength, and the crack path
deviations are mainly controlled by the local strength mismatch. If the local fracture
resistance properties could not be obtained and used for cracks in heat-affected zone
(HAZ), interface and near interface zone, the use of the J-resistance curves of base
metals or weld metals following present codes will unavoidably produce
non-conservative (unsafe) or excessive conservative assessment results.
Very limited experimental data are available in the literature on fracture beha-
viour of bi-metallic pipe weld joints. Hence, a systematic study was carried out by
the authors on the fracture behaviour of straight pipe weld joints having circum-
ferential through-wall crack.
2 Material Properties
The bi-metallic pipe weld specimens were made of base materials SA 508 Gr.
3 Cl-1 low alloy steel and SA 312 Type 304LN stainless steel, and buttering
material was nickel-rich Inconel alloy (ENiCrFe-3) conforming to AWS A5.11/A
5.11M-2010 [5]. The SA 508 Gr. 3 Cl-1 material conformed to ASTM A508/
A508M-14 [6], and the SA 312 Type 304LN stainless steel conformed to ASTM
A312/A312M-14b [7]. The chemical composition and the mechanical properties of
steels used in bi-metallic pipe weld joints are given in Tables 1 and 2.
422
Table 1 Chemical composition of steels used in bi-metallic pipe weld joints

Elements Chemical compositions (%)
Low alloy carbon steel Stainless steel ENiCrFe-3
Tested ASTM A508/A508M-14 Tested values ASTM A312/ Tested AWS A5.11/
values [8] Grade 3 Class 1 [6] [8] A312M-14b [7] values [8] A5.11M-2010 [5]
Carbon 0.190 0.250 (max) 0.023 0.035 0.038 0.100 (max)
Manganese 1.300 1.200–1.500 0.820 2.000 6.750 5.000–9.500
Phosphorous 0.018 0.250 (max) 0.024 0.045 0.014 0.030 (max)
Sulfur 0.002 0.250 (max) 0.001 0.030 0.015 0.015 (max)
Silicon 0.230 0.400 (max) 0.460 1.000 0.870 1.000 (max)
Nickel 0.700 0.400–1.000 8.17 8.000–12.000 66.000 59.000 (min)
Chromium 0.170 0.250 (max) 18.13 18.000–20.000 16.590 13.000–17.000
Molybdenum 0.440 0.450–0.600 0.260 – 0.800 –
Copper 0.130 0.200 (max) 1.070 – 0.080 0.500 (max)
Titanium – 0.015 (max) – – 0.029 1.000 (max)
Aluminum 0.020 0.025 (max) 0.027 – 0.062 –
Niobium – – – – 2.100 1.000–2.500
Tantalum – – – – – 1.000–2.500
Iron 96.8 – 71.015 – 6.240 10.000 (max)
G. Raghava et al.
Table 2 Mechanical properties of steels used in bi-metallic pipe weld joints

Properties Carbon steel Stainless steel
Tested ASTM A508/A508M-14 Tested ASTM A312/
values [8] Grade 3 Class 1 [6] values [8] A312M-14b [7]
Yield strength, 570 450 208 205 MPa
ry (MPa)
Ultimate tensile 694 620–795 549 515 MPa
strength, ru (MPa)
Percentage 23 16 81 35 (L)
elongation – – – 25 (T)
L Longitudinal; T Transverse
3 Experimental Investigations
3.1 Details of Bi-metallic Pipe Weld Joints
The bi-metallic pipe weld joints were of 324 mm nominal outer diameter having
circumferential through-wall crack. The specimens were made of SA 508 Gr.3 Cl-1
low alloy steel (ferritic) pipe on one half and SA312 Type 304LN stainless steel
(austenitic) pipe on the other half. The length of the specimens was 4000 mm, and
the average thickness was 24 mm. Initial notch angle was either 60° or 90°; the
corresponding notch length was 170 or 254 mm. Table 3 gives details of the
bi-metallic pipe weld joints.
3.2 Details of Fatigue Pre-cracking
Prior to the fracture tests, fatigue pre-cracking of the bi-metallic pipe weld specimens
was carried out under four-point bending, under load control, using a ±1000 kN
capacity servo-hydraulic actuator. The inner and outer spans during fatigue
pre-cracking of the pipes were 972 and 3500 mm, respectively. The load ratio was
0.1, and the frequency was maintained in the range of 0.5–1.8 Hz. The maximum
cyclic load was around 20% of the Theoretical Plastic Collapse Load (TPCL) of the
pipe weld specimen. The TPCL was calculated using the following Eq. (1):

16rf R2m t h 1 h
PL ¼ cos sin ð1Þ
ZL 4 2 2
Flow stress (rf) is defined as the mean of yield strength and ultimate tensile
strength. Minimum yield strength and tensile strength were considered for calcu-
lation of collapse load. Fatigue pre-cracking was carried out till the crack growth in
the circumferential direction reached approximately 2 mm at both the notch tips.
Table 4 gives details of inner and outer spans, frequency and magnitude of cyclic
424
Table 3 Details of bi-metallic pipe weld joints

Specimen ID Carbon steel Stainless steel l (mm) Details of notch
D (mm) t (mm) D (mm) t (mm) 2C (mm) W (mm) 2h (o)
QMSP-1-60TWC-Fe 325 24.09 323 23.96 4000 169 2.12 59.77
QMSP-2-60TWC-CW 325 24.72 324 24.01 4013 174 2.35 61.44
QMSP-3-90TWC-CW 325 24.14 324 23.72 4007 254 2.37 89.69
QMSP-4-60TWC-CB 325 23.99 324 23.37 4008 175 3.25 61.79
QMSP-5-60TWC-CHAZ 326 25.13 325 24.50 4011 172 3.00 60.55
QMSP-6-60TWC-Au 326 24.21 326 23.77 4012 170 2.38 59.75
QMSP-7-90TWC-CHAZ 325 24.37 324 23.99 4012 255 3.00 90.05
Type of notch Through-wall; Fe Crack is located at ferritic region; CW Crack is located at centre of weld; CB Crack is located at centre of buttering; CHAZ
Crack is located at centre of heat-affected zone; and Au Crack is located at austenite region
G. Raghava et al.
Table 4 Details of fatigue pre-cracking of bi-metallic pipe weld joints
Specimen ID Span (mm) Frequency Cyclic load Number of Crack length* Crack angle*, 2h
(Hz) (kN) cycles (mm) (o)
Inner Outer Min. Max. Tip Tip
span span A B
QMSP-1-60TWC-Fe 972 3500 0.5 16 160 17,500 2.92 2.95 61.85
QMSP-2-60TWC-CW 972 3500 0.5 16 160 29,500 2.85 2.18 63.22
QMSP-3-90TWC-CW 972 3500 0.5 12 120 24,000 2.67 3.21 91.77
QMSP-4-60TWC-CB 972 3500 0.5 16 160 45,000 1.71 0.85 62.70
QMSP-5-60TWC-CHAZ 972 3500 0.5 16 160 17,500 2.83 2.27 62.35
QMSP-6-60TWC-Au 972 3500 0.5–1.5 12 120 105,250 2.91 7.13 63.29
Monotonic Fracture Studies on Bi-metallic Pipe Weld Joints …
QMSP-7-90TWC-CHAZ 972 3500 0.5 12 120 21,000 0.75 2.00 90.74

*After fatigue pre-cracking
425
±1000 kN
capacity actuator
Pipe specimen
Fig. 1 A view of fatigue pre-cracking of a bi-metallic pipe weld joint
loading during fatigue pre-cracking of bi-metallic pipe weld joints. The pipes were
simply supported with hinge support at one end and roller support at the other end.
A steel distribution beam was used to apply cyclic load at two points along the span
to simulate four-point bending with constant bending moment between the two
loading points. Figure 1 shows a view of fatigue pre-cracking of a bi-metallic pipe
weld joint. Digital camera was used to monitor surface crack growth during fatigue
pre-cracking.
3.3 Fracture Tests
3.3.1 Test Set-up Arrangement
After fatigue pre-cracking, the welded pipe specimens were subjected to monotonic
loading with intermittent unloading to carry out fracture test. The experimental
set-up used for carrying out fracture tests was the same as that used for carrying out
fatigue pre-cracking. A view of fracture test set-up for a bi-metallic pipe weld joint
is shown in Fig. 2. Fracture tests were carried out using ±1000 kN/±2000 kN
capacity servo-hydraulic actuator. The servo-hydraulic actuator was fixed to a steel
reaction frame, which was connected to the strong floor of the laboratory. The
actuator system consists of an inbuilt LVDT for measuring displacement and a load
cell for measuring the applied load. Figure 3 shows close-up view of a bi-metallic
pipe weld joint and fracture test set-up.
Fig. 2 Fracture test set-up for bi-metallic pipe weld specimen QMSP-1-60TWC-Fe
Fig. 3 Close-up view of the bi-metallic pipe weld specimen and fracture test set-up
3.3.2 Instrumentation and Data Acquisition
During the fracture tests, the pipe weld specimens were instrumented to obtain the
following data:
• Applied load from inbuilt load cell in servo-hydraulic actuator
• Load-line displacement (LLD) from inbuilt LVDT in servo-hydraulic actuator
• Crack mouth opening displacement (CMOD) using Clip gauge
• Surface crack growth using Image Processing Technique
• Deflection of bi-metallic welded pipe specimen using LVDT
• Angular rotation of pipe using inclinometer
• Circumferential deformation (ovality) using ovality ring with LVDTs.
The applied load and load-line displacement were measured directly using strain
gauge-based load cell and inbuilt LVDT of the actuator. Crack mouth opening
displacement was measured using specially fabricated clip gauges. Strain
gauge-based clip gauges with opening of 5–30 and 15–60 mm range were used, and
these were calibrated prior to the tests to the required range of displacement. The
clip gauges were connected to the data acquisition system, and the data were
continuously acquired. Circumferential deformation (ovality) measurements were
also made at the stainless steel portion of the pipe using an ovality set-up. The
ovality set-up consisted of an ovality ring and four LVDTs. IPT was used to
measure surface crack growth at both the crack tips. The IPT consisted of three
CCD cameras interfaced to a computer system with image analysis software.
Two CCD cameras were focused towards the crack tips. A grid of 5 mm spacing
was made on the welded pipe specimen near the crack tips to obtain the crack
growth data. The third CCD camera was focused towards the control console of the
actuator to record the images of load and displacement corresponding to the crack
growth at various stages of loading. The images at the two crack tips gave the
surface crack growth. The deflection of the pipe weld specimen was measured by
means of ±100 mm range LVDTs kept along the span. The LVDTs were con-
nected to MGC Plus data logger interfaced to a computer. Totally, five post-yield
strain gauges were mounted on the pipe specimen to measure the strains during the
fracture tests. Out of the five strain gauges, four strain gauges were mounted
adjacent to the two notch tips, and the remaining one strain gauge was mounted on
the opposite side of the notch.
3.3.3 Details of Loading
The bi-metallic welded pipe specimens were subjected to monotonic loading with
intermittent unloading. The static monotonic load was applied in displacement
control at the rate of 0.04 mm/s initially till the displacement value reached 20 mm.
Then, the specimen was unloaded in displacement control at the rate of 0.04 mm/s
till the displacement value dropped by 2 mm. Subsequently, the specimen was
loaded till the increment in displacement value reached 10 mm (i.e. 30 mm from
beginning). The procedure was continued in this manner till the end of the test.
During the fracture tests, load, load-line displacement, crack mouth opening dis-
placement, ovality in the austenitic steel pipe at 245 mm from notch plane and
deflection of the specimens (at five locations) were continuously monitored. Surface
crack growth was monitored using Image Processing Technique (IPT).
Additionally, angular rotation at both the loading supports was also recorded with
respect to load and load-line displacement.
Table 5 gives the results of fracture tests on bi-metallic pipe weld joints [9–12].
Figure 4a–c shows the load versus load-line displacement curves for the joints.
Figure 5a–c shows the load versus crack mouth opening displacement curves.
Presence of initial notch of length 254 mm reduced the collapse load of the
specimen by 30% when compared with the specimen having initial notch of length
174 mm at the centre of the weld. The collapse load of the weld joint having crack
in the buttering layer is 12% less compared to that with crack in the heat-affected
zone. Increase in the angle of crack in the heat-affected zone from 60° to 90°
reduced the collapse load of the weld joint by 19%. In the case of specimen having
initial crack in the HAZ towards ferritic region, crack grew towards the fusion line
between buttering and ferritic region. In the case of specimen having initial crack in
the austenitic region, crack growth was in the plane of the notch till the end of the
test. In the remaining specimens, i.e. specimens having initial crack in the centre of
the weld, centre of the heat-affected zone and centre of the buttering, propagation of
the crack was towards the stainless steel region. Figure 6 shows deflected shape of
the specimen QMSP-6-60TWC-Au at the end of fracture test, and Fig. 7 shows
close-up views of crack tip locations at the end of fracture test. Final crack lengths
(including initial notch length) for the specimens at the end of fracture test varied
from 247 to 455 mm. Results of the studies will be useful in developing fracture
assessment methodologies which are required to ensure structural integrity of
bi-metallic pipe weld joints.
430
Table 5 Results of fracture tests on bi-metallic pipe weld joints

Specimen ID Span (mm) MMLa Load-line displacementb CMODb Crack lengthc Final crack angle,
(kN) (mm) (mm) (mm) 2h (o)
Inner Outer Tip A Tip B
span span
QMSP-1-60TWC-Fe 1134 3700 1001.02 210.50 17.10 45.21 30.15 86.42
QMSP-2-60TWC-CW 972 3700 804.97 96.24 14.02 92.28 87.50 126.71
QMSP-3-90TWC-CW 972 3700 562.60 58.70 13.30 96.13 89.70 157.39
QMSP-4-60TWC-CB 972 3700 753.40 91.90 12.10 81.91 83.18 120.10
QMSP-5-60TWC-CHAZ 972 3700 854.70 140.00 17.90 46.22 55.37 96.32
QMSP-6-60TWC-Au 1134 3700 758.40 124.20 25.10 106.53 108.69 135.41
QMSP-7-90TWC-CHAZ 972 3700 691.40 95.90 16.40 120.17 80.29 160.34
c
MML Maximum Measured Load; aAs recorded in data logger; bValues corresponding to MML; As obtained from IPT
G. Raghava et al.
Fig. 4 Load versus load-line displacement curves for the bi-metallic pipe weld joints
Fig. 5 Load versus CMOD curves for the bi-metallic pipe weld joints
Fig. 6 Deflected shape of the

specimen QMSP-6-60TWC-
Au at the end of fracture test
Fig. 7 Close-up views of

crack tip locations at the end
of fracture test for the
specimen QMSP-6-60TWC-
Au
5 Summary and Conclusions
Fracture studies were carried out under monotonic bending on seven numbers of
bi-metallic pipe weld joints of 324 mm nominal outer diameter having circumfer-
ential through-wall crack. The specimens were made of SA 508 Gr.3 Cl-1 low alloy
steel (ferritic) pipe on one half and SA312 Type 304LN stainless steel (austenitic)
pipe on the other half. The initial through thickness notch was located in the
different regions of the weld joints such as weld centre, buttering, heat-affected
zones and base metals (low alloy steel and stainless steel). Based on the results of
the studies, the following conclusions are made:
• In the case of specimens having notch at the weld centre, collapse load of the
specimen with initial notch length of 254 mm reduced by 30% in comparison
with that of the specimen with initial notch length of 174 mm.
• The collapse load of the weld joint having crack in the buttering layer is 12%
less in comparison to that with similar crack in the heat-affected zone, the initial
notch length being approximately 175 mm in both the cases. Increase in the
initial notch angle in the heat-affected zone from 60° to 90° reduced the collapse
load of the weld joint by 19%.
• In the case of specimen having initial crack in the HAZ towards ferritic region,
crack grew towards the fusion line between buttering and ferritic region.
• In the case of specimen having initial crack in the austenitic region, crack
growth was in the plane of the notch till the end of fracture test.
• The specimens having initial crack in the centre of the weld, centre of the
heat-affected zone and centre of the buttering, propagation of the crack was
towards the stainless steel region.
• Results of these studies will be useful in developing fracture assessment
methodologies which are required to ensure structural integrity of bi-metallic
pipe weld joints.
Acknowledgements The authors from CSIR-SERC thank Prof. Santosh Kapuria, Director and
Dr. K. Balaji Rao, Advisor (Management), CSIR-SERC, Chennai, for their valuable guidance,
encouragement and support in the R&D activities. The assistance rendered by the technical staff of
the Fatigue and Fracture Laboratory, CSIR-SERC in conducting the experimental investigations is
gratefully acknowledged. This paper is published with the permission of the Director,
CSIR-SERC, Chennai.
References
1. W. Brocks, A. Cornec, W. Schönfeld, Fracture mechanics assessment of a bi-metallic welded

joint, in Proceedings of the 14th European Conference on Fracture Mechanics (ECF 14),
Krakow, Poland, 2002
2. C. Faidy, G. Chas, S. Bhandari, M.P. Valeta, R. Hurst, A. Youtsas, P. Nevasmaa, W. Brocks,
D. Lidbury, C. Weisner, Structural integrity of bi-metallic components BIMET, Brite-Euram
Final Project Report, 2001
3. C. Faidy, G. Chas, S. Bhandari, M.P. Valeta, R. Hurst, A. Youtsas, P. Nevasmaa, W. Brocks,
D. Lidbury, C. Weisner, BIMET: structural integrity of bi-metallic components. Nucl. Engg.
Desig. 209(1–3), 79–85 (2001)
4. G. Wang, H. Wang, F. Xuan, S. Tu, Local fracture behavior and integrity assessment of a
dissimilar metal welded joint in nuclear power systems, in: Proceedings of 13th International
Conference on Fracture, Beijing, China, 6–21 June 2013
5. AWS A5.11/A 5.11M—2010, Specification for nickel and nickel-alloy welding electrodes for
shielded metal arc welding, American Welding Society. An American National Standard
(2010)
6. ASTM A 508/A 508M—14, Standard specification for quenched and tempered
vacuum-treated carbon and alloy steel forgings for pressure vessels. ASTM International
(2014)
7. ASTM A 312/A 312M—14b, Standard specification for seamless, welded, and heavily cold
worked austenitic stainless steel pipes. ASTM International (2014)
8. S. Kumar, P.K. Singh, I.A. Khan, V. Bhasin, R.K. Singh, K.K. Vaze, K.N. Karn, J. Derose, S.
P. Srivastava, A.K. Sinha, Evaluation of tensile and fracture properties of dissimilar metal
pipe weld: low alloy ferritic steel and austenitic stainless steel. Technical Report, BARC/
2014/E/2014, Bhabha Atomic Research Centre, Mumbai (2014)
9. M. Saravanan, S. Vishnuvardhan, G. Raghava, P. Gandhi, Monotonic and cyclic fracture
studies on bi-metallic pipe weld joints. Sponsored research report no. R&D 03-SSP
14541-SR-1, Jan 2015(CSIR—Structural Engineering Research Centre, Chennai)
14541-SR-2, June 2015 (CSIR—Structural Engineering Research Centre, Chennai)
11. M. Saravanan, S. Vishnuvardhan, G. Raghava, P. Gandhi, ‘Monotonic and cyclic fracture
14541-SR-3, Sept 2015 (CSIR—Structural Engineering Research Centre, Chennai)
14541-SR-4 and Final, Dec 2015 (CSIR—Structural Engineering Research Centre, Chennai)
Part VII
Experimental Techniques
Assessment of Strain in a Corrosive
Environment of Structural Steel
A. Cinitha, P. K. Umesha and K. Kesavan
Abstract Corrosion damage not only causes structural failure, but also leads to
maintenance and repair costs. Improving the effective sensor technology for corro-
sion detection can reduce the economic loss and provide safe structures with longer
service. This paper presents an evaluation technique developed to assess the pro-
gressive strain in a corrosive environment using fiber Bragg grating (FBG) sensors.
The progress in corrosion process of structural steel coupons under stressed and
unstressed condition in a laboratory environment is studied. The strains are mea-
sured with electrical resistance strain gauges and FBG sensors and are compared.
Strain in coupons under stressed and unstressed condition is measured, and distinct
strain behavior is brought out along with reliability of measurements in adverse
environment. The central wavelength shift and peak power of FBG are analyzed
with interrogator, and the capability of FBG sensor to monitor the progress in
corrosion is brought out.

Keywords Corrosion Fiber Bragg grating sensor Electrolyte
Electrical resistance strain gauge
Nomenclature
n Effective refractive index of the core
pe Effective photo-elastic coefficient
Pi.j Coefficients are the Pockel’s coefficients of the stress-optic tensor
DT Incremental increase in temperature
DM Incremental increase in mass loss
DR Iresistance change in the gauge in ohms
Ro Original or unstrained gauge resistance
aT Coefficient of thermal expansion
e Unit engineering strain in the specimen surface
m Poisson’s ratio
A. Cinitha (&) P. K. Umesha K. Kesavan

e-mail: cinitha@serc.res.in

https://doi.org/10.1007/978-981-10-7197-3_36
K Grating pitch
De Incremental increase in strain
a, b, c Constant coefficients
kB Bragg wavelength
1 Introduction
Fiber optic-based sensing technology has shown remarkable progress and accuracy
level in structural health monitoring of bridges, towers, industrial plants, silos,
offshore platforms, etc. These optical sensors, which are superior in durability,
corrosion resistance, waterproof, and explosion proof, provide numerous advan-
tages over conventional electrical resistance sensors such as their small size, their
low weight, or their immunity to electromagnetic interference. Fiber Bragg grating
(FBG) sensors are one of the advanced devices in this technology, especially to
measure strain and deformations in severe environments. FBG-based sensors are
capable to measure the physical parameters such as strain, pressure, and tempera-
ture [1–3] and had reviewed advances in fiber Bragg sensor technologies. Although
the conventional electric resistance-based sensors were well established and are still
under continuous development, fiber grating sensors have exhibited their distinct
advantages and competitiveness in developing, e.g., sensor array multiplexed in
serial, remote sensors for harsh environment monitoring, and ultrasensitive
miniature optical sensors, besides their well-known merits including electrically
passive operation, immunity to electromagnetic interference, self-referencing, and
high sensitivity [4], demonstrated simultaneous measurement of strain and tem-
perature using FBG sensors. The corrosion phenomena of structural steel involve
mainly two elements: the material and its environment. In particular, corrosion is
defined as the deterioration of a material, usually a metal, which results from a
reaction with its environment, causing the degradation of both. The local pH and
temperature influence corroding chemical reactions and affect the rate of corrosion.
In the present study, comparative measures of laboratory simulated corroded cou-
pons bonded with FBG and conventional electrical resistance sensors are carried
out. Corrosion is induced by galvanostatic method on structural steel coupons. The
variation in temperature and pH of electrolyte is monitored with WTW probes.
Initially, the response of the FBG sensor is validated with electrical resistance strain
gauge under ambient condition, for two sets of tension coupon test. Further to the
validation, four sets of experiments are carried out in a corrosive condition by two
different methods, case 1: unstressed and case 2: stressed and the results are
discussed.
Assessment of Strain in a Corrosive Environment … 439
2 Fiber Bragg Grating (FBG) Sensors
The FBGs used are fabricated by creating a periodic modulation of the refractive
index along a photosensitive silica fiber, reflectivity of more than 95%, and a
bandwidth of 0.2 nm. Fiber Bragg gratings are used as strain or temperature, pH
sensors in which variations of those parameters are transformed to a shift of the
Bragg wavelength reflected back as a Gaussian profile signal. In general, fiber
Bragg gratings are periodic structures that are imprinted directly into the core of
glass optical fiber by powerful ultraviolet radiation. Such structure consists of a
periodically varying refractive index over typically several millimeters of the fiber
core. The specific characteristic of FBG for sensing applications is that their peri-
odicity causes them to act as wavelength-sensitive reflectors. During imprinting
process, the intensity of the ultraviolet illumination is made to occur in a periodic
fashion along the fiber core. At a sufficiently high power level, local defects are
created within the core, which then give rise to a periodic change in the local
refractive index. This change in refractive index (RI) created is permanent and
sensitive to a number of physical parameters, such as pressure, temperature, strain,
and vibration. Thus, by monitoring the resultant changes in reflected wavelength,
FBG can be used for sensing applications to measure various physical quantities.
Fiber Bragg grating sensor the response arises from two sources, namely the
induced change in pitch length (K) of the grating and the perturbation of the
effective core refractive index (n). The wavelength of the reflected spectrum band is
defined by the Bragg condition [5], given by the expression
kB ¼ 2nK ð1Þ
where n is effective refractive index of the core and K is the grating pitch.
K fiber Bragg grating consists of a longitudinal, periodic variation in the
refractive index of the core of an optical fiber as shown in Fig. 1.
Fig. 1 Schematic of the

in-fiber grating structure of a
fiber bragg sensor
3 Fiber Bragg Grating Sensing Technique
Fiber Bragg grating (FBG) is a passive optical component obtained by recording a

local, longitudinal period modulation of the refractive index in the optical fiber
core, as depicted in Fig. 2a. Due to the local change of the refractive index, any
light propagating along the fiber core suffers partial reflection at each of the grating
layers. As a consequence of the periodicity of the index modulation, constructive
interference of the reflected light takes place for those wave vectors that satisfy the
Bragg condition given in Eq. 1. This implies that part of the incident spectrum is
not transmitted, but reflected, by grating as shown in Fig. 2a. The technical details
of the FBG used for the present study are given in Table 1. Since the wavelength
reflected by the grating is a function of n and K, variations of the temperature and
variations of the strain due to loss of metal (i.e., subsequent change in geometry of
corroded surface) attributed by change in pH, i.e., or of the strain to which the
grating is submitted, give rise to appreciable wavelength change, which can be
considered as indicated in Fig. 2b. Where a, b, c constant coefficients and DT, DM,
De are incremental increase in temperature, metal loss factor, and strain, respec-
tively. The principle of operation of FBG sensing consists of monitoring the Bragg
wavelength and correlating wavelength change, B, to the variations in the measured
value.
The length of the FBG, which is an integral part of the fiber, is normally few
millimeters. When light from a broadband source is launched in the fiber, the FBG
reflects wavelength intensity that satisfies Eq. 1 while the rest of the wavelengths
Fig. 2 Scheme of fiber Bragg grating operation

Table 1 Technical details of FBG

Center CW Reflectivity 3 dB band Fiber type Grating
wavelength tolerance width length
1540 nm ±1 nm 0.70% 0.25 ± 0.05 nm SMF28-C 10 mm
Polymide
get transmitted. The reflected wavelength kB is called Bragg wavelength and is

determined by the period of the grating ðKÞ. The Bragg wavelength is sensitive to
changes in either the fiber-core refractive index or the period of the grating. This
makes it an excellent sensor for measurement of mechanical or thermal perturba-
tions. Unlike intensity-modulated sensors which need a very stable light source and
interferometric sensors which need a highly coherent laser, fiber Bragg
grating-based sensors generally require a broadband source and a high-resolution
wavelength shift detection system. Fiber Bragg gratings have been reported for the
measurements of strain temperature, dynamic magnetic field, etc. The sensing
principle is that the Bragg wavelength will vary with the change of these parameters
experienced by the fiber. In the case of load-induced strain, both physical elon-
gation of the optical fiber (and the corresponding change in the grating pitch) and
the change in the refractive index of the fiber due to photo-elastic effects take place.
As a result, the shift in Bragg wavelength occurs. The shift in Bragg wavelength
with strain can be expressed as
DkB ¼ ð1 pe ÞkB e ð2Þ
where e is the applied strain and pe is an effective photo-elastic coefficient term, for
the present study it has been taken as 0.2.
The sensing of strain and temperature due to physical elongation of the sensor
and change in fiber index due to photo-elastic effect is expressed [1], as
0 2 3 1

2 dn
@ n 4 dT 5 A
DkB ¼ 2nK 1 ½P12 mðP11 þ P12 Þ e þ aT þ DT ð3Þ
2 n
where e is the applied strain, Pi,j coefficients are the Pockel’s coefficients of the
stress-optic tensor, m is Poisson’s ratio, and aT is the coefficient of thermal
expansion
of the fiber material and is the temperature change. The factor
n2
2 ½P12 mðP11 þ P12 Þ has a numerical value of 0.2. However, there is not
much studies are carried out to sense the change in fiber index due to combined
effect of change in strain, temperature, and pH in a corrosive environment.
4 Electrical Resistance Strain Gauge
The electrical resistance strain gauge is a physically simple device, which can be
easily applied in a straightforward manner for elementary measurements of surface
strains. In the present study, electrical foil strain gauge of gauge length 2 mm is
fixed to the coupon using cyanoacrylate adhesive. The foil gauge, Fig. 3, is
essentially a small printed circuit on photographic plate. The details of the foil
gauge used for the present study are shown in Table 2. The gauge factor is 2.0. In
order to use such a gauge for detecting a strain of 1 µe (which corresponds to that
produced by a 206.84 kN/m2 stress in steel for a simple uniaxial loading), a change
of resistance DR of 240 µΩ needs to be measured with an indicator.
The relative change in resistance is given by
L
R¼q ð4Þ
A
The strain sensitivity can be defined as a basic bulk property of the

strain-sensitive alloy used in a strain gauge. When this metal is formed into a grid
and provided with attachment points for lead wires, the gauge will exhibit a dif-
ferent relationship between resistance change and applied strain. The term ‘gauge
factor’ (GF) [6] is used to quantify this relationship and is defined as
Fig. 3 Typical foil strain gauge
Table 2 Details of foil strain gauge

Gauge length Gauge width Backing length Backing width Resistance
(mm) (mm) (mm) (mm) (Ω)
2 1.5 6.5 3 120
Note The present gauge backing is made of epoxy resin with thickness 0.03 mm which exhibits
excellent electrical insulation performance. The backing has self-temperature compensation for
steel ranging from 0 to 80 °C
DR=Ro DR=Ro
GF ¼ ¼ ð5Þ
DL=Lo e
where DR is resistance change in the gauge in ohms, Ro is original or unstrained

gauge resistance, and e is unit engineering strain in the specimen surface under the
gauge grid DL/Lo.
However, the use of electric strain gauges in harsh environment is not advisable
due to their low sensitivity to respond against variation in temperature and pH.
5 Experimental Set-up
Experiments are carried out with fiber Bragg grating-based sensors and conven-
tional electrical resistance strain gauges to monitor strains induced during simulated
corrosion of unstressed and stressed coupons under galvanic method. The tension
coupons are cut from tubes of diameter 76.1 mm and are corroded by using
accelerated corrosion process. The corrosion process is carried out by keeping the
test specimen as anode and steel plate cut from the same parent metal as cathode.
They are placed in the electrolyte of 3.5% NaCl solution. The specimen (coupons
are 250 mm long with a mid-gauge length of 60 mm and width as 12.5 mm, as per
ASTME8 M [7] specifications, Fig. 4) to be corroded is treated as anode, and a
structural steel specimen of slightly large area of exposure than anode (80 mm
75 mm) is considered as cathode. Anode is connected to the +ve terminal.
A specimen cut from the same parent metal was used as cathode and connected to
the −ve terminal. Calomel electrode is used to measure the initial and final potential
under zero current condition. The width of the gripping edge is 75 mm, and width
of the gauge length area is 12.5 mm. The galvanic corrosion of the gauge portion is
completely achieved by passing current of 3.4 A for duration of 300 min. The
gauge length of the coupon is bonded with FBG sensor of center wavelength
1540 nm and grating length 10 mm and conventional electrical resistance strain
gauge of 2 mm gauge length on the opposite face of corrosion. Initially, the
response of FBG sensors is validated with electrical strain gauge (ESG) under
ambient condition through a series of tensile test, namely FBG-TT-01, FBG-TT-02,
ESG-TT-01, and ESG-TT-02 as shown in Fig. 5. The corrosion is carried out by
Fig. 4 Structural steel coupons glued with FBG and Electrical resistance strain gauge
FBG-TT-01 ESG-TT-01
FBG-TT-02 ESG-TT-02
500
400
Stress (N/mm2) 300
200
100
0
0 10000 20000 30000 40000
Strain
Fig. 5 Comparison of strain measurements of FBG and ESG
Fig. 6 Experimental set-up for unstressed condition
two methods: case 1 is unstressed condition, Fig. 6, and case 2 is stressed condition,
Fig. 7. Four sets of coupons, namely FBG-ST-01, FBG-ST-02, FBG-ST-03, and
FBG-ST-04, are the specimens used for unstressed condition, and FBG-SST-05,
FBG-SST-06, FBG-SST-07, and FBG-SST-08 are the respective specimens used
for stressed condition. Under stressed condition, the tension coupon specimen is
gripped in a 25-ton UTM and applied 10% of the ultimate load to generate stress
under the test condition.
An optical sensor for steel corrosion monitoring based on etched FBG sputtered
with iron film showed intrinsically sensing capability as their resonant wavelengths
vary with the perturbation induced change of the grating period and the fiber
effective index [8]. In the present study, the strains measured by FBG sensor glued
to steel coupons in a progressive corrosive environment are considered as the sum
total of low strains generated due to metal loss followed by irregular changes on the
exposed surface and due to variation in temperature. While the performance of
ordinary strain gauge used is guaranteed till 23 °C with 50% relative humidity, the
apparent strain is guaranteed for a range of 0–80 °C. And variation in pH is having
Fig. 7 Experimental set-up for stressed condition
influence on rate of corrosion. During the accelerated corrosion process, it is

observed that temperature as well as pH of the electrolyte varies. The temperature
and pH characteristics of the electrolyte during corrosion process are also measured
with WTW probes.
The observed strain versus time behavior of FBG sensors and electrical resistance
strain gauge under induced corrosion process under stressed and unstressed con-
dition are shown in Figs. 8, 9 and 10. From the experimental investigations, it is
observed that the strain sensed by the electrical resistance strain gauge varies from
−28 to 500 µe, and strain sensed by FBG sensors varies from 0 to 800 µe for
duration of 5 h. The observed strain measurements confirm that there is no physical
damage due to applied load to the specimen. The detected strains in all cases by
sensors are due to metal loss from the corroding surface along with variation in pH
and temperature. The entire gauge region of the specimen is expected to be com-
pletely corroded within 5 h (i.e., 300 min). Initially, under unstressed condition,
both FBG and electrical resistance strain gauge were initialized, whereas under
stressed condition, both FBG and electrical resistance strain gauge show initially a
strain value of 283 le due to applied load and initialized. As the corrosion process
progresses, the electrical resistance strain gauge shows apparent strain, i.e., com-
pensation for temperature component during unstressed condition (case 1), whereas
under stressed condition, it is capable to sense strain induced due to combined effect
of abrupt reduction in cross-sectional area and temperature. Compared to electrical
resistance strain gauges, FBG sensors show its versatility to sense strains induced
due to abrupt changes in cross-sectional area and chemical changes due to variation
Fig. 8 Strain versus time

from FBG sensor under
unstressed condition
Fig. 9 Strain versus time

from FBG sensor under
stressed condition
in pH and temperature in a severe corrosive environment. The strain sensed by FBG

sensor consists of three components, the induced strain due to stress (i.e., strain
induced due to abrupt change in cross-sectional area under stressed condition), the
strain induced due to change in temperature, and the low strain induced due to metal
loss (in a varying pH and temperature environment). Under unstressed condition,
the strain sensed by FBG is due to change in temperature and metal loss due to
corrosion (in a varying pH and temperature environment). The sensed strain by
FBG sensor for stressed condition (case 2), Fig. 12, is found to be higher than
unstressed condition (case 1), Fig. 11, and this may be due to higher rate of cor-
rosion in a stressed condition. After 130 min (2.17 h) of test, FBG sensors under
case 1 show dropping behavior under unstressed condition, and this indicates the
depletion of sensor from surface of the specimen. Almost 50% of the initial weight
of the metal from the surface of anode is expected to be removed due to electro-
chemical process within 2.5 h. Then the formation of irregularities on the surface of
anode observed to be high, and this may be correctly sensed by the FBG. The
typical variation in temperature and pH observed during the test is shown in
Figs. 13 and 14, respectively.
Fig. 10 Strain versus time electrical strain gauge under stressed and unstressed condition
Fig. 11 Comparison of FBG-ST-01 FBG-ST-02

unstressed FBG strain FBG-ST-03 FBG-ST-04
measurements 500
400
Strain (με)
300
200
100
0
0 100 200 300
Time (Minutes)
Fig. 12 Comparison of FBG -SST-05 FBG-SST-06

stressed FBG strain FBG-SST-07 FBG-SST-08
measurements 1000
800
Strain (με)
600
400
200
0
0 100 200 300
Time (Minutes)
Fig. 13 Typical temperature

versus time
Fig. 14 Typical pH versus

time
7 Conclusion
In the present study, experiments are carried out by inducing galvanic corrosion on
structural steel coupons. The coupons are corroded under two conditions, viz.,
unstressed (case 1) and stressed (case 2). The strain measurements in coupons under
stressed and unstressed conditions are monitored with FBG sensors and electrical
resistance strain gauges. From the studies, it is concluded that FBG is suitable to
measure strain due to corrosion, since it is capable to measure strains due to
temperature (0–50 °C) and other interference (variation in pH from 11.4 to 14). The
electrical resistance strain gauge is incapable to sense the induced strain on a metal
surface exposed to corrosive environment with varying temperature and pH. From
the FBG, it is concluded that the strain induced during unstressed condition is
combined effect of temperature and variation in corrosion rate due to change in pH.
Resultant strain in stressed condition is combined effect of strain due to stress,
temperature strain, and variation in corrosion rate due to change in pH. Thus,
FBG-based sensors are reliable than other conventional electrical resistance-based
strain gauges. FBGs can be used to assess the onset of the corrosion before major
impairment in the surface of the metal specimen.
Acknowledgements This paper is published with the permission of the Director, CSIR-Structural
Engineering Research Centre, Chennai, India.
References
1. A.D. Kersey, M.A. Davis, H.J. Patrick, M. LeBlanc, K.P. Koo, C.G. Askins, M.A. Putnam, E.
J. Friebele, Fiber grating sensors. J. Lightwave Technol. 15(8), 1442–1463 (1997)
2. L. Suzhen, Z. Wu, L. Zhou, Health monitoring of flexural steel structures based on distributed
fiber optic sensors. Struct. infrastruct. Eng. 6(3), 303–315 (2010)
3. A.P. Zhang, S. Gao, G. Yan, Y. Bai, Advances in optical fiber Bragg grating sensor
technologies. Photonic Sens. 2(1), 1–13 (2012)
4. C. Fernandez-Valdivielso, I.R. Matias, F.J. Arregui, Simultaneous measurement of strain and
temperature using a fiber Bragg grating and a thermochromic material. Sens. Actuators, 107–
116 (2002)
5. B.D. Gupta, Fiber Optic Sensors Principles and Applications (New India Publishing Agency,
New Delhi, 2006)
6. R.L. Hannah, S.E. Reed, Strain Gage User’s Handbook (Elsevier Applied Science, New York,
1997)
7. ASTM E8/E8M-11, Standard Test Methods for Tension Testing of Metallic Materials. ASTM
International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA, 19428–2959
US
8. Zhang Ning, Wei Chen, Xing Zheng, Hu Wenbin, Min Gao, Optical sensor for steel corrosion
monitoring based on etched fiber Bragg grating sputtered with iron film. IEEE Sens. J. 15(6),
3551–3556 (2015)
A Test Stand for Studying Subcritical
and Critical States of Full-Size Reinforced
Concrete Structures
Igor Shardakov, Irina Glot, Aleksey Shestakov, R. Tsvetkov

and V. Yepin
Abstract Reinforced concrete is one of the basic building materials. Therefore, it

is essential to study the behavior of reinforced concrete structures in subcritical and
critical states. In this paper, we discuss the performance of a test stand designed to
study deformation processes in the elements of full-size models of reinforced
concrete structures subjected to quasi-static and dynamic loadings. The design of
the stand and its main technical characteristics have been developed based on the
results of numerical simulations of the evolution of deformation processes in the
tested structure in going from elastic to inelastic state, which finally leads to its
destruction. These simulations allowed us to design the foundation and load frame
of the stand, to define the parameters of loading devices able to provide a wide
range of deformations, to choose apparatus for recording the successive stages of
damage accumulation and macroscopic fracture occurrence.
Keywords Experimental stand Reinforced concrete Fracture mechanics

Full-scale testing
1 Introduction
The operation safety of reinforced concrete (RC) buildings and constructions is of

great practical importance. For such constructions, the elaboration of models, which
can properly describe the deformation processes in subcritical and critical states, is
still an open question. This is due to the fact that the critical states in structures
develop according to a rather complex spatial–temporal scenario. Normally, a
deformed structure consists of a great number of inhomogeneous elements, inter-
relations between these elements are rather complex, and the structural character-
istics of reinforced concrete can vary widely. The deformation processes taking
I. Shardakov (&) I. Glot A. Shestakov R. Tsvetkov V. Yepin

Institute of Continuous Media Mechanics UB RAS, Korolev Street, 1,
614013 Perm, Russia
e-mail: shardakov@icmm.ru

https://doi.org/10.1007/978-981-10-7197-3_37
452 I. Shardakov et al.
place in separate structural elements, for example in beams, are extensively

investigated in the literature [1, 2]. However, the available information on testing
separate structural elements is not sufficient to create a model, which can adequately
describe the behavior of the entire structure. To do this would require experiments
based on the full-scale model, which is a reinforced concrete structure consisting of
structural units (several storys and several typical units disposed in a horizontal
plane). For such experiments, it is essential to design a test stand of a scale suitable
for generating, visualizing, and identifying subcritical and critical strain states.
Worldwide, there are only few laboratories, where full-scale structural testing
experiments can be conducted. For example, the European Laboratory for Structural
Assessment (ELSA) in Italy carries out research in the area of improving safety of
constructions [3]. Studies of the seismic characteristics of multi-story concentrically
braced frames described in [4–6] are of immediate interest to experimentalists. This
interest has arisen in connection with the problems of using structures in the regions
of high seismic activity. The description of static full-scale structure tests is rarely
found in the literature. A series of loading tests performed on a 1:2 scale replica of a
Cypriot single-story structure are described in [7].
Creation of an experimental stand for testing full-scale constructions will make it
possible to generate subcritical and critical strain processes in structures and their
components under various quasi-static loading scenarios. Registration of these
processes will enable adjustment of the existing mathematical models describing
the deformation behavior of precast and cast-in-situ reinforced concrete
constructions.
In this paper, we present the results of mathematical simulation of the defor-
mation behavior of a RC structure and its components affected by the external force
and kinematic factors. The RC structure is a physical model, which will be tested on
a stand, which is currently under development. The proposed mathematical model
considers the process of deformation from the initial elastic state to the state of
complete failure. Analysis of the results of numerical simulation allows us to
develop one of the possible variants of the structural scheme of the test stand and to
estimate its parameters.
2 The Model of the Concrete Structure
In our investigation, we have studied the fragment of a 1:2 scale four-story building
(Fig. 1e). The model object comprises 24 typical structural cells combined into four
levels. The total height of the structure is 6 m, and base proportions are 6 9 m.
A multi-element factor is assumed for all three directions (height, length, and
width). By considering this factor, we can determine a true character of the spatial
distribution of deformation, especially at the stage of subcritical and critical states
of the RC structure. Note that the chosen scale (1:2) allows us to use standard
A Test Stand for Studying Subcritical and Critical States … 453
Fig. 1 Scheme of the RC structure (e) and its main elements: a “one column”; b “2 columns–
beam”; c “9 columns–12 beams”; d “9 columns–12 beams–4 floor slabs”
concrete with a known mass fraction ratio to manufacture model structural units.
Using of smaller-scale models made of this material can not guarantee that the
deformation process occurring in them be identical to the processes in the original
object.
Figure 1 presents the fragments of the tested model structure from the simplest
to the most complex from the viewpoint of their design. For each fragment and the
structure as the whole, we have simulated the process of deformation from the
initial elastic state to the state of complete failure and loss of bearing capacity.
Quasi-static loading was realized kinematically (due to the prescribed displacement
of separate fragments of the structure).
Mathematical modeling of the deformation process, including the cracking stage,
was performed within the framework of solid mechanics. Cracking in concrete was
described as a sharp reduction in the material stiffness in a localized region. Based
on the governing tensor relations, we determined the direction of propagation of the
localized, reduced-stiffness region, the location of this region, and the mechanisms
responsible for reducing material stiffness. The main theoretical principles under-
lying this approach were described in [8]. A jump-wise reduction of material
stiffness represents a peculiar feature of cracking in concrete. The deformation
process in the metal reinforcement was modeled in the context of the elastoplastic
theory [9]. The numerical implementation of the model was carried out using the
finite-element method in the framework of the ANSYS package.
3 The Results of Numerical Analysis
3.1 Uniaxial Tension—Compression of One Column
We have modeled the deformation process in a RC column by the relative dis-

placement (e) of the column ends along its axis. Two diagrams (F–e) were obtained,
Fig. 2 Calculation results: a (F–e) compression diagram for the column; b tension diagram
respectively, for compression (Fig. 2a) and tension (Fig. 2b). Here, F is the
resultant response of the column to the relative displacement of its ends. On
compression of the sample (Fig. 2a) from its initial state, point 3 (e ¼ 0; F ¼ 0), the
force increases linearly up to point 2 (e ¼ 900 106 ; F ¼ 1100 kN). After the
force passes point 2, the concrete destroys and the total force is exerted upon the
reinforcement, which thus loses its stability and is unable to withstand extra loads—
point 1 (e ¼ 920 106 ; F ¼ 31 kN).
The process of tension (Fig. 2b) proceeds linearly to point 4 (e ¼ 64 106 ;
F ¼ 78 kN); then, the concrete destroys, and the entire force is exerted upon the
reinforcement—point 5 (e ¼ 66 106 ; F ¼ 8 kN). With increasing deformation,
the force increases until the reinforcement material enters into the plastic strain zone.
After the plastic deformation, the reinforcement experiences rupture.
3.2 Deformation of the System “2 Columns–Beam”
The beams and columns are connected in such a way as to form the frame of the
building. Due to uneven settlement of columns, the beam experiences a bending
moment. This gives rise to tension zones, where the fracture in the form of cracks is
realized. In these zones, the tension force is exerted upon the reinforcement only.
A further increase in uneven settling leads either to reinforcement rupture, or to
failure of the concrete in its compressed region. This strongly reduces the stiffness
of the material and even may cause its full failure. The simplest system representing
this mechanism consists of 2 columns and 1 beam (Fig. 1b). The loading test
involves vertical displacement (U) of the foundation of the first column, while the
second column remains fixed.
Fig. 3 Calculation results obtained for the system “2 column–beam” under differential settlement
conditions: a diagram showing how the force exerted on the column changes at given
displacement (numbers—key deformation points); b fracture due to tension (points 3–5), and
fracture due to compression (points 6–9)
Based on the results of numerical simulation, we have plotted the curve (F–
U) (Fig. 3a). Here, the force F is the resultant response of the end of the right-hand
column directed along its axis. The numbers at this curve show key deformation
points. Rupture areas corresponding to these deformation stages are shown in black
in Fig. 3b. At points 3–5 on the curve (F–U), fracture is caused by tension and at
points 6–9 by compression. Before applying displacements, the structure is under its
proper weight—point 1 (U ¼ 0; F ¼ 1:9 kN). Linear increase in the displacements
up to point 2 (U ¼ 1:2 mm; F ¼ 3:7 kN) proceeds without ruptures. However, the
subsequent points 3 (U ¼ 1:6 mm; F ¼ 4:0 kN) and 4 (U ¼ 2:0 mm; F ¼ 4:1 kN)
are associated with the generation and propagation of cracks in the extended zones,
which increase essentially at point 5 (U ¼ 3:6 mm; F ¼ 4:5 kN). At point 6
(U ¼ 38:9 mm; F ¼ 8:2 kN), the compressed region becomes the site of nucleation
of the fracture zone (right-hand low part of the beam), which on approaching point 7
(U ¼ 38:9 mm; F ¼ 8:2 kN) becomes more pronounced. The concrete in the second
compressed zone located in the left-hand upper part of the beam destroys at point 8
(U ¼ 44:4 mm; F ¼ 8:6 kN). After that, the force exerted on the column decreases
until point 9 (U ¼ 49:7 mm; F ¼ 7:8 kN) is reached. Further increase in the dis-
placement leads to complete loss of load-bearing capacity.
3.3 Deformation of the System “9 Columns–12 Beams”
The next fragment of higher complexity is the system consisting of 9 columns and
12 beams (Fig. 1c). It indicates a combined action of beams and columns of the first
story at vertical displacement U of the central column foundation (Fig. 4a). As in
Fig. 4 Calculation results obtained for the system “9 columns–12 beams”: a deformed shape of
the system at relative vertical displacement of the central column; b diagram force (F) versus
displacement (U)
the previous calculation, a plot of force F generated in the column versus the given
displacement of its foundation is constructed (Fig. 4b).
The fracture of beams is similar to that observed in the system “2 columns–beam.”
The key deformation points are shown in Fig. 4b: 1 (U ¼ 0; F ¼ 6:0 kN)—system
under its proper weight; 2 (U ¼ 0:8 mm; F ¼ 10:7 kN)—fracture initiation in the
extended zones of beams; 3 (U ¼ 45:0 mm; F ¼ 44:6 kN)—maximum force and
initiation of fracture in the compressed zones of beams; 4 (U ¼ 62:9 mm;
F ¼ 39:2 kN)—loss of bearing capacity.
3.4 Deformation of the System “9 Columns–12 Beams–4

Floor Slabs”
A contribution of floor slabs to the operation of the examined structure is considered

by studying the behavior of the system “9 columns–12 beams–4 floor slabs”
(Fig. 1d). This system differs from the previous one by the presence of four floor
slabs. The deformed shape of the system due to relative vertical displacement of the
central column foundation is presented in Fig. 5.
Figure 6a presents the plot of force F applied to the central column versus the
prescribed displacement U. The key deformation points marked on the graph are as
follows: 1 (U ¼ 0; F ¼ 34:1 kN)—system is under its proper weight; 2
(U ¼ 0:5 mm; F ¼ 69:5 kN)—splitting of joints between the floor slabs accom-
panied by simultaneous fracture of the concrete in the extended zone; 3
(U ¼ 3:1 mm; F ¼ 126:8 kN)—maximum force achievement, initiation of fracture
in beams; 4 (U ¼ 4:7 mm; F ¼ 110:5 kN)—complete fracture of floor slabs; 5
(U ¼ 13:3 mm; F ¼ 107:8 kN)—fracture of the concrete in the compressed zone
and loss of bearing capacity. In Fig. 8b, the areas of destroyed concrete in the slabs
formed at the loading stages corresponding to points 2, 3, and 4 are shown in black.
In contrast to the same system but without floor slabs, the maximum force exerted
Fig. 5 Deformed shape of the system “9 columns–12 beams–4 floor slabs” at relative vertical
displacement of the central column
Fig. 6 Calculation results for the system “9 columns–12 beams–4 floor slabs”: a diagram force
versus displacement (numbers—key deformation points); b fracture areas (black color—top view)
in compliance with points on the plot (F–U)
on the column increases by a factor of 2.8, and the displacement of columns at the
moment when the force reaches a maximum decreases by 14 times. Hence, it can be
concluded that the floor slabs have a strong impact on deformation and fracture of
the structure.
3.5 Deformation of the Four-Level Structure
The structure is deformed due to the vertical displacement of the foundation of one
internal column with respect to other fixed ones (Fig. 7a). Analysis of the results of
numerical simulation yields the plot (F–U), Fig. 7b. Here, the force F is the axial force in
the foundation of the internal column, for which the displacement U is specified. The
key deformation points marked on the curve are as follows: 1 (U ¼ 0; F ¼ 147 kN)—
system at its proper weight; 2 (U ¼ 0:5 mm; F ¼ 260 kN)—appearance of first
Fig. 7 Results of modeling of the deformation process in the four-story structure: a deformed
shape of the structure, b diagram force versus displacement
cracks in the extended zones of beams; 3 (U ¼ 5:2 mm; F ¼ 621 kN)—generation of

cracks in floor slabs corresponds to the maximum force in the column; 4
(U ¼ 11:6 mm; F ¼ 590 kN)—fracture of the compressed zones of beams.
4 The Structure Diagram of the Test Stand and Its

Parameters
Numerical simulations of deformation of the model reinforced concrete structure

subjected to subsequent loading up to its destruction allowed us to develop the
structure of the test stand for physical experiments with large-scale structures.
Figure 8 presents the structure diagram of the test stand with a model construction
inside it. The dimensions of the model construction are 6 6 9 m. Overall
dimensions for the force loop of the test stand and the reinforced concrete foun-
dation are given in figure.
Based on the results of numerical experiments, we have evaluated parameters for
the loading units of the stand. It has been established that the range of force
generated by the loading unit in order to reliably reach the critical state of the model
should be 0–1000 kN. Table 1 gives the values of estimated critical loads (forces
and displacements) applied to the components of the model construction at different
loading stages. To provide a controlled transition of the model from elastic to
subcritical and critical states, it is necessary to ensure accurate dosing of loading.
Simulations showed that the accuracy of prescribing the displacement of separate
elements of the model construction should be as high as 0.1 mm.
The analysis of the sites of fracture initiation found from numerical experiments
of the construction allowed us to determine the appropriate locations of strain
sensors and to fit the parameters of the registration apparatus (sensor types, mea-
surement accuracy). This information will be used to perform physical experiments
on studying the evolution of critical states in the model reinforced concrete
construction.
Fig. 8 Test stand with scale model of the structure: 1—reinforced concrete foundation, 2—force
loop, 3—scale model of building
Table 1 Forces and displacements at various deformation stages in the assembled structure and
its fragments
Structure Limiting values of Limiting values of Max. force at all
element elastic deformation elastic deformation deformation process
F (kN) U (mm) F (kN)
“2 column– 3.7 1.2 8.6
beam”
“9 column–12 10.7 0.8 44.6
beams”
“9 column–12 69.5 0.5 126.8
beams–4 floor
slabs”
Four-story 260 0.5 621.0
structure
5 Conclusion
Based on the mathematical simulation of quasi-static deformation processes taking

place in the model RC structure and its fragments, we determined the following:
• linear dimensions of the scale model;
• specific forces, deformations, and displacements at which one can realize sub-
critical and critical deformation processes in the assembled RC structure and its
components;
• scenarios for spatial distribution of fracture in the assembled RC structure and its
components;
• required accuracy of the prescribed kinematic loadings.
The obtained preliminary estimates served as a basis for the development of the
structural stand for investigating deformation processes in reinforced structures.
A set of loading units able to provide a controlled transition of the model con-
struction to the critical state were proposed. The strain registration system was
designed.
Acknowledgements The work is supported by Russian Science Foundation (project No. 14–29–
00172).
References
1. C. Casandjian, N. Challamel, C. Lanos, J. Hellesland, Reinforced Concrete Beams, Columns

and Frames: Mechanics and Design (ISTE and Wiley, 2013)
2. Ł. Skarżyński, J. Tejchman, Experimental investigations of fracture process in concrete by
means of X-ray micro-computed tomography. Strain 52, 26–45 (2016)
3. P. Negro, D.A. Bournas, F.J. Molina, Pseudodynamic test on a full-scale 3-storey precast
concrete building: global response. Eng. Str. 57, 594–608 (2013)
4. D.P. McCrum, B.M. Broderick, An experimental and numerical investigation into the seismic
performance of a multi-storey concentrically braced plan irregular structure. Bull. Earthq. Eng.
11, 2363–2385 (2013)
5. K.C. Tsai, P.C. Hsiao et al., Pseudo-dynamic tests of full-scale CFT/BRB frame—part 1:
specimen design, experiment and analysis. Earthq. Eng. Struct. D 37, 1081–1098 (2008)
6. K.C. Tsai, P.C. Hsiao et al., Pseudo-dynamic tests of full-scale CFT/BRB frame—part 2:
seismic performance of buckling-restrained braces and connections. Earthq. Eng. Struct. D 37,
1099–1115 (2008)
7. R. Illampas, D.C. Charmpis, I. Ioannou, Finite element simulation of the structural response of
adobe masonry buildings subjected to lateral loading, in Proceedings SAHC (Mexico, 2014),
12 p
8. K.J. William, E.P. Warnke, Constitutive model for the triaxial behavior of concrete. Proc. Int.
Assoc. Bridge Struct. Eng. 19, 1–30 (1974)
9. L.M. Kachanov, Foundations of Theory of Plasticity (North-Holland, Amsterdam, 1971)
Mode I SIF Determination of Orthotropic
Laminates with Double-Ended Cracks
Using a Single-Strain Gage
Debaleena Chakraborty, D. Chakraborty and K. S. R. K. Murthy
Abstract The experimental determination of mode I SIF (KI ) in orthotropic

composites is significant, and strain gage-based approaches for SIF determination
are quite economic and easy to use. Strain gage techniques that have been devel-
oped for measurement of KI of orthotropic laminates are based on the solutions of
single-ended crack theories. The present investigation aims at demonstrating the
application of a single-strain gage technique (based on single-ended crack theory)
recently proposed by the authors to double-ended cracked configurations for
accurate measurement of KI. Finite element-based numerical analyses have been
conducted on carbon-epoxy orthotropic laminates having center-cracked configu-
rations. The existence and determination of optimal gage locations for accurate
measurement of KI have also been shown through numerical studies. Numerical
results show that accurate values of KI for double-ended cracked orthotropic lam-
inates can be obtained with the help of a single-strain gage, when placed within the
optimal zone.

Keywords Orthotropic laminate Strain gage Mode I SIF Double ended cracks
Nomenclature
a11 ; a12 Material properties depending on Young’s modulus, Poisson’s ratio, and
shear modulus
a22 ; a21 Material properties depending on Young’s modulus, Poisson’s ratio, and
shear modulus
a; b Parameters dependent on material properties
Ei Young’s modulus along ith direction ði ¼ L; T Þ
GLT Shear modulus in L–T plane
L; T Principle material directions (longitudinal and transverse)
D. Chakraborty (&) D. Chakraborty K. S. R. K. Murthy

Department of Mechanical Engineering, IIT Guwhati, Guwahati 781039, India
e-mail: debaleena@iitg.ernet.in

https://doi.org/10.1007/978-981-10-7197-3_38
462 D. Chakraborty et al.
1 Introduction
A substantial amount of work has been reported in the field of fracture mechanics of
composite materials in the last few decades. Irwin [1] and Wu et al. [2] reported on
the feasibility of applying the concepts of linear elastic fracture mechanics (LEFM)
to homogenous, isotropic materials to composites following appropriation to
account for the directional properties of these materials. This leads to the exami-
nation of crack instability of these materials using stress intensity factor (SIF) as a
criteria to measure the strength of a cracked body.
The experimental determination of SIF using strain gages is quite popular due to
its cost-effectiveness and feasibility in handling. The Dally and Sanford [3]
single-strain gage technique is the first of its kind for homogeneous isotropic
materials which enables the determination of mode I SIF ðKI Þ using a single-strain
gage and necessitates a single-strain gage to be placed along a line (decided by
angle, h and called gage line) and oriented along / as shown in Fig. 1.
A significant emphasis has been laid on the experimental determination of SIFs of
cracked composite panels since they are of great utility not only in complex con-
figurations but also in validating the numerical and experimental results [4–9].
Cerniglia et al. [6] formulated an over deterministic method for estimation of KI in
orthotropic materials by employing large number of strain gages. Shukla and
coworkers [7] were the first to propose a single-strain gage technique for the
determination of KI for orthotropic laminates. However, instead of a three-parameter
series like Dally and Sanford, they used a two-parameter series representation of the
strain field around the crack tip thereby limiting the allowable radial distance for
pasting the strain gage. Khanna and Shukla [8] employed the technique proposed by
Shukla et al. [7] in fracture analysis of a dynamic crack in orthotropic composites.
A straightforward extension of the Dally and Sanford technique to orthotropic
materials has been developed by Chakraborty et al. [9], where the benefits of using a
three-parameter series can be availed and the gage could be placed at such a distance
from the crack tip that not only the singular solution remains valid in that zone but
the strain readings are also unaffected by strain gradients and 3-D effects.
Fig. 1 Crack tip local

coordinate system and
location of strain gage
Mode I SIF Determination of Orthotropic Laminates … 463
It may be pointed out that till date single-ended crack theories have been
employed to develop experimental methods for measurement of mode I SIF of
orthotropic materials using strain gages [6–9] and tested only on single-ended
cracked configurations. However, these single-ended crack theories, can be suc-
cessfully applied to double-ended crack configurations when the distance between
the neighboring crack tips is such that their stress fields don’t interact with each
other [10].
Only recently Chakraborty et al. [11] extended the strain gage technique by
Chakraborty et al. [9] (which is based on single-ended crack theory) to the
double-ended cracks for the accurate estimation of KI in laminated orthotropic
composites having internal cracks. Using the same technique, an attempt has been
made in the present work to determine KI of a ½902 =010S carbon-epoxy laminate
having a center-cracked configuration. Numerical simulations have been presented
using the laminate properties of ½902 =010S carbon-epoxy laminates.
2 Theoretical Formulations
In this section, the extension of the Dally and Sanford technique [3] to orthotropic
materials and a theoretical basis for ascertaining the limit of the maximum radial
distance for possible positioning of a strain gage has been laid out. Using
three-parameter strain series, the normal strain component eaa at an angle / with the
crack axis at a point P (Fig. 1) located by r and h within the singularity dominated
zone can be obtained as
A0 pffiffi
eaa ¼ pffiffi f1 ðr; h; /Þ þ A1 r f2 ðr; h; /Þ þ B0 f3 ðh; /Þ ð1Þ
r
where f1 ; f2 , and f3 , the coefficients of A0 ; A1 , and B0 , are functions of h; / and

material properties of the orthotropic laminate ðEL ; ET ; GLT and mLT Þ and is given
by
8 9
> 1 1 mLT mTL a b 1 1 >
>
> cos tan ðð b þ a Þ tan hÞ >
>
pffiffiffi 2 < ET 1 þ mLT 2a 2 =
f1 ðh; /Þ ¼ 4 cos h þ ðb þ aÞ sin h
2

>
> 1 m 1 1 > >
>
:
LT
pffiffiffiffiffiffiffi
1
sin tan ððb þ aÞ tan hÞ >
;
GLT ð1 þ mLT Þ mTL 2a 2
8 9
> 1 1 mLT mTL a þ b cos 1tan1 ððb aÞ tan hÞ
> >
>
>
< >
=
pffiffiffi ET 1 þ mLT 2a 2
þ 4 cos2 h þ ðb aÞ sin2 h
>
> 1 mLT 1 1 > >
>
: þ pffiffiffiffiffiffiffi sin tan1 ððb aÞ tan hÞ >
;
GLT ð1 þ mLT Þ mTL 2a 2
ð2Þ
8 9
> cosð12 tan1 ððb þ aÞ tan hÞÞ >
> ffiffi ða bÞ >
1 1 mLT mTL 1 < 4ðcos2 h þ ðb þ aÞ2 sin2 hÞ =
p
f2 ðh; /Þ ¼
1 þ mLT 2a >
> cosð1 tan1 ððbaÞ tan hÞÞ >
ð a þ bÞ >
ET
: þ pffiffi 2 2 ;
4ðcos h þ ðbaÞ2 sin2 hÞ
8 9 ð3Þ
sin½1 tan1 ððb þ aÞ tan hÞ
> >
< pffiffi 2
>
>
=
1 mLT 1 4ðcos2 h þ ðb þ aÞ2 sin2 hÞ
þ pffiffiffiffiffiffiffi
GLT ð1 þ mLT Þ mTL 2a > > sin½1 tan ððbaÞ tan hÞ >
1
>
: pffiffi 2 2 ;
4ðcos h þ ðbaÞ2 sin2 hÞ
and
bh i
f3 ðh; /Þ ¼ ða þ bÞ2 ðb aÞ2 a11 cos2 / þ a12 sin2 / ð4Þ
2a
It could be shown that there are specific values of h and / for which the
coefficients of A1 and B0 could be made zero. Putting those values in Eq. (1), we get
8 2 39
cosð12 tan1 ððb þ aÞ tan hÞÞ
>
> ffiffi
6 4ðcos2 h þ ðb þ aÞ2 sin2 hÞ ða bÞ >
7>
p
>
> >
>
>
> 1 1mLT mTL 1 6 7 >
>
>
> E2 1 þ mLT 2a 4
ð Þ 5 >
>
>
> þ pffiffi4 cos2 h þ ðbaÞ2 sin2 h ða þ bÞ >
cos 1
tan 1
ðð ba Þ tan hÞ
>
< =
2
1 ð Þ
eaa ¼ pffiffi A0 2 3
r >
> sinð1 tan1 ððb þ aÞ tan hÞÞ >
>
>
> h i pffiffi 2 >
> ð5Þ
>
> mLT pffiffiffiffiffi 1 6
6 4ðcos2 h þ ðb þ aÞ2 sin2 hÞ 7 >
>
>
> þ 1 7 >
>
>
> G ð1 þ m Þ m 2a 4 sinð12 tan ððbaÞ tan hÞÞ
1 5 >
>
: 4 cos2 h þ ðbaÞ2 sin2 h ;
12 TL TL
p ffiffi
ð Þ
A0
) eaa ¼ pffiffi C
r
where C is a constant for a given value of h; / and material properties. Taking

logarithm on both sides of Eq. (5), we get
1
lnðeaa Þ ¼ lnðr Þ þ lnðC 0 Þ ð6Þ
2
A plot of Eq. (6) between lnðeaa Þ and lnðr Þ leads to a straight line of slope equal
to −0.5 with an intercept of lnðC 0 Þ. If rmax is the extent of valid three-parameter
zone, theoretically, the straight line property should break beyond r [ rmax as more
than three parameters will be needed in Eq. (1) to estimate the eaa . Thus, by placing
a single-strain gage as shown in Fig. 1 at a radial distance r within rmax from the
crack tip along the line at an angle of h and oriented at an angle of /, the measured
strain eaa can be equated to Eq. (5) to obtain the value of unknown coefficient A0 .
By applying the standard definition of mode I SIF, KI , it can be easily shown that
A0 is related to the SIF by the relation
pffiffiffiffiffiffi
KI ¼ 2pA0 ð7Þ
3 Numerical Simulation
A center-cracked orthotropic ½902 =010S carbon-epoxy laminate configuration sub-

jected to uniform tensile stress with the material properties as EL ¼ 67:77 GPa; ET ¼
142:7 GPa; mLT ¼ 0:01; GLT ¼ 4:304 GPa is considered (Fig. 2a). Following the
procedure described in Sect. 2, the values of / and h for which the coefficients B0 and
A1 become zero are found to be 84 and 61 , respectively.
Finite element analysis is carried out using ANSYS®. Figure 2b shows the
analysis domain with boundary conditions used for the numerical analysis, and
Fig. 2c shows the typical finite element mesh considered after proper convergence
study. Eight-noded isoparametric elements (PLANE 183) have been used for finite
element discretization of the plate, and quarter point elements (QPE) have been
Fig. 2 a Orthotropic center-cracked plate, b analysis domain, c FE mesh of the model

used around the crack tip. The mesh has been designed such that the nodes of
several elements are made to lie along the gage line. According to the present
technique, a single-strain gage is to be placed at an appropriate location along the
gage line and oriented along / in order to measure the linear strain eaa .
From the finite element results, the linear strain ðeaa Þ and radial distance ðr Þ are
computed for all the nodes along the gage line. Figure 3 shows the plot of lnðeaa Þ
versus lnðr Þ for all the nodal values along the gage line. Crack tip point is not
plotted as the radius of this point is zero. It can be seen that the plot consists of a
linear plot followed by a nonlinear one as predicted by the theory. The radial
distance till which the plot is linear and beyond which the plot changes from a
linear one to a nonlinear one is rmax or the extent of the three-parameter zone or the
optimal radial location for pasting the strain gage. A line of slope −0.5 is super-
posed on the plot of lnðeaa Þ versus lnðr Þ. Considering this line to be the exact
solution, absolute percentage relative error at all values of radius is being calcu-
lated. Finally, rmax is estimated as the radius at which the error is less than 1% (as
one goes from right to left in Fig. 3). The analytical value of mode I SIF of this
configuration is given by [7]
pffiffiffi
KI ¼ YI ða=bÞr a ð8Þ
where r is the applied stress, a is the crack length, and YI is the specimen geometric
factor given by [12]
YI ¼ 1:77 þ 0:227ða=bÞ 0:510ða=bÞ2 þ 2:699ða=bÞ3 ð9Þ
For this configuration with crack length to width ratio ða=bÞ of 0.4 and loaded at
r ¼ 100 MPa, the reference value of mode I SIF determined using Eq. (8) is KI ¼
pffiffiffiffi
27:6 MPa m and rmax is found to be 22.42 mm. Radial locations are selected
within and outside the simulated rmax to establish the importance of radial posi-
tioning of the strain gage. The percent relative errors in measured KI determined
Fig. 3 Linear and nonlinear variation of lnðeaa Þ with lnðr Þ for the nodes along the gage line
Table 1 Simulated mode I pffiffiffiffi

r ðmmÞ eaa KI ðMPa mÞ % Relative error
SIF of the orthotropic panel at
optimal and nonoptimal 10.04 1.29E−03 28.11 1.8
positions 16.06 1.05E−03 28.2 2.1
25.01 8.82E−04 33.25 20.47
30.06 8.28E−04 34.22 23.98
using Eq. (7) from the strain readings at those locations obtained from the finite
element analysis are shown in Table 1. The percentage relative error in KI mea-
sured at those locations is computed as
KReferencesolution Kmeasuredorsimulated
%Rel. error ¼ 100 ð10Þ
KReferencesolution
Results in Table 1 clearly show that it is possible to accurately determine KI

(order of error 2%) for center-cracked orthotropic laminates using a single-strain
gage if the gage is placed within rmax . On the other hand, placing a strain gage
outside rmax leads to highly inaccurate values of KI of about 24%. These results
substantiate that the present technique of determination of KI using a single-strain
gage can be used for accurate determination of KI even for double-ended cracked
orthotropic composite specimens if the gages are placed within the optimal locations.
4 Conclusion
The single-strain gage technique for determination of mode I SIF of cracked

orthotropic materials has been numerically verified for carbon-epoxy laminates with
a double-ended cracked configuration thereby establishing the successful imple-
mentation of the single-ended crack theory to double-ended cracks for orthotropic
materials. Numerical results show that accurate estimation of KI of the
center-cracked carbon-epoxy laminate is possible for strain gages well within the
optimal locations. On the other hand, high errors of the order of around 24% could
be observed if the gages are placed outside the estimated rmax for that configuration.
These observations establish the fact that proper positioning of strain gages as per
the predetermined optimal locations is essential for accurate estimation of SIFs.
References
1. G.R. Irwin, Analytical aspects of crack stress field problems. T&AM Report No. 213,
University of Illinois, Urbana (1962)
2. E.M. Wu, On the application of fracture mechanics to orthotropic plates. T&AM Report
No. 248, University of Illinois, Urbana (1963)
3. J.W. Dally, R.J. Sanford, Strain gage methods for measuring the opening mode stress
intensity factor. Exp. Mech. 27, 381–388 (1987)
4. M.C. Baik, S.H. Choi, J.S. Hawong, J.D. Kwon, Determination of stress intensity factors by
the method of caustics in anisotropic materials. Exp. Mech. 35, 137–143 (1995)
5. M. Mojtahed, L.W. Zachary, Use of photoelasticity to determine orthotropic stress-intensity
factor. Exp. Mech. 27, 184–189 (1987)
6. D. Cerniglia, V. Nigrelli, A. Pasta, Experimental and numerical determination of stress
intensity factor in composite material, in Conference Proceedings ICCM-12, Paper 932,
Europe (1999)
7. A. Shukla, B.D. Agarwal, B. Bhusan, Determination of stress intensity factor in orthotropic
composite materials using strain gages. Eng. Fract. Mech. 32, 469–477 (1989)
8. S.K. Khanna, A. Shukla, Development of stress field equations and determination of stress
intensity factor during dynamic fracture of orthotropic composite materials. Eng. Fract. Mech.
47, 345–359 (1994)
9. D. Chakraborty, K.S.R.K. Murthy, D. Chakraborty, A new single strain gage technique for
determination of mode I stress intensity factor in orthotropic composite materials. Eng. Fract.
Mech. 124–125, 142–154 (2014)
10. K. Bearden, J.W. Dally, R.J. Sanford, Experimental determination of for short internal cracks.
J. App. Mech. 68, 937–943 (2001)
11. D. Chakraborty, K.S.R.K. Murthy, D. Chakraborty, Determination of in orthotropic laminates
with double ended cracks using a single strain gage technique. Theory Appl. Frac. Mech. 82,
96–106 (2016)
12. H. Tada, P.C. Paris, G.R. Irwin, The Stress Analysis of Cracks handbook (ASME, New York,
2000)
Mechanical Testing of Elevated
Temperature PMC, Metallic,
and CMC Coupons
Erik A. Schwarzkopf and Michael J. Shepard
Abstract A cornerstone of structural integrity is the proper use and characteriza-

tion of structural materials. Improper materials characterization or use of a material
system outside its designed use range (intentionally or unintentionally) can have
perilous results for structural integrity. As gas turbine engines and other applica-
tions drive toward higher operating temperatures for structural elements, the
materials characterization and design allowable generation activities become
increasingly difficult. We discuss elevated temperature testing requirements and the
development of testing solutions aligned with these requirements. Mechanical
testing systems for elevated temperature testing for polymeric matrix composites
(PMCs), metallics, and ceramic matrix composites (CMCs) were developed. Test
capability for temperatures as high as 1500 °C was demonstrated. The development
of these systems will be briefly reviewed with particular focus on design aspects,
system performance, and general usability of a system targeting testing of cylin-
drical metallic specimens at temperatures up to 1200 °C.
Keywords High temperature PMC CMC Metals Mechanical testing
1 Introduction
The global marketplace is driving toward transportation and energy products and
systems that are increasingly efficient. In this drive toward increasing efficiency,
two principle themes are dominant with regard to material selection. The first theme
is the utilization of alternative materials that are superior in terms of their
mechanical properties on a density normalized (specific) basis. The second theme is
increasingly high operational temperatures for energy conversion processes where
this will improve efficiency (combustion). These material selection themes can
E. A. Schwarzkopf (&) M. J. Shepard

MTS Systems Inc., 14000 Technology Drive, Eden Prairie, MN 55344, USA
e-mail: Erik.Schwarzkopf@MTS.com

https://doi.org/10.1007/978-981-10-7197-3_39
470 E. A. Schwarzkopf and M. J. Shepard
often involve substitutions of materials of entirely different classes. Relevant

examples of such substitutions from an aerospace context include ceramic matrix
composites (CMCs) for nickel alloys, titanium alloys for nickel alloys, and polymer
matrix composites (PMCs) for titanium or aluminum alloys.
The ramifications of these material selection decisions with respect to materials
testing requirements at a wide range of temperatures have been discussed and
summarized previously [1–7]. Mechanical testing systems must be capable of
achieving appropriate testing temperatures and mechanical loading conditions. The
temperature requirements tend to be a primary consideration due to the complexity
that elevated temperatures bring for general system function and durability.
Summarizing briefly, for PMCs, titanium alloys, and aluminum alloys, a nominal
use temperature range of 23–425 °C is appropriate. For high-temperature metals, a
temperature range of nominally 800–1200 °C covers a useful range of service and
processing temperatures. And finally for CMCs in structural applications, testing at
temperatures between 800 and 1500 °C are required.
For each of these temperature ranges and families of materials, appropriate
systems for mechanical testing are required. In addition to a load frame, a
mechanical testing system will require a method of holding and applying load to
test specimens (grips), a method for getting the test specimen to appropriate tem-
perature (furnaces or environmental chambers, induction heating), and often some
method of characterizing strain (extensometry, strain gages, etc.). Each of these
elements contributes to the proper execution of the mechanical test and the accuracy
of the collected data. The performance of these individual components can be
interrelated, so a systems approach is appropriate.
This paper will focus on experimental system evaluation results for a testing
system designed for testing cylindrical, metallic specimens at temperatures up to
1200 °C.
1.1 High-Temperature Grips
To understand how interrelated component issues manifest during test setup, con-
sider a typical specimen and a typical loading history. Metallic parts used in turbine
engines often have duty cycles that include “through-zero” loading. Accurate ele-
vated temperature and property measurement of these materials require minimal
thermal gradient in the specimen gage section and the ability to maintain positive
specimen contact with the grip while maintaining good specimen alignment during
the entire loading history [8, 9]. Traditionally, this means that both the grips and
specimen are controlled to the desired test temperature to minimize temperature
gradients. For cylindrical metallic specimens, traditional grip end designs are
threaded or “button-head” to allow sufficient compressive preloading of the grip
end to allow through-zero testing of the gage section. These “hot” grips become
problematic when the application temperature for metallic specimens exceeds about
Mechanical Testing of Elevated Temperature … 471
1000 °C [10]. At these highest temperatures, the stresses in the grips exceed the
strength of commonly available materials and also leave the grip structures sus-
ceptible to creep distortion or microstructural instability.
“Cold” grips, where the specimen is gripped outside of the furnace, are often
used in these more aggressive temperatures, but require a long specimen to extend
outside of the furnace, and result in significant temperature gradient in the speci-
men. Experimental economy, especially with costly developmental materials, does
not generally allow these longer specimens, and the thermal gradients can become
significant, adding experimental scatter and potentially reducing test yield due to
failures outside the gage section.
“Warm grips” are a compromise. Multi-zone furnaces are used to keep the
specimen gage length at the desired test temperature while the grip and specimen
ends are heated as close as possible to the test temperature while still maintaining
the strength of the grips. These grips also allow significantly shorter specimen
dimensions and greatly reduced thermal gradients. The trade-off, essentially, is an
increase in grip and furnace complexity in exchange for an increase in test quality.
Three zone furnaces for materials testing are common, but “warm grips” usually
require special integrated cooling of the grip, since the grip is so close to the very
hot center zone of a furnace.
The advantage of “warm grips” becomes more apparent when testing CMC
specimens at the highest application temperatures approaching 1500 °C [11].
CMCs, like PMCs, typically are tested as flat specimens. Unlike the metallic
specimens, these flat, “dog-bone” specimens do not have threaded ends or buttons
for end preloading. The most common solution is to load the specimen faces in
shear using wedges. A very effective solution is to use hydraulically actuated
wedges to grip the flat faces of these specimens. CMC and PMC specimens often
have very poor through-thickness (short transverse) strength so uniformly applied
pressure from hydraulic wedge faces on the gripped region minimizes specimen
damage during gripping. often lack compressive strength across their thinnest cross
section, and the evenly applied pressure from the hydraulic wedges protects the
fibers in the polymer or ceramic matrix. In addition to preventing specimen damage,
hydraulic grips also maintain constant gripping pressure as the specimen reaches
testing temperature. Other solutions exist for gripping flat specimen, but many of
the other solutions are limited to tensile only loading, or apply even displacement,
rather than even pressure, which changes as both the grip and coupon changes their
shape during heating.
The wedges used for flat composite specimen are retained by a wedge chamber,
and at temperatures above 1000 °C, no commonly available material can easily
withstand the hoop stress caused by the wedges in the chamber. As a result, “warm
grips” are used for CMC testing at test temperatures between 1000 and 1500 °C.
“Hot grips,” with the grips at the same temperature as the specimen, are used for
PMC testing at temperatures below 425 °C.
1.2 Furnaces and Chambers
Furnaces and chambers must be paired with gripping technology to optimize ele-
vated temperature testing results and minimize data variability. For the highest
temperatures, where “warm grips” are required, furnaces typically have a special
central zone with special elements providing the energy density to reach these
highest 1200–1500 °C temperatures. For these highest temperatures, MoSi2 heating
elements are typical. These elements require slightly more complex controlling
electronics for laboratory furnaces, but provide higher energy densities and higher
temperatures than more traditional silicon carbide or resistive alloy heating ele-
ments used in most laboratory furnaces.
The high power density of MoSi2 furnace elements allow the heated zone of the
furnace to be smaller, allowing testing professionals to use smaller specimens. The
hot, center zone of these very high-temperature furnaces is typically about 50 mm
allowing for a 25 mm gage length. The outer zones of these 1200 °C plus furnaces
typically use resistive alloy elements due to their cost-effective and robust nature.
Elevated temperature testing of PMCs uses “hot grips,” since the test temperature is
typically much less than 371 °C, the highest Tg temperatures for the composite resins.
These MTS model 647 All-Temperature grips grips are general purpose wedge grips
designed for working temperatures up to about 540 °C. As general purpose grips, they
can accommodate a wide range of specimen geometries, with a concomitant increase
in size of the wedge chamber. The preferred heating subsystems used for these PMC
tests are referred to as an environmental chamber and are usually comprised of a large
heated box chamber (1/4 cubic meter heated space) to accommodate the hydraulic
wedge grips and provide ample volume for various fixtures associated with PMC
testing. Environmental chambers offer a capability to include test effects such as
humidity and allow for controlled venting of noxious gasses that can be given off by
composites during elevated temperature testing. The chamber for these PMC tests is
often larger than the furnaces required for higher temperature metal or CMC tests.
Although larger equipment usually is less efficient, the thermal mass of the grips and
chamber leads to very stable test environments.
1.3 Contact and Optical Extensometry
The most common methods for accurate strain measurement during mechanical
testing are contact extensometry and strain gages. Contact extensometry works well
for testing with more conventional furnaces. With a smaller furnace, extensometers
can be situated outside the chamber with their rods extending through furnace
penetrations to contact the specimen [12–14]. With environmental chambers used
for PMC testing, the larger chamber makes using contact extensometers more
difficult. Due to the size of the environmental chamber, the extensometer rods
become too long and cause additional measurement variability. A more compact
extensometer could be used inside the chamber but the high temperatures needed
for some experiments would not be compatible with most available extensometers.
Strain gages are available that can perform at elevated temperatures, but the
inconvenience and expense of strain gage application is not particularly desirable,
especially for high volume testing.
Increasing use is being made of video extensometry/digital image correlation.
These methods are generally available are non-contact and can be situated outside
the chamber if an optical quality window of sufficient size is available. The method
requires a specimen with significant surface texture or user-applied targets or pat-
terns. Uniform lighting is particularly important. At higher temperatures, the glow
from heated specimens can reduce contrast and interfere with the video quality
needed for precision measurements. For the PMC system discussed, a custom blue
LED lighting solution was utilized in conjunction with filters to minimize any loss
of contrast due to specimen heating.
2 Results
Representative, preliminary results are reported for the testing system capable of
testing cylindrical, metallic specimens at temperatures up to 1200 °C. This system
is depicted in Fig. 1 after over 1000 h of elevated temperature use. Similar char-
acterizations are complete or in-process for the previously discussed systems
operating at 425 and 1500 °C.
2.1 System Alignment
Proper system/grip alignment is critical to ensure excessive bending strains are not
induced in the test specimen, reducing test accuracy. The relevant specifications for
alignment are ASTM E1012 and ISO 23788. The alignment of the MTS 1200 °C
test system, installed on a custom fabricated servo-hydraulic load frame, was
assessed using the MTS 709 alignment system, as per ASTM E1012. The system
was found to be alignable to better than ASTM E1012 Class 5, the best available
classification. This level of alignability has been demonstrated with both a
button-head and threaded specimen end configuration.
Figure 2 depicts one relevant experiment from this alignment exercise. In this
figure, the bending strains associated with three banks (upper, middle, and lower) of
four strain gages on the gage section of an alignment specimen are plotted. For
compliance with ASTM E1012 Class 5 alignment requirements, all data must fall
under the bending strain limits indicated by the dotted lines. The system was found
Fig. 1 1200 °C test system (grips and furnace) installed on servo-hydraulic load frame
compliant with E1012 Class 5, as indicated in Fig. 2, as well with repeated removal
and rotation of the specimen, as required by ASTM E1012.
2.2 Thermal Gradients in Specimen Gage Section
To evaluate thermal gradients achievable in typical specimen gage sections at

1200 °C, representative cylindrical dog-bone specimens were fabricated in three
different total specimen lengths, all as per the recommendations of ASTM E606—
Standard Test Method for Strain-Controlled Fatigue. The specimens were ther-
mocoupled to allow for an assessment of the thermal gradient achievable in the
specimen gage section. Figure 3 depicts the thermal gradients achieved over the
specimen gage sections as a function of time for the longest specimen (152.4 mm
total length). Table 1 depicts the thermal gradients achieved for all specimen
Fig. 2 Bending strain versus applied axial strain for ASTM E1012 1200 °C system alignment
Temperature (°C)
-38.1 -25.4 -12.7 0 mm 12.7 25.4 38.1

Position from Center of Gage Length (mm)
Fig. 3 Thermal gradients over time for 152.4 mm specimen

Table 1 Gage section thermal gradients achieved for three representative specimen lengths at
1200 °C
Total specimen length Gage length Gage gradient (°C) ASTM E606 target (°C)
(mm) (mm)
101.6 25.4 7 14.7
127 19.0 3–5 14.7
152.4 12.7 2–3 14.7
lengths. The gradients are significantly better that the ±1% recommended in E606,
demonstrating excellent thermal gradient performance.
2.3 Grip Durability
Since the load-bearing elements of the grip are operating at significantly elevated
temperatures, durability needs to be thoroughly evaluated. No failures were expe-
rienced through 990 h of elevated temperature operation. Historically, in grips
operating for extended periods at elevated temperatures, gross creep distortion had
been observed in structural elements subjected to the most aggressive combinations
of temperature and stress. In the case of this grip design, the grip inserts (sometimes
called grip nuts) experienced the most challenging creep exposures (highest tem-
peratures under sustained load). Measurements were made throughout the testing
campaign to document gross creep distortion (Fig. 4). No measureable creep dis-
tortion was observed, and operation of these threaded elements was routine
throughout all 990+ hours of use.
Fig. 4 Creep assessment in worst-case structural elements (no gross creep distortion indicated)
Fig. 5 Minor oxidation distress in non-structural elements
On non-structural elements, subjected to temperatures nearly as high as the gage

section, some modest oxidation was observed. The observed oxidation was not
sufficient to inhibit function or warrant replacement (Fig. 5).
3 Conclusions
High-temperature testing considerations for three mechanical testing systems have

been discussed—a system for PMCs (up to 425 °C), a system for metals (up to
1200 °C), and a system for CMCs (up to 1500 °C). Equipment and associated
trade-offs are discussed particularly with respect to alignment, thermal gradient,
durability, and strain measurement, with specific examples being provided for the
1200 °C system under development.
The importance of appropriate gripping solutions was reviewed. A proper grip
needs to effectively couple loads into a specimen without causing damage in the
gripped section. The grip needs to align properly so that any bending moments
induced in the specimen are strictly minimized. And finally, it needs to be able to
survive the testing temperatures sufficiently well to provide a satisfactory service
life. This may require the grip to be made of high-temperature materials and, in
extreme cases, have integrated water and/or air cooling.
The selected thermal solution (furnace, chamber, etc.) needs to be sufficient to
reach the targeted temperatures with the lowest possible thermal gradient across the
specimen gage section. It also needs to accommodate any needed accessories such
as contacting or video extensometry.
Finally, if strain measurement is required for the experiment, the system archi-
tecture, particularly the heating solution, needs to be configured to accommodate it.
The strain measurement solution itself needs to be employed in a manner that will
not subject it to a damaging environment or reduce its accuracy significantly.
Acknowledgements The continuing assistance of the US Air Force Research Laboratory’s

Materials and Manufacturing Directorate under Cooperative Research and Development
Agreements 13-210-RX-01 and 13-210-RX-02 is kindly acknowledged. Particular thanks are
extended to Drs. Andrew Rosenberger and Jonathan Spowart. Additional valuable support was
provided by Mr. Phillip Blosser and Ms. Jennifer Pierce of the University of Dayton Research
Institute, as well as Mr. Larry Zawada of Universal Technologies Corporation.
References
1. E.A. Schwarzkopf, Evaluating tradeoffs in high-temperature testing. Adv. Mater. Process.

174, 16–18 (2016)
2. Z. Fawaz, Quality control and testing methods for advanced composites materials in
aerospace engineering, in Advanced Composite Materials for Aerospace Engineering:
Processing, Properties, and Applications, ed. by S. Rana, R. Fangueiro (Woodhead
Publishing/Elsevier Science, 2016), pp. 429–452
3. U. Burger, L. Rochat, Aspects of damage tolerance and fatigue of CFRP structural
components. SAE Int. J. Aerosp. 8(2), 292–302 (2015)
4. A.L. Gyekenyesi, M.G. Castelli, J.R. Ellis, C.S. Burke, A study of elevated temperature
testing techniques for the fatigue behavior of PMCS: Application to T650-35/AMB21. NASA
Technical Memorandum 106927 (1995)
5. J. Mills-Brown, K. Potter, S. Foster, T. Batho, The development of a tensile testing rig for
composite laminates. Compos. A 52, 99–105 (2013)
6. D.C. Larsen, S.L. Stuchly, The mechanical evaluation of ceramic fiber composites, in Fiber
Reinforced Ceramic Composites, ed. by K.S. Mazdiyasni (Noyes Park Ridge, NJ, 1990),
pp 182–221
7. D.M. Dawson, R.F. Preston, A. Purser, Fabrication and materials evaluation of high
performance aligned ceramic fiber-reinforced, glass-matrix composite, ed. by W. Smothers.
11th Annual Conference on Composites and Advanced Ceramic Materials: Ceramic
Engineering and Science Proceedings, vol. 8, Issue 7/8, (Wiley, Hoboken, NJ, USA, 1987)
8. K. Staerk, High temperature axial strain-controlled LCF\TMF fatigue testing of flat-sheet
specimens, in Fatigue and Durability Assessment of Materials, Components and Structures,
ed. by M.R. Bache, P.A. Blackmore, J. Draper, J.H. Edwards, P. Roberts, J.R. Yates
(Sheffield, UK, 2003), pp. 389–397
9. P.B.S. Bailey, A.D. Lafferty, Specimen gripping effects in composites fatigue testing—
Concerns from initial investigation, sXPRESS. Polym. Lett. 9(5), 480–488 (2015)
10. R.J. Greene, G.S. Hartman, A.H. Rosenberger, Detection of small cracks in nickel-based
superalloys at elevated temperature, in 2004 SEM X International Congress & Exposition on
Experimental & Applied Mechanics (2004)
11. G.A. Hartman, L.P. Zawada, S.M. Russ, Techniques for elevated temperature testing of
advanced ceramic composite materials, in Fifth Annual Hostile Environments and High
Temperature Measurements Conference. Society for Experimental Mechanics, Costa Mesa,
CA (1988)
12. J.Z. Gyekenyesi, J.H. Hemann, Optical strain measuring techniques for high temperature
tensile testing. NASA Contractor Report 179637 (1987)
13. J.Z. Gyekenyesi, P.A. Bartolotta, An evaluation of strain measuring devices for ceramic
composites. NASA Technical Memorandum 105337 (1991)
14. L.P. Zawada, Longitudinal and transthickness tensile behavior of several oxide/oxide
composites. Cerm. Eng. Sci. Proc. 19(3), 327–339 (1998)
In Situ Measurement of Deformation
Under Tension of ABPBI
and Its Composites
Abhijeet Dhiman, F. G. S. Wasim, M. Neergat

and Krishna N. Jonnalagadda
Abstract Fuel cell devices convert chemical energy directly into electrical energy
using hydrogen as fuel and are appreciated for generating clean energy, but are not
yet commercially significant. Research in this area is in progress to make hydrogen
fuel cells commercially feasible with good efficiency and reliability. Operating fuel
cells below 100 °C temperature lead to poisoning of platinum electrodes due to
small traces of carbon mono-oxide. This effect can be reduced significantly at
temperatures above 150 °C. However, the commonly used Nafion membrane
cannot be used at such temperatures due to the degradation in conductivity as a
result of dehydration. One solution to this problem is to substitute commonly used
Nafion membrane by other polymer electrolytic membrane, which can be operated
at high temperatures. Recently, Poly (2,5-benzimidazole) (ABPBI) has been
developed as a potential polymer electrolyte membrane (PEM) for fuel cell appli-
cations. ABPBI membrane doped with strong acids like phosphoric acid enhances
conductivity at high temperatures. Furthermore, operating at high temperatures
poses challenges to mechanical integrity and durability of the membrane. The
mechanical endurance is one of the limiting factors for the long-term durability of
PEM-based fuel cells. The degradation in performance of membrane is believed to
be the result of mechanical and chemical effects acting together. Taking mechanical
effects in brief, variation in temperature and humidity during the operation of fuel
cell produces strains in membrane. Also, high strains are generated while starting
fuel cell from cold state to operating temperature of 150 °C. In this work, the
mechanical response of ABPBI polymer, ABPBI polymer doped with PWAZrO2
inorganic filler, and phosphoric acid-doped ABPBI polymers was characterized.
Mechanical behavior was extracted via in situ tensile experiments on 20–40 µm
polymer specimens. It was found that the addition of filler increased the stiffness of
A. Dhiman (&) M. Neergat K. N. Jonnalagadda

Indian Institute of Technology Bombay, Powai, Mumbai 400076, India
e-mail: abhi.dhiman.iitb@gmail.com
K. N. Jonnalagadda
e-mail: krishnajn@iitb.ac.in
F. G. S. Wasim
IITB-Monash Research Academy, Powai, Mumbai 400076, India

https://doi.org/10.1007/978-981-10-7197-3_40
482 A. Dhiman et al.
the membrane while acid-doped membranes with and without fillers showed a
significant decrease in the stiffness and an increase in ductility.

Keywords Fuel cell Polymer electrolytic membrane Poly (2,5-benzimidazole)

Digital image correlation In situ tensile testing
1 Introduction
Fuel cell devices convert chemical energy directly into electrical energy using
hydrogen as fuel. Polymer electrolyte fuel cells are constructed using polymer
electrolytic membrane (PEM) as the proton conductor which separates the anode
and the cathode. Three main functions of PEM include acting as medium for ion
conductivity, providing barrier for separating reactant gases, and support structure
for electrode catalysts [1]. Over the years, several membranes materials have been
developed, of which Nafion is the most commonly used polymer electrolytic
membrane or proton exchange membrane. For effective conduction of protons,
Nafion has to be hydrated with water, which limits its use up to the temperature of
80 °C [2]. Above this temperature, conductivity of Nafion reduces drastically as the
result of dehydration. Furthermore, platinum electrodes used in PEMFCs are very
sensitive to corrosion in the presence of even small traces of carbon mono-dioxide,
which are present along with reactant gases or fuel. This poisoning of platinum
electrodes can be reduced by operating at higher temperatures in range of 150–
200 °C [3]. Also, operating at higher temperatures improves heat and thermal
management of fuel cell with increase in reaction rates [4]. During the past decade,
there has been extensive research on developing polymer electrolyte membranes,
which could exhibit adequate proton conductivity at low humidity levels. But high
cost and low durability are the two barriers for their commercial use.
Acid-doped polymer membranes are believed to be possible substitution of
Nafion due to thermal stability and ionic conductivity at high temperatures. poly-
benzimidazole (PBI) is a commercially available polymer, which shows appreciable
conductivity up to 200 °C when doped with strong acids, such as phosphoric acid
and sulfuric acid [5–7]. The family of polybenzimidazole is not limited to PBI. The
base polymer studied in this work, poly (2,5-benzimidazole) (ABPBI), is the
simplest among benzimidazole-type polymers since it can be prepared from single
and inexpensive commercial monomer (3,4-diaminobenzoic acid) (DABA) by
condensation in polyphosphoric acid (PPA) or methanesulfonic acid (MSA).
Compared to PBI, the ABPBI polymer has a higher affinity toward phosphoric acid
because of the absence of phenylene ring. High amounts of phosphoric acid doping
are required in ABPBI to achieve high proton conductivity at high temperature
same as the case of PBI. However, the mechanical properties of PBI-based mem-
branes have been shown to deteriorate by doping with strong acids due to swelling,
plasticizing nature of acid [8], and acid leaching under continuous usage. To
increase resistance to deformation, ABPBI membranes have been doped with
In Situ Measurement of Deformation Under Tension … 483
inorganic filler like ZrO2 [9]. Impregnation of inorganic filler shows increase in
stiffness but also results in decrease in strain-to-failure due to poor adhesion
between nano-fillers and polymer matrix.
The durability of membranes is affected by the combined mechanical and
chemical loads acting together during the operation of the fuel cell. Variations in
temperature and humidity are the main factors for deterioration in mechanical
response of PEM membrane as they cause swelling and internal stresses. At the
molecular scale, the fluctuation in humidity level can induce changes in morphol-
ogy as well as lead to macroscopic dimensional changes, which can be quite
significant. Huang et al. [10] showed that the adverse effects of relative humidity
cycling on mechanical properties in Nafion. At the start of a PEM fuel cell, the
acid-doped ABPBI undergoes large changes in humidity and temperature.
Furthermore, RH fluctuations during operation of fuel cell can affect the overall
performance and durability. Hence, it is also important to study the mechanical
behavior of polymer electrolytic membrane subject to different RH conditions.
Various techniques have been developed over the past few decades to estimate
the mechanical properties of polymers. The reported values for modulus of ABPBI
varied depending upon the type of testing method, viz., tensile test, indentation and
varied in the range of 1.19–2.4 GPa [9, 11, 12]. Tensile strength was reported in
range of 69.9–121 MPa [12–14], and strain-to-failure was reported between 8 and
65% [12–14] for ABPBI polymer. Determination of the elastic modulus and frac-
ture strength for polymeric thin films using tensile testing method has been reported
to be a better technique [15, 16]. In situ tensile experiments provide accurate
measurements of mechanical properties of thin films over a wide range of loading
conditions. In this work, in situ optical experiments were combined with DIC to
extract local deformation fields used to compute the mechanical properties of
polymer at room temperature of 25 ± 0.5 °C and humidity of 55 ± 5%. The
advantage of this technique is its ability to measure local strains with minimally
invasive specimen preparation, which is not affected by strain localization, necking,
and cracks in the gauge section of the specimen. In this work, the mechanical
behavior in the form of stress–strain curves has been extracted for ABPBI mem-
branes in dry as well as wet conditions with and without PWAZrO2 inorganic filler.
All the experiments were conducted in situ, and therefore, accurate measure of
strain was obtained under large extensions to failure.
2 Experiments
2.1 Materials
Phosphotungstic acid (PWA), zirconium dioxide nano-particles (ZrO2), and

3,4-diaminobenzoic acid (DABA) 97% from Sigma-Aldrich and phosphorous
pentoxide (P2O5), methanesulfonic acid (MSA), phosphoric acid (PA), and
ammonium hydroxide (NaOH) from Merck were used without further treatment.
De-ionized (DI) water was obtained from Direct-Q Millipore de-ionizer.
2.2 Synthesis of Poly (2,5-Benzimidazole) (ABPBI)
ABPBI was synthesized by following the method reported elsewhere [14]. The
synthesis procedure is as follows: in a round bottom flask, 3 g of P2O5 was dis-
solved in 22 mL of MSA at a temperature of 150 °C for 1 h under nitrogen. The
temperature was raised to 160 °C and 2 g of DABA was added slowly, and the
reaction was carried out under nitrogen purging. After 30 min, the dark purple
viscous solution was precipitated in de-ionized water and neutralized with 10%
NaOH solution. The polymer was again washed several times with DI water and
dried in oven at 80 °C. The obtained filaments of polymer were crushed and used
for further characterization (Fig. 1).
2.3 Synthesis of Phosphotungstic Acid-Grafted Zirconium

Dioxide Particles (PWA/ZrO2)
About 400 mg of phosphotungstic acid was dissolved in 50 mL DI water and 2 g of

ZrO2 nano-particles were added and magnetically stirred for 24 h [17]. The reaction
temperature was increased to 100 °C and the dispersion was allowed to dry under
continuous stirring. The dried powder was calcined at 200 °C for 2 h and stored in
airtight bottle for further use.
2.4 Preparation of Membrane
The ABPBI filaments were dissolved in a known amount of MSA for 10 h on a

magnetic stirrer to make a 3 wt% solution [14]. The homogeneous viscous solution
was poured on a petri dish and heated to 180 °C on a hot plate under a fume hood
until the solvent completely evaporated. The petri dish diameter was appropriately
chosen to get the desired membrane thickness. Membranes of thickness varying
Fig. 1 Synthesis of ABPBI from DABA

between 10 and 40 µm were cast to perform different experiments. The dried

membranes were immersed in DI water to remove any traces of acid. Further,
membranes were dried in vacuum oven at 60 °C for 6 h.
2.5 Preparation of Composite Membranes
Composite membranes of 2 and 5 wt% PWA/ZrO2 nano-particles were prepared in

the following way: a known amount of PWA/ZrO2 nano-particles were sonicated in
5 mL MSA to get a uniform dispersion of the particles and mixed with the ABPBI
solution to get the desired loading. The mixture was magnetically stirred for 3 h and
sonicated well before casting on the petri dishes; uniform, 10–40 µm thick mem-
branes obtained.
2.6 Phosphoric Acid Doping of the Membranes
The ABPBI and composite ABPBI membranes dried in the oven at 80 °C for 24 h,
and the dry weight of the membrane was recorded. The dry membranes were
immersed in 60 wt% phosphoric acid for 3 days and the surface acid was wiped
with a tissue and dried in oven at 80 °C to get the acid-doped membranes. The
percentage intake of phosphoric acid doping was calculated by recording the weight
change before and after phosphoric acid doping. All the membranes showed an acid
uptake of *80%.
2.7 Specimen Preparation for Tensile Experiments
For tensile experiments, the ABPBI and ABPBI composite specimen geometry was
designed for holding between two symmetric clamps, which leads to end effects
near the boundaries [18]. These end effects result in non-uniform stress distribution
near boundaries. To reduce the end effects, tensile test specimen was prepared in the
shape of a dog bone by carefully cutting them from the membrane using a template,
whose shape was patterned by EDM wire cut machining to ensure precision.
Figure 2a shows the pattern and specimen prepared. The dimensions of dog
bone-shape specimen used for the experiments are shown in Fig. 2b with 5 mm
gauge length and 0.7 mm width. The dimensions of specimen were chosen as per
the relation mentioned in literature [18]. To verify the dimensions selected, an
elastic finite element (FE) Analysis was done to ensure uniform stress distribution
in gauge length under uniaxial tensile loading. As observed from Fig. 2c, uniform
Fig. 2 a Template and specimen, b dimension of dog bone-shape specimen, and c axial stress and
maximum principal stress distribution under uniaxial tension
stress distribution is obtained at gauge section. There were regions of stress vari-
ation at ends of gauge section, but the magnitude of stresses here is less as com-
pared to those in the gauge section. Thus, from this analysis, the region of gauge
section can be considered under uniaxial loading.
2.8 In Situ Tensile Experiment
The tensile tests were carried out on a modified micro-tensile setup developed in
our laboratory [16]. The apparatus, shown in Fig. 3, mainly consists of an optical
microscope, camera, piezoelectric actuator, load cell, positioning stages, data
acquisition system (DAQ), and computer. During the experiment, optical images
were captured using a microscope at 50 magnification and a Sony monochrome
CCD camera with 1024 780 pixels. Conventional strain measurement techniques
cannot be used because of small dimensions of the specimen. So, digital image
Fig. 3 In situ optical micro-tensile setup
correlation (DIC) was employed to calculate full-field deformation on the sample

surface [19–21].
Digital image correlation technique requires the surface of the specimen with
speckle pattern, which gives a random distribution of gray values when images are
taken with a monochrome camera. The casted ABPBI and PWAZrO2 composite
polymer membrane were found to have dense and random speckle pattern Fig. 4,
which was suitable for extracting deformation fields over the gauge section.
However, pure ABPBI membrane has a pattern, which is quite large for quality DIC
results. To get more accurate results, a small and dense pattern of silicon powder is
sprayed over the surface [16]. Figure 4c shows the speckle pattern made from
silicon powder. The specimen was gripped between clamps on the side of actuator
and load cell. Fine adjustments were made using the translational and rotational
Fig. 4 Speckle pattern of a ABPBI, b ABPBI/PWAZrO2, and c silicon powder over ABPBI
stages for alignment of specimen. An actuator applied the displacement at constant

velocity and simultaneously the specimen is continuously imaged, which were
post-processed to get strain fields and strains using VIC-2D DIC software.
3.1 In Situ Tensile Testing
The optical images of the thin film specimens obtained during the tensile exper-
iments were analyzed with the DIC software. Figure 5 shows the full-field dis-
placement contours calculated using DIC software for ABPBI with 2 wt%
PWA-ZrO2. Note that all the experiments in this manuscript are conducted at strain
rate range of 0.0002/s. The displacement contours were aligned with the loading
axis (u), and this confirmed good alignment for uniaxial experiment during the
experiments.
Figure 6 shows the stress−strain response of ABPBI, ABPBI with 2 wt%
PWA-ZrO2, and ABPBI with 5 wt% PWA-ZrO2. These results suggest that the
elastic modulus and yield strength increase with increasing percentage of
PWA-ZrO2 nano-particle fillers. It can be concluded that the incorporation of
PWA-ZrO2 enhances the anti-deformation capacities and stiffness of membranes.
The decrease in strain-to-failure with increasing PWA-ZrO2 percentage might be
due to week interfacial strength between the composite particulate and membrane
matrix. Table 1 summarizes the mechanical properties measured from tensile
experiments, where the Young’s modulus and ultimate tensile strength
(UTS) increased with filler content, while the failure strain showed a monotonic
decrease.
Doping of ABPBI and its composite membranes with phosphoric acid up to 80%
by weight shows decrease in mechanical properties like elastic modulus and yield
strength due to swelling and plasticizing nature of phosphoric acid [22]. Due to
weakening of the intermolecular forces by inclusion of the acid within the mem-
brane structure and its swelling leads to mechanical deterioration [1]. However,
Fig. 5 Displacement contours along the loading axis (u) for ABPBI with 2 wt% PWA-ZrO2
Fig. 6 Stress versus strain

response of ABPBI and
PWA-ZrO2 composite ABPBI
membranes
impregnation of PWA-ZrO2 nano-particles results in increase of elastic modulus

and yield strength as shown in Fig. 7. Note here that experiments on acid-doped
pure ABPBI membranes were not possible as the pattern deposited externally on the
samples either before or after acid doping did not adhere well to the surface and
hence surface deformation using DIC could not be measured.
Table 1 Mechanical properties of ABPBI and ABPBI composites

Specimen Young’s modulus Ultimate tensile strength Strain-to-failure
(GPa) (MPa) (%)
ABPBI 1.703 ± 0.029 31.056 ± 6.898 11.2
ABPBI/PWAZrO2 (2%) 2.417 ± 0.083 39.922 ± 4.647 4.2
ABPBI/PWAZrO2 (5%) 3.001 ± 0.088 47.679 ± 3.985 3.6
PA-ABPBI/PWAZrO2 1.386 ± 0.070 26.484 ± 6.177 8.4
(2%)
PA-ABPBI/PWAZrO2 1.814 ± 0.052 36.916 ± 0.643 5.4
(5%)
Fig. 7 Stress versus strain

response of acid-doped
ABPBI and PWA-ZrO2
composite ABPBI membranes
4 Conclusion
From stress–strain curves, it can be concluded that incorporation of PWA-ZrO2

fillers, the mechanical properties like Young’s modulus and yields strength can be
enhanced significantly, whereas doping with phosphoric acid degrades the prop-
erties because of swelling and plasticizing nature of acid. Hence, mechanical
properties of polymer electrolytic membrane are crucial for overall performance of
fuel cell.
References
1. R. He, Q. Li, A. Bach, J.O. Jensen, N.J. Bjerrum, Physicochemical properties of phosphoric
acid doped polybenzimidazole membranes for fuel cells. J. Membr. Sci. 277(1), 38–45 (2006)
2. Q. Li, J.O. Jensen, R.F. Savinell, N.J. Bjerrum, High temperature proton exchange
membranes based on polybenzimidazoles for fuel cells. Prog. Polym. Sci. 34(5), 449–477
(2009)
3. S. Gottesfeld, J. Pafford, A new approach to the problem of carbon monoxide poisoning in
fuel cells operating at low temperatures. J. Electrochem. Soc. 135(10), 2651–2652 (1988)
4. C. Yang, P. Costamagna, S. Srinivasan, J. Benziger, A. Bocarsly, Approaches and technical
challenges to high temperature operation of proton exchange membrane fuel cells. J. Power
Sour. 103(1), 1–9 (2001)
5. J. Wainright, J.-T. Wang, D. Weng, R. Savinell, M. Litt, Acid-doped polybenzimidazoles: A
new polymer electrolyte. J. Electrochem. Soc. 142(7), L121–L123 (1995)
6. Y.-L. Ma, J. Wainright, M. Litt, R. Savinell, Conductivity of pbi membranes for
high-temperature polymer electrolyte fuel cells. J. Electrochem. Soc. 151(1), A8–A16 (2004)
7. R. Savinell, E. Yeager, D. Tryk, U. Landau, J. Wainright, D. Weng, K. Lux, M. Litt, C.
Rogers, A polymer electrolyte for operation at temperatures up to 200 °C. J. Electrochem.
Soc. 141(4), L46–L48 (1994)
8. J. Lobato, P. Canizares, M. Rodrigo, J. Linares, J. Aguilar, Improved polybenzimidazole films
for h3po4-doped pbi-based high temperature pemfc. J. Membr. Sci. 306(1), 47–55 (2007)
9. H. Zheng, M. Mathe, Enhanced conductivity and stability of composite membranes based on
poly (2,5-benzimidazole) and zirconium oxide nanoparticles for fuel cells. J. Power Sour. 196
(3), 894–898 (2011)
10. X. Huang, R. Solasi, Y. Zou, M. Feshler, K. Reifsnider, D. Condit, S. Burlatsky, T. Madden,
Mechanical endurance of polymer electrolyte membrane and pem fuel cell durability.
J. Polym. Sci., Part B: Polym. Phys. 44(16), 2346–2357 (2006)
11. E.A. Franceschini, H.R. Corti, Elastic properties of nafion, polybenzimidazole and poly [2,
5-benzimidazole] membranes determined by afm tip nano-indentation. J. Power Sour. 188(2),
379–386 (2009)
12. H. Zheng, L. Petrik, M. Mathe, Preparation and characterisation of porous poly (2,
5benzimida-zole)(abpbi) membranes using surfactants as templates for polymer electrolyte
membrane fuel cells. Int. J. Hydrogen Energy 35(8), 3745–3750 (2010)
13. J.J. Linares, C. Sanches, V.A. Paganin, E.R. Gonzalez, Poly (2, 5-bibenzimidazole)
membranes: physico-chemical characterization focused on fuel cell applications.
J. Electrochem. Soc. 159(7), F194–F202 (2012)
14. H.-J. Kim, S.Y. Cho, S.J. An, Y.C. Eun, J.-Y. Kim, H.-K. Yoon, H.-J. Kweon, K.H. Yew,
Synthesis of poly (2, 5-benzimidazole) for use as a fuel-cell membrane. Macromol. Rapid
Commun. 25(8), 894–897 (2004)
15. K. Wouters, P. Gijsenbergh, R. Puers, Comparison of methods for the mechanical
characterization of polymers for mems applications. J. Micromech. Microeng. 21(11),
115027 (2011)
16. C. Robin, A. Vishnoi, K.N. Jonnalagadda, Mechanical behavior and anisotropy of spin-coated
su-8thin films for mems. J. Microelectromech. Syst. 23(1), 168–180 (2014)
17. M.A. Naik, S. Samantaray, B. Mishra, Phosphotungstic acid nanoclusters grafted onto high
surface area hydrous zirconia as efficient heterogeneous catalyst for synthesis of octahydro-
quinazolinones and b-acetamido ketones. J. Cluster Sci. 22(2), 295–307 (2011)
18. I.M. Ward, J. Sweeney, Mechanical properties of solid polymers, (Wiley 2012)
19. A.D. Mulliken, Low to high strain rate deformation of amorphous polymers: experiments and
modeling, Ph.D. thesis, Massachusetts Institute of Technology, 2004
20. F.-P. Chiang, R.-M. Juang, Laser speckle interferometry for plate bending problems. Appl.
Opt. 15(9), 2199–2204 (1976)
21. M. Sutton, W. Wolters, W. Peters, W. Ranson, S. McNeill, Determination of displacements
using an improved digital correlation method. Image Vis. Comput. 1(3), 133–139 (1983)
22. A. Sannigrahi, D. Arunbabu, T. Jana, Thermoreversible gelation of polybenzimidazole in
phosphoric acid. Macromol. Rapid Commun. 27(22), 1962–1967 (2006)
Determination of Effectiveness
of Viscoelastic Dampers in Controlling
the Seismic Response of Piping System
by Shake Table Testing
R. K. Verma, P. N. Dubey and G. R. Reddy
Abstract Piping systems are considered as lifeline of the industrial units. The
earthquake load is one of the important design basis accidental loadings, considered
in the design of the nuclear power plant piping systems. To meet the contradictory
requirements of thermal and earthquake loads, snubbers are used. Snubbers allow
free thermal expansion in case of thermal loads and act as rigid support in the case
of earthquake load. However, the structure of snubber is complex, provides less
damping, expensive, and requires frequent maintenance. In addition to this,
mechanical snubber may pose locking problem and hydraulic snubber, if used, may
leak and may not cater the intended function when required. Leakages in hydraulic
snubbers and locking of mechanical snubbers have been observed in nuclear power
plants (NPPs) worldwide. In view of this, passive dampers are encouraged to
control the seismic response of piping in NPPs. Passive energy dissipation devices
absorb and dissipate a substantial portion of the energy imparted to the piping
system by earthquake shaking. Viscoelastic damper is a kind of passive energy
dissipating device. To determine the effectiveness of commercially available vis-
coelastic dampers in controlling the seismic response of piping systems, shake table
testing has been performed at CPRI, Bangalore. A 6″ NB schedule 40 piping loop
fitting to the 3 m 3 m size of CPRI shake table has been designed and fabricated.
Shake table testing of piping loop subjected to sinusoidal loading in the form of sine
sweep with a sweep rate of 1 octave per minute to get the response of the piping
with/without viscoelastic dampers has been carried out. Sweep was from 1 to
20 Hz, which was the range of interest as frequencies of the most of the piping
systems lie in this range. Piping loop was also subjected to response spectra with
0.15 and 0.45 g Zero Period Acceleration (ZPA) to get the response of the piping
due to earthquake loading. To determine the variation of stiffness and damping
coefficient, viscoelastic damper has also been tested at different frequencies. It has
been observed from the test results that the viscoelastic dampers are effective in
controlling the seismic response of piping systems.
R. K. Verma (&) P. N. Dubey G. R. Reddy

Reactor Safety Division, Bhabha Atomic Research Centre, Trombay,
e-mail: rajeev@barc.gov.in; rajeev.ballia@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_41
494 R. K. Verma et al.
Keywords Viscoelastic damper Snubber Shake table Sine-sweep test

Site-specific ground response spectrum
1 Introduction
Chemical and nuclear power plant piping systems are generally supported on spring
hangers, rod hangers, U-clamps, snubbers, and friction supports, etc. In normal oper-
ating condition, piping system is subjected to deadweight, pressure, and thermal loads.
On the other hand, during an earthquake, it will also be subjected to abnormal earth-
quake loads in addition to the normal loads. It is well known that the requirement of
piping characteristics for the thermal and the earthquake loads is not the same. In case of
thermal load, piping should be flexible enough to allow thermal expansion, and for the
earthquake loads, it should be rigid enough to limit the earthquake-induced displace-
ments. Therefore, it is a good design practice to have a balance between stiffness and
flexibility.
Conventional supports such as rod hangers increase the stiffness and reduce the
flexibility in vertical direction. These are preferable if thermal expansion of the
piping in the vertical direction is low. Spring hangers are generally called deadweight
supports and have low stiffness. These supports are simple to design and require
absolutely no maintenance. Snubbers are good for both the thermal and earthquake
loads. Snubbers allow free thermal expansion in case of thermal loads and act as rigid
support in the case of earthquake load. However, the structure of snubber is complex;
it provides less damping, is expensive, and requires frequent maintenance. In addi-
tion to this, mechanical snubber may pose locking problem and hydraulic snubber, if
used, may leak and may not cater the intended function when required [1].
In view of above, passive dampers are encouraged to control the seismic
response of piping in newer power plants. One such energy-absorbing device
discussed in the present work is viscoelastic damper.
2 Viscoelastic Damper
Viscoelastic dampers are devices for dissipating energy. They are used for reduction of
vibration in piping systems and together with helical steel springs in the vibration
isolation of massive industrial equipment such as presses and forging hammers.
Viscoelastic damper typically consists of a moving part immersed in highly viscous
fluid. The moving part is in the form of a rod (piston). The damper piston can move in all
directions, and damping forces are developed as a result of shearing action and dis-
placement in the fluid. Schematic of a viscoelastic damper has been shown in Fig. 1.
The dynamic characteristics of a viscoelastic damper depend primarily on the properties
of the viscous fluid and secondarily on the geometry of the device viz. shape of piston,
gap between piston and cylinder. The viscoelastic damper exhibits viscoelastic
behavior i.e., behavior that incorporates both elastic and viscous characteristics.
Furthermore, the properties of viscoelastic dampers are strongly frequency dependent
[2, 3].
Determination of Effectiveness of Viscoelastic Dampers … 495
Fig. 1 Schematic of a viscoelastic damper
3 Testing of Viscoelastic Damper
To characterize the commercially available viscoelastic dampers, tests have been

conducted at different frequencies to determine the variation of stiffness and
damping coefficient with frequencies. Figure 2 shows the force versus displacement
graphs of viscoelastic dampers at different frequencies. The stiffness and damping
Fig. 2 Force versus displacement at frequencies 1, 2, 3, and 4 Hz

Table 1 Variation of stiffness and damping coefficient with frequencies

Frequency (Hz) Stiffness (N/mm) Damping coefficient (Ns/mm)
1.00 212.18 54.46
1.50 271.18 42.29
2.00 311.60 36.89
3.00 393.69 28.52
3.50 433.43 27.29
4.00 485.68 23.91
coefficient at different frequencies has been calculated from force versus dis-
placement graphs. Variations in stiffness and damping coefficient with frequencies
have been listed in Table 1. It can be observed from Table 1 that the properties of
viscoelastic dampers are strongly frequency dependent.
4 Fractional-Derivative Maxwell Model for Viscoelastic

Damper
Makris and Constantinou [4] observed that the frequency dependency of the
mechanical properties of the high viscous dampers varied as frequency was raised
to fractional rather integer powers. This suggests that differentials of fractional order
could be used in modeling of such dampers.
Similar observations prompted Gemant [5] to first propose fractional-derivative
models for viscoelastic materials. The shear stress–strain relationship in the
fractional-derivative Maxwell model is
s þ kDr ½s ¼ l Dq ½c ð1Þ
in which s = shear stress; c = shear strain; l and k = generalized material con-

stants; Dr[f(t)] = fractional derivative of order r of the time-dependent function f.
For motion in the vertical direction, Makris and Constantinou observed that the
damper fluid is primarily subjected to shearing action. Accordingly, the force–
displacement relationship in vertical motion is expressed as
P þ kDr ½P ¼ C0 Dq ½u ð2Þ
where P = force; u = displacement; C0 = damping coefficient at zero frequency

that could be measured in an oscillatory test at very low frequency. By applying the
Fourier transform to Eq. 2, the relationship between force amplitude PðxÞ e and
displacement amplitude ~uðxÞ is given by
Table 2 Maxwell parameters Maxwell parameters Values obtained from the test
obtained from the test
C0 80,000
q 1
r 0.7
k 0.2
PðxÞ ¼ ½K1 ðxÞ þ iK2 ðxÞuðxÞ ð3Þ
where i = imaginary unit; and K1 and K2 are storage and loss stiffness, respectively.
qp rp qp
C0 xq cos 2 1 þ kx cos 2
r
þ C0 kxq þ r sin rp
2 sin 2
K1 ¼ rp ð4Þ
1 þ k x þ 2kx cos 2
2 2r r
qp rp qp

C0 xq sin 2 1 þ kx cos 2
r
C0 kxq þ r sin rp
2 cos 2
K2 ¼ rp ð5Þ
1 þ k x þ 2kx cos 2
2 2r r
K2 ðxÞ
C ðx Þ ¼ ð6Þ
x
Here, C ðxÞ is the damping coefficient.

The experimental data was used to estimate the parameters r and k of
fractional-derivative Maxwell model, described by Eq. 2. This was done by setting
the parameter q to unity and curve fitting of storage stiffness (K1) and damping
coefficient (C) of tested damper by fractional Maxwell model. The constant C0 was
found by projecting the damping coefficient curve to zero frequency. The values so
obtained are summarized in Table 2. Figure 3 shows the variation of stiffness and
damping coefficient with frequencies of viscoelastic damper.
Fig. 3 a Variation of damping coefficient versus frequency; b variation of stiffness versus

frequency
5 Shake Table Testing of Piping Loop With/Without

Viscoelastic Damper
To study the effectiveness of viscoelastic dampers in controlling the seismic

response of piping systems, shake table testing has been performed at CPRI,
Bangalore. A 6″ NB Schedule 40 piping loop has been designed to accommodate it
on the shake table of 3 m 3 m in size. Figure 4 shows the piping layout for shake
table testing at CPRI, Bangalore. Masses of 250 kg each have been attached at D
and E locations to bring the frequency down to spectral peak in order to attract
maximum seismic excitation. The fundamental frequency of the piping loop is
3.75 Hz obtained from the finite element analysis.
Testing of piping loop with/without viscoelastic dampers has been carried out.
Piping loop was subjected to a sinusoidal loading in the form of sine sweep with a
sweep rate of 1 octave per minute to get the response of the piping. Sweep was from
1 to 20 Hz, which was the range of interest as frequencies of the most of the piping
systems lie in this range. Acceleration time histories (responses) have been mea-
sured at B, C, I, and J. Figure 5 shows the response of the piping loop with/without
viscoelastic dampers for a sinusoidal loading of 0.2 g amplitude at C. The funda-
mental frequency of the piping loop obtained from the test is 3.64 Hz.
An artificial time history corresponding to Tarapur Atomic Power Station
(TAPS) site-specific ground response spectrum has been generated as shown in
Fig. 6. This time history has been used as an input to shake table for testing of the
Fig. 4 Piping loop for shake table testing at CPRI, Bangalore

Fig. 5 Response of piping loop for a sinusoidal loading of 0.2 g amplitude at C. a Without
dampers; b with dampers
Fig. 6 Input time history (0.15g PGA)
piping loop with/without viscoelastic dampers. Testing has been done for two levels
of earthquake, viz., 0.15 and 0.45g PGA, to get the response of the piping due to
earthquake loading.
Figures 7 and 8 shows the response of the piping loop with/without viscoelastic
dampers for an earthquake load corresponding to 0.15 and 0.45g PGA at C,
respectively.piping loop with/without visco
Fig. 7 Response of piping loop for an earthquake load corresponding to 0.15 g PGA at C.
a Without dampers; b with dampers
Fig. 8 Response of piping loop for an earthquake load corresponding to 0.45 g PGA at C.
a Without dampers; b with dampers
6 Conclusions
Shake table testing of a commercially available viscoelastic damper has been done
to study the effectiveness of the damper in controlling the seismic response of
piping systems. Based on the above study, the following conclusion can be made.
1. The properties (stiffness and damping coefficient) of viscoelastic dampers are
strongly frequency dependent.
2. The fractional-derivative Maxwell model may be used to model the viscoelastic
damper used in the present study.
3. Viscoelastic dampers are effective in controlling the response of the piping
systems subjected to sinusoidal and earthquake loadings.
References
1. J. Jonczyk, P. Gruner, Loads of piping systems due to malfunctions of snubbers. Nucl. Eng.
Des. 130, 411–433 (1991)
2. V.V. Kostarev, A.M. Berkovski, A.J. Schukin, in Upgrading of Dynamic Reliability and Life
Extension of Piping by Means of High Viscous Damper Technology. Conference on
Transactions of PVP ASME, Boston, (1999).
3. F.P. Barutzki, Extending the service life of piping systems through the application of viscous
fluid dampers. GERB Vib. Control Syst. Inc (2002).
4. N. Makris, M.C. Constantinou, Fractional-derivative Maxwell model for viscous dampers. J.
Struct. Eng., ASCE 117(9), 2708–2724 (1991)
5. A. Gemant, A method of analyzing experimental results obtained from Elasto viscous bodies.
Physics 7(8), 311–317 (1936)
Experimental Study of Deformation
Processes in Reinforced Concrete Beams
Strengthened with a Carbon Fiber Sheet
Anton Bykov, Igor Shardakov, Aleksey Shestakov and Irina Glot
Abstract In practical terms, repair and strengthening of reinforced concrete

structures with composite materials is an issue of great importance. However, how
this strengthening affects the strength of reinforced structures has not been ade-
quately justified. The specific features of the deformation behavior of strengthened
structures are mainly determined by the debonding of the composite from the
concrete surface. Experimental data describing the debonding process are practi-
cally absent. This paper presents the results of experimental studies of deformation
processes taking place in bending reinforced beams strengthened with carbon fiber
sheets. A thermography technique was used to detect the debonding in the exam-
ined structures. The results obtained made it possible to evaluate the effect of the
strengthening of carbon fiber sheets on the deformation process and bearing
capacity of reinforced concrete beams.
Keywords FRP RC beams Strengthening IC debonding Infrared

thermography
1 Introduction
External reinforcement methods of concrete bent elements with composite materials

are already widely used in the world [1–4]. It is well known that when loading a
concrete beam reinforced with a composite, delamination of the composite from the
concrete base occurs due to the opening of cracks [4–7]. However, the deformation
behavior of beams under real loads is insufficiently investigated. In practice, the
A. Bykov
Perm National Research Polytechnic University, Komsomolski Av., 29,
614990 Perm, Russian Federation
I. Shardakov (&) A. Shestakov I. Glot
Institute of Continuous Media Mechanics UB RAS, Korolev Street, 1, 614013
Perm, Russia
e-mail: shardakov@icmm.ru

https://doi.org/10.1007/978-981-10-7197-3_42
504 A. Bykov et al.
technology of restoration of the working capacity of beams is used, in which cracks

have formed during loading. These include injecting of cracks with special repair
materials and subsequent gluing with a carbon fiber sheet. But there are almost no
experimental data describing the features of the deformation process, accompanied by
delamination of the composite in the reinforced concrete beams, which have been
restored in unloaded condition or directly under load. Works [8–11] are devoted to the
issues concerning the use of the infrared thermography technique in order to detect
defects in the area of the debonding of the external reinforcement from the concrete
surface. However, the use of the infrared thermography technique to register
delaminations that occur by a cohesive scenario (with a violation of the integrity of the
surface layer of concrete) in these works is not mathematically justified.
This paper presents the results of a study of deformation processes in reinforced
concrete beams strengthened with CFRP before load application and during loading
after the appearance of the first cracks and their injection. To study temperature
distribution on the beam surface, the infrared thermography technique was used.
A developed mathematical model is demonstrated to describe the temperature
changes in the system of “a carbon fiber sheet–epoxy resin–concrete–delamination–
concrete,” which simulates the concrete base with a glued composite and voids in
the surface layer of concrete. Based on this model, numerical experiments were
performed which allowed us to determine the rational parameters of the infrared
photography to register delamination of carbon fiber layer from the concrete base.
2 Testing Samples
Twenty-two rectangular cross-sectional sample beams of width 120 mm, height

220 mm, and a total length of 1290 mm were prepared for the study. The samples
were divided into two groups: group B1 of concrete strength class B20 and group
B2 of concrete strength class B35. In our investigation, the following test series
were performed: the “a” series with non-strengthened beams (reference samples),
the “b” series with beams strengthened in the unloaded state prior to their instal-
lation on the test bench (Fig. 1a), and the “c” series with beams strengthened under
loading (Fig. 1b).
It should be noted that the test of the “c” series samples was held in two stages: stage
1—loading of the unstrengthened beam until the appearance of cracks and their
opening to 1 mm in width; stage 2—additional loading of the beam after repair and
strengthening until its destruction. Repairing of the cracks and strengthening of the
beam with a carbon fiber sheet, without removing the load from it, took place between
the first and second stages. When repairing the cracks, they were widened, grouted
with a repair compound, and injected with a low-viscosity epoxy resin, as shown in
Fig. 2. As the repair compound, a mixture of 0.5 mm sand fraction and the epoxy
adhesive Sikadur-330 in the ratio of 1:1 were used. A carbon fiber sheet
SikaWrap-230 and the epoxy resin Sikadur-330 were used for strengthening purpose.
The reinforcement scheme of the beams with carbon fiber and steel is shown in Fig. 3.
Experimental Study of Deformation Processes in Reinforced … 505
Fig. 1 a Bonding of the sheet onto the “b” series beams; b bonding of the sheet onto the “c”
series beams
Fig. 2 Epoxy crack injection
Fig. 3 Reinforcement scheme of beams with steel and a carbon fiber sheet
506 A. Bykov et al.
3 The Modeling of Heat Transfer in the System

of “Carbon Sheet–Epoxy Resin–Concrete–
Delamination–Concrete”
To simulate the heat transfer processes in the beam covered with a strengthening
layer, an initial boundary value problem was set, describing one-dimensional
non-stationary temperature field in the system of “a carbon sheet–epoxy resin–
concrete–delamination–concrete,” which takes into account the specific of
debonding the composite layer with a concrete fragment (cohesive scenario). The
design scheme of the beam cross section is shown in Fig. 4. The mathematical
formulation of the problem is described by a heat conduction equation (1),
boundary (2), and initial (3) conditions.
@Ti ki @ 2 Ti
¼ ; i 2 ð1; 5Þ; x 2 ðli1 ; li Þ; l0 ¼ 0; ð1Þ
@t ci qi @x2

@T1 Q; t ¼ th @Ti @Ti þ 1 @T5
k1 ¼ ; ki ¼ ki ; Ti ¼ Ti þ 1 ; i 2 ð1; 4Þ; x ¼ li ; k5 ¼ 0; x ¼ l5 ;
@x 0; t ¼ t [ th @x @x @x
ð2Þ
Ti ðx; 0Þ ¼ 0; i ¼ ð1; 5Þ; x 2 ðli1 ; li Þ; ð3Þ
where the subscript i denotes the number of the layer, and in accordance with
Fig. 4, T ¼ Tðx; tÞ—temperature; t—time; x—spatial coordinate; c—specific heat;
p—density; and k—thermal conductivity coefficient.
It is possible to achieve high accuracy in simulation of temperature changes in
the given multilayer system if we know the actual thermal characteristics of its
materials. To determine the thermophysical parameters of the materials of the
Fig. 4 Design diagram of the

problem for the beam cross
section: 1 carbon sheet, 2
epoxy resin, 3 delaminated
concrete, 4 delamination (air
gap), 5 concrete base
Fig. 5 Evolution of the

temperature increment on the
concrete surface during
heating and cooling
system “a carbon sheet–epoxy resin–concrete–delamination–concrete,” a prelimi-

nary experiment was performed. Using an external heat source of the given
capacity, the surface of the composite bonded onto the concrete was heated. After
exposure for a certain time, the source was removed, then the multilayer system was
cooled, and non-stationary temperature distribution occurred. Temperature changes
on the CFRP surface were recorded using an infrared imager. As a result, the time
dependence of the temperature on the CFRP surface was obtained. A similar
dependence was obtained in the numerical experiment. Then, such thermal
parameters of the elements were chosen, which have provided a satisfactory
agreement between the results of calculation and experiment.
Figures 5 and 6 display graphs of the temperature changes on the surface of the
entire concrete and of the multilayer system during heating and cooling, respec-
tively. Each figure shows two graphs: The first one was obtained experimentally
and the second one using the mathematical model. Table 1 shows the thermal
parameters of the system “a carbon sheet–epoxy resin–concrete–delamination–
concrete.” The values chosen during the preliminary experiment are marked with *.
The other parameters, listed in the table, correspond to the reference values. The
heat source capacity and heating duration, allowing to register delamination of
0.05 mm thickness at depth of 1 mm, were also determined in the simulation. They
are equal to 926 W/m2 and 9 s, respectively.
The subsequent numerical experiments let us establish time dependence of the
temperature increment on the sample surface during heating and cooling for the
samples having delamination and without it (Fig. 7). It was found that the evolution
of the temperature increments of the entire sample and samples with delamination is
different. We denote the difference between the temperature increments of the entire
sample and the sample with a defect corresponding to a certain point of time as a
temperature signal. This quantity is an informative parameter, which gives an
indication of the presence of delamination. Numerical simulations have shown that
508 A. Bykov et al.
Fig. 6 Evolution of the temperature increment on the surface of the multilayer system during
heating and cooling
Table 1 Parameters of the system elements

System elements Specific heat Thermal Density Thickness
J/(kg grad) conductivity (kg m3) (mm)
W/(m grad)
Carbon sheet (1) 720 256 1600 0.1
Epoxy resin (2) 1300* 0.17* 1100 0.9
Delaminated concrete (3) 710* 0.63* 2400 1**
Delamination–air gap (4) 1000 0.026 1225 0.05**
Concrete base (5) 710* 0.63* 2400 13
the value of the temperature signal is not constant: It is greater in the cooling step
than in the heating one, and it reaches its maximum at one point tm (Fig. 8). As
followed from the numerical simulation, for the tested samples, the temperature
signal must be recorded during the cooling step after heating for 9 s. The maximum
temperature signal is recorded at the 19th second from the beginning of observation
(or in the 10th second from the beginning of cooling).
The maximum temperature signal characterizes the sensitivity of this method to
the presence of a delamination defect. Figure 9 in the coordinates “thickness of the
defect–occurrence depth of the defect” shows a family of lines, separating the area
of delamination sizes that may be detected using the infrared shooting from the field
of “invisible” delaminations. For every duration of heat impulse, the detectable
delamination area lies below the corresponding curve as shown in Fig. 9.
As shown by numerical experiments, as the heat time increases, the area of
visible defects becomes greater. However, when observing in the 19th second after
Fig. 7 Change in the

temperature increment on the
surface of carbon fiber during
heating and cooling: 1
without delamination, 2 with
delamination of thickness
0.05 mm, a depth of
occurrence of 1 mm
Fig. 8 Change in the

difference of the increment in
the temperatures shown in
Fig. 4
Fig. 9 Curves that determine

the areas of possible
delamination registration for
different heating duration
510 A. Bykov et al.
the start of heating, heating longer than 15 s does not lead to further expansion of
the field of defects visible for thermography. Therefore, the heating, characterized
by parameters determined by the results of the mathematical modeling, allows to
register delaminations more than 1 mm thick, lying at a depth less than 2 mm under
the concrete surface.
4 The Technique of Testing Beams
The tests were carried out on a specially designed and manufactured test bench, as
shown in Fig. 10a, according to the scheme of a four-point bending. When testing
the “c” series, the beam was fixed in a bent shape at the time of repair works,
including the injection of the cracks and solidification of the CFRP layer (about
10 days). Fixation was performed using a special telescopic strut which provided a
constant bending moment for a long time.
The loading of the beams was performed by a successive increasing quasistatic
load with a step of 2 kN, representing 4–6% of the breaking load. At each stage of
loading, a 5–10-min stop was done, during which the pattern of cracks was
recorded, as well as their opening width. During the stop, the stretched surface of
the beam was subjected to heating of 926 W power and 9-s duration and subsequent
Fig. 10 a Test bench: 1 the bottom beam of the test bench, 2 the top beam of the test bench, 3
twin rack, 4 rack, 5 mowing, 6 a sample beam, 7 hinged movable support, 8 hinged immovable
support, 9 hydraulic jack, 10 distribution bar, 11 loadmeter; b the scheme of the infrared
photography: 12 the thermal imager FLIR T620, 13 halogen lamp, 14 mirror
cooling. The scheme of the equipment arrangement in the infrared photography is

shown in Fig. 10b. Thermographic measurements were supplemented with a set of
mechanical measurements. At each load step, strains in specific sections of the
beam were measured by strain gauges, deflections of the beam were measured using
displacement sensors, and cracks on the beam surface were recorded.
The main provisions of thermogram recording technique were as follows:
• Along the length of the beam, five locations of the infrared imager mounted on a
tripod were fixed, so that five thermograms, sequentially lined up, would give a
panoramic picture of the carbon fiber strip.
• Thermal markers, necessary for accurate combining of individual thermograms
in a panoramic image, were attached to the lower surface of the beam.
• The shooting was made through the reflection in the mirror; thus, it ensured
safety of the expensive equipment when shooting on the steps close to the beam
destruction.
• Prior to loading, the beams performed the infrared shooting, which gave the
image of the beam without delamination.
• The heating of the carbon fiber strip was carried out with a halogen lamp. When
turned off, the lamp was put aside to ensure uniform cooling of the beam
surface;
• The heating time and the observation time were monitored with the help of a
stopwatch with an accuracy of 0.5 s.
The primary thermograms of the beam surface (Fig. 11a) were obtained during
the tests, which were then transformed by a specially developed algorithm and
implemented in the MATLAB package. Let T ðx; yÞ denote the primary temperature
which for an arbitrary point ðx; yÞ on the CFRP surface sets a difference between the
temperatures at the initial moment and in the 19th second after the beginning of
observation. The so-called normalized thermograms were calculated on the first step
of the algorithm (Fig. 11b):
Fig. 11 a Primary thermograms; b normalized thermograms; c temperature contrast maps;

d binary defect maps
512 A. Bykov et al.
T0 ðx ; y Þ
Tjn ðx; yÞ ¼ Tj ðx; yÞ ; ð4Þ
Tj ð x ; y Þ
where Tjn ðx; yÞ and Tj ðx; yÞ are the normalized and primary temperatures at the jth
stage of loading at the point with coordinates ðx; yÞ; T0 ðx ; y Þ and Tj ðx ; y Þ are the
primary temperatures at the initial stage and jth stage of loading at the point ðx ; y Þ
where there is no delamination until the destruction of the beam.
Then, the values of the current temperature contrast were calculated Cj ðx; yÞ
(Fig. 11c), reflecting the difference of the values of the normalized temperature at
arbitrary point at the jth stage of loading Tjn ðx; yÞ and the primary temperature at the
same point T0 ðx; yÞ at the initial stage:
Tjn ðx; yÞ T0 ðx; yÞ

Cj ðx; yÞ ¼ 100%: ð5Þ
T0 ðx; yÞ
The increased values of the current temperature contrast indicate the presence of
delamination in the observed area of the beam. The magnitude of the threshold
values of the temperature contrast C , separating the defect-free regions from the
defect ones, can be obtained by estimating the average contrast value C j and the
standard deviation rj in those areas of the thermograms, where the presence of
defects is excluded. The estimation of these values, obtained according to the data
of the tests of all the tested beams with a reinforcing layer, made it possible to find
the following magnitude of the threshold value: C ¼ Cj þ 3rj ¼ 27:9%. The areas,
where the current temperature contrast value does not exceed the threshold value,
were identified as defect-free and the rest as areas with delamination. Figure 11d
shows the binary defect maps corresponding to successive steps of loading.
5 Results
The “a” series samples were destroyed due to rupture of the tensile reinforcement
and the fragmentation of the compression concrete area. The “b” and “c” series
samples were destroyed due to rupture of the carbon fiber sheet in the middle part of
the span. The average values of the limiting bending moment of the “a” series
beams are 6.83 and 7.12 kNm for B1 and B2 groups, respectively. For the rein-
forced series “b” beams, the limiting bending moment equals to 10.47 and
10.76 kNm for B1 and B2 groups, respectively. For the reinforced series “c”
beams, the limiting bending moment is 10.29 and 11.00 kNm for B1 and B2
groups, respectively.
Figure 12 shows the dependence of the maximum deflection of the beam on the
bending moment obtained when testing the three series of beams: beams without
Fig. 12 a Bending moment–deflection relationships for group B1; b bending moment–deflection

relationships for group B2
any reinforcement layer (the “a” series); beams that preliminary reinforced with a
carbon fiber sheet (the “b” series); and beams which underwent the reinforcing
procedure while testing (the “c” series). The comparison of the graphs obtained in
the “a” and “c” series clearly demonstrates an increase in the bearing capacity of the
beams that have undergone the preliminary reinforcement with the CFRP sheet. The
maximum bending moment, which such beams can stand, has happened to be
37–39% higher than the reference samples. The graphs clearly depict the formation
stage of the first cracks in the concrete: It corresponds to a sharp change in the slope
angle of the curves.
It is found experimentally that on average, the first crack in reinforced beams
appears when deformation in the carbon fiber reaches 170–250 lm/m. After the
appearance of the first crack, the deformations of the carbon fiber rapidly increase to
1000–2500 lm/m depending on the distance of the strain gauge from the crack.
Thus, after the appearance of the first crack, the compatibility of deformations of the
concrete base and carbon fiber sheet is violated. A certain area appears where the
concrete can no longer perceive tensile deformation and gets destroyed, and the
carbon fiber is deformed without breaking. This area, in fact, is the area of
delamination, and the moment of reaching limit deformations in the concrete is the
onset of delamination. However, delamination, associated with the appearance of
the first crack, is not visible in the infrared photography. Delamination that is
noticeable in the infrared range occurs after crack opening up to 0.3–0.5 mm. For
the “b” series beams, delamination develops in the area between the principal main
cracks, and for the “b” series beams, it develops between the new crack of the
second generation and the old crack of the first generation (the dashed line), as
shown in Fig. 13.
The set of curves showing dependencies of carbon fiber sheet deformations on
the value of the relative area of delamination is given in Fig. 14.
The presented dependences are close to linear ones. When the limit state of the
beam is achieved, the area of delamination in the preliminary reinforced beams (the
514 A. Bykov et al.
Fig. 13 a A pattern of cracks and a binary defect map for «b» series; b a pattern of cracks and a
binary defect map for «c» series
Fig. 14 Dependence of carbon fiber sheet deformations on the value of the relative area of
delamination
“b”-series) is in 2.1–2.3 times less than in the reinforced under load beams (the “c”
series), at the same relative deformation of the carbon fiber.
6 Conclusions
A mathematical model was developed to describe the temperature changes in the

system of “a carbon sheet–epoxy resin–concrete–delamination–concrete”, which
takes into account the characteristic features of cohesive delamination of the
composite from the concrete base. This model allowed us to determine the
parameters of the infrared shooting to register delamination of the carbon fiber of
thickness 1 mm from the concrete base as follows: heating power—926 W/m2;
heating duration—9 s; and the optimal observation moment–at the cooling stage
after the heating in the 19th second after the start of observation. Infrared shooting
in this mode makes it possible to detect delaminations of a thickness exceeding
1 mm, which lie at a depth down to 2 mm.
It is shown that for the composite strengthened beams, which are destroyed due
to the rupture of CFRP, their bearing capacity does not depend on the point when
strengthening was performed. The bearing capacity of previously reinforced beams

and beams reinforced during loading after the appearance of the first cracks and
their grouting are higher by 37–39 and 38–49%, respectively, than the unreinforced
beams.
It was found that for the beams reinforced during loading, the relative area of
delamination increases in 2.1–2.3 times in comparison with the beams reinforced
before load application.
Acknowledgements This research was supported by the Russian Science Foundation, project
No. 14-29-00172.
References
1. J.G. Teng, Failure modes of FRP-strengthened restructures, in 26th Conference on Our World
in Concrete & Structures: 27–28 Aug 2001, Singapore, pp. 627–634
2. A.A. Bykov, A.V. Kalugin, New model for evaluation of FRP debonding strain for Russian
design code. Adv. Mater. Sci. Eng. 2013. http://dx.doi.org/10.1155/2013/130162
3. M.R. Esfahani, Flexural behaviour of reinforced concrete beams strengthened by CFRP
sheets. Eng. Struct. 29(10), 2428–2444 (2007)
4. H. Saadatmanesh, RC beams strengthened with GFRP plates. I: experimental study. J. Struct.
Eng. 117(11), 3417–3433 (1991)
5. J. Yao, J.G. Teng, J.F. Chen, Experimental study on FRP-to-concrete bonded joints. Compos.
B Eng. 36(2), 99–113 (2005)
6. S.K. Woo, J.J. Kim, K. Byun, Y. Song, Bond-slip parameter determination procedure of RC
flexure member strengthened with prestressed CFRP plates. KSCE J. Civ. Eng. 17(1), 179–
191 (2013)
7. R.J. Gravina, S.A. Hadigheh, S. Setunge, Bond and force transfer of FRP materials bonded to
concrete using sitecure system, in Proceedings of the 3rd Asia-Pacific Conference on FRP in
Structures, Hokkaido, 2012
8. J. Tashan, R. Al-Mahaidi, Detection of cracks in concrete strengthened with CFRP systems.
Compos. B Eng. 64, 116–125 (2014)
9. J.R. Brown, H.R. Hamilton, Quantitative infrared thermography inspection for FRP applied to
concrete using single pixel analysis. Constr. Build. Mater. 39, 1292–1302 (2013)
10. C.W. Hu, J.K.C. Shih, R. Delpak, D.B. Tann, Detection of air blisters and crack propagation
in FRP strengthened concrete elements using infrared thermography, in Proceedings of the
3rd Annual Thermographers’ Conference «InfraMation 2002», Orlando, 2002
11. S.A. Keo, F. Bracheleta, F. Breabana, D. Defera, Defect detection in CFRP by infrared
thermography with CO2 laser excitation compared to conventional lock-in infrared
thermography. Compos. B Eng. 69, 1–5 (2015)
Partial Delamination Detection
and Quantification in Composite
Laminates Using Laser Doppler
Vibrometer
Yashdeep P. Nimje and Gangadharan Raju
Abstract Vibration-based structural health monitoring techniques are widely used

for detection and quantification of delamination in composite structures. In this
work, non-contact vibration measurement technique namely laser Doppler
vibrometer (LDV) is used for damage characterization of partial delamination in
composite beams. The modal information such as natural frequencies and mode
shapes of composite beam are obtained using LDV and validated with finite ele-
ment models for comparison with the obtained experimental results. Mode
shape-based damage detection algorithms such as fractal dimension approach are
used for quantification of delamination in composite beams. In the present work, a
modified generalized fractal dimension technique is proposed to locate delamina-
tion of different sizes and shapes in the composite beam. Different types of partial
delamination like C-type, I-type in composite beam specimens were fabricated and
experimentally studied using LDV. The modified fractal dimension technique was
then applied to the mode shapes results of the various partial delaminations to
quantify the damage.
Keywords Laser Doppler vibrometer Delamination Composites laminates

Fractal dimension technique
1 Introduction
Over the last two decades, there is a growing need for real-time monitoring of
structures made of composite materials to avoid catastrophic failures and provide
confidence on the usage of these high-performance materials into aerospace, marine
Y. P. Nimje (&) G. Raju

Department of Mechanical & Aerospace Engineering, Indian Institute of Technology
Hyderabad, Kandi 502205, Telangana, India
e-mail: me14mtech11038@iith.ac.in
G. Raju
e-mail: gangadharanr@iith.ac.in

https://doi.org/10.1007/978-981-10-7197-3_43
518 Y. P. Nimje and G. Raju
and wind energy sectors. Unlike metals, failure mechanisms are quite complex in
composites involving fibre breakage, fibre pull-out, matrix cracks, fibre-matrix
debonding and delamination. Among the various failure modes, delamination is a
barely visible damage induced by transverse impact and results in significant
reduction of the strength and stiffness of the composite. Delamination can alter the
dynamic responses such as natural frequencies, mode shapes, and these changes can
be extracted to estimate damages in the structure by modal analysis [1, 2].
Vibration-based structural health monitoring (SHM) is one of the major
approaches for inspection and structural integrity evaluation. Reliable modal-based
methods are essential to detect and quantify damages for implementation of robust
SHM systems. The challenges are to identify the changes in the physical parameters
due to delamination type of damage and correlate them with the corresponding
measured modal parameters [3]. The present research work focuses on developing
the relationship between the dynamic response of healthy and damaged composite
specimens for damage identification and quantification. Vibration-based SHM has
gradually moved from accelerometers for sensing to embedded piezo sensors, fibre
optic sensors and non-contact measurements like laser Doppler vibrometer
(LDV) to capture the modal response of structures. Lestari et al. [4] proposed a
combined analytical and experimental modal approach to locate delamination in
composite beams using curvature mode shapes. They used surface bonded piezo-
electric wafers to capture the mode shape of composite structures. Qiao et al. [5]
used LDV for damage detection in composite plates using modal curvature and
developed algorithms based on global smoothing method fractal dimension tech-
nique to locate the damage. Shang et al. employed LDV to perform modal testing of
composites and later proposed damage detection based on subset selection tech-
nique to locate the delamination. In this work, LDV measurement system is used to
capture the mode shapes of the various delaminated composite structures. The
acquired mode shapes were then analysed using the modified fractal dimension
technique for locating the delamination and quantification of their size.
Furthermore, numerical finite element analysis of partial delaminated composite
structures is carried out to validate the damage detection algorithms.
2 Finite Element Modelling of Composites
Numerous works have been reported on the vibration studies on delaminated

composite structures. Della and Shu [6] have presented a comprehensive review of
the various analytical and numerical approaches on the vibration of delaminated
composites. Most of the works rely on finite element modelling for modal analysis of
delaminated composites [1, 2, 7–9]. Finite element analysis was carried out using
commercial software ANSYS 12 to model the healthy and delaminated composite
structures in order to extract their vibration behaviour, i.e. natural frequency and
mode shapes. A three dimensional 8 noded ANSYS element SOLID185 is used to
model the composite beam. To model the delamination using finite elements, the
Partial Delamination Detection and Quantification in Composite … 519
approach proposed by Ihesiulor et al. is used. In this approach, two volumes were
created above and below the delamination and meshed using brick elements. Next,
the nodes situated along the interface of healthy segments were merged together by
introducing contact elements, while nodes in the interface of delaminated region
were not merged. Contact elements (TARGET170/CONTACT173) were introduced
between the surfaces to prevent separation and inter-penetration. Reduced integra-
tion was carried out to avoid the shear locking problem when analysing thin
structures. A mesh sensitivity study was conducted to determine the number of
element to be employed for healthy as well as delaminated beam. For each laminae,
finite element mesh density of 100, 20 and 1 elements are chosen along the length,
width and thickness direction, respectively. Block Lancoz method was then used to
extract the modal parameters. Torsion and in-plane bending mode were discarded,
and only, flexural bending modes were used in this study. Edge or partial delami-
nation is a serious concern in helicopter blades and turbine blades. In the present
work, the objective was to numerically investigate the effect of partial delamination
on the modal characteristics of the composite structure. The schematic diagram of
the various types of delamination numerically studied using finite element analysis is
shown in Fig. 1.
For validation purpose, the numerical examples of partial delaminated composite
beam given by Kumar et al. [7, 10] are investigated. For the numerical study,
graphite/epoxy composite specimen with stacking sequence [0/90]2s and dimen-
sions given by length = 0.127 m, width = 0.0127 m and thickness = 0.254 mm
was chosen. Material properties of the laminae are given by E11 = 134 GPa,
E22 = 10.3 GPa, E33 = 10.3 GPa, G12 = 5 GPa, G23 = 3.59 GPa, G31 = 5 GPa,
c12 = 0.33, c23 = 0.37, c13 = 0.33, density = 1480 kg/m3. Finite element mod-
elling of C-type and I-type delamination was carried out in ANSYS by varying the
size of delamination in width-wise direction and keeping the length to be constant
of 50.8 mm. The nodes associated along the interface of healthy region are merged
together with the help of contact elements (TARGET170/CONTACT173), and the
nodes in delaminated region were set free.
From Table 1, it is inferred that natural frequency results obtained using FEA
closely match with the results given by Kumar et al. [7, 8]. Next, the mode shape
information is further studied using signal processing algorithms for location of the
delamination in composite structure.
Fig. 1 Schematic diagram of beam having delamination a through width delamination, b multiple
delamination, c C-type delamination, d I-type delamination
Table 1 Channel-type delamination at first interface having 50.8 mm lengthwise delamination

and I-type delamination at first interface of 76.2 mm lengthwise delamination
Percentage of width-wise delamination First layer C-type First layer I-type
(%) delamination, natural delamination, natural
frequency (Hz) frequency (Hz)
Kumar FEA Kumar FEA
results results
0 81.46 81.40 81.46 81.40
30 81.32 80.93 81.51 81.34
50 80.15 80.19 81.12 81.19
70 78.01 79.09 80.31 80.94
90 76.12 78.99 80.48 80.69
100 76.04 75.43 68.24 64.69
3 Damage Detection Method—Fractal Dimension

Technique
Various type of damage detection techniques are available such as damage index
method [3], gapped smoothening method [3] which require the information of
healthy structures for damage detection. Generalized fractal dimension technique is
another widely used approach which does not require healthy structure information
for damage location. Fractal dimension technique was proposed by Hadjileontiadis
and Douka [11] and used for locating the crack in beams. Later, it was modified by
Qiao et al. [5] and applied to locate delamination in composite plates. Delamination
in the structure will affect the local stiffness of the structure, which in turn will
introduce small changes in their mode shapes. Evaluating fractal dimension on
mode shapes data will enhance the small localized change and will help in detecting
and locating the delamination. Generalized fractal dimension was defined by Qiao
et al. [5] and it is given by:
log ðnÞ
GFD ¼ 10 ð1aÞ
log10 dlss þ log10 ðnÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ds ¼ ðyi þ k yi Þ2 þ S2 ðxi þ k xi Þ2 ð1bÞ
M qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X
ls ¼ ðyi þ j yi Þ2 þ S2 ðxi þ j xi Þ2 ð1cÞ
j¼1
where n is the number of nodes used in a single window; x and y are the system
coordinates, and S is the scale factor which needs to be chosen carefully in order to
detect the delamination successfully. In Eq. (1b), ds is the maximum distance in the
curve; ls is the total length of the curve; k is the size of the window. However, Qiao
et al. [5] did not give a proper explanation on choosing the scaling factor S for
various mode shapes. In this work, the scaling factor was removed to compute the
fractal dimension and a modified expression is given by:
log ðnÞ
MGFD ¼ 10 ð2aÞ
log10 DLss þ log10 ðnÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ds ¼ ðDyi þ k Dyi Þ2 þ ðDxi þ k xi Þ2 ð2bÞ
M qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X
Ls ¼ ðDyi þ j Dyi Þ2 þ ðDxi þ j Dxi Þ2 ð2cÞ
j¼1
where in Eq. (2b), Dxi is the normalized nodal displacement with respect to max-
imum nodal displacement in x-direction; Dyi is the normalized nodal displacement
with respect to maximum nodal displacements in y-direction. M is the number of
nodes present in the single window. Ds is the difference in the normalized nodal
displacement between the first and last node of the chosen window; k is size of the
window; x is the out of plane deformation; y is vertical distance of the node. To
locate the delamination and compute the size of the delamination, GFD and MGFD
were applied to mode shapes acquired from ANSYS modelling. Firstly, through
width, delamination case taken for the study and the details of the specimen are
shown in Fig. 2a. GFD and MGFD were applied to the first mode shape of the
(a) L = 24mm
W =50mm
L = 220mm
(b) (c)
Fig. 2 a Schematic diagram of beam having delamination at the centre, b first mode by using
fractal dimension, c first mode by using modified fractal dimension technique
composite beam, and the results are shown in Fig. 2b, c, respectively. Both the
approaches are effectively capturing the delamination location and size as shown in
Fig. 2.
Subsequently, both the approaches were applied to study the delamination at the
tip of the beam. The results of GFD and MGFD are presented in Fig. 3. GFD
approach is not able to identify the delamination location at the tip of the beam,
whereas MGFD can capture the delamination location and size more accurately.
Next, the minimum size of the delamination that can be identified by both the
approaches was investigated. Figure 4 shows that MGFD can capture delamination
of size 2.2 mm, whereas GFD fails to locate it.
GFD and MGFD were then applied to study the multiple delamination case, and
the results are presented for the first mode shape in Fig. 5. GFD was able to locate
the delamination near the fixed end, but was not able to capture the delamination at
the tip (Fig. 5b). MGFD was able to capture the delamination near the fixed end and
the tip of the beam. Both the approaches were then applied to the fourth mode
shape, and the results are shown in Fig. 6.
MGFD overcomes all the above limitations of GFD, and the results in this
section show that it can capture location and size of delamination more accurately.
Thus, the proposed MGFD was then applied to study the experimental mode shapes
obtained by LDV in order to identify the delamination.
(a) L = 174mm L =24mm
W = 50mm
L = 220mm
(b) (c)
Fig. 3 a Schematic diagram of beam having delamination at the tip, b first mode by using fractal
dimension, c first mode by using modified fractal dimension technique
(a)
2.2mm
W= 50mm
L =220mm
(b) (c)
Fig. 4 a Schematic diagram of beam having delamination at the centre, b first mode by using
fractal dimension, c first mode by using modified fractal dimension technique
(a)
L =24mm L =24mm
W = 50mm
L = 220mm
(b) (c)
Fig. 5 a Schematic diagram of beam having multiple delamination, b first mode by using fractal
dimension, c first mode by using modified fractal dimension technique
Fig. 6 a Fourth mode by using fractal dimension, b fourth mode by using modified fractal
dimension technique
4 Experiment Setup
Laser Doppler vibrometer (LDV) is a non-contact vibration measurement technique

capable of measuring the movements of the scanning points in Z-direction, i.e. out
of plane displacement. The system works on the principle of Doppler effect and
interferometry for vibration measurement. The schematic diagram of the experi-
mental setup to capture the modal information of composite beams using LDV is
shown in Fig. 7.
A periodic chirp signal in the frequency range of 10 Hz–50 kHz is used for the
excitation of the composite beam. To improve the reflective ability of the composite
specimen, a thin layer of reflective white spray paint was used. The beam was
excited by mounting on an electronic shaker as shown in Fig. 7. In this study, glass
fibre reinforced plastic (GFRP) laminates were chosen for experimental study.
The GFRP specimens were fabricated using hand layup and vacuum bag technique.
For the creation of delamination between layers, Teflon tape of thickness 0.09 mm
was inserted between the laminas.
Fig. 7 Laser Doppler vibrometer

Table 2 Experimental and numerical results of various delaminated GFRP specimens

Mode Healthy Full-width C-type I-type Multiple
No. specimen (Hz) delamination delamination delamination delamination (Hz)
(Hz) (Hz) (Hz)
FEA Exp FEA Exp FEA Exp FEA Exp FEA Exp
First 16.84 17.18 16.82 16.60 16.83 16.60 16.83 16.06 16.79 15.625
Second 105.47 107.81 105.43 105.46 105.45 105.46 105.45 105.40 102.96 104.50
Third 295.37 287.57 284.96 279.06 291.58 283.06 291.68 – 291.62 270.50
Fourth 578.78 582.50 573.74 580.89 577.92 581.08 577.84 – 564.31 577.80
5 Experimental Results for GFRP Composite Specimen
Experimental study was carried out on the unidirectional GFRP laminate [0]4
specimen having dimension of 220 mm 50 mm 1.16 mm. The material
properties of GFRP laminae were obtained by ASTM standard tests and are given
by E11 = 42.1 GPa, E22 = E33 = 10.3 GPa, G12 = 3.84 GPa, G23 = 3.97 GPa,
G31 = 3.84 GPa, c12 = 0.27, c23 = 0.37, c31 = 0.27 and density = 2100 kg/m3. The
C-type and I-type delamination composite specimens were fabricated as per the
information given in the reference [7, 8]. Experiments were carried out on the
various delaminated specimens using LDV, and the natural frequencies obtained are
given in Table 2.
The results in the table show that both the experimental and finite element mod-
elling results of the delaminated GFRP specimen are close to each other.
Subsequently, MGFD was applied to the first mode shape of the delaminated speci-
men, and the results are given in Fig. 8. Thus, MGFD technique can be effectively
used for locating and quantifying the partial delamination in GFRP specimens.
Fig. 8 a Full-width delamination specimen, b multiple delamination specimen, c C-type

delamination specimen, d I-type delamination specimen
6 Conclusion
In this work, we investigated experimentally and numerically the modal properties

of various partially delaminated specimens. Finite element modelling of partial
delaminated specimen was successfully carried out to validate the results available
in the literature. Modified fractal dimension technique was proposed and applied to
various delamination cases. MGFD and GFD techniques were then applied to both
numerical and experimental mode shapes of various delaminated specimens.
MGFD technique was accurate in locating and estimating the actual size of the
partial delamination in GFRP specimens. In future, we want to extend the MGFD
technique to locate delamination in composites plate structures.
Acknowledgements I would like to thank Dr. Ashok Pandey (SENSOR and ACTUATORS
laboratory, IIT Hyderabad) for allowing to use the LDV facility and perform my experiments.
References
1. O.K. Ihesiulor, K. Shankar, Z. Zhang, T. Ray, Delamination detection with error and noise
polluted natural frequencies using computational intelligence concepts. Compos. Part B 56,
906–925 (2014)
2. Z. Zhang, K. Shankar, T. Ray, E.V. Morozov, M. Tahtali, Vibration-based inverse algorithms
for detection of delamination in composites. Compos. Struct. 102, 226–236 (2013)
3. W. Fan, P. Qiao, Vibration based damage identification methods a review and comparative
study. Struct. Health Monit. 10(1), 83–29
4. W. Lestari, P. Qia, S. Hanagud, Curvature mode shape based damage assessment of carbon/
epoxy composite beam. J. Intel. Mat. Syst. Struct. 18, 189–208 (2007)
5. P. Qiao, K. Lu, W. Lestari, J. Wang, Curvature mode shape-based damage detection in
composite laminated plates. Compos. Struct. 80, 409–428 (2007)
6. C.N. Della, D. Shu, Vibration of delaminated composite laminates: a review. Appl. Mech.
Rev. 60(1), 1–20 (2007)
7. S. Keshav Kumar, Delamination modelling and detection in composite structure, PH.D.
thesis, Indian Institute of Science, Bangalore
8. S. Keshav Kumar, R. Ganguli, D. Harursampath, Partial delamination modelling in composite
beams using a finite element method. Finite Elem. Anal. Des. 75, 1–12
9. P. Mujumdar, S. Suryanarayan, Flexural vibration of the beams with delamination. J. Sound
Vib. 125(3), 441–461 (1988)
10. M.H.H. Shen, J.E. Grady, Free vibrations of delaminated beams. AIAA J. 30(5), 1361
11. L.J. Hadjileontiadisa, E. Doukab, A. Trochidisc, Fractal dimension analysis for crack
identification in beam structures. Mech. Syst. Signal Process. 19, 659–674 (2005)
FEA and Experimental Analysis
on Buckling and Post-buckling Behavior
of CFRP Composite Panel Using Digital
Image Correlation Technique
Muhammad Shuaib NK, Ramji M, Naresh Reddy Kolanu

and Gangadharan Raju
Abstract Finite element and experimental study have been carried out to inves-
tigate the buckling and post-buckling behavior of carbon fiber reinforced polymer
(CFRP) composite flat panels. For the experimental study, panels were manufac-
tured using hand lay-up followed by vacuum bagging process. The composite
panels were then subjected to uniaxial compression loading on an MTS test system
with the specimen simply supported on all the sides using specially designed fix-
ture. Digital image correlation technique (DIC) was used for capturing whole field
strain, axial displacement, out-of-plane deflection, and mode shapes. Experimental
buckling and post-buckling results are compared with the finite element analysis.

Keywords Buckling Composites Carbon fiber reinforced polymer (CFRP)
Digital image correlation (DIC)
Nomenclature
a Length of the plate
b Width of the plate
t Thickness of the plate
Dij Plate bending stiffnesses
w Out-of-plane deflection
N Line pressure load
Lcr Critical buckling load
Muhammad Shuaib NK Ramji M (&) Naresh Reddy Kolanu

Engineering Optics Lab, Department of Mechanical & Aerospace Engineering,
Indian Institute of Technology Hyderabad, Hyderabad 502205, India
G. Raju
ND & E Lab, Department of Mechanical & Aerospace Engineering,
Indian Institute of Technology Hyderabad, Hyderabad 502205,
Telangana, India

https://doi.org/10.1007/978-981-10-7197-3_44
528 M. S. NK et.al
1 Introduction
Composite materials have a wide range of applications because of their high

strength/stiffness to weight ratio, excellent fatigue, and possibility of tailored
properties. Starting from traditional application areas such as military aircraft, use
of composites has grown rapidly to impact various engineering and industrial fields
including automobiles, civil structures, sports equipment, and even marine struc-
tures. As the range of application increases, it is very important to have a deep
understanding of all the mechanical behaviors such as stiffness, strength, and sta-
bility behavior of the composites. As composites are commonly used in mass
reduction applications, most of the structures are thin in nature, and there is a higher
chance for buckling under compressive loading. Because of this thin nature of
composite structures, buckling is an important design driver for their safe and
reliable design.
Earlier Chen and Bert [1] have done a numerical study on optimal design of
simply supported rectangular laminated plates of composite material, subjected to
uniaxial compressive loading. Whitney [2] and Bao et al. [3] developed an exact
solution for critical buckling of solid rectangular orthotropic plates with all edges
simply supported and two edges simply supported/two edges clamped, respectively.
Jana and Bhaskar [4] have done few important studies on buckling of simply
supported rectangular plates under uniform and non-uniform uniaxial compression
for thin-walled structures. Yeh and Fang [5] investigated laminated plates under
pure bending analytically and experimentally. Later, Radu and Chattopadhyay [6]
investigated dynamic instability associated with composite panels using higher
order shear deformation theory. Shukla and Kreuzer [7] have done significant work
on estimating the critical/buckling loads of laminated composite rectangular plates
under in-plane uniaxial and biaxial loadings for different combinations of simply
supported, clamped, and free boundary conditions.
Although we get critical loads using load versus displacement graph from testing
machine in experiments, getting whole field strain, and out-of-plane deflection is
not possible using strain gauges, noncontact laser displacement meters [8] or using
linear variable displacement transducers [9] where we get the deflection of point or
multiple points. Recently developed 3-D DIC technique can be used to capture the
whole field deflection behavior which provides insight into the physics behind
buckling of plates. Ruran [10] has used 3-DIC technique to determine the delam-
ination buckling and growth behaviors of a cross-ply composite laminate. In this
study, 3-D DIC technique has been used to analyze pre-buckling and post-buckling
behavior such as, whole field strain, axial displacement, and out-of-plane deflection
of composite panel.
FEA and Experimental Analysis on Buckling and Post-buckling … 529
Fig. 1 Loading and

boundary condition
2 Problem Description
Finite element and experimental studies are carried out for an eight-layered carbon
fiber reinforced polymer composite (CFRP) rectangular flat plate having dimen-
sions of length (a) = 400 mm, width (b) = 200 mm, and thickness (t) = 1.8 mm.
The plate is analyzed for in-plane uniaxial compressive loading, and the boundary
conditions are shown in the Fig. 1. Plate with quasi-isotropic lay-up sequence
[±45°/90°/0°]s is chosen for the study, and all the edges are assumed to be simply
supported.
3 Finite Element Analysis
Finite element study has been carried out using commercial finite element software
ANSYS Version 15.0. An 8-node shell element (Shell-281) having 6 degree of
freedom per node is chosen for performing the buckling and post-buckling analysis
of composite panels. Mesh convergence study was performed, and the element size
was chosen to be 10 mm 10 mm, with a total number of 800 elements over the
complete plate of dimension 200 mm 400 mm. Material properties for CFRP
composites are obtained from material characterization tests conducted as per
ASTM standards. Material properties obtained from characterization tests are listed
in Table 1.
Block Lanczos mode extraction method was used to compute the critical
buckling load of the composite plates. The results of buckling analysis were then
used to perform the post-buckling analysis. An initial imperfection or a small lateral
load is necessary to initiate the instability of the structure which leads to buckling.
Table 1 CFRP material Material properties Values

properties
Longitudinal modulus E11 98.41 GPa
Transverse modulus E22 6.3 GPa
In-plane shear modulus G12 1.53 GPa
Out-plane shear modulus G23 1.91 GPa
In-plane Poisson’s ratio #12 0.23
Out-plane Poisson’s ratio #23 0.30
530 M. S. NK et.al
The first mode shape from Eigen buckling analysis has been used as the initial
imperfection with a scaling factor of 10% of the thickness of the plate. In the
post-buckling regime, the strain displacement relationship is nonlinear and requires
nonlinear solvers to solve the resulting finite element matrix equations. Newton–
Raphson method with automatic load stepping option was chosen to perform the
post-bucking analysis.
3.1 FEA Results
Initially, finite element analysis is done for CFRP composite having 8 layers with a
thickness 0.25 mm for each layer and with material properties listed in Table 1.
Comparison between different quasi-isotropic lay-up has been done to find out the
best lay-up sequence having higher critical buckling load and maximum
post-buckling stiffness.
From Fig. 2, quasi-isotropic [45°/−45°/90°/0°]S sequence is found out to be
having higher critical buckling load and maximum post-buckling stiffness, and
further studies have been carried out for the same lay-up sequence. The first three
critical buckling loads and their corresponding mode shapes are shown in
Fig. 3a–c, respectively.
Subsequently, post-buckling analysis was carried out on the composite panels to
study the reduction of stiffness of the plate after buckling and also the out-of-plane
displacement behavior of the panels. The results of the post-buckling analysis are
shown in Fig. 4.
Fig. 2 End shortening with

uniform loading for different
quasi-isotropic lay-up
sequences
Fig. 3 Different mode shapes for buckling of quasi-isotropic panel from finite element analysis,
a 1st buckling mode shape, critical buckling load, Ncr 1
¼ 3550:5 N, b 2nd mode shape,
Ncr ¼ 3779:9 N, c 3rd mode shape, Ncr ¼ 4499:5 N
2 3
Fig. 4 Post-buckling behavior from nonlinear buckling analysis, a end shortening with uniform
compressive loading, b variation in maximum out-of-plane deflection with uniform compressive
loading
4 Experimental Study
Experimental studies using 3-D DIC technique have been carried out to determine
the in-plane and out-of-plane deformation and variation of axial load with the
applied compression load.
532 M. S. NK et.al
4.1 Specimen Fabrication
CFRP composite panels having a length of 400 mm, width 200 mm, and thickness
1.8 mm and having both unidirectional ð0 Þ8 and quasi-isotropic [45°/−45°/
90°/0°]S lay-up sequences were made as shown in Fig. 5, for the experimental
studies. Composite panels were fabricated using hand lay-up followed by vacuum
bagging technique which gives an excellent fiber volume fraction. Unidirectional
carbon fiber mat having a density of 200 grams per square meter is used as the
reinforcement and CY-230 Epoxy resin mixed with HY-951 hardener in a weight
ratio of 10:1 is used as the matrix. Weight of resin taken is same as the weight of
fiber to form laminate and then kept it under vacuum. Aluminum plate of thickness
3 mm is placed on top to get uniform thickness throughout the panel. A burn-off
test has been conducted to determine the constituent content of CFRP made by
vacuum bagging process, and the volume fraction of fiber in the composite is found
out to be 0.53, and the weight fraction is 0.61.
4.2 Fixture Design and Fabrication
A fixture specially designed to realize the simply supported boundary condition and
loading described in Fig. 1 has been fabricated using EN-08 steel to perform the
experiment. Fixture consists of 4 parts, bottom plate, top plate, and 2 side supports
(see Fig. 6). Side supports are bolted to bottom plate and can slide through the
rectangular cutout made on top plate. 60 V-grooves are made on inner surfaces of
both plates and side supports to constrain out-of-plane deflection of all the sides of
testing panel.
Fig. 5 Casting of composite panel using vacuum bagging technique

Fig. 6 a Complete assembled fixture for buckling test, b two side supports with V-grooves,
c complete fixture with composite flat panel loaded
4.3 Experimental Setup
CFRP panels with zero degree unidirectional and quasi-isotropic [45°/−45°/

90°/0°]S lay-up sequences were tested using MTS servo-hydraulic cyclic testing
machine of 100 kN capacity. Panels were painted and speckled after vacuum
bagging and machining for getting speckled images for 3-D DIC analysis during
testing. Displacement controlled loading has been given to composite panels at a
rate of 0.4 mm per minute, and images captured using two grasshopper CCD
cameras (Point Grey-GRAS-5055 M-C) having a spatial resolution of
2448 2048 pixel at the rate of 2 images per second. Images are captured using
VIC snap software and post-processed using VIC 3-D software from correlated
solution. Figure 7 shows the typical experimental setup used for the present study.
Fig. 7 Experimental setup for buckling study of composite flat panels

534 M. S. NK et.al
Fig. 8 Experimental results obtained for a 8-layered quasi-isotropic [45°/−45°/90°/0°]S laminate

using 3-D DIC technique, a mode shape obtained at 4.5 kN, b applied load versus end shortening,
and c applied load versus maximum out-of-plane deflection
4.4 Result
Post-processing of captured images obtained from experiment was done using VIC
3-D software. Buckling mode shape, in-plane axial displacement, and post-buckling
results are extracted and shown in Fig. 8. There is a reduction in axial in-plane
stiffness and a drastic increase in out-of-plane deflection at load near 3100 N, which
can be taken as critical buckling load.
5 Results, Comparison, and Discussion
Mode shapes, critical buckling load, axial displacement, and post-buckling results
obtained from experiments using 3-D DIC are compared with FEA.
Mode shapes are showing the same trend in both FEA and experimental analysis
for UD ply as well as quasi-isotropic ply. Number of half sine waves in first mode
shape for UD and quasi-isotropic lay-ups are one and two, respectively, along with
the loading axis. For UD laminates at 2.0 kN load, the maximum out-of-plane
deflections are matching with an error percentage of 15.2% (Fig. 9a), whereas in
quasi-isotropic lay-up, both minimum and maximum deflections are matching with
an error percentage less than 10% (Fig. 9b). Change in maximum out-of-plane
deflection with respect to loading has been compared with both UD and
quasi-isotropic lay-ups in Fig. 10.
The plot of applied compression load versus out-of-plane deflection for both UD
and quasi-isotropic panels is shown in Fig. 10. For UD laminate, the post-buckling
results show the out-of-plane behavior of FEA as well as experimental study that
are close to each other. However, for quasi-isotropic, experimental critical buckling
Fig. 9 Experimental (DIC) and finite element (FEA) comparison of out-of-plane deflection in
mm, a unidirectional lay-up at 2.0 kN load, b quasi-isotropic lay-up at 4.5 kN load
Fig. 10 Experimental (DIC) and finite element (FEA) comparison of variation of out-of-plane
deflection with loading, a unidirectional lay-up, b quasi-isotropic lay-up
load is slightly less than the FEA value, though both FEA and experimental
post-buckling results are showing similar trend.
End shortening of panel with compressive loading in FEA and experiment has
also been compared for quasi-isotropic lay-ups. From Fig. 11a, there is a significant
difference in experimental post-buckling stiffness compared to FEA for uniform
compressive loading, which was later identified to be slight tilting (0.17 mm) of the
testing fixture toward left side. This condition was then simulated in FEA by
replacing the uniform loading with parabolic compressive loading, and the com-
puted post-buckling stiffness is matching with the experimental results. The effect
of tilting is clearly visible from Fig. 11c, where the in-plane axial displacement
contours are tilted instead of straight horizontal lines.
536 M. S. NK et.al
Fig. 11 a Experimental (DIC) and finite element (FEA) comparison of end shortening of
quasi-isotropic lay-up, b schematic diagram of loading, c in-plane axial displacement from
experiment
6 Conclusion
Study has been done for understanding the pre-buckling and post-buckling behavior
of the CFRP flat panel under in-plane axial compressive loading with all the sides
simply supported. Finite element study using Eigen buckling followed by nonlinear
analysis has been done to obtain the buckling behavior of CFRP panel including
mode shapes, critical buckling load, and axial/out-of-plane deflections, which are
used to validate the experimental results obtained from 3-D DIC technique. From
finite element study, [45°/−45°/90°/0°]S lay-up shows the highest post-buckling
stiffness and has got a higher critical buckling load among all quasi-isotropic lay-up
sequences. The composite plate fabricated for experimental study of quasi-isotropic
panel is of the above sequence. From experimental and finite element study,
comparing UD and quasi-isotropic lay-up sequences, even though the UD lay-up
sequence is having a very high pre-buckling and post-buckling stiffness, the critical
buckling load is very less compared to the quasi-isotropic lay-up. Experimentally
determined mode shapes, critical buckling load, and out-of-plane deflection are to
match well with finite element estimates for both UD and quasi-isotropic lay-up
sequences. Although there was a noticeable reduction in experimentally determined
results of plane axial stiffness in comparison with FEA, which is later resolved by
exactly replicating the experimental loading condition in the finite element mod-
eling. As the CFRP panel shows excellent stiffness toward compression, even slight
non-parallelism (in range of micro meters) between top and bottom side of fixture
leads to experimental errors in terms of in-plane axial stiffness. Making them
parallel and perfectly aligning both the sides of the loading fixture would exert a
uniform load and would generate exact stiffness in experimental studies.
References
1. T.L.C. Chen, C.W. Bert, Design of composite-material plates for maximum uniaxial
compressive buckling load. Proc. Okla. Acad. Sci. 56, 104–107 (1976)
2. J.M. Whitney, Structural Analysis of Laminated Anisotropic Plates (Technomic Publishing,
Lancaster, PA, 1987)
3. G. Bao, W. Jiang, J.C. Roberts, Analytic and finite element solutions for bending and
buckling of orthotropic rectangular plates. Int. J. Solids Struct. 34(14), 1797–1821 (1997)
4. P. Jana, K. Bhaskar, Stability analysis of simply-supported rectangular plates under
non-uniform uniaxial compression using rigorous and approximate plane stress solutions.
Thin-Walled Struct. 44, 507–516 (2006)
5. M.K. Yeh, L.B. Fang, Bending behavior of delaminated composite plates with contact effects.
Compos. Struct. (1997) (Elsevier)
6. A.G. Radu, A. Chattopadhyay, Dynamic stability analysis of composite plates including
delaminations using a higher order theory and transformation matrix approach. Int. J. Solids
Struct. 39(7), 2000
7. K.K. Shukla, Y. Nath, E. Kreuzer, K.V. Sateesh Kumar, Buckling of laminated composite
rectangular plates. J. Aerosp. Eng. 18, 215 (2005)
8. H. Suemasu, T. Irie, T. Ishikawa, Buckling and post buckling behavior if composite plates
containing multiple delaminations. J. Compos. Mater. 43(2), 191–202 (2009)
9. H. Gu, A. Chattopadhyay, An experimental investigation of delamination buckling and post
buckling of composite laminates. Compos. Sci. Technol. 59(6), 903–910 (1999)
10. J.T. Ruan, F. Aymerich, J.W. Tong, Z.Y. Wang, Optical evaluation on delamination buckling
of composite laminate with impact damage. Adv. Mater. Sci. Eng. 2014, 390965 (Hindawi
Publishing Corporation)
Part VIII
Fatigue
High Cycle Fatigue Strength of Spring
Steel with Small Scratches
Yoshiro Nishimura, Masahiro Endo, Keiji Yanase, Yuichi Ikeda,

Susumu Miyakawa and Nobuyuki Miyamoto
Abstract When the compression coil springs are subjected to high cyclic stress,
fatigue crack sometimes initiates and propagates from small scratch or defect
produced in the manufacturing process. However, there are a limited number of
studies on the fatigue behavior of high-strength spring steel in the presence of small
scratch. In this study, the fatigue tests were conducted to examine the uniaxial
fatigue behavior of a high-strength spring steel (JIS G 3561, SWOSC-V) in the
pffiffiffiffiffiffiffiffiffi
presence of small circumferential notches. Based on area parameter model, the
sensitivity of hard steels, including a high-strength spring steel, to small scratches
was qualitatively and quantitatively examined.
Keywords Spring steel Scratch Notch effect Fatigue limit area
1 Introduction
These days, a number of components in the automotive vehicles have various types
of compression coil springs. Further, because of the strong demand for better
performance of automotive vehicles, the use of high-strength spring steels has been
increasing [1]. When the compression coil springs are subjected to high cyclic
stress, fatigue crack sometimes initiates and propagates from small scratch or defect
Y. Nishimura (&) S. Miyakawa N. Miyamoto

Materials Engineering R&D Division, DENSO Corporation, 1-1 Syowa-cho,
Kariya City 448-8661, Japan
e-mail: yoshirou_nishimura@denso.co.jp
M. Endo K. Yanase Y. Ikeda
Department of Mechanical Engineering, Fukuoka University, 8-19-1 Nanakuma Jonan-ku,
Fukuoka City 814-0180, Japan
M. Endo K. Yanase
Institute of Materials Science and Technology, Fukuoka University, 8-19-1 Nanakuma
Jonan-ku, Fukuoka City 814-0180, Japan

https://doi.org/10.1007/978-981-10-7197-3_45
542 Y. Nishimura et al.
produced in the manufacturing process. As well recognized, fatigue is by far the

most common cause of failure for engineering components. Therefore, its preven-
tion is a major preoccupation for a number of industries. The fatigue strength of
engineering components can be significantly lowered by the presence of surface
scratches, surface finish flaws, nonmetallic inclusions, casting defects, etc.
Accordingly, the ability to predict the effects of those defects on fatigue life and
fatigue strength is crucial in engineering design.
In this study, to gain a fundamental understanding of the fatigue behavior of a
spring steel (JIS G3561, SWOSC-V) with a small scratch, the fatigue tests were
conducted in the high cycle fatigue regime. The uniaxial fatigue behavior was
investigated by using a rotating bending fatigue testing machine. The effect of
scratch was investigated by introducing the small circumferential notch on the
specimen surface. Finally, the effects of small scratches were studied by using the
area parameter model ([2], cf. [3–5]).
2 Experiment
The investigated material was an oil-tempered Si–Cr steel wire used for valve
springs (JIS G3561, SWOSC-V). SWOSC-V is a super clean steel, especially
intended for the manufacture of valve springs and other springs that require
excellent fatigue properties and excellent relaxation properties at moderately high
working temperature. Chemical compositions of the investigated material are listed
in Table 1. The mechanical properties of the material are controlled through
patenting, drawing, and oil-tempering process, and they are shown in Table 2.
Figure 1 shows the specimens manufactured from the wires with a diameter of
9 mm by turning and grinding process. The center of specimen is finished by using
paper polishing, electro-polishing, and colloidal silica to reduce the roughness and
the residual stress on the specimen surface. The surface roughness of the specimen
was negligibly small, and it had no effect on fatigue behavior. The residual stress in
the axial direction was measured with an X-ray diffraction, and the residual stress of
the specimen was within ±20 MPa on the specimen surface. By using a
micro-Vickers hardness tester, the Vickers hardness, HV, was measured both in
cross-sectional and longitudinal section of a specimen. As shown in Fig. 2, the
anisotropy on HV was not observed, and HV = 533 kgf/mm2 was obtained. To
investigate the effect of scratch on the fatigue properties, a small 60o V-shaped
circumferential notch with depth of t = 50, 100 lm and notch root radius of
q = 15 ± 5 lm were introduced on the specimen surface (cf. Fig. 1b).
Table 1 Chemical compositions of SWOSC-V (in % weight)

C Si Mn Cr Cu P S
0.51 * 0.59 1.20 * 1.60 0.50 * 0.80 0.50 * 0.80 * 0.2 * 0.025 * 0.025
High Cycle Fatigue Strength of Spring Steel with Small Scratches 543
Fig. 1 Geometries and sizes of test specimens (in mm)
Fig. 2 Results of
measurement for Vickers
hardness
Table 2 Mechanical properties of SWOSC-V

0.2% proof stress (MPa) Tensile strength (MPa) Reduction of area (%)
1400 * 1570 1610 * 1760 40 *
2.1 Smooth Specimen
Based on the applied bending moment, M, the nominal stress amplitude, ra , is

defined as:
M pd 3
ra ¼ ; where Z ¼ ð1Þ
Z 32
Here, d is the diameter of specimen. Figure 3 shows the S-N data of smooth
specimen at room temperature, and its fatigue limit is experimentally determined as
Fig. 3 S-N data of smooth

specimen
rw0 ¼ 850 MPa. On the other hand, fatigue limit of smooth specimen can be
predicted by the following equation [2]:
rw0 ¼ 1:6HV 0:1HV ð2Þ
Here, rw0 is in MPa and HV is in kgf/mm2. By substituting HV ¼ 533 in Eq. (2),

rw0 ¼ 853 MPa is obtained, which is nearly equal to rw0 ¼ 850 MPa that is
determined by experiment. The relationship of Eq. (2) holds when fatigue crack
mainly initiates from slip bands on the material surface, and Eq. (2) is usually
applicable for HV < 400. On the other hand, when HV > 400, the defects and
inclusions inside the materials frequently influence the initiation of fatigue crack.
Namely, fatigue limit of high-strength steel becomes sensitive to the presence of
defects and inclusions, and fatigue limit cannot be properly predicted with Eq. (2).
In this study, however, fatigue cracks initiate from the surface despite HV > 400,
and fatigue limit is reasonably predicted with Eq. (2). This fact indicates that the
size of inclusions in SWOSC-V is significantly small, and fatigue limit is controlled
by the bulk material property of SWOSC-V (i.e., HV). In the literature, Akiniwa
et al. [6] reported that fatigue cracks in SWOSC-V were initiated in all specimen at
the surface, and no inclusion or granular facet areas were detected at the crack
initiation site. Thus, they concluded that since the resistance of SWOSC-V against
crack propagation from internal inclusions is relatively large, the presence of
inclusion does not decrease fatigue strength up to the giga cycle fatigue regime.
Finally, as shown in Fig. 3, the knee point emerges at relatively small number of
cycles, N 3:0 105 cycles.
2.2 Notched Specimen
Figure 4 shows a preliminary result of notched specimen, in which the effect of

residual stress at the notch root is highlighted. Fatigue limit with a small circum-
ferential notch can be predicted by the following equation [1]:
1:43ðHV þ 120Þ pffiffiffiffiffiffiffiffiffi pffiffiffiffiffi

rw ¼ pffiffiffiffiffiffiffiffiffi 1=6 ; where area ¼ 10 t ð3Þ
ð areaÞ
Here, area (in lm) represents the square root of the defect projected onto the
plane perpendicular to the loading direction, and t (in lm) is the notch depth (Fig. 1b).
As shown in Eq. (3), in the case of small circumferential notch, fatigue limit is gov-
erned by the notch depth, t, not by the notch root radius, q. To calculate the applied
stress, the outer diameter of specimen is used for d in Eq. (1). It is noted that the
accuracy of Eq. (3) is within 10% for a wide variety of steels [2]. As shown by Fig. 4,
before electro-polishing, the specimens with t = 50 lm did not fail until N 107
cycles though much higher stresses (20 and 30%, respectively) than the predicted
fatigue limit were applied. Interestingly, after electro-polishing, those specimens
instantly failed under the same applied stresses. Figure 5 shows the non-propagating
crack observed at the notch root before electro-polishing. The existence of
Fig. 4 Effect of electro-polishing on the fatigue strength (t = 50 lm)

Fig. 5 Non-propagating
cracks observed at the notch
root before electro-polishing
(stress amplitude is +30% of
the predicted fatigue limit, cf.
Fig. 4)
non-propagating crack indicates the presence of compressive residual stress at the

notch root due to machining, and the residual stress affects the fatigue crack growth
behavior and fatigue limit. After a slight electro-polishing, the previously observed
non-propagating crack disappeared, thus the depth of crack would be very small (2–
5 lm). Accordingly, though the affected zone is small, the removal of residual stress at
the notch root is essential to properly evaluate the effect of notch for the present
material. Therefore, electro-polishing was applied to remove a layer of 10–15 lm
thickness at the notch root, and the fatigue test was conducted accordingly.
Figure 6 shows the comparison of S-N data for smooth and notched specimens.
When a specimen endured the load cycles of N 1.2 107 cycles, it is regarded
Fig. 6 Comparison of S-N data

Fig. 7 Non-propagating
cracks observed at the notch
root (t = 50 lm) (stress
amplitude is ±0% of the
predicted fatigue limit, cf.
Fig. 6)
as a run-out specimen. For instance, based on Fig. 6, fatigue limit is experimentally

determined as rw = 402 MPa for t = 50 lm that is equal to the prediction with
Eq. (3). It is widely recognized that fatigue limit of notched components is fre-
quently dictated not by crack initiation, but by the critical condition for the prop-
agation of a short crack at the root of a notch. Therefore, the boundary between
propagation and non-propagation separates the safe from the potentially unsafe
fatigue regime (cf. [7–9]). Figure 7 shows the non-propagating crack observed at
the fatigue limit for t = 50 lm, and such a crack was not observed when the stress
amplitude was reduced to ra = 381 MPa (−10% of predicted fatigue limit, cf.
Fig. 6). Accordingly, ra = 381 MPa is smaller than the crack initiation threshold
for t = 50 lm, and the non-propagating crack exists in the limited range of stress
amplitude.
Figure 8 shows the variations of fatigue limit associated with the size of cir-
cumferential notch. In addition to SWOSC-V, the experimental data of S45C
(HV = 400, unpublished data), S50C (HV = 375 378, [2]), and maraging steel
Fig. 8 Variations of fatigue limit associated with the size of circumferential notch
(HV = 720, [2]) are plotted. As shown, the overall accuracy of ±10% still holds for
these hard steels. However, if any, as the material hardness increases, the experi-
mental data tend to deviate from the prediction line at smaller area or t. Such a
behavior is typically highlighted for maraging steel (symbol ⋆ in Fig. 8). To
investigate this behavior for SWOSC-V, the fatigue tests with notched specimens
will be further conducted.
3 Conclusion
In this study, the fatigue tests were conducted to examine the uniaxial fatigue
behavior of high-strength spring steel, SWOSC-V. To study the effects of small
scratch, the small circumferential notch was carefully introduced to the specimen
surface. The present investigation confirms that the area parameter model is
applicable to predict the fatigue strength of SWOSC-V and other hard steels.
However, the tendency of deviation of experimental data for hard steels from the
prediction line is observed as the size of notch increases.
References
1. N. Ibaragi, Developments in automotive valve spring wire rods. Kobe Steel Eng. Rep. 50, 27–
30 (2000). (in Japanese)
2. Y. Murakami, Metal Fatigue: Effects of Small Defects and Nonmetallic Inclusions (Elsevier,
New York, 2002)
3. M. Endo, K. Yanase, Effects of small defects, matrix structures and loading conditions on the
fatigue strength of ductile cast irons. Theor. Appl. Fract. Mec. 69, 34–43 (2014)
4. K. Yanase, M. Endo, Multiaxial high cycle fatigue threshold with small defects and cracks.
Eng. Fract. Mech. 123, 182–196 (2014)
5. B.M. Schönbauer, K. Yanase, M. Endo, VHCF properties and fatigue limit prediction of
precipitation hardened 17-4PH stainless steel. Int. J. Fatigue 88, 205–216 (2016)
6. Y. Akiniwa, S. Stanzl-Tschegg, H. Mayer, M. Wakita, K. Tanaka, Fatigue strength of spring
steel under axial and torsional loading in the very high cycle regime. Int. J. Fatigue 30, 2057–
2063 (2008)
7. M. Endo, A.J. McEvily, Fatigue crack growth from small defects under out-of-phase combined
loading. Eng. Fract. Mech. 78, 1529–1541 (2011)
8. K. Yanase, M. Endo, Analysis of the notch effect in fatigue. J. ASTM Int. 9, 1–13 (2012)
9. K. Yanase, K. Shojima, C. Ogata, High-cycle fatigue threshold behaviors in notched plates. Int.
J. Damage Mech 22, 1006–1022 (2013)
Estimation of Load Sequence Effect
on Fatigue Crack Growth Life According
to Various Prediction Models
A. N. Savkin, A. A. Sedov, A. V. Andronik and K. A. Badikov
Abstract Possible variability of endpoints on the da/dN-DK curve is analyzed, and

its influence on crack growth life is shown. Basics of the fatigue crack growth
model of material with a single crack are presented. The proposed model calculates
local stress at a remoted point close to the crack tip. The model establishes a
relationship between the stresses and the near-threshold range of stress intensity
factor. Tests were carried out for the low crack growth rates on aluminum alloy
2024-T3 and low carbon steel with rare small overloads of different sequence, and
as a result, a-N and da/dN-DK plots were received. Fatigue predictions were done
using existing Elber, Barsom, Wheeler, Willenborg models, NASGRO and
FASTRAN software and the proposed model. It is shown that the proposed model
of local stress calculation at the distance from crack tip is able to predict fatigue life
considering load sequence of no less accurate than the best of existing models.
However, application area is restricted to small overloads, high cycle ratio, and low
crack growth rate at the moment.
Keywords Fatigue crack growth Load sequence Prediction models

Local stress calculation
1 Introduction
Growth of macrocracks usually may be described by the da/dN-DK curve, which is

a plot of the crack growth rate da/dN function of the stress intensity factor range
(SIF) DK (Fig. 1). Three regions can be noted on the da/dN-DK curve [1]: (1) the
region of the near-threshold crack growth rates (I); (2) the Paris regime (II); (3) the
region of stable tear preceding fracture (III). Nonlinear form of the da/dN-DK curve
is conditioned by various dominated crack growth mechanisms on the various
A. N. Savkin A. A. Sedov (&) A. V. Andronik K. A. Badikov

Department of Strength of Materials, Volgograd State Technical University,
Lenina Av., 28, Volgograd, Russian Federation 400005
e-mail: alexander.a.sedov@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_46
550 A. N. Savkin et al.
Fig. 1 da/dN-DK curve and

corresponding fractographies
for the aluminum alloy
2024-T3 [2]
regions that is affirmed by fractography analysis [2]: (1) brittle microfracture occurs
on the I region; (2) cyclic slip and shear prevails on the II region; (3) quasistatic
fracture by cleavage, strain localization, and microvoid coalescence is typical for
the III region. Endpoints of the da/dN-DK curve are important characteristics of the
fatigue resistance of material [1]: (1) threshold range of SIF DKth; (2) fracture
toughness Kc.
These properties DKth and Kc are often taken constant during calculation, but
variable in practice. Though inaccuracy in taking Kc as an input value for calcu-
lation leads to the insignificant divergency of the test and prediction results of crack
growth life till fracture of an order 102…104 cycles, then inaccuracy in taking DKth
causes a huge error of an order 105…106 cycles; thus, the precise DKth value is a
more critical factor at the high cycle fatigue.
2 Near-Tip Stress Model for DKth
An interesting model of the variable DKth was proposed by Sunder [3], which has
found an empirical correlation between DKth and the local stress at the distance r*
from the crack tip as a result of the tests with constant amplitude loading and
periodical overloads to define DKth for the aluminum alloy (Fig. 2). To avoid the
stress singularity at the crack tip, Sunder estimated the local stress r* at the distance
from the crack tip r* obtained likewise correction of plastic zone derived by Irwin
(Fig. 3a) [3]:

1 Kc2
r ¼ : ð1Þ
2 p r f ef E
where rf—local fracture stress, ef—local fracture strain.

Estimation of Load Sequence Effect on Fatigue Crack Growth Life … 551
Fig. 2 Relationship between

the threshold range of SIF
DKth and the local stress near
crack tip r* for the aluminum
alloy 2024-T3 [3]
Fig. 3 Local stress r* near crack tip at the distance r* (a); graphical representation of
combination of the Neuber rule and Ramberg-Osgood equation (b)
Evidently, given estimation allows to look inside of the plastic zone, but
restricted by the condition r* rf. Using the Neuber rule [4] and the
Ramberg-Osgood equation (Fig. 3b) [5], Sunder succeeded to associate the SIF
K or its range DK with the local stress r* at the distance from crack tip r* for
monotonic and cyclic loading correspondingly [3]:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
u
u 10 !
r r
K ¼ t2p r E r
n
þ ; ð2Þ
E K0
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
u
u
10 !
Dr Dr
DK ¼ t2p r E Dr
n
þ2 ; ð3Þ
E 2K 0
where K and DK—SIF and its range for the current point and load cycle
correspondingly;
E—elastic modulus;
r* и Dr*—stress and its range for the current point and load cycle
correspondingly;
n′ и K′—factors of cyclic hardening in Ramberg-Osgood equation.

In such a manner, Sunder established the unambiguous relationship between the
current loading and the threshold range of SIF DKth. Above-mentioned formulas
and relationships are sufficient to formulate the fatigue crack growth prediction
model; therefore, thee interest appears for comparison analysis of efficiency of
proposed and existing models. Classical models of Elber, Barsom, Wheeler,
Willenborg, and contemporary models of NASGRO [6] and FASTRAN [7] were
elected for this analysis. The special interest is resided in the generalized and
modified Willenborg models in NASGRO and the crack closure model in
FASTRAN. The listed models belong to the following classes [8]: (1) crack closure
model; (2) probabilistic models; (3) models considering plasticity close to the crack
tip.
It is a common knowledge that overloads and underloads cause various influence
on the fatigue crack growth rate da/dN. Overloads usually retard crack propagation,
but underloads can accelerate it [9]. Overloads exert various influence on the crack
opening force Pop and the threshold range of SIF DKth (Fig. 4) [10]. Therefore,
according to the contemporary crack closure models, an overload and an underload
cause increases of the Pop, but DKth grows after the overload and falls after the
underload by the proposed model. As well as overloads and underloads cause
various cyclic stress–strain response near the crack tip during the following loading
(Fig. 5): mean stress during the baseline loading is less after overloads than after
underloads and it is considered in the proposed model (Fig. 2). This various
description must find its way in the crack growth life prediction for the fatigue crack
growth. Correlation between DKth and r* was obtained during the tests with
Fig. 4 Influence of various overloads on the crack opening force Pop and the threshold range of
SIF DKth [10]
Fig. 5 Various cyclic stress–strain response near the crack tip during the baseline loading after
overloads (a) and underloads (b)
overloads thus the interest appears to carry out similar tests with overloads of the
various sequence.
Conventional aluminum alloy AA 2024-T3 was used as a test material, the
duralumin of the Al–Cu–Mg dopant system. Besides, Cosmo Magna company has
afforded the low carbon alloy steel for manufacturing of a suspension underframe in
Ford Fiesta and Fusion automobiles. C(T)-specimens of the nonstandard geometry
were used for the tests (Fig. 6a): (1) specimen thickness B is 5 mm instead of usual
12.5 mm; (2) hole diameter D is 6 mm instead of usual 12.5 mm.
It is suggested that the variable threshold range of SIF DKth effects on crack
growth life, but not the crack closure; thus, the high cycle ratio for the baseline block
of the constant amplitude loading R 0.7 and small overloads Kol/Kmax 1.5
were applied. According to the above-mentioned recommendations for the alu-
minum alloy 2024-T3, the test schedule was planned, including: (1) tests with
periodical underload–overload; (2) tests with periodical overload–underload;
(3) constant amplitude loading. The tests with overloads and underloads consisted of
overload cycle and baseline constant amplitude loading block. The tests were con-
ducted for the following overload levels: (1) Kol/Kmax = 1.5, i.e., 50%;
Fig. 6 Geometry of test specimen (a); load sequence with overload–underload (b)
(2) Kol/Kmax = 1.4, i.e., 40%; (3) Kol/Kmax = 1.3, i.e., 30%. The cycle ratio R and the
maximal load Pmax for the baseline loading block were different, but the load range
was the same ΔP 0.5 kN due to the next conditions: (1) R = 0.7, Pmax = 1.7 kN;
(2) R = 0.75, Pmax = 2 kN. After the overloads with frequency Fol = 0.1 Hz,
baseline loading was completed with the frequency Fbl = 150 Hz during Nbl = 104
cycles, and loading was repeated afterward. The underloads were in tensile loading
zone (Kul/Kmax = 0.15). The constant amplitude loading was applied for the cycle
ratios: (1) R = 0.75; (2) R = 0.7; (3) R = 0.5; (4) R = 0.3; (5) R = 0.1. However, the
cycle ratio R varied, but the maximal load Pmax = 5 kN of the baseline loading was
constant at the tests with the low carbon steel: (1) R = 0.75; (2) R = 0.7; (3) R = 0.5.
Applied designations are illustrated on Fig. 6b.
The preliminary calculation of DKth at the proposed tests resulted in surface plots
(Fig. 7) describing alteration of the threshold range of SIF DKth and the effective
range of SIF DKeff at the various cycle ratio R and the overload level Kol/Kmax
Fig. 7 Relationship between the threshold range of SIF DKth, the overload level Kol/Kmax, and the
maximal SIF of the baseline constant amplitude loading Kmax (a); relationship between the
effective range of SIF DKeff and the threshold range of SIF DKth at the cycle ratios for the baseline
loading R = 0.7 (b), R = 0.75 (c), R = 0.8 (d)
during the crack propagation. The proximity of surfaces (DKeff ! DKth,

DKeff > DKth) is typical for the near-threshold crack growth rates, where excess of
surface for the threshold range of SIF DKth (DKeff DKth) means the crack growth
delay.
As the results of all tests, the generalized da/dN-DK curves were obtained for the
aluminum alloy 2024-T3 (Fig. 8) and the low carbon steel (Fig. 9). Fatigue crack
growth rates for the tests with the last overload and underload at overload block
differ approximately by an order. It leads to the significant divergence at the fatigue
crack growth prediction and cannot be explained using the crack closure
phenomenon.
For the crack growth rate estimation, the modified Forman-Mettu equation was
used [6]:
p
DKth
da 1 DK
¼ C DKeff q ; ð4Þ
m eff
dN 1 Kc
Kmax
where ΔKeff—effective range of SIF. It is proposed in the tests with overloads that
the crack closure influence is negligibly small, so we can accept ΔK = ΔKeff.
The Paris factors c and m were defined during the constant amplitude tests, p and
q—during tests with overloads and underloads. All these factors were constant
during the crack growth life prediction unlike the DKth. Assuming the variable
character of p factor, it was found that the higher DKth, the higher p value takes
place as a result of the approximation procedure and vice versa. Moreover, the
linear relationship between DKth and p is typical for the tests with underload–
overload sequence during the overload cycle. Thus, modification may appear for the
further versions of the fatigue crack growth rate equation. Despite this fact for the
purpose of forecasting averaged p factor value was chosen.
Fig. 8 Generalized
da/dN-DK curve for the
aluminum alloy 2024-T3
Fig. 9 Generalized
da/dN-DK curve for the low
carbon steel
Comparison of the prediction and experimental results by the various models is

given on Fig. 10. Results are shown for the aluminum alloy on Fig. 10a, b and for
the steel on Fig. 10c, d.
Correlation factors between the experimental and prediction results of the fatigue
crack growth life are presented on Tables 1 and 2 for the aluminum alloy and the
Fig. 10 Correlation between the experimental and estimation results of the fatigue crack growth
life for the aluminum alloy 2024-T3 (a, b) and the low carbon steel (c, d)
Table 1 Correlation factors between experimental and estimation results of the fatigue crack
growth life for the tests on the aluminum alloy 2024-T3
r
Crack closure model (Elber) 0.346
Probabilistic model (Barsom) 0.267
Crack tip plasticity model (Wheeler) 0.387
Crack tip plasticity model (Willenborg) 0.507
Generalized Willenborg model (NASGRO) 0.52
Modified generalized Willenborg model (NASGRO) 0.537
Crack closure model (FASTRAN) 0.58
Proposed model for local stress estimation at the distance from crack tip 0.89
Table 2 Correlation factors r

r between experimental and
estimation results of the Crack closure model (Elber) 0.556
fatigue crack growth life for Probabilistic model (Barsom) 0.739
the tests on the low carbon Crack tip plasticity model (Wheeler) 0.38
steel Crack tip plasticity model (Willenborg) 0.46
Generalized Willenborg model (NASGRO) 0.45
Modified generalized Willenborg model (NASGRO) 0.563
Crack closure model (FASTRAN) 0.751
steel correspondingly. Correlation factor for the tests on the aluminum alloy is
greater for the proposed model than for others. The tests to define the threshold
range of SIF DKth at the steel are not accomplished yet, so the corresponding
calculation results are not given. Crack closure model in FASTRAN seems to be
also effective, but the fact that the proposed model showed the better accuracy than
the commercial models reveals a very promising position of the first one.
3 Conclusions
1. The novel fatigue crack growth model based on the local stress estimation at the
distance from the crack tip was formulated. The model predicts the fatigue crack
growth life at the region of the near-threshold crack growth rates more effective
than the existing models, assuming smallness of overloads.
2. Commercial software, for example, NASGRO and FASTRAN, uses hybrid
models for fatigue crack growth life estimation to combine the advantages of
various models and for the forecasts, which are more precise. Sometimes they
work well, however, as it is shown at the current paper, some results leave much
to be desired. Low values of the correlation factors between the forecasting and
experimental results of the crack growth life according to the NASGRO and
FASTRAN models are represented.
Acknowledgments This paper was financially supported by RFBR grant 17-08-01648 A and RF
President grant MK-943.2017.8.
References
1. J. Schijve, Fatigue of Structures and Materials (Springer, Delft, 2009), p. 623

2. R. Sunder, Unraveling the science of variable amplitude fatigue. J. ASTM Int. 9(1), 32 (2012)
3. R. Sunder, Engineering application of threshold stress intensity, in ASTM-ESIS Annual
Symposium on Fracture Mechanics, USA, FL, 2013, pp. 24–48
4. H. Neuber, Kerbspannungslehre (Springer, Berlin, 2000), p. 326
5. T. Siretanu, A. Mitu, M. Giuclea, O. Solomon, D. Stefanov, Analytical method for fitting the
Ramberg-Osgood model to given hysteresis loops. Proc. Roman. Acad. Series A. 15(1), 35–
42 (2014)
6. J. Maierhofer, R. Pippan, H.-P. Gänser, Modified NASGRO equation for physically short
cracks. Int. J. Fatigue 59, 200–207 (2014)
7. M. Vormwald, Classification of load sequence effects in metallic structures. Procedia Eng.
101, 534–542 (2015)
8. J.A. Bannantine, J.J. Comer, J.L. Handrock, Fundamentals of Metal Fatigue Analysis
(Prentice Hall, New Jersey, 1990), p. 271
9. M. Sander, H.A. Richard, Fatigue crack growth under variable amplitude loading Part II:
analytical and numerical investigations. Fatigue Fract. Eng. Mater. Struct. 29(4), 303–319
(2006)
10. A.N. Savkin, A.V. Andronik, R. Abhilash, Crack closure detection using photometrical
analysis. Period. Polytech. Mech. Eng. 59(3), 114–119 (2015)
SHM—Prognostic Analysis of Tapered
Attachment Lugs Under Fatigue Loading
K. Bharath, B. V. Sravan Kumar, L. Chikmath and B. Dattaguru
Abstract Over the years, the flight safety has become a matter of prime concern
due to several fatigue failures of structural components and the need to prevent
them. Among these, large number of fatigue failures was initiated at major dis-
continuities such as the inevitable structural joints. Due to this, structural health
monitoring (SHM) at these critical locations has become a crucial aspect in the
modern damage tolerance-based design concept. Lug joints are often preferred in
large-scale structural components which act as primary load-carrying members and
are more susceptible to failure due to high stress concentration and geometric
irregularities. The primary issue in the analysis of these joints is the loss of contact
between the pin–hole boundary during loading and possible change in contact
conditions during fatigue loading. This paper deals with the behaviour of tapered
attachment lug joints under fatigue loading and finally presents prognostic analysis
to estimate remaining life of the joints at any instant of fatigue cycles.
Keywords Crack growth Fasteners Fatigue Lug joints Prognosis

SHM
Nomenclatures
2a Hole diameter
a0 and b0 Crack growth material constants
ac Critical crack length
b Fatigue strength exponent
C and m Material constants
c and s Regions of contact/separation
K. Bharath L. Chikmath (&) B. Dattaguru

Department of Aerospace Engineering, Jain University, Bengaluru, India
e-mail: lokamanyachikmath@gmail.com
B. V. Sravan Kumar
National Aerospace Laboratories, Bengaluru, India

https://doi.org/10.1007/978-981-10-7197-3_47
560 K. Bharath et al.
GI Mode-1 Strain energy release rate

KI Mode-1 Stress intensity factor
KIc Mode-1 Fracture toughness
Kt Stress concentration factor
L and W Length and width of the lug
P Pin load
R and RP Stress and load ratios
Ri and Ro Inner and outer radii of the lug
b=2 Taper angle
rt and rr Tangential and radial stresses
rbr Bearing stress
srh Shear stress
1 Introduction
Safety of an aircraft is a matter of grave concern in all forms of design concepts.

Joints which are inevitable in large-scale structural systems like aircraft are often
prone to failures predominantly under cyclic loading which drives the development
of effective and reliable computational methods relevant to fatigue and fracture
analysis. Hence, location of sensors for integrated vehicle health monitoring
(IVHM) system at critical joints in aircraft structures has become one of the future
requirements [1]. The science and technology of structural health monitoring
(SHM), which is an important part of integrated vehicle health management
(IVHM), comprises of two major aspects viz., one is the diagnosis, which deals
with the detection of crack-like flaws in structures followed by prognosis, which
involves the assessment of crack, crack growth behaviour and further estimation of
residual strength and residual life of the structure [2].
Large-scale structural components are assembled in parts, and lugs are used for
transferring loads at major joints such as the wing–fuselage junction. These lug
joints are generally used as primary load-carrying members, and the ease of
assembly and disassembly is an added advantage to prefer fasteners in metallic
structures. In this current work, aluminium lug with push fit rigid pin is analysed,
which nearly corresponds to a steel pin in an aluminium lug with the modulus ratio
of pin to lug being equal to three. The effect of the relative rigidity of the pin and the
lug on the stress intensity factor is negligible [3]. Unlike interference and clearance
fits whose separation/contact regions between pin and the lug hole vary with
varying pin load which leads to nonlinear moving boundary value problem, push fit
poses a linear elastic problem with constant contact/separation region between pin
and the lug for monotonically varying applied pin load [4]. However, in push fit
too, load reversals during fatigue loading cause different contact configurations
when the pin is pulled and pushed alternatively. The prognosis in these cases is
SHM—Prognostic Analysis of Tapered Attachment … 561
based on physical models, but it is interesting since one has to be aware that these
are not simple linear problems, but they could be nonlinear.
The literature on lug joints is vast, and in this paper, only a few of those past
contributions relevant to the current work are reviewed. The classical contribution
of Bickley [5] is still valid, and most of the industry practices assume that the load
transfer at the pin–hole interface is over half of the hole boundary, and the load is
transferred in a cosine distribution. With the advent of finite element methods, more
accurate contact stress analysis [6] showed that while Bickley’s assumptions are
nearly valid for push fits, whereas for interference and clearance fit joints contact/
separation zones need to be evaluated. Economic analysis using inverse formula-
tions is presented in a series of papers, and these are reviewed in [3, 7]. For the
current configuration of tapered lug joints (including the straight lug as a limiting
case) with a push fit, results from the literature [8, 9] are used in the current work
for comparative purposes.
Tapered attachment lugs with axially symmetric pin load are analysed in the
current work. Generally, lugs are tapered in order to take the pin load effectively.
Straight lug is the limiting case of zero taper, and the earlier studies [10] for straight
lug will be used for comparison purpose. Most of these contributions have not
addressed the issue of fatigue crack initiation and crack growth in these joints
including the effect of changing contact conditions. The present work highlights the
prognostic analysis of these lug joints and describes the computational methods to
estimate residual life of these critical joints. Finite element (FE) method is used for
this purpose with four-node quadrilateral elements in MSC PATRAN/NASTRAN
[11] FE software package. Through the stress analysis, critical locations are iden-
tified in these lug joints, and the fatigue crack initiation life is estimated with help of
Basquin’s relation, Goodman diagram and Miner’s rule [12, 13] for a prescribed
load cycle using locally developed software. Further Paris law including Elber
crack closure model [14] is adopted in computing crack growth life, and the fracture
parameters required for crack growth formulae are estimated using Modified Virtual
Crack Closure Integral (MVCCI) technique [15, 16].
2 Methodology
Tapered attachment lugs, with push fit where pin diameter is equal to the hole
diameter of the lug, are analysed. The lug joints are analysed for loads along
positive x-axis (called Pull) loads and along negative x-axis (called Push) loads for
different taper angles (Fig. 1). There is partial loss of contact for an infinitesimally
small pin load, and these contact/separation regions do not change with mono-
tonically increasing load level either in Pull or Push loads leading to linear elastic
problems. However, the regions of separation/contact change when the fatigue
loading is of Pull–Push type with R < 0.
Pin is assumed to be rigid and the lug is of aluminium T3 2024. The rigid pin
nearly corresponds to a steel pin in aluminium lug. Further, the interface between
the pin and the hole is assumed to be smooth. Fatigue analysis for the above lug
models for different taper angles is conducted for the Pull–Pull (Rp 0) and the
Pull–Push load cycles (with 0 > Rp > 1), and the fatigue crack initiation life is
estimated at critical locations identified from the stress analysis. Typical load cycles
applied are shown in Fig. 2. The variation of contact/separation region between the
lug and the pin for Pull and Push pin load cases is shown in Fig. 3. In case of push
fit, the contact/separation regions for any load level are nearly 180° [4].
2.1 Boundary Conditions and Finite Element Modelling
The region of contact/separation between the hole and the pin remains invariant
with monotonic increasing pin load, and this region of contact is nearly between
90°: 0°: −90° and 90°: 180°: −90° for Pull and Push pin loads, respectively, for
axial loading. In the remaining region, the pin and the hole boundary separate from
each other. The far end of the lug joints are supported by rollers, and the rigid body
motion is suppressed at point A in both x- and y-directions (Fig. 1). The boundary
conditions are as given below.
Region of contact ðhc Þ 90 : 0 : 90 Ux ¼ Uo ; srh ¼ 0 ð1Þ
Region of Separation ðhs Þ 90 : 180 : 90 rr ¼ 0; srh ¼ 0 ð2Þ
The inequality constraint in contact region is rr 0 and in the separation region

is Ur 0.
At the far end at x ¼ L Ux ¼ 0 and sxy ¼ 0 ð3Þ
Fig. 1 Typical tapered

attachments lug
Fig. 2 Typical load cycles applied on the lug
Fig. 3 Contact/separation regions c and s for pull and push loads
p
Z2
Equilibrium eqn : rr rt cos h dh ¼ P; þ P ¼ Pull load; P ¼ Push load
p2
ð4Þ
The regions of contact and separation are determined by the satisfaction of the
inequality constraints. Even though the region of separation/contact is specified
from literature as given in the above equations, the inequality constraints are
checked during the analysis.
For the case of Push loading, the region of contact changes to 900: 1800: −900
and the region of separation will be 90°: 0°: −90°. The boundary conditions in the
region of contact and separation remain the same except the regions contact/
separation are to be interchanged as in Eqs. (1) and (2) given above. Uo is the rigid
body displacement of the pin. Boundary conditions at the far end are given in
Eq. (3). Equilibrium equation for Pull and Push load case is given in Eq. (4). For
each specified values of the load P in x-direction, the equilibrium is checked by
summing up the x reactions at the far end.
Table 1 Convergence study of FEM

Tapered lug
(Taper angle 22.5°)
No. of nodes Normalized maximum stress (rmax/rbr) for the pin load, 1000 N
9112 4.785
10,008 5.196
10,856 5.442
11,789 5.686
12,052 5.812
12,071 5.8912
12,078 5.9304
12,082 5.9304
12,085 5.9306
12,089 5.9312
Finite element modelling of the above lug models is carried out in MSC
NASTRAN/PATRAN [11] finite element software package using four-node
quadrilateral shell elements. The load on the pin is applied by using RBE–2 ele-
ments in NASTRAN/PATRAN package. This imposes a multipoint constraint
(MPC) on all the nodes in contact with the hole boundary with total load on all the
nodes in x-direction is imposed as P, the specified total load on the pin. The
movement of the centre of the pin yields the rigid body displacement of the pin Uo
which can be estimated from the results. Convergence study was carried out on a
tapered lug of taper angle 22.5° to select suitable number of degrees of freedom.
Convergence study results are given in Table 1, and it was decided to run all
tapered lugs with 12,078 nodes. The typical finite element model of the taper
attachment lugs is shown in Fig. 4.
2.2 Fatigue Crack Initiation Life
Fatigue crack initiation life analysis is carried at the critical nodes identified from
the stress analysis using well-known equations, and they are reproduced here for the
sake of continuity. Typical crack initiation location is shown in Fig. 5. The elastic
part of the Basquin’s relation (Eq. 5) is used to estimate the damage per cycle,
considering the mean stress effect and S-N life approach for the respective fatigue
load cycles [12, 13] as given below.
" #ð1bÞ
1 ra
Nf ¼ ð5Þ
2 r0f rm
Fig. 4 FEM model of

tapered lug (taper angle 22.5°)
Fig. 5 Crack initiation

location
where b is fatigue strength exponent, and rf′ is fatigue strength coefficient [8].
Fatigue crack initiation life is estimated for the load ratio (Pmin/Pmax) Rp = 0,
Rp = −1 and Rp = −2 until the cumulative damage index at the critical node for
these applied load cycles becomes one. Damage index is estimated using Miner’s
rule [13] given in the Eq. (6).
X
N
ni
D¼ ð6Þ
i¼1
Ni
where n is the number of alternating stress of i types, and Ni is total number of

cycles to failure of i types, and damage occurs when the damage index reaches the
value of one.
2.3 Crack Growth Analysis
A crack which cannot be detected with 95% probability (2 mm [17]) is assumed at

the maximum tensile stress location identified from the stress analysis. Crack
growth configuration is shown in Fig. 6. Crack growth life is estimated using Paris
law updated with Elber crack closure effect [14] as given in the Eq. (7). C and m are
the material constants [18], a′ and b′ are the constants in the Elber [14] model.
da
¼ CðDKeff Þm ð7Þ
dN
where,
pffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
DKeff ¼ Kmax Kcl; Kcl ¼ rcl pa; Kmax ¼ rmax pac;
Keff ¼ ða0 þ b0 RÞKI ð8Þ
where rcl and Kcl are stress and SIF at crack closure load, respectively.
Fracture parameters are estimated using MVCCI (Modified Virtual Crack
Closure Integral) technique [15, 16]. Four-node quadrilateral elements near the
crack tip is shown in Fig. 7, from which strain energy release rate (GI) and stress
intensity factor (KI) are estimated using Eqs. (9) and (10), respectively, where E is
the Young’s modulus of the material. Assuming plane stress conditions
1
GI ¼ ½Fz w ð9Þ
2Da
pffiffiffiffiffiffiffiffiffiffiffiffiffi
KI ¼ GI E ð10Þ
Fig. 6 Crack growth

configuration
Fig. 7 Quad element at the

crack tip to estimate fracture
parameters
3 Results and Discussions
3.1 Variation of Maximum Tensile Stress
The effect of taper on the maximum tensile stress for both Pull and Push loadings is
shown in Fig. 8. The results for straight lug [10] are also plotted in these and
subsequent figures to visualize the results for the limiting case of b = 0. It is seen
from the figure that as expected the maximum tensile stress is linear with applied
loading confirming that the push fit case is a linear problem. However, the primary
result is that the maximum tensile stress decreases with the angle of taper for the
case of Pull loading. The same trend is followed even in case of Push loading,
wherein it is noticed that lug joint with higher taper angle posses lesser tensile
stresses followed the lower taper angles. Also it is important to notice that the
stresses are too low for Push load to that of Pull load. The results for straight lug
[10] are also plotted in the subsequent figures to visualize the results for the limiting
case of b = 0.
3.2 Stress Concentrations Effects
Stress concentration factor (SCF) (rmax/rbr where bearing stress rbr = P/2Rit and
t = thickness) due to applied pin load (both Pull and Push) for different Ro/Ri ratios
of the lug is estimated. Since the SCF is obtained by normalizing the maximum
stress value with bearing stress it could have a high value. So it is shown in a
logarithmic (natural log) scale along the y-axis. For Pull loading (Fig. 9a), it is seen
that with increasing Ro/Ri (keeping Ro constant and making the hole smaller) ratios
the stress concentration factor decreases due to increase in bearing stress as Ri
decreases (rbr a 1/Ri). From Fig. 9a, it is observed that the straight lug has higher
stress concentration factor in comparison with tapered attachment lugs for Pull
loading. Obviously, when the taper angle is increased keeping Ro constant, the load
Fig. 8 Variation of maximum tensile stress for applied pull and push load cases
flow will be smoother into a wider lug, and this causes lesser SCF. For the Push
load case (Fig. 9b) too, SCF decreases with increase in taper angle. It is also seen
that there is decrease in SCF with increase in Ro/Ri ratios as in case of Pull load.
Also there is significant drop in SCF in Push load case when compared with Pull
load. Figure 9c shows the comparison between empirical solution available in the
literature given in the Eq. (11) for the lug joints and the current results for tapered
lug for the taper angle of 22.5°. The results of FEM vary within the 2.5% of actual
values of literature and empirical solutions [9].

b0
0:6751000
rmax b0 R0
Kt ¼ ¼ 2:75 1 ð11Þ
rbr 135 Ri
3.3 Radial and Tangential Stress Distributions
Radial stress distributions (normalized with rbr) around the pin—the lug–hole
interface are plotted in (Fig. 10a, b) for Pull and Push loads, respectively, for
different taper angles and Ro/Ri equal to 2.5. For Pull loading (Fig. 10a), radial
Fig. 9 a SCF for pull load, b SCF for push load, c comparison of SCF for taper lug
stress distribution is throughout smaller for higher taper angles in the contact region
in comparison with smaller taper angles. A similar trend is seen in case of Push
loading also (Fig. 10b). The results taken from Ref. [10] straight lug show highest
radial stress distribution in the contact region in comparison with tapered lugs
Fig. 9 (continued)
irrespective of pin loading directions. The radial stresses for the case of Push loads
are larger compared to the case of Pull loads for the same load level.
Tangential stresses (hoop stresses) are computed around the pin–hole interface
and are normalized with the bearing stress for Pull and Push load cases (Fig. 11a, b).
In case of Pull load shown in Fig. 11a, there are two points at 90° and −90° where the
tangential stresses are the highest. At these maximum tensile tangential stress points,
straight lug has higher values in comparison with tapered lugs. These peak values
decrease with taper angle. Figure 11b shows the distribution of tangential stress
around the pin–hole interface for Push loading which also has two major peak points
of stress at same locations, but their magnitude is much smaller than for the case of
Pull load, which is again higher for straight lug and decreases with increase in taper
angle. The fatigue loading with Pull–Push type will be stressed alternating between
the two peaks from Fig. 11a, b with the configuration of contact/separation changing
when the load passes through zero value.
3.4 Fatigue Crack Initiation Life
Stress history at a point is monitored due to applied load/displacement for which the
fatigue analysis is carried out till the crack initiation. The critical locations are
identified through the stress analysis in particularly from the tensile tangential stress
Fig. 10 a Radial stress distribution at the hole boundary for pull load, b radial stress distribution
at the hole boundary push load
Fig. 11 a Tangential stresses around the hole boundary for pull load, b tangential stresses around
the hole boundary for push load
distributions around the lug–hole interface (Fig. 11a, b). From these distributions,
the maximum tensile tangential stress points are identified as the potential crack
initiation and crack growth locations. Figure 12 shows load cycles for Rp = 0 and
−1. Respective local stress variation due to these load cycles at the maximum
tensile stress locations are qualitatively shown in the figures. It is seen for both Pull
and Push loadings, the tangential stress at the critical locations is tensile, but of a
smaller magnitude for Push loads. It is observed that for Rp = −1, the local max-
imum tensile stress at the critical nodes is too small for the Push load in comparison
with Pull load, and hence in case of Rp = −1, the contribution towards the fatigue
failure is not significant. The crack initiation life is shown only for Rp = 0, and it is
verified that it does not vary in a significant manner with the negative Rp ratios.
The lug configurations are analysed for varying Pmax with Rp = 0, and the results
are compared for various taper angles. Initially, for lower loads, all the lugs possess
very high life and damage is negligible. With the progressively increasing load
Pmax, there is a drop in the fatigue life with increase in damage. Straight lug shows
less fatigue life in comparison with tapered lugs. The life estimation is shown only
for Rp = 0 in Fig. 13. It is seen that there is clear though not significantly large
increase in fatigue crack initiation life with increase in taper angle.
Fig. 12 Local stress variations due to applied fatigue load cycles [Rp = Pmin/Pmax]
Fig. 13 Fatigue crack initiation life for load ratio R = 0
3.5 Crack Growth Analysis
The contribution of Push load to crack growth is negligible compared to Pull load.
So the crack growth results are presented for the case of Rp = 0. The results are
mainly directed to bring out the effect of taper angle which is the main parameter in
the current paper. Initial crack length of 2 mm is assumed as mentioned earlier as
the crack length which can be detected with 95% confidence level. The analysis is
carried out for a fixed Pmax and for the load cycle (Rp = 0). The critical crack length
for unstable crack growth is plotted for different taper angles in Fig. 14, and the
crack growth life is shown in Fig. 15. It is seen that critical crack length for straight
lug is relatively small when compared with tapered lugs since the effect of stress
intensity factor range ðDKÞ decreases with taper angle. In Fig. 15, it is seen that
straight lug has the least crack growth life, and the crack growth life increases with
taper angle due to higher stress concentration factors in straight lug compared to
tapered lugs.
Fig. 14 Critical crack length for different taper angles of the lug
Fig. 15 Crack growth life

4 Conclusion
Lug joints play the key role in large-scale structural components which are used as
primary load-carrying members. Health monitoring of these lug joints seeks major
attention since their failure can be catastrophic. Tapered attachment lugs with rigid
pin are analysed for the push fit case where in contact/separation region do not
change with monotonically varying pin load. The major issue in the analysis is the
loss of contact between the pin and the hole. During load reversals, the contact
angles will change, and this has to be accounted in the analysis. Critical locations
where maximum tensile tangential stresses occur are identified from finite element
analysis.
The SCF with respect to bearing stress decreases in the case of both Pull loading
and Push loading with (Ro/Ri). The fatigue crack initiation life and crack growth life
are estimated from which one can derive the remaining life of the structure at any
given instant during fatigue loading. The stress concentration around the lug
decreases with taper angle. This results in tapered lugs with higher taper angle show
that better crack initiation and crack growth life are achievable when compared with
straight lugs.
Acknowledgements The work presented in this paper is carried out under the project Structural
Damage Mitigation sanctioned by Aeronautical R&D Board. The investigators profusely thank the
Structures Panel, Aeronautical R&D Board, India, for sanctioning the project.
References
1. M. Esperon-Miguez, P. John, I.K. Jennions, A review of integrated vehicle health

management tools for legacy platforms: challenges and opportunities. Int. J. Prog.
Aerosp. Sci. 56, 19–34 (2013)
2. S. Gopalkrishnan, M. Ruzzene, S. Hanagud, Computational techniques for structural health
monitoring. Springer, London (2011)
3. P.D. Mangalgiri, T.S. Ramamurthy, B. Dattaguru, A.K. Rao, Elastic analysis of pin joints in
plates under some combined pin and plate loads. Int. J. Mech. Sci. 29(2), 577–585 (1987)
4. W.G. Bickley, The distribution of stress round a circular hole in a plate. Philos. Trans. R. Soc.
Lond. Ser. A, 227(128), 383–415
5. H.G. Harris, I.U. Ojalvo, R.E. Hooson, Stress and deflection analysis of mechanically
fastened joints. AFFDL-TR-70-49 (1971)
6. A.K. Rao, Elastic analysis of pin joints. Comput. Struct. 9, 125–144 (1978)
7. B. Dattaguru, Effect of non-linear behaviour of joints on the damage tolerance analysis in
aerospace structures. Proc. Indian Natl. Sci. Acad. (Special Issue—Part A) 79(6), 553–562
(2013)
8. K. Kathiresan, T.R. Brussat, J.L. Rudd, Damage tolerance assessment of aircraft attachment
lugs. J. Eng. Fract. Mech. 23(6), 1067–1084 (1986)
9. T.M. Hsu, Analysis of cracks at attachment lugs. J. Aircraft 18(9), 755–760 (1981)
10. L. Chikmath, Prognostic analysis of fastener in lug joints. Ph.D. Thesis, Department of
Aerospace Engineering, Jain University, Jakkasandra, 562112, India (2017)
11. The MSC NASTRAN Documentation. MSC.Software Corporation, CA, USA (2005)
12. S. Suresh, Fatigue of materials (Cambridge University Press, UK, 1998)

13. T.L. Anderson, Fracture mechanics: fundamentals and applications (CRC Press, Taylor &
Francis Group, Boca Raton, 2005)
14. W. Elber, The significance of fatigue crack closure. In: Damage tolerance in aircraft structure.
ASTM STR 486, 230–242 (1971)
15. E.F. Rybicki, M.F. Kanninnen, A finite element calculation of stress intensity factors by a
modified crack closure integral. J. Eng. Fract. Mech. 9, 931–938 (1977)
16. T.S. Ramamurthy, T. Krishnamurthy, K. Badrinarayana, K. Vijaykumar, B. Dattaguru,
Modified crack closure integral method with quarter point elements. Mech. Res. Commun. 13
(4), 179–186 (1986)
17. J. Haapalainen, E. Leskela, Probability of detection simulations for ultrasonic pulse-echo
testing. In: 18th World Conference on Non-destructive Testing, Durban, South Africa,
pp. 16–20, Apr 2012
18. J. Gillbert Kaufman, Properties of aluminum alloys—fatigue data and the effects of
temperature-product form and processing. ASM International Data Handbook (2008)
Experimental Study on Fatigue–Vibration
Interactions of Cyclically Loaded FRP
Composite Cantilever Beams Made
of Glass/Carbon-Layered Arrangements
R. Murugan, R. Ramesh and K. Padmanabhan
Abstract Woven fabric-reinforced polymer composites are widely used in the

structural applications like aircraft, automotive, marine and machine tool structures
due to their excellent integrity, conformability and balanced mechanical properties
within the fabric plane. It is observed that the structural members used in such
applications experiencing vibration and considerable cyclic loading in operations
are inevitable in dynamic conditions. Therefore, the study of the interaction of
vibration characteristics of composite structures due to the fatigue loading needs
much attention to meet the industrial requirements. In this study, the interaction of
fatigue loading and vibration characteristics of epoxy-based glass/carbon hybrid
beams was evaluated by measuring the variation in the modal stiffness. The two
types of hybrid layer arrangements using glass and carbon woven fabrics, GCCG
and CGGC, are preferred for the present investigation, by considering the balanced
modulus property in longitudinal and transverse direction of the specimens.
A displacement-controlled fully reversed flexural fatigue testing rig was designed
and developed with required control circuits and test specimens to find the inter-
action of flexural fatigue loading on vibration characteristics of fibre-reinforced
composite beams. The vibration characteristics and interaction of fatigue loading on
vibration characteristics of thin-walled woven fabric glass/carbon hybrid composite
beams made of different stacking sequences for promoting good dynamic stability
were investigated.

Keywords Hybrid composite Stacking sequence Vibration characteristics
Interaction of fatigue loading on vibration characteristics
R. Murugan
Panimalar Institute of Technology, Poonamallee, Chennai 600123, India
R. Ramesh (&)
Sri Venkateswara College of Engineering, Sriperumbudur 602117, India
e-mail: rameshraja@svce.ac.in
K. Padmanabhan
School of Mechanical and Building Sciences, VIT University, Vellore 632014, India

https://doi.org/10.1007/978-981-10-7197-3_48
580 R. Murugan et al.
1 Introduction
Fibre-reinforced polymeric (FRP) composites are widely used in the structural

applications like aircraft, automotive, marine and machine tool structures due to its
high strength/stiffness-to-weight ratio. The recent fabrication of structures in
polymer composites contains reinforcing fibres in the form of fabrics and properly
formulated polymeric matrices. In woven fabrics, the fibre rovings are interlaced
with two mutually orthogonal directions to one another which promote excellent
integrity and conformability and balanced mechanical properties within the fabric
plane. Due to these advantages, woven fabric composites become popular and
widely used in all such mentioned applications. It is also observed that the structural
members used in such applications experiencing vibration and considerable cyclic
loading in operations are inevitable in dynamic conditions. Therefore, it is very
important to understand and evaluate the influence of cyclic loading on vibration
characteristics in terms of improving its dynamic stability of composite structures.
The anisotropic nature of FRP material makes the evaluation of its fatigue
behaviour more complicated. This heterogeneous and anisotropic nature of FRP
leads to the formation of different stress levels within the material during cyclic
loading and thereby forms the various combinations of damage modes such as
matrix cracking, fibre breakage, delamination, debonding and ply failure [1]. It is
commonly accepted that FRP degradation of performance during cyclic loading is
associated with a reduction in the composite material stiffness [2]. Research on
fatigue of composite materials is conducted earlier by performing tension–tension
fatigue and tension–compression fatigue, since damage is developing more or less
equally in all layers of the composite specimen.
Very few authors only reported the bending fatigue experiments. Weber and
Schwartz [3] monitored the bending fatigue behaviour of unidirectional carbon/
epoxy composite strands using the changes in resistivity and in bending modulus
[3]. Van Paepegem and Degrieck [4] investigated the fatigue performance of plain
woven glass/epoxy composite specimens with two different configurations such as
[#0°]8 and [#45°]8. They showed that the two specimen types, although being made
of the same material, have quite different damage behaviour and that the stiffness
degradation follows a different path [4]. Belingardi et al. [5] performed
displacement-controlled bending fatigue tests with stress ratio of R = 0.1 up to
1 million cycles on a carbon–glass hybrid composite beams. It was found that the
amount of stiffness reduction was a function of the magnitude of applied fatigue
loading on the specimen [5]. Lian and Yao [6] experimentally studied the fatigue
behaviour of six kinds of E-glass/epoxy composite laminates with different lay-up
sequences and obtained the stiffness degradation models in longitudinal, transverse
and in-plane shear direction [6].
Most of the earlier investigations have been concerned with damage-induced
changes in stiffness as found from observed variations in resonant frequencies of
composite specimens. Little work was done in continuously monitoring the vari-
ation in stiffness as a function of the number of cycles to failure. The main objective
Experimental Study on Fatigue–Vibration Interactions … 581
of the present work is to investigate the vibration characteristics and the interaction
of fatigue and vibration characteristics on thin-walled woven fabric glass/carbon
hybrid composite beams used for automotive and aircraft structural applications.
2 Fabrication of Hybrid Composite Specimen
The hybrid composite specimens were fabricated using hand lay-up method.
E-Glass and T300 carbon fibres in plain woven fabric form with areal density of
600 g/m2 were used as reinforcement. Thermoset resin of grade LY556 was pre-
ferred as matrix. A uniform volume fraction of Vf = 0.5 was controlled for the two
types of hybrid laminates. The casting was cured for 24 h in a compression
moulding machine with a nominal pressure of 2.5 MPa at room temperature. After
curing, the hybrid laminates were cut into beam of size 250 25 2.4 mm.
Two different layer sequences H1 and H2 as shown in Table 1 were preferred in
fabricating hybrid laminates. In hybrid layer arrangement, H1, the two inner layers
are preferred as carbon fabrics and two outer layers as glass fabrics. For H2 hybrid
arrangement, the layering control was done by placing two carbon fabrics in outer
layers and two glass fabrics in inner layers. The layer arrangement of hybrid
laminate was selected in consideration of balanced modulus property along the
longitudinal and transverse direction with respect to neutral axis of the plate.
Figure 1 shows the fabricated four-layered woven fabric hybrid composite beams
made up of glass and carbon fabrics. Table 1 shows the dimensions and symbols
used for the hybrid composite specimens considered.
Table 1 Symbol used and its corresponding layer arrangement of glass/carbon hybrid composite
beams
Symbol Specimen description Layer sequence Dimensions (l * w * t)
(mm)
H1 Four-layered hybrid laminate 250 * 25 * 2.4
Outer—glass and inner—carbon GCCG
H2 Four-layered hybrid beam 250 * 25 * 2.4
Outer—carbon and inner—glass CGGC
l—length; w—width; t—thickness
Fig. 1 Image showing the two types of woven fabric glass/carbon hybrid composite beams
fabricated by hand lay-up technique
3 Experimental Conditions
3.1 Static Deflection and Flexural Test
For measuring the static deflection, the beam specimen is fixed firmly by using
clamping device on the work table. The standard weights are applied at the free end
in different magnitude levels. This experimental arrangement is used to find the
optimum span length for fatigue study of the hybrid samples and also to measure
the loss of stiffness of the hybrid beams subjected to fatigue load.
Flexural test was performed as per ASTM D790 for all samples with a recom-
mended span-to-depth ratio of 16:1 [7]. The size of the specimen is 127 mm
12.5 mm 2.4 mm. The flexural test was conducted on closed-loop
servo-hydraulic universal testing machine INSTRON 3382 with feed rate of
1.2 mm/min at room temperature. Five specimens were tested for each type of
laminate, and the average value was calculated.
3.2 Impulse Response Test
The vibration characteristics of the hybrid composite beams measured using

impulse excitation technique [8]. Figure 2 shows the arrangement and equipment
involved in conducting the vibration test. For conducting the vibration test, one end
Fig. 2 Photograph showing experimental test set-up for vibration measurement (1) composite
specimen (2) impact hammer (B&K Type 5800B4) (3) accelerometer (B&K Type 3055B2)
(4) data acquisition card (B&K Type Photon+) (5) PC with RT Pro Software showing FRF
of the specimen is rigidly fixed by work clamping device and the other end was left
as free end for offering fixed-free boundary condition. A beam length of 136 mm
was set as suspended length for impulse testing. An instrumented impact hammer
(B&K Type 5800B4) with a load cell was used to excite induce the excitation of
flexural vibrations of the beam. The response of the structure was measured using
an accelerometer (B&K Type 3055B2) at the free end of the beam as shown in
Fig. 2. The signals of the impact hammer and the accelerometer were amplified and
digitised through a high-speed data acquisition board (B&K Type Photon+) and
finally processed using FFT software (RT Pro Photon+), to obtain frequency
response functions (FRFs) of the tested component. The measured FRFs were used
to extract the modal frequency values of each vibration mode.
3.3 Imposing Reversed Cyclic Bending
The experimental set-up developed specially for flexural fatigue tests on cantilever
beam specimens is shown in Fig. 3. The outcoming shaft of the motor has a
rotational speed of 1440 rpm. The power is transmitted by a single-stage gearbox to
a second shaft, which provides a fatigue testing frequency of 10 Hz. The influence
of frequency on fatigue loading can be assumed to be small in this range of values
[8]. To investigate the effect of fatigue load on vibration characteristics of the two
different stiffness samples, H1 and H2, an optimum span length of the composite
specimens was arrived as 136 mm by conducting static deflection test. In the
present investigation, stress-controlled fatigue test is preferred to compare the effect
Fig. 3 Experimental test set-up for applying completely reversed bending fatigue load on hybrid
composite specimen
Table 2 Experimentally evaluated flexural properties of woven fabric laminates [9]

Specimen Flexural Flexural Area MOI Flexural rigidity
strength (MPa) modulus (GPa) (mm4) (N mm2) 106
H1 491 21.23 28.8 0.62
H2 507 42.08 28.8 1.23
of fatigue load on vibration characteristics of the hybrid beams with two different
stacking sequences H1 and H2. Since the flexural strength of two hybrid beams is
approximately equal to 500 ± 10 MPa, 20% of flexural strength of hybrid beams,
i.e. 100 MPa, is taken as the limiting stress for stress-controlled fatigue test.
However, the flexural modulus values are different for the hybrid beams H1 and H2
as shown in Table 2; to achieve the reference stress range for the two hybrid beams,
deflections of the beam set in the test rig need to be controlled properly. Deflections
for the two hybrid beams are theoretically calculated by using the standard beam
deflection formula and bending equation. The experimental test conditions for
fatigue loading are as follows:
Test specimen type H1 (GCCG) and H2 (CGGC)

Free length, L 136 mm
Test frequency, f 10 Hz
Induced deflection, d ± 17 mm for H1 and ±12 mm for H2
Induced bending stress, rb −100 to +100 MPa
Stress ratio, R = rmax/rmin −1
In this experiment, the vibration characteristics were measured after every 5,000
cycles up to a limited number of 200,000 cycles. The modal frequency values were
measured experimentally by impulse excitation technique for all 5,000 cycles and
tabulated.
4.1 Flexural Properties of Hybrid Composite Laminates
The experimental flexural properties of hybrid samples are reported in Table 2

based on the earlier investigation by Murugan et al. [9]. Figure 4 shows the load–
deflection curves of hybrid composite laminates obtained by flexural test. Flexural
modulus (Eb) and flexural rigidity (EbI) were evaluated for the tested laminates and
reported in Table 2. The variation in the flexural rigidity of hybrid laminates
confirms that the stacking sequence controls the flexural rigidity. In case of hybrid
Fig. 4 Load–deflection
curves of hybrid composite
laminates obtained from
flexural test
laminates, the layer arrangement H2 offered a noticeable increase in flexural

modulus as compared to the other arrangement H1. The load–deflection charac-
teristic curves of H1 and H2 samples tested confirm its variation in performance. In
transverse bending, the flexural modulus of outer layer of composite beam plays an
important role [10]. Hybrid laminate with carbon fibre plied as external layer (H2)
offers high resistance against bending.
4.2 Analysis of Static Deflection
Loss of stiffness due to fatigue loading of the hybrid beams, H1 and H2, was
measured by static deflection test set-up. Deflection of the hybrid beam at free end
for common load of 0.5 kg was measured before and after subjected to 2 105
cycles of fatigue load. Table 3 shows the corresponding deflection values and
calculated beam stiffness values. The percentage loss of stiffness evaluated from
experimental deflection after 2 105 cycles found to be 76.03% for H1 beam and
45.14% for H2 beam. Table 3 reveals that loss of stiffness of hybrid beam H1 is
nearly twice that of hybrid beam H2. It reveals the improved dynamic stability of
hybrid beam H2 over the other hybrid arrangement H1.
Table 3 Stiffness degradation evaluated and compared at known 0.5 kgf load level
Specimen Measured deflection at Stiffness performance at Stiffness
0.5 kgf load (mm) 0.5 kgf load (N/mm) degradation (in %)
At 0 After 2 105 At 0 After 2 105
cycle cycle cycle cycle
H1 3.88 16.2 1.264 0.303 76.03
H2 2.94 5.36 1.668 0.915 45.14
4.3 Evaluating the Modal Stiffness of Hybrid Composite

Beams Subjected to Limited Fatigue Loading
Figure 5 shows the comparison of frequency response plots of hybrid beams, H1

and H2, evaluated at various interval of fatigue loading. It reveals that the vibration
response of hybrid beams changed due to fatigue loading and lowers the modal
frequency values. Frequency response function plots of both beams were shifted
towards left due to fatigue load which indicates the fall in vibration response of
hybrid beams tested. Table 4 shows the variation of vibration characteristics such as
modal frequency of hybrid beams H1 and H2 for first three modes arrived from
experimentally recorded FRF function for all 5,000 cycles of fatigue loading. The
frequency at which the resonant set attained for hybrid beam H2 shows higher range
than other beam H1 for all levels of fatigue loading. It reveals that the stacking
sequence of glass and carbon fabric layers influences appreciably its vibration
characteristics during fatigue loading. Figure 6 shows the modal stiffness decay in
(a)
(b)
Fig. 5 Comparison of FRF plots of hybrid beams obtained at 0 cycles, after 1 105 cycles and
2 105 cycles of fatigue loading a FRF plots of H1 beam b FRF plots of H2 beam
Table 4 Variation of modal frequency values at successive modes of H1 and H2 hybrid beams
subjected to a limited fatigue cycling of 2 105 cycles
No. of cycles (105) Mode 1 Mode 2 Mode 3
Frequency Frequency (Hz) Frequency (Hz)
(Hz)
H1 H2 H1 H2 H1 H2
0.0 33.9 46.9 302.5 386.8 858.5 962.4
0.5 23.9 45.4 265.0 383.5 632.5 888.8
1.0 22.5 43.1 266.3 366.2 648.5 823.2
1.5 22.5 38.1 255.0 342.9 620.0 805.7
2.0 20.0 36.6 248.5 339.8 698.5 883.8
Fig. 6 Comparision of experimental modal stiffness decay curves at successive modes of hybrid
beams H1 and H2 during fatigue loading
woven fabric hybrid composite beams, H1 and H2, due to limited fatigue loading.
A rapid drop in modal stiffness was observed at the beginning stage of the fatigue
loading for both the hybrid beams and followed by a region of gradual reduction
region.
The bending stiffness of composite structures decreased steadily with increasing
number of fatigue cycles because of diminishing elastic properties of the constituent
materials during fatigue [11, 12]. The percentage loss of stiffness evaluated from
experimental static deflection test of hybrid beams after limited fatigue loading in
Sect. 4.2 also confirms this attribute. The degree of degradation of modal stiffness
of H1 specimen during cyclic loading in all successive modes was higher than the
other hybrid layer arrangement H2. In the H1 specimens, several fibres are broken
at the clamped end of the specimen and a sort of ‘hinge’ is formed [13]. On the
other hand, the damage level at the clamp end of H2 specimen is not observed to be
as severe as the H1 specimen. High flexural strength and stiffness of the hybrid
layer arrangement H2 (Table 3) offer more resistance against the flexural bending
than the other layer arrangement H1. In this study, effect of fatigue loading on the
free vibration characteristics of hybrid beams was both understood and evaluated in
terms of the modal stiffness decay. It was found that the change in vibration
characteristics of composite specimens could be considered as an indication of

variation in the strength and stiffness of the composite beams due to fatigue loading.
4.4 SEM Observation and Analysis of Damage Over Glass/

Carbon Hybrid Beams After Subjected to Limited
Fatigue Load
A scanning electron microscopy (SEM) analysis was performed on samples from

tested specimens under completely reversed bending fatigue loading. Figure 7
shows the closer view of fractured surfaces obtained from optical microscopy at the
fixed end after the fatigue tests of the hybrid beams H1 and H2, respectively. From
these macroimages of failed surfaces, it is revealed that damage during bending
fatigue is immense at the fixed end due to the occurrence of high bending stresses.
The samples were obtained from this fixed end of the hybrid composite beams H1
and H2 for SEM analysis. SEM images of failed surfaces are shown in Fig. 8.
Earlier investigation by Dew-Hughes and Way [14] revealed that the possible
stages of fatigue damage in FRP materials are local debonding, joining up of
debonds, matrix cracking and fibre fracture where debonding is considered as the
main criterion for failure [14].
Figure 8a, b show SEM images of surface damage induced at the fixed end of
hybrid beam H1 due to limited cycles of fatigue loading. During fatigue loading,
first fibre breakage occurs at stresses exceeding the strength of the weakest fibre. An
isolated fibre breakage causes shear stress concentration at the interface around the
tip of the broken fibre. The interface may then fail, leading to debonding of the fibre
from the surrounding matrix. Figure 8a shows the fibre–matrix debonding, severe
delamination of the matrix and more numbers of longitudinal fibre breakages near
Span Length Span Length
Fig. 7 Macroimages showing failure surfaces of hybrid beams H1 and H2 at the fixed end after
limited fatigue load
Fig. 8 SEM images of the fractured surfaces of hybrid composite laminate H1 after 2 105
cycles of flexural fatigue loading
the fixed end of hybrid beam H1. Figure 8b reveals the transverse fibre breakage of
H1 specimen and transverse matrix striations in agreement with fatigue loading.
The presence of high strength carbon fibres at the surface of H2 beam promotes the
fibre breakage and crack propagation predominantly through the epoxy matrix.
SEM images of the fatigue-damaged surfaces of hybrid composite beams revealed
that hybrid sample H1 (GCCG) had relatively severe fibre breakages than H2
(CGGC) samples. The microstructural damages such as matrix cracking, fibre
breakage and fibre–matrix debonding formed during fatigue loading caused the
degradation of modal stiffness in the initial stage. SEM analysis of failed surfaces of
hybrid beams H1 and H2 confirms the presence of all types of damages due to
fatigue loading. Earlier research on fatigue behaviour of FRP composites by
Manjunatha et al. [15], Wicaksono and Chai [16] and Sudevan et al. [17] showed a
similar trend for the stiffness decay curve [15–17].
5 Conclusions
Completely reversed bending fatigue experiments were successfully conducted to

understand the interaction between fatigue and vibration of the hybrid beams H1
and H2. The modal stiffness of both hybrid composite beams decreases with
increasing number of fatigue cycles. Comparative results showed that the degree of
degradation of modal stiffness in H1 beam is higher than the other arrangement H2.
The stiffness loss evaluated by experimental static deflection test for fatigue
imposed hybrid beam also confirms the appreciable difference in stiffness degra-
dation. The morphology study clearly revealed that the localised variation in
damage intensity among the two different stacking sequences caused the difference
in vibration performance. Evaluation of free vibration characteristics of hybrid
beams before and after fatigue damage shows that the carbon fabric plied as
outermost layers offered enhanced dynamic stability than the outer layers replaced
by low modulus glass fabric (GCCG). The obtained fundamental layering sequence,
CGGC, is effective and beneficial for structural applications and should be extended
by this plying form condition to meet out the designed dimension.
References
1. P.K. Mallick, Fibre Reinforced Composites, Materials and Manufacturing and Design. CRC
Press, Taylor & Francis Group (2008)
2. Raif Sakin, Irfan Ay, Ramazan Yaman, An investigation of bending fatigue behavior for
glass-fiber reinforced polyester composite materials. Mater. Des. 29, 212–217 (2008)
3. Ingo Weber, Peter Schwartz, Monitoring bending fatigue in carbon-fiber/epoxy composite
strands: a comparison between mechanical and resistance techniques. Compos. Sci. Technol.
61, 849–853 (2001)
4. W. Van Paepegem, J. Degrieck, Experimental set-up for and numerical modelling of bending
fatigue experiments on plain woven glass/epoxy composites. Compos. Struct. 51, 1–8 (2001)
5. G. Belingardi, M.P. Cavatorta, C. Frasca, Bending fatigue behavior of glass–carbon/epoxy
hybrid composites. Compos. Sci. Technol. 66, 222–232 (2006)
6. Wei Lian, Weixing Yao, Fatigue life prediction of composite laminates by FEA simulation
method. Int. J. Fatigue 32, 123–133 (2010)
7. ASTM Standard D790-10, Standard Test Method for Flexural Properties of Unreinforced
and Electrical Insulating Materials. ASTM International, West Conshohocken, PA (2003)
8. D.J. Ewins, Modal Testing: Theory, Practice and Application. Research Studies Press,
Hertfordshire (2000)
9. R. Murugan, R. Ramesh, K. Padmanabhan, Investigation of the mechanical behavior and
vibration characteristics of thin walled glass/carbon hybrid composite beams under a
fixed-free boundary condition. Mech. Adv. Mater. Struct. 23(8), 909–916 (2016)
10. K.S Pandya, C. Veerraju, N.K. Naik, Hybrid composite made of carbon and glass woven
fabric under quasi-static loading. Mat. Design 32, 4094–4099 (2011)
11. S. Mahdi, A.J. Kinloch, F.L. Matthews, M.A. Crisfield, Fatigue performance of undamaged
and repaired composite sandwich beams. Proc. Inst. Mech. Eng. Part L: J. Mater. Design
Appl. 217, 229–246 (2003)
12. A. El Mahi, A. Bezazi, Describing the flexural behaviour of cross-ply laminates under cyclic
fatigue. Appl. Compos. Mater. 16, 33–53 (2009)
13. W. Van Paepegem, I. De Baere, E. Lamkanfi, J. Degrieck, Monitoring quasi-static and cyclic
fatigue damage in fibre-reinforced plastics by Poisson’s ratio evolution. Int. J. Fatigue 32,
184–196 (2010)
14. D. Dew Hughes, J.L. Way, Fatigue of fibre—reinforced plastics: a review. Composites 4(4),
167–173 (1973)
15. C.M. Manjunatha, Taylor A.C. Sprenger, A.J. Kinloch, The tensile fatigue behaviour of a
glass fibre reinforced plastic composite using a hybrid toughened epoxy matrix. J. Compos.
Mater. 44(17), 2095–2109 (2010)
16. S. Wicaksono, G.B. Chai, A review of advances in fatigue and life prediction of
fibre-reinforced composites. J. Mater. Design Appl. 227(3), 179–195 (2012)
17. Deepika Sudevan, Raghu V. Prakash, M. Kamaraj, Post-impact fatigue response of CFRP
laminates under constant amplitude and programmed FALSTAFF spectrum loading. Procedia
Eng. 101, 395–403 (2015)
Investigation of Cyclic Stress–Strain
Behavior in FCC Single Crystals
Nipal Deka and Krishna N. Jonnalagadda
Abstract The plastic behavior in polycrystalline materials depends on the defor-

mation of individual grains or single crystals. In this work, cyclic stress–strain
behavior was studied in single crystal FCC material to understand the effect of
strain localization. Numerical studies were accomplished utilizing crystal plasticity
finite element method (CPFEM) on a single crystal with a notch. Under, two types
of localization modes, namely growth of extrusion–intrusion bands and the strain
accumulation in shear localized regions near the crack tip, are commonly noticed.
The process of continuous strain accumulation in one certain direction is often
termed as strain ratcheting. Importantly, this particular behavior is generally
counted as one of the major causes of fatigue damage and also considered as very
vital in understanding the fatigue crack nucleation and its propagation. In the
current work, the strain ratcheting behavior of FCC single crystals with a notch
subjected to cyclic loads was studied using CPFE methodology with suitable
incorporation of nonlinear kinematic hardening law.
Keywords Cyclic loading T-stress FCC single crystal CPFE

Nomenclature
c_ 0 Reference shear rate
sa Resolved shear stress of the slip system a
va Back stress for the slip system a
ga Critical resolved shear stress
m Rate sensitivity parameter
hab Hardening matrix
h0 Initial hardening rate
hs Saturation hardening rate
s0 Initial strength of the slip system
N. Deka (&) K. N. Jonnalagadda

Department of Mechanical Engineering, Indian Institute of Technology Bombay,
e-mail: nipal_deka@iitb.ac.in

https://doi.org/10.1007/978-981-10-7197-3_49
594 N. Deka and K. N. Jonnalagadda
ss Saturation strength of the slip system

E Young’s modulus
t Poisson’s ratio
K1 Stress intensity factor
T T-stress
log kp1 Maximum principal logarithmic plastic strain
1 Introduction
Metals are most extensively used in many engineering applications such as aero-
space structures, automobile components, nuclear reactors, and
micro-electromechanical systems. At the microscopic level, all these metals consist
of crystals, which are mechanically anisotropic. Constitutive behavior at the scale of
the microstructure plays a significant role in many important phenomena including
fatigue failure, fracture. It is also well acknowledged that factors like single crystal
behavior and distribution of grain orientation control the local stress–strain field [1].
While the first one controls the microscopic stress–strain response, second one
controls the macroscopic response of metals. It is well understood that under fatigue
loading microscopic deformation results in the nucleation and growth of cracks. At
the microscale level, crack growth occurs through grains or along grain boundary in
more commonly employed polycrystalline materials. The fracture behavior in
polycrystalline materials is governed by the stress and plastic strain fields existing
near a stationary crack tip in elastic–plastic solid [2]. And if the crack opening
displacement is far smaller than the size of the grain, the process zones are limited
within a single grain [2]. The need to understand the fracture mechanisms in
polycrystalline materials under fatigue loading also requires the study of plastic
deformation around the crack tip in a single crystal. An enhanced outlook of the
deformation and stress fields near the crack in a single crystal would certainly assist
the appreciation and forecast of the fracture of those materials under fatigue
loading. Therefore, the study of consequences of local stress–strain field on phe-
nomena like fatigue and fracture in the context of single crystal is very essential so
as to forecast the overall behavior of the polycrystal. Beyond nucleation, the growth
of fatigue cracks is also affected by the stress state near the crack tip. In order to
understand the effect of constraint on the crack tip fields, mode I finite deformation
elastoplastic simulations need to be conducted.
Generally, two types of events namely the growth of extrusion–intrusion bands
[3] and the strain accumulation in shear localized regions near the crack tip [4] are
commonly noticed under cyclic loading. The process of continuous strain accu-
mulation in one certain direction is often termed as “strain ratcheting” [3]. Most
importantly, this particular behavior is generally accounted as the major cause of
fatigue damage and considered as very vital in understanding the fatigue crack
Investigation of Cyclic Stress–Strain Behavior … 595
nucleation and its propagation. In the current work, cyclic stress–strain behavior for
aluminum single crystal under zero-tension cyclic load was studied with crystal
plasticity finite element (CPFE) framework with suitable incorporation of nonlinear
kinematic hardening rule. By including the T-stress in the modified boundary layer
simulation, the effect of constraint on the development of stress and strain fields
under cyclic load was analyzed.
2 Crystal Plasticity Model
The computational advantages of crystal plasticity finite element (CPFE) models to

solve crystal plasticity problems under complex loading conditions and deformation
mechanisms have developed CPFE as one of the most versatile tools. The flexibility
to incorporate a variety of constitutive formulations for plastic flow and hardening
enables CPFE to capture small-scale deformation mechanisms like slip, twining, in
simulation.
2.1 Constitutive Model
The cyclic evolution of back stress component was modeled implementing the flow
rule as [5],
a
s va 1=m

c_ ¼ c_ 0
a sign ðsa va Þ ð1Þ
ga
where sa was resolved shear stress of the slip system a, va represented the back
stress component of the slip system a, and initial value of it was chosen to be zero.
The term c_ 0 was the reference shear rate. The exponent m was the rate sensitivity
parameter. The value of m tending to zero reduced the model to be rate independent
and ga could be interpreted as critical resolved shear stress.
Rate of hardening of slip system will depend upon shear rate on slip systems,
which was given as follows:
X
12
g_ a ¼ hab c_ b ð2Þ
b¼1
The term hab was the hardening matrix. The diagonal terms of this matrix
signified self-hardening while non-diagonal terms signified latent hardening. Thus,
hardening matrix captured the interaction between different slip systems.
PAN model [6] was used to define isotropic hardening which was implemented
as follows:

hab ¼ hðcÞ q þ ð1 qÞdab ð3Þ
where hðcÞ takes the form given by [6],

ðh0 hs Þc
hðcÞ ¼ hs þ ðh0 hs Þsech2 ð4Þ
ss s0
with c was the total slip accumulated on all the active slip systems and q was the
latent hardening ratio. The initial and the saturation hardening rates were repre-
sented by material parameters h0 and hs , respectively. s0 and ss represented the
initial and saturation strength of the slip system, respectively.
The evolution of back stress on each slip system, va , was modeled according to
the nonlinear kinematic hardening rule [5],
v_ a ¼ Crate ðCsat c_ a va jc_ a jÞ no sum on a ð5Þ
where the rate of growth and the saturation of the back stress were controled by
Crate and Csat , separately.
2.2 Numerical Simulation Using CPFEs
A large circular disk along with a notch is modeled considering a very large value
of the ratio of the radius of the disk to the notch diameter, Rb00 such as 70,000 with
notch diameter, b0 ¼ 1 unit [2]. The displacement components based on the elastic
mode I, plane strain crack tip field [7] are applied on the outer boundary of the
circular disk which are functions of stress intensity factor, K1 and T-stress, are given
by,
1
R0 2
1þt h h 1 t2
u1 ¼ K1 cos j 1 þ 2 sin2 þ TR0 cos h ð6Þ
2p E 2 2 E
1
R0 2
1þt h h tð 1 þ tÞ
u2 ¼ K1 sin j þ 1 2 cos2 TR0 sin h ð7Þ
2p E 2 2 E
where E and t are Young’s modulus and Poisson’s ratio and j ¼ 3 4t for plane
strain. In the present simulation, a FCC single crystal (aluminum) was considered
and the crack geometry was chosen to be (010) [101] with the crack on plane
(010) and the crack propagation direction to be [101], which was same as Rice’s
analytical analysis [8].
Cyclic displacements were applied on the remote boundary where KI and T-
stress were varied from zero to Kmax and zero to Tmax , respectively, as
displayed in
pffiffiffiffiffi
Fig. 1b. The normalized value for Kmax was considered to be Kmax = s0 b0 ¼ 80
Investigation of Cyclic Stress–Strain Behavior … 597
Fig. 1 a Circular disk used in the FE analysis. b Displacement applied in the modified boundary
layer analysis for cyclic case
[2] and three different values of normalized Tmax , Tmax =s0 ¼ 1, 0 and +1, which
include the T-stress span for numerous fracture geometries [9].
A User Material (UMAT) code was written with finite element analysis software,
ABAQUS [10] by implementing these flow rules and hardening laws. Results of the
finite element analysis are presented in the following section.
3 Results and Discussions
Cyclic evolution of plastic strain fields close to the crack tip is presented in Fig.
2
by plotting the maximum principal values of logarithmic plastic strain, log kp1
simulated after completing 20 cycles for three values of T-stress. It can be observed
that the negative values of T-stress affected drop of constraint around the crack tip
with the increase of plastic strain accumulation. Strain accumulation was quite
substantial in the localization band at 55° in comparison to the other two bands at
90° and 125°. Very small accumulation was observed in the band oriented at 90° for
negative magnitude of T-stress. However, there was fairly noticeable strain accu-
mulation at 90° and 125° band for nonnegative T-stress. In fact, the contribution of
125° band in the plastic strain distribution was more visible for the T ¼ s0 case.
Again, there was barely any plastic strain near the crack face for negative T-stress,
which was certainly not the case for zero and positive T values. It was also wit-
nessed that the strain accumulation along the slip band at 55° for negative T-stress
was relatively high in magnitude and was reduced with the increase of T-stress. This
kind of accumulation of strain in the slip bands in one direction with number of
cycles was affiliated to “strain ratcheting.”

Fig. 2 Contour plots of plastic strain, log kp1 (for three different cases of T, a T ¼ s0 , b T ¼ 0,
and c T ¼ s0 after 20 cycles
References
1. F. Roters, P. Eisenlohr, L. Hantcherli, D. Tjahjanto, T. Bieler, D. Raabe, Overview of

constitutive laws, kinematics, homogenization and multiscale methods in crystal plasticity
finite-element modeling: theory, experiments, applications. Acta Mater. 58(1), 1152–1211
(2010)
2. S.D. Patil, Constraint effects on stationary crack tip fields in ductile single crystals. Ph.D.
thesis, IISC, 2009
3. J.M. Finney, C. Laird, Strain localization in cyclic deformation of copper single crystals. Phil.
Mag. 31(2), 339–366 (1975)
4. S. Flouriot, S. Forest, L. Remy, Strain localization phenomena under cyclic loading:
application to fatigue of single crystals. Comput. Mater. Sci. 26, 61–70 (11th International
workshop on computational mechanics of materials)
5. J.D. Clayton, Homogenization and incompatibility fields in finite strain elastoplasticity. Ph.D.
thesis, Georgia Institute of technology, 2002
6. D. Peirce, R.J. Asaro, A. Needleman, Material rate dependence and localized deformation in
crystalline solids. Acta Metall. 31, 1951–1976 (1983)
7. M.L. Williams, On the stress distribution at the base of a stationary crack. J. Appl. Mech.
24(1), 109–114 (1957)
8. J. Rice, Tensile crack tip fields in elastic-ideally plastic crystals. Mech. Mater. 6(4), 317–335
(1987)
9. J. Rice, Limitations to the small scale yielding approximation for crack tip plasticity. J. Mech.
Phys. Solids 22(4), 17–26 (1974)
10. ABAQUS, ABAQUS documentation (Dassault Systmes, Providence, RI, USA, 2011)
Analysis of the Plastic Zone Size Ahead
of the Crack Tip in Notched Components
and Its Application to Fatigue Problems
Masahiro Endo and Keiji Yanase
Abstract A method based upon the Dugdale model is proposed to calculate the
plastic zone size (pzs) at the crack tip of a Mode I crack emanating from notch of
arbitrary shape. To provide a generalized method, a crack at the notch root was
modeled as an edge crack in a semi-infinite plate subjected to the same stress as that
generated near the notch root in the absence of a crack. The calculated pzs ahead of
the tip of the crack emanating from circular and elliptical holes was in reasonable
agreement with the accurate values reported in the literature. By making use of the
calculated pzs, the near-threshold behavior of crack emanating from notch under
cyclic loading condition was simulated by using McEvily’s fatigue crack growth
equation. Subsequent comparison with the available experimental data revealed that
the proper consideration of plastic zone size can capture the phenomena and trends
observed in the experimental data with reasonable accuracy.
Keywords Dugdale model pzs Stress intensity factor Notch effect

High-cycle fatigue
1 Introduction
In principle, there is a finite, albeit small, region around a crack tip in which the
state of stress is adequately characterized by the stress intensity factor (SIF). In
essence, the linear elastic stress analysis of crack predicts infinite stresses at the
crack tip. In real materials, however, stresses at the crack tip are finite due to
inelastic material deformation, such as plasticity in metals, which leads to the
relaxation of crack-tip stresses. It is well established that, under small-scale yielding
M. Endo (&) K. Yanase

Department of Mechanical Engineering, Fukuoka University, Fukuoka City 814-0180, Japan
e-mail: endo@fukuoka-u.ac.jp
M. Endo K. Yanase
Institute of Materials Science and Technology, Fukuoka University, Fukuoka City 814-0180,
Japan

https://doi.org/10.1007/978-981-10-7197-3_50
600 M. Endo and K. Yanase
conditions, SIF can reasonably account for the extent of stress intensity in the
vicinity of cracks. Therefore, SIF is widely used to characterize the crack growth
behavior in the high-cycle fatigue regime. In practice, there are circumstances in
which the small-scale yielding condition is not satisfied. For instance, applying SIF
to small cracks emanating from the notch root may lead to a hazardous prediction
because the plastic zone size (pzs) is comparable to the crack length (i.e.,
small-scale yielding condition is not satisfied), due to the fact that the notch root is
highly stressed by the stress concentration.
In reality, the stresses, strains, and displacements in the vicinity of crack asso-
ciated with the plastic deformation are difficult to calculate. The finite element
method (FEM) is a convenient tool, but a special care to obtain reliable results must
be exercised for the mesh size, loading step, etc. On the other hand, applying simple
corrections to linear elastic fracture mechanics is a very convenient and attractive
alternative. Concerning Mode III, a modified crack length, cmod , which is the sum of
physical crack length c and half of the pzs, x (i.e., cmod ¼ c þ x=2), can account for
the effect of plasticity [1, 2]. Strictly speaking, this relation does not hold for Mode I
and II cracks. Nonetheless, this and similar approaches are widely used to address
crack problems in the presence of plasticity. For instance, in the Dugdale model [3],
the elastic–plastic behavior is modeled by superposing two elastic solutions: a
through crack under remote tension and a through crack with closure stresses at the
crack tip. In the literature, the Dugdale model was applied for a center crack in an
infinite plate [3], an edge crack in a semi-infinite plate [4], and a crack emanating
from a circular hole [5].
The procedures for analyzing constant amplitude fatigue under small-scale
yielding condition are well established, though a number of uncertainties remain.
For instance, large-scale yielding and short cracks introduce additional level of
complexities that are presently not fully understood. In this work, a generalized
method is proposed based on the Dugdale model, which is an effective approach for
studying the elastic–plastic behavior of crack emanating from notch. As an appli-
cation, the proposed method is combined with McEvily’s equation [6] to simulate
the fatigue crack growth behavior emanating from a circular hole in the high-cycle
fatigue regime. As will be shown, the proper consideration of plastic zone size can
capture the phenomena and trends observed in the experimental data with rea-
sonable accuracy.
2 Dugdale Model for Notch Problem
2.1 Dugdale Model
In the Dugdale model [3], a long, slender plastic zone at the crack tip in a
non-hardening material is considered. This strip-yield plastic zone is modeled by
assuming a crack length 2ðc þ xÞ, where x is pzs, with a closure stress equal to
Analysis of the Plastic Zone Size … 601
(a) Circular hole (b) Elliptical hole
Fig. 1 Crack emanating from a hole and plastic zone
(a) Notch stress field in (b) Notch stress field (c) Closure stress field
the absence of cracks imposed on the crack imposed on the crack
Fig. 2 Application of linear superposition with Dugdale model for Fig. 1a
yield stress of the chosen material, rY , at the crack tip. Since the stresses are finite in
the plastic zone, there cannot be a stress singularity at the crack tip. Accordingly,
the value of x must be calculated, such that SIFs due to remote tension and closure
stress cancel each other.
Let us consider an extension of the Dugdale model for notch problem. Figure 1
shows a crack emanating from a circular hole and an elliptical hole, respectively.
By using the principle of linear superposition, Fig. 1a can be effectively decom-
posed into the respective figures in Fig. 2. Given that only Fig. 2b, c contains the
crack, the following condition for SIF must be satisfied according to the Dugdale
model:
KA þ KB ¼ 0 ð1Þ
where KA is the SIF due to the notch stress field of ryy (Fig. 2b) and KB is the SIF
due to the closure stress of rY (Fig. 2c). By solving Eq. (1), the value of pzs or x
can be calculated. In practice, for an arbitrary notch shape, it is difficult to calculate
x since the closed-form solution for Eq. (1) is not available. Therefore, in what
follows, an approximate method is presented by making use of the solution for an
edge crack in a semi-infinite plate.
2.2 Approximation with an Edge Crack in a Semi-Infinite

Plate
For the circular hole (Fig. 1a), the notch stress field on the x-axis is given as [7]:
" #
a2 3a4
ryy ðxÞ ¼ ry1 1 þ þ ð2Þ
2ðx þ aÞ2 2ðx þ aÞ4
where ry1 is the remote tensile stress and the origin of x-axis is, for convenience,
moved to the edge of hole. It is noted that when l a, the problem of Fig. 2b is
equivalent to that of Fig. 3. To correctly account for the arbitrary notch stress field, the
following calculation with a pair of concentrated forces, P, is convenient [8] (Fig. 4):
pffiffiffiffi
2P pl
K ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi f1 þ f ðvÞg ð3Þ
p l 2 x2
Fig. 3 Approximation of the

edge crack with notch stress
field by a circular hole
Circular hole Notch stress field

by a circular hole
Fig. 4 Edge crack with a pair

of concentrated forces [8]
y
x
0
x
P
where
f ðvÞ ¼ ð1 v2 Þð0:2945 0:3912v2 þ 0:7685v4 0:9942v6 þ 0:5094v8 Þ;

ð4Þ
v ¼ x=l
Therefore, by substituting Eq. (2) into Eq. (3), KA for Fig. 2b is calculated as:
Zl pffiffiffiffi
2ryy ðxÞ pl
KA ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi f1 þ f ðvÞg dx ð5Þ
p l 2 x2
0
Similarly, KB for Fig. 2c is given as:
Zl pffiffiffiffi
2rY pl
KB ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi f1 þ f ðvÞg dx ð6Þ
p l 2 x2
c
By substituting Eqs. (5) and (6) into Eq. (1), x can be calculated via a numerical
method (e.g., bisection method). To correctly address the presence of an elliptical
hole (Fig. 1b), the following expression [9] can be used for Eq. (5):

1 k
ryy ðxÞ ¼ ry1 f þ
2
F1 ðfÞ F2 ðfÞ ð7Þ
f2 1 k1
where
( )
1 k 1 kþ3 2
1 kþ1
F1 ðfÞ ¼ 2 f ;
ðf 1Þ2 2 k þ 1 k 1 ðk 1Þ2
! ð8Þ
4f2 k k
F2 ðfÞ ¼ 2 f2
ðf 1Þ3 k2 1 ðk 1Þ2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a xþ x 2 n2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k¼ ; f¼ ; n¼ a2 b2 ð9Þ
b n
Figure 5 shows the results for a circular hole (Fig. 1a). Overall, the present
calculation shows a moderate agreement with the accurate results reported by Rich
and Roberts [5]. Further, Fig. 6 shows the results for the elliptical hole (Fig. 1b), in
which x is calculated in the absence of crack (i.e., l ¼ x in Fig. 1b). In other
words, the pzs due to the stress concentration by an elliptical hole is calculated in
Fig. 6. Compared to the accurate results reported by Nisitani [10], a good agree-
ment is observed.
The pzs of a center crack in an infinite plate under remote tension is analytically
given as [3]:
Fig. 5 Plastic zone size with

a circular hole [5]
Fig. 6 Plastic zone size with

an elliptical hole [10]

p ry1
xcenter ¼ c sec 1 ð10Þ
2 rY
On the other hand, the pzs of an edge crack in a semi-infinite plate is analytically
given as [4]:

p ry1
xedge ¼ 1:1215 c sec
2
1 ð11Þ
2 rY
In principle, at c=a ! 0, x ¼ xedge should hold (i.e., consideration of edge

crack is proper). However, at c=a ! 1, x ¼ xcenter should hold (i.e., consideration
of edge crack is no longer proper). The present calculation shows that x ¼ xedge
holds at both c=a ! 0 and c=a ! 1 due to employing the solution for the edge
crack (Figs. 3 and 4). This means that the following modification is necessary at
c=a ! 1 (cf. Eqs. 10 and 11):
x
xmod ¼ ð12Þ
1:12152
In this study, the following form is considered for convenience:
x 2 c
xmod ¼ ; where A ¼ ð1:12152 1Þ tan1 B þ 1 ð13Þ
A p a
As shown in Fig. 7, the parameter B governs the rate of convergence for A. For
instance, by minimizing the error in Fig. 5, B = 4 is obtained for the circular hole.
Fig. 7 Variations of
correction factor, A
As depicted in Fig. 8, a use of Eq. (13) provides results that are in good agreement
with the accurate results reported by Rich and Roberts [5].
3 Fatigue Crack Growth Simulation with the Extended

Dugdale Model
As an application of the extended Dugdale model, let us consider the fatigue crack
growth behavior emanating from a circular hole in the high-cycle fatigue regime.
According to McEvily et al. [6], the fatigue crack growth equation at a stress ratio
of R = −1 is rendered as:
dc
¼ A M2 ð14Þ
dN
where
hpffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffii
M ¼ 2pre F þ Y pcF ra 1 ekðcre Þ Kopmax DKeffth ð15Þ
Here, ra is the amplitude of remote cyclic stress, k is a material constant that

governs the rate of crack closure development, Kopmax is a value of SIF when a
macroscopic crack opens, and DKeffth is the effective range of the SIF at the
threshold level. Based on Eq. (15), so-called inherent crack length, re , is estimated
by considering the fatigue limit of a smooth specimen, rw0 , at R = −1 as follows:
Fig. 8 Modified plastic zone

size with a circular hole [5]
2
1 DKeffth 1 p rw0
re ¼ pffiffiffiffiffiffiffiffi ; where F0 ¼ sec þ 1 ð16Þ
1:12 pF0 Drw0 2 2 rY
In general, re is in the order of a few microns. Further, F is the elastic–plastic

correction factor for crack length based on the Dugdale model (cf. Eq. 13):
F ¼ 1 þ xmod =2c ð17Þ
Finally, Y is the geometric correction factor for SIF, which is illustrated for the
circular hole in Fig. 9 [11]. In essence, Eq. (15) considers the effects of a crack tip
of finite radius [12], the Kitagawa effect [13], and the crack closure. Based on
Eq. (14), the threshold condition of dc/dN = 0 can be found by setting M = 0.
Namely, by calculating the stress level of ra for a specific crack length of c to
satisfy M = 0, the unique threshold behavior for a given notch geometry can be
predicted.
The fatigue crack emanating from a circular hole was previously investigated by
El Haddad et al. [14]. Table 1 provides the material constants used for the pre-
diction, whereby comparisons between the present predictions and the experimental
data are depicted in Fig. 10. It is noted that one prediction (respective solid lines)
considers the effect of stress concentration for pzs (i.e., F in Eq. 15), while the other
prediction (respective dashed lines) neglects such an effect. In the dashed prediction
lines, the initial stresses (i.e., crack initiation stress) at c ¼ re denote the maximum
threshold stress. This means that the crack initiation directly leads to fatigue failure
without any evidence of crack arrest. On the other hand, concerning the solid
prediction lines for a ¼ 0:2 and 0:48 mm, the maximum stresses at the points A and
Fig. 9 Variations of
geometric correction factor,
Y [11]
Table 1 Material constants (Material: CSA G40.11 steel)

pffiffiffiffi pffiffiffiffi
Kopmax (MPa m) DKeffth (MPa m) rY (MPa) rw0 (MPa) k (m1 )
5.5 3.0 376 276 6000
B are greater than the respective crack initiation stresses (i.e., stresses at c ¼ re ).
This finding indicates that the crack initiation does not lead to fatigue failure and the
non-propagating crack exists. However, when the hole radius is large
(a ¼ 4:8 mm), the crack initiation leads to fatigue failure because the crack initi-
ation stress itself is the maximum threshold stress. Consequently, non-propagating
cracks are not observed for a ¼ 4:8 mm. Accordingly, in Fig. 10, the points A, B,
and C represent the predicted fatigue strength for the respective hole radii. Overall,
the findings reported here confirm that the proper consideration of plastic zone size
(i.e., the solid line) captures the phenomena and trends in the experimental data with
reasonable accuracy.
Fig. 10 Comparisons
between the predictions and Stress concentration is considered for pzs
Stress concentration is not considered for pzs
experimental data [14]
re = 4.3µm
4 Conclusion
In this study, the Dugdale model was extended and was applied in order to
investigate the notch problem in the high-cycle fatigue regime. In principle, the
proposed method takes advantage of the solution for an edge crack to calculate the
plastic zone size. As an application, the proposed method was combined with
McEvily’s equation to simulate the fatigue crack growth behavior emanating from a
circular hole in the high-cycle fatigue regime. The present prediction captures the
phenomena and trends observed in the experimental data with reasonable accuracy.
References
1. G.R. Irwin, Naval research laboratory report No. 5486, Naval Research Laboratory,
Washington, D.C. (1960)
2. C.K.H. Dharan, B.S. Kang, I. Finnie, Finnie’s Notes on Fracture Mechanics (Springer, New
York, 2016)
3. D.S. Dugdale, Yielding of steel containing slits. J. Mech. Phys. Solids 8, 100–104 (1960)
4. H. Nisitani, Y. Murakami, Interaction of elasto-plastic cracks subjected to a uniform tensile
stress in an infinite or a semi-infinite plate. Proc. Int. Conf. Mech. Behav. Mater. 1, 346–356
(1972)
5. T. Rich, R. Roberts, Plastic enclave sizes for internal cracks emanating from circular cavities
within elastic plates. Eng. Fract. Mech. 1, 167–173 (1968)
6. A.J. McEvily, D. Eifler, E. Macherauch, An analysis of the growth of short fatigue cracks.
Eng. Fract. Mech. 40, 571–584 (1991)
7. S.P. Timoshenko, J.N. Goodier, Theory of Elasticity (McGraw-Hill International, New York,
1982)
8. R.J. Hartranft, G.C. Sih, Alternating method applied to edge and surface crack problems.
Mech. Fract. 1, 179–238 (1973)
9. Y. Murakami, Theory of Elasticity and Stress Concentration (Wiley, West Sussex, 2017)
10. H. Nisitani, Elastic-plastic stress in a semi-infinite plate having an elliptical arc notch with an
edge crack under tension or longitudinal shear, Proc. Int. Cong. Fract. 2, 1–513 (1973)
11. Y. Murakami et al., Stress Intensity Handbook (Pergamon Press, Oxford, 1987)
12. G.R. Irwin, First Symposium on Naval Structure and Mechanics (Pergamon Press, Oxford,
1960), pp. 557–594
13. H. Kitagawa, S. Takahashi, Applicability of fracture mechanics to very small cracks or the
cracks in the early stage. Proc. Int. Conf. Mech. Behav. Mater. 627–631 (1976)
14. M.H. El Haddad, T.H. Topper, K.N. Smith, Prediction of non-propagating cracks. Eng. Fract.
Mech. 28, 573–584 (1979)
Effect of Loading Frequency
on the Fatigue Crack Growth Rate of Type
304L(N) Material
Mukhar Sharma, Punit Arora, P. K. Singh, D. K. Sahoo,

Gargi Choudhuri and V. Bhasin
Abstract The aim of this study was to study the effect of loading frequency on the
fatigue crack growth rate (FCGR) behaviour of Type 304L(N) austenitic stainless
steel material. Compact tension (C(T)) specimens machined from the solution
annealed pipe were used for carrying out the FCGR tests as per the standard
procedure of ASTM E647. Constant amplitude loading method with varying fre-
quency (0.01–10 Hz) was followed for cyclic loading during the tests. The tests
were carried out at load ratio 0.1 and ambient temperature. Experimental results
show that FCGR increases with decrease in frequency for a given stress intensity
factor range. Increase in FCGR is very significant from high to low frequency.
Fracture surface examinations of tested specimens were carried out using scanning
electron microscope to understand the increased crack growth at lower frequency.
The fracture surfaces revealed the presence of facets which have feature of brittle
fracture. These facets could be the locations for the formation of deformation-
induced-martensite (DIM). The facets are relatively harder (320 HV) as compared
to matrix of the material (155 HV) which may have led to higher crack growth rate.
Area covered by these facets is more for fracture surface tested at lower frequency.
FCGR of present test study is also compared with FCGR curve for austenitic
stainless steel given in ASME Section XI. ASME FCGR curve is comparable with
that of the FCGR curve of loading frequency 10 Hz. However, ASME FCGR
would lead to over-prediction of fatigue crack growth life for lower loading fre-
quency scenario.
Keywords Fatigue crack growth rate Deformation-induced martensite

Loading frequency
M. Sharma D. K. Sahoo
Board of Radiation and Isotope Technology, Navi Mumbai 400703, India
P. Arora (&) P. K. Singh G. Choudhuri V. Bhasin
Bhabha Atomic Research Centre, Mumbai 400085, India
e-mail: punit@barc.gov.in

https://doi.org/10.1007/978-981-10-7197-3_51
612 M. Sharma et al.
Nomenclature
C(T) Compact tension
TPB Three-point bend
FCG Fatigue crack growth
FCGR Fatigue crack growth rate
DIM Deformation-induced martensite
a Crack length
W C(T) width
K Stress intensity factor (SIF)
DK Stress intensity factor range
HCF High-cycle fatigue
SEM Scanning electron microscope
L Longitudinal
C Circumferential
Pmax Maximum load in one cycle
Plim Limit load
DP Load range
OD Outer diameter
R Load ratio
N Number of cycles
da/dN Crack growth rate
COD Crack opening displacement
ry Yield strength
ru Tensile strength
rf Flow strength (0.5 ( ry + ru))
SS Stainless steel
1 Introduction
Fatigue causes progressive and localized damage in a material due to repeated or

fluctuating strains, at the nominal stresses that have maximum values less than the
yield strength of the material. The cracks in the piping components of nuclear
power plants (NPP) may get generated in due course of their service life or might
have gone undetected during the pre-service inspection of components. The
accurate assessment of residual fatigue crack growth life of cracked components
serves two objectives that are life extension programme of power plants and
demonstration of leak before break criterion. In NPP, different service level loads
are categorized based on the frequency of their occurrence during its design life.
Effect of Loading Frequency on the Fatigue Crack … 613
The studies are required to generate fatigue crack growth rate (FCGR) parameters
on small-scale specimens such as compact tension (C(T)), three-point bend
(TPB) specimens under different loading frequencies.
Over recent years, austenitic stainless steels have been extensively used for
various engineering components specially in nuclear power plants because of their
high corrosion resistant and an excellent combination of mechanical properties
like high strength and high ductility. Now-a-days, Type 304L or Type 304L(N)
stainless steel is being used in place of Type 304 stainless steel as these low
carbon grades are less susceptible to formation of chromium carbide which may
lead to failure due to the intergranular stress corrosion cracking. Due to meta-
stable nature of austenitic stainless steel, it tends to form relatively harder
martensite upon large deformation.
The strain that occurs ahead of the crack tip during fatigue crack growth in a
metastable material induces a partial transformation to martensite, which alters both
the microstructure and the stress state at the crack tip and should, therefore, change
the fatigue crack growth rate. The martensite deformation causes a deformation in
shape that is substantiated by surface relief effect [1] and a volume change that is
dependent on composition and is 2% in 304-type stainless steels [2, 3].
Deformation-induced martensite formation is influenced by many factors like steel
chemistry, deformation mode, strain, stress, strain rate, grain size, grain orientation
and temperature [4]. Although, researchers have concluded that there is no effect of
frequency on the fatigue crack growth rate in air environment and room tempera-
ture. There have been research studies on the influence of the deformation-induced
martensite transformation under monotonic tensile loading and low-cycle fatigue in
304 series of materials. Das et al. [4] investigated the behaviour of this material
under cyclic loading, with increase in strain amplitude, the formation of
deformation-induced martensite increases. Roa et al. [5] studies showed that for-
mation of deformation-induced martensite is deleterious for low-cycle fatigue
regime but small amount of deformation-induced martensite (DIM) is beneficial for
high-cycle fatigue (HCF) regime. While in another study by Das et al. [6], under
monotonic loading at lower strain levels, the amount of martensite increases very
rapidly with increase in true strain in a linear proportional manner, whereas it is
observed that at low true strain value, higher the strain rate, more is the martensitic
transformation, while at high true strain value, reduction in strain rate favours the
formation of deformation-induced martensite. But the effect of the influence of
loading frequency (or strain rate) on fatigue crack growth rate behaviour is not
studied much and is an open area for investigations.
For better understanding and accurate assessment of FCG life of austenitic
stainless steels, influence of loading frequency/strain rate on its fatigue behaviour
must be established. In this context, the aim of present paper is to study the effect of
loading frequency/strain rate on fatigue crack growth rate (FCGR) behaviour of
Type 304L(N) stainless steel material. Experiments are conducted on Type 304L
(N)SS material using constant load amplitude method of standard ASTM E647
procedures [7]. The effect of cyclic loading frequency on FCG life is brought out.
Table 1 Chemical composition of present Type 304L(N) stainless steel (in wt%)
Cr Ni Mn C Si P S N Fe
18.26 8.45 0.79 0.021 0.33 0.004 0.003 0.1 Remainder
Table 2 Tensile properties of Type 304L(N) material

ry (MPa) ru (MPa) % El
240 560 64.2
% El: total percentage elongation
2 Experimental Procedures
2.1 Material Details
Austenitic stainless steel, Type 304L(N) stainless steel in the form of extruded pipe
of outer diameter (OD) 324 mm and thickness 25 mm in the solution annealed
condition as per the specification of ASME Section II [8] was used for machining of
standard compact tension (C(T)) specimens.
The chemical composition and tensile properties of the material are shown in
Tables 1 and 2, respectively.
2.2 Specimen Details
The C(T) specimens were designed as per the standard test procedure of ASTM
E647 [7]. The machined notch was put in circumferential direction with respect to
pipe axis and was named as LC orientation. The LC orientation means loading
direction is in longitudinal (L) direction, and crack growth is in circumferential
(C) direction with respect to axis of the pipe. The C(T) specimen with its dimen-
sions is shown in Fig. 1a, and orientation with respect to pipe axis is shown in
Fig. 1b.
2.3 Fatigue Pre-cracking
The pre-cracking for all the specimens was carried out as per the requirements of
ASTM E647 [7]. The pre-cracking and subsequent fatigue crack growth rate
(FCGR) tests were carried out at servo-hydraulic machine of maximum load
capacity as ±100 kN and travel of ±60 mm. The loading frequency was main-
tained as 10 Hz, and the load ratio (R) (the ratio of minimum load to maximum load
in a cycle) was kept as 0.1. The maximum stress intensity factor (K) during
Fig. 1 a Typical schematic of C(T) specimen and b layout of the C(T) specimen as machined
from full-scale pipe of 324 mm OD and 25 mm thickness of Type 304L(N) material
pre-cracking was less than that during FCGR test to rule out retardation effects. The
stress intensity factor range at the end of pre-cracking was varying from 19 to
22 MPa-√m for various tests. The ratio of crack length (a) to width (W) was nearly
0.3 after pre-cracking.
2.4 Fatigue Crack Growth Rate (FCGR) Tests
FCGR tests were conducted on pre-cracked specimens using constant load ampli-
tude method with a maximum and minimum load values of 16 and 1.6 kN,
respectively. The test loading frequency was varied from 0.01 to 10 Hz to bring out
the effect of loading frequency on the fatigue crack growth rate. The crack length
(a) based on the compliance-crack size correlation [7] was recorded with number of
fatigue cycles (N). The derived FCGR parameter (da/dN) from a-Vs-N curve was
then correlated with crack driving force, that is, stress intensity factor range (DK).
The stress intensity factor range (ΔK) was calculated from the crack length, cyclic
load and geometric factor [9]. Test parameters are shown in Table 3.
The crack length (a) with number of cycles (N) curves for different loading fre-
quencies are plotted in Fig. 2. All the curves are plotted from the data generated in
the tests, considering same initial reference crack length as 17 mm. This figure
shows that the crack growth life under slower loading frequency (0.01 Hz) is
significantly shorter than that under 10 Hz frequency.
Table 3 Tests and loading parameters for the fatigue crack growth rate tests
Test name Units SSB-f10 SSB-f1 SSB-f0.1 SSB-f0.01
Test temperature °C 25–27 °C 25–27 °C 25–27 °C 25–27 °C
Notch r (tip radius) mm 0.2 0.2 0.2 0.2
geometry a0 (initial notch mm 12.5 12.5 12.5 12.5
depth)
W (total width) mm 50 50 50 50
b (remaining mm 37.5 37.5 37.5 37.5
ligament)
B (thickness) mm 18 18 18 18
Material ry (yield strength) MPa 240 240 240 240
Properties ru (ultimate MPa 560 560 560 560
strength)
rf (flow stress) MPa 400 400 400 400
η* – 0.277 0.277 0.277 0.277
Plim (limit load) kN 108.8 108.8 108.8 108.8
FCGR test Pmax kN 16.00 16.00 16.00 16.00
parameters R (stress ratio) – 0.1 0.1 0.1 0.1
Pmin kN 1.60 1.60 1.60 1.60
Frequency Hz 10 1 0.1 0.01
Number of cycles DN 113,347 26,780 6594 4680
ai (initial crack mm 14.5 17 16 14.6
length)
af (final crack mm 25.00 26.25 25.50 20.25
length)
DKi (initial DK) MPa- 19.6 22.3 21.2 19.7
√m
DKf (final DK) MPa- 34.6 37.4 35.6 26.4
√m
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2a2 4a
*g ¼ b þ b þ 2 2a b þ1
Figure 3 shows the variation of fatigue crack growth life for same initial crack
length and final crack length with loading frequency. This comparison is made for
the initial crack size of 17 mm to final crack size of 21 mm, subjected to load range
(DP) of 14.4 kN and Pmax as 16 kN (i.e. R = 0.1).This figure shows that the effect
of frequency on FCG life becomes predominant as we decrease the loading fre-
quency from 10 Hz (high-frequency regime) to 0.01 Hz (low-frequency regime).
There is approximately 70% reduction in FCG life on lowering the frequency from
10 Hz to 1 Hz.
Fig. 2 Crack size (a) versus the number of load cycles (N) curves for different loading
frequencies
∆P=14.4kN, Pmax=16kN, 17≤a(mm)≤21
Fig. 3 Fatigue life for the same crack growth (starting from 17 to 21 mm) with the loading
frequency
The fatigue crack growth rate (da/dN) is plotted with crack driving force (DK) in
Fig. 4, for different frequencies. It is apparent from this figure that FCGR under
slow frequency is significantly higher as compared to that under high frequency.
This test investigation is in line with the tensile test studies carried out by Das et al.
[6]. Das et al. have brought out in their studies that for higher true strain value, the
decrease in the strain rate favours the formation of deformation-induced martensite.
Although the above study is for the monotonic tensile loading and there is no direct
correlation with the cyclic loading, but in an approximate way, similar observation
was found in the present test results. This implies that the formation of martensite
promotes the crack growth rate by providing a brittle facet to the crack. Model
proposed by James and Jones [10] also shows that in ASS at room temperature,
there is a slight increase in FCGR on decreasing the loading frequency. In subse-
quent studies, Shih and Chen [11] also proposed a modified power law relation for
ASS and concluded increase in FCGR with decrease in loading frequency. This is
also supported by the study of Khan [12] that mentioned the strain hardening effect
generated due to the formation of deformation-induced martensite accelerates the
crack growth rate. However, Khan studies were carried out on Type-301 stainless
steel material. The present test results are consistent with the literature data.
Subsequent to FCGR test studies, the cause of higher crack growth rate was
brought out through microstructural investigations using SEM. Specimens were cut
open, and the fatigue fracture surface was examined under scanning electron
microscope. The microscopic examinations of fatigued surfaces revealed the
Fig. 4 Fatigue crack growth rate (da/dN) versus stress intensity factor range (ΔK) under different
frequencies
Fig. 5 Typical SEM image of microstructure of fracture surface of specimen tested under 10 Hz
frequency
presence of discrete planar facets. These facets typically indicate the occurrence of
brittle mode of fracture at the localised regions. Figures 5 and 6 show the typical
SEM images of the fractured surfaces tested under loading frequency of 10 and
1 Hz, respectively. Figure 6 denotes typical brittle facets, and they are found to be
occupying more area than that under 10 Hz frequency (Fig. 5) for same DK viz.
same location from notch tip for given load range of 14.4 kN (Table 3). The
hardness of region near to fracture plane was measured to be 320 HV in SSB-f0p1
specimen (0.1 Hz test) on transverse free surface plane. However, untested speci-
men had hardness of nearly 150 HV. The facet-like features on fracture surface
(Figs. 5 and 6) and hatched-like impressions on transverse etched planes (Figs. 7
and 8) near to fracture plane have been characterized as DIM [4]. These observa-
tions indicate the formation of DIM near to crack plane leading to higher crack
growth rate for low rate of loading tests. The features belonging to
deformation-induced martensite are clearly visible as hatched region on the trans-
verse free surface plane (perpendicular to crack surface) using optical microscopy
in Fig. 7 for 10 Hz loading frequency test. These features are depicted in Fig. 8 for
0.1 Hz loading frequency test. Here also, the extent of DIM (hatched region) is
higher for lower loading frequency. The results of optical microscopy are consistent
with facets observed under SEM.
The ASME Section XI has the fatigue crack growth rate curve for austenitic
stainless steel material in air environment at room temperature [13]. The fatigue
crack growth rate as per ASME Section XI lies below the tests conducted under
lower frequency range (0.01–1 Hz). This comparison is shown in Fig. 4. For any
engineering component having part through thickness crack subjected to
Fig. 6 Typical SEM image of microstructure of fracture surface of specimen tested under 1 Hz
frequency showing facets
Fig. 7 Typical optical image of etched transverse free surface (perpendicular to crack surface) of
specimen tested under 10 Hz frequency at 12 mm location from notch tip
Fig. 8 Typical optical image of etched transverse free surface (perpendicular to crack surface) of
specimen tested under 0.1 Hz frequency at 12 mm location from notch tip
low-frequency fatigue loading, if fatigue life is predicted using ASME prescribed

FCGR curve, this would lead to over-prediction of fatigue crack growth life of
cracked component.
4 Conclusions
Study on the effect of loading frequency on fatigue crack growth rate (FCGR)
behaviour of Type 304L(N) austenitic stainless steel material can be summarized
as:
1. FCGR increases with decrease in frequency for a given stress intensity factor
range. Increase in FCGR is very significant from high to low frequency.
2. Fracture surface examinations revealed the presence of facets which have feature
of brittle fracture. These facets could be the locations for the formation of
deformation-induced martensite (DIM). Area covered by these facets is more in
the specimens tested at lower frequency.
3. ASME FCGR curve is comparable with that of the FCGR curve of loading
frequency 10 Hz. ASME FCGR curve lies below the FCGR curves generated
for frequencies 0.01–1 Hz in present study. Use of ASME FCGR curve for
application where loading frequency is lower would lead to over-prediction of
FCG life.
References
1. J.F. Breedis, W.D. Robertson, Acta Metall. 10, 1077–1088 (1962)

2. H. Fiedler, B. Averbach, M. Cohen, Trans. ASM 47, 276–290 (1955)
3. R. Read, Acta Metall. 10, 865–877 (1962)
4. A. Das, S. Sivaprasad, P.C. Chakraborti, S. Tarafder, Morphologies and characteristics of
deformation induced martensite during low cycle fatigue behaviour of austenitic stainless
steel. Mater. Sci. Eng. A 528(2011), 7909–7914
5. J.J. Roa, G. Fargas, E. Jimenez-Pique, A. Mateo, Deformation of mechanism induced under
high cycle fatigue tests in austenitic stainless steel. Mater. Sci. Eng. A 597, 232–236 (2014)
6. A. Das, S. Das, Experimental investigation on martensite transformation and fracture
morphologies of austenitic stainless steel. Int. J. Plast. 25, 2222–2247 (2009)
7. Standard Test Methods for Measurement of Fatigue Crack Growth Rate, ASTM E 647-05
8. ASME, Boiler Pressure Vessel Code, Section II: Materials. Edition 2001
9. T.L. Anderson, Fracture Mechanics, Fundamental and Applications (Taylor and Francis
Publishers, Oxford, 2005)
10. L.A. James, D.P. Jones, Fatigue crack growth correlations for austenitic stainless steels in air,
in Conference on Predictive Capabilities in Environmentally Assisted Cracking, ASME PVP,
vol. 99, 1985, pp. 353–414 (Presented at the Winter Annual Meeting of the American Society
of Mechanical Engineers)
11. Y.S. Shih, J.J. Chen, The frequency effect on the fatigue crack growth rate of 304 stainless
steel. Nucl. Eng. Des. 191(1999), 255–230
12. Z. Khan, Stress-induced martensitic transformation in metastable austenitic stainless steels:
effect on fatigue crack growth rate. ASM Int. JMEPEG 5, 201–208 (1996)
13. ASME Boiler Pressure Vessel Code, Section XI, In-service Inspection of Nuclear Power Plant
Components, Edition 2010
Fatigue Life Characterization
of Additively Manufactured Acrylic like
Poly-Jet Printed Parts
J. A. Suresh, Gurunathan Saravana Kumar, Palaniappan Ramu

and Jayaganthan Rengaswamy
Abstract Application of additively manufactured components is gaining popu-

larity in functional application. Several polymer-based additive manufacturing
processes such as fused deposition modelling (FDM), selective laser sintering
(SLS), stereolithography apparatus (SLA) and poly-jet printing to name a few cater
to automotive to medical applications. Reliability in service of these components
has to be ascertained for using them for functional application. The components
made from additive manufacturing have inherent anisotropy due to the process
characteristics such as build orientation and laser scanning/deposition tool paths.
Literature reports fatigue life characterization mostly in FDM parts. In this work, a
fatigue life analysis was carried out on components manufactured by poly-jet
printed parts. Also, two build orientations were used to study the effect of this
process parameter on the tensile strength and fatigue life. Other process parameters
such as layer thickness, finishing type and polymer material were kept constant.
A standard test specimen as per ASTM standard D7791–12 was prepared using
Stratasys® Eden 350 V™ poly-jet printer using Veroclear 720™ material in two
build orientations. The specimens were tested for tensile test and tensile–tensile
fatigue cycles to failure at different stress levels. The test results were analysed to
understand the fatigue life characteristics, and it was found that the process
parameter variation has significant effect on the part strength (tensile) and reliability
(fatigue failure). The fractured surface was analysed using SEM images to cor-
roborate the inferences from test results.
J. A. Suresh G. S. Kumar (&) P. Ramu J. Rengaswamy

Department of Engineering Design, Indian Institute of Technology Madras,
Chennai 600036, India
e-mail: gsaravana@iitm.ac.in
J. A. Suresh
e-mail: rajsresh@iitm.ac.in
P. Ramu
e-mail: palramu@iitm.ac.in
J. Rengaswamy
e-mail: metarj@iitm.ac.in

https://doi.org/10.1007/978-981-10-7197-3_52
624 J. A. Suresh et al.
Keyword Additive manufacturing Mechanical test Fatigue

Part build orientation
1 Introduction
Additive manufacturing (AM) is a unique alternative to traditional manufacturing

techniques that embraces the concept of reducing product development cycles and
shortening the time-to-market of new products [1]. AM has become a popular
method to produce not only prototypes and moulds, but increasingly AM is being
used to fabricate components for engineered products for a range of industrial
applications. AM production of multimaterial components and devices is of con-
siderable interest, as it allows the fabrication of elastomeric hinges, integrated grips
and other products where material changes can achieve enhanced functionality [2],
reduced part counts and streamlined assembly. The ability to selectively use several
materials with different mechanical properties offers design freedom for producing
final parts with tunable failure locus and resistance to fracture. Reliability in service
of additively manufactured components has to be ascertained for using them for
functional application. The components made from additive manufacturing have
inherent anisotropy due to the process characteristics such as build orientation and
laser scanning/deposition tool paths which lower the mechanical strength [3]. Also,
since the part is fabricated layer by layer, staircase effect [4] reduces the surface
finish [5] and also causes poor form tolerance [6].
The poly-jet direct 3D printing (PJD-3DP) system is an AM process that builds
physical models from computer models with good surface finish by adding layers of
photopolymer resin. This process uses simultaneous jetting of modelling materials
to create physical prototypes [7]. The Stratasys® poly-jet process is a direct 3D
printing process (direct 3DP) in which layers of photopolymer are printed onto the
build platform via inkjet print heads (as described in Fig. 1). During printing, UV
lamps (located on either side of the print head) cure the recently deposited pho-
topolymer. The direct 3DP process differs from the more common indirect 3DP
processes in that the build material is being jetted, rather than jetting a binder into a
bed of build material.
Because there is no powder bed in direct 3DP processes, multiple materials can
be printed in the same build—similar to the manner in which multiple colours can
be printed with traditional inkjet printers. Stratasys® has developed a number of
different photopolymers with differing degrees of stiffness, strength and trans-
parency for various applications [8]. Several literatures are available in character-
ization of parts fabricated with poly-jet process such as surface finish [5] and part
strength [4], to name a few. Literature reports fatigue life characterization mostly in
FDM parts [9]. Since the literature on fatigue life analysis of poly-jet parts is limited
[10], the present work was carried out to understand the same. Tensile and fatigue
tests were carried out on ASTM D7791–12 specimens fabricated using Veroclear
720™ model material. Veroclear 720™ is a rigid photopolymer, nearly colourless
Fatigue Life Characterization of Additively Manufactured … 625
Fig. 1 Additive manufacturing using poly-jet process. Courtesy Stratasys®
material featuring proven dimensional stability for general purpose, fine-detail

model building and visual simulation of transparent thermoplastics like PMMA.
The authors also evaluated the effect of two different orientations on tensile and
fatigue life for comparison. Rest of the paper discusses the details of methodology,
results and inferences from the study.
2 Materials and Methods
Stratasys® Eden 350 V™ rapid prototyping machine was used to build parts to
study tensile and fatigue characteristics and the effect of two part build orientation
on the same. The models were built with the photopolymer Fullcure Veroclear
720™ model material and Fullcure 705™ support material supplied by Stratasys®
company. Process parameters such as layer thickness, type of finish and build
orientation are important factors affecting the surface finish and mechanical char-
acteristics of poly-jet AM components [11]. Out of these factors, only build ori-
entation is considered for the present study. Thus, for the present experimental
study, these parameters were chosen as described in Table 1. The machine can
Table 1 Process parameters S. No. Factors Level-1 Level-2

chosen for experiment and
their levels 1 Layer thickness, t (µm) 16 –
2 Type of finish Matte –
3 Build orientation, h (º) 0, 90, 0 90, 0, 0
Fig. 2 a CAD model of tensile and fatigue coupon, b samples fabricated corresponding to two
orientations (L1 and L2)
build parts in two layer thicknesses (16 and 32 µm) and two different finish types
(matte and glossy). The parts (in this case tensile coupons) were built in two
orientations as described in Table 1. The angle h in (º) is a vector that describes the
built orientation; angle made by the longitudinal axis (pull direction) of the samples
to the build table axes (X, Y and Z). In all two combinations of layer thickness, types
of finish and build orientation were obtained. Henceforth, specimen built with
level-1-built orientation (specimen axis parallel to X-axis) will be designated as L1
and the other as L2. The size of the specimens (shown in Fig. 2a) was designed as
per ASTM D7791–12. The fabricated parts are shown in Fig. 2b. For each com-
bination (L1 and L2), two parts (A and B) were built to get statistical concurrence.
The uniaxial tensile test was conducted using displacement controlled testing
with a controlled ramp rate of 2 mm/min in a servo-hydraulic UTM. The fatigue
tests were conducted with force-controlled tension–tension mode. The load form
chosen was sine wave with a frequency of 20 Hz. The fatigue tests were conducted
till the failure, and the fractured surfaces were examined by SEM analysis to reveal
the fracture details.
3.1 Tensile Strength
The tensile tests were performed at ambient conditions to obtain the tensile prop-
erties. The specimens used for this study were prepared to ASTM D638 standard as
shown in Fig. 2. The specimens were first cleaned thoroughly to be free from the
support material and any other foreign material. Two samples were tested in each
case.
The results of tensile tests are provided in Table 2, and the stress strain graph is
shown in Fig. 3. It can be observed that poly-jet printed parts exhibit very close
UTS and yield point stress. It is also observed that the UTS of the L2 samples (built
in perpendicular orientation L2) has slight improvement when compared with L1
sample (built in parallel orientation). The elongation of L1 specimens is longer than
the L2 samples. This could be attributed to continuous polymerization of L1 along
the axis of the specimen as compared to discontinuous polymerization (more than
one pass) for the L2 samples.
3.2 Fatigue Strength
Fatigue strength of the material for both orientations was tested under tension–
tension with force-controlled loading cycles axially. S-N plots were obtained from
these tests. The fatigue experiments were carried out using a servo-hydraulic uni-
versal testing load frame. The test system is computer controlled with provision for
data acquisition. The experiments were conducted in the ambient conditions. The
tensile test results were used to select the stress levels for the fatigue tests. The
R-ratios (ratio of the minimum stress over the maximum stress in a loading cycle)
were selected in the range from 0.05 to 0.1 for the experiments. The specimen was
tested until 106 loading cycles to obtain the fatigue limit at chosen stress levels.
Figure 4 shows the S-N curve for the L1 and L2 oriented samples. Four alter-
nating stress levels were used in steps. Three samples were tested for each stress
levels for both types of orientations. The fatigue limit is obtained at 8 MPa for the
parallel-oriented specimens as against 9 MPa for perpendicular-oriented samples.
Hence, there is a small improvement in fatigue life in perpendicular-oriented
samples.
628
Table 2 Tensile properties of poly-jet printed parts

Specimen Repeat Yield Average yield Tensile Average tensile % Average %
code strength (MPa) strength (MPa) strength (MPa) strength (MPa) Elongation elongation
L1 (parallel A 31.85 31.88 32.72 32.57 127 126.5
orientation) B 31.92 32.43 126
L2 (perpendicular A 34.15 34.13 34.35 34.31 113 113.5
orientation) B 34.12 34.27 114
J. A. Suresh et al.
Fig. 3 Stress–strain (engineering) characteristics of poly-jet printed parts
Fig. 4 S-N plot of poly-jet printed parts
3.3 Microstructural Observations
The samples for optical microscopy were sectioned from the printed samples for
both parallel and perpendicular orientation of the samples. The sectioning of the
samples was made methodically to preclude any damage or scratches that could
occur on surface of the printed samples. The microstructure of the samples was
characterized by an optical microscope (Carl Zeiss Axioskop series). The
Fig. 5 Microstructure of printed samples at 50 a parallel and b perpendicular oriented
calibration of the microscope was done as per the ASTM E1951−02 requirements.
The samples were viewed at various magnifications to understand the influence of
surface characteristics on fatigue strength of the printed material. The images
obtained under different magnifications are shown in Figs. 5 and 6. The optical
micrographs in Fig. 5a, b reveal the horizontal and vertical shades representing the
L1 (parallel oriented) and L2 (perpendicular oriented) samples, respectively. The
shades indicate the parallel and perpendicular micro-irregularities on surfaces of the
samples. However, the optical micrographs in Fig. 6a, b show the presence of
surface irregularities in both orientation of the samples, and especially microcracks
were observed at the parallel-oriented samples. The microcracks were observed on
entire surface on both sides of the sample.
The cross-sectional SEM images taken at parallel- and perpendicular-oriented
samples are shown in Fig. 7a, b. It reveals that the irregularities such as tiny voids
and microcracks are relatively less accounting for higher tensile and fatigue strength
of the printed samples in the perpendicular-oriented samples as compared to
parallel-oriented samples.
Fig. 6 Microstructure of printed samples at 400 a parallel and b perpendicular oriented

Fig. 7 SEM images of fatigue fracture surface a parallel oriented, b perpendicular oriented and
c higher magnification of fracture surface
Since the printing is carried out layer-by-layer deposition of Veroclear in the

present work, the interface with various geometrical features and irregularities is
inevitable as observed in the present work. The orientation of cracks and applied
load in the parallel-oriented samples could facilitate the crack propagation rate
slightly faster than the perpendicular-oriented samples. The crack branching and
fibre pull out and its misorientation could slow down the crack movements in
perpendicular-orientated samples as evident from the SEM cross-sectional samples.
4 Conclusion
Poly-jet printer was used to fabricate three-dimensional parts using Veroclear

polymer in two different build orientations in the present work. The tensile and
fatigue tests of the printed samples were performed to evaluate its strength and
fatigue life in both parallel and perpendicular orientations. The mechanisms of
improvement in strength and fatigue life of samples were substantiated using optical
and SEM characterization of surface morphology and fracture surface. The
perpendicular-oriented samples exhibit higher tensile and fatigue strength as
compared to parallel-oriented samples as evident from the surface morphological
and cross-sectional characterization of samples made by SEM. The multipass
deposition and polymerization of printed resins in the poly-jet printing accounts for
the anisotropy in strength and fatigue life seen in the present work. The interface
with varying surface roughness and voids’ formation in the printed parts ought to be
tailored for improving the fatigue life. The study is a pilot study, and the results
need to be augmented with experiments involving larger sample size for statistical
validity.
References
1. Objet Inc., 3D printing & rapid prototyping by Objet Ltd., in 3D Printing & Rapid
Prototyping by Objet Ltd. (2012). Available: http://www.objet.com/. Accessed 12 July 2012
(Online)
2. I. Gibson, D.W. Rosen, B. Stucker, Additive Manufacturing Technologies: Rapid Prototyping
to Direct Digital Manufacturing (Springer, 2009)
3. B. Caulfield, P. McHugh, S. Lohfeld, Dependence of mechanical properties of polyamide
components on build parameters in the SLS process. J. Mater. Process. Technol. 182, 477–
488 (2007). ISSN 0924-0136
4. O.S. Es-Said, J. Foyos, R. Noorani, M. Mendelson, R. Marloth, Effect of layer orientation on
mechanical properties of rapid prototyped samples. Mater. Manuf. Process. 15(1), 107–122
(2000)
5. D. Ahn, J.H. Kweon, S. Kwon, J. Song, S. Lee, Representation of surface roughness in fused
deposition modeling. J. Mater. Process. Technol. 209, 5593–5600 (2009)
6. B. Lee, J. Abdullah, Z. Khan, Optimization of rapid prototyping parameters for production of
flexible ABS object. J. Mater. Process. Technol. 169, 54–61 (2005). ISSN 0924-0136
7. S. Ashley, Rapid prototyping systems. Mech. Eng. 113(4), 34 (1991)
8. T. Wohlers, T. Gornet, History of Additive Manufacturing. Wohlers Report (2014). http://
wohlersassociates.com/history2014.pdf
9. J. Lee, A. Huang, Fatigue analysis of FDM materials. Rapid Prototyp. J. 19(4), 291–299
10. E. Brandl, U. Heckenberger, V. Holzinger, D. Buchbinder, Additive manufactured
AlSi10 Mg samples using selective laser melting (SLM): microstructure, high cycle fatigue,
and fracture behavior. Mater. Des. 34, 159–169 (2012)
11. K. Puebla, K. Arcaute, R. Quintana, R.B. Wicker, Effects of environmental conditions, aging,
and build orientations on the mechanical properties of ASTM type I specimens manufactured
via stereolithography. Rapid Prototyp. J. 18(5), 374–388 (2012)
Part IX
Material Processing
Influence of Machining Process on Surface
Integrity and Fatigue Life of a Turbine
Rotor Blade
Benudhar Sahoo, S. K. Panigrahi and R. K. Satpathy
Abstract Turbine blades of a military gas turbine are one of the most
safety-critical components as per MIL-STD-1783 as its failure cannot be contained
by engine casing in case of failure. Statistic tells, the major failure of the turbine
blade is due to fatigue. Further, both the surface integrity and structural integrity of
a turbine blade only can be assessed by full-scale testing. This study deals with a
turbine blade made of a wrought nickel base alloy AP220BD of a military gas
turbine engine. The primary shape of the blade is achieved by closed die forging
followed by machining of the aerofoil profile. Two machining processes, namely
(i) Non-conventional machining ECM process and (ii) Conventional CNC milling,
have been adopted to get the finished shape of the blade. Aluminising coating by
pack cementation method was deposited to provide oxidation resistance. Surface
integrity of the blade has been evaluated by visual inspection of the surface,
roughness, surface residual stress and microhardness survey from the surface to
subsurface. X-ray diffraction method has been adopted to measure the residual
stress. Full-scale fatigue testing of the turbine blades machined by both the pro-
cesses has been conducted following incremental amplitude Armstrong method.
Comparison of fatigue life has been carried out statistically using Student’s t test.
The fractographic study has been made both under low magnification and high
magnification. SEM could reveal the initiation point and crack propagation mode.
Turbine blades machined by ECM gave a higher tensile surface residual stress and
lower fatigue life. Blades machined by ECM reveal multiple crack initiation while
blades machined by CNC milling shows few cracks. There is a distinct separation
of progressive and overload failure region during the fractographic study. However,
the ECM process needs to be optimised, and subsequent surface treatments also
need to be improved to have enhanced fatigue life.
B. Sahoo (&) R. K. Satpathy

DRDO, RCMA, Sunabeda, Koraput 763002, India
e-mail: bsahoo543@gmail.com
S. K. Panigrahi
Defence Institute of Advanced Technology (DU), Pune, India

https://doi.org/10.1007/978-981-10-7197-3_53
636 B. Sahoo et al.

Keywords Turbine rotor blade High cycle fatigue Electrochemical machining

Residual stress Armstrong method HCF ECM CNC
Nomenclature
HCF High cycle fatigue
ECM Electrochemical machining
CNC Computer numerical control
1 Introduction
Turbine rotor blades of a military gas turbine are designed against high cycle
fatigue (HCF) with the major source of HCF is vibration [1]. As per MIL-5007E,
the rotor blades are subjected to vibration fatigue test at its resonance condition for
2 106 cycles for qualifying blade life [2]. Turbine rotor blades are made of
nickel-base alloys for their outstanding high-temperature strength and excellent
manufacturability. Because of the lengthier configuration of low-pressure turbine
rotor (LPTR) blade, undue loading due to vibration is an important consideration
for its life. Non-conventional machining process such as electrochemical machining
(ECM) was adopted as a metal removal process for its faster removal rate. With the
significant advancement in computer numerically controlled (CNC) machining
process, it is replacing the non-conventional machining process for its improved
accuracy and repeatability. Also, it is becoming an alternative as a green manu-
facturing process against the malign effect of ECM process on the environment.
The present investigation deals with the influence of ECM process and CNC
milling process on surface integrity and fatigue life of a turbine rotor blade.
2 Experimental Detail
Wrought nickel base alloy AP220BD of Russian origin having a typical chemical
composition in weight percentage: Ni-15Co-10Cr-6Mo-6 W-2.5Ti-4.3Al-0.6
V- 0.03C in the cold-rolled condition has been used for manufacturing the LPTR
blade by envelope forging. Two machining processes, namely ECM and CNC mil-
ling, have been adopted to generate the aerofoil profile from the blade forging. In ECM
process, both the suction surface and pressure surfaces are machined at the same
setting followed by grinding to remove the surface unevenness. Sodium chloride is
being used as the dielectric fluid during ECM process with the stainless steel elec-
trodes. Aerofoil surface of turbine rotor blade has been generated by 5-axis CNC
milling process using a machine of make HX-253S Switzerland with the help of ridgid
cam software (RCS). The finished shape was achieved in three passes using carbide
Influence of Machining Process on Surface Integrity … 637
tool followed by buffing to achieve a mirror finish. The blades are provided with an
aluminising coating of 30–40 lm thickness to resist oxidation and high-temperature
corrosion. The coating was deposited by diffusion technique using low-activity pack
cementation process. The turbine blades are put in a metal retort in a pack consists of
inert oxide Al2O3, halide salt NH4Cl and FeAl powder [3].
Vibration fatigue testing of blades using incremental amplitude method sug-
gested by Dr. Armstrong is based on Miner’s hypothesis. The stress is simulated by
the af level where a—amplitude of vibration and f—frequency of vibration. The
initial af is 1.0 ft/s while the each increment af is 0.5 ft/s and the dwell period at
each “af” level is 30 min. The test is continued till failure of the blade.
2.1 Test and Checks
As fatigue is predominantly a surface phenomenon, surface roughness and micro-

hardness checks have been carried out to characterise the surface [4–6]. Full-scale
fatigue testing of a turbine blade by vibration fatigue method following Armstrong
staircase has been adopted to evaluate its life.
3 Results and Analysis
The topography of the surface of the blade influences HCF life most. The surface
texture is also affected by the machining parameters. The undulation in the surface
acts as a stress concentrator and initiates fatigue failure. The hardness of the blades
indicates the strength and the surface residual stress. The two critical surface fea-
tures influencing fatigue life, i.e. surface roughness and microhardness were mea-
sured on the blades manufactured by both the routes.
3.1 Surface Roughness
See Table 1.
3.2 Microhardness
The full aerofoil profile was covered by measuring hardness at seven locations as
depicted in Fig. 1 using microhardness tester make: TUKON TM1162 with a load
of 100 g. The hardness survey was carried out from the top of the coating into the
blades at a depth of 20 µm till attaining 100 µm as shown in Fig. 2, and the
638 B. Sahoo et al.
Table 1 Surface roughness of blades produced by ECM and CNC milling process
(a) ECM process: surface roughness (Rz) in µm
Root Middle Top
Suction surface 2.055–2.51 1.65–1.829 3.737
Pressure surface 1.432–1.80 2.418–2.665 3.63
(b) CNC milling: surface roughness (Rz) in µm
Root Middle Top
Suction surface 0.915 1.13–1.865 1.34
Pressure surface 1.201 1.073 2.46–2.82
Fig. 1 Sketches showing blade aerofoil for hardness measurement a sections of blade and
b locations for hardness measurement
hardness impressions are shown in Fig. 3. Pressure surface exhibited higher

hardness compared to the suction surface. The maximum hardness on suction
surface and pressure surface is found to be 758 and 830 HV, respectively.
Maximum hardness was observed on the coating layer. The maximum hardness and
hardness gradient over the blades manufactured by ECM are found to be more
compared blades machined by CNC milling process as reflected in Fig. 2.
(a) (b)
Fig. 2 Variation of microhardness in the aerofoil of turbine rotor blades a electrochemical

machining b CNC milling
Aluminising coating being brittle provides the highest hardness in comparison

with the substrate, i.e. nickel base wrought alloy AP220BD. Aluminising coating
gets deposited by a diffusion process, and the surface energy of the substrate does
influence the rate of diffusion. As ECM and CNC milling yield different levels of
surface energy, the same will be reflected in the coating and to a lesser degree in
hardness. But HCF life gets significantly affected by the surface and subsurface
condition, therefore, different machining operations such as ECM and CNC milling
results in different fatigue performance.
3.3 Residual Stress
Surface residual stress was determined by the X-ray diffraction residual stress
analyser, Proto automated XRD system and service manufacturing, using Mn-Ka
tube with detector angle of 156.32° and collimator of higher diameter in the range
of 3–5 mm. The turbine rotor blade was divided radially into three sections, namely
root, middle and tip. Also, it was divided laterally into three sections, namely
leading edge, trailing edge and middle portion as shown in Fig. 4. A full profile of
the blade has been mapped for residual stress measuring nine points each on
pressure surface and suction surface.
Residual stress on the pressure surface ranges from −93 to 277 MPa for blades
produced by ECM route and from −685.5 to 176.4 MPa for blades produced by
CNC milling. Residual stress on the suction surface varies from −83 to −225 MPa
and −131.8 to −368.9 MPa for blades produced by ECM route and CNC milling
routes, respectively. The maximum stress was observed near the fir-tree root
towards the leading edge. Ozel and Ulutan reported that residual stress increases
with an increasing cutting tool radius and coated tools show less residual stress at
the surface than at subsurface [7]. Residual stress varies during lifetime and
depending on thermal energy or mechanical energy imparted to the surface. Under
the application of mechanical and thermal energy, elastic deformations associated
Fig. 3 Microhardness impressions on turbine rotor blades showing coating and blade a electro-
chemical machining b CNC milling
640 B. Sahoo et al.
Trailing edge
Fir-tree root Suction
surface
A B C
Pressure
surface
Leading edge
Fig. 4 Schematic diagram showing section of a turbine blade for residual stress measurement
with residual stress relax in case of plastic deformation. Under the application of
elastic load, tensile strength remains stable but this increases when the load is
increased close to yield strength of the material [8].
3.4 Fatigue Life Evaluation of Turbine Rotor Blades
Various machining processes modify the surface in terms of residual stress pattern,
plastic deformation, micro-cracking, etc. which in turn influence the fatigue life.
The turbine rotor blade is subjected to vibratory fatigue testing in bending mode
under constant stress. It has been reported by Pervey and Koster that relative
behaviour of materials under combined tension–tension loading is approximate to
the cantilever bending [9]. Full-scale fatigue testing under resonance condition is a
well-proven method for evaluating the HCF life of turbine blades [10]. The max-
imum stress at the root of a turbine rotor blade during fatigue testing under can-
tilever loading is expressed in Eq. (1) [11] by
pffiffiffiffiffiffi
rmax ¼ 2pyk Eqaf ð1Þ
where
a leading edge tip amplitude
f frequency of vibration
k radius of gyration of the root section of the blade
y distance from neutral axis to the point of maximum stress
E Young’s modulus
q density
Table 2 Fatigue test results (a) ECM process

of blades a ECM and b CNC
milling process S. Frequency (Hz) Failing Time at failed af
No. “af” (min)
1 334.7 3 25
2 335 3.5 15
3 335.5 3 10
4 334.4 3 20
5 331.7 3 25
6 331.5 3 10
(b) CNC milling
1 328 3 25
2 344 3.5 10
3 341 3.5 15
4 333 3.5 10
5 337 3 20
6 337.5 3 30
Therefore, “af” (amplitude frequency) is a convenient way of comparing the

fatigue life of blades produced by different manufacturing process suggested by
Armstrong [11]. Full-scale fatigue testing under resonance condition is a
well-proven method for evaluating the HCF life of turbine blades [9]. The turbine
rotor blade has been subjected to fatigue testing by Armstrong incremental
amplitude method. It is based on cumulative damage rule of Miner’s hypothesis. In
this method, a single specimen is tested till failure with incremental amplitude
having subjected to an equal number of cycles at each test amplitudes. In the
present investigation, the initial “af” level was 1 ft/s. The natural frequency failing
af and time at failed “af” has been tabulated at Table 2 for blades produced by ECM
and CNC milling routes, respectively.
3.4.1 Estimation of Fatigue Life
The fatigue life has been evaluated by incremental amplitude Armstrong method.
Using Miner’s hypothesis of cumulative damage rule, incremental amplitude fati-
gue life has been converted into life at constant “af” in Eq. (2) [12].
h 1

S N c ¼ Sr Nsc ðNr =Ns Þ þ ðr 1=rÞÞc
1
1
1
ic ð2Þ
þ ðr 2=rÞÞc þ þ ð2=rÞÞc þ ð1=rÞÞc
where
642 B. Sahoo et al.
S Stress level at which life is calculated, here 1 ft/s

N Life at constant stress level S
Sr Final or failure amplitude level in ft/s
Nr No. of reversal completed at failure amplitude
Ns No. of reversal completed at other incremental amplitude
r No. of steps from 1.0 “af” to failure, i.e. (maximum amplitude/incremental
amplitude)
(1/c) fatigue exponent coefficient.
3.4.2 Sample Calculation for Log Life of Blade
The blade sl no. 1 produced by ECM process was having its natural frequency of
vibration as 334.7 Hz, and the failed “af” was 3.0 ft/s with a dwelling for 25 min at
the failed amplitude. Using Eq. (2), the log life at unit “af” level was found to be
9.194103.
Similarly, the fatigue log life for all the sample six blades produced by ECM and
CNC milling batch has been calculated.
3.4.3 Statistical Analysis of Fatigue Life
The fatigue life lies in a log-normal distribution and the sample size, i.e. six (06) is
small compared to the population size of the blades. Therefore, the analysis of
variance (Students “t”-value) has been adopted to compare the fatigue life of the
blades produced by both the processes.
The “t” value is calculated using Eq. (3) [12] with the mean and standard
deviation (SD) of the turbine rotor blades produced by both the routes

Xa Xb
t¼
p 1 p N SD2 þ N SD2 ð3Þ
þ 1 a a b b
Na Nb ðNa þ Nb 2Þ
Xa Mean life of blades produced by ECM

Xb Mean life of blades produced by CNC Milling
SDa Standard deviation of blades produced by ECM
SDb Standard deviation of blades produced by CNC Milling
Na Sample size of blades produced by ECM
Nb Sample size of blades produced by CNC Milling
The mean log life ðX a Þ and standard deviation ðSDa Þ of the blades produced by
ECM process are 9.251717 and 0.507564, respectively. The mean log life ðX b Þ and
standard deviation ðSDb Þfor the blades produced by CNC milling process are
9.713233 and 0.528702, respectively. The degree of freedom is ten.
The “t” value with 95% probability and degree of freedom ten is found to be
from the statistical chart as 2.23. The calculated value of “t” for the blades produced
by ECM route and CNC milling (for 1 in 10,000 at 955 confidence intervals) is
found to be 1.24 and 1.24, respectively. This is less than the reference value of 2.4
indicating that there is no significant difference in fatigue life of the blades produced
by both the routes.
Statistically, both the processes are comparable; however, the blades produced
by CNC route provide superior fatigue life in comparison with blades produced by
ECM route. It is also confirmed from the literature that there is approximately 30–
40% reduction in fatigue life of components produced by ECM process.
4 NDT of Turbine Blade Post-fatigue Test
Blades after fatigue test have been subjected to FPI (Fluorescent penetrant
inspection) to reveal the cracks. It was found that cracks are initiating near the root
of the blade on the LE (leading edge) on all the twelve blades as shown in Fig. 5.
Fig. 5 Cracks at the leading edge on failed blade after vibration fatigue test a ECM, b CNC
milling
644 B. Sahoo et al.
5 Fractograpghy
Turbine blades post-fatigue test were cut along the failed plane to examine the
fatigue origin and crack propagation. There is a separation between progressive
failure region and overload failure region. The progressive failure is designated as
fatigue failure region as shown in Fig. 6. There is a half-moon cut region of the
blade where fatigue crack propagated and fatigue is the dominant damage mech-
anism. The origin of fatigue is at the pressure surface near leading edge. The
magnified view of the fatigue cracks was shown in Fig. 7 for clarity.
Fig. 6 Fracture surface of the blade after fatigue test for both the manufacturing processes at low
magnification a ECM and b CNC milling
Fig. 7 Fatigue origin in turbine blade of different machining processes a and c ECM and b and
d CNC milling
Striations
Overload
Region Eye
Fatigue failure
Region
Fig. 8 Sketch showing various features of fatigue failure of turbine blade observed under SEM
with arrow showing direction of propagation of fatigue crack
Fig. 9 SEM fractograph showing cracks in turbine blades manufactured by different machining
processes a ECM showing multiple cracks and b CNC milling showing only one crack
Sketch showing fatigue failure region, overload failure region and a half-moon
cut shape of the fatigue region with its eye on the pressure surface near leading edge
has been shown in Fig. 8. This is a typical fatigue feature of HCF failure.
Examination of overload region of turbine blade machined by ECM route
revealed many secondary cracks compared only one crack in blade machined by
CNC milling route as shown in Fig. 9.
The surface roughness of pressure surface of blades machined by ECM process
is found to be less compared to the suction surface while the reverse trend was
observed on blades machined by CNC milling process. However, in both the cases,
fatigue failure has from the pressure surface near leading edge. From the above
observations, it can be inferred that influence of surface roughness is not significant
on fatigue life of turbine blade tested at room temperature. Similar findings have
been reported by Pervey and Koster on HCF life [9]. Microhardness is influenced
by the residual stress. The average microhardness of the suction surface is found to
be less compared to that of pressure surface. Residual stress on the suction surface
is of compressive in nature while that of pressure surface is tensile. The literature on
the relation between residual stress with microhardness reveals that there is an
increase in hardness value with compressive stress for medium carbon steel
646 B. Sahoo et al.
investigated by Tosha [13]. However, the crack location by NDT is experiencing

maximum the tensile stress of pressure surface. The origin of fatigue is at the region
of maximum tensile stress. The fatigue failure region looks like a half-moon cut
with its eye on the pressure surface confirms a typical HCF failure.
6 Conclusions
Following conclusions were made from the present investigation:

(a) The maximum level of af the blades survived is 3.5 ft/s. Out of the six blades
produced by ECM route, only one blade could survive up to af of 3.5, while
four out of six by CNC Milling route.
(b) The origin of fatigue is near the leading edge at the pressure surface. There is
separation between progressive failure and over load failure region.
(c) Higher level of residual tensile stress of the blades machined by ECM route
caused lower fatigue life during blade testing.
(d) The maximum stress location approximately coincides with the crack location
as detected by dye penetrant test (DPT).
(e) The half-moon cut region is a typical fatigue feature of turbine blades failing
under high cycle fatigue (HCF).
Turbine rotor blades produced by CNC milling route are found to be superior in
terms of fatigue life.
Acknowledgements The authors acknowledge the support rendered by HAL, Koraput. The
encouragement and guidance from the Chief Executive (A) and Group Director (propulsion),
CEMILAC, Bangalore, are sincerely acknowledged.
References
1. B.A. Coweles, High cycle fatigue in aircraft gas turbines-an industry perspective. Int. J. Fract.
80, 147–163 (1966)
2. Military Specification, MIL-E-5007D, pp. 30–31 (1973)
3. C.T. Sims, W.C. Hagel, The super alloys. Chapter-(12), in Coatings and Protection (Wiley
Science Publication), pp. 341–370
4. D. Ulutan, T. Ozel, Machining induced surface integrity in titanium and nickel alloys: a
review. Int. J. Mach. Tools Manuf. 51, 250–280 (2011)
5. M.C. Hardy, C.R.J. Herbert, W. Li, D.A. Axinte, A.R.C. Sharaman, A. Encinas-Oropesa,
P.J. Withers, Characterising the integrity of machined surfaces in a powder nickel alloy used
in aircraft engines. Procedia CIRP 13, 411–416 (2014)
6. R.M. Saoubi, J.C. Outeiro, O.W. Dillon Jr, J.S. Jawahir, A review of surface integrity in
machining and its impact on functional performance and life of machined products. Int.
J. Sustain. Manuf 1(1/2), 203–236 (2008)
7. T. Ozel, D. Ulutan, Prediction of machining induced residual stress in turning of titanium and
nickel based alloys with experiment and finite simulation. CIRP Ann.-Manuf. Technol. 61,
547–550 (2012)
8. A. Madariaga, D.J. Arrazoa, J.A. Esnaola, J. Ruiz-Hervias, P. Munoz, Evolution of residual
stress induced by machining in a nickel based alloy under static loading at room temperature.
Procedia CIRP 13, 175–180 (2014)
9. P.S. Pervey, W.P. Koster, Effect of surface integrity of fatigue of structural alloys at elevated
temperature. ASTM STP 520, 522–531 (1973)
10. J. Tommy, J. Seidt, M.-H. Herman Shen, T. Nicholas, Development of a novel vibration-base
fatigue testing methodology. Int. J. Fatigue 26, 477–486 (2004)
11. E.K. Armstrong, R.E. Stevenson, J. R. Aeronaut. Soc. 64(592), 117–130
12. Fatigue testing of turbine blade schedule Adour 811 Report No. E/EKA/34960
13. K. Tosha, Influence of residual stress on the hardness number in the affected layer produced
by shot peening, in Second Asia-Pacific Forum on Precision Surface Machining and
Deburring Technology, Seoul, Korea, pp. 48–54 (2002)
Spark Plasma Sintering Process as a Tool
for Achieving Microstructural Integrity
P. Parameswaran, T. Ezhilarasi, Srinivasa R. Bakshi, V. Thomas Paul

and E. Mohandas
Abstract The fact that real materials are not perfect crystals, is critical to materials
engineering as the presence of crystalline defects is the most important feature of
the microstructure which influences the mechanical properties. Exposure of mate-
rials to high temperature for long duration would result in structural failure, when a
sub-size crack grows into a critical level. This paper concentrates on the novel
application of spark plasma sintering (SPS) technique in achieving microstructural
integrity of materials by crack closure using the superior capability of SPS for
annealing the defects in materials. Due to the presence of applied compressive
stress, expansion is restricted and brings about the closure of cracks. The crack
surfaces then come in contact with each other, and energization between the crack
surfaces causes them to bond. The presentation would bring about the nature of
bonding achieved through SPS when two model systems, diffusion bonding of
stainless steel discs and stainless steel with ferroboron powder, were considered and
highlighted its applicability through systematic optimization.
Keywords Spark plasma sintering Diffusion bonding Microstructural integrity
P. Parameswaran (&) T. Ezhilarasi V. Thomas Paul E. Mohandas

Metallurgy and Materials Group, Indira Gandhi Centre for Atomic Research,
Kalpakkam 603102, Tamil Nadu, India
e-mail: param1961@gmail.com
T. Ezhilarasi
e-mail: ezhila.t@gmail.com
V. Thomas Paul
e-mail: thomaspaulv@gmail.com
E. Mohandas
e-mail: mohandas.eladath@gmail.com
S. R. Bakshi
Department of Metallurgical and Materials Engineering, Indian Institute of Technology
Madras, Chennai 600036, India
e-mail: sbakshi.iitm@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_54
650 P. Parameswaran et al.
1 Introduction
The properties of materials are greatly affected by their microstructure, which may
contain defects ranging from point defects to three-dimensional defects. Prolonged
high-temperature exposure of materials may lead to structural failure, when a
microcrack grows into a critical level. This paper concentrates on the novel
application of spark plasma sintering (SPS) technique in achieving microstructural
integrity of materials by crack closure using the superior capability of SPS for
annealing the defects in materials through diffusion bonding.
Since the bonding process involved in crack closure that occurs between two
different planar surfaces is basically driven by mass diffusion, model studies were
carried out to simulate the crack closure problem. Diffusion bonding employing
pressure and temperature, the interface region formed would exhibit a gradation in
microstructure. This gradient could be reduced if the heating rate and holding time
were reduced. In this respect, spark plasma sintering is expected to provide a good
alternative method for diffusion bonding [1]. We have shown using model exper-
iments that cracks could be closed by SPS by a diffusion bonding phenomena.
Further, by using the process, composite discs of SS with ferroboron powders
sandwiched in between are subjected to SPS in order to study the feasibility of
obtaining compacts that have sufficient strength and porosity for use as control rod
materials in fast reactors. The studies show that SPS can be a fast process for
achieving these objectives.
Present study aims at manipulation of diffusion bonding parameters: tempera-
ture, holding time and applied pressure and examination on the achievement of
structural integrity in terms of uniformity in microstructure and hardness.
2 Experimental Details
When spark discharge appears in the gap between the materials of compact, a local
high-temperature state occurs. This causes vaporization and melting of the surfaces
during the SPS process. Constricted shapes or “necks” are formed around the
contact area between the particles. These necks gradually develop, and plastic
deformation progresses during sintering resulting in a sintered compact of over 99%
density. Since only the surface temperature of the particles rises rapidly by
self-heating, particle growth of the starting powder materials is limited to a great
extent.
A stack of 4 SS discs of 500 µm thickness was tightly held by an initial pressure
by placing them in a graphite die. Experimentation was carried out in a vacuum
chamber at different temperatures of 900, 1000 and 1100 °C. A pressing force of
30 MPa is applied at the beginning of the SPS cycle and held constant throughout
the process. The sample was heated to the desired temperature at the heating rate of
100 °C/min and held at that temperature for the dwell time of 5 min. For the second
Spark Plasma Sintering Process as a Tool for Achieving … 651
Fig. 1 Recording of process parameters
case, FeB powder is placed between stainless steel discs and sintered at a tem-
perature of 900 °C keeping the other process parameters same as the previous
experiment. A typical SPS processing schedule is shown in Fig. 1 along with the
recording of process parameters during the process.
The bonded specimens were sectioned and prepared for metallography by
conventional processes of successive grinding and polishing, followed with elec-
trolytic etching in oxalic acid. Scanning electron microscopy (SEM) studies were
carried out by M/s FEI-XL30 SEM. Further, a hardness profile across the sample
was generated by measuring at an interval of 100 µm using Vickers microhardness
tester employing an applied load of 100 g.
3.1 Case Study 1: Similar Materials
The process parameters during the experiment were monitored and given in Fig. 1,
which clearly shows that the pressure and vacuum were constant throughout the
process. On the other hand, a minor decrement in the ram displacement of
*500 µm suggests that bonding is achieved as the initial height being 5 mm is
reduced to 4.5 mm. When the discs were kept in contact, the gap was present which
gets closed and resulted in bonding. The overall reduction from the geometry of the
compact in terms of the height is 500 lm which in comparison with original height
of 2 mm indicates a 25% reduction. In order to assess the quality of the bonding,
cross sections of the same were studied using SEM. Figure 2a–d depicts the sec-
ondary electron images of stainless steel discs treated at 900 °C. It can be observed
that bonding is present between the discs with uniform microstructure at the
interface. However, presence of pores (Fig. 2b) is observed in the interface region.
A close SEM examination of grains at the regions proximate to edges of the
specimen showed the evidence of sensitization as shown in Fig. 2d. Austenitic
stainless steels which have undergone treatment in the temperature range between
500 and 900 °C or have been cooled slowly from annealing temperatures (1000–
1200 °C) become sensitized [2]. During extended exposure of the materials to this
temperature range, chromium-rich carbides tend to precipitate at grain boundaries
(GB). Significant carbide precipitation can result in severe chromium depletion near
the grain boundary, referred to as sensitization. Exposure of sensitized material to
aggressive environments can lead to a breakdown in passivity and consequent
intergranular degradation termed as intergranular corrosion (IGC) and intergranular
stress corrosion cracking (IGSCC).
Fig. 2 Micrographs of diffusion bonding of SS discs sintered at 900 °C (pressure—30 MPa,

heating rate—100 °C/min and holding time—5 min) showing: a overall view of the sectioned
specimen, b and c interface region with pores, d grain boundary sensitization indicated by arrow
In order to understand the present observation, the chosen time of 5–10 min for
SPS was compared with TTS diagram for SS316, which is a plot of exposure time
versus temperature necessary for sensitization. The C-shaped curve [3] separates the
sensitized and non-sensitized regions. The nose of this curve specifies the critical
temperature at which the minimum time for sensitization (tmin) which is 5 min in
the present case. The hashed portion of the figure indicates the process window
employed in the present experiment. It can be inferred that processing time falls
within the sensitization region of the plot (Fig. 3).
Figure 4 shows micrographs from SS disc joints prepared by sintering at
1000 °C. It can be seen that bonding line is observed along the interface line with
good adherence and absence of porosity (Fig. 4c). In the presence of pressure and
electric current, localized necking occurs faster due to joule heating [4].
Consequently, the temperature rises very fast (faster than conventional sintering and
hot pressing), and the densification is completed within few minutes. The coarse
grain structure and twinning were also seen clearly. The evidence of grain boundary
indicates the formation of metallic bonding between the atoms of each side of the
interface [5]. Twinning indicates that the material is subjected to high amount of
stress during the bonding process.
On comparing the micrographs Figs. 2d and 4d, it is clear that the severity of
sensitization increases with sintering temperature. It is very clear that the sensiti-
zation has occurred up to a depth of about 150 µm from the surface of the speci-
men. This is due to the carbon transport at high temperature from graphite mould to
steel specimen which is in intimate contact during the process, as similar to the
Fig. 3 TTS diagram for 316SS adopted from [3]; hashed area shows processing window of the
present study
Fig. 4 Micrographs from diffusion bonded SS discs sintered at 1000 °C (pressure—30 MPa,
heating rate—100 °C/min and holding time—5 min) showing a over all view of sintered joint at
low magnification, b severely sensitized region at the edges, c interface with good bonding but,
d coarse grains and annealing twins, e depth of carbon diffusion
carburization of stainless steel at temperatures above 595 °C reported in the liter-

ature [6, 7].
Additionally, extensive twinning was also present in the grains as indicated by
arrows in Fig. 4c. Twinning indicates that the material is subjected to high amount
of stress. The occurrence of annealing twins which is profound at higher processing
temperature is beneficial to grain refinement. The twin boundary is a strong obstacle
to dislocation slip, which provides an effective strengthening similar to grain
boundaries. Twin boundaries prevent dislocation slip so that dislocation gets piled
up at the twin boundaries. Thus, the presence of extensive twin boundaries would
enhance the mechanical properties [8, 9].
Figure 5 shows the microstructural changes that occurred when the process
temperature is 1100 °C. Figure 5a indicates fine grains suggesting imminence of
recrystallization in the region proximate to the mould. Microcracks were also
present in the region as marked by arrows. There was also an extensive grain
growth along with twinning (Fig. 5b, c) in the central region of the final compacted
specimen. Further, Fig. 5b indicates that at 1100 °C the specimen got heavily
sensitized as depicted by formation carbide films along the grain boundaries.
Irrespective of process temperature, the sensitization effect was at all temperatures.
Fig. 5 Micrographs of SS discs sintered at 1100 °C (pressure—30 MPa, heating rate—100 °C/
min and holding time—5 min) a heavily sensitized area with cracks (arrows) along the grain
boundaries, b annealing twins, c good bonding in the interface region (box) along with grain
growth (arrows)
It is apparent that carburization effect occurred through the surface areas close to
graphite mould [6, 7].
Figure 6 gives the hardness profile measured across the sintered joints processed
at three different temperatures. The effect of temperature on hardness was found to
be marginal. Uniform hardness is seen in samples treated at 900–1000 °C in the
central region. Also, it is observed that the average hardness value decreases with
the increasing processing temperature. The low hardness values at high tempera-
tures indicate that the material lost its strength due to sensitization effect. Based on
microstructure and hardness values, it is optimized that 1000 °C is the best-suited
process temperature considering the integrity of the joint. Similar results were
reported by Li et al. [5], by carrying out through vacuum diffusion bonding.
However, the present results demonstrate saving in time and reduced temperature of
processing in comparison with diffusion bonding as it can be kept below 5 min with
possible increase in pressure to 50 MPa which would form future studies [5].
Considering dominance of sensitization and grain growth in the steels studied, it is
suggested that keeping the time of SPS less than 5 min with a temperature lower
than 1000 °C to avoid grain growth would be optimum. In order to achieve better
compaction, the pressure may be enhanced to *50 MPa in future studies.
Fig. 6 Hardness profile across the bonded SS at different temperatures; note nearly uniform
hardness seen in samples treated at 900–1000 °C
3.2 Case Study 2: Dissimilar Material
In the case of second part of the experiment, i.e. bonding of dissimilar materials of
FeB with SS disc, sintering was carried out at 900 °C at a heating rate of
100 °C/min with a pressing force of 30 MPa for the duration of 5 min. Figure 7
depicts the microstructure of the sectioned specimen. Adherence was observed
between SS and FeB at the interface regions of centrally located discs (marked R1
in Fig. 7a). However, the ends exhibited a poor adhesion (marked R2). The
microstructure of FeB powder region exhibited lots of open and closed pores, which
indicates that good compaction is not achieved, possibly due to insufficient pres-
sure. However, at the interface region, the adhesion was good as no cracks or pores
were seen (R1 in Fig. 7). Hardness profile across the bonded SS disc is given in
Fig. 8. Profile showed a nearly constant hardness values throughout the region.
Further studies have to be carried out with varying pressure and temperature to
arrive at better integrity.
Fig. 7 Micrograph showing the cross section of dissimilar bonding of FeB with SS sintered at
900 °C (a) R1—defect free bonded region, R2—crack along the interface; details of R1 and R2 in
(b and c), respectively
Fig. 8 Hardness profile

along the FeB-SS bond
measured at an interval of
100 µm
4 Conclusions
• Systematic studies of Spark Plasma Sintering have been carried out to under-
stand the effect of sintering parameters on bonding on two systems. (i) With two
similar materials of SS discs and (ii) with dissimilar materials (FeB and SS) of
different forms namely metal and powder in nature.
• It was understood that the densification/bonding is influenced by the sintering
parameters, i.e. temperature, pressing force and the holding time.
• When sintering on similar material was carried out at different temperatures, the
overall compaction of 25% was achieved which is basically measured in terms
of the reduction in the compact height before and after the experimentation.
• The integrity was understood in terms of distinct microstructural features like
(i) sensitization, (ii) recrystallization and grain growth, (iii) formation of car-
bides films along the grain boundary, (iv) precipitation at the surface region and
(v) hardness.
• Hardness profile indicates that the bonded region has the same strength as the
base metal. A marginal variation in the hardness value was observed with
varying process temperature.
• Considering sensitization and dominance of grain growth, it is suggested that
keeping the time of SPS less than 5 min at a temperature lower than 1000 °C to
avoid grain growth. In order to achieve better compaction, the pressure may be
enhanced to *50 MPa in future studies.
• In the case of dissimilar materials, a good adhesion was achieved at the central
interface region. However, the microstructure of FeB powder region is found to
have lots of open pores and closed pores which indicate that good compaction is
not achieved. It could be due to insufficient pressure used in the process. Further
experimentation with increase in applied pressure is essential
Acknowledgements The authors would like to thank Mr. Vasanth, Research Associate,
Department of Metallurgical and Materials Engineering, Indian Institute of Technology Madras,
Chennai for his sincere support in carrying out the sintering process. The authors thank Dr. A.K.
Bhaduri, Director, Metallurgy and Materials Group, IGCAR, Kalpakkam for his constant
encouragement and support during the course of the present work.
References
1. T. Hungrı́a, J. Galy and A. Castro, Spark plasma sintering as a useful technique to the
nanostructuration of piezo-ferroelectric materials, Adv. Eng. Mater. 11(8) (2009)
2. M. Dománková, E. Kocsisová, I. Slatkovský, P. Pinke, The microstructure evolution and its
effect on corrosion properties of 18Cr-12Ni-2,5Mo steel annealed at 500–900 °C. Acta
Polytech. Hung. 11(3) (2014)
3. Practical Guidelines for the Fabrication of High Performance Austenitic Stainless Steels, ISBN
978-1-907470-10-3. Published by the International Molybdenum Association (IMOA),
London, UK
4. M. Gendre, A. Maître, G. Trolliard, A study of the densification mechanisms during spark

plasma sintering of Zirconium (oxy−) carbide powders. Acta Mater. 58, 2598–2609 (2010)
5. S.-X. Li, F.-Z. Xuan, T. Shan-Tung, Fatigue damage of stainless steel diffusion-bonded joints.
Mater. Sci. Eng. A 480, 125–129 (2008)
6. B. Fleischer, J.H. DeVan, J.H. Coobs, in Graphite-Stainless Steel Compatibility Studies, Doc.
No. ORNL-TM-338 (1962)
7. S. Schwarz, R. Oberacker, M.J. Hoffmann, in Microstructure and Mechanical Properties of
Maraging Steel Processed by Spark Plasma Sintering (SPS), Euro PM 2008—Sintering
8. J.-H. Shin, J.-W. Lee, Effect of twin intersection on the tensile behaviour in high austenitic
stainless steel. Mater. Charact. 91, 19–25 (2014)
9. C.X. Huang, G. Yang, Y.L. Gaob, S.D. Wu, Z.F. Zhang, Influence of processing temperature
on the microstructures and tensile properties of 304L stainless steel by ECAP. Mater. Sci. Eng.
A 485, 643–650 (2008)
Part X
Structural Health and Condition
Monitoring
On Increasing Quality of Measurement
Results at Testing State of Construction
Materials
I. P. Miroshnichenko and I. A. Parinov
Abstract The results of the development of new software for processing the results
of measuring the intensities of optical fields of interference patterns, created by
optical means for measuring small linear and angular displacements of control
object surfaces, are discussed. These measurement means are proposed on the base
of a laser two-way interferometer with aligned branches; examples of their use are
given. The offered software was certificated in Russia due to the state registration of
corresponding computer programs. This software allows one to improve the quality
of measurement results in solving various actual practical problems of studying the
properties and processes of defect formation in new materials, diagnosing the state
of construction materials of the power elements of goods by acoustic nondestructive
testing at all stages of their life cycle.

Keywords Laser interferometer Interference pattern Measurement of small

displacements Control object surface
1 Introduction
At present, the introduction of modern methods and means of laser interferometry

opens new possibilities in the solution of various actual practical problems of
studying the properties and processes of defect formation in new materials (in-
cluding layered, polymer, and composite) and also in the diagnostics of the state of
construction materials for power elements of goods at all stages of their life cycle.
One of the most preferred for solving scientific and production problems is a
laser two-way interferometer with combined branches, proposed in [1, 2].
Experimental studies of the functional characteristics of above-mentioned inter-
I. P. Miroshnichenko (&)
Don State Technical University, Rostov-on-Don, Russia
e-mail: ipmir2011@yandex.ru
I. A. Parinov
Southern Federal University, Rostov-on-Don, Russia

https://doi.org/10.1007/978-981-10-7197-3_55
664 I. P. Miroshnichenko and I. A. Parinov
ferometer, modified for the contactless measurement of the small displacements of

control object surfaces in the diagnostics of the state of materials by acoustic
nondestructive testing methods, were described in detail in works [3–5].
New mathematical models and original software for modeling the intensities of
the optical fields of interference patterns, created by above interferometer, modified
for solving measurement problems, were developed. This software was certificated
in Russia due to the state registration of computer programs. Numerical simulation
for various variants of optical schemes taking into account the features of their
constituent parts (geometric characteristics, beam splitter type, etc.) was carried out,
the results and analysis of which are described in detail in works [6, 7].
Based on the results of numerical and experimental modeling, new optical
facilities for contactless measurements of small linear and angular displacements of
the control object surfaces, methods, and means for recording the intensities of the
optical fields of interference patterns, unambiguously associated with the measured
displacements, were proposed. Moreover, the methods and means of protecting the
developed measuring means from the influence of external and internal destabi-
lizing influences on measurement results were developed. Technical solutions,
implemented in the proposed measuring facilities, were patented in Russia. The
functional characteristics of the prototypes of these measuring instruments were
numerically and experimentally investigated and grounded. The results of the
development and scientific calculation and experimental justification of the pro-
posed optical measuring means were described in detail in works [8–10]. Works
[11, 12] describe new scientifically grounded method and an optical device for
measuring small spatial (linear and all angular) displacements of the surfaces of
control objects. Works [9, 10] present technical solutions, patented in Russia,
allowing the use of the developed optical means as components of mobile diag-
nostic setups and measuring controlled small displacements without loss of func-
tional properties and accuracy of characteristics at the site of operation of the
diagnosed objects.
The above-mentioned measuring facilities have been successfully used in
experimental studies of defect formation processes in thin samples of tape
high-temperature superconductors, as well as in the development of methods for
monitoring their quality during manufacture. Descriptions of experimental mea-
suring devices are given in the works [13, 14]. Experimental operation of the above
optical measuring means has shown that one of the directions for improving the
quality of measurement results can be the development and use of new software for
processing the results of measuring the intensities of optical fields of interference
patterns that are unambiguously associated with measurable small displacements of
the surfaces of control objects.
This paper describes the results of the development and use of this software in
solving measurement problems.
On Increasing Quality of Measurement Results … 665
2 Software for Processing the Measurements Results

of the Optical Field Intensities of Interference Patterns
and Examples of Its Use
It is developed software in MathCad 2000 Professional and above versions for

processing the measurement results of the optical field intensities of interference
patterns. These patterns are created by optical means for measuring small linear and
angular displacements of the control object surfaces, proposed on the base of a laser
two-way interferometer with combined branches.
This software includes a set of computer programs, developed on the base of the
same typical algorithm. These programs model a certain variant of information
collection, connected with the intensity of optical field of interference pattern (a
single photodetector device, a group of photodetectors, a matrix of photoreceivers
with predetermined geometric characteristics, etc.). This information is collected
from predetermined areas of the interference pattern (in one or several rings of the
interference pattern, in the selected region of the interference pattern, and so on).
The software also uses various methods of statistical processing of information
obtained.
In each computer program of the software, a typical algorithm for simulating the
process of measuring small displacements of the control object surfaces by optical
interference measuring means is implemented. The essence of this algorithm can be
visually shown in the simplest example of measuring a small linear displacement of
the control object surface by a single photodetector device (e.g., a photodiode of
FD9 type) installed in one of the rings of the interference pattern.
The initial data for processing is a set of images of interference patterns (n = 1, 2
…, N), presented in standard graphic formats. Each (nth) of them corresponds to a
certain moment of measurement (to small linear displacement) during registration
of the image of the interference pattern. In this process, it is performed a repro-
duction of the given value of a small linear displacement of the control object
surface, as well as the geometric characteristics of the photodetector device.
In a whole, this algorithm, implemented in the software, consists of the fol-
lowing steps:
1. Visualization and analysis of the intensity of the optical field of the initial
interference pattern (n = 1), characterizing the initial position of the surface of
the control object, whose appearance is represented in Fig. 1.
2. Visualization and analysis of the intensities of optical fields of all interference
patterns, included in a set of images (n = 1, 2, …, N, e.g., N = 31), corre-
sponding to certain measuring moments (to linear displacements) during reg-
istration (the appearances of interference patterns n = 1, 2 … 9 are shown in
Fig. 2).
3. The choice of the region in the initial interference pattern (n = 1), restricted by
the given geometric characteristics of the photodetector device, by applying
horizontal and vertical lines (marking) and its visualization (see Fig. 3).
Fig. 1 Appearance of the original interference pattern (n = 1)
n=1 n=2 n=3
n=4 n=5 n=6
n=7 n=8 n=9

…………………………………………………….
Fig. 2 Appearances of interference patterns n = 1, 2, …, 9

Fig. 3 Appearance of the

marking on the original
interference pattern (n = 1)

original interference pattern
(n = 1) with the selected
region
4. Visualization of the selected region of the original interference pattern (n = 1),

restricted by the given geometrical characteristics of the photodetector device
(see Fig. 4).
5. Specification and visualization of the selected area in all interference patterns
(n = 1, 2, …, 31), belonging the set by using the parameters of the original
interference pattern (the appearances of the interference patterns n = 1, 2, …, 9
are shown in Fig. 5).
n=1 n=2 n=3
n=4 n=5 n=6
n=7 n=8 n=9

…………………………………………………….
Fig. 5 Appearances of interference patterns n = 1, 2, …, 9 with selected areas
6. Determination of the summary intensity Is for the chosen region for all inter-
ference patterns, belonging the set (n = 1, 2, …, 31), and constructing the
dependence of the change in the calculated summary intensity Is for the chosen
region for each of the registration moments (see Fig. 6).
7. Processing the results obtained using, for example, various built-in tools that
implement methods of statistical processing and forecasting and known soft-
ware. Figure 7 shows, for example, the dependence of the summary intensity Is
for the selected region of interference patterns n = 1, 2, …, 31 after statistical
processing using the built-in tools of MathCad 2000 Professional software.
The developed software also includes computer programs for processing the
results of intensity measurements obtained with the aid of two photoreceivers,
mounted in adjacent interference pattern rings (see Fig. 8), and also when recording
the intensity from the selected region of the interference pattern, characterized by
maximum contrast (see Fig. 9).
Fig. 6 Dependence of summary intensity Is for selected region of interference patterns n = 1, 2,

…, 31
Fig. 7 Dependence of summary intensity Is for selected region of interference patterns n = 1, 2,

…, 31 after statistical processing
The programs described above, included in the software, have been tested and
used in processing the results of experimental studies of the functional character-
istics of promising optical measuring facilities.
Further development of the proposed algorithm and software will be connected
with the improvement of standard unified algorithms for modeling various options
for measuring and processing the results of measurements of small spatial dis-
placements of control object surfaces and also will be directed to automation of the
processing test data.

(n = 1) with the selected
regions for the registration
variant using two
photoreceivers

(n = 1) with selected region
for the registration variant of
intensity from the selected
area
3 Conclusions
New software has been developed for processing the measurement results of the
optical field intensities of interference patterns, created by optical means, for
measuring small linear and angular displacements of control object surfaces, pro-
posed on the base of a laser two-way interferometer with combined branches.
The proposed software has been certified in Russia due to the state registration of
corresponding computer programs and allowed improving the quality of mea-
surement results by solving various actual practical problems of studying the
properties and processes of defect formation in new materials, diagnosing the state
of construction materials of the power elements of goods by acoustic nondestructive
testing methods of control at all stages of their life cycle.
The proposed software can also be successfully used to simulate the measure-
ment of small linear and angular displacements of control object surfaces by cre-
ating new methods for processing the intensities of the optical fields of interference
patterns, created by optical measuring means, developed on the base of laser
interferometers of various types.
The software and technical solutions, described in the article, were presented at
the International Innovation Salons INVENTIONS GENEVA 2017 (Switzerland)
and EUROINVENT 2017 (Romania), where were awarded gold medals.
Acknowledgements The study was performed under partial support of the Russian Foundation
for Basic Research (No. 16-08-00740). I. P. Miroshnichenko and I. A. Parinov acknowledge
financial support of the Russian Ministry of Education and Sciences in the framework of
“Organization of Scientific Research” Government Assignment.
References
1. A.N. Matveev, Optics (High School, Moscow, 1985), 351 p (in Russian)
2. A.S. Batrakov, M.M. Butusov, G. P. Grechka, Laser Interferometry Systems
(Radiocommunication, Moscow, 1981), 456 p (in Russian)
3. I.P. Miroshnichenko, A.G. Serkin, Meas. Tech. 49(5), 22 (2006)
4. I.P. Miroshnichenko, A.G. Serkin, Russ. J. Nondestr. Test. 43(4), 234 (2007)
5. I.P. Miroshnichenko, A.G. Serkin, Russ. J. Nondestr. Test. 44(5), 318 (2008)
6. I.P. Miroshnichenko, A.G. Serkin, V.P. Sizov, Meas. Tech. 50(1), 10 (2007)
7. I.P. Miroshnichenko, A.G. Serkin, V.P. Sizov, J. Opt Technol. 75(7), 437 (2008)
8. I.P. Miroshnichenko, I.A. Parinov, E.V. Rozhkov, A.G. Serkin, V.P. Sizov, in Piezoelectrics
and Related Materials: Investigations and Applications, ed. by I.A. Parinov. (Nova Science
Publishers, New York, 2012), p. 239
9. I.P. Miroshnichenko, I.A. Parinov, E.V. Rozhkov, V.P. Sizov, V.A. Shevtsov, in Physics and
Mechanics of New Materials and their Applications, ed. by I.A. Parinov, S.-H. Chang (Nova
Science Publishers, New York, 2013), p. 145
10. I.P. Miroshnichenko, I.A. Parinov, E.V. Rozhkov, S.-H. Chang, in Advanced Materials—
Manufacturing, Physics, Mechanics and Applications, ed. by I.A. Parinov, S-H. Chang, V.Y.
Topolov. Springer Proceedings in Physics, vol 175 (Springer Cham, Heidelberg, 2016) p. 341
11. I.P. Miroshnichenko, I.A. Parinov, IOP Conf. Ser.: Mater. Sci. Eng. 209, 8 (2017)
12. I.P. Miroshnichenko, I.A. Parinov, J.-K. Wu, W.-L. Hong, M.-Y. Yeh, in Advanced Materials
—Techniques, Physics, Mechanics and Applications, ed. by I.A. Parinov, S.-H. Chang, M.A.
Jani. Springer Proceedings in Physics, vol. 193 (Springer Cham, Heidelberg, 2017), p. 591
13. I.P. Miroshnichenko, I.A. Parinov, E.V. Rozhkov, A.G. Serkin, Metallurgist 50(7–8), 408
(2006)
14. I.P. Miroshnichenko, A.G. Serkin, Metallurgist 54(3–4), 189 (2010)
Damage Growth Study in Unidirectional
CFRP Composites Using Infrared
Thermography
Sukanta Das, Naresh Reddy and Gangadharan Raju
Abstract Structural integrity evaluation of composite structures requires a com-

prehensive understanding of the various failure mechanisms like matrix cracking,
fiber breakage, fiber-matrix interface failure, delamination, and their evolution
across micro-to-macro scales. Non-destructive testing and evaluation (NDT&E)
techniques are applied to composite structures for discerning the various damage
mechanisms and provide information for numerical life estimation tools. Among
the NDT methods, infrared thermography is one of the modern non-invasive
techniques suited for real-time monitoring and gives in situ information regarding
the onset of damages and its growth. In this work, the aim is to investigate damage
initiation and progression in unidirectional carbon fiber reinforced plastic (CFRP)
composite under static loading condition. The passive thermography (PT) and
active thermography techniques are employed in damage growth studies in CFRP
laminate using an infrared camera. The PT experiments were carried out to examine
the damage information in the thermograms under monotonic ramp loading. Image
processing of the thermal images obtained using thermography is used to get
qualitative information about the damage evolution in unidirectional CFRP
specimens.
Keywords Non-destructive testing and evaluation Passive thermography

Damage study Carbon fiber reinforced plastic
1 Introduction
Composite materials are increasingly used in the field of aerospace, marine, auto-
mobile, and wind energy applications due to their high specific strength, stiffness
properties, and tailorability options to design efficient structures. Recently,
S. Das (&) N. Reddy G. Raju

Department of Mechanical & Aerospace Engineering, Indian Institute of Technology
Hyderabad, Kandi, Telangana 502285, India
e-mail: me14mtech11036@iith.ac.in

https://doi.org/10.1007/978-981-10-7197-3_56
674 S. Das et al.
advanced carbon fiber-reinforced polymer matrix composites have been widely

used in primary and secondary load-bearing aircraft structures. Composites usage in
critical structures necessitates a clear understanding of their mechanical behavior
under static and dynamic loading. Unlike metals, composite materials are inho-
mogeneous and anisotropic, which makes their failure more complicated, and
understanding those failures leads to more efficient design.
Infrared (IR) thermography is one of the many techniques used to “see the
unseen.” As the name implies, it uses the distribution (suffix -graphy) of surface
temperatures (prefix thermo-) to assess the structure or behavior of what is under the
surface [1]. Infrared thermography is one of the most modern, efficient
non-destructive techniques, used in aerospace, mechanical, electrical, and civil
engineering. The fundamental principle of this method is that the infrared sensor
senses the emitted infrared radiation from the focused object and processes it in the
form of a surface temperature distribution. Image processing of the acquired ther-
mal images provides qualitative and quantitative information about the presence of
damage in structures. With the advent of technology, thermal imaging has evolved
and has become a useful tool to identify the critical zone well before the failure [1].
There are two approaches in infrared thermography: passive, in which the features
of interest are naturally at a higher or lower temperature than the background and
active, in which an external energy source is required to produce a thermal contrast
ðDTÞ between the area of interest and the background [2].
A good number of the researchers have worked on this topic, and most of them
were interested in cross-ply glass fiber reinforced plastic (GFRP) under dynamic
loading [3, 4], but very few researchers worked on unidirectional (UD) CFRP
composite under static loading [5, 6]. In this work, the objective is to investigate
damage initiation and progression in UD CFRP composite under static loading
condition. The passive thermography experiments were carried out under monotonic
ramp loading to examine the damage initiation information in the thermograms. The
image processing of the recorded thermal images is adopted to get qualitative
information about the damage evolution in UD CFRP samples. For the validation,
the identified damages in the passive thermogram were correlated with the micro-
scopic surface texture images for studying local micro-damages at the lamina.
2 Materials and Test Setup
2.1 Materials and Specimen Preparation
The material used in this study is a carbon/epoxy laminate with 60% volume fiber
content. To maintain the material properties constant for all the test specimens,
coupons are prepared from the same plate. The plate is made with eight layers of
unidirectional dry woven fabric. An epoxy resin of Araldite CY230 mixed with the
Damage Growth Study … 675
Fig. 1 Tensile test specimen’s dimensions
hardener of Araldite HY951 in a ratio of 10:1 was used to impregnate the dry fabric.
The laminated plate was fabricated using hand lay-up followed by vacuum bagging
techniques to improve the fiber volume fraction. The sealed wet laminate was cured
for 24 h at room temperature of 28.5 °C and a vacuum pressure of—87 kPa. The
specimens were endowed with 2 mm-thick aluminum (tt) adhesively bonded tabs,
ensuring the shear forces transferred through the adhesive, and minimize any stress
concentration due to the gripper pressure (700 psi). The schematic of the test
specimen is shown in Fig. 1. For preparing the test coupon, carbide diamond-coated
milling cutter of diameter 6 mm was used to reduce the machining effects at the
edges of coupon.
2.2 Test Setup
The tensile test was carried out on unidirectional CFRP composite coupons under
displacement control mode in MTS Landmark 370, having a maximum load
capacity of 100 kN. During tensile testing, the IR thermal camera from FLIR
SC5500 equipped with a cooled indium antimonide detector which facilitated to
record the smallest of temperature differences. It is working in the waveband range
of 2.5–5.1 lm and produces thermal images of 320 256 pixels. The camera has
Focal Plane Array (FPA) InSb sensor with a capacity of 25 mK thermal sensitivity
[7]. The micro-cracks, delamination, fiber damages, and also other damages were
observed and recorded using this infrared (IR) camera.
For passive thermogram recording, the IR camera was placed approximately
510 mm far from the surface of the specimen loaded into the MTS machine. The
experimental test setups are shown in Fig. 2. The camera was connected to a
computer for recording the thermogram data. The recording and post-processing of
the data were done using Altair software. This software allows highlighting the
damaged zones by performing a pixel-by-pixel analysis of the temperature data. To
reduce the environmental noise in the recorded thermogram, a non-transparent
plastic sheet covered the whole experimental setup (i.e., MTS machine and the IR
camera). The test room was maintained at a constant temperature of 23 °C.
For inspection of the damaged specimens, active thermography adopted using
the same camera additionally used two halogen lamps having a maximum capacity
of 2.5 kW for the heat wave. For all active thermography tests, the lamps and
camera have been kept at a distance of 700 and 550 mm from the specimen,
respectively. All the active thermography tests were conducted using IrNDT
676 S. Das et al.
Fig. 2 Test setups a passive thermography, b Active thermography
software with lock-in and transient thermography module supported by automation

technology.
For a microscopic analysis of the preidentified damage sections from the image
processing thermogram, Olympus STM6 optical microscope with the objective lens
“MPlanFL N 10/0.30” having a capacity of 10 magnification with a numerical
aperture value of 0.3 was used.
3 Experimental Tests
3.1 Static Loading
Static monotonic tensile tests were performed in the displacement control mode
with a crosshead speed of 1 mm/min. The three specimens were loaded on an MTS
universal testing machine, and during each tensile test, the surface thermogram of
the specimens was recorded by the thermal camera. The aim was to capture any
onset of micro-damages indication in the temperature trend and correlate it with the
material’s stiffness. Three specimens were loaded up to stress level of 500, 700, and
900 MPa, respectively, and the corresponding temperature data acquisition fre-
quency was set as 120, 85, and 55 Hz.
3.2 Active Thermography
After tensile loading, every loaded specimen was inspected through active (i.e.,
lock-in and transient pulse techniques) thermography to capture the macro-damages.
A set of different thermal excitation frequencies starting from 2, 1, 0.34, and 0.2 Hz
with corresponding thermal periods of 5, 5, 4, and 4 was selected for the lock-in
thermography. For transient pulse thermography, a rectangular pulse width of 1.5,
2.25, and 3 s with a period of 10, 15, and 20 s, respectively, was adopted. The
recorded thermograms were evaluated using pulse phase techniques to plot the phase
diagram. The image acquisition frequency of 300 Hz was chosen for both the
techniques. The aim of these tests was to plot the phase diagrams of the loaded
specimen and correlate phase delay response of the damage location with the passive
thermogram’s damage indications.
3.3 Microscopic Analysis
For the validation of the thermogram techniques, the specimens were cut along the
transverse direction at passive thermogram’s damage indications and inspected the
sections under the optical microscope to capture the micro—macrodamages. The
damage sections of the specimen were cut using carbide diamond-coated milling
cutter of diameter 3 mm. To reduce machining defects, a different set of cutting
parameters, i.e., cutting speed and depth of cut were chosen to be applied on an
unloaded specimen. After analyzing all the microscopic images of different
machining sections of the healthy specimen, it was concluded that the least number
of defects was observed for the cutting speed of 1000 RPM with the 0.5 mm depth
of cut. These optimum cutting parameters were adopted to cut the different loaded
specimens for microscopic analysis.
4 Test Results and Discussions
4.1 Experimental Results
The tensile test result shows (Fig. 3) an approximately linear behavior of the stress–
strain at the initial stage of the test. Later, the linear stress–strain relation starts
changing its slope when the specimen reaches to a stress value of 270 MPa (Fig. 4).
This change in slope phenomena can be correlated with the starting of different
damage mechanisms, which continue until it reached the ultimate strength.
Thermogram of the loaded specimens is allowed to do both qualitative and
quantitative analysis of the damages. For qualitative analysis, it was tough to locate
the hottest zone of the specimen from the raw thermograms. At low-stress level, the
raw thermograms did not indicate any the thermal contrast between damage initi-
ation temperature and the specimen temperature. To enhance the damage zone,
corresponding to the hottest zone of the specimen, an image processing was done.
For image processing, thermograms were first normalized by dividing the first
frame of the recorded film and then performed subtraction operation between the
678 S. Das et al.
Fig. 3 Temperature and stress variation plot with respect to time for specimen 03
Fig. 4 Linear variation of stress and temperature in the first phase (thermoelastic effect)
current frame and the previous frame of the normalized film [5]. These arithmetic
operations have been performed using Altair software. Figure 5 represents the
maximum normalized subtracted temperature of the small area of interest
(AoI) versus time. A threshold value of 0.006 of the “maximum normalized sub-
tracted temperature” was chosen for identifying the damages in the specimen. All
Fig. 5 Maximum normalized subtracted temperature timing graph for specimen 01
the frames which are above the threshold value were inspected for the hottest zone
and calculated for the damage initiation. Figures 6, 7, and 8 represented the set of
thermograms, which were above the threshold value specimens 01, 02, and 03,
respectively. Every thermogram within Figs. 6, 7, and 8 represents the evolution of
damages with time. For example, in Fig. 6 representing the key thermograms of
specimen 1, the first thermogram represents the first critical temperature rise appear
in the specimen at 20.12 s. This frame is considered for the damage initiation of the
UD CFRP composite under static loading, and the corresponding stress value is
considered as damage stress.
For quantitative analysis, temperature variation within the specimen due to the
external load can be obtained using the thermoelastic effect. All the tested speci-
mens showed a similar temperature variation with time. The temperature profile can
be schematized in two phases. The first phase is characterized by an approximately
linear variation of the temperature with time. The linear increase in temperature
observed in CFRP specimens is in agreement with the thermoelastic effect
ðDT ¼ km Trm Þ, assuming the negative coefficient of thermal expansion (CTE) in
the fiber direction. The second phase is characterized by a nonlinear variation in the
temperature with time. This can be attributed to the initiation and growth of local
micro-damages arising from the pre-existing manufacturing defects [3, 5, 6]. These
damages can be confirmed by micrographic analyses using an optical microscope.
The temperature and the stress trends are represented together on the same graph
to illustrate the thermoelastic effect due to an external load. In Fig. 4, the first phase
is characterized by a linear trend, which can be described with an equation obtained
through a linear regression analysis. The end of the first phase might be correlated
to the end of the linear thermoelastic behavior of the material and beginning of the
micro-damages in the material. Hence, the stress value corresponds to the end of the
linear thermoelastic phase, considered as damage initiation stress. The average
value, obtained from static tests, was equal to 221 MPa and represents the damage
initiation stress point ðrd Þ for the unidirectional CFRP specimens [3, 5].
680 S. Das et al.
Fig. 6 Normalized subtracted thermogram with damage indications for specimen 01 at a different
time (t)
time (t)
For active thermography, both lock-in and transient pulse thermography were
done for all specimens. The lock-in test result shown in Fig. 9a represents the phase
information on loaded specimen 01 (500 MPa) and healthy specimen. It is clearly
observed from the phase diagram of the loaded specimen and healthy specimen that
damages are located the top portion of the specimen 01. After pixel-by-pixel
analysis of the phase value, we summarize that the damages location of the
time (t)
(a)
(b)
Loaded specimen 01 (500 MPa)
(2) (1)
Healthy specimen
Zoomed part of the damage

location from figure 9(a)
(c)
Fig. 9 Phase diagram and temperature variation of point (1) and (2) for specimen 01. a Phase plot
using lock-in thermogram, b zoomed part of the damage location from the (a), c temperature
variation of point (1) and (2) over time
682 S. Das et al.
Table 1 Damage dimensions from microscopic images for specimen 01

Section no. Damage Damage Damage Damage Damage
no length (pixel) length (mm) width (pixel) width (mm)
1 1 570 0.342 138 0.0828
2 474 0.2844 234 0.1404
3 504 0.3024 186 0.1116
4 1392 0.8352 357 0.2142
2 1 951 0.5706 135 0.081
2 705 0.423 189 0.1134
3 1 823 0.4938 159 0.0954
2 816 0.4896 108 0.0648
3 672 0.4032 273 0.1638
Fig. 10 Microscopic images. a Damage 01 and 02 after image stitching adjacent to damaged
image (b) and (c). b Microscopic image of damage 01 at 10 zoom for Sect. 1 of specimen 01.
c Microscopic image of damage 02 at 10 zoom for Sect. 1 of specimen 01. Dimensions of the
damages are given in Table 1
specimen 01 is nearly same with passive thermogram damage indications. The

temperature variation in the damage point (1) and health point (2) corresponds to
loaded specimen 01 and healthy specimen which is shown in Fig. 9c. The deviation
in cooling profile between healthy and loaded specimen indicates the damage
present at the point (1). The above analysis clearly indicates the damages for the
specimen 01 and agrees with the respective passive thermogram damage
indications.
To validate passive thermogram damage indications, specimens were cut along
the damage indications and inspected under a microscope. For specimen 01, a list of
the damages and their corresponding dimensions were mentioned in Table 1.
Figure 10 represents the microscopic image of one of the damaged sections of
specimen 1. The microscopic image of the section was able to highlight local
damages. In Fig. 10b, c, damages are representing the local damages.
5 Conclusions
Quasi-static tensile experiments were carried out on unidirectional CFRP specimens

to study the damage initiation and progression using IR thermography. Both passive
and active thermography techniques were used to monitor the damage progression
in UD laminates. Passive thermography results were able to capture the thermoe-
lastic effect showing a temperature increase with loading within the elastic limit.
Also, using PT, we were able to identify the damage initiation stress point (ap-
proximately 221 MPa) beyond which the temperature variation becomes nonlinear
with load showing the growth of damage in CFRP specimens. Further, image
processing operations were performed on thermograms acquired using PT, and a
threshold value was adopted to identify the damages in the CFRP specimens.
Active thermography techniques, namely lock-in and transient pulse thermography,
were then used to validate the damage identified using PT. Finally, the specimens
were cut at the damage location defined by PT, and the micrographic images were
captured using an optical microscope. Also, an experimental study was carried out
to determine the optimal cutting speed and depth of cut to minimize the machining
damage. The microscopic images obtained at the damaged section agreed with the
results achieved by PT.
Acknowledgements The authors wish to thank prof. (Asso.) M. Ramji, for allowing to access
material characterization laboratory, IIT Hyderabad, and also like to thank Mr. M. Pravin, com-
posite manufacturing Lab, for helping tensile specimen preparation, Swaraj Kumar Tripathy,
M. Tech. student, IIT Hyderabad, and Divya Selvaraj, a summer intern student from NIT Trichy
for helping in microscopic analysis.
References
1. X.P.V. Maldague, Theory and Practice of Infrared Technology for Nondestructive Testing
(Wiley, New York, 2001)
2. X.P.V. Maldague, Introduction to NDT by active infrared thermography. Mater. Eval. 60(9),
1060–1073 (2002)
684 S. Das et al.
3. C.C.F.L.L. Vergani, C. Colombo, F. Libonati, L. Vergani, Fatigue damage in GFRP. Int.

J. Struct. Integr. 3(4), 424–440 (2012)
4. W. Harizi, S. Chaki, G. Bourse, M. Ourak, Composites: part B mechanical damage assessment
of glass fiber-reinforced polymer composites using passive infrared thermography. Compos.
Part B 59, 74–79 (2014)
5. B. Rodriguez, C. Galleguillos, R. Fernández, F. Lasagni, Passive infrared thermography for
damage monitoring during structural testing of CFRP parts, in Proceedings of 16th European
Conference Composite Materials, pp. 22–26 (2014)
6. N.I. Baurova, W. Hao, O. Xiao, Microstructure of carbon fiber and carbon reinforced plastic.
J. Mater. Sci. Chem. Eng. 1, 28–32 (2013)
7. FLIR® Systems, FLIR SC5000 Series. FLIR® Systems. Available: http://www.flir.co.uk/cs/
display/?id=42577. Accessed 10 May 2016 (Online)
Online Damage Detection in Metallic
Cylindrical Shells Using Guided Wave
Modes by Time Reversal Method
P. Jayesh and Christudas R. Bijudas
Abstract Online damage detection and structural health monitoring (SHM) of

aerospace structures are having great scientific and economical importance, since
maintenance costs can be considerably reduced by introducing condition-based
structural maintenance instead of time-linked preventive maintenance. Active
structural health monitoring using guided wave modes is gaining importance after
the introduction of lightweight piezoelectric wafer transducers (PWTs) since the
mass penalty associated with these systems is considerably less. Multiple research
papers are generated worldwide on the use of guided Lamb wave modes for active
defect detection in thin platelike structures. Since so many civil and aerospace
structures are cylindrical in shape, studies on the use of guided wave modes for
SHM of cylindrical shells are gaining importance. Studies on the use of guided
wave modes for SHM of cylindrical shells and pipes are done by a few authors by
pulse echo method. In this chapter, the methods to generate different guided wave
modes in metallic isotropic cylindrical shells, dispersion characteristics of wave
modes in cylindrical shells, their interaction with defects, etc., are studied.
Moreover, the structural health monitoring of cylindrical shells using the time
reversibility concept adopted from guided Lamb wave mode is also studied.
Numerical modelling of propagation of guided wave modes in plates and cylin-
drical shells is done using finite element methods, and comparisons are made.
Experimental evaluation on the propagation of guided waves in cylindrical shells is
carried out. Dispersion characteristics of guided wave modes in cylindrical shells
and the signature of guided wave modes for different defects are studied by
numerical modelling and experiments. A tomography scheme is worked out which
can predict the presence of defects and its propagation, which can be used for
condition-based structural maintenance of aerospace structures. A comparison
between the numerical modelling and experimental studies is done, and damage
indices based on time-reversed reconstruction are generated.
P. Jayesh C. R. Bijudas (&)

Indian Institute of Space Science and Technology, Thiruvananthapuram 695547, India
e-mail: jayeshpkalappila@gmail.com

https://doi.org/10.1007/978-981-10-7197-3_57
686 P. Jayesh and C. R. Bijudas

Keywords Cylindrical shells Guided wave modes Structural health monitoring

Time reversibility Tomogram
Nomenclature
SHM Structural health monitoring
PWT Piezoelectric wafer transducer
1 Introduction
Online damage detection and prognosis of cylindrical shells, especially pressurized

vessels, fascinated the researchers worldwide over the last century. A breakthrough
in structural health monitoring happened after the introduction of lightweight, cheap
and extremely thin piezoelectric wafer transducers (PWTs). Guided wave modes
can be generated by supplying electric signal to the PWTs. SHM systems consist of
permanently embedded PWTs in structures, and the output signals are monitored
and analysed to get the required information on the health of the structure.
Giurgiutiu [1] earlier introduced the idea of embedded ultrasonic structural radar
(EUSR) for damage detection in plates. In [2, 3], authors extended this concept to
cylindrical shells. Muc and Stawiarski [4] give a comparison between experimental
and numerical studies on the propagation of elastic guided waves in cylindrical
shells. Li et al. [5] studied the effect of pressurized fluid in the wave propagation
characteristics in cylindrical pressure vessels and also investigated the mode at
which the attenuation due to interaction with fluid is less. Gresil et al. [6] give
detailed guidelines on the usage of finite element method on the wave propagation
problems. Ing and Fink [7] carried out experiments on self-focusing process of the
time reversal mirrors and time recompression for detection of flaws. Poddar and
Giurgiutiu [8] used a method to avoid boundary reflections in experimental studies,
by applying modelling clay near the shell boundaries. Sohn [9] concluded that
baseline data-based damage detection techniques will have limited practical
applications unless otherwise new methods are discovered to correctly compensate
for the environmental and operational variabilities. Qiang and Shenfang [10] fil-
tered out the scattered signals from defect locations and time reversed it so that
these waves got focused at scatterers. In order to separate the scattered signal,
authors introduced a new unique PWT arrangement and a time window function is
then used to separate out the scattered signals from the source and boundary
reflection signals. Harley et al. [11] used time reversal technique to effectively
remove the coherent noise and also to increase the signal-to-noise ratio. Doing this,
the sensitivity of the damage detection technique is considerably increased. Park
et al. [12] attempted the near perfect reconstruction of the original signal after time
reversal. A narrowband excitation is used with limited frequency components, so
that after time reversal reconstruction uniform scaling of different frequency
Online Damage Detection in Metallic Cylindrical Shells … 687
components is possible. Hence, the original input signal can be reconstructed.

Wang et al. [13] described a synthetic time reversal imaging method for structural
health monitoring of plates. Ying et al. [14] carried out time reversal wave prop-
agation experiments in pipes, so that they could able to collect the signatures for
different kinds of defects in a pipe. Ying et al. [14] experimentally simulated welded
butt joints, effect of internal pressure in pipe, pipe with lateral defect, pipe with
longitudinal defect and pipe with corrosion-like defect. In this chapter, the propa-
gation characteristics of elastic guided waves in cylindrical shell structures in
longitudinal and circumferential directions are evaluated experimentally, numeri-
cally and analytically. Time reversible recompression of guided waves is used as a
basis for a new tomography scheme which can detect the presence and growth of
damages in cylindrical shell structures.
2 Experimental and Numerical Studies
2.1 Experimental Studies
Experimental investigation is carried out on a cylindrical shell structure of diameter

0.6 m and length 1.2 m. The cylindrical shell is made of 1.5-mm-thick AA 8081
material in H14 temper condition. Agilent 33522A arbitrary wave form generator is
used for generating the eight peaked tone burst. A Hanning window is used to get a
narrow frequency band of the input signal. The oscilloscope used is Agilent DSO
X2004A, while 0.5-mm-thick 7-mm2 PWT 5A sourced from Sparkler Ceramics is
used for exciting the waveform. PWT transmitter/receiver pairs are attached to the
cylindrical shell in longitudinal and circumferential directions. Experimental set-up
and schematic diagram showing the location of attached PWTs are shown in Fig. 1.
The output signal on careful analysis reveals that the input wave packet is
divided into multiple submodes due to mode separation and dispersion. The first
wave mode is a symmetric wave mode, and it is longitudinal L (0, 2) mode of the
cylindrical shell. The second wave mode is an antisymmetric one, and it is the
longitudinal L (0, 1) mode. This can be further understood by plotting the
Fig. 1 a Experimental set-up and b locations of PWT in experimental set-up

Fig. 2 Group velocity dispersion plot (left) and longitudinal (right) circumferential
displacement behaviour at two points, inside and outside the cylindrical shell.
Group velocities of the wave modes for different frequencies are estimated and
compared with the results of analytical [15] and experimental predictions as shown
in Fig. 2. The analytical estimation of wave propagation characteristics in the
cylindrical shell structure is done using a closed-form solution method as described
by Sanderson [15]. Usually, the scatterings from the defects sink into the reflections
from the boundaries and make it difficult to achieve the damage detection in
cylindrical shell structures. The reflections from the boundaries are dampened by
placing pottery clay over the boundaries of the structure. The schematic diagram of
the pottery clay attached to the boundaries of the cylindrical shell is given in Fig. 3.
A typical plot showing comparison of output voltage signals after applying the clay
Fig. 3 Schematic diagram showing the application of pottery clay over the cylindrical shell
boundaries
Fig. 4 Comparison for

output voltage signals without
and with pottery clay covered
the boundaries—100 kHz
over the boundaries is shown in Fig. 4. It can be seen that the reflections from the
boundaries are almost completely eliminated. Fast Fourier transform (FFT) of the
output signals is done to estimate the frequency content, and it is confirmed that no
higher multiples of the input frequency are present in the output signals. This is an
indication of the damage-free structure.
Time reversal of the output wave components is carried out, and the amplified
time-reversed signals are transmitted from the receiver PWT. It is found that the
original input signal is reconstructed with near perfect accuracy. To compare the
original signal and reconstructed signal, a quantity known as correlation factor is
used which is defined as ‘correlation factor = Covariance(x, y)/(Sqrt(Variance
(x)) * Sqrt(Variance(y)))’. The observed correlation factor for defect-free wave
propagation paths is as high as 99.5%. Fully reconstructed recompressed signal
along with the original initial input signal (amplitude normalized) is shown in
Fig. 5.
An experiment is conducted on the proposed tomography scheme to detect the
presence of damage on the cylindrical shell structure. For this, a 1.5-mm-thick
100-mm-square cylindrical shell piece is attached to the surface of the cylinder
Fig. 5 Comparison between TR output signal and original input signal (for 300 mm a axial and
b circumferential wave travel)—100 kHz
Fig. 6 a Experimental set-up with 600 mm 600 mm 9 PWT array and b the possible wave
propagation paths
using epoxy adhesive (Araldite) at one edge of the metallic piece (covering almost
25% of the surface area). The metallic piece is placed such that the two cylindrical
surfaces are in contact. When a wave is propagating, there will be a slapping action
between the metallic surfaces and it will simulate an intentionally introduced
damage. Experimental set-up with 600 mm 600 mm 9 PWT array is used as
shown in Fig. 6a. The possible wave propagation paths between the PWT elements
are given in Fig. 6b.
Time reversal reconstruction is attempted in all the wave paths, and it is found
that the reconstruction is very poor in the case of paths with artificially introduced
damages as shown in Fig. 7. The correlation factor observed for these paths is as
less 15%, while it is more than 99% in case of all other defect-free paths. To
enhance the imaging process, a numerical quantity labelled as inverse correlation
factor (100-correlation factor in %) is used. A computer program is generated for
this end use. Coordinate locations of the PWT array and the correlation factors for
different paths are input to the program. The number of pixels in the cylindrical
shell image constructed by the computer program can be adjusted across axial and
Fig. 7 Experimental time

reversal reconstruction for a
wave propagation path with
damage
Fig. 8 a Constructed tomogram and b different steps in construction of the tomogram
tangential directions to change accuracy of damage imaging. The program will find
the pixels through which each wave propagation paths are passing through and allot
the corresponding inverse correlation factors to all the pixels associated with the
path. Inverse correlation factors are assigned to pixels in other wave propagation
paths also. The pixels at which the wave propagation paths are crossing, the inverse
correlation factors will be added up. When multiple paths passing through a defect,
all the inverse correlation factors will added up to a value higher than the sur-
rounding pixels. Then, these pixels at the defect location will be highlighted with a
different colour. The constructed tomogram is shown in Fig. 8a. The entire process
is schematically shown in Fig. 8b. The tomogram gives the exact location and
visualization of the defect.
2.2 Numerical Simulations Using Finite Element Method
Both axisymmetric and plane strain analyses are carried out using Ansys 14.5 FEM
software. The 0.6-m-diameter, 1.2-m-long and 1.5-mm-thick cylindrical shell is
modelled. Two PWTs are modelled on outer surface of the cylindrical shell. The
bottom and top nodes of the piezoelectric devices modelled are voltage coupled.
The load input is given as a voltage signal in one of the PWTs at the topside master
node. The mode separation and dispersion characteristics as observed in experi-
ments are reconfirmed using numerical modelling. The input wave packet is sep-
arated into multiple subwave modes. The first wave mode is L (0, 2) mode, while
the second one is L (0, 1) mode. The group velocity dispersion plot is estimated
from 2D FE analyses. FE analyses repeated with plane strain analyses. Propagation
of waves in the circumferential direction can be better visualized using plane strain
analysis. The FE models are shown in Fig. 9. The propagation of the wave packet at
different time intervals is shown in Fig. 10 (both symmetric and antisymmetric
modes are there, even though the antisymmetric mode is clearly visible due to large
out of plane displacements). FE analyses are carried out to simulate the entire
tomography experiment. All the wave propagation paths are modelled, and the time
Fig. 9 FE models for a axisymmetric analyses and b plane strain analyses
Fig. 10 Propagation of wave through the cylindrical shell at different time intervals for
a axisymmetric analyses and b plane strain analyses
reversal reconstruction is attempted. The results are matching with the experimental
predictions. FE model is shown in Fig. 11a. The propagation of wave packet
through the defect is shown in Fig. 11b. Figure 12 shows the numerically simulated
time reversal reconstruction attempted for a path with artificially introduced dam-
ages. It can be seen that the time reversal reconstruction is poor.
Fig. 11 FE analysis for the propagation of wave packet through the path with damage a FE model
and b propagation of wave through the damage
Fig. 12 Comparison
between the numerically
simulated TR output signal
and original input signal over
a typical propagation path
which contains the defect
(poor reconstruction)
3 Conclusions
• Near perfect time reversal reconstruction by time recompression is carried out in

cylindrical shell type structures.
• All the experimental wave mode packets are matching with the finite element
predictions.
• Dispersion plot for the cylindrical shell structure considered is plotted using
experimental evaluation and compared with analytical and numerical predic-
tions. Group velocities of different modes are matching with FE predictions for
all the frequencies considered.
• A tomography scheme is developed based on the time reversal reconstruction of

guided wave modes which becomes a basis for a baseline-free structural health
monitoring technique.
References
1. V. Giurgiutiu, Tuned lamb wave excitation and detection with piezoelectric wafer active
sensors for structural health monitoring. J. Intell. Mater. Syst. Struct. 16 (2015)
2. V. Giurgiutiu, J. Bao, Embedded-ultrasonics structural radar for in situ structural health
monitoring of thinwall structures, in AIAA (2004)
3. V. Giurgiutiu, L. Yu, Embedded ultrasonic structural radar with piezoelectric wafer active
sensors for damage detection in cylindrical shell structures, in AIAA (2004)
4. A. Muc, A. Stawiarski, Modeling damage in cylindrical shells using elastic wave based
techniques, in 18th International Conference on Composite Materials
5. F. Li, F. Liu, X. Sun, H. Li, G. Meng, Propagation of guided waves in pressure vessels.
J. Wave Motion (2014)
6. M. Gresil, V. Giurgiutiu, Y. Shen, B. Poddar, Guidelines for using the finite element method
for modeling guided lamb wave propagation in SHM processes, in 6th European Workshop
on Structural Health Monitoring
7. R.K. Ing, M. Fink, Time reversed lamb waves. IEEE Trans. Ultrason. Ferroelectr. Freq.
Control 45(4) (1998)
8. B. Poddar, V. Giurgiutiu, Experimental validation of analytical model for lamb wave
interaction with geometric discontinuity. Proc. SPIE 9437, 795–803 (1997)
9. H. Sohn, Effects of environmental and operational variability on structural health monitoring.
Philos. Trans. Roy. Soc. (2007)
10. W. Qiang, Y. Shenfang, Baseline free imaging method based on new PWT sensor
arrangements. J. Intell. Mater. Syst. Struct. 20 (2009)
11. J. Harley, N. O’Donoughue, J. States, Y. Ying, J. Garrett, Y. Jin, J. Moura, I. Oppenheim, L.
Soibelman, Focusing of ultrasonic waves in cylindrical shells using time reversal, in Seventh
International Workshop on Structural Health Monitoring, 2009
12. H. Park, H. Sohn, K. Law, C. Farrar, Time reversal active sensing for health monitoring of a
composite plate. J. Sound Vib. 302 (2007)
13. C. Wang, J. Rose, F. Chang, A synthetic time reversal imaging method for structural health
monitoring. Smart Mater. Struct. (2003)
14. Y. Ying, J. Harley, J. Garrett, Y. Jin, J. Moura, N. O’Donoughue, I. Oppenheim, L.
Soibelman, Time reversal for damage detection in pipes, in SPIE conference (2010)
15. R. Sanderson, A closed form solution method for rapid calculation of guided wave dispersion
curves for pipes. Wave Motion (2014)
Author Index
A Choudhuri, Gargi, 611

Akshitha, G., 103 Cinitha, A., 267, 437
Alfaori, Qusay, 169
Andronik, A.V., 549 D
Anilchandra, A.R., 195 Daga, Rajesh, 369
Arora, Punit, 611 Das, Sukanta, 673
Dattaguru, B., 559
B Deka, Nipal, 593
Bade, Simhachalam, 303 Dhiman, Abhijeet, 481
Badikov, K.A., 549 Dubey, P. N., 493
Bakshi, Srinivasa R., 649 Duong, L.V., 291
Balasubrahmanian, A.K., 395
Bharathanatha Reddy, R., 65 E
Bharath, K., 559 Endo, Masahiro, 541, 599
Bhasin, V., 419, 611 Ezhilarasi, T., 649
Bhat, Praveen, 385
Bhaumik, S.K., 65 G
Bijudas, Christudas R., 685 Gandhi, P., 419
Biswas, Swati, 117 Gangadharan, R., 527
Bojja, Ramesh, 27, 195 Glot, Irina, 451, 503
Boniface, Vinodkumar, 279 Goyal, Sunil, 353
Bose, S., 205 Gujar, Shriram, 77
Bykov, Anton, 503
H
C Holdsworth, Stuart R., 1, 341
Chaitanya Sagar, T., 253
Chakraborty, D., 461 I
Chakraborty, Debaleena, 461 Ikeda, Yuichi, 541
Chandan Reddy, V.R., 407 Ingole, S.M., 395
Chandra Kishen, J.M., 147 Iyer, Nagesh R., 157
Chaudhry, V., 395
Chebanenko, V.A., 291 J
Chen, Zhen, 1 Jagannathan, N., 27, 195
Cherpakov, A.V., 291 Jaladeen, S., 325
Chikmath, L., 559 Jayaram, V., 205
Chinthapenta, Viswanath, 253 Jayesh, P., 685

https://doi.org/10.1007/978-981-10-7197-3
696 Author Index
Jensen, Hanna, 169 R

Jensen, Morten, 169 Raghava, G., 419
Jonnalagadda, Krishna N., 481, 593 Raghavan, Kotur S., 53
Raghavendra, K., 65
K Raghu Prasad, B.K., 157
Kalyanasundaram, Valliappa, 341 Rahul, G.R., 205
Kesavan, K., 437 Raju, Gangadharan, 517, 673
Khaderi, S.N., 315 Ramachandra Murthy, A., 157
Khan, I.A., 419 Ramesha, H., 15
Kodag, Pradeep B., 215 Ramesh, R., 579
Kumar, Gurunathan Saravana, 623 Ramji, M., 235, 315, 527
Kumar, Jivan, 117 Ramu, Palaniappan, 623
Kumar, Suranjit, 419 Ranganath, V.R., 225
Kumar, Suresh, 225 Ravi, S., 353
Reddy, G.R., 493
L Reddy, Naresh, 527, 673
Laha, K., 353 Rengaswamy, Jayaganthan, 623
Lakshmana Rao, C., 303 Rozhkov, E.V., 291
Lakshminarayana, H.V., 15
Lakshminarayana, S.C., 93 S
Sahoo, Benudhar, 635
M Sahoo, D.K., 611
Madan, M., 65 Sahoo, Kanhu Charan, 353
Mahendra Babu, N.C., 15 Sahoo, P.K., 77
Manjunatha, C.M., 27, 39, 195 Sakthivel, K., 27
Manjuprasad, M., 39, 77 Samal, Mahendra Kumar, 369
Master, Bakul, 385 Saravanan, M., 419
Miroshnichenko, I.P., 663 Sarkar, Rosy, 325
Miyakawa, Susumu, 541 Satishkumar, V.N., 117
Miyamoto, Nobuyuki, 541 Satpathy, R.K., 635
Mohandas, E., 649 Savkin, A.N., 549
Mohan Kumar, M., 103 Saxena, Ashok, 169
Muktibodh, U.C., 395 Schwarzkopf, Erik A., 469
Murthy, K.S.R.K., 461 Sedov, A.A., 549
Murugan, R., 579 Selvaraj, P., 407
Seshadri, Matta, 235
N Seshagirachari, M., 195
Narendra Babu, S.N., 117 Shailesh Rao, A., 103
Neergat, M., 481 Shardakov, Igor, 129, 451, 503
Nimje, Yashdeep P., 517 Sharma, Anil Kumar, 407
Nishimura, Yoshiro, 541 Sharma, Mukhar, 611
Shepard, Michael J., 469
P Shestakov, Aleksey, 129, 451, 503
Padmanabhan, K., 579 Shuaib, Muhammad, 527
Palani, G.S., 267 Simon, Keerthy M., 147
Panbarasu, K., 225 Singh, P.K., 419, 611
Panigrahi, S.K., 635 Sivaranjani, T., 39
Parameswaran, P., 353, 649 Soloviev, A.N., 291
Parinov, I.A., 291, 663 Sravan, B.V., 559
Patil, Prataprao, 315 Srinivasan, N., 103
Pavan Kumar, D.V.T.G., 39 Srivatsa, Shivashankar R., 183
Prakash, R.V., 225 Sujata, M., 65
Author Index 697
Suresh, J. A., 623 Venkatesh, S., 93

Suresh Kumar, R., 407 Verma, R.K., 493
Vesmawala, Gaurang R., 137, 215
T Vignesh, S., 303
Thomas Paul, V., 649 Vinayaka, H.L., 183
Tsvetkov, R., 129, 451 Vishnuvardhan, S., 419
U W
Umesha, P. K., 267, 437 Wasim, F.G.S., 481
V Y
Vaghela, Ajaysinh R., 137 Yanase, Keiji, 541, 599
Varughese, Byji, 93 Yepin, V., 129, 451
Velusamy, K., 325, 407
Subject Index
A Cohesive energy and second order elastic

Actuator assembly, 96, 100 constants, 253, 255, 256, 258, 259, 262,
Additive manufacturing, 625 263
Aluminium alloy, 303–305 Cohesive zone law, 235, 237
Aneurysm, 169–171, 176–178, 180 Collagen degradation, 169–171, 173–180
Aorta, 169–171, 173, 175–179 Compliance, 195, 197, 199
Armstrong method, 635, 641 Composite, 94–97, 100, 195–197, 199, 201,
ASME, 54, 55, 61, 63 528
AUSC, 354 Composite laminate, 281, 518
Composite structure, 183, 184, 193
B Compression member, 267, 268, 270, 274,
Ballistic impact, 183–185, 191, 193 275, 277
Band structure, 253, 255, 256, 261 Computer Numerical Control (CNC), 635, 636,
Beach marks, 120 638, 639, 641–646
Bearings, 117, 118, 122, 124 Concrete, 137, 138, 140, 141, 144, 147, 148,
Biaxial columns, 215, 216 151–155
Bi-metallic pipe weld joints, 421–426, Control object surface, 664, 665, 669, 670
429–433 Control plug, 325, 326, 330, 332, 334, 335,
Bimorph, 291–293 337, 339
Buckling, 279–281, 284–288, 528–531, Corrosion, 267–270, 272–275, 277, 438,
533–536 443–445, 448
Bulkhead, 106, 107, 113, 114 CPFE, 595, 596
Buttering, 420, 421, 429, 432, 433 Crack, 15–18, 21–26, 104, 107, 109–111, 113
Crack growth, 157, 159, 160, 163–165, 560,
C 561, 566, 573, 574, 576
Cantilever, 291–293, 295, 297 Cracking, 216, 222
Carbon Fiber Reinforced Plastic (CFRP), 529, Crash cooling, 326
530, 532, 533, 536, 673–675, 679, 683 Crashworthiness, 304
Carbon nanotubes, 138 Creep, 369, 370, 373–381
Catastrophic failure, 399, 405 Creep cracking, 1, 3–11
CBR, 326, 335 Creep curve prediction, 348
CFRP composite laminates, 78, 79, 89 Creep fatigue damage assessment, 370,
CFRP composites, 225, 232 373–380
Characterization, 158 Creep-fatigue interaction, 341, 342
Circumferential joint, 15–18, 20, 21, 23, 25 Creep strength degradation, 343
CMC, 470, 471, 477 Critical microcrack, 147–149, 151–155

https://doi.org/10.1007/978-981-10-7197-3
700 Subject Index
Crushable foam, 304 Finite element analysis, 107, 355, 356

Cyclic loading, 594 Finite Element Method (FEM), 292
Cyclic softening, 342, 343, 347, 348 Four point bending, 423
Cylindrical shells, 685–693 Fractal dimension technique, 518, 520, 526
Fractography, 226
D Fracture, 104, 105, 107, 114
Damage study, 674 Fracture energy, 157–159, 164
Damage tolerance, 104, 105 Fracture mechanics, 138, 453
DBA, 54–56, 58 Frequency, 297–300
Deformation induced martensite, 613, 618, FRP, 504, 507, 511, 513, 514
619, 621 Fuel cell, 482, 483, 490
Deformation monitoring, 130, 131, 135 Full-scale testing, 452
Delamination, 195–198, 200–202, 279–288, Fuselage, 15–17
518–522, 524–526
Design, 53–55, 57, 60–63 G
Design in-place analysis, 386, 387, 391 Gas turbine blades, 65, 66, 74
DFT, 253–258, 262 Gas turbine engine, 118
Diffusion bonding, 650, 652, 655 Glass fiber, 27, 29
Digital Image Correlation (DIC), 235, 239, Goodman and Gerber mean stress corrections,
487, 528, 531, 533–536 48–51
Dimples, 121, 122 Graphene, 253, 255, 256, 258–264
Double ended cracks, 463, 467 Ground surface subsidence, 130
Double sided stepped lap joint, 235, 245–247, Guided wave modes, 685, 686, 694
249
Dugdale model, 599–601, 606, 607, 609 H
Hanger and support, 369, 371, 372, 374, 379
E Hardness measurement, 355, 363
Electrical resistance strain gauge, 438, 442, Harmonics, 295
443, 445, 448 Heat-affected zone, 421, 429, 432, 433
Electro Chemical Machining (ECM), 636, 638, High Cycle Fatigue (HCF), 408, 417, 600, 606,
639 609, 636, 646
Electrolyte, 438, 443, 445 High energy piping, 369–371
Energy absorption, 304, 310–312 High-R, 1, 3, 5–11
Experimental stand, 452 High temperature, 1, 3, 4, 6, 7, 9, 11, 470, 472,
473, 477
F Hybrid composite, 579–590
Failure analysis, 216 Hydrostatic level, 130, 133, 134
Fast breeder reactor, 408, 415
Fasteners, 560 I
Fatigue, 65–68, 71, 73, 76, 117, 118, 121, 124, IC debonding, 504, 506
125, 560–562, 564, 565, 570, 573, 574, Infrared thermography, 504
576, 624, 626, 627, 630–632 Inner vessel, 325, 326, 330, 335, 337–339
Fatigue crack growth, 552, 555, 557 In situ tensile testing, 488
Fatigue crack growth rate, 613–615, 618, 619, Interaction of fatigue loading on vibration
621 characteristics, 579
Fatigue life, 40, 48–51, 77–79, 84, 85, 87–90 Interfacial transition zone, 147
Fatigue limit, 543–547 Interference pattern, 664, 665, 667, 670
Fatigue loading, 157, 159, 163 Interlaminar fracture toughness, 226
FCC single crystal, 596 Intermediate Heat Exchanger (IHX), 325, 326,
FE analysis, 100 330, 332, 337–339
Feature level test, 94–97, 100 Irradiation damage, 396, 398
FEM, 183, 185, 186, 191, 315, 318–321
Fiber bragg grating sensor, 439 J
Fiber reinforced polymer, 215 Jacket structure, 386, 387, 389–394
Subject Index 701
K Polymer electrolytic membrane, 482, 483, 490

Kinetic theory of fracture, 77, 78, 80, 83, 89 Power, 291, 292, 296, 297, 300
DKth, 1–11 Prediction models, 552
Prognosis, 560
L Proof mass, 291–297, 299, 300
Laser Doppler Vibrometer (LDV), 518, 524 Pzs, 599, 600, 602, 604, 605, 607
Laser interferometer, 671
Leak Before Break (LBB), 403, 404 Q
Level fluctuation, 410, 411, 415 Quantum Espresso, 253, 256–258, 260, 261,
Life prediction, 353, 359, 362, 366 263
Loading frequency, 613–615, 617–619, 621
Load sequence, 553, 555 R
Local stress calculation, 550, 551 RC beams, 504
Longitudinal joint, 15–17, 19, 21, 25 Reinforced concrete, 451, 452, 458, 459
Lug joints, 560–562, 568, 576 Remaining life, 157, 158, 160, 163–165
Residual strength, 15–17, 22, 23, 25, 26, 109
M Residual stress, 635, 637, 639, 640, 645
Main landing gear, 41–44, 44, 48–51 Resin infusion, 225–229, 232
Matrix cracking, 77, 78 Retrofitting, 215
Maximum absolute principal stress, 50, 51 Rigid line inclusion, 315–317, 319–322
Measurement of small displacements, 664, 665 RSR, 385, 389
Mechanical properties, 138 Rupture, 169–171, 173, 175–178, 180
Mechanical test, 625
Metals, 470, 477 S
Microstructural integrity, 650 Scratch, 541, 542, 548
Mode I SIF, 462, 465–467 Shake table, 493, 498, 500
Monotonic fracture, 426, 429 SHM, 560
Multiaxial creep, 363, 366 SHPB, 304, 306–308
Multicontinuum theory, 77, 78, 80, 83, 89 Silica nanoparticle, 27–30, 34, 37
Sine-sweep test, 493, 498
N Single Edge Notched Clamped Beam
Non-destructive Testing and Evaluation (SENCB), 205–207, 209, 213
(NDT&E), 673 Singularity, 315, 317, 319, 320
Nonlinear Finite Element Analysis, 183, 193 Site-specific ground response spectrum, 498
Notch effect, 542, 545, 546, 548, 600, 601 Slow crack growth, 206, 213
Snubber, 493, 494
O Spark Plasma Sintering (SPS), 650, 658
OGDHR, 325, 326, 329, 338 Spectrum fatigue, 27, 28, 31–33, 35, 37
Onset-of-growth, 195–197, 199–201 Spectrum load, 196–201
Orthotropic laminate, 462, 463, 467 Spring steel, 541, 548
Overheating, 65, 75 Stacking sequence, 579, 584, 586, 590
Oxide induced crack closure, 1, 3, 5, 6, 9, 11 Strain energy release rate, 235, 237, 238,
241–243, 249
P Strain gage, 462, 464, 466, 467
Part build orientation, 625 Strain intensity factor, 316–319, 321, 322
Passive thermography, 673, 674, 683 Strength, 53, 54, 56–58, 60, 62, 63
Piezoelectric generator (PEG), 291–300 Strengthening, 504, 506
Piping code, 369, 371, 378, 379 Stress analysis, 369–371, 373–376, 378
Pitting corrosion, 267, 268, 273, 274, 277 Stress intensity factor, 15–17, 21–24, 205–209,
Plastic collapse, 396, 400, 405 211–214, 599
PMC, 470–472, 477 Stress linearization, 55–57, 59
PMMA, 205, 206, 208, 209, 212–214 Stress relaxation, 341–350
Poly (2,5-Benzimidazole), 481 Stress rupture, 65, 74, 75
Polymer composite, 28, 33 Stress spectra, 46, 48–51
702 Subject Index
Striations, 121 U
Structural health monitoring (SHM), 685–687, Ultimate strength analysis, 387–389
694 Ultra High Strength Concrete (UHSC),
Structural integrity, 16, 415, 417 157–159, 159, 160, 164
Superalloy, 67, 68, 73, 74 Uniform corrosion, 267, 268, 272, 273
T V
Tearing instability, 396, 400, 402, 405 Vibration characteristics, 579–584, 586–588,
Tear strap, 107, 109–113 590
Tension softening, 157, 158, 160, 164, 165 Visco-elastic damper, 493–495, 496–500
Test design, 94 V(K) curve, 206
Thermal movement, 369, 372, 379 Voltage, 291, 296–300
Through-wall notch, 423, 424
Time reversibility, 685 W
Tomogram, 691 Wind turbine blade, 280, 281, 288
T-stress, 595–597
Turbine rotor blade, 636, 638–642, 646

2018 Book AdvancesInStructuralIntegrity

Uploaded by

Copyright:

Available Formats

2018 Book AdvancesInStructuralIntegrity

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2018 Book AdvancesInStructuralIntegrity

Uploaded by

Copyright:

Available Formats

Raghu Prakash

ISBN 978-981-10-7196-6 ISBN 978-981-10-7197-3 (eBook)

© Springer Nature Singapore Pte Ltd. 2018

Printed on acid-free paper

This Springer imprint is published by Springer Nature

I am delighted to note that the ﬁrst International Structural Integrity Conference

Bangalore, India Baldev Raj

The conference was sponsored by several agencies such as Department of

Chennai, India Raghu Prakash

High-R Fatigue Crack Growth Threshold Stress Intensity

A Study on the Effect of Crack Stopper for Enhanced Damage

Part II Civil Structures

Part III Composites

Studies on Fracture Features and Fracture Toughness of CFRP

Part IV Computational Mechanics

Stress Analysis for Integrity Assessment of High-Energy

Part VI Energy and Transportation

Part VII Experimental Techniques

Determination of Effectiveness of Viscoelastic Dampers in

Part VIII Fatigue

Effect of Loading Frequency on the Fatigue Crack Growth

Part IX Material Processing

Part X Structural Health and Condition Monitoring

Dr. Raghu Prakash is currently working as Professor

Indian Institute of Science, Bangalore. He is a member

Vikram Jayaram is Chair of the Division of

Dr. Ashok Saxena most recently (until January 3,

He received his MS and Ph.D. degrees from

Stuart R. Holdsworth and Zhen Chen

Abstract While knowledge relating to the determination and practical application

Keywords High-R High temperature DKth Oxide-induced crack closure

S. R. Holdsworth (&) Z. Chen

© Springer Nature Singapore Pte Ltd. 2018 1

1 Background and Introduction

High-cycle fatigue crack growth (HCFCG) behaviour may conveniently be con-

2 High-Temperature Fatigue Crack Growth Thresholds

As will be considered later, evidence of creep cracking in certain circumstances

3 Oxide-Induced Crack Closure

Table 1 Summary of oxide-induced crack closure data (550 °C)

Fig. 3 Variation of oxide

The degree of consistency is evident in Table 1 and Fig. 4. An alternative and

An additional mechanism interaction which may be encountered in

high-cycle fatigue cracks may go undetected by examination only of da/dN(DK)

Table 2 Summary of Steel k′ b C

5.1 da/dN(DKth) Criterion Sensitivity: Oxidation

Fig. 8 Influence of time

5.2 da/dN(DKth) Criterion Sensitivity: Creep Cracking

H. Ramesha, N. C. Mahendra Babu and H. V. Lakshminarayana

Abstract Fracture mechanics provides a methodology for prediction, prevention

H. Ramesha (&) H. V. Lakshminarayana

© Springer Nature Singapore Pte Ltd. 2018 15

acr Critical half crack length

Fuselage of commercial transport aircraft is a complicated structure designed to

Longitudinal and circumferential joints of fuselage structure are considered for

Analysis is carried out in following steps

2.1 Longitudinal Joint

A segment of fuselage structure shown in ﬁgure is considered for the analysis.

2.2 Circumferential Joint

Figure 2a shows the details of a circumferential joint conﬁguration in a fuselage

Fig. 2 a Circumferential joint conﬁguration; b Circumferential joint in a curved panel with a