Al-Maadhidi MASc S2022

Analysis of Retrofitted Concrete Columns using 3D
Elastic-Plastic Modeling
Zaineb Al-Maadhidi
A Thesis
in
The Department
of
Building, Civil and Environmental Engineering
Presented in Partial Fulfillment of the Requirements
for the Degree of
Master of Applied Science (Civil Engineering) at
Concordia University
Montréal, Québec, Canada
October 2021
© Zaineb Al-Maadhidi, 2021

C ONCORDIA U NIVERSITY
School of Graduate Studies
This is to certify that the thesis prepared
By: Zaineb Al-Maadhidi
Entitled: Analysis of Retrofitted Concrete Columns using 3D Elastic-Plastic
Modeling
and submitted in partial fulfillment of the requirements for the degree of
Master of Applied Science (Civil Engineering)
complies with the regulations of this University and meets the accepted standards with respect to
originality and quality.
Signed by the Final Examining Committee:
Chair
Prof. Khaled E. Galal
Examiner
Prof. Khaled E. Galal
Examiner
Prof. Ashutosh Bagchi
Supervisor
Dr. Emre Erkmen
Approved by
Ashutosh Bagchi, Chair of
Department of Building, Civil and Environmental Engineering
October 2021
Mourad Debbabi, Dean of
Faculty of Engineering and Computer Science
Abstract
Analysis of Retrofitted Concrete Columns using 3D Elastic-Plastic Modeling
Zaineb Al-Maadhidi
Fiber Reinforced Polymer (FRP) jacketing is frequently used to enhance the performance of
concrete columns. The strength and ductility of a concrete column increase after FRP jacketing
mainly because of concrete’s improved behaviour under confinement. To achieve a reliable retrofit
design, engineers often need to understand how much the performance is improved after FRP jack-
eting. For that purpose, numerical modelling is often necessary. Phenomenological elastic-plastic
models are widely used for the numerical modelling of concrete because of their capability of rep-
resenting 3D concrete behaviour considering permanent inelastic deformations. This research aims
to implement an elastic-plastic model to simulate FRP-jacketed concrete columns. The 3D material
model is validated against existing experimental data and comparisons with the results of mod-
els developed in ABAQUS software. It is shown that the proposed modelling approach is capable
of providing an accurate behaviour of square concrete columns confined with reinforcements and
FRP jackets under compression. After the validation of the model, a parametric study is conducted
to illustrate the consequence of partial wrapping on the behaviour of retrofitted square concrete
columns and to test the effect of FRP and concrete properties on the confined concrete strain. The
obtained results from the parametric study are then used to acquire a suitable confined concrete
strain equation/formula using a nonlinear regression technique. The core of this technique is based
on an optimization method that finds the optimal coefficients for the proposed equation, helping in
choosing the equation with the best fitting performance to the confined strain that is gained from the
3D material model. This proposed equation is recommended to improve the Canadian Standards by
describing the impact of material properties on the confined concrete strain.
iii
Acknowledgments
I would like to express my appreciation to my supervisor Dr. Emre Erkmen who was present to
provide assistance with his knowledge, effort, time, and guidance. He was always there to support
me during the research program.
I would like to acknowledge and express my sincere gratitude to my parents for their uncondi-
tional and endless support, inspiration, and encouragement with my studies and work. Even though
we have been far away from each other, their prayers and wishes were with me every second of this
journey. I doubt I would never do this without them. Special thanks to my brothers and cousins for
their endless motivation and help.
I would like to thank the Canadian Standards Association (CSA) for their appreciated scholar-
ship. Thank you for choosing me with the other students from all over Canada and allowing me to
contribute to improving the Canadian Standards.
Last but not least, many thanks to my friends, colleagues and everyone who was there to en-
courage me with their kind words, much appreciated!
iv
Thesis Published Work
Zaineb Khalid and Emre Erkmen. (2021, May 26-29). “Analysis of Retrofitted Concrete
Columns using 3D Elastic-Plastic-Damage Modelling, ”Canadian Society for Civil Engineering
Annual Conference. 2021. https://csce2021.ca/wp-content/uploads/2021/05/
CSCE-Updated-Program 210529.pdf.
v
Contents
List of Figures ix
List of Tables xiii
1 Introduction 1
1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Literature Review 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Models for FRP Confined Concrete Columns . . . . . . . . . . . . . . . . . . . . 6
2.2.1 Compressive Strength Models . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.2 Stress-Strain Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 Finite Element Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Literature Review Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Theories of Plasticity 20
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Elasto-Plastic Material Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
vi
3.2.1 Stress-Strain Relationship . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2.2 Frictional Block Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.3 The Tangent and Plastic Modulus . . . . . . . . . . . . . . . . . . . . . . 24
3.2.4 The Yield Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.5 Plastic Potential and Plastic Flow Rule . . . . . . . . . . . . . . . . . . . 28
3.2.6 Consistency Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.7 Principle of Plastic Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.8 Strain Hardening and Perfect Plasticity . . . . . . . . . . . . . . . . . . . 30
3.3 Plasticity Model Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.1 Yield Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.2 Hardening and Softening Functions . . . . . . . . . . . . . . . . . . . . . 33
3.3.3 Potential Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4 3D Material Model (FORTRAN Code) 36
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.2 Finite Element Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 Boundary Condition, Loading, and Meshing . . . . . . . . . . . . . . . . . . . . . 38
4.4 Description of the Material Models and Parameters . . . . . . . . . . . . . . . . . 39
4.4.1 Concrete Compression Model Implemented in FORTRAN . . . . . . . . . 39
4.4.2 Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.4.3 FRP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5 Finite Element Model (ABAQUS) 45
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Finite Element Type and Meshing . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.3 Boundary Condition and Loading . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.4 Description of Material Model and Properties . . . . . . . . . . . . . . . . . . . . 47
5.4.1 Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4.2 Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4.3 FRP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
vii
6 Numerical Results 56
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.2 Validation of the Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.2.1 Reinforced Concrete Column . . . . . . . . . . . . . . . . . . . . . . . . 56
6.2.2 Plain Concrete Columns Confined with FRP . . . . . . . . . . . . . . . . 57
6.2.3 Reinforced Concrete Columns Confined with FRP . . . . . . . . . . . . . 58
6.3 Results of the 3D Material Model . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.4 Parametric Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
7 Implementation of Optimization Method to Obtain the Confined Strain Equation 64
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
7.2 Gaps and Limitations in the Canadian Standards . . . . . . . . . . . . . . . . . . . 65
7.3 Discussing the Gaps and Limitations in the Canadian Standards . . . . . . . . . . . 66
7.4 The Proposed Nonlinear Regression Algorithm to Obtain the Maximum Confined
Concrete Strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.4.1 Performance Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.4.2 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.5 Summary Outcome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
8 Conclusion and Future Work 82
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8.2 Summary Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
8.3 Recommendations for Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Bibliography 84
viii
List of Figures
Figure 1.1 Damaged column due to an earthquake . . . . . . . . . . . . . . . . . . . 2
Figure 2.1 External steel jacket . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Figure 2.2 External FRP jack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Figure 2.3 Confinement pressure distribution for noncircular section . . . . . . . . . . 7
Figure 2.4 stress-strain curves for confined concrete. . . . . . . . . . . . . . . . . . . 8
Figure 2.5 Modified distribution of confinement stress . . . . . . . . . . . . . . . . . . 10
Figure 2.6 Loading configuration considered in the validation. . . . . . . . . . . . . . 11
Figure 2.7 Test setup of the specimens including a) ties at spacing 95mm b) ties at
spacing 200mm c) specimen with FRP layers . . . . . . . . . . . . . . . . . . . . . 18
Figure 3.1 Stress-strain curve for annealed material. . . . . . . . . . . . . . . . . . . . 22
Figure 3.2 The behavior of a steel bar at different stress stages. . . . . . . . . . . . . . 23
Figure 3.3 Types of stress-strain models . . . . . . . . . . . . . . . . . . . . . . . . . 23
Figure 3.4 Frictional block model for a linear elastic-plastic material. . . . . . . . . . . 24
Figure 3.5 Plastic strain and tangent modulus. . . . . . . . . . . . . . . . . . . . . . . 24
Figure 3.6 The von Mises and Tresca yield surfaces in biaxial stress space. . . . . . . . 27
Figure 3.7 The von Mises and Tresca cylinders in principal stress space. . . . . . . . . 27
Figure 3.8 The direction of the plastic strain. . . . . . . . . . . . . . . . . . . . . . . . 28
Figure 3.9 The model of associative plasticity that obeys the principle of plastic work. . 29
Figure 3.10 Perfectly plastic materials. . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Figure 3.11 Isotropic hardening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Figure 3.12 Kinematic hardening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
ix
Figure 3.13 The split of hardening function into the hardening and softening part. . . . . 34
Figure 4.1 Column Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Figure 4.2 Finite element compatibility. . . . . . . . . . . . . . . . . . . . . . . . . . 37
Figure 4.3 Boundary conditions and loading of the concrete column. . . . . . . . . . . 38
Figure 4.4 Laminates across thickness. . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Figure 5.1 Finite element types used in ABAQUS. . . . . . . . . . . . . . . . . . . . . 46
Figure 5.2 Meshed column in ABAQUS. . . . . . . . . . . . . . . . . . . . . . . . . . 46
Figure 5.3 Boundary condition and loading in ABAQUS. . . . . . . . . . . . . . . . . 47
Figure 5.4 Lam and Teng’s stress-strain model for FRP-confined concrete. . . . . . . . 48
Figure 5.5 Reinforcement configuration in ABAQUS for BS2. . . . . . . . . . . . . . 51
Figure 5.6 Lamina with unidirectional fibres. . . . . . . . . . . . . . . . . . . . . . . 52
Figure 5.7 Material properties required for FRP in ABAQUS. . . . . . . . . . . . . . . 53
Figure 6.1 Stress-strain curve for reinforced concrete column (BS2). . . . . . . . . . . 57
Figure 6.2 Stress-strain curve for plain concrete column wrapped with one FRP layer
(BC1). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Figure 6.3 Stress-strain curve for plain concrete column wrapped with five FRP layers
(BC5). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Figure 6.4 Stress-strain curve for RC column wrapped with one layer of FRP (BS1C1). 58
Figure 6.5 Stress-strain curve for RC column wrapped with three layers of FRP (BS1C3). 59
Figure 6.6 Stress-strain curve for RC column wrapped with five layers of FRP (BS1C5). 59
Figure 6.7 Stress-strain curve for RC column wrapped with three layers of FRP (BS2C3). 59
Figure 6.8 Stress-strain curve for RC column wrapped with five layers of FRP (BS2C5). 60
Figure 6.9 Stress-strain curves for retrofitted concrete columns using the 3D material
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Figure 6.10 Stress-strain curve for BS2C5 with different mesh sizes using 3D material
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Figure 6.11 Comparing the behavior of partially wrapped plain concrete column (BC1)
with different spacing with fully wrapped and not wrapped cases using 3D material
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
x
Figure 6.12 Comparing the behavior of partially wrapped reinforced concrete column
(BS2C5) with fully wrapped and not wrapped cases using 3D material model. . . . 62
Figure 6.13 Different vertical spacing of FRP strips on the concrete column. . . . . . . . 63
Figure 7.1 Maximum confined concrete strain vs number of layers of FRP from 3D
material model and the Canadian Standards. . . . . . . . . . . . . . . . . . . . . . 66
Figure 7.2 Stress-strain curves for plain concrete column wrapped with different num-
ber of FRP layers using the 3D material model. . . . . . . . . . . . . . . . . . . . 67
Figure 7.3 Stress-strain curves for reinforced concrete column wrapped with different
number of FRP layers using the 3D material model. . . . . . . . . . . . . . . . . . 67
Figure 7.4 Changing some of the material parameters simultaneously in the structural
analysis program. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Figure 7.5 Maximum strain obtained from 3D material model vs maximum strain ob-
tained from Equation 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
xi
tained from Equation 10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Figure 7.18 The Mean Square Error for all proposed equations. . . . . . . . . . . . . . 80
xii
List of Tables
Table 2.1 Mechanical properties of the materials. . . . . . . . . . . . . . . . . . . . . 17
Table 2.2 Experimental program. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Table 5.1 Mechanical properties of the carbon fiber. . . . . . . . . . . . . . . . . . . . 55
Table 5.2 Mechanical properties of the resin epoxy matrix. . . . . . . . . . . . . . . . 55
Table 5.3 The mechanical properties of the lamina composites as used in ABAQUS. . . 55
Table 7.1 The material parameters and the corresponding confined concrete strain and
strength. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Table 7.2 The results obtained from nonlinear regression algorithm with respect to each
possible equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
xiii
Chapter 1
Introduction
1.1 Overview
The performance of the concrete columns is the primary factor affecting the structural response
under extreme loading such as an earthquake. Upgrading the columns during their lifetime may
be required due to damage or deterioration. Inadequate column behavior can happen due to in-
appropriate design or construction, and due to environmental factors such as heat expansion or
earthquake events as shown in Figure 1.1. To prevent or avoid further failures, the columns may
need retrofitting.
In recent times, using Fiber Reinforced Polymer (FRP) composites as wraps or jackets for con-
crete columns has become a very popular technique to improve concrete column’s performance. It
has been found that FRP jacketing increases the strength and ductility of the columns Rousakis &
Karabinis (2012) Parvin & Brighton (2014). In addition, it is a simple installation so it can be used
to retrofit the concrete columns which are damaged after construction.
In order to achieve a reliable design, engineers often need to understand the failure mechanisms
of the structure. Describing the behavior of confined concrete under multiaxial compressive loading
is one of the most challenging scopes in structural engineering. The elastic-plastic constitutive laws
are not enough to model the load resistance and deformation capacity of confined concrete under
multiaxial loads. Accordingly, more advanced sophisticated constitutive models are required.
1
Figure 1.1. Damaged column due to an earthquake Ilki (2006).
1.2 Problem Statement
Considering the collapse mechanisms and understanding the residual load capacity of the struc-
ture before and after possible retrofitting are required to be able to decide whether to continue with
the operations, retrofit or demolish the structure.
Numerical modeling is often necessary to predict the behavior of the columns before and after
applying the confinement techniques. Consequently, the design parameters can be improved and
premature failure can be avoided. Besides, it is very important to use appropriate design standards
for concrete columns reinforced with FRP composites for both fully and partially wrapped columns.
1.3 Objective
The global objective of the research work is to efficiently and accurately model concrete columns
after retrofitting. To achieve this objective the following tasks are followed:
(1) Adapt a 3D material model to capture the behavior of FRP confined reinforced concrete
columns based on the coupled Elasto-Plastic model.
(2) Predict the load capacity of FRP reinforced concrete columns.
2
(3) Develop a finite element model using ABAQUS software for validation purposes.
(4) Verify the accuracy of the numerical strategy by comparing the outcomes with experimental
results from previous work and with ABAQUS results.
(5) Recommend an optimization-based nonlinear regression to fill the gaps and limitations in the
Canadian Standards (CSA S806-12) that are related to reinforced concrete columns confine-
ment.
1.4 Methodology
In order to fulfil the objectives of this research, computational technology is developed to pre-
dict the behavior of both full and partial FRP wrapped concrete columns. A 3D material model
is adopted based on the plasticity model to simulate the behavior of concrete beyond elasticity
Sarikaya & Erkmen (2019). 3D modeling approaches are used to capture 3D stress distributions
and their effect on material behavior. Also, the 3D material model has the ability to capture the
confinement effect on the concrete structures.
The numerical model is developed by using the FORTRAN programming language. It is able
to capture many possible modes of failure such as concrete crushing. It can easily be extended to
capture reinforcement buckling and FRP tearing, however, the case studies were created such that
these modes do not occur and therefore not considered in the analysis of this research.
The accuracy and efficiency of the numerical model are verified by comparing the outcomes
of the model with experimental results from previous work. Moreover, a finite element model is
developed in ABAQUS software for the confined concrete columns. The ABAQUS has the ability
to analyze reinforced concrete columns that are wrapped with FRP. Accordingly, this software is a
sufficient way to validate the results from the numerical strategy.
Filling in the gaps of limitations in the Canadian Standards by developing a nonlinear regression
algorithm based on an optimization method to obtain the best suitable equation to describe the effect
of FRP and concrete properties on the value of the maximum confined concrete strain. The non-
linear regression depends on the data acquired by the parametric study using the structural analysis
program that is linked with the 3D material model in the FORTRAN program.
3
1.5 Outline of the Thesis
This thesis is organized into eight chapters including this chapter.
• Chapter 2 reviews the confinement techniques and explains briefly some of the existing for-
mulation and models for FRP confined columns. As well as a brief description of previous
publications and works related to this field such as finite element modeling. The details of the
case study are also provided at the end of this chapter.
• Chapter 3 describes the theories of plasticity including the behavior of elasto-plastic ma-
terials and the component of plasticity models which are the yield surface, hardening and
softening functions, and potential surface.
• Chapter 4 describes the 3D material model including the concrete compression model which
is implemented in FORTRAN codes.
• Chapter 5 describes the finite element model in ABAQUS software in detail involving the
concrete, reinforcement, and FRP.
• Chapter 6 presents the main findings and results from the numerical methodology (FOR-
TRAN code and ABAQUS). In this chapter, the 3D material model is validated against
ABAQUS and experimental outcomes from previous studies. A parametric study is also con-
ducted in terms of testing the behavior of partially wrapped square concrete columns under
compression.
• Chapter 7 summarizes the Canadian Standards gaps and limitations. In addition, it provides
a nonlinear regression algorithm that helps in defining the change in the maximum confined
strain after applying FRP layers by using the outcomes of the parametric study. The best equa-
tions which are recommended for the Canadian Standards to obtain the maximum confined
strain are presented at the end of this chapter.
• Chapter 8 summarizes the main findings and observations from this study. Also, it provides
recommendations for future research work.
4
Chapter 2
Literature Review
2.1 Introduction
Confining concrete is a common practice in structural retrofitting applications, especially for
bridge piers and columns. Those structures are strengthening due to unsuitable design or errors
in the construction phase, and due to the modification of the structure use and the corresponding
design loads. It has been shown very early that lateral confining stress can lead to a significant
improvement in the strength of the concrete Richart et al. (1929). There are many ways to provide
confinement to the concrete such as steel jackets, hoops, and FRP wrapping Rousakis & Karabinis
(2012) Zhang et al. (2020) as shown in Figure 2.1 and Figure 2.2.
There is a difference between the lateral confinement provided by the FRP jacket and the steel.
The lateral confinement pressure added by steel is constant following the yielding of steel. While
the confinement pressure provided by the FRP jacket increases with the lateral strain of concrete as
FRP does not yield. Furthermore, the FRP jackets apply a linear confining pressure which increases
gradually until concrete ruptures as a result of its lateral dilation, whereas the reinforced steel exerts
a constant confining pressure after yielding.
Recently, applying Fiber Reinforced Polymer (FRP) composites as wraps or jackets for concrete
columns has become a well-known technique to enhance concrete column’s behavior. It has been
shown that FRP jacketing rises the strength and ductility of the concrete columns Ozbakkaloglu
& Saatcioglu (2007) Ilki et al. (2008) Rousakis & Karabinis (2012) Parvin & Brighton (2014).
5
Figure 2.1. External steel jacket Zhang et al. Figure 2.2. External FRP jack Rousakis &
(2020). Karabinis (2012) .
Moreover, it is a simple installation, lightweight, corrosion resistance, cost-effective, and a very
good technique for retrofitting columns that have been already constructed Bakis et al. (2002) Cheng
& Karbhari (2006)Basalo et al. (2012). Accordingly, FRP is superior in many aspects compared to
other types of jacketing such as steel jacketing and concrete jacketing.
There are two types of FRP; glass FRP (GFRP) and carbon FRP (CFRP). Both types were used
effectively in confining concrete columns as full jackets or partial confining strips Ghernouti &
Rabehi (2011). Recently, It has been proved that FRP partial wrapping technique is an economi-
cal alternative to FRP full wrapping and this is working very well in enhancing concrete columns
performance Zeng et al. (2018).
2.2 Models for FRP Confined Concrete Columns
The complexity of constitutive models for concrete varies from simple such as empirical and
uniaxial models to complex such as plasticity and plasticity-damage models. In both aspects, an
experimental calibration of the parameters is required but it is less in the case of the plasticity
models. Accordingly, uniaxial models are more experimentally dependent since they are scenario-
based models.
A large number of researchers have studied and simulated the behavior of the FRP confined
6
concrete columns. Most of the existing models are for circular sections since they are causing a
uniform confining pressure which can be estimated based on the mechanical properties of the FRP
and the diameter of the section. While the confining pressure around the perimeter of an FRP
rectangular and square columns is not uniform. Accordingly, it is more difficult to use a mechanical
solution to formulate the pressure distribution.
Several of the existing models for rectangular sections are quite similar to the circular section
except for the shape factor introduced for non-uniform confinement and the definition of the equiva-
lent diameter of the rectangular section. Figure 2.3 illustrates the confinement behavior at the corner
of the rectangular section.
Figure 2.3. Confinement pressure distribution for noncircular section Pham & Hadi (2014).
Although many kinds of research have been carried out to study the behavior of plain concrete
columns confined with FRP only, few models were proposed to account for the steel contribution
and its interaction with FRP in the confined columns. In recent years, the implementation of FRP
to rehabilitate and strengthen existing reinforced concrete (RC) columns has been widely used. In
addition to the efficiency of FRP in increasing axial strength, shear capacity, and energy absorption
capacity of reinforced columns, it protects the longitudinal steel bars from buckling. It has been
found that the axial strength of columns confined with steel bars and FRP jackets is much higher
than the axial strength of columns confined with FRP only J.-Y. Lee et al. (2004). Figure 2.4 shows
the stress-strain relationships of confined concrete.
7
Figure 2.4. stress-strain curves for confined concrete J.-Y. Lee et al. (2004).
2.2.1 Compressive Strength Models
As it has been proven before, the strength of a concrete column can be strongly enhanced after
applying FRP confinement. Accordingly, many types of research have been carried out to eval-
uate the confinement effect on the column’s strength. Jianguo et al. (2008) developed a model
to predict the concrete strength of rectangular columns that are confined with Fiber Reinforced
Polymer (FRP). In addition, a design formula was suggested to calculate the ultimate load of the
FRP-wrapped rectangular columns based on their investigated results. The confined core concrete
strength fcc was assumed to be a function of the uniaxial compressive strength of unconfined con-
crete and the effective confinement pressure fr as given in the equation below:
fcc = fc + kfr (1)
where fcc is the unconfined concrete strength and k is the confinement effectiveness coefficient,
which is suggested to be a constant of 2.3 by fitting and regressing the test data from previous stud-
ies. Jianguo et al. (2008) also proposed a formula to calculate the ultimate load-carrying capacity
8
Pu of FRP strengthens columns as shown in the following expression:
Pu = Pcu + Psu (2)
where Pcu is the compressive strength that is carried by the concrete and Psu is the compressive
strength that is carried by the longitudinal reinforcing steel bars. A good correlation was obtained
between the experimental results and the results from the proposed model and the design formula.
Pham & Hadi (2014) proposed a model to predict the strength of FRP confined rectangular
concrete columns by focusing on the stress concentration at the corners of the section. Thus, the
following equations were suggested to formulate a linear relationship between the normalized com-
′ and the normalized confining pressure f :
pressive strength of confined concrete fcc l
′
fcc fl,e
′
= 0.68 + 3.91 ′ (3)
fco fco
fl,e = fl kc (4)
πr
kc = (5)
b + h − r(4 − π)
′ is the unconfined concrete strength, r is the corner radius,

where kc is the corner effect ratio, fco
b and h are the column sides, and fl,e is the effective confining pressure. They concluded that the
used model for stress prediction fits very well with the experimental results.
Although, the majority of the available models in the literature for rectangular and square con-
fined columns with FRP focused on the confinement stress at corners only, Moodi et al. (2018) con-
sidered both the corners and neighbourhood areas as shown in Figure 2.5.
Hence, Moodi et al. (2018) modified the model proposed by Lam & Teng (2003a) using the whole
algorithm and a widespread database to estimate the compressive strength of confined rectangular
and square concrete columns with FRP. The effective strain coefficient of FRP kε and section shape
coefficient ka is considered as an individual coefficient (kε )new . Moreover, the modified model
considered different increased strength values for samples with a compressive strength of less than
35 MPa or more than 35 MPa. The final proposed model after the optimization is summarized as
9
Figure 2.5. Modified distribution of confinement stress Moodi et al. (2018).
below:
′ ′ fl,a ′
fcc = fco (1 + 4.485(kε )new ′
) , fco ≤ 35M P a (6)
fco
′ ′ fl,a ′
fcc = fco (1 + 2.478(kε )new ′ ) , fco > 35M P a (7)
fco
πr + 0.0524b + 0.01h
(kε )new = (8)
b + h − r(4 − π)
′ is the unconfined concrete strength, f ′ is the confined concrete strength,f

where fco cc l,a is the con-
fining pressure, b and h are the column sides, and r is the corner radius.
Moodi et al. (2018) also used Response Surface Methodology (RSM) to estimate the compres-
sive strength of confined columns with FRP. It has been shown that RSM predicts compressive
strength more accurately when compared with other models. The correlation coefficient has in-
creased by about 34% and 26% for RSM and the modified model respectively.
Zignago et al. (2018) modified a material constitutive model to predict the load capacity of re-
inforced concrete confined columns. This model is capable to model the simultaneous confinement
effect of reinforcement and FRP wrapping of reinforced concrete columns subjected to different
loading conditions; axial concentric load, eccentric axial load, and a combination of axial and cyclic
lateral loads as shown in Figure 2.6
10
Figure 2.6. Loading configuration considered in the validation Zignago et al. (2018).
Zignago et al. (2018) also calculated the lateral confinement pressure fl′ as the summation of
confinement pressure from transverse reinforcement steel fl,steel and FRP jackets fl,F RP as shown
in the following equation:
1 1
fl′ = fl,steel + fl,F RP = ks ρs σs + kf ρf σf (9)
2 2
where ks is the coefficient of steel confinement effectiveness, ρs is the transverse steel reinforcement
ratio, σs is the stress of the transverse reinforcement, and ρf is the stress of the FRP. The numerical
simulation of the model was compared with the experimental results which were collected from the
literature. It has been found that the new model provides an excellent estimation of the load-carrying
capacity.
2.2.2 Stress-Strain Models
In general, the stress-strain models for concrete are classified into two groups; which are design-
oriented models and analysis-oriented models. In the design-oriented models the compressive
strength, ultimate axial strain, and stress-strain behavior of FRP confined concrete are estimated
11
using closed-form equations based on experimental results. Whereas, the stress-strain curves of
FRP confined columns of the second group are created using the incremental numerical procedure.
The analysis-oriented models have the advantage of considering the interaction between the con-
fined concrete and the material that is used for the confinement. However, it is not used in the direct
design because of the complexity of the incremental process.
Lam & Teng (2003a) proposed a new design-oriented stress-strain model based on existing test
data and observations. They showed that the assumption that the rupture of FRP happens when the
stress in the jacket reaches the tensile strength of the FRP material is invalid for confined columns
with FRP. Accordingly, it was proved that the stress-strain model for confined columns with FRP
must be based on the actual hoop rupture strain of FRP instead of the ultimate material tensile strain.
Lam & Teng (2003a) also studied different existing design-oriented stress-strain models then
they improved the existing models by considering the effect of jacket stiffness. A new equation was
proposed to calculate the compressive strength based on the largest databased that collected from
the literature:
′
fcc fl,a
′
= 1 + 3.3 ′ (10)
fco fco
′ ,f ′ , and f
where fco cc l,a are the unconfined concrete strength, confined concrete strength, and con-
finement pressure provided by the FRP layer respectively. The ultimate strain equation was also
proposed taking into account the FRP efficiency factor:
εcu fl,a εf rp 0.45

= 1.75 + 5.53 ′ ( ) (11)
εco fco εco
where εco is the axial strain at the compressive strength of unconfined concrete, εcu is the axial
strain of the confined concrete, and εf rp is the hoop rupture strain. It has been concluded that the
new oriented design stress-strain model is simple and suitable for direct use in design.
Fattah (2018) reviewed a variety of existing models of concrete columns confined with lateral
steel and FRP that are available in the literature. Their performance was assessed based on ex-
perimental cases of square columns using stress-strain diagrams. Also, a statistical analysis was
12
conducted for theoretical peak strength and ultimate strain obtained from the existing models to im-
prove their performance. Fattah (2018) also proposed a new model that overcomes the shortcomings
′ , and the ultimate strain ε :
found in the reviewed models to calculate the peak strength fcc cu
fl′
fcc = fc′ (1 + kc kR1 ) (12)
fc′
fl′ −0.6 −0.2
kc = 2.66( ) w (13)
fc′
s ds Es
w= + (14)
ds dLs Esl
where fc′ is the unconfined concrete strength,fl′ is the effective confining pressure, kc is the strength
enhancement parameter, w is a parameter for the longitudinal steel to lateral steel ratio, ds is the lon-
gitudinal steel diameter, dLs is the lateral steel diameter, s is the tie spacing and kR1 is a parameter
of the corner radius to width ratio. The ultimate strain is defined as:
fl′ 0.9
εcu = εl + 0.75w−0.95 ( ) 1.1(5−ηf ) kR2 f or ηf ≥ 5 (15)
fc′
f′
εcu = εl + 0.75w−0.95 ( l′ )0.9 2(5−ηf ) kR2 f or ηf < 5 (16)
fc
where εl is the lateral concrete strain, kR2 is a parameter of the corner radius to width ratio, and ηf
is the number of FRP layers. It was concluded that the proposed model has better correlations to the
experimental curves than the existing models in the literature.
It is well known that in FRP- confined concrete the end restraints prevent the confined columns
from the lateral expansion and generate a non-uniform strain of the FRP jacket over the column
height. Accordingly, the peak axial stress and strain are underestimated. In order to eliminate the
effect of end restraints, Teng et al. (2015) recalibrated the stress-strain model and proposed the
following equation to calculate the lateral strain:
εc −εl 0.7 −εl σ1

= {[1 + 0.75( )] − exp[−7( )]} × (1 + 8 ′ ) (17)
εco εco εco fco
where εc and εl are the axial strain and the lateral strain of concrete respectivly, εco is the axial strain
′ is the unconfined concrete strength, and σ
of unconfined concrete at the compressive strength, fco 1
13
is the lateral confining pressure.
Due to the change in the confinement effectiveness coefficient, the axial strain of concrete at
the peak of axial stress of concrete under lateral confining pressure is calculated from the following
equation:
ε∗cc σ1
= 1 + 20 ′ (18)
εco fco
where ε∗cc is the axial strain at the peak axial stress of concrete under a lateral confining pressure.
′
Teng et al. (2015) also modified the expression of the compressive strength of confined concrete fco∗
as shown below:
′ ′
fcc∗ = fco + 4σ1 (19)
The results obtained from the finite element analysis were compared with experimental stress-strain
curves. It has been shown that there is a good agreement between the results, so the provided finite
element approach has the ability to predict the behavior of reinforced concrete and plain concrete
columns confined with FRP.
2.2.3 Finite Element Modeling
During the first decade of the 21st century, several attempts have been done to model concrete
columns that are confined with FRP using the finite element method. This method has the ability
to deal with nonlinear geometries and capture the interaction of different materials. On the other
hand, using finite element modeling of confined columns is a challenging task since the definition
of the concrete material model must be defined accurately to represent the volumetric behavior of
concrete subjected to triaxial stress states.
For finite element structural modeling purposes, several uniaxial stress-strain relationships have
been developed for confined concrete including the works of Saatcioglu & Razvi (1992) Attard
& Setunge (1996) Isleem et al. (2018) . Such models require a prior estimate of the confinement
14
pressure which however depends on the local stress distribution. On the other hand, elastic-plastic-
damage based constitutive laws have been applied extensively for the description of the progressive
failure of concrete.
3D elasto-plastic damage models have the capability of capturing the confined concrete behavior
naturally as a result of the stress distribution as the stress-strain relations are developed considering
the multi-axial states. Furthermore, the evolution of the internal parameters in phenomenological
plasticity and damage models can be related to underlying mechanisms within the material in a ther-
modynamically consistent manner, e.g., development of slip bands and dislocation systems causing
plastic deformations or void nucleation and crack development causing stiffness degradation. Initial
attempts to merge elastoplastic and damage constitutive models can be found in Lemaitre (1985)
Simo & Ju (1987) . Other coupled plasticity and damage models include the works of Klisiński &
Mroz (1988), Benallal et al. (1988), Lubliner et al. (1989), Yazdani & Schreyer (1990), Meschke et
al. (1998), Grassl & Jirásek (2006), and Voyiadjis et al. (2008).
An efficient coupled elastoplastic damage model that is capable of simulating the behavior of
plain concrete was also developed by J. Lee & Fenves (1998). Recently, Sarikaya & Erkmen (2019)
showed that the J. Lee & Fenves (1998) model can be captured as a special case within the frame-
work of Armero & Oller (2000) by imposing a kinematic condition between the strain components.
In both Sarikaya & Erkmen (2019) and J. Lee & Fenves (1998) a single failure surface, poten-
tial function and hardening/softening criterion can be adopted in order to characterize the inelastic
behavior of concrete, providing an efficient computational framework.
Some other attempts for elasto-plastic-damage modeling of concrete can be found in Grassl et
al. (2002); Grassl & Jirásek (2006); Papanikolaou & Kappos (2007). ABAQUS has included Elasto-
Plastic-Damage models developed by Lubliner et al. (1989) and Lam & Teng (2003a) as modeling
options.
Mirmiran et al. (2000) developed a nonlinear finite element model to analyse confined concrete
with FRP. This model was generated based on a non-associative Drucker-Prager plasticity model
which considers the pressure sensitivity of the materials. The concrete core was modelled as a solid
element while the FRP jacket was modelled as linear-elastic membrane shell elements. A parametric
program inside ANSYS software was developed for generating the mesh of different geometries and
15
material properties. It has been concluded that Drucker-Prager plasticity has the ability to predict
the axial stress-strain response of the FRP confined columns effectively.
Chakrabarti et al. (2008) proposed a nonlinear finite element model to analyse the behavior of
plain and reinforced concrete columns wrapped with FRP sheets. They used the ANSYS finite
element software to simulate the model for the analysis of square and circular concrete columns and
to predict the behavior of strengthened columns. The model was also used for a parametric study
to measure the effect of FRP thickness, stiffness, and orientation on the confinement and concrete
strength. The model was validated with other finite element and experimental results from previous
works.
Teng et al. (2015) presented a three-dimensional finite element analysis for modeling the behav-
ior of both; FRP -confined plain concrete and reinforced concrete circular columns. They obtained
a local stress-strain model for concrete under uniform confinement to generate the input parameters
for Yu et al. (2010) plastic damage model.
Yu et al. (2010) developed a modified damage plasticity model to simulate the behavior of con-
fined concrete under non-uniformed confinement. The modification of the model was within the
theoretical framework of the damage plasticity model that is available in ABAQUS software. Yu
et al. (2010) modified the damage parameter, strain hardening/softening and flow rule which are all
confinement dependent. They also defined an effective confining pressure which is a special char-
acteristic of non-uniformly confined concrete. One of the limitations of this model is the simulation
of concrete confinement of the non-circular columns since it depends on some assumption
Hany et al. (2016) presented a modified concrete damage plasticity model (CDPM) for both
normal and high strength FRP confined concrete. This modified model has the ability to predict the
monotonic axial stress-axial strain behavior of confined concrete columns with FRP and reinforce-
ment, as well as the lateral dilation of FRP confined concrete. After comparing the results of the
modified model (CDPM) with experimental results of FRP confined concrete samples that available
in the literature, it has been shown that there is a good agreement between the finite element results
of the proposed model and the tested data.
Farahmandpour et al. (2017) also developed a damage plasticity model for concrete to describe
the confinement sensitivity and dilation characteristics of concrete under a triaxial stress state. Most
16
of the existing constitutive models have many limitations, for example, the strain and stress can-
not be fully analysed. Therefore, finite element code is required to generate a triaxial constitutive
model of concrete. It has been concluded that the proposed model is valid through steel and FRP
confinement with a wide range of confinement stiffness based on experimental validation.
2.3 Case Studies
In order to perform numerical modeling to predict the behavior of concrete columns before and
after retrofitting, experimental data were obtained from the literature Rousakis & Karabinis (2012).
This experimental program consisted of forty-two prismatic concrete columns. Most of the columns
were strengthen by internal steel reinforcement and external FRP wrapping
All the tested columns have the same geometry of square cross-section of 200 mm, 320 mm
height, and 30 mm corner radius. The specimens that were selected in this research for validation
purposes were subjected to monotonic compression load until failure. The mechanical properties of
concrete, steel reinforcement and carbon FRP (CFRP) are shown in Table 2.1.
Table 2.1. Mechanical properties of the materials.
Compressive strength 25.5 MPa

Concrete
Elastic modulus 19.3 GPa
Nominal yield stress 500 MPa
Steel
Elastic modulus 200 GPa
Tensile modulus 240 GPa
CFRP
Thickness of each layer 0.117 mm
Strain at failure 0.155
The reinforced concrete columns contain transverse and longitudinal reinforcement. The longi-
tudinal reinforcement has a 14 mm diameter. Whereas, the transverse reinforcement has a diameter
of 8 mm. Two spacing distances were used for transverse ties; 200 mm and 95mm, which sym-
bolized S1 and S2, respectively. For the FRP wrapping, unidirectional carbon FRP sheets were
implemented to some specimens for providing an external strength. Each sheet has a 300mm width
17
and the sheets are cut into a suitable length based on the number of layers that are used with an
overlap of 200 mm only for the last external layer.
The application of the FRP composite required the preparation of the concrete surface. Accord-
ingly, a coat of two components of epoxy resin was applied carefully on the concrete surface to
ensure the proper underlay for the wrapping procedure. Two-component epoxy resin was also used
to glue the composite sheets together by hand.
The following Table 2.2 shows the details and notation of the specimens used for the numerical
validations and Figure 2.7 shows the experimental test setup of the specimens including the steel
configuration.
Table 2.2. Experimental program.
Specimen label Steel bars & ties CFRP sheet layers Jacket nominal thickness mm
B6 - - -
BS1 4Φ14 & 8Φ/200 - -
BS2 4Φ14 & 8Φ/95 - -
BC1 - 1 0.117
BC5 - 5 0.585
BS1C1 4Φ14 & 8Φ/200 1 0.117
BS1C3 4Φ14 & 8Φ/200 3 0.351
BS1C5 4Φ14 & 8Φ/200 5 0.351
BS2C1 4Φ14 & 8Φ/95 1 0.117
BS2C3 4Φ14 & 8Φ/95 3 0.351
BS2C5 4Φ14 & 8Φ/95 5 0.585
Figure 2.7. Test setup of the specimens including a) ties at spacing 95mm b) ties at spacing 200mm
c) specimen with FRP layers Rousakis & Karabinis (2008).
18
2.4 Literature Review Summary
This chapter presented background information from the literature review about confining con-
crete techniques and the models which are used to model confined concrete with FRP. After review-
ing some of the models that exist in the literature, two main points were noticed. The first point is
most of the existing models are for confined concrete for circular column cross-sections. Secondly,
few models were proposed to account for the steel contribution and its interaction with FRP in the
confined columns. It is important to focus more on models for confined concrete that model square
or rectangular cross-sections and capture the interaction of different materials such as FRP and steel
reinforcement
19
Chapter 3
Theories of Plasticity
3.1 Introduction
The theory of linear elasticity is useful to model the materials that undergo small deformation
that will return to their original state after removing the load. The energy consumed during the
elastic deformation is stored as elastic strain energy which is recovered completely due to load
removal.
Most of the materials suffer a permanent deformation when subjected to a load beyond its critical
value which is called the yield stress. This deformation happens due to the dislocation and migration
of the grain boundaries on the micro-level and it is irreversible. The distortion of an element is an
energy dissipation process.
The elastic analysis does not give any information about the behavior of the materials after they
reach the yield stress point. Therefore, the elastic model does not give the actual load that the
structure can carry before collapsing.
From that perspective, the theory of plasticity was introduced to describe the irreversible de-
formation that remains upon complete unloading. Moreover, plastic analysis calculates the actual
failure load of the structure. The failure load is critically greater than the elastic load capacity of the
structure.
This chapter will illustrate the fundamental elements of plasticity theory. These elements in-
clude general stress-strain relations, yield condition, flow rule, consistency condition, the principle
20
of maximum plastic work, and isotropic and kinematic hardening.
3.2 Elasto-Plastic Material Behavior
3.2.1 Stress-Strain Relationship
The behavior of loaded materials is described in terms of the stress and strain relationship.
There are two different ways to represent the applied stress. First calculating the stress-based on the
original cross-sectional area Ao of the specimen which is called nominal stress σn Beer & Johnston
(1992):
F
σn = (20)
Ao
while the other way is the true stress σ which is calculated based on the current cross-sectional area
A.
F
σ= (21)
A
In which the area and the force are both changing with time. In the elastic range, there is very small
elongation. Accordingly, the change in cross-sectional area is negligible so both definitions of stress
are more or less equivalent. The strain is defined as the ratio of the change in length of the specimen
to its original length Beer & Johnston (1992):
L − Lo
ε= (22)
Lo
where L is the current length and Lo is the original length which is continually changing.
The stress-strain diagram is obtained by plotting strain (ε) as abscissa and stress (σ) as an ordi-
nate. The stress-strain curve for annealed material is divided mainly into two parts; elastic range and
plastic range as illustrated in Figure 3.1. In the elastic region (from 0 to A), the material recovers
all strains after unloading. The curve is linear from 0 to A′ with slope E, which is known as Young
Modulus so the material obeys Hook’s law.
21
Figure 3.1. Stress-strain curve for annealed material.
Point A′ is called the proportional limit while point A is the yield point where a small amount
of permanent deformation is observed. As long as the material behaves elastically, the arrangement
of the particles remained the same. Beyond the yield point, the material is classified as either brittle
or ductile. For ductile materials, when yielding ends and load increases, the curve continues to rise
from A to B but with a slope less than curve 0A′ , this behavior is called strain hardening. If the
material is unloaded at point B it will return along path BC parallel to the original elastic line. This
is known as elastic recovery.
The strain which remains upon unloading is irreversible and called plastic deformation which is
due to the distortion of the particle arrange. Further increases in the load lead the curve to point D
which is the ultimate stress value. After point D, the cross-sectional area starts to decrease. This
behavior is called necking where the strength is dropped. Therefore, the material becomes weak
and failure is occurred due to the development of shear stress. The change in the shape of a steel
bar in different stress stages is shown in Figure 3.2.
There are many types of stress-strain models for elastic and plastic deformation as proposed in
Figure 3.3. When a material deforms but still can take a load, it is called Elasto-plastic material and
it follows the behavior of the first stress-strain curve (a) where both elastic and plastic curves are
assumed linear. In (b) the work hardening is negligible so the yield stress is constant after yielding
Chung & Lee (2018), such models are called perfectly plastic. The linear hardening behavior is
represented in (c) and the rigid-perfectly plastic behavior is shown in (d).
22
Figure 3.2. The behavior of a steel bar at different stress stages Kelly (2019).
(a) Linear (b) Elastic/ (c) Rigid/ linear (d) Rigid/

elastic-plastic. perfectly-plastic. hardening. perfectly-plastic.
Figure 3.3. Types of stress-strain models
3.2.2 Frictional Block Models
The visual behavior of plastic materials can be shown by frictional block models Simo & Hughes
(2006). The linear elastic-plastic model with linear strain hardening combines hardening spring with
stiffness H, in parallel with the friction block. After yielding, an ever increases in stress is needed to
be applied to keep the block moving and the elastic strain keeps occurring due to further elongation
of the spring.
After that, the stress splits into two parts, one is called the yield stress which is carried by the
moving block and the other one is known as the overstress which is carried by the hardening spring.
23
Figure 3.4 demonstrates the frictional block model for a linear elastic-plastic material with linear
strain hardening; (a) stress-free, (b) elastic strain, (c) elastic and plastic strain, and (d) unloading
behavior, where H is the plastic modulus.
Figure 3.4. Frictional block model for a linear elastic-plastic material.
3.2.3 The Tangent and Plastic Modulus
In the elastic region, the stress and strain are relating to each other by Hook’s law (σ = Eε)
where E is the Young modulus. In the plastic region, the slope of the stress-strain curve is K which
is known as tangent modulus that will change during plastic deformation as shown in Figure 3.5. At
Figure 3.5. Plastic strain and tangent modulus.
24
any point of strain, the increment of stress dσ is related to the increment of strain dε through Simo
& Hughes (2006) Borja (2013):
dσ = Kdε (23)
After yielding, the total strain increment splits into elastic εe strain and plastic strain εP :
dε = dεe + dεp (24)
The stress and plastic strain increment are relating to each other through the plastic modulus H as
shown below:
dσ = HdεP (25)
Finally, the tangent modulus is related to plastic modulus by the following equations:
1 1 1
= + (26)
K E H
σ σ−Y dσ EH
ε = εe + ε P = + →k= = (27)
E H dε E+H
Note: when:
• H = 0 Perfect plastic
• H > 0 Hardening
• H < 0 Softening
3.2.4 The Yield Criteria
A condition that defines the limit of elastic behavior and the beginning of plastic deformation
under any possible combination of stress is called yield criterion or yield condition. This law is
25
not only applied to the first loading state, but also to reloading an element that unloaded from the
previous loading state.
For the simple case of one-dimensional loading, the yield criterion is defined by the value of
the stress that will cause permanent deformation. In other words, the yield criterion is graphically
illustrated by a point. For two-dimensional loading, the yield will occur when the combination of
the applied stress touches a curve between two loading directions Borja (2013).
In the case of three-dimensional loading, the plastic deformation will happen when the applied
stress touches the yield surface. In general, the yield criterion is represented by the yield surface,
if the applied load stays inside the surface the material behaves elastically. Once the stress touches
the yield surface, plastic deformation will be produced. It has been found that the hydrostatic stress
does not affect yielding, therefore the yielding depends only on the magnitude and direction of the
deviatoric stress vector J2 Han & Reddy (2012).
Many criteria were applied to determine whether a material has yielded, Tresca and von Mises
were commonly used.
Tresca and von Mises Yield Criterion
This yield criterion is known as the theory of maximum shear strength. It is applicable for ductile
materials such as metal. Tresca criterion assumed that the material yields when the maximum shear
stress reaches a particular value. The yield surface was simulated as a hexagon as shown in Figure
3.6. The plastic strain increment vector is directed along the normal to the side of the hexagon when
the stress lies on one of the sides Krabbenhøft (2002). Therefore, plastic deformation is a pure shear
in the direction of the maximum shear stress. Mohr’s circle is used to solve for maximum shear
stress and the material will fail if:
σ1 − σ3 ≥ σ0 (28)
where σ1 and σ3 are the maximum and minimum normal stress, and σ0 is the failure of stress in
uniaxial loading.
Von Mises criterion (also known as the maximum distortion energy) is based on the Tresca
26
hypothesis but it takes into account that the hydrostatic stresses do not affect the material failure.
Thus, only the deviatoric part of the stress tensor will change the shape and use it to construct the
yield function for von Mises plasticity criterion Han & Reddy (2012).
The von Mises criterion is also referred to as J2 plasticity since it can be written in the coordinate-
independent format in terms of the second invariant J2 of the deviatoric part of the stress tensor. The
von Mises stress states that two stress states with equal distortion energy have an equal von Mises
stress Han & Reddy (2012). The yield surface has a circular cylinder shape as shown in Figure 3.7.
Figure 3.6. The von Mises and Tresca yield surfaces in biaxial stress space.
Figure 3.7. The von Mises and Tresca cylinders in principal stress space.
27
3.2.5 Plastic Potential and Plastic Flow Rule
The elastic strain increment is directly related to the stress increment by Hook’s law. Therefore,
it is important to relate plastic strain increment to the stress increment and current stress. The ratios
of components of the plastic strain rate can be determined from the following equation Ibrahimbe-
govic (2009) Borja (2013):
∂g
dεpij = dλ (29)
∂σij
where dλ is a positive scalar also it is called slip rate. This scalar is equal to zero in elastic unloading
and g is known as plastic potential which may be the same as yield surface.
If the plastic potential is the same as the yield surface, the plastic flow rule is called an associated
flow rule (f = g). Otherwise, it is called a non-associated flow Han & Reddy (2012).
The direction of plastic flow defines by the potential function which is usually the same as the
yield function. The plastic flow rule states that during loading which causes the plastic deformation,
all stresses should stay on the yield surface. As a result, the direction of the stress increments is
tangential to the yield surface Borja (2013).
Besides, the direction of the elastic strain is also tangential to the yield surface because it pro-
duces stress increments. Whereas the plastic strain increments do not produce stress so their direc-
tions are normal to the yield surface as shown in Figure 3.8.
Figure 3.8. The direction of the plastic strain.
28
3.2.6 Consistency Condition
In order to obtain a whole relationship between stress and strain, the consistency condition
should be assumed. For perfectly plastic materials (no hardening), consistency condition states that
during loading the stress remains on the yield surface Borja (2013).
For strain hardening solids, the consistency condition means that the stress remains on the new
yield surface (expanded, contracted, or translated). In other words, plastic loading is known as a
consistency condition where loading from a plastically deforming state leads to another plastically
deforming state Han & Reddy (2012) Borja (2013).
3.2.7 Principle of Plastic Work
Plastic work is also known as maximum plastic dissipation or the rate of plastic work. It states
that the plastic work that is done in a given plastic strain rate has the maximum value. In particular,
it is greater than the frictions work that is done by any state of stress not exceeding the yield limit.
The principle of maximum plastic dissipation is a vital component of the theory of plasticity.
The values of the internal variables for plasticity can be obtained from the principle of plastic work.
In the elastic range, the plastic dissipation is equal to zero. As a result, the values of the internal
variables are zero Han & Reddy (2012). There are two considerable consequences of the principle of
maximum plastic dissipation. First, the convexity of the yield surface (smooth surface). Secondly,
the normality law which states that the plastic strain rate is normal to the yield surface Figure 3.9.
Figure 3.9. The model of associative plasticity that obeys the principle of plastic work.
29
3.2.8 Strain Hardening and Perfect Plasticity
In a one-dimensional (uniaxial test) case, the specimen will deform after yielding and then
harden. Figure 3.10 illustrates the perfectly plastic materials where the material deforms at the yield
point with no hardening.
Figure 3.10. Perfectly plastic materials.
In the multiaxial case, strain hardening is when the yield surface changes in size, or location
which depends on the complete history of plastic deformation since the previous loading. With the
increase of stress beyond the yield point more plastic deformation will be produced and the material
becomes stronger and more difficult to deform. There are two main hardening rules; isotropic
hardening and kinematic (or anisotropic) hardening.
Isotropic Hardening
The rule of isotropic hardening assumed that the yield surface expands or contracts uniformly
about centre of the original yield surface without a change in the shape as shown in Figure 3.11
Borja (2013).
The yield surface will be described by the following equation:
f = f (σij ) − R(α) = 0 (30)
where R represents the size of the yield surface depending on the plastic strain through the hardening
internal variable α.
30
Figure 3.11. Isotropic hardening.
Kinematic Hardening
The rule of kinematic hardening assumed that the yield surface translates in the stress space
during plastic flow as demonstrated in Figure 3.12. However, the shape and size of the yield surface
will remain the same Ibrahimbegovic (2009) Borja (2013). It is important to note that the elastic
range doesn’t change and is simply translated. To capture the behavior of the kinematic harden-
ing, back stress notation will be introduced. The yield surface will be described by the following
Figure 3.12. Kinematic hardening.
31
equation:
f = f (σij − αij ) − R0 = 0 (31)
where αij represents the coordinates of the centre of the yield surface (back stress) and R0 is a
material constant that reflects the size of the original yield surface.
3.3 Plasticity Model Component
Many researchers suggested different theories and formulations for concrete modeling. It has
been found that plasticity models are the most appropriate to capture the nonlinearities that grow as
concrete loaded. The advantage of the plasticity theory is the ability to split the strain into elastic
and plastic, the latter one is used to describe the inelastic behavior of concrete under compression.
The main plasticity model formulations are; yield surface (loading surface), the hardening and
softening function, and plastic potential function.
3.3.1 Yield Surface
The yield/loading surface is used to describe the concrete triaxial stress state during plastic
flow in terms of parameters that determine the size and shape of the surface. The yield surface
proposed by Menetrey & Willam (1995) was the most popular in the literature. This yield surface
formula is an extension of the Hoek and Brown failure criterion. It is described in terms of Haigh-
Westergaard in stress space. Haigh-Westergaard coordinates are (ξ, ρ, θ), where ξ is the hydrostatic
stress invariant, ρ is the deviatoric stress invariant, θ is the deviatoric polar angle. The yield surface
proposed by Menetrey & Willam (1995) is given by the following equation:
√ ρ ξ
f = ( 1.5ρ)2 + qn (k)m[ √ r(θ, e) + √ ] − qn (k)qs (k) ≤ 0 (32)
6 3
The terms of the equation will be explained in detail in the coming chapter.
Piscesa et al. (2017) modified the Menetrey & Willam (1995) loading surface equation by intro-
ducing an additional variable α to control the frictional parameter and it is known as the frictional
32
driver parameter. It is used to accurately predict the peak and residual stresses for concrete under
confinement. The new general form of loading surface is written as:
√ ρ ξ
f = ( 1.5ρ)2 + qn (k)αm[ √ r(θ, e) + √ ] − qn (k)qs (k) ≤ 0 (33)
6 3
The frictional parameter m will be modified by α based on the demanding peak stress and residual
stress. The driver parameter is calculated by a linear interpolation function based on the softening
function:
α = (αpeak qs (k) + αres (1 − qs (k))) (34)
where αpeak is the frictional driver parameter at residual, while αres is the frictional driver parame-
ters at peak and both of the parameters are calculated using the following equations:
′ ′
(fcc − fr )2 − fc2 2.5(fc )0.79 2
αpeak = ′2 ; fr = 0 → α peak = − (35)
mfr fc mft m
(fres − fr )2
αres = ; fr = 0 → αres = 0 (36)
mfr fc′ 2
Note that fcc and fres are the peak stress and residual stress for concrete under confinement respec-
tively.
3.3.2 Hardening and Softening Functions
Hardening and softening functions can be described by means of the change of the location
and size of the yield surface, that change is controlled by the hardening/softening parameter (k).
Generally, the hardening parameter is selected to be the length of the plastic strain vector. Notwith-
standing, this parameter has no ability to outline the increase of plastic deformation under multiaxial
33
compression stress. Therefore, many studies were done in the literature to modify the strain hard-
ening material.
Grassl et al. (2002) presented a novel hardening law by using the volumetric plastic strain as
a hardening parameter instead of the length of the plastic strain. Moreover, this hardening law is
associated with non-linear plastic potential and a yield surface based on Hoek and Brown failure
criteria. Accordingly, the novel hardening law is suitable for modeling the load resistance and
deformation capacity of concrete in uniaxial, biaxial, and triaxial compression.
Grassl et al. (2002) proposed one hardening/ softening function q(k) which is consisting of two
functions (Figure 3.13); one representing the hardening qh(k) and the other one representing the
softening qs(k) as shown below:
q(k) = qh(k)qs(k) (37)
Figure 3.13. The split of hardening function into the hardening and softening part.
Since Grassl et al. (2002) doesn’t propose a separate function for the hardening and softening,
many studies were carried out to propose such a function.
Carrazedo et al. (2013) proposed new equations to define the hardening and softening functions
in the ascending and descending branches based on the stress-strain model. Those functions are
simpler than others and can model the behavior of a different type of concrete. The hardening
parameter is selected following the suggestion of Grassl et al. (2002). It has been shown that the
volumetric strain can be decoupled in elastic and plastic components and the plastic volumetric
34
strain is similar for all concrete types at the peak stress. Thus, the volumetric plastic strain is not
used directly as a hardening parameter, a normalized parameter was used instead as shown below:
εpv
k= p (38)
εvo
In the above equation, εpv is the volumetric plastic strain while εpvo is the volumetric plastic strain at
peak stress for uniaxial compression.
Carrazedo et al. (2013) proposed an equation for the descending branch:
C3
qh (k) = 1; qs (k) = (39)
(k − 1)2 + C3
where C3 is a parameter that affects the shape of the hardening function in the descending branch.
Note that changeover between the hardening qh (k) and the softening qs (k) curve occurs when
qh = qs = 1 to ensure continuity and obtain a smooth transition.
3.3.3 Potential Surface
Most of the quasi-brittle materials have a non-associated flow rule, concrete is one of them. In
the situation of non-associated flow, the plastic potential function plays an important role in the
plastic constitutive models. In general, the plastic potential function is used to govern the direction
of the plastic strain vector. Hence, it controls the accumulation of the plastic volumetric strain.
Grassl et al. (2002) suggested a quadratic form for the plastic potential surface using the coor-
dinates in the Haigh-Westergaad stress space.
Papanikolaou & Kappos (2007) proposed a new function for the plastic potential surface that
characterise by Lode-angle (θ) dependency and an adjustable order (n). This plastic potential
surface allows the calibration of the strain under equibiaxial compression. Besides, it has a good
ability to observe the softening behavior of concrete.
Piscesa et al. (2017) also introduced a plastic potential function with plastic dilation rate control
parameter αpo in uniaxial and triaxial settings, which allows the plastic potential function to have a
linear or non-linear plastic dilation rate.
35
Chapter 4
3D Material Model (FORTRAN Code)
4.1 Introduction
Computational technology is proposed to predict the behavior of FRP fully and partially wrapped
concrete columns. The numerical 3D material model is developed by using the FORTRAN pro-
gramming language. The 3D model is adopted and developed by Sarikaya & Erkmen (2019) based
on elastic-plastic model to simulate the behavior of concrete beyond elasticity. This model has the
ability to capture the confinement effect on the concrete structures. Also, each component of the
modelled column (reinforcement, FRP and concrete bulk) can be introduced independently into the
analysis as shown in Figure 4.1.
(a) Concrete bulk. (b) FRP layer. (c) Reinforcement.
Figure 4.1. Column Components.
36
4.2 Finite Element Type
The concrete column model contains the concrete bulk, FRP jacketing as well as steel rein-
forcement which requires different types of elements to work together as they will be connected at
the nodes. In the model developed, a four-node orthotropic shell-type element is used for the FRP
sheets, a two-node beam-type element is used for the steel reinforcement and an 8-node solid ele-
ment is used for the concrete bulk. The shell and solid elements contain drilling degrees of freedom
so that all elements have compatible 6-DOFs as illustrated in Figure 4.2. All finite element types
have been implemented and assembled in FORTRAN.
Figure 4.2. Finite element compatibility.
37
4.3 Boundary Condition, Loading, and Meshing
All the modelled columns in the program were fixed in the bottom in all directions while the
nodes on the top were released except the middle point where the load was applied as shown in
Figure 4.3. The load was applied based on the displacement control method on the middle top node
on the direction Z. It was applied with a minus sign to follow the right direction. The displacement
was increasing by 0.01 mm each step until failure. Each time the corresponding lamda λ was
tabulated in order to obtain the load-deflection curve. The lamda is representing the load factor.
Figure 4.3. Boundary conditions and loading of the concrete column.
38
Meshing is the process where complex geometry is divided into simple elements by generating
nodes and connecting them together. All the columns were modelled with 160 elements (32 mm
mesh size). In order to test the model against mesh sensitivity, one column was modelled many times
with different mesh sizes. The values are 40 mm, 53 mm, and 64 mm respectively. The behavior of
the column with different mesh sizes was compared with the behavior of the 160 element-model.
4.4 Description of the Material Models and Parameters
4.4.1 Concrete Compression Model Implemented in FORTRAN
The plastic model that determines the envelope curve for stress-strain relationship consists of a
potential surface, hardening law, which describes the deformation capacity in multiaxial compres-
sion, and a yield surface. The yield surface proposed by Menetrey & Willam (1995) was employed
which is given as:
√ ρ ξ
ϕp (ξ, ρ, θ, kp ) = ( 1.5ρ)2 + qh (kp )m[ √ r(θ) + √ ] − qh (kp )qs (kp ) ≤ 0 (40)
6 3
where qh and qs controls the shape and location of the loading surface and m can be written as:
fc2 − ft2 e
m=3 (41)
fc ft e + 1
In which fc is the uniaxial compressive strength, ft is the uniaxial tensile strength taken herein as
0.09 fc . Both the plastic potential and the yield surface are constituted by using the unified co-
ordinates in the Haigh–Westergaard stress space, which are based on the stress invariants. The three
co-ordinates ξ, ρ and θ can be expressed as:
I1
ξ=√ (42)
3fc
√
2J2
ρ= (43)
fc
39
√
3 3 J3
cos3θ = (44)
2 J 3/2
2
where:
I1 = σ11 + σ22 + σ33 (45)
1
J2 = [(σ11 − σ22 )2 + (σ22 − σ33 )2 + (σ33 − σ11 )2 ] + τ12
2 2
+ τ23 2
+ τ31 (46)
6
I1 3 I1
J3 = −( ) + ( )2 (σ11 + σ22 + σ33 )
3 3
I1
+( )(τ12 τ12 + τ13 τ13 + τ23 τ23 + σ11 σ22 + σ11 σ33 + σ22 σ33 ) − τ13 τ13 σ22
3
−τ23 τ23 σ11 − τ12 τ12 σ33 + 2τ12 τ13 τ23 + σ11 σ22 σ33 (47)
The eccentricity e defined in the Menetrey & Willam (1995) model can be written as:
1+ϵ
e= (48)
2−ϵ
in which:
ft fb2 − fc2
ϵ= (49)
fb fc2 − ft2
where fb is the equibiaxial compressive strength taken herein as 1.5fc−0.925 . In Equation (40), r(θ)
is the polar radius as:
ν(θ)
r(θ) = (50)
s(θ) + t(θ)
In which:
ν(θ) = 4(1 − e2 )cos2 θ + (2e − 1)2 (51)
40
s(θ) = 2(1 − e2 )cosθ (52)
t(θ) = (2e − 1)[4(1 − e2 )cos2 θ + 5e2 + 4e]1/2 (53)
Hardening and softening of concrete can be simulated by varying the shape and location of the
loading surface during plastic flow. The variation is controlled by the hardening/softening parameter
kp . During the hardening range qh in Equation (40) for concrete can be selected as Papanikolaou &
Kappos (2007):
s
εpvo − kp 2
qh (kp ) = ko + (1 − ko ) 1−( ) (54)
εpvo
where:
σco
ko = (55)
fc
In which σco is the uniaxial concrete stress at the onset of plastic flow. In Equation (54), εpvo is the
threshold value for the volumetric plastic strain at uniaxial concrete strength defined as:
fc
εpvo = (1 − 2v) (56)
Ec
where Ec and v are the Young’s modulus and Poisson’s ratio for concrete, respectively. During
softening range qs in Equation (40) for concrete can be selected as Papanikolaou & Kappos (2007):
1 2
qs (kp ) = ( n1 −1 2 ) (57)
1+ ( n2 −1 )
kp εpvo +t fc
where n1 = ,n
εpvo 2
= εpvo
and t = 15000 . Note that fc is considered in MPa. The potential
function is again written in Haigh-Westergaard stress space and adopted herein from Grassl et al.
41
(2002) expressed as:
ρ ρ ξ
Θp (ξ, ρ, qp ) = −A( p )2 − B p +p (58)
qh (kp )qs(kp ) qh (kp )qs(kp ) qh (kp )qs(kp )
in which:
ψ2 − ψ1
A= (59)
2(ρ1 − ρ2 )
ψ1 − ψ2
B = ρ1 − ψ1 (60)
(ρ1 − ρ2 )
In Equations (59) and (60), ρ1 and ρ2 are the normalized deviatoric stress indicators at uniaxial and
triaxial compressive strength, respectively as:
r
2
ρ1 = (61)
3
r
2 |fcc − σpc |
ρ2 = (62)
3 fc
where fcc is the triaxial compressive strength taken herein as 4.333 fc and σpc is the lateral stress
taken herein as fc . On the other hand, ψ1 and ψ2 are the inclinations of the plastic strain vector
under uniaxial and triaxial compressive strength, respectively given as:
√ |ε3pu − ε1pu |
ψ1 = 2 (63)
εpvo
and
√ |ε3pc − ε1pc |
ψ2 = 2 (64)
εpvo
In the above equation ε3pu is the axial plastic strain component at uniaxial compressive strength,
42
which can be calculated as:
fc
ε3pu = εc − (65)
Ec
and ε1pu is the lateral plastic strain component at uniaxial compressive strength, which is calculated
as:
εpvo − ε3pu
ε1pu = ε2pu = (66)
2
In the equation above, εc is the total strain in the axial direction at uniaxial compressive strength.
The axial plastic strain component at triaxial compressive strength ε3pu is defined as:
1
ε3pu = εcc − (fcc − 2vσpc ) (67)
Ec
and ε1pc is the lateral plastic strain component at triaxial compressive strength, that is obtained from:
εpvo − ε3pc
ε1pc = ε2pc = (68)
2
σpc
For concrete it can be assumed that εcc = εc (1 − 17 fc ), e.g., Papanikolaou & Kappos (2007)
and generally σpc is taken as σpc = fc , and thus in Equation (67), εcc becomes εcc = 18εc . The
damage parameter φ is updated after every converged step. For this purpose, we have adopted the
relationship given in Grassl & Jirásek (2006) defined as:
kp
−C p
φ = (1 − e εvo ) (69)
In which C is a parameter that is to be calibrated based on stiffness degradation.
It is important to note that we have adopted the potential function proposed by J. Lee & Fenves
(1998), which is a special case of Equation (58) obtained by selecting A = 0 and B = − α1p , i.e.
Θp (ξ, ρ) = −Bρ + ξ − a (70)
43
We have used B = −6.25, i.e., αp = 0.16 as suggested by J. Lee & Fenves (1998).
4.4.2 Reinforcement
The longitudinal and transverse reinforcement were considered as elastic perfectly plastic mod-
els. The steel nominal yield stress and the modulus of the elasticity were used in the program to
model the reinforcement behavior.
4.4.3 FRP
The FRP layers were modelled as unidirectional lamina composites. The thickness of the layers
was divided into two parts from the centre of the composite as shown in Figure 4.4. The FRP
material properties were entered in the program including the tensile modulus of elasticity, shear
modulus, and the Poisson’s ratio.
Figure 4.4. Laminates across thickness Erkmen & Gottgens (2018).
44
Chapter 5
Finite Element Model (ABAQUS)
5.1 Introduction
The finite element software ABAQUS is used to model the behavior of concrete columns con-
fined with FRP wraps. The ABAQUS has the ability to analyze reinforced concrete columns
wrapped with FRP. Accordingly, this software is a sufficient way to validate the results from the
numerical strategy based on the coupled plastic damage model.
5.2 Finite Element Type and Meshing
In order to model the concrete column in ABAQUS, different element types have been used for
each part of the column. Concrete, steel and FRP are the major materials used in the model. The
bulk concrete is modelled as a homogeneous 8-node 3D brick element (C3D8R) and the longitudinal
and transverse steel are modelled as linear truss element (T3D2) as shown in Figure 5.1. To model
the FRP jacket, a shell element is used (S8R).
To model the interaction between the concrete and the reinforcement, an embedded region con-
straint is used. The embedded contact region is used to make sure that the number of translational
degrees of freedom (DOF) at a node on the embedded element is identical to the number of trans-
lational degrees of freedom at a node on the host element (Compatible DOF). The reinforcement
was embedded in the concrete which is considered the host region. Therefore, the concrete and
45
Figure 5.1. Finite element types used in ABAQUS ABAQUS (2008).
reinforcement share the same node where a perfect bond was assumed. The interaction between the
concrete and the FRP jacket was considered as cohesive behavior contact.
All elements should have a compatible degree of freedom and share the same node. Accordingly,
all the used elements in the model were assigned the same mesh size to ensure that the results
obtained from the finite element model are accurate. The mesh size that used in the model is 32 mm
to achieve good results with reasonable simulation speed as shown in Figure 5.2.
Figure 5.2. Meshed column in ABAQUS.
5.3 Boundary Condition and Loading
All the modelled columns in ABAQUS were fixed in the bottom in all directions and released
in the top except the top middle point where the load was applied as shown in Figure 5.3. In order
46
to obtain the load-deflection behavior of the modelled columns, a static monotonic load was placed
on the mid-top of the column. The loading was applied until failure using the displacement control
technique. Displacement increments were adjusted to 3mm in each step.
Figure 5.3. Boundary condition and loading in ABAQUS.
5.4 Description of Material Model and Properties
5.4.1 Concrete
The elastic performance of concrete was determined based on the elastic modulus and Poisson’s
ratio. The values of those parameters were used as specified in the experimental data. For the
inelastic behavior, ABAQUS uses the concrete damage plasticity (CDP) constitutive model. This
model considers two main failure mechanisms, which are tensile cracking and compressive crushing
ABAQUS (2008).
The CDP model in ABAQUS forms from plastic behavior, compressive behavior, and tensile
behavior. The compressive behavior of concrete requires determining the relationship between the
yield stress and inelastic strain. The CDP model is primary developed for reinforced concrete struc-
tures. Thus, a design-oriented stress-strain model for concrete Lam & Teng (2003b) was imple-
mented.
47
Plastic Behavior of Concrete
To define the plasticity model of concrete, there are some fundamental parameters which need
to be defined. Those parameters are the dilation angle (ψ), the plastic potential eccentricity (e),
the ratio of initial equibiaxial compressive yield stress to initial uniaxial compressive yield stress
(f bo /f co ), the ratio of the second stress invariant on the tensile meridian which controls the shape of
the yield surface (Kc ), and the viscosity (u). The dilation angle was used as 31 based on calibration.
The eccentricity e, (f bo /f co ), kc , and u were defined as 0.1, 1.16, 2/3, and zero respectively Demir
et al. (2016).
Compressive Behavior of Concrete
To obtain the compressive behavior for confined concrete with FRP, a design-oriented stress-
strain model is used Lam & Teng (2003b). The first part of the stress-strain curve is a parabolic
line and the second part is a straight line as shown in Figure 5.4. The slope of the parabola line
is the modulus of elasticity of the unconfined concrete. The parabolic line meets the straight line
smoothly so there is no change in slope between the two portions. For simplicity, the intersection
′ . The stress-strain curve
point of the two lines represents the unconfined compressive strength fco
′ and the ultimate axial strain ε of
ends at a point where both the ultimate compressive strength fcc cu
the confined concrete are reached.
Figure 5.4. Lam and Teng’s stress-strain model for FRP-confined concrete Lam & Teng (2003b).
48
The Lam and Teng stress-strain model for confined concrete is given by the following equations
Lam & Teng (2003b):



Ec εc − (Ec −E2 )2 2

′
4fco εc 0 ≤ εc ≤ εt
σc (71)
′ +E ε

fco

c c εt ≤ εc ≤ εcu
where:
′
2fco
εt = (72)
Ec − E2
′ − f′
fcc co
εt = (73)
εcu
In which σc is the axial stress, εc is the axial strain of confined concrete, Ec is the elastic modulus of
unconfined concrete, εt is the axial strain at the transition point, and E2 is the slope of the straight
line of the curve.

′ is obtained from Samaan et al. (1998):
In this model fcc
′
fcc fl
′
= 1 + k1 ′ (74)
fco fco
where fl is the confining pressure and it can be obtained from:
2σj t 2Ef rp εj t
fl = = (75)
d d
In which σj is the FRP jacket hoop stress, t is the total thickness of FRP, d is the diameter of the
confined concrete core, Ef rp is the FRP modulus of elasticity in hoop direction, and εj is the hoop
tensile strain of the FRP at failure. The hoop stress σj is related to the hoop strain εj and it can be
obtained from Equation (76) as applied in the confining pressure formula:
σj = Ef rp εj (76)
49
In the developing confinement model, it has been suggested that the hoop tensile strain εj should
be taken as the actual hoop rupture strain measured in the FRP jacket εh,rup (εj = εh,rup ) rather than
considering it as FRP material ultimate tensile strain εrup as existed in most of the FRP confined
concrete models Karbhari & Gao (1997) Saafi et al. (1999) Pessiki et al. (2001). Also, the value of
k1 in Equation (74) is considered as 3.3 Lam & Teng (2003b).
FRP efficiency factor Kε has been proposed to relate the actual hoop rupture strain εh,rup to the
FRP material ultimate tensile strain εrup Pessiki et al. (2001):
εrup = Kε εf rp (77)
It has been shown that the value of the efficiency factor Kε is varying with the type of the FRP.
Based on database collected from the literature, it has been found that the average value of Kε is
0.586 for carbon FRP (CFRP) Lam & Teng (2003a).
To calculate the ultimate axial strain of uniformly confined concrete εcu , the following equation
is used:
εcu fl εh,rup 0.45

= 1.75 + k2 ′ ( ) (78)
εco fco εco
In which εco is the axial strain at the compressive strength of unconfined concrete, and k2 is the
strain enhancement coefficient. The value of those two parameters is 0.002 and 12 respectively
Lam & Teng (2003a).
This design- oriented stress-strain model was extended to model rectangular concrete columns
confined with FRP. Therefore, some modifications were applied to the compressive strength and
ultimate axial strain formulas by adding shape factors ks1 and ks2 as shown below:
′
fcc fl
′
= 1 + k1 ks1 ′ (79)
fco fco
εcu fl εh,rup 0.45

= 1.75 + k2 ks2 ′ ( ) (80)
εco fco εco
50
Those shape factors are proposed to take into consideration the effective confined concrete area
in rectangular columns since the rectangular sections have non-uniform confinement stress. The
strength enhancement ks1 and the strain enhancement ks2 shape factors are defined as follow Lam
& Teng (2003b):
b Ae
ks1 = ( )α (81)
h Ac
h Ae
ks2 = ( )β (82)
b Ac
where h ≥ b, and Ae /Ac is the effective confinement area ratio. Based on some experimental data,
the value of the two exponents α and β have been found to be 2 and 0.5 respectively Lam & Teng
(2003b).
5.4.2 Reinforcement
The behavior of steel was modelled as elastic perfectly plastic model. The parameters which
were used to define the model are modulus of elasticity, yield stress, and Poisson’s ratio. Figure 5.5
illustrates the reinforcement arrangement for BS2 in ABAQUS.
Figure 5.5. Reinforcement configuration in ABAQUS for BS2.
51
5.4.3 FRP
FRP is considered in ABAQUS as unidirectional lamina composites. A unidirectional lamina is
considered as a homogenous body and consisting of continuous fibres strongly bonded in a matrix
as shown in Figure 5.6. The longitudinal fibres are the load-carrying elements while the matrix
provides protection to the fibres and distributes the load between fibres Cristescu et al. (2003).
The FRP jacket is modelled as linear elastic orthotropic material, which means that the mechanical
Figure 5.6. Lamina with unidirectional fibres Cristescu et al. (2003).
properties of the material are inconsistent in different directions.
To determine the composite in-plane elastic properties which are required for FRP in ABAQUS
(Figure 5.7), the ‘mechanics of material approach’ is used. This approach assumes that the compos-
ite is void-free, the fibres have the same shape and size, they are distributed equally, and fibres are
perfectly bonded to the matrix. Accordingly, the overall prosperities of the composite are obtained
based on volume fractions for both fibres (f ) and matrix (m).
Longitudinal Young modulus E1 is estimated based on the rule of mixtures by assuming the
52
Figure 5.7. Material properties required for FRP in ABAQUS.
same strain in the fibre and matrix Kollar & Springer (2003) as follows:
E1 = Em Vm + Ef 1 Vf (83)
whereas, the transverse Young modulus E2 is determined from the following relation that is often
called the inverse rule of mixtures:
1 Vm Vf
= + (84)
E2 Em Ef 2
The overall Poisson’s ratio v12 can be obtained by using the same assumption of the rule of
mixtures as follows:
v12 = vm Vm + vf 12 Vf (85)
where Vm is the volume fraction of the matrix, Vf is the volume fraction of the fibre, Em and Ef
are the Young moduli for the matrix and fibre respectively. The rest of the Poisson’s ratios in the
53
different directions are determined by using Maxwell relation Tsai (1988) as given below:
vij vji
= (86)
Ei Ej
Since the composite material is subjected to shear stress, shear moduli are required for model-
ing transverse shear deformation in all directions. The overall in-plain shear modulus G12 can be
estimated by assuming that the fibre and matrix have the same shear stress Cristescu et al. (2003).
Thus, the shear modulus of the composite material is calculated from the following:
1 Vm Vf
= + (87)
G12 Gm Gf 12
where Gm and Gf are the shear moduli for the matrix and fibre respectively. The subscripts 1,2, and
3 represent the laminate coordinate system. Axis 1 is parallel to the fibres, axis 2 is perpendicular
to the fibres, and axis 3 is perpendicular to the plain of Lumina as illustrated in Figure 5.6.
Usually it is assumed that the unidirectional lamina can be treated as transversely isotropic
material Soden et al. (2004). Accordingly, the independent elastic constants are minimized to five
since E2 = E3 , G12 = G13 , v12 = v13 and G23 = E2 /2(1 + v23 ) Tsai (1988).
The FRP sheets are modelled as lamina composites in ABAQUS. The mechanical properties of
the composites are obtained from the experimental data Rousakis & Karabinis (2012).
Since there are not enough experimental data available about the composite’s in-plane elastic
properties, some reasonable assumptions were made based on the type of the FRP sheet for numer-
ical modeling purposes. The carbon C-sheet 240 is made from carbon fibres and epoxy resins as
a matrix to support the fibres. The mechanical properties for the carbon fibres and the resin epoxy
matrix that were considered in the ABAQUS modeling Daniel et al. (2006) are shown in Table 5.1
and Table 5.2.
To obtain the mechanical properties of the unidirectional lamina that are entered in ABAQUS
software, the rule of mixtures is used as explained in the equations above. Regarding the volume
fractions of both fibres and matrix, two assumptions were made. The assumptions are; the ratio
between Ef /Em = 30 and the matrix volume fraction is 0.3, accordingly the fiber volume fraction
is 0.7 Matthys (2000).
54
Table 5.1. Mechanical properties of the carbon fiber.
Property Carbon fibre (AS4)

Density (g/cm3 ) 1.81
Longitudinal modulus,Ef 1 (GPa) 235
Transverse modulus, Ef 2 (GPa) 15
Axial shear modulus, Gf 12 (GPa) 27
Transverse shear modulus, Gf 21 (GPa) 7
Poisson’s ratio vf 12 0.2
Table 5.2. Mechanical properties of the resin epoxy matrix.
Property Epoxy matrix (HY6010)

Density (g/cm3 ) 1.17
Young’s modulus,Em (GPa) 3.4
Shear modulus Gm (GPa) 1.26
Poisson’s ratio vm 0.36
Table 5.3 illustrates the mechanical properties of the lamina composites based on; the mechan-
ical properties of the fibres and matrix, assumptions that were made, and the equations that were
derived based on the rule of mixture.

Table 5.3. The mechanical properties of the lamina composites as used in ABAQUS.
Property E1 (GPa) E2 (GPa) v12 G12 (GPa) G13 (GPa) G23 (GPa)
CFRP 165.52 7.4127 0.248 3.78 3.78 2.969
55
Chapter 6
Numerical Results
6.1 Introduction
In this chapter, the outcomes of the 3D material model and the finite element model in ABAQUS
are presented. All the results are graphically illustrated in terms of stress-strain curves. The results
are obtained by converting the load-deflection curves into stress-strain curves. The vertical axis
represents the stress which is calculated by dividing the load by the column’s cross-section area.
While the horizontal axis represents the strain which is computed by dividing the displacement by
the length of the column. Each figure has three lines, those lines are representing the behavior of the
concrete columns obtained from experiments, the 3D material model, and the ABAQUS software
respectively.
6.2 Validation of the Numerical Model
The 3D material model was validated against the experimental results which are obtained from
a previous study and another numerical finite element model that is generated by ABAQUS.
6.2.1 Reinforced Concrete Column
Figure 6.1 shows the performance of a reinforced concrete (RC) column with longitudinal and
transverse steel bars. It is clear from the figure that the results of the 3D material model are in good
56
agreement with the experimental and the finite element model in ABAQUS. Therefore, this model
has the ability to simulate the mechanical behavior of reinforced concrete columns.
Figure 6.1. Stress-strain curve for reinforced concrete column (BS2).
6.2.2 Plain Concrete Columns Confined with FRP
Figures 6.2 and 6.3 illustrate the stress-strain curves for plain concrete columns confined with
one and five carbon FRP layers respectively. The performance of the columns based on the 3D
material model is in good harmony with the experimental and ABAQUS results. Therefore, the
model can simulate the performance of confined concrete columns with different numbers of FRP
layers.
Figure 6.2. Stress-strain curve for plain concrete column wrapped with one FRP layer (BC1).
57
Figure 6.3. Stress-strain curve for plain concrete column wrapped with five FRP layers (BC5).
6.2.3 Reinforced Concrete Columns Confined with FRP
Figures 6.4-6.8 show the stress-strain curves at mid-span for retrofitted reinforced concrete
columns with FRP. The model was tested for five different reinforced concrete column types; the re-
inforcement configuration and the number of FRP layers were different in each case. It is clear from
the figures that the 3D numerical model can capture the behavior of reinforced concrete columns
after FRP retrofitting since its performance is in good agreement with the experiment and ABAQUS
outcomes.
Any difference between the results that are obtained from the 3D material model and ABAQUS
could be due to using different material models for concrete and different finite element types. For
example, to model the reinforcement in the 3D material model, a beam element was used while in
ABAQUS, a truss element was used.
Figure 6.4. Stress-strain curve for RC column wrapped with one layer of FRP (BS1C1).
58
Figure 6.5. Stress-strain curve for RC column wrapped with three layers of FRP (BS1C3).
Figure 6.6. Stress-strain curve for RC column wrapped with five layers of FRP (BS1C5).
Figure 6.7. Stress-strain curve for RC column wrapped with three layers of FRP (BS2C3).
59
Figure 6.8. Stress-strain curve for RC column wrapped with five layers of FRP (BS2C5).
6.3 Results of the 3D Material Model
In order to compare the behavior of confined and unconfined concrete columns, the validated
3D material model was used. Figure 6.9 demonstrates the mechanical behavior of three different
column types; unconfined column, confined columns with FRP jackets, and reinforced columns
with FRP jackets. It is clear from the results that the load-carrying capacity of the columns increased
when applying the FRP jackets. In addition, there is a direct relationship between the strength of
the column and the number of applied FRP layers; the strength is increasing with the increase of the
thickness of the FRP jacket (number of layers).
The enhancement in the column strength was 15% when applying one layer of the FRP and
80.7% when using five FRP layers. Applying reinforcement and one layer of FRP improved the
load-carrying capacity of the column by approximately 9 MPa. The maximum increase in the
axial stress of the column was reached when applying reinforcement and five layers of FRP to the
unconfined column, the strength was almost doubled.
Furthermore, as observed from the results, the confined columns with FRP layers show more
ductile behavior when compared with the unconfined case. The ductility can be remarked by the
increase in the axial confined strain.
60
Figure 6.9. Stress-strain curves for retrofitted concrete columns using the 3D material model.
6.4 Parametric Study
A parametric study was conducted to test the sensitivity of the mesh size on the stress-strain
curve of a confined reinforced concrete column with five layers of FRP using the 3D material model.
In Figure 6.10, the numerical results of the stress-strain curve for BS2C5 are illustrated with mesh
sizes of 64mm, 53mm, 40mm, and 32mm. It is clear from the results, that the mesh size does not
have a major effect on the behavior of the retrofitted column when using the 3D material model.
In addition, the 3D material model was used to model partially wrapped plain concrete (BC1)
and reinforced concrete (BS2C5) columns. Their behavior was compared with fully and unwrapped
(unconfined) cases as shown in Figure 6.11 and Figure 6.12 respectively. The width of the FRP layer
was 32 mm and the vertical spacing of the FRP strips was 32mm which started from the bottom of
the column. Another spacing distance of 96mm was used in the case of the plain concrete column.
The FRP strips configuration is graphically demonstrated in Figure 6.13 for 32 mm and 96mm
vertical spacing. It was concluded that the 3D model is capable of modeling partially wrapped
columns. It is obvious from the outcomes that the partial wrapping technique can enhance the
strength of the columns. In addition, it was noticed that when FRP strips spacing decreases, the
load-carrying capacity of the column rises.
61
Figure 6.10. Stress-strain curve for BS2C5 with different mesh sizes using 3D material model.
Figure 6.11. Comparing the behavior of partially wrapped plain concrete column (BC1) with
different spacing with fully wrapped and not wrapped cases using 3D material model.
Figure 6.12. Comparing the behavior of partially wrapped reinforced concrete column (BS2C5)
with fully wrapped and not wrapped cases using 3D material model.
62
(a) 32mm spacing. (b) 96mm spacing.
Figure 6.13. Different vertical spacing of FRP strips on the concrete column.
63
Chapter 7
Implementation of Optimization Method

to Obtain the Confined Strain Equation
7.1 Introduction
Using building design codes is very important to ensure adequate structural safety and quality.
Engineers depend on the design codes and standards to provide guidance and minimum require-
ments for sufficiently designed elements. Recently, many countries and institutions have established
codes, guidance, and standards for designing confined reinforced concrete columns with FRP.
For example, the American Concrete Institute (ACI 440.2R,2017) has introduced a section with
four pages length for designing confined concrete columns with FRP, the German Committee for
Structural Concrete (DAfStb- Guideline, 2012) has a section with five pages length, and the Stan-
dardization Administration of the People’s Republic of China (GB 50608, 2010) has considered the
design of confined concrete columns with a section of eight pages in length. Whereas, the Cana-
dian Standards Association (CSA S806, 2012) is limited to one-page length. The number of pages
reveals the significance of considering the confinement effect in several standards.
This chapter presents an optimization method procedure that is recommended to the Canadian
Standards with respect to the maximum confined concrete strain. A design formula is proposed
to describe the effect of the concrete and FRP mechanical parameters on the value of strain. To
obtain this effect numerically, a parametric study is conducted using the 3D material model. The
64
data involved in this parametric study are used to define the designed formula with the aid of the
optimization algorithm, performing what is known as nonlinear regression.
7.2 Gaps and Limitations in the Canadian Standards
While reviewing and comparing the above codes, it has been found that the Canadian Standards
(CSA S806-12) has gaps and limitations in designing confined reinforced concrete columns with
FRP as summarized below:
• CSA S806-12 is limited to fully wrapped columns only so there is no consideration for par-
tially wrapped columns.
• CSA S806-12 is not providing a formula to calculate the maximum confined concrete strain
after applying FRP. It always considers the confined strain as the peak concrete strain value
which is 0.0035. However, an accurate value for confined concrete strain is important to
predict the behavior of confined columns under axial compressive load.
• CSA S806-12 does not specify any design model for the confined concrete.
• When wrapping rectangular columns, the code is limited to a ratio of 1.5 to h/b (h/b <=
1.5).
• For rectangular sections confined with transverse FRP laminates, section corners shall be
rounded to a radius not less than 20 mm before the FRP application.
• CSA S806-12 is limited to the ultimate limit state so there is no consideration for serviceabil-
ity limit state.
• The load combination that considered in CSA S806-12 is a combination of axial compression
and bending (not applicable for axial compression only)
• CSA S806-12 considers a strength reduction factor of 0.75 for FRP jackets without consider-
ing the type of composite material (e.g. Carbon FRP or Glass FRP).
65
7.3 Discussing the Gaps and Limitations in the Canadian Standards
The current work is focusing on presenting recommendations to fill some of the gaps and limi-
tations in the Canadian Standards that are related to the design of concrete columns reinforced with
FRP composites. In this chapter, a parametric study is conducted using the 3D material model in
order to provide a design formula for the maximum confined concrete strain that is wrapped with
FRP layers using nonlinear regression.
The outcomes of the 3D material model show that the model has the ability to model par-
tially wrapped columns as illustrated in section 6.4 while the Canadian Standards is limited to fully
wrapped columns only. The load capacity of the column increases when the spacing of the FRP
strips decrees. Also, the results of the 3D model show that the maximum confined concrete strain is
increasing with the number of applied FRP layers (Figure 7.1) not as specified in the CSA S806-12
as a fixed value, which is equal to the peak concrete strain 0.0035.
Figure 7.1. Maximum confined concrete strain vs number of layers of FRP from 3D material
model and the Canadian Standards.
Figure 7.2 shows the stress-strain curves of the unconfined plain concrete column and the perfor-
mance of the column when adding different numbers of FRP layers. It is clear from the results that
there is a direct relationship between the number of applied FRP jackets and the maximum confined
concrete strain value. Moreover, the same relationship is applied for the reinforced concrete column
66
as demonstrated in Figure 7.3. The maximum confined strain increased from 0.0042 to 0.0073 after
adding five layers of FRP to the reinforced concrete column. Accordingly, the effect of the FRP
layers (confinment effect) on the maximum confined concrete strain should be addressed clearly.
Figure 7.2. Stress-strain curves for plain concrete column wrapped with different number of FRP
layers using the 3D material model.
Figure 7.3. Stress-strain curves for reinforced concrete column wrapped with different number of
FRP layers using the 3D material model.
67
The ACI 440.2R-17 considered the effect of the applied FRP layers to the concrete columns on
the maximum concrete strain (εcc ) as specified in the following equation:
flj εju k4
εcc = εc0 k2 + εc0 k3 ( ) (88)
fc εc0
where fc is the unconfined concrete strength, εc0 is the peak unconfined concrete strain, εju is the
FRP’s rupture strain, flj is the confinement pressure provided by FRP as shown in Equation (89),
and k2 − k4 are factors with values of 1.5, 12, and 0.45 respectively.
2tj Ej εju
flj = (89)
D
where tj is the thickness of the FRP, Ej is the modulus of the composite material, and D is the
diameter of the circular cross-section.
The Chinese guidance (GB 50608, 2010) used the following approach for finding the confined
concrete strain:
s
fljd
εcc = 0.0035 + 0.015 , fljd ≤ 0.6 εF RP,k (90)
fcd
where fljd is the design FRP jacket strength, fcd is the design concrete strength, and εF RP,k is the
characteristic maximum strain for an FRP sheet.
The German standards (DAfStb- Guideline, 2012) implemented the same approach as used in
fl j fl jk
the ACI 440.2R-17 but with different factors and fc was replaced by fc . In which fljk is the
characteristic FRP jacket strength, and k2 − k4 are 1.75, 19, and 0 respectively.
7.4 The Proposed Nonlinear Regression Algorithm to Obtain the Max-
imum Confined Concrete Strain
In this part of the work a nonlinear regression is performed, where data obtained from a paramet-
ric study is fit to a model and then expressed as a mathematical function. The nonlinear regression
method is illustrated in Algorithm 1. This method starts by defining the Inputs which are:
68
• Geometry parameters of the column include: the dimensions of the column, boundary condi-
tions, and mesh sizing.
• Material parameters which are: concrete unconfined strength, concrete modulus of elasticity,
concrete Poisson’s ratio, unconfined concrete strain, the thickness of the FRP layer, tensile
and shear modulus of FRP layer and the Poisson’s ratio of FRP.
• A combination of thirteen possible equations for the maximum confined concrete strain for
square columns under compressive load.
Algorithm 1: The Nonlinear Regression Algorithm

Input : Geometry parameters, Material parameters , Possible Equations
Output: Regression parameters
1 Set the inputs for the tested column in the 3D material model program ;
2 Set the parametric study using structural analysis program ;
3 Run the structural analysis program ;
4 Obtain the output data in tabulated format ;
5 Load selected parameters (from Step 2) and the maximum confined strain and strength
(from Step 4) into the Matlab code ;
6 Set possible number of N equation forms for the maximum confined concrete strain ;
7 for i ← 1 to N do
8 Define a set of initial conditions (M) for the parameter coefficients related to
Equation i ;
9 Run the optimization code to obtain the optimal parameter coefficients for initial
condition of Equation i ;
10 Select regression parameters that correspond to the best performance ;
11 end
12 Compare the performance of the N equations ;
13 Rank the N equations ;
14 Return regression parameters of the best equation
The terms of the equations were suggested based on the literature and design standards review-
ing Ozbakkaloglu et al. (2013). There are unlimited numbers of equations that can be recommended
to be tested by the optimization procedure. In the current research, thirteen equations for calculating
the maximum confined concrete strain under compressive load were recommended to be fitted to
the 3D material model results as shown below:
ε
k2 + log(tf rp ) ∗ Ef ju
rp Econc
Eq 1 : ε′cc = k1 + + k4 ∗ log( ′ ) (91)
(b ∗ f c)k3 fcc
69
ε
rp
Eq 2 : ε′cc = k1 + (92)
(b ∗ f c)k3
ε
rp Econc
Eq 3 : ε′cc = k1 + + k3 ∗ log( ′ ) (93)
(b ∗ f c) fcc
ε
rp Econc ∗ εju
Eq 4 : ε′cc = k1 + + k3 ∗ log( ′
) (94)
(b ∗ f c) fcc
Econc
Eq 5 : ε′cc = k1 + k2 ∗ log(tf rp ) + k3 ∗ log( ′
) (95)
fcc
Econc
Eq 6 : ε′cc = k1 + k2 ∗ tf rp + k3 ∗ ′
(96)
fcc
k2 ∗ tf rp ∗ Ef rp k3
Eq 7 : ε′cc = k1 + ( ) (97)
b ∗ fc
tf rp ∗ Ef rp k2 fc
Eq 8 : ε′cc = (k1 ∗ ) + k3 ′ (98)
b ∗ fc fcc
Ef rp ∗ fc
Eq 9 : ε′cc = k1 + k2 ∗ tf rp + k3 ∗ ( ′
) (99)
Econc ∗ fcc
Ef rp ∗ log(fc )
Eq 10 : ε′cc = k1 + k2 ∗ tf rp + k3 ∗ ( ′
) (100)
Econc ∗ fcc
Ef rp ∗ log(fc )
Eq 11 : ε′cc = k1 + k2 ∗ tf rp + k3 ∗ ( ′ )
) (101)
Econc ∗ log(fcc
ε
k3 + k4 ∗ log(tf rp )Ef ju
rp k6
Eq 12 : ε′cc = k1 + (k2 ∗ ) (102)
k5 ∗ b ∗ f c
70
2 ∗ tf rp ∗ Ef rp ∗ εju εju k3
Eq 13 : ε′cc = 0.0035 ∗ k1 + 0.0035 ∗ k2 ( )( ) (103)
b ∗ fc 0.0035
The steps for the nonlinear regression algorithm are as follows:
• Step 1: Is to Set the inputs for the 3D material model. These inputs are the column geometry
and the material parameters.
• Step 2: In order to Set the parametric study, a structural analysis program that is written on the
FORTRAN Language is used. This program is working in parallel with the 3D material model
based on changing some material parameters simultaneously in the model’s input folders as
shown in the given example in Figure (7.4) to test their effect on the maximum confined
concrete strain. The parameters that were alternating are the thickness of the FRP layer,
unconfined concrete strength, the elastic modulus of concrete, and the tensile modulus of
FRP.
• Step 3: Run the structural analysis program.
• Step 4: Obtaining the corresponding maximum confined concrete strain and strength from
the outputs and tabulated in Table 7.1.
• Step 5: Load the thickness of FRP layer tf rp , unconfined concrete strength fc , elastic
′ , the
modulus of concrete Econc , tensile modulus of FRP Ef rp , confined concrete strength fcc
rupture strain of the FRP εju , and width of the column cross-section b into the Matlab code.
• Step 6: Set the possible number of equations (Equation 91 - 103) in the Matlab code.
• Step 7- Step 8: Define a set of initial values for the parameter coefficients related to each
equation.
• Step 9: Run the optimization code to obtain the optimal parameter coefficients for the Equa-
tions. The optimization method used in this work is based on least squares optimization al-
gorithm and it is illustrated in fitnlm (n.d.), which was developed by MathWorks developers.
The method proved its efficiency and reliability to perform linear and nonlinear regression.
The Matlab function mdl = f itnlm(tbl, modelf un, beta0) is used to fit the model specified
71
by modelf un to variables in the table or dataset array tbl, and returns the nonlinear model
mdl. then, f itnlm estimates model coefficients using an iterative procedure starting from the
initial values in beta0. The term tbl represents the material parameters and the corresponding
confined concrete strain and strength shown in Table 7.1. The term modelf un represents the
possible equations. The term beta0 is the initial values for the coefficient in the proposed
equation.
• Step 10: The optimization method calculates the R2 value, P-value, and Mean Square Er-
ror for tuned coefficient, and then Select regression parameters that correspond to the best
performance.
• Step 12- Step 13: Compare and Rank the equations based on their performance.
• Step 14: Return regression parameters of the best equation.
Figure 7.4. Changing some of the material parameters simultaneously in the structural analysis
program.
72
Table 7.1. The material parameters and the corresponding confined concrete strain and strength.
Parameter 1 Parameter 2 Parameter 3 Parameter 4 Parameter 5

εcc tf rp (mm) fc (MPa) Econc (MPa) Ef rp (MPa) ′ (MPa)
fcc
1 0.0044 0.117 -25.5 19300 240000 29.33
2 0.00534 0.351 -25.5 19300 240000 36.61
3 0.00578 0.585 -25.5 19300 240000 44.31
4 0.00618 0.785 -25.5 19300 240000 51.2625
5 0.0043 0.117 -25.5 20000 240000 29.23
6 0.0052 0.351 -25.5 20000 240000 36.38
7 0.0059 0.585 -25.5 20000 240000 44.22
8 0.00625 0.785 -25.5 20000 240000 51.12
9 0.00418 0.117 -25.5 20500 240000 29.175
10 0.00512 0.351 -25.5 20500 240000 36.209
11 0.0058 0.585 -25.5 20500 240000 44.05
12 0.006188 0.785 -25.5 20500 240000 50.85
13 0.00415 0.117 -25.5 21500 240000 29.078
14 0.005125 0.351 -25.5 21500 240000 35.925
15 0.00575 0.585 -25.5 21500 240000 43.624
16 0.00625 0.785 -25.5 21500 240000 50.68
17 0.00378 0.117 -25.5 23000 240000 28.9
18 0.0048 0.351 -25.5 23000 240000 35.523
19 0.00525 0.585 -25.5 23000 240000 42.635
20 0.00581 0.785 -25.5 23000 240000 49.587
21 0.0044 0.117 -25.5 19300 220000 29.036
22 0.0052 0.351 -25.5 19300 220000 35.692
23 0.0057 0.585 -25.5 19300 220000 42.809
24 0.00625 0.785 -25.5 19300 220000 49.306
25 0.0044 0.117 -25.5 19300 260000 29.629
26 0.00556 0.351 -25.5 19300 260000 37.608
27 0.006125 0.585 -25.5 19300 260000 46.298
28 0.0045 0.117 -25.5 19300 280000 29.925
29 0.00556 0.351 -25.5 19300 280000 38.535
30 0.00625 0.585 -25.5 19300 280000 47.998
31 0.00453 0.117 -25.5 19300 300000 30.219
32 0.00575 0.351 -25.5 19300 300000 39.534
73
The material parameters and the corresponding confined concrete strain and strength (Continue).
Parameter 1 Parameter 2 Parameter 3 Parameter 4 Parameter 5

εcc tf rp (mm) fc (MPa) Econc (MPa) Ef rp (MPa) ′ (MPa)
fcc
33 0.00625 0.585 -25.5 19300 300000 49.49
34 0.0035 0.117 -20 19300 240000 23.096
35 0.00434 0.351 -20 19300 240000 28.917
36 0.004656 0.585 -20 19300 240000 35.105
37 0.00515 0.785 -20 19300 240000 41.007
38 0.00518 0.117 -30 19300 240000 34.422
39 0.00625 0.351 -30 19300 240000 42.915
40 0.0044375 0.117 -25.5 19500 240000 29.3123
41 0.00475 0.117 -27 19300 240000 31.039
42 0.005 0.117 -29 19300 240000 33.3036
43 0.0054 0.117 -31 19300 240000 35.566
44 0.0035 0.117 -20 19300 220000 22.86
45 0.00431 0.117 -25 19300 220000 28.472
46 0.005156 0.117 -30 19300 220000 34.077
47 0.003531 0.117 -20 19300 260000 23.33
48 0.00437 0.117 -25 19300 260000 29.05
49 0.00522 0.117 -30 19300 260000 34.766
50 0.0036 0.117 -20 19300 300000 23.8
51 0.00444 0.117 -25 19300 300000 29.63
52 0.00531 0.117 -30 19300 300000 35.46
53 0.00475 0.117 -27 19300 240000 31.04
54 0.005109 0.117 -29 19300 240000 33.3
55 0.0054 0.117 -31 19300 240000 35.56
56 0.00418 0.351 -20 19300 220000 28.148
57 0.00528 0.351 -25 19300 220000 35.028
58 0.00593 0.351 -30 19300 220000 41.74
59 0.00428 0.351 -20 19300 260000 29.605
60 0.00544 0.351 -25 19300 260000 36.882
61 0.0043 0.351 -20 19300 300000 31.04
62 0.00547 0.351 -25 19300 300000 38.69
63 0.00509 0.117 -27 19300 500000 34.94
64 0.005625 0.117 -30 19300 500000 38.74
74
7.4.1 Performance Measures
In this work P- values, R2 , Mean Square Error (MSE) and graphical analysis of the data set are
used to identify whether the used equations with their optimal coefficients are suitable to express
the relation between maximum strain and material properties.
P-value is a conditional probability that identifies whether the obtained results are significant
or not when the null hypothesis is assumed to be true. The null hypothesis is a hypothesis that
is used in statistics to show that there is no difference between certain specified populations. For
example, there is no statically significant relationship between the type of food I feed the cows and
the growth of the cows. In other words, the coefficients in the equation are not important, there is no
significant effect. The alternative hypothesis is the opposite of the null hypothesis which means that
there is a difference between two or more variables. For example, the coefficients in the equation
are important and have a significant effect. When the p-value is less than a chosen significant level
then the null hypothesis is rejected. This significant level is type 1 error (α) which is the probability
of rejecting the null hypothesis when it is true and usually it specified as 0.05. The test hypothesis
is characterized as below:
H0 (Null hypothesis): βi = 0 [coefficient is not important]
H1 (Alternative hypothesis): βi ̸= 0 [coefficient is important]
The R-squared (R2 ) or the coefficient of determination is a statistical measure of how close the
data are to the fitted regression line. While the Mean Squared Error (MSE) measures the average
of the squares of the error that indicates how close the regression line is to the set of points. The
graphical analysis of the data set is illustrated in the next subsection.
7.4.2 Results and Discussion
The results from the nonlinear regression are represented as:
- The R2 value, coefficient optimal parameters, and P-values for each equation as shown in
Table 7.2.
- The maximum strain obtained from the 3D material model vs maximum strain obtained from
each equation is graphically illustrated in Figure 7.5 to Figure 7.17. The black dots indicate
75
the material model strain obtained from the parametric study using the 3D material model,
while the red dots represent the fitted strain from each proposed equation.
- The Mean Square Error for all proposed equations as demonstrated in Figure 7.18
After comparing the outcomes of the proposed algorithm, it was shown that there are three
equations provided the best fitting, these equations are:
• Eq1 which has the maximum coefficient of determination value of 0.979, and minimum MSE
value of 1.35e−08 and the P-values for the four optimal coefficients are less than 0.05, which
means that all the coefficients are significant to Eq1 . The graphical analysis of the data set
for this equation that is shown in Figure 7.5 demonstrates an accurate fitting to the material
model strain data set.
• Eq5 has a coefficient of determination value of 0.957, MSE of 2.66e−08, and all its parameter
coefficients are significant. Figure 7.9 shows that this equation provides a suitable fitting to
the data set.
• Eq6 has a coefficient of determination value of 0.956 and MSE of 2.72e − 08. Based on the
P-values, this equation has three coefficients, only two of them are significant to the equation.
Figure 7.10 shows that this proposed equation demonstrates a good fitting to the material
maximum strain.
The equations that showed the worst fitting are:
• Eq7 which has a low coefficient of determination value of 0.378, and the higher MSE of
3.88e − 07. The P-values demonstrate that only one of its coefficients is significant to the
equation. Figure 7.11 shows that this equation has a bad fitting to the material model strain
data set.
• Eq10 which also has a low coefficient of determination value of 0.491 and a high MSE of
3.18e − 08. The graphical analysis of the data set for this equation in Figure 7.14 confirms
this assessment.
76
• Eq13 has the lowest coefficient of determination with the value of 0.374 and a high MSE of
3.84e − 07. Figure 7.17 reveals the worst fitting among all equations.
Table 7.2. The results obtained from nonlinear regression algorithm with respect to each possible
equation.
Equation# Initial values R2 Optimal parameters P-value MSE

12 k1=0.024818 4.6966e-33
0.01 k2=-51.879 0.00022232
Eq.1 0.979 1.35e-08
0.5 k3 =1.1999 7.961e-40
70 k4=-0.0029206 5.4989e-23
1 k1=0.0091448 5.2125e-41
Eq.2 120 0.892 k2=-24.77 4.248e-14 6.71e-08
0.5 k3=1.005 4.183e-83
0.003 k1=0.011276 7.1299e-17
Eq.3 0.01 0.889 k2=-21.836 3.6696e-14 6.30e-08
7 k3=-0.00036508 0.032744
0.003 k1=0.010595 1.0062e-22
Eq.4 0.01 0.889 k2=-21.836 3.6696e-14 6.30e-08
-7 k3=-0.00036508 0.032744
0.003 k1= 0.030397 1.4556e-40
Eq.5 12 0.957 k2 =0.00012541 0.0040699 2.66e-08
7 k3 =-0.0040304 4.4035e-35
0.003 k1=0.0090297 1.6775e-50
Eq.6 12 0.956 k2=-0.00023872 0.077188 2.72e-08
7 k3=-6.7948e-06 1.4235e-34
0.003 k1 =0.0041323 1.3381e-06
Eq.7 12 0.378 k2 =4.2841e-06 0.89377 3.88e-07
7 k3=0.68949 0.25234
0.1 k1 =0.021655 1.5444e-16
Eq.8 2.7 0.733 k2 = -0.13069 1.4786e-21 1.67e-07
1.2 k3 =-0.015856 9.1023e-14
0.003 k1=0.0032483 1.5131e-08
Eq.9 12 0.517 k2= 0.0030953 8.2621e-09 3.01e-07
0.2 k3= 8.5339e-05 0.030504
1 k1=0.0049043 3.9461e-14
Eq.10 1.2 0.491 k2=0.0018967 0.00010302 3.18e-07
1 k3=-0.00037153 0.23275
1 k1=0.003337 2.6166e-11
Eq.11 12 0.527 k2=0.0026876 2.7188e-11 2.95e-07
1 k3=7.235e-05 0.015174
0.0035 k1 = -0.4033 6.6333e-11
1 k2=-0.097571 0.99112
2 k3=-0.19835 0.97342
Eq.12 0.669 2.06e-07
3 k4=-1.6628e+05 0
12 k5=79027 0
0.5 k6=0.2263 0.97386
1.5 k1=1.2577 1.1434e-40
Eq.13 12 0.374 k2=0.078382 1.1178e-08 3.84e-07
0.45 k3= -0.56192 1.9197e-82
77
Figure 7.5. Maximum strain obtained from 3D Figure 7.6. Maximum strain obtained from 3D
material model vs maximum strain obtained material model vs maximum strain obtained
from Equation 1. from Equation 2.
78
79
Figure 7.17. Maximum strain obtained from 3D material model vs maximum strain obtained from
Equation 13.
Figure 7.18. The Mean Square Error for all proposed equations.
7.5 Summary Outcome
An optimization-based nonlinear regression algorithm was proposed to test the performance of
the proposed equations for the maximum confined concrete strain. The objective function of the
optimization method that is performed in Matlab is to minimize the mismatch between the strain
calculated from each equation (ε′cc ) and the obtained strain from the 3D material model (εcc ).
Equation 13 was selected from the American standards (ACI 440.2R-17) to be tested in the
nonlinear regression. The only change that was done is replacing the diameter D with the width of
column b since the proposed equation in the ACI 440.2R-17 is for circular cross-section and in this
work, the column’s cross-section is squared. It has been found that the coefficients k1 -k3 in Eq13
80
are 1.257, 0.0078 and -0.562 respectively to perform best fitting to the 3D material model results.
However, the values of those coefficient in ACI 440.2R-17 are as follows k1 = 1.5, k2 = 12, and k3 =
0.45.
Based on the performance measure of each equation, it has been shown that the following equa-
tions are providing the best fitting to maximum confined concrete strain that is obtained from the
3D material model:
ε
rp Econc
Eq 1 : ε′cc = k1 + + k4 ∗ log( ′ )
(b ∗ f c)k3 fcc
Econc
Eq 5 : ε′cc = k1 + k2 ∗ log(tf rp ) + k3 ∗ log( ′
)
fcc
Econc
Eq 6 : ε′cc = k1 + k2 ∗ tf rp + k3 ∗ ′
fcc
81
Chapter 8
Conclusion and Future Work
8.1 Introduction
This study is conducted to capture the behavior of retrofitted concrete columns with Fibre Re-
inforced Polymer (FRP). Given this, a 3D material model based on elasto-plastic models is adopted
using the FORTRAN language. 3D elasto-plastic models are particularly suitable for the modeling
and analysis of retrofitted concrete columns as confinement pressure is generally unknown before
the analysis, which depends on the size, shape, loading, boundary conditions as well as material
properties.
This research is mainly focused on the validation of a 3D elasto-plastic model. Accordingly,
finite element analysis software ABAQUS is used to develop a numerical finite element model for
the confined concrete columns. The parameters of the concrete constitutive model are introduced in
detail in the 3D material model and in the ABAQUS model. Also, the same material properties and
boundary conditions of the column are used in both models. The 3D material model is verified by
comparing its outcome with the ABAQUS finite element model and experimental results from the
literature.
A parametric study is generated to illustrate the applicability of the proposed 3D constitutive
model in modeling partially wrapped square concrete columns and to test the effect of some material
properties on the confined concrete strain. The outcomes of the parametric study are recommended
to improve a part of the Canadian Standards which is related to concrete columns reinforced with
82
FRP composites. Consequently, a nonlinear regression algorithm is proposed to obtain an equation
to calculate the maximum confined strain based on optimization method. This developed algorithm
tested the performance of multiple equations and classified these equations based on their respective
R2 values, P-values, MSE, and graphical behavior.
8.2 Summary Discussion
The major findings and observations from this study are summarized as follows:
• The 3D material model was validated against the ABAQUS model and the experimental re-
sults in simulating the behavior of retrofitted columns with FRP composites under monotonic
axial compressive loading.
• The proposed 3D material model has the ability to clearly predict the performance of fully
and partially wrapped concrete square columns with FRP under compressive load.
• The FRP jacket which is applied to the concrete column increases the strength and the ductil-
ity of the column.
• The strength of partially wrapped square concrete columns rises as the FRP strips vertical
spacing reduces.
• Mesh size does not have a significant effect on the column behavior when modeling the col-
umn using the 3D material model.
• The material properties of the concrete and FRP affect the value of the maximum confined
strain.
• There is a direct relationship between the number of the FRP layers applied to confine the
concrete column and the value of the maximum confined concrete strain.
• Thirteen possible equations for the maximum confined strain were proposed to be tested using
a nonlinear regression; with the aid of the optimization method, only three equations were
recommended for the Canadian standards based on the higher R2 and least P-value and MSE.
83
8.3 Recommendations for Future Work
Based on the work done in this research, the following are recommended for future studies in
the field of modeling retrofitted concrete columns with FRP:
• Test the 3D material model for cyclic loading.
• Propose equations to calculate the maximum confined strength and strain for partially wrapped
columns using the proposed nonlinear optimization method.
• Utilizing artificial intelligence approach to perform the nonlinear regression.
84
References
ABAQUS, I. (2008). Abaqus analysis user’s manual, version 6.8. ABAQUS Providence, RI.
Armero, F., & Oller, S. (2000). A general framework for continuum damage models. i. infinitesimal
plastic damage models in stress space. International Journal of Solids and Structures, 37(48-50),
7409–7436.
Attard, M., & Setunge, S. (1996). Stress-strain relationship of confined and unconfined concrete.
Materials Journal, 93(5), 432–442.
Bakis, C. E., Bank, L. C., Brown, V., Cosenza, E., Davalos, J., Lesko, J., . . . Triantafillou, T.
(2002). Fiber-reinforced polymer composites for construction—state-of-the-art review. Journal
of composites for construction, 6(2), 73–87.
Basalo, F. J. D. C., Matta, F., & Nanni, A. (2012). Fiber reinforced cement-based composite system
for concrete confinement. Construction and Building Materials, 32, 55–65.
Beer, F. P., & Johnston, E. (1992). Mechanics of materials. new york7 mcgraw-hill. Inc.
Benallal, A., Billardon, R., & Doghri, I. (1988). An integration algorithm and the corresponding
consistent tangent operator for fully coupled elastoplastic and damage equations. Communica-
tions in Applied Numerical Methods, 4(6), 731–740.
Borja, R. I. (2013). Plasticity: modeling & computation. Springer Science & Business Media.
Carrazedo, R., Mirmiran, A., & de Hanai, J. B. (2013). Plasticity based stress–strain model for
concrete confinement. Engineering structures, 48, 645–657.
85
Chakrabarti, A., Chandra, A., & Bharagava, P. (2008). Finite element analysis of concrete columns
confined with frp sheets. Journal of reinforced plastics and composites, 27(12), 1349–1373.
Cheng, L., & Karbhari, V. M. (2006). New bridge systems using frp composites and concrete: a
state-of-the-art review. Progress in Structural Engineering and Materials, 8(4), 143–154.
Chung, K., & Lee, M.-G. (2018). Basics of continuum plasticity. Springer.
Cristescu, N. D., Craciun, E.-M., & Soós, E. (2003). Mechanics of elastic composites. Chapman
and Hall/CRC.
Daniel, I. M., Ishai, O., Daniel, I. M., & Daniel, I. (2006). Engineering mechanics of composite
materials (Vol. 1994). Oxford university press New York.
Demir, A., Caglar, N., Ozturk, H., & Sumer, Y. (2016). Nonlinear finite element study on the
improvement of shear capacity in reinforced concrete t-section beams by an alternative diagonal
shear reinforcement. Engineering Structures, 120, 158–165.
Erkmen, R. E., & Gottgens, B. (2018). A shell element for buckling analysis of thin-walled
composite-laminated members. International Journal of Structural Stability and Dynamics,
18(02), 1850021.
Farahmandpour, C., Dartois, S., Quiertant, M., Berthaud, Y., & Dumontet, H. (2017). A concrete
damage–plasticity model for frp confined columns. Materials and Structures, 50(2), 1–17.
Fattah, A. A. E. (2018). New axial stress-strain model of square concrete columns confined with
lateral steel and frp. Composite Structures, 202, 731–751. doi: 10.1016/j.compstruct.2018.03
.085
fitnlm. (n.d.). Retrieved from https://www.mathworks.com/help/stats/fitnlm
.html;jsessionid=d3a8b63ce136c1287ff60b24ca62
Ghernouti, Y., & Rabehi, B. (2011). Frp-confined short concrete columns under compressive
loading: experimental and modeling investigation. Journal of reinforced plastics and composites,
30(3), 241–255.
86
Grassl, P., & Jirásek, M. (2006). Damage-plastic model for concrete failure. International journal
of solids and structures, 43(22-23), 7166–7196.
Grassl, P., Lundgren, K., & Gylltoft, K. (2002). Concrete in compression: a plasticity theory with
a novel hardening law. International Journal of Solids and Structures, 39(20), 5205–5223.
Han, W., & Reddy, B. D. (2012). Plasticity: mathematical theory and numerical analysis (Vol. 9).
Springer Science & Business Media.
Hany, N. F., Hantouche, E. G., & Harajli, M. H. (2016). Finite element modeling of frp-confined
concrete using modified concrete damaged plasticity. Engineering Structures, 125, 1–14.
Ibrahimbegovic, A. (2009). Nonlinear solid mechanics: theoretical formulations and finite element
solution methods (Vol. 160). Springer Science & Business Media.
Ilki, A. (2006). Frp strengthening of rc columns (shear, confinement and lap splices). BULLETIN-
FIB, 35, 123.
Ilki, A., Peker, O., Karamuk, E., Demir, C., & Kumbasar, N. (2008). Frp retrofit of low and
medium strength circular and rectangular reinforced concrete columns. Journal of Materials in
Civil Engineering, 20(2), 169–188.
Isleem, H. F., Wang, D., & Wang, Z. (2018). Modeling the axial compressive stress-strain be-
havior of cfrp-confined rectangular rc columns under monotonic and cyclic loading. Composite
Structures, 185, 229–240.
Jianguo, C., Jian, F., Jing, W., & Yunfeng, C. (2008). Behavior of frp-wrapped rectangular concrete
columns. In 2008 international conference on information management, innovation management
and industrial engineering (Vol. 3, pp. 445–448).
Karbhari, V. M., & Gao, Y. (1997). Composite jacketed concrete under uniaxial compres-
sion—verification of simple design equations. Journal of materials in civil engineering, 9(4),
185–193.
Kelly, P. (2019). Mechanics lecture notes: An introduction to solid mechanics. Retrieved January,
25, 2019.
87
Klisiński, M., & Mroz, Z. (1988). Description of inelastic deformation and degradation of concrete.
International journal of solids and structures, 24(4), 391–416.
Kollar, L. P., & Springer, G. S. (2003). Mechanics of composite structures. Cambridge university
press.
Krabbenhøft, K. (2002). Basic computational plasticity. University of Denmark.
Lam, L., & Teng, J. (2003b). Design-oriented stress-strain model for frp-confined concrete in
rectangular columns. Journal of reinforced plastics and composites, 22(13), 1149–1186.
Lam, L., & Teng, J. G. (2003a). Design-oriented stress–strain model for frp-confined concrete.
Construction and building materials, 17(6-7), 471–489.
Lee, J., & Fenves, G. L. (1998). Plastic-damage model for cyclic loading of concrete structures.
Journal of engineering mechanics, 124(8), 892–900.
Lee, J.-Y., Oh, Y.-J., Park, J.-S., & Mansour, M. Y. (2004). Behaviors of concrete columns confined
with both spiral and fiber composites. In 13th world conference on earthquake engineering
(Vol. 13).
Lemaitre, J. (1985). A continuous damage mechanics model for ductile fracture.
Lubliner, J., Oliver, J., Oller, S., & Onate, E. (1989). A plastic-damage model for concrete. Inter-
national Journal of solids and structures, 25(3), 299–326.
Matthys, S. (2000). Structural behaviour and design of concrete members strengthened with exter-
nally bonded frp reinforcement (Unpublished doctoral dissertation). Ghent University.
Menetrey, P., & Willam, K. (1995). Triaxial failure criterion for concrete and its generalization.
Structural Journal, 92(3), 311–318.
Meschke, G., Lackner, R., & Mang, H. A. (1998). An anisotropic elastoplastic-damage model for
plain concrete. International journal for numerical methods in engineering, 42(4), 703–727.
Mirmiran, A., Zagers, K., & Yuan, W. (2000). Nonlinear finite element modeling of concrete
confined by fiber composites. Finite elements in analysis and design, 35(1), 79–96.
88
Moodi, Y., Mousavi, S. R., Ghavidel, A., Sohrabi, M. R., & Rashki, M. (2018). Using response
surface methodology and providing a modified model using whale algorithm for estimating the
compressive strength of columns confined with frp sheets. Construction and Building Materials,
183, 163–170.
Ozbakkaloglu, T., Lim, J. C., & Vincent, T. (2013). Frp-confined concrete in circular sections:
Review and assessment of stress–strain models. Engineering Structures, 49, 1068–1088.
Ozbakkaloglu, T., & Saatcioglu, M. (2007). Seismic performance of square high-strength concrete
columns in frp stay-in-place formwork. Journal of Structural Engineering, 133(1), 44–56.
Papanikolaou, V. K., & Kappos, A. J. (2007). Confinement-sensitive plasticity constitutive model
for concrete in triaxial compression. International Journal of Solids and Structures, 44(21),
7021–7048.
Parvin, A., & Brighton, D. (2014). Frp composites strengthening of concrete columns under various
loading conditions. Polymers, 6(4), 1040–1056.
Pessiki, S., Harries, K. A., Kestner, J. T., Sause, R., & Ricles, J. M. (2001). Axial behavior of
reinforced concrete columns confined with frp jackets. Journal of composites for Construction,
5(4), 237–245.
Pham, T. M., & Hadi, M. N. (2014). Stress prediction model for frp confined rectangular concrete
columns with rounded corners. Journal of Composites for Construction, 18(1), 04013019.
Piscesa, B., Attard, M., Samani, A., & Tangaramvong, S. (2017). Plasticity constitutive model for
stress-strain relationship of confined concrete. ACI Structural Journal, 114(2).
Richart, F. E., Brandtzæg, A., & Brown, R. L. (1929). Failure of plain and spirally reinforced
concrete in compression (Tech. Rep.). University of Illinois at Urbana Champaign, College of
Engineering . . . .
Rousakis, T. C., & Karabinis, A. I. (2008). Substandard reinforced concrete members subjected to
compression: Frp confining effects. Materials and structures, 41(9), 1595–1611.
89
Rousakis, T. C., & Karabinis, A. I. (2012). Adequately frp confined reinforced concrete columns
under axial compressive monotonic or cyclic loading. Materials and structures, 45(7), 957–975.
Saafi, M., Toutanji, H., & Li, Z. (1999). Behavior of concrete columns confined with fiber reinforced
polymer tubes. Materials Journal, 96(4), 500–509.
Saatcioglu, M., & Razvi, S. R. (1992). Strength and ductility of confined concrete. Journal of
Structural engineering, 118(6), 1590–1607.
Samaan, M., Mirmiran, A., & Shahawy, M. (1998). Model of concrete confined by fiber composites.
Journal of structural engineering, 124(9), 1025–1031.
Sarikaya, A., & Erkmen, R. (2019). A plastic-damage model for concrete under compression.
International Journal of Mechanical Sciences, 150, 584–593.
Simo, J. C., & Hughes, T. J. (2006). Computational inelasticity (Vol. 7). Springer Science &
Business Media.
Simo, J. C., & Ju, J. (1987). Strain-and stress-based continuum damage models—i. formulation.
International journal of solids and structures, 23(7), 821–840.
Soden, P. D., Hinton, M. J., & Kaddour, A. (2004). Lamina properties, lay-up configurations and
loading conditions for a range of fibre reinforced composite laminates. In Failure criteria in
fibre-reinforced-polymer composites (pp. 30–51). Elsevier.
Teng, J., Xiao, Q., Yu, T., & Lam, L. (2015). Three-dimensional finite element analysis of reinforced
concrete columns with frp and/or steel confinement. Engineering Structures, 97, 15–28.
Tsai, S. W. (1988). Composites design. Think Composites, P. O. Box 581, Dayton, Ohio 45419,
USA, 1988. 583.
Voyiadjis, G. Z., Taqieddin, Z. N., & Kattan, P. I. (2008). Anisotropic damage–plasticity model for
concrete. International journal of plasticity, 24(10), 1946–1965.
Yazdani, S., & Schreyer, H. (1990). Combined plasticity and damage mechanics model for plain
concrete. Journal of engineering mechanics, 116(7), 1435–1450.
90
Yu, T., Teng, J., Wong, Y., & Dong, S. (2010). Finite element modeling of confined concrete-ii:
Plastic-damage model. Engineering structures, 32(3), 680–691.
Zeng, J., Guo, Y., Li, L., & Chen, W. (2018). Behavior and three-dimensional finite element
modeling of circular concrete columns partially wrapped with frp strips. Polymers, 10(3), 253.
Zhang, D., Li, N., Li, Z.-X., & Xie, L. (2020). Experimental investigation and confinement model
of composite confined concrete using steel jacket and prestressed steel hoop. Construction and
Building Materials, 256, 119399.
Zignago, D., Barbato, M., & Hu, D. (2018). Constitutive model of concrete simultaneously confined
by frp and steel for finite-element analysis of frp-confined rc columns. Journal of Composites for
Construction, 22(6), 04018064.
91

Al-Maadhidi MASc S2022

Uploaded by

Copyright:

Available Formats

Al-Maadhidi MASc S2022

Uploaded by

Document Information

Original Title

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

Al-Maadhidi MASc S2022

Uploaded by

Copyright:

Available Formats

Analysis of Retrofitted Concrete Columns using 3D

Building, Civil and Environmental Engineering

Presented in Partial Fulfillment of the Requirements

for the Degree of

Master of Applied Science (Civil Engineering) at

Montréal, Québec, Canada

© Zaineb Al-Maadhidi, 2021

This is to certify that the thesis prepared

By: Zaineb Al-Maadhidi

Entitled: Analysis of Retrofitted Concrete Columns using 3D Elastic-Plastic

and submitted in partial fulfillment of the requirements for the degree of

Master of Applied Science (Civil Engineering)

originality and quality.

Signed by the Final Examining Committee:

Analysis of Retrofitted Concrete Columns using 3D Elastic-Plastic Modeling

to implement an elastic-plastic model to simulate FRP-jacketed concrete columns. The 3D material

describing the impact of material properties on the confined concrete strain.

me during the research program.

their endless motivation and help.

contribute to improving the Canadian Standards.

courage me with their kind words, much appreciated!

Columns using 3D Elastic-Plastic-Damage Modelling, ”Canadian Society for Civil Engineering

Annual Conference. 2021. https://csce2021.ca/wp-content/uploads/2021/05/

List of Tables xiii

1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.5 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2 Models for FRP Confined Concrete Columns . . . . . . . . . . . . . . . . . . . . 6

2.2.1 Compressive Strength Models . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2.2 Stress-Strain Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.2.3 Finite Element Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.4 Literature Review Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3.2 Elasto-Plastic Material Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.2.2 Frictional Block Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2.3 The Tangent and Plastic Modulus . . . . . . . . . . . . . . . . . . . . . . 24

3.2.4 The Yield Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.2.5 Plastic Potential and Plastic Flow Rule . . . . . . . . . . . . . . . . . . . 28

3.2.6 Consistency Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.2.7 Principle of Plastic Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.2.8 Strain Hardening and Perfect Plasticity . . . . . . . . . . . . . . . . . . . 30

3.3 Plasticity Model Component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.3.1 Yield Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.3.2 Hardening and Softening Functions . . . . . . . . . . . . . . . . . . . . . 33

3.3.3 Potential Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4 3D Material Model (FORTRAN Code) 36

4.2 Finite Element Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.3 Boundary Condition, Loading, and Meshing . . . . . . . . . . . . . . . . . . . . . 38

4.4 Description of the Material Models and Parameters . . . . . . . . . . . . . . . . . 39

4.4.1 Concrete Compression Model Implemented in FORTRAN . . . . . . . . . 39

5 Finite Element Model (ABAQUS) 45

5.2 Finite Element Type and Meshing . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5.3 Boundary Condition and Loading . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5.4 Description of Material Model and Properties . . . . . . . . . . . . . . . . . . . . 47

6.2 Validation of the Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . 56

6.2.1 Reinforced Concrete Column . . . . . . . . . . . . . . . . . . . . . . . . 56

6.2.2 Plain Concrete Columns Confined with FRP . . . . . . . . . . . . . . . . 57

6.2.3 Reinforced Concrete Columns Confined with FRP . . . . . . . . . . . . . 58

6.3 Results of the 3D Material Model . . . . . . . . . . . . . . . . . . . . . . . . . . 60

6.4 Parametric Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61