Aci Bond & Development of Reinforcement

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
COPYRIGHT 2003; ACI International (American Concrete Institute) `ÿ Document provided by IHS Licensee=Aramco HQ/9980755100, User=, 07/23/2003
00:34:55 MDT Questions or comments about this message: please call the Document
Policy Management Group at 1-800-451-1584.
~~
Obb2949 05425bb 258
Bond and Development

of Reinforcement
A Tribute to Dr. Peter Gergely
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Editor international
Roberto Leon SP- 180
W Obb2949 0 5 4 2 5 b ï 194
DISCUSSION of individual papers in this symposium may be submitted in

accordance with general requirements of the AC1 Publication Policy to AC1
headquarters at the address given below. Closing date for submission of discus-
sion is June 1, 1999. All discussion approved by the Technical Activities Com-
mittee along with closing remarks by the authors will be published in the
NovemberíDecember 1999 issue of either AC1 Structural Journal or AC1 Mate-
rials Journal depending on the subject emphasis of the individual paper.
The Institute is not responsible for the statements or opinions expressed in its
publications. Institute publications are not able to, nor intended to, supplant
individual training, responsibility, or judgment of the user, or the supplier, of
the information presented.
The papers in this volume have been reviewed under Institute publication proce-
dures by individuals expert in the subject areas of the papers.
Copyright 1998 @
AMERICAN CONCRETE INSTITUTE

P.O. Box 9094
Farmington Hills, Michigan 48333-9094
All rights reserved including rights of reproduction and use in any form or by
any means, including the making of copies by any photo process, or by any
electronic or mechanical device, printed or written or oral, or recording for
sound or visual reproduction or for use in any knowledge or retrieval system or
device, unless permission in writing is obtained from the copyright proprietors.
Printed in the United States of America
Editorial production: Jane Carroll
Library of Congress catalog card number: 98-86927
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
PREFACE
During the AC1 Spring Convention 1997, AC1 Committee 408, Bond and
Development of Reinforcement, organized four sessions intending to assess
the state-of-the-art in bond research, practical applications, and code develop-
ments. The sessions were co-sponsored by the Technical Group 2/5-Bond
Models of the Comité Euro-Internationale du Béton (CEB).
The symposium drew a very large number of abstracts from its open cail for
papers. The final selection of the abstracts was based on both the technical
merit and relevance of the topic. The majority of the presenters have provided
written contributions to this volume, which is intended to serve as the third
major conference on bond research in the past 15 years. The first major con-
ference was organized in Scotland, UK, in 1982, and the second in Riga,
Latvia, in 1992. Although these conferences are not formally connected, their
intent has been to provide a forum for discussion of advanced mechanistic
bond models, experimental research aimed both at model verification and
development of design provisions, and practical design problems associated
with bond and development of reinforcement.
The sessions held in Seattle, Washington, were organized into a symposium

honoring Dr. Peter Gergely, a long-time member of Committee 408, who had
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
recently passed away. This volume is dedicated to Dr. Peter Gergely, an out-
standing engineer, educator and friend whose joie de vivre will not be forgot-
ten by those who came into contact with him.
The editor would like to acknowledge the assistance of CEB 2/5 chairman, Dr.
Rajels Tepfers of Chalmers University of Technology, Göteborg, Sweden, in
organizing the symposium. Dr. Tepfers is retiring after a long and distin-
guished career in structural engineering that includes pioneering work in the
area of bond and development of reinforcement.
Roberto Leon, Chairman, AC1 Committee 408

School of Civil and Environmental Engineering
Georgia Institute of Technology
Atlanta, February 1998
Obb2943 05425bî Tb7 D
A Tribute to Dr. Peter Gergely
The symposium held during the AC1 meeting in Seattle in March 1997 and
this Special Publication were organized to honor the contributions Dr. Peter
Gergely made to the American Concrete Institute, especially his work in bond
and anchorage, and his leadership of AC1 Committee 408, Bond and Develop-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ment of Reinforcement.
Peter Gergely was a professor of structural engineering at Corneil University

until the time of his death on August 25th, 1995, after a long and courageous
battle against cancer. He was born in Budapest, Hungary, on February 12th,
1936. He entered the Technical University of Budapest in 1954 but left in
1956 to join the freedom fighters in the Hungarian Revolution. He was one of
the last resistance fighters to escape before Soviet troops invaded. He fled to
Austria and then immigrated to Canada, where he received a bachelor’s
degree in civil engineering from McGill University in Montreal in 1960. He
received a doctorate from the University of Illinois in 1963. Immediately
afterwards, he joined the faculty of Cornell University.
Peter Gergely is survived by his wife Kinga, son Zoltan, and daughter and
son-in-law, Illa and David Burbank. One of his greatest pleasures was to taik
about his grandson, Istvan, who he was sure would carry on his love of chess.
Peter Gergely had an exceptional zest for life and took great pride in his Hun-
garian ancestry. He often spoke of his early school years in Budapest and
about how lucky he was to have had wonderful, inspirational teachers, espe-
cially in mathematics and literature. These early experiences with great teach-
ers inspired his teaching and his superb ability to motivate students to achieve
their best. He was one of the most successful educators in the Cornell College
of Engineering, and in 1995, he received a college award for his outstanding
dedication to teaching.
V
W û b b i ? ï 4 9 0542570 787
Peter made major contributions to a wide variety of structural engineering

problems. His research led to many important advances in understanding the
mechanics of reinforced and prestressed concrete, and he worked to develop
results that could be used to improve building codes. One of his early projects
on cracking in concrete members resulted in an equation that has been in the
AC1 Building Code since 1971. He served on many AC1 committees and was
chairman of AC1 Committee 408 from 1974 to 1980. Many of the provisions
for development length and splices now in the Building Code were proposed
and moved forward during his chairmanship.
In recognition of his outstanding contributions, Dr. Gergely received many

AC1 honors. He was elected a Fellow of the American Concrete Institute
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(1974); he received the AC1 Delmar Bloem Distinguished Service Award
(1981), was a co-recipient of the AC1 Wason Award for Most Meritorious
Paper (1993), and received the AC1 Structural Research Award (1994). He
was a co-recipient of the ASCE State of the Art of Civil Engineering Prize
(1974) and the ASCE Raymond C. Reese Research prize (1976). Of ail his
honors, the one that meant the most to him was the Honorary Doctorate he
received in 1992 from his alma mater, the Technical University of Budapest,
given “for his outstanding international activities in advancing the develop-
ment of his profession of mechanics and reinforced concrete.”
His love of learning, joy in discovery, passion for music, chess, and all things
Hungarian were evident to ail who knew him. His dedication to teaching and
to structural engineering will continue to impact his students and his col-
leagues. AC1 will m i s s his unique talents and leadership abilities.
James O. Jirsa, Chairman

Civil Engineering Department
The University of Texas at Austin
Austin, January 1998
VI
CONTENTS
EVALUATION OF BOND PERFORMANCE IN REINFORCED

CONCRETE STRUCïURES
by S. L. McCabe and S. J. Pantazopoulou .............................. 1
BOND AND SPLïïïING: A VEXING QUESTION

by P. G. Gambarova, G. P. Rosati and C. E. Schumm . . . . . . . . . . . . . . . . . . . . 2 3
CONTRIBUTION OF CONCRETE BETWEEN CRACKS AT INELASTIC

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
STEEL STRAINS AND CONCLUSIONS FOR THE OPTIMIZATION

OF BOND
by R. Eligehausen, J. Ozbolt and U. Mayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 5
A DILATATIONAL-INTERFACE MODEL FOR BOND

byJ.V.Cox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
TENSION STIFFENING AND CRACKING BEHAVIOR

IN HIGH-STRENGTH CONCRETE
by G. Creazza, R. Di Marco, S. Russo and E. Siviero . . . . . . . . . . . . . . . . . . . 105
BOND UNDER REPEATED LOADING

byG.L.Balázs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
BOND MODELLING OF PRESTRESSING STRAND

by J. A. den Uijl ................................................ 145
CONFINEMENT ROLE IN ANCHORAGE CAPACITY

by E. Giuriani and G. A. Plizzari ................................... 171
BOND OF RIBBED BARS MODELED THROUGH CONCRETE

CONFINEMENT
by A. J. Bigaj, J. A. den Uijl and J. C. Walraven. ...................... 195
PULL-OUT BOND BEHAVIOR OF RIBBED BARS IN NORMAL

AND HIGH-STRENGTH CONCRETE WITH VARIOUS CONFINEMENTS
by B. Engström, J. Magnusson and Z. Huang ........................ .215
PREVENTING BRITTLE FAILURE OF TENSION SPLICES

IN HIGH-STRENGTH CONCRETE
by A. Azizinamini. ............................................. .243
Obb2949 0542572 551
BOND OF EPOXY-COATED REINFORCEMENT

IN NORMAL AND HIGH-STRENGTH CONCRETE
by T. Grundhoffer, P. A. Mendis, C. W. French and R. Leon. . . . . . . . . . . . . . 261
OVERVIEW OF RESEARCH TO IMPROVE THE DEVELOPMENT

CHARACTERISTICS OF REINFORCING BARS
by D. Darwin, J. Zu0 and M. L. Tholen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,299
BOND SLIP OF BRIDGE COLUMN REINFORCEMENT ANCHORED

IN CAP BEAMS
by S . Sritharan, J. M. Ingham, M. J. N. Priestley and F. Seible . . . . . . . . . . . . 319
PULLOUT OF HORIZONTAL REINFORCEMENT EBEDDED

IN MASONRY
by E. Vintzileou and N. Psilla. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
EFFECTS OF TRANSVERSE REINFORCEMENT AND BONDED LENGTH

ON THE SIDE-BLOWOUT CAPACITY OF HEADED REINFORCEMENT
by R. A. DeVries, J. O. Jirsa and T. Bashandy. . . . . . . . . . . . . . . . . . . . . . . . .367
PERFORMANCE OF CONCRETE BRIDGE DECKS AND SLABS

REINFORCED WITH EPOXY-COATED STEEL
UNDER REPEATED LOADING
by H. O. Hasan, D. B. Cleary and J. A. Ramirez ...................... .391
DESIGN RECOMMENDATIONS FOR EPOXY-COATED

REINFORCEMENT
by J. Cairns, J. O. Jirsa and S. L. McCabe . . . . . . . . . . . . . . . . . . . . . . . . . . . .405
FUNDAMENTAL ANALYSIS OF RC LAP SPLICES

by T. Ichinose, T. Hayashi and W. Lin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .431
BOND PROPERTIES OF HIGH-STRENGTH

FIBER REINFORCED CONCRETE
byB.AarupandB.C.Jensen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
BOND STUDIES OF REINFORCING BARS

IN SILICA FUME CONCRETE
by B. S. Hamad and M. S. Itani . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .473
BOND OF FRP REINFORCEMENT IN CONCRETE:

A STATE-OF-THE-ART IN PREPARATION
by R. Tepfers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,493
VIII
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
IObb2949.:0542573 496 I
SP 180-1
Evaluation of Bond Performance in Reinforced

Concrete Structures
by S. L. McCabe and S. J. Pantazopoulou
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
SvnoDsis: The increasing significance of performance-based criteria in modem
structural design has motivated new considerations in bond design of conventional
reinforcing steels, relating to more reliable assessment of both the demand and the
supply sides of the anchorage/development design problem. Accurate identification
of the required anchorage lengths needed to ensure strain compatibility, by proper
consideration of the conditions affecting bond, is necessary to limit slippage of the
steel relative to the concrete. While minimum development lengths calculated by
designers imply that the bar is fully anchored, it is well established by expenmental
observation that in practice there is always some bar slip.
Recent research results from around the world provide the basis for improved
understanding of the effects on bond performance of critical parameters such as
confinement, spacing, and material properties. Much of this work has been empirical
in nature and the applicability of empirical design expressions in calculations is
limited. Nonlinear finite element calculations and other sophisticated analysis re-
quires more information as to how the bond failure proceeds than simply an upper
limit.
This paper will summarize the available information that exists both within
North America through AC1 and within the CEB as to the viable approaches and
philosophies that can be applied to the bond problem. The range of appIication of
the various techniques will be identified as will limitations and needs for more re-
search.
Keywords: Anchorage; bond; bond models; design; detailing; development; finite

element method; plasticity; reinforcement
1
2 McCabe and Pantazopoulou
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Steven L. McCabe is Associate Professor of Civil Engineering at the Universi-
ty of Kansas. His research interests include bond and development of reinforcement,
earthquake engineering, finite element methods and damage mechanics, and the
development of new reinforcing systems. He is an active member of CRSI, ASCE,
CEB and AC1 among others. Currently, McCabe is the chair of AC1 Committee 439,
Steel Reinforcement, a member of Building Code subcommittee 3 18-B, Development
and Reinforcement, and a member of the CEB Bond Models Task Group.
Stavrvoula J. Pantazopoulou is an Associate Professor of Civil Engineering at
the University of Toronto. She is active in AC1 and CEB serving on AC1 Committees
368, Earthquake Resisting Concrete Structurai Elements and Systems; AC1 447, Finite
Element Analysis of Reinforced Concrete Structures; AC1 408, Bond of Reinforce-
ment, and the CEB Bond Models Task Group. Her research interests include bond
and development of reinforcement, development design and analysis methods for
earthquake resisting structural systems and the application of finite element in prob-
lems of continuum mechanics.
INTRODUCTION
In the traditional capacity design approach used in North America for rein-
forced concrete structures, designing for bond was considered to be a question of
providing adequate strength along the bar anchorage for development of the bar
yield force. Therefore, bond, as well as all other tension-failure controlled problems
such as shear, were dealt with in design by establishing allowable stress limits for
concrete, with stresses calculated from an admissible state of equilibrium without
consideration of deformations. When needed, calculations of ductility capacity were
performed after dimensioning and were associated with purely flexural response.
Hence, of the two coordinates in the load-deformation response recorded in anchor-
age tests, the AC1 code writing bodies have focused only on the load capacity side
(with particular reference to yielding of reinforcement). The usual approach taken in
the past was to synthesize a summary of experimental data for bond strength using
empirical relations that subsequently formed the basis of the design requirements.
These empirical relations attempt to describe the parametric dependence of bond
strength on an array of important design variables. Although this process is fraught
with debate, within AC1 the main point of concern has really only been in selecting
the form of the design expressions. The array of significant variables used, as well as
the underlying philosophy of the approach, is generally accepted as a necessary
accessory of the capacity design framework. Note that whereas deformation (slip)
was universally acknowledged in the reported tests which formed the basis of current
design requirements for bond, no systematic attempt had been made to process the
available information in a manner analogous to the efforts placed on the side of
strength.
The emergence of new performance based design approaches have refocused
the problem of bond, since at the heart of these approaches lie considerations about
deformation demand and supply, which inevitably bring forth questions about
compatibility of deformations between concrete and the anchored bar throughout the
various stages of response. For the design to meet certain predetermined perfor-
mance requirements, it may no longer be adequate to simply assess the strength of a
given anchorage without at the same time knowing the levels of deformation that
would have to develop in order for the given strength to be realized, and whether
these deformations can be realized in the structure without completely altering the
assumed load path (redistribution). For this reason, code writing bodies which have
already explicitly adopted performance-based design approaches, such as the CEB,
require modeling of the complete load deformation envelope of the bond-slip re-
Obb2949 0542575 260.b
Tribute to Peter Gergely 3
sponse of a given anchorage as reference for design purposes. Complete bond-slip

relationships have been obtained either by empirical modeling of tests, or by analyti-
cal (mechanistic, numerical, etc.) methods. These approaches will be critically re-
viewed in the following sections with the objective to identify the salient features,
range of applicability and common points between alternative methods. The scope of
the review is limited to straight bar anchorages and will concentrate on monotonic
loading, although many of the observations that can be made are common to both
monotonic and cyclic loading.
REVIEW OF AVAILABLE MODELING APPROACHES
The array of important variables controlling the mechanics of bond both
directly or indirectly appears from familiar experimental evidence to be quite exten-
sive and has not been completely defined as of yet. The parametric complexity of the
problem discourages formulation of straightforward mathematical descriptions. Even
when restricting the scope to straight bar anchorages under monotonically increasing
loads, the complexity of the problem persists: Position of the bar at casting, bar
profile (rib geometry), transfer length, material properties, direct or incidental con-
finement of the surrounding concrete, and geometric boundary conditions all seem to
not only have a profound influence on the shape and coordinates of the load defor-
mation response, but also determine the mode of failure (splitting, pullout, or combi-
nation of the two modes).
Depending on the type of idealization and simplifying assumptions used in
modeling the load deformation envelope for bond, published models can be classified
as follows:
(a) mechanistic vs. empirical models
(b) global versus localized models
(c) splitting vs. pullout models
(d) numerical or hand-calculation models
CODE APPROACHES AND MODELS
The CEB design model and all ACI-related models are global, hand-calcula-
tion approaches: the influence of bond is evaluated by considering the free body
equilibrium of a segment of the anchored reinforcing bar (Fig. 1). The critical
section is at one end of the segment, at which point the applied bar stress is known. A
distribution of bond stresses along the segment is postulated based on experimental
results. The magnitude of bond stress is obtained from local bond stress - slip rela-
tionships that are often referred to as constitutive laws for bond. Note that this term is
necessarily used loosely in this context, as bond is not an intrinsic physical property
but the mechanical response of the bar - concrete assembly under certain specific
boundary conditions. The force in the other end of the bar segment is obtained
directly from equilibrium considerations. If the characteristic stress-strain relation-
ships of steel have a positive definite tangent stiffness throughout the strain range,
then it also is possible to obîain the deformations and the distribution of stresses along
the anchored bars.
(1) Models Adopted bv AC1 Committees 318 and 408:
Various AC1 committees (408, 318) have adopted several simplified versions
of the above model over the years. In all cases the length of the segment considered
involves the entire anchorage length, Ld, whereas the known stress input at the critical
section is usually the yield or ultimate stress capacity of the bar (Ld corresponds to the
length required for the stress in the bar to become zero at the unloaded end). The
intensity of bond stresses ub is taken as constant along the anchored length, set equal
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Ob62149 0 5 4 2 5 7 b L T 7 m
to an allowable maximum value multiple of the square root of f'c in order to acknowl-
edge the fact that bond failure is associated with tension failure of concrete in the
principal directions. Note that by definition, bond stress is proportional to the rate of
change of bar stress along the anchored length, therefore as a variable it has only
local physical significance. Therefore the assumption of constant bond stress distri-
bution is a simplifying idealization adopted for the sake of mathematical conve-
nience. Experimental evidence suggests that whereas this assumption leads to accept-
able results for normal weight concrete and uncoated bars, it cannot adequately model
the behavior of epoxy coated bars, or of bars anchored in high strength concrete, (In
these cases the bond stress distribution appears to be highly nonlinear with peak
intensity at the loaded end of the bar.)
The apparent advantage of the AC1 approaches is two fold. First, despite the
overall empiricism inherent in the assumptions made, there is an underlying physical
model, which explicitly establishes equilibrium in order to define the required Ld (in
design) or in order to define allowable values for (from experiments, where Ld is
known). Second, the resulting expressions are simple enough for routine hand
calculations. The obvious disadvantages of this approach are, first in the underlying
assumptions of constant bond stress, which, as was mentioned in the proceeding, are
unrealistic in the case of high strength concrete or epoxy coated steel. Secondly, the
model provides no detailed information about the behavior of the bar under load;
rather, the result is a binary answer as to the adequacy of the anchorage length for the
development of the yield strength of the reinforcement. In a performance driven
design framework it is necessary to predict structural behavior at specified load
demands. Here it is not sufficient to know whether the anchorage length can or
cannot develop the bar force capacity. For example, if the member is stressed up to
the yield point or slightly beyond yield, then the bar slip relative to concrete will be
limited and perhaps not important. In contrast, if the system is subjected to large
inelastic deformations, then the demands on the bar and surrounding concrete will be
such that significant bar slip may develop. This slippage will result in reduced
apparent stiffness of the member and reduced energy dissipation capacity in the
overall system. The assumption of perfect bond leads to higher estimates of the
structural stiffness and to lower displacements than the actual values, the more so for
systems undergoing cyclic load reversals. Sozen (1974) observed that the plastic
rotation of a concrete joint may be doubled due to bond slip. Because this amount of
increase in rotation affects significantly the story drift, it is important in design to
limit bar slip to small values. In the AC1 408 report (1992), it is recommended that
bond stresses be limited to about 80% of the ultimate values when designing for
cyclic loads. The objective of this recommendation is to limit the demand on the bars
- and consequently the intensity of bond stress - to values below the levels that would
cause deterioration of the surrounding concrete.
The AC1 Desien Equations for Bond: Over the recent years, AC1 design
expressions have progressively evolved from simple formulae to the more complex
but also more accurate expressions of today. The process was primarily driven by an
ever-expanding database of published experiments. Typically, best-fit lines have
been used to describe the relationship between test parameters and performance.
However the procedure was packaged, the resulting design expressions were intended
to ensure (based on the experimental evidence) that the anchorage length was suffi-
ciently long to fully develop the bar nominal yield force, while maintaining the bond
stresses below allowable limits. The reason why converging to the design expressions
of today has taken several iterations is that it has been shown over time, that the
experimental values are sensitive to specimen shape and support conditions during
testing. The perceived significance of other conditions, such as position at casting,
has never been properly linked to physical properties of concrete such as porosity
and composition. With such vital information missing from the database, and with the
ongoing emergence of new qualities of concrete materials, the relative significance of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542577 033 M
this parameter on the anchorage length is continually changing. An added layer of

complication in the process of deriving design expressions from a collection of tests
has been related to the mode of failure of the tests considered; note that whereas the
most challenging aspect of bond design is to protect against splitting failures, until
recently the majority of available tests in the literature had ended as puiiout failures.
Uniform specifications for designing and testing of bond specimens have been
proposed only recently by the research community.
Early expressions for bond simply divided the yield force by the bond area
surrounding the bar:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
in the AC1 318-56 Building Code, the limiting bond stress was taken as 0.10
f, for bottom deformed bars and 0.07 fCfor top bars, with an upper limit on cylinder
strength of 3500 psi. Thus an average bond stress of 350 psi (2.41 m a ) was as-
sumed. Ferguson (1965) recommended the use of even lower bond stresses than
permitted by AC1 318 in applications of low cover (Le., a limit of 0.05 f, for smaller
sized bars and 0.04 f, for No.9, 10 and 11 bars). For Grade 60 (420 MPa) No. 8
bar embedded in 4000 psi (30 MPa) concrete, a development length of 30 inches
(762 mm) would be required by this formula at a bar stress of 24 ksi (170 MPa) - the
typical value employed in allowable stress design in 1956. Thus, here was the source
of the 30 bar diameter rule of thumb used for many years in concrete design and
construction in North America.
This method for computing development lengths was used until the early
seventies when strength design replaced the allowable stress design. A primary study
that drove the development of improved equations was the work of Orangun, Jirsa and
Breen (1975, 1977). Design was based on the following empirical relationship:
A f
.Jrc U.S. customary units (2)
'1
U,, = 0.1 +0.25- C +4.15 db
db
-+L,
Af
tryt
41.52cdb
1& SI. Units (3)
The above equation models bond strength as a linear function of bar diame-
ter. The term 3C/db reflects the confining influence of cover and the negative effects
of close bar spacing on bond. The square root power of f, is used in order to
indicate that bond strength is a function of the tensile capacity of concrete and
increases at a slower rate than cylinder compressive strength. Passive confinement is
accounted for in a separate tem that is proportional to the confining steel area. After
simplification of terms, and by ignoring the confinement term, the above equation
may be written in terms of the required development length:
0.04Abf
L, = U.S. Customary Units (4)
M Obbi2949 0542578 T 7 T
0.0% f
L = Y S.I. Units
d
with an upper bound on bond stress of 625 psi (4.31 MPa) using AC1 318 limits.
Similar work by AC1 Committee 408 has led to the following equation for the
*
required development length of Grade 420 (60 ksi) reinforcement:
5500A
-
U.S. Customary Units (7)
Ldb -
where K is the smaller of
(a) 0.5db + C, + Ku or
(b) 0.5db + C,
I
+ - ; but no larger than 3d,
In the AC1 318-95 Building Code, the bond expressions have been reevaluat-
ed and restated in terms of bar diameter. The basic equation is,
( aB
L =by
For standard configuration, this equation reduces to
for No. 7 bars and larger with at least one bar diameter of cover, c, and minimum
stirrups provided, Ktp The coefficient in the denominator is 25 for smaller bars.
Using an upper bound of 10,000 psi (69 MPa) for the compressive strength of
concrete, the larger bar series has a required development length of 30 bar diameters.
This corresponds to a bond stress of 500 psi (3.45 MPa).
From the preceding review it is evident that the AC1 philosophy over the years
has been consistent. The objective has been to limit bond stresses to values in the
range of 500 - 600 psi (3.45 - 4.14 MPa), which are considered conservative based
on experience from experimental studies. For different configurations in terms of
bar spacing, cover or confining steel, adjustment factors are used to represent the
influence of these parameters on the required development length. For example, the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2949 0542579 906 W_ _
Km term in the AC1 318-95 equation and the associated cover terms represent these
factors. This approach was calibrated with research results and was originally devel-
oped by the AC1 Committee 408 (1990).
In curve-fitting the available database of bond tests, the AC1 408 approach has
+
been to include a factor in the analysis to account for localized uncertainties in the
+
material properties and configuration. A value of 0.80 in the denominator is used
to increase the required development length. On the other hand, the AC1 318 ap-
+
proach has been to avoid compounding the factors that appear in the bond expres-
sion and in flexure, shear or other applications. For this reason, the 318 expressions
have a factor close to 1.0. It was mentioned in the introduction that an explicit
relationship for bond stress vs. slip is not explicitly addressed in the AC1 documents
relating to bond. This is justified, since the overall approach of the AC1 philosophy is
to provide rules to preclude significant amounts of plasticity; in this range of re-
sponse, detailed mechanistic models are not essential. Elaborate analytical models for
bond have been developed by the research community, but are mostly used in the
study of problems involving a significant degree of nonlinearity, and fall beyond the
scope of the AC1 framework of design.
(2) The CEB Approach in Modelinp the Problem of Bond:
The CEE! Model Code (1990) has taken a somewhat different approach to
design: it is required that performance be checked within defined limit states. Two
primary limit states are considered, the Service Limit State (SU)and the Ultimate
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Limit State (ULS). The SLS represents a comparable state to that implied by the
serviceability checks within the AC1 318 Building Code. Straightforward design rules
are provided; the design objective is to ensure minimum levels of performance for
crack control, deflections and other items.
The ULS represents a plastic limit state where the structure and its components
are assumed to undergo significant inelastic deformations. Additional checks are
needed to ensure the design satisfies pertinent performance and capacity require-
ments. The use of computer-based analysis, and in particular the Finite Element
Method, is a hallmark of this approach.
A significant example of the differences between the SLS and ULS is that of
moment redistribution. When considering the ULS, the Model Code allows up to
50% of the negative moment to be redistributed to the positive part of the moment
diagram. Such levels of redistribution are not allowed by the AC1 318 code, where
the maximum redistribution is 10%.
The moment redistribution provisions have an important corollary and that is
the basic performance of the reinforcing steel. European reinforcing steels are
generally comparable to ASTM A 706 steel. They are weldable, low-alloy steels of
good quality and material traceability. The significant difference is the relative lack
of strain hardening in these Euro-steels as compared to U.S. steels. For example, a
Class A steel with a 500 Mpa nominal yield has a required ratio of tensile strength to
yield strength of 1.08. Within ASTM-A 706 it is required that the ratio of ultimate to
yield strength be a minimum of 1.25. From a design perspective this means that a
member forming a plastic hinge will have a relatively constant amount of moment
capacity as the hinge is loaded and rotation occurs. In a similar situation but using
US. steel the moment capacity will increase substantially as the steel hardens. The
implication is that the increase in capacity will provide additional strength when
needed under severe demands. In addition, considerations prevail relating to in-
creased need for confinement as compared to the fist onset of yielding as the internal
loads are changing within the member. Therefore, the different amounts of moment
redistribution permitted by the two codes complement the relative characteristics of
E 0 b b 2 9 4 7 0542580 628 m
the reinforcing steels considered in the two cases.

In practical applications forces and moments within the structure may be
computed either by nonlinear analysis, linear analysis, linear analysis with redistribu-
tion, or by plastic analysis. Each of these options has its own set of requirements and
models that must be followed. It is not the intent of this paper to restate the entire
Model Code; however it is evident that the Model Code is compatible with many
alternative approaches of analysis and design.
Compared to the AC1 approaches, another source of difference is the applica-
tion of factors both to increase the level of loads on the structure for design (load
factors) but also the reduction of material strengths by other factors. The design
strength of reinforcing steel is defined as
where the value of the partial factor ys is 1.15 for sustained loads. In the specific case
of bond and development, similar comments can be made. The CEB requires that the
designer check the bond stress to ensure that levels are not excessive.
CEB Desion Rules: The CEB defines the basic length to transfer the yield
force of a bar as
d f
L =byd
4fbd
where db is the bar diameter in mm, and fyd is the bar design yield force. Term fbd
represents the design bond stress of concrete, given by
fbd = qlq2q3fl'd (12)

where q i is a geometry factor taken as 2.25 for ribbed bars, q 2 is an orientation factor
for bond (a value of 1.0 is used in most cases). q 3 is a bar size factor set at 1.0 for 32
mm and smaller bars and taken as (132 - db)/iûû for larger bars. The term f t i is the
design tensile strength of concrete, defined as the characteristic tensile strength
divided by 1.50.
The design anchorage length is determined as
L = ala2a3a4a5LdAs,cai ~ L
d, net A d, min
s,eff
In the above, factors aiadjust the length based on:

i = 1: Accounts for the effect of bar form, for straight bars = 1.0
i = 2: Accounts for the presence of welded transverse bars = 0.7
i = 3: Accounts for confinement provided by the concrete cover:
For straight bars = 0.7 2

{
1 -0.15[~~-d,J
db 1' ''O
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542583: 564.m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
i = 4: Accounts for confinement provided by traverse steel:
where K = 0.1 for bars in stimp comers and 0.05 for bars away from
stirrup comer
i = 5: Effect of pressure perpendicular to the splitting plane
For high bond bars, the product a3 a4 a5 = 0.7 as a limit. For a typical 25
mm diameter bar in a 30 MPa concrete, analogous to a No. 8 U.S.bar in a 4000 psi
concrete, the required development length would be about 780 mm or 31 inches.
An attractive feature of the CEB approach is that it supplies the designer with
a simplified bond model that may be used in practical applications. The model is
based on research results and uses a nonlinear law to relate the applied bond stress, 2,
to bar slippage, s. The model is illustrated in Fig. 2. The characteristic values SI, s2
and s3 are based on the state of confinement of the concrete and the quality of bond
that is thought to be present.
The four segments of the relationship are defined as follows:
z =f ~ fors3 a s (17)
Terms for these equations are defied in Table 1. The parameter z max
represents the bond strength. Once the strength is attained, the bar slips at constant
stress up to a value of slip, s2, of 0.6 mm for unconfined concrete and 3.0 mm for
confined concrete. Beyond that point the bond resistance decays with increasing bar
slip and it reaches a residual value of z f at a slip value s3 of approximately 2.5 mm.
1O McCabe and Pantazopoulou
m Obb2949 0542582 4TO W
ANALYTICAL BOND MODELS

The heart of the issue in bond modeling is the bar slippage that occurs while
the system is being loaded. If compatibility of strain between bar and concrete is
maintained, then no relative movement occurs (Le., no bar slip). This behavior is not
found in actual members where there is always some bar slip as the reinforcement is
loaded. Moreover, the assumption of perfect bond can be unconservative in analysis.
Some shear failures are in fact bond failures that cause a reduction of shear capacity
as a secondary effect (McCabe 1997). From a practical viewpoint, it is of interest to
determine, for a given reinforcing arrangement, the amount of bond-slip likely to
develop and its effect on performance. Note that if the slippage is too great then
stiffness is reduced causing large member displacements. This behavior has been
welt-documented in experimental studies of monotonic and cyclically loaded struc-
tures and members. In the case of monotonic loading, local bond-slip relationships
are difficult to measure experimentally. Results are generally reported as force at the
loaded end vs. slip at either the loaded or unloaded end. Darwin et al. (1992) report-
ed that the amount of actual slip measured under load was significant, even at low bar
stress levels. Adhesion of the bar to concrete is seen to be destroyed nearly immedi-
ately. Following loss of adhesion, the amount of slip increases with increasing bar
force in a nonlinear manner leading to complete loss of resistance to slippage at about
0.002 inches (0.05 mm) for a 25 mm diameter bar. The behavior noted is specimen
and configuration dependent with changes in confinement, stirrups and cover, and bar
size having large effect on the amount of slippage and the slope of the curve. Thus,
the actual bar slip vs. bar stress relationship is complex and one that is quite difficult
to characterize experimentally in a manner that can be readily used in design. Rather
the information is of more interest to researchers.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
A traditional approach taken in modeling the behavior is to use experimental

data and represent the experimental trends by rules. These empirical approaches are
very practical and may be the only workable model for large scale computations
where complete structural systems must be modeled. However, more robust analytical
models of the bar-concrete interface behavior have been proposed; these are derived
from basic principles and to a large extent they avoid the limitations implicit in
empirical models. Studies have included three primary approaches (AC1 408 1992):
(1) solve for bond performance based on an assumed set of equilibrium and compati-
bility equations; (2) assume a function to model bond-stress distribution along the
bar; and (3) use non-associated plasticity theory or fracture mechanics to model the
interface between the steel and concrete where the bond develops.
Models that fall in categories (1) and (2) represent the background of the
existing code methods and as such, have been implicitly described in the preceding
sections. Alternative modeling approaches that attempt to mathematically define the
mechanics of the bond are reviewed below:
la) Mechanics-Based Models for Bond--Recently, a number of experimental
and analytical studies have been presented, seeking a more fundamental definition of
the local bond slip relationships (Gambarova et al. 1989, Giuriani et al. 1991, Cox
and Henmann 1992, Rosati and Schumm 1992, Rostasy et al. 1987). Giuriani et al.
(1991) considered the development of an axially loaded bar, confined by transverse
stirrups and anchored in concrete which has already developed splitting cracks.
Provided that some reinforcement is crossing the splitting plane, bond action is still
possible past the development of splitting. Based on the work of Gambarova et al.
(1989), the local bond stress-slip relationship is expressed as a function of the crack
opening, w:
Obb2747 0542583 337 m
where,
and um,wthe maximum bond stress that can be reached for a given crack opening w;
um,0the maximum bond stress that can be reached if there is no splitting crack (w =
O); y1, yz, ß1 and & are coefficients determined from experimental data. For a given
crack opening w, the stress developing in the transverse reinforcement crossing the
crack is estimated as follows:
where the coefficients R, al, and ao are related to the assumed anchorage conditions
of the transverse reinforcement (Giurianni et al. 1991). Additional confining action,
such as that provided by the residual tensile strength of cracked concrete was also
accounted for in the model, and was expressed as a function of the crack opening and
the maximum aggregate size, d,:
where 00 the radial confining stress at w = O and k an empirical constant. The local
bond stress is also related to the total radial force produced by the anchored rein-
forcement for a given splitting crack opening:
where W is the total cross section of stirrup legs crossing the splitting plane divided by
the cross section of the principai bar &, and B is the total area of concrete in the
splitting plane divided by Ab. The nonlinear system of equations presented here is
solved numerically to establish the relationship between bond stress, crack width, siip,
and transverse confinement. The pitfall of the model is in the many empirical con-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stants required for the definition of the various stress components; these were ob-
tained from a series of specially designed experiments by Gambarova et al. (1989).
The merit of the model is in that it takes a fundamental approach to the mechanics of
bond, and therefore provides a crisp mathematical relationship between the important
design variables.
IObb2949 0542584 273 m
(b) A Plasticity Model for Bond--Cox and Hermann (1992) used an alterna-
tive analytical approach in order to derive local bond stress - slip relationships from
consideration of the mechanics of anchored bars. Key to this formulation is that
bond stress and slip are fully coupled with radial confinement stress and dilation. As
before, the objective was to produce a bar scale model that would be able to repro-
duce rib phenomena, while at the same time be usable in member scale computations.
In this study, a two degree of freedom non-associated plasticity model for bond was
formulated based on the experimental results of Malvar (1991) and Gambarova et al.
(1989). Plasticity was chosen as the framework for studying bond because of the
existence of a contact interface between bar and concrete, with Coulomb-like friction-
al behavior.
The model relates local slip and radial dilation to bond stress and radial
confinement stress in a finite thickness region around the bar- so called process zone.
The generalized stress and strain vectors are defined as,
where z and CT are the shear and normal stresses at the interface, db the bar diameter,
and 6, and 6, represent the slip and the radial dilation, respectively (i.e., these are
relative displacement quantities between the inner and the outer surface of the inter-
face zone). Using experimental data, the following elastic moduli were derived,
relating the stress vector Q with the vector of relative displacements q:
o.1 - 0.012~gn(S~)H(-6~-
-E)
- O.O12~gn(6t>H(-6~-- E ) 0.04
In the above, E, is the elastic modulus of concrete, H is the heaviside step

function, and E denotes numerical zero. The physical interpretation of the terms in
the property matrix is as follows: The diagonal terms are obtained either from experi-
mental measurements or from calibration studies of anchorage tests. The off-diago-
na1 terms are small, reflecting that prior to the occurrence of plastic deformation,
negligible radial stress will arise if shear deformation is imposed at the interface
(similar reasoning for the other entry). As used, the heaviside function will set the
off-diagonal terms to zero in the case of no-slip (because in that case the problem
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
reduces to one of radial contraction, with no resultant shear force along the bar axis).
The sgn function adjusts the direction of the stress vector to the direction of applied
slip, since the stress state considered is a contact problem.
The strength of the interface is regulated by a yield surface:
and a non-associated flow rule:
[qPIT = <h>[sgn(z), 91 (25)

In these equations, ft represents the tensile strength of concrete, C(d) and p(d)
are the isotropic and kinematic hardening functions respectively, g is the rate of
OLL12949 0542585 LOT m Tribute to Peter Gergely 13
plastic dilation with respect to plastic slip, and h is the loading index. Parameters C,
We, and p are functions of a damage measure d, defined as the ratio (&P/sr)< 1, where
s, the rib spacing (the upper limit 1 was enforced based on the results of Eligehausen
et al. (1983) which show that bond is completely failed and only the frictional mis-
tance remains when slip is equal to rib spacing). The selected flow rule has the
following parametric behavior: For small values of the normai stress 6,failure is
dominated by longitudinal and transverse cracking and the rib is able to slide along
the interface surface causing small amount of crushing and large dilation. For large
values of the radial confining pressure failure is dominated by local crushing and by
shearing of the concrete between ribs, and in this case, the amount of dilation is
limited. The merit of this formulation is that it only requires two input property
variables for concrete (Le., tensile strength and elastic modulus), and two geometric
properties of the reinforcement (diameter and rib spacing). The model was used
successfully to correlate several independent experimental studies.
[c) Detailed Modeling of the Bar Profile - Fracture Mechanics Amroach
Ideally, many of the limitations and influences on the observed bond behavior which
are caused by the experimental setup could be avoided or quantified if a detailed
mathematical model of an anchored bar embedded in concrete could be used to
supplement the experimental studies. A complete model accounts for the exact
geometry of the bar surface (such as spacing and height of the ribs), the constitutive
properties of concrete under triaxial states of stress, and represents accurately the
boundary conditions of both concrete and steel in all directions. This is possible only
with three dimensional finite element idealizations of the anchored length of the
embedded bar. In such analyses, local bond - slip relationships are no longer input
properties of the model, but rather, they emerge from the computations as a result of
prescribed geometry, material properties, and global boundary conditions. Although
the computational demands of the finite element models of the type described in this
section render them impractical for large scale modeling of complete siructural
systems, yet, results from analysis of anchored bar examples could be particularly
useful in deriving characteristic bond stress - slip relationships for global models, as
was done in the past with experimental results.
Several 3-D finite element models have been published in recent years,
including work by Ingraffea et al. (1983), Ozbolt and Eligehausen (1992), Pijaudier-
Cabot et al. (1991), Mainz et al. (1992), Rots (1989), Eligehausen (1989), Brown et
al. (1993). The most difficult aspect in the development of these models is in the
definition of constitutive relations for the stress-strain response of plain concrete. To
reduce the amount of computational effort required for a solution, most published
finite element models consider the nonlinearity in the material behavior only for a
narrow zone of elements which includes the contact region between reinforcement
and concrete. Stress concentrations, which arise at the interface of the ribs with the
interlocking concrete, cause localized damage and irreversible energy absorption.
This region is referred to in the literature as interface zone, fictitious interface
(Pijaudier-Cabot et al. (1991), Rots (1989), Eligehausen (1989)), or process zone
(Ingraffea et al. (1983), Malvar (1991), Cox and Hermann (1992)) and is modeled
by nonlinear elements with special constitutive properties. It is common practice to
model concrete as linear elastic outside the interface zone.
Because the process of stress transfer from rib to concrete invariably occurs
by cracking and crushing of the surrounding concrete, it has been argued that model-
ing bond warrants a fracture mechanics approach (Rots (1989), and Eligehausen
(1989). Maim et al. (1992), Brown et al. (1993)). Formulation of the models in this
framework classifies the deformation mechanisms that contribute to bond-slip into
four types, i.e., elastic deformations, secondary cracking (cone shaped), longitudinal
splitting cracking (radial), and crushing in front of the ribs (Ingraffea et al. (1983),
Rots (1989)). Of those, it is commonly assumed that secondary and longitudinal
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Ob62949 0542586 04b m
14 McCabe and Pantazopouiou
cracking are the most important contributions. These are closely related to the
occurrence of fracture (cracks). In numerical analysis, they can be accounted for by
(1) lumping all contributions to the fictitious interface, or (2) by modeling the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
individual contributions in detail (in a narrow layer of interface elements which is

treated as a continuum with all the failure mechanisms that go with this assumption,
i.e., crushing, softening, cracking, etc.). The latter approach is computationally very
intensive and hence not appropriate for practical applications. Research by Ozbolt
and Eligehausen using a micro plane model offers the promise of solving this type of
problem with a fracture mechanics theory but in a more efficient manner since the
actual characteristics of the bars do not need to be modeled (Ozbolt and Eligehausen,
1992).
Through such analytical formulations, bond strength may be studied as a
function of rib height, rib shape, concrete cover, lead length, embedded length, and
transverse reinforcement. While the effort in researching models to predict bond
performance has been significant, the AC1 318 Building Code and other design codes
used in the US have not yet considered such analysis options and instead rely on the
experimentally based anchorage equations to determine adequacy of embedment.
No official endorsement of any particular approach in modeling bond and bar
slippage is provided.
INFLUENCE OF BOND ON STRUCTURAL PERFORMANCE
Determining the performance of reinforced concrete structures under load is
of great practical importance, particularly when considering dynamic load conditions
such as strong ground motion or explosion. From a design perspective it is desirable
that the structure survives such extreme load conditions with no loss of strength
although it may undergo significant displacements and rotations at members and
connections. To achieve this objective the designer needs pertinent computational
and modeling tools that will enable determining the characteristics of structural
response during loading.
Consider for example the familiar practice of moment redistribution. It can
be an attractive design alternative to move a portion of the negative moments from
support locations and to relocate this demand into the positive moment portion of the
span. During response to lateral load, flexural members seek moment capacity at the
negative moment locations (supports), due to large rotational demands at these points.
If insufficient capacity is available, a plastic hinge will form at the corresponding
member end, whereas the residual load not accounted for is carried by positive
moments. It was mentioned earlier that at present, AC1 practice limits this redistribu-
tion to about lo%, while CEB allows designers to redistribute up to 50% of the total
negative moment demand. However, the designer is required to verify that the system
can indeed support this level of demand along with the resulting rotations of the
hinge.
Observations on Bond Modeling
From the foregoing discussion it is evident that both AC1 and CEB have
provided a comprehensive set of simple rules that govern the design of reinforced
concrete structural systems. These rules only require hand calculations and define the
required anchorage lengths for development of the ultimate bar force. The objective
of this requirement is to ensure that given the detailing and the section geometry, the
engagement of the bar deformations in concrete are sufficient to anchor the bar when
it develops its full force capacity.
In determining the anchorage status the basic question is stated in the form of
empirical equations developed over the years from experimental studies of bond.
9 0662949 0542587 T82 m
Whereas the equations are simple, the behavior certainly is not; however, the equations
give the sense that the reinforcing bars truly are anchored in every sense of the word
(i.e., there is no movement of the bar as it supports load). Though the concept of a
perfectly bonded bar in concrete is convenient in design, it is not representative of the
true behavior of the bar under load. As the load increases bond-slip occurs, which
may become large enough to push the bar deformations (ribs) into the surrounding
material resulting in dilation of concrete and a resulting tensile stress field. If the
slippage continues, then at some point fracture initiates as a result of stress concentra-
tion near the first loaded rib. The fracture energy is released instantaneously (brittle
response) whereas the fracture surface grows through the thickness and along the bar.
The splitting crack develops along a minimum energy path in concrete, from the bar
surface to the nearest free surface of concrete (i.e., through the cover).
Thus, the checks that are made by the designer using either the AC1 or CEB
code equations are an effort to preclude large enough bond slippage to cause a
splitting fracture of the cover over the reinforcement. The code equations are an
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
attempt to deter this slippage by providing a long enough bar embedment, or large
enough confinement through cover or transverse steel.
Experience has shown that these equations lead to reasonable designs and that
research and design expertise has provided a framework that generally provides a
correct assessment of the state of stress and slippage of the bars being developed.
However, several issues are becoming increasingly relevant. One is that much of the
data that were used in deriving these equations were obtained from specimens of low
to moderate strength concrete whereas high strength concrete is more widely used
today. Bond in high strength concrete has been shown to be quite different in that
less redistribution of load occurs in that case from one deformation (rib) to another.
The result is less ductile bond performance and a greater tendency to "unzip" the bar
from the concrete as the bond failure begins. An other issue is that many new rein-
forcing systems are becoming important such as welded wire, high strength bars,
headed bars; all these products behave differently from the classical deformed bars
(350-520 MPa yield strength) that were used in the past for much of the supporting
experimental work. Today we are essentially beginning to use a new set of materials
that behave differently than those in use even as recently as 10 years ago. Because of
the nature of the design problem with such new materials, accurate determination of
bond behavior becomes even more important than before, since a localized bond
failure would be more likely to result in a large scale problem in a member, connec-
tion or in the structure as a whole.
It also has become apparent that the existing equations are truly only applica-
ble when designing new structures under conventional loads. The checks provided by
the code equations may not be sufficient in assessing existing structures that have
been significantly modified for some new use or subjected to an unforeseen overload
such as from blast or earthquake. In a case such as this there are no simple design
alternatives. One option is to apply the Finite Element Method (FEM) or other
computer-based structural models in order to assess how the system behaves under
load. This type of analysis can provide essential information as to how the system
moves and redistributes load. A key ingredient is modeling of the nonlinearity that
occurs as yielding of the steel and cracking of the concrete occurs. The application
of nonlinear FEM has been the source of ongoing research and additional work is
needed to provide straightforward, practical tools for use by designers. However, it is
indeed becoming more and more apparent that nonlinear FEM may be the best
practical alternative when evaluating structures that have been overloaded, or loaded
in a manner unforeseen in the original design.
The accuracy of the overall analysis lies in modeling of the phenomena at the
steel-to-concrete interface. The assumption of perfect bond, in essence the lack of a
IObb29Y9 0542588 919
bond model, is an easy way to model the structure but produces unconservative
results; in that case the steel-concrete system is assumed to work at a perfect efficiency
transmitting the strains of the steel directly into the concrete. A major problem is lack
of consensus of how exactly to model the bond. The general view has been that any
individual bond model is capable of providing good results so long as the model has
been calibrated against experimental data. Thus, given a set of data the model may be
tuned to match these values. While this approach may be an acceptable step in
process of developing new modeling techniques, it is not proper for design; here
efficient methods for determining conservative estimates of anchorage capacities are
required.
Therefore, the assessment of structural performance using nonlinear FEM will
depend in part on removing much of the guesswork that exists at the present time in
modeling of bond. This need is one that can be filled by organizations like AC1 and
CEB. There is a clear need for guidance and for clear concise modeling rules for
bond. This need will only become greater as more and more structural engineers turn
to nonlinear analysis methods to determine the adequacy of members and structures
under load. At the present time, designers may choose between utilizing the present
CEB bond model which is over a decade old and developing case-specific models
consistent with their own interpretation of the literature. Those using the AC1 guide-
lines for bond will find no information as to bond modeling in the Building Code and
will find only passing reference to the concept in the various bond reports of AC1
Committee 408.
Oversumlv of Bond Resistance or How Much Bond is Too Much?
The possibility of oversupply of bond resistance has become recently an
important consideration of performance based design. In the past, it was generally
assumed that there never can be too much bond and that improvements in bond
through changes in the bar geometry or other factors, will only improve the perfor-
mance of the structure. However, this is not supported by either theory or experi-
ment: note that if bond is improved beyond the ability of the member to sustain it, the
mode of failure will change from pullout to splitting. In such circumstances there is
too much pullout resistance for the member as detailed. To remedy this situation
additional transverse reinforcement would be necessary in order to confine the bars
and thereby increase the splitting bond capacity. This may mean additional cost that
must be weighed against the gains in performance. It may indeed be the case that the
additional bond is not needed and that the structural performance is acceptable if the
bar slips, provided that the slippage is controlled and accounted for in the design.
With regards to the performance of plastic hinges, it has been suggested that
increases in bond may lead to decreases in plastic hinge rotation (referred to in the
literature as rotational capacity). Load redistribution in structures subjected to severe
excitations occurs because of ductile performance of those hinges. Thus, any loss of
ductility resulting from increases in bond may be counterproductive and may lead to
premature failure. This topic is in need of systematic investigation, because of the
many types of reinforcing steels available . Existing data is not sufficient to determine
whether with North American and European steels there is a potential upper limit to
the bond that is advisable with all steels or if a limit exists for only certain types of
steels, if at all.
CONCLUSIONS
The conclusion is that we are moving into a new era of practice. The designer
is faced with using many new tools for the design and construction of structures.
These tools include new materials such as high strength concrete and new types of
reinforcing systems, both of which may be outside the bounds of the data used to
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662749 0542589~855D
develop the existing bond relationships. Moreover, new analysis tools principally
include computer-based methods. While linear elastic models used to determine
behavior under nomial loads will continue to represent the vast majority of applica-
tions, it will become more and more commonplace for practicing engineers to rely on
nonlinear methods for assessing structurai behavior. A key ingredient will be the
accurate modeling of all aspects of structural response, including the very real ehe-
nomena of bond slip.
It has been shown in a number of studies that the assumption of perfect bond
in reinforced concrete members subjected to large demands is not an accurate as-
sumption. Moreover, it may indeed be unconservative as well. Thus, the analysis of
reinforced concrete systems under large demands requires the type of model provid-
ed within the CEB framework to accurately determine system response. It is incum-
bent on the standards and code writing organizations to provide a unified approach
for bond modeling in such applications to ensure accuracy and safety in the analysis.
The question can be asked whether any form of bond modeling in large scale
analysis is practical for design. At this time, the practicality is limited to those situa-
tions where detailed determination of post yield behavior is needed. Also, the practi-
cality is limited by the analysis platform that the model is analyzed on as well as the
software. Certainly, using analytical bond models in structural design is not a practi-
cal alternative for typical applications. It still is a primary element in the determina-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
tion of displacements and drift. However through ongoing research, it is likely that
the application of bond models to performance determination will become a reason-
able altemative for designers even for typical structures. This change is due in part to
research and in part to improvements in computing hardware. The future will no
doubt include design and analysis software that will enable designers to predict -
accurately - bond stresses and bar- slip in an easy and straightforward manner.
REFERENCES
AC1 Committee 318 (1992). Building Code for Reinforced Concrete (AC1 318-89)
(Revised 1992) and Commentary - AC1 318R-89, American Concrete Institute,
Farmington Hills, MI, 353 pp.
AC1 Committee 318 (1995). Building Code Requirements for Structural Concrete
(AC1 318-95) and Commentary - AC1 318R-95, American Concrete Institute,
Farmington Hills, MI, 369 pp.
AC1 Committee 408 (1990). Suggested Development, Splice, and Standard Hook
Provisions for Deformed Bars in Tension (AC1 408.1R-90), American Concrete
Institute, Farmington Hills, MI.
AC1 Committee 408 (1992). State of the Art Report on Bond under Cych$ Loads
(AC1 408.2R-92), American Concrete Institute, Farmington Hills, MI, 32 pp.
Brown C., Darwin D., and McCabe S., (1993). “Finite Element Fracture Analysis of
Steel Concrete Bond,” Research Report No. SM 36, The Univ. of Kansas Center for
Research, Lawrence, Kansas, 100 pp.
Committee Euro-Intemational du Beton (CEB) (1990). CEB-FZP Model Code,
Lausanne, Switzerland.
Cox J. V., and Hermann L., (1992). “Confinement Stress Dependent Bond Behavior,
Part Ií: A Two Degree of Freedom Plasticity Mode,” Proceedings, International
Conference, Bond in Concrete, CEB, Riga, Latvia, Vol. 3, pp. 11-11:11-20.
Eligehausen R. E., P.Popov, and V. W.Bertero, (1983). “Local Bond Stress-Slip
Relationships of Deformed Bars under Generalized Excitations,” Report No. UCB
/EERC-83/23, University of California Berkeley, 169 pp.
I0662949 0542590 577
Eligehausen R., (1989). “Anchorage to Concrete,” Chapter 13 in Fracture Mechanics

of Concrete Structures - From Theory to Applications, (pages 263-280).
Ferguson, P. M. and J. N. Thompson, (1965). “Development Length of High
Strength Reinforcing Bars,” AC1 Journal, V. 62, No. 1, Jan., pp. 71-94.
Gambarova P., Rosati G., and Zasso B., (1989). “Steel to Concrete Bond after Con-
crete Splitting: Test Results,” Materials and Structures, 22, pp. 35-47.
Gambarova P., Rosati G., and Zasso B., (1989). “Steel to Concrete Bond after Con-
crete Splitting: Constitutive Laws and Interface Deterioration,” Materials und Struc-
tures, 22, pp. 347-356.
Giuriani E., Plizzari G., and Schumm C., (1991). “Role of Stirrups and Residual
Tensile Strength of Cracked Concrete on Bond,” J. of Structural Engineering, ASCE,
117(1), January, pp. 1-18.
Ingraffea A., Gerstle W., Gergely P., and Saouma V., (1984). “Fracture Mechanics of
Bond in Reinforced Concrete,” J. of Structural Engineering, ASCE, 110(4), April, pp.
871-890.
König G., and Tue N., (1993). “Method to Solve Bond Differential Equation System
in Prestressed Concrete,” J. of Structural Engineering, ASCE, 119(lo), October, pp.
2194-2206.
Mainz J., Stock1 S., and Kupfer H., (1992). “F.E. Calculations Concerning the Bond
Behavior of Deformed Bars in Concrete,” Proceedings, International Conference
Bond in Concrete, CEB, Riga, Latvia, Vol. 3, pp. 12-17:12-26.
Malvar L. J., (1992). “Bond of Reinforcement Under Controlled Confinement,” AC1
Materials J., 89(6), Nov.-Dec., pp. 593-601.
McCabe, S. L. (1997). “Size Effect of Reinforced Concrete Members under Shear,”
CEB Bulletin 237, Laussanne, Switzerland, pp. 171-183.
Orangun, C. O., J. O. Jirsa, and J. E. Breen (1975). “The Strength of Anchored Bars:
A Reevaluation of Test Data on Development Length and Splices,” Research Report
No. 15#-3F, Center for Highway Research, Univ. of Texas at Austin, Jan., 78 pp.
Orangun, C. O., J. O. Jirsa, and J. E. Breen (1977). “Reevaluation of Test Data on
Development Length and Splices,” ACI Journal, V. 74., No. 3, Mar., pp. 114-122.
Ozbolt J., and Eligehausen R., (1992). “Numerical Simulation of Cycling Bond-Slip
Behavior,” Proceedings, Intemational Conference, Bond in Concrete, CEB, Riga,
Latvia, Vol. 3, pp. 12-27:12-33.
Pijaudier-Cabot G., Mazars J., Pulikowski J., (1991). “Steel-Concrete Bond Analysis
with Nonlocal Continuous Damage,” J. of Structural Engineering, ASCE, 117(3),
Mar., pp. 862-882.
Rosati, G. and Schumm, C.,(1992). “Modeling of the Stress Intensification at a
Pointed Notch Tip under Tension and Bending,” Studi e Ricerche, Politecnico di
Milano, V. 13, pp. 241-259.
Rostasy F. S., and Scheuermann J., (1987). “Verbundverhalten einbetonierten
Betonrippenstahls bei extrem tiefer Temperatur,” Deutscher Ausschuss fur
Stahlbeton, H& 380, pp. 43-105.
Rots J., (1989). “Bond of Reinforcement,” Chapter 12 in Fracture Mechanics of
Concrete Structures - From Theory to Applications (pages 245-262), RILEM, (Editor:
L. Elfgren), Chapman and Hall, 407 pp.
Sozen, M. A. (1974). “Hysteresis in Structural Elements,” Applied Mechanics in
Earthquake Engineering, ASME AMD, Vol. 8, pp, 63-98.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obhi2!949 05425qI 403
TABLE 1-CEB BOND MODEL PARAMETER DEFINITIONS (MODEL CODE 1990)
Good All other Good All other

bond bond bond bond
conditions c011ditio11s conditions conditions
0.6 mm 0.6mm I .O mm I .O min

0.6 min 0.6 111111 3.0 mm 3.0 I1lRl
I .O min 2.5 inni Clcar rib Clwr rib
spacing spacing
0.4 0.4 0.4 0.4
2.OJL.i 1 .0JL 2.5dL.k Iq k r
0. I 5Tmi 0.15r,,;,, O.407,",, 0.40T,,,,,
'e
e
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
Iüb62949 0542592 3 4 T W
aY
.-c
E?
O
Y-
.-r
E
c
O
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
Obb2949 0542573 2ôb

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Bond and Splitting: A Vexing Question
by P. G. Gambarova, 6. P. Rosati and C. E. Schumm
Svnopsis: Splitting does always occur in some way prior to bond failure, in the
form of either partial splitting (quite often undetected) or full splitting, the latter
being the subject of several recent papers, owing to the importance of cover
splitting in RlC elements. Starting from the test results on fully-split specimens
(iike those by the authors on special specimens having a fabricated crack) it is
possible to formulate suitable bond stress-confinement stress relationships. These
models can be introduced into the limit-analysis models developed lately for the
description of partial splitting up to the onset of full splitting and bar pull-out
in short anchorages. In this way, a linkage between the bar-concrete pressure
(studied here through a limit-analysis elastic-cohesive model) and the bond stress
is established, in order to evaluate the ultimate bond capacity and to investigate
the transition from a splitting-type failure to a pull-out failure. At the same
time, such important topics as concrete tensile strength and fracture energy,
crack cohesion and localization, concrete cover and bar diameter, fiber content
and externa1 pressure can be incorporated into the model. A set of diagrams
showing the bond capacity and crack numberlopeninglpenetration versus
concrete cover is presented, and the design implications of both the theoretical
and experimental results are discussed.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Keywords: Bond (concrete to reinforcement); bond strength; reinforced concrete

structures; strength; splitting (cracks); test
23
ObbZ947 054Z595 059
24 Gambarova, Rosati and Schumm
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Pietro G . Gambarova has b e m u profirsor since 1981 ut Milan Universiiy of
Technology, where his .ccicnt(fic intí)re.ct.r urc cnncmtruted on reinforced
concrete nwchanrcJ, bond, high-perf¿irniunce cí)ncrm.\ and high-temperature
effects on cementitious compo;\ite.c; he is also active in the Task Group TC 2/5
(Bond Models) qf CEB Commission 2 (Marerial and Behaviour Modelling). Prof.
Gambarova is a graduate of Milan Univiirsiiy qf Technology and was a visiting
scholar at Northwestern University (Evanston, IL) in 1978 and 1982.
Gianpaolo Ros& has been an assistant professor since 1992 at Milan University
of Technology, where he rcceived both his M S and PhD degrees, the lauer with
a thesis on tension stiffening and crack cohesion; his main .field of research
includes bond, concrcte .fracture mechanic;\ and very high-strength concretes.
Dr. Rosati is a member of CEB Task Group TG 2/5 (Bond Models).
Cristiano E. Schumm studied structural engineering ut Milan University qf

Technology (MS and PhD degrees), whew he became un ussistunt prqfessor in
1993; his main field o f revearch includes punching in FRC slabs, limit-analysis
models for bond, and concrete,fracturc mechanics. Dr. Schumm is u member of
CEB TG 217 (Concrete Tensile Strongrh in R/C Design).
INTRODUCTION AND NATURE OF PROBLEM
Concrete splitting and concrete-steel bond are so deeply intertwined that bond
failure is always accompanied by a more or less extended splitting. The latter
makes bar piill-out easier [I], but goes often undetected because the radial
cracks remain limited to the concrete layer closest to the bar (Fig. 1).
In spite of its local nature, splitting affects the structural behavior from
the point of view of safety (structural strength and stiffness diminish owing to
splitting) and durability. Furthermore, splitting is destined to become even more
important as high-strength concrete (HSC, more brittle and prone to cracking)
enters the scene and is increasingly used in a variety of structures, not to speak
of epoxy-coated and non-metallic rebars, which can develop a larger wedging
action against the cover, owing to the lack of chemical adhesion (in both cases)
and to the high Poisson ratio (in the latter case).
Though through-splitting (often resisted by transverse reinforcement or
external confinement) is the most severe situation, partial splitting is definitely
more subtle and difficult to study, since inany aspects of concrete mechanics
come into play, such as concrete tensile strength and multiaxial behavior, crack
cohesion, concrete-bar friction, crack penetration (or extension) and opening,
and crack stability in the radial direction [2-91.
In this paper concrete splitting and bar pull-out are treated from the
viewpoints of both experimental investigation and inatliematical modelling. Since
both shear and compressive stresses develop at the bar-concrete interface, the
tests are instrumental in formulating a suitable bond stress-confinement stress
relationship, where the crack opening or the confinement are a predetermined
parameter. The test specimens are generally presplit or cast with a fabricated
crack (Fig. 2) [lo-121.
Q5425qb Tq5
Fig. 1 -Bond and splitting in RIC: (a) experimental evidence (moiré holography,
Baranek, IBBC-TNO, Delft, 1980), and (b) mathematical modelling (limit-analysis
models [2-4,6-8, 191)
(ci) crack width A,, w

bar slip 4 --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 2-Bond-confinement interaction: different test philosophies; short

anchorages: (a,b) single fabricated cracked plane (crack number n= 2), and (cl
multiple splitting; (a) constant crack width, and (b,c) constant confinement; (d)
long anchorage
Obb29Lt9 0542597 721 9
2 I0.4Mpa
3 17.2 MPa
4 24.IMPa
5 31.0Mpa
10
Bar slip ô (mm) Bar slip 6 (mm)
Fig. 3-Bond response (f, = 38-40 MPa; LI 0 = 3.50-3.75): (a) passive

confinement [71 (1.a pure pull-out tests: severe confinement, sleeve thickness
1O mm; 2.a splitting tests: moderate-severe confinement, spiral pitch 14,28 mm,
0 s = 6mm), and (b) active confinement [lo] (the different values refer to the
nominal pressure acting on the bar).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb29Li9 0542598 8bì3 m
The formulation of the bond-confinement relationship is a first, necessary
step in modelling the failure of a short anchorage. It is treated first in the paper,
as required also by the limit-analysis models still under development [4,6-81
which do not address explicitly the evaluation of the bond stress and focus their
atîention on bar-concrete pressure (Le., confinement stress on the bar owing to
the slipinduced wedging action).
In the second part of the paper, on the basis of a few fundamental
assumptions regarding the concrete (considered as a two- or three-phase
material, partial splitting included, Fig. lb), the cohesion of the radial cracks,
the static admissibility at crack tip and the kinematic admissibility of the crack
pattern (splitting cracks), the following quantities are calculated by maximizing
bar-concrete pressure with respect to crack penetration: (a) the maximum
pressure attainable at the bar-concrete interface, p-; (b) the "critical" crack
penetration, s,, or 4, (beyond which cracking becomes unstable in the radial
direction); and (c) the number and opening of the radial cracks, n and 8, (or w).
As a result, once the opening of the splitting cracks and the pressure along the
bar are known, the bond stress-confinement stress relationship allows for
evaluating the maximum bond stress (= bond strength or capacity, 73,for any
bar and concrete diameter, and for any applied external pressure.
Since the experimental bond-confinement relationships are governed by
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
crack opening in low-confinement conditions (large crack width), and by the
crushing and shearing-off of the concrete along the bars, under severe
confinement (small crack width), the limit-analysis models can describe both
"splitting failures" and "pull-out failures", once the above mentioned
relationships are introduced. This last issue is relevant not only to a better
understanding of the two fundamental failure modes of bond, but also to the
detailing in WC design, since - for instance - overreinforcing an anchorage
region in the transverse plane may be even detrimental to bond, owing to such
side-effects as bar congestion.
The limit-analysis models [6,8,12] are also instrumentai in describing
crack localization in the radiai direction, which is an important and still open
issue. As a matter of fact, bond efficiency depends on the width of the splitting
cracks, but the width is in turn a function of the crack number! Moreover, the
knowledge of the crack number is badly needed also as an input data for F.E.
modelling of tension stiffening and anchorages; previously adopted idealizations
have included a predefined number of splitting cracks [4,5,7], smeared cracks
without crack cohesion [16] and a "process zone", within the framework of
elastoplasticity [171.
Beside partial splitting and bond capacity, there is full splitting, when
bond capacity is guaranteed only by the transverse confinement, either active or
passive. Through cracks may be caused by the applied loads (generally in the
ultimate load situation) or by other factors, such as shrinkage and exposure to
high temperatures (which may occur also in the working load situation). .
However, only by knowing the bond stress/confinement stress relationship is it
possible to infer the ultimate capacity of a short anchorage from either the width
of the through cracks, or the maximum available confinement (which is strictly
related to the amount of transverse reinforcement [15]).
=
~
a Obb2949 0542599 7 T 4
TEST RESULTS ON BOND AND SPLI’ITING
Bond stress-confinement stress laws for cover splitting can be formulated if

suitable test results are available. Four basic parameters are at issue: pull-out
force, confinement action, bar slip and crack width (i.e. the width of the
splitting cracklcracks). Since test results have to be clear and possibly
unaffected by the structural setting, short anchorages should be preferred, to
ensure a well-defined bond behavior, with a uniform distribution of both the
static and kinematic parameters.
In the last fifteen years, different approaches have been adopted, ail
based on planar or axisymmetric specimens, generally reinforced with a single
bar.
In the planar specimens tested in Milan in the 1980s and early 1990s (see
Refs. in [12]), a pull-out force was applied to a single, short bar embedded in
a concrete thick plate (Figs. 2a and 4), whose forinwork had special separators,
to create a fabricated splitting crack. The 4 basic parameters were measured
directly; in most cases the crack width was kept constant by adjusting the
confinement throughout the test, and only in a few cases the confinement was
kept constant; in ail tests the confinement was controlled by means of special
mechanical devices.
Similar tests were carried out by Modena et al., on either ordinary or
epoxy-coated bars subjected to constant confinement (Fig. 2b, [i i]).
With regard to axisymmetric specimens, both active and passive
confinements were adopted. In the former case (Malvar [lo], Fig. 3b) the
cylindrical specimens were enclosed with a hydraulically-tightened ring
(confining ring, Fig. Lc), radial precracking was forced at the beginning of each
test, and the radial dilation was measured; the crack number was not prefixed
(2 to 4 cracks). In the latter case (Noghabai [7]), both outer steel sleeves and
inner steel spirals were used (Fig. 3a), and both normal-strength and high-
strength concretes were studied. In Noghabai’s tests the specimens were initially
unsplit, and the splitting cracks formed during the test (crack number: 2-5).
Going back to planar specimens, those tested by Plizzari et al. [is] are
unique since they allow studying the formation and propagation of splitting
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
along the reinforcement of a long anchorage (one or two bars). The confinement
was provided by 3 two-leg stirrups (Fig. 2d), and the moirè technique was used
for an early detection of the cracks.
Summing up, all tests have their strong and weak points: axisymmetric
tests with passive confinement are realistic and instrumental in checking the
soundness of mathematical models; the same holds for long anchorages. On the
other hand, only by testing precracked bidimensional specimens (as in Figs. 2a,
b,d) can the four basic parameters (crack opening, bar slip, bond stress and
confinement) be controlled separately, in order to investigate their interaction.
Of course, more complex tests with a more realistic bar arrangement are needed
too [13] for the validation of F.E.codes.
Finally, when considering the different test series, comparing their
results is always a difficult task, since critical assumptions are required. For
instance, in the case of Noghabai’s splitting tests (passive Confinement, Figs. 3a
and 8), the spirals were assumed to be at the onset of yielding, and the
Obb2949 0542b00 24b ~ m
extension of the radial cracks was put equal to 1.5 times the bar diameter, while
a linear distribution was adopted for the cohesive stresses.
BEHAVIORAL LAWS
The test results at constant crack opening [I23 (Figs. 5,6) show that the
mechanical characteristics of bond are unfavorably affected not only by crack
opening, but also by the diameter of the bars. Small-diameter bars have an edge
over medium- and large-diameter bars, with reference to bond strength, initial
bond stiffness, confinement-slip coupling and initial friction coefficient at the
bar-concrete interface. Nevertheless, the test results are far from exhaustive,
since - for instance - no attention has been devoted so far to the effects of bar
diameter under a given and constant Confinement (active confinement). On the
other hand, in the case of passive confinement (Fig. 3a, steel sleeve [7], severe
confinement and negligible splitting), the scanty experimental data show a
favorable effect of bar diameter.
Fig.4-Pull-out tests: specimens with a

single fabricated crack plane
(2 cracks) 1121; specimen size:
300 x300 x 60-90 mm
ao
Fig. 5-Pull-out tests at constant crack width 1121: bond stress-bar slip curves ( o test results
and interpolating curves, Eq. 41; 0 -14,18,24 mm; U 0 3.0,2.9,2.4 t10 = 4.3,5.0,3.8 -
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
- Obb29Yî 0542bOL L A 2 W
Fig. 6-Pull-o~t
tests at constant
0.3
o.2 --_-1.75 initial bond

20
stiffness; and (c)
(c) initial friction
o s io 15 20 o 5 io 1s 20 o s io 15 20 coefficient
(d,4)xio3
Bond streneth and stiffness
Owing to the more uniform contact between a small-diameter bar and the
concrete (which is quite close to a mortar in the transition zone froin the bar to
the undisturbed concrete), both the bond strength and the initial bond stiffness
are definitely higher for small-diameter bars (tests at constant crack width, Fig.
6).The following relationships fit the test results shown in Fig. 5 quite well:
r 1
where: k, = 0.907 + 0.5 D [1 - 0.81 DI, D = d,/d,, with d,, = 18 min,
W = 6,/db , a, = 30x10-' , a, = SOx10-3, a, = lSOx10-3,
D = 1/D = ddd, , a, = 2 . 6 1 ~ 1 0 .,~ a, = 2 0 0 ~ 1 0 - ~ ,
with 6, = crack opening (or width, Fig. 2),

6: = 6, - 6: = (total bar slip) - (initial free slip),
S)db = 0.57(6,/db): 6: is the rigid slip ensuing from crack opening (see
Refs. in [12]).
On the whole, the initial bond stiffness (Fig. 6b) seems to be more sensitive to
both crack opening and bar diaineter than bond strength (Fig. 6a).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The initial friction coefficient k, is a slowly-decreasingfunction of crack

opening, as shown by the linear regression in Fig. 6c (solid line). However, the
dispersion of the test results is so large that k, may be considered as
independent of crack opening (1.75 is the mean value and f 0.22 is the
standard deviation).
The linear regression on the test results [12] leads to the following
relationship:
k, = 1.25+0.48D (3)
As for 6, (opening or width of a splitting crack), assuming a uniform
distribution in the radial direction (Fig. 2) is a simplification, which disregards
the elastic deformation of the embedment.
Obb2949 0542602 OLï
Bond-slip relationship
The test results on small-, medium- and large-diameter bars [12] make it
possible to formulate a bond-stresshar-slip relationship, taking into account
Eqs.1 and 2 for bond strength and stiffness:
where:
m = 1.36 + 0.015 db (1 - zr).
In spite of its apparent complexity, Eq. 4 is easy to handle, once R,(W), r,,,/f,
and m have been evaluated, starting from the bar diameter and the crack
opening. Of course, in Eq. 4 the dimensionless crack opening W is a parameter
and not a current variable. Consequently, the use of Eq. 4 as a relationship
between the bond stress and the crack opening requires some care. In Fig. 5,
Eq. 4 is plotted for 3 bar diameters and 4 values of the dimensionless crack
opening: on the whole the fitting of test results is satisfactory.
Bond-confinement interaction
The test results (see Refs. in 1123) show that bond-confinement interaction is
characterized by closed envelopes which have the following properties:
- little, but not negligible, initial cohesion (in Fig. 7 for u=O), mostly
independent of crack opening (r0, Eq. 5);
- frictional behavior ensuing from bar wedging action and concrete restrained
dilatancy; the frictional behavior is represented by the mostly linear branch
following the initial cohesion; its slope is very close to the friction c0eff.k;
- extended damage in the concrete surrounding the bar, with little wedging
action and concrete dilatancy, leading eventually to the shearing-off of the
concrete keys between the steel lugs; such behavior is represented by the
flattening-out of the envelope;
- greater bond strength T , , ~for smaller bars, as already pointed out.
The above-mentioned behaviors (initial cohesion, dry friction and extended
damage) suggest a trilinear approximation of the bond-confinement envelopes
or a polynomial formulation:
with ~,lf, = 0.042, k, = 1.45, k,, = 1.26 + 11.10 (6,1úb) + 2190 @,Id$.
The polynomial approximation should be stopped when the bond stress reaches
its maximum.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542603 T55
0.6I
I O'
O 10 20 30 40 O 0.2 0.4 0.6 0.8
o (MPa) o i f,
Fig. 7-Bond stress-confinement stress envelopes: (a) test results and fitting
(small-diameter bars, 0 = 14 mm), and (b) proposed law (Eq. 5)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
BOND STRENGTH VERSUS CONFINEMENT
In Fig. 8 several test results [7,10-121, at constant confinement (variable crack

opening), at constant crack opening (variable confinement) and with passive
confinement are represented, in terms of bond strength and confinement stress.
As already pointed out in previous papers (see for instance Ill]), for
small and medium confinement levels (df, < 0.25) there is a definitely linear
relationship between the bond capacity and the confinement stress, after concrete
splitting.
The linear regression on the test results regarding sinali-, medium- and
large-diameter bars (db= 14-24 min) leads to Eq. 6.
At higher confinement levels ( d f , = 0.25-0.80) the dispersion of the test
results is considerable, but the results show clearly that the bond strength does
not keep up with the confinement level. Such an occurrence is confirmed also
by the results at constant crack width, which are quite close to those at constant
confinement. In other words, it does not matter whether the bond peak-stress
T,,,,, is reached during a loading process at constant contïneinent (a = constant)
or at variable confinement, since T,,,,, is mostly related to the actual confinement
stress. This is nothing new, since the test results quoted in [12] show that the
higher the confinement level, the less the path-dependency of bond.
The linear regression on the test results regarding medium-diameter bars
(d, = 19 min, Fig. 8) leads to Q. 7:
T,,~.J. = 0.075 + 1.20 a([, (u([. 50.25) (6)
T,~~,J[. = 0.293 + 0.247 a([, (a(f;.>0.25) (7)
Eqs. 6 and 7 (full and dot-dashed straight lines in Fig. 8) represent two different
behaviors regarding the failure of a bar after concrete splitting.
m Obb27Y7 0542b04 971
O Authors (constant confinement); Modena (constant confinement);

0.0 Authors (constant crack width); v Noghabai (Iiwsive confineinent);
----- Maivar (constant confinement):
Tests S P1-PS; O P6-PlO; - Linear Regression ,
( O @)
A PCl-PCS _____ Linear Regression (0)
Fig. 8-Ultimate bond capacity as a function of the confinement level; as a rule,

the normal stress refers to the peak of the bond stress-bar slip curve [7,12]
At low and medium confinement levels (a/f, < 0.20-0.25) the bond-slip
response is characterized by the steady opening of the splitting cracks and by
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the ensuing increase of the radiai and circumferential strains [ 10,l i]; thus, the
bond strength is greatly affected by the Confinement levei, which tends to keep
the cracks closed.
On the other hand, at high confinement levels (a//f,>O.25) the bond-slip
response is characterised by very limited radiai strains in the concrete
surrounding the bar [lo], which implies a limited crack opening; as a result, the
bond strength is little affected by the confinement level.
In the former case the failure of bond is due to a combination of crack
opening and concrete crushing underneath the lugs of the bar, while in the latter
case a veritable pull-out occurs, with partial or negligible splitting and with the
lugs crushing and furrowing the concrete surrounding the bar, or shearing off
the concrete keys between the lugs.
The value a/f, = 0.25 (which is the threshold of "true" pull-out) agrees
quite well with the value a/f, = 0.30 quoted by Nagatomo and Kaku [ 181.
At the close of this part of the paper, it is worth noting that the term
"confinement" is used here to indicate the normal stress at the bar-concrete
interface (a, Ü or pb in Figs. lb and 2a,b,c). This stress has nothing to do with
the external pressure pc (if any, Fig. lb), which - of course - tends to increase
the confinement stress.
~ ~
0662949 0542b05 828

LIMIT-ANALYSIS MODELS
Elastic-Cohesive Model
Several limit analysis models [4,6,7,8] have recently been developed (Fig. 9) to
predict the capacity of a short anchorage subjected to pull-out and to describe the
pattern of the splitting cracks and the hoop action in the concrete cover. These
models are based on the elastic-cracked axial-symmetric model introduced by
Tepfers in the Seventies [2].
Tepfers' model, though simple and based on clear and sound assumptions,
does not encompass a few important aspects of the splitting failure, such as (a)
crack cohesion (which is related to concrete softening in tension), (b) crack number,
(c) the bidimensional nature of the state of sîress in the transition region between
the cracked core and the solid "ring" (tension + compression), and (d) the favorable
effects of external pressure (as occurs in proximity to a direct support).
Tepfers
(1 979)
VI
Reinhardt
van der Veen
(1990)
[41
Rosati
Schumm
(1992)
[61
Noghabai Elastic-cohesive
(1995) model
[71 (n=2)
ECH
D w,-w
Rosati Eìastic-plastic- o w= f, 2-
Schumm cohesive w, D, + kw
(1995) model
See Refs. in (n var.) Proc. 2"* Int. Workshop on HPFR
u91 Cement ComDosites. Ann Arbor
I
~ ~~~~~
EPC ((Mi, USA), June 1995, pp.349-381.
Fig. 9-limit analysis models to predict confining capacity and crack pattern of
concrete cover
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb2949 0542bOb 7b4 W
Several models have been developed lately to introduce crack cohesion

(l3g,9), which in mm requires the introduction of crack kinematics and number,
since the crack width is the mling parameter of any cohesive law. As a muit, a
statically-admissible approach (as in Tepfers’ model) is no longer sufficient, since a
suitable assumption on crack kinematics must be introduced.
All of the models which have been developed since 1990 (Reinhardt and
van der Veen [4], Noghabai [7] and Rosati-Schumm [6,8])are based on the
assumption of partial cracking in the cover and perfect elasticity in both the outer
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ring and the concrete keys between the spiitting cracks. They are also based, in the
case of no external pressure, on a uni-axial failure criterion, with ultimate tensile
sbiain E, equal to m.
Different stress displacement laws have been adopted for crack cohesion
(exponential [4]; non linear polynomial [4]; linear [7]; hyperbolic or linear [6,8]).
The law adopted here for concrete softening [20] contains some parameters
related to concrete fracture, such as concrete tensile strength fa and critical crack
opening w,:
D w,-w
o,(w) = fct3-
w, Da + K W
where o, is the cohesive tensile stress (i.e. hoop stress along a spiiüing crack), w
and w, are the crack opening and its critical value (a,,,(w=w,)=û), Da is the
maximum aggregate size and K is a shape parameter related to fracture energy Gf :
As for crack kinematics, different assumptions may be introduced for the

circumferential elongation in the cracked are, leading to many relationships for
the ratio ~6 between the elongation of any circumferential fiber (Rb 5 r a ) and the
elongation at the interface between the cracked core and the outer elastic ring (r=Q
(Eig.10):
in particular, the following assumptions may be adopted, according to the

actual bar-to-concrete interface behavior:
- constant circumferential elongation, with each fiber subjected to the same stress
Obb2949 0542b07 bTO W
which develops at the tip of the splitting cohesive cracks (€,(r)=f&) (the
equiIibrium in the hoop direction is not satisfied inside each concrete key) [4];
- constant circumferential elongation, with each fiber subjected to the same stress
which develops along the splitting cohesive cracks (c&(r)=c&(r))(the equilibrium
in the hoop direction is satisfied) [6-81;
- variable circumferential elongation, i.e. according to the hoop-strain distribution
of an "equivalent" elastic case, or to a hyperbolic distribution.
Depending on the kinematic assumption, different crack-width and cohesive
stress distributions are obtained (Fig. 10) and, by imposing the equilibrium along
any radial, partially-split section, the bar-concrete pressure can be formulated as
follows:
cohesive elashc cohesive elashc

,stress stress mess scress
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
crack mdm v crackwidm
0 Rb % % 0 Rb Rc Rrn
rebar cracked concrete solid concrete rebar cracked concrete solid concrete
I r Rc
ar(r,Rc) = -+-C 1 ar(r,Rc)=->I
c
2 2R,
Fig. 1O-Crack-width (w), and cohesive (0,)and elastic (ot)stress distributions

for different kinematic conditions a
o,(r=R,l
Fig. 11-Sketch of a concrete key between two radical cracks (a); plots of hte
confining pressure Pb and crack-tip cohesive stress o, as a function of crack
penetration P, (b)
0bb2949 0542608 537
In [6,8] the crack number is not given a-priori, but is one of the unknowns of
the problem. By definition, it must be the number no (Fig. 11) that satisfies
both sîatic consistency (positive cohesive stresses smaller than fa or equal to fct
at the crack tip) and kinematic consistency (positive crack opening):
I 1
static
condition
local
equilibrium
1
in the radial
solid
constitutive
relationships
concrete
cracked
concrete
y
l) faIl simmT1
kinematic
condition
ar =-
6, (Rc)
failure
criterion
% =-
E~ *polar
criterionof
localization
n&integer2
uniform 1
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
direction ot = Ec&, w = w(o, ) 0I I Rb 2

R c .
crack width cohesivestress
concrete key betweentwo radial cracks:

equilibrium in the radial direction
consistent stresses and consistent crack widths

I
1
static condition kinematic condition
02 Dw(r,Rc,n) I fa w(r,R,,n) 2 O
, ,
CRACK NUMBER consistent pressure radial pressure
N
(first integer 2 n4
+ +
stabilitv condition
dpt(Rc)/dRc = O
CRACK PENETRATION RADIAL PRESSURE
Fig. 12-Flow-chart of the ECH model by Rosati and Schumm, see refs. in [191
W Obb2943 0542609 473 m
38 Gambarova. Rosati and Schumm
According to experimental evidence, it seems better to adopt the first

integer number N of radial cracks bigger than no(&), so that the consistent radial
pressure becomes:
By maximizing the pressure with respect to crack extension

b
( dp (R ,) / dR = O ) and by giving the integer value N to the crack number n, the
-
effective crack penetration RcNand the effective confining capacity pb of
concrete cover are found [6,8].
It is worth noting that the effects of extemal pressure have also been
introduced [8]: bond capacity increases, but not beyond an upper limit,which is
related to the compressive strength of the concrete in the radial direction; the crack
number, critical extension and maximum width decrease because of the external
pressure.
To evaluate bond strength on the basis of the bar-concrete pressure Pb

obtained with the above-formulated model, a bond stress - radial pressure
relationship (Eq.5 ) is needed. This relationship depends on the crack width at
the bar-to-concrete interface and it has been calibrated with reference to a single
splitting plane, characterized by two polar symmetric radiai cracks (N=2). For
this reason, an equivalent crack width 6, has to be evaluated for every radial
crack pattern (N22). For the sake of simplicity [12], its value can be calculated
as being equal to:
A comparison of the proposed bond model with experimental results by

Malvar [lo], obtained for different and predetermined values of the external
pressure pe, is shown in Fig. 14a: of course the burt-concrete pressure Pb could
not be measured during the test, and is one of the unknowns of the proposed
model.
Similarly, Fig. 14b shows the comparison of the present model with
Orangun’s formula:
where no cohesive contribution is accounted for (f, = compressive strength, C =

concrete cover, 1, = lap of the splice). In particular, the solid line corresponds to
Eq. 15, the dotted and dashed lines correspond to normal concrete (Gf= 0.08
N/mm and w, = 0.15 mm) and to fiber-reinforced concrete (Gf= 5 N/mm and w,
= 3 mm) respectively.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Ob62949 0542630 395
-A x71MPa
----AB03
---O431
Pe --m2s6
01.72
--- I O 4 0
--.-.00.86
Fig. 13-Comparison of proposed bond model with Malvar's test results (101 (a),
and Orangun, Jirsa and Breen's equation (AC1 JOURNAL, V.74, No. 3, 1977) (b)
Elastic-plastic-cohesive model
The fundamental aspects of the elastic-plastic-cohesive model developed from

the elastic-cohesive model presented in [6,8] (Fig. i4a) are as follows:
1.. The concrete mass surrounding the bar consists of a three-phase material with
an inner radially-cracked and cohesive core, an intermediate plastic ring and
an outer elastic ring; the intermediate ring is a recent development, since the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
hoop action in [6,8] was developed only by the elastic ring and by the
cohesion of the radial cracks in the original model.
2. The hoop action in the elastic and plastic rings (with the contribution of
external pressure p,., if any [8])prevents the piill-out of the bar as long as the
penetration of the radial cracks R, is less than or equal to a "critical" value
which maximizes bar-concrete pressure pi, and, consequently, puli-out
strength. Beyond this critical value, the equilibrium of the bar is no longer
possible, since the radial cracks become unstable and spread across the whole
cover.
3. The solid parts of the radially-cracked core behave elastically and are subject
to uniaxial compression in the radial direction (v=O).
4. The radial cracks have some residual strength in tension (crack cohesion
[20]) and can transfer tensile stresses, as shown by the softening branch of
the stress-displacement curve (Eq. 8, Fig. 14c). In the case of fiber-
reinforced concrete, both f,, and w, depend on fiber content.
5 . The radial cracks stabilize if, and when, the hoop stress at their apex is equal
to concrete strength in uniaxial tension (uniaxial criterion) or in tension-
compression (biaxial criterion): tension-compression interaction is detrimental
m 0662949 0542611 O21
to both tensile and compressive strength, and should be introduced (biaxial
criterion) in the case of high external pressure [SI).
6 . Crack width in the concrete core and bar-concrete pressure can be formulated
in terms of a few geometric and mechanical parameters by imposing (a) the
equilibrium condition along any radially-cracked section, and (b) the
kinematic condition in the cracked core, which governs the shape of the
radial cracks, i.e. the dependence of the width on the radial coordinate r.
7. In the uncracked region, the hoop strains are assumed to have an elastic-type
distribution according to an elastic constitutive law, energetically equivalent
to the actual elastic-plastic law. However, the hoop stresses are introduced
according to the actual elastic-plastic behavior of the concrete (Fig. 14c), and
as such they appear in the equilibrium equation.
8. The critical crack extension, representing the maximuin exploitation of bond,
is obtained by maximizing bar-concrete pressure pi, with respect to crack
extension, for any given crack number.
9. The crack number is defined as the nuinber corresponding to the smallest
value of the pressure maximums, in fiiltïllinent of both static and kinematic
consistency. Smaller or greater crack numbers, albeit statically and
kinematically consistent, prevent the maximum exploitation of bond in the
former case, or may cause unstable situations in the latter case.
By applying this model, the behavior of a bar embedded in plain
concrete (Fig. 14c, curve 1) or in high-voluine fiber-reinforced concrete (Fig.
14c, curves 2 and 3) can be compared. In Fig. 14c, the stress-strainlstress-
displacement curves do not refer to any specific normal or fiber-reinforced
concrete, but are three possible and realistic, albeit arbitrary, constitutive laws.
The three curves have the saine initial slope (fiber content does not significantly
alter the elastic behavior), slightly different peaks (fibers add some strength in
tension), different plateaus (High-Perforinance Fiber-Reinforced Cement
Composites have a reinarkable pseudo-plastic behavior, depending on fiber
content) and slightly different softening branches, representing crack
localization. The extension of the plastic plateau Ac is O, 7 c l , 15 c i respectively
(cl = peak strain in plain concrete = fc,/E,.).
Fig. 14b shows how rapidly bond behavior tends to approxiinate to fully
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
plastic behavior, because of the contribution of the fibers (fiber content should
be in excess of 4-5% by volume).
Fig. 15 shows the variation of crack width (at the bar-concrete interface,
Fig. E a ) , crack number and critical penetration (Fig. 15b) as a function of
concrete cover.
In summary, high fiber-contents (which turn the behavior of concrete
under tension from elastic-brittle to elastic-plastic-softening) guarantee much
better bond characteristics for the same concrete cover, as well as thinner and
less extended splitting cracks, and - in the end - greater strength and durability.
CONCLU DI NCi REM ARKS
Concrete splitting is something that designers, builders and researchers have to

learn to live with, even more since the new materials entering into the scene,
m Ob62949 0542b12 Tbô
Tribute t o Peter Gergely 41
ECR ECH EPC PLS
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 14-(a) 2-D Limit Analysis Models for cover splitting: ELS = ELaStic,
ECR = Elastic-CRacked, ECH = Elastic-CoHesive, EPC = Elastic-Plastic-Cohesive,
and PLS=PLaStic, see also Figure 9; (b) bar-concrete pressure and bond strength
versus concrete cover (ECR" = Elastic-CRacked with bond cohesion 131); (c) typical
stresslstrain and stresslcrack-opening laws for plain concrete (1) and fiber-
reinforced concretes (21, (3) (vf > 5 percent by volume- high-volume fiber content)
6
n
5
z /---
__-----
8 0.904
HPFRCC
0.000
o
O 1 2 3 4 5 6 O 1 2 3 4 5 6
CONCREE COVER RATIO C/hb CONCRmE COVER RATIO C/&
Fig. 15-Plots of maximum crack-width (a), and crack numberlcrack penetration

(b) versus concrete cover
= Obb2îqî 0542b13 î T 4
such as high-strength concretes and non-metallic bars, favor cover splitting,

owing to their lack of toughness in the former case (silica-fume concretes) and
to their reduced chemical adhesion and enhanced wedging action in the latter
case. Other factors too play a relevant role, like bar rusting, concrete shrinkage
and concrete exposure to extreme environmental conditions (low and high
temperatures). However, the mixed effects of the environment-related factors
need still to be thoroughly investigated.
Owing to the number of parameters affecting cover splitting, further efforts
are needed to work out the interface equations (bond stress-confinement stress
laws) valid for each specific situation, as well as the constitutive laws of the
concrete, since splitting is not a pure interface problem.
Once the above equations have been formulated, the limit-analysis models
come into play. In spite of their 2-D description of the 3-D bond problem, these
models are capable of shedding light on each specific parameter, and allow a
comparison between different materials (concrete and reinforcement). In this
sense, 2-D models are instrumental in the calibration of the code provisions
regarding high-strength and fiber-reinforced concretes, and special bars, with
reference to bar cover and interspace, splice length and anchorage capacity.
Beside the parameters related to the concrete embedment. the issue of the
confinement needs special attention, and new or improved formulations for both
active and passive confinement (transverse reinforceinent) are necessary. With
regard to this point, the 2-D Limit-Analysis models are promising because their
favorable effects as crack arrestors can be investigated in a relatively easy way.
Finally, the crack number should be carefully considered too, since more
cracks (generally thinner and less extended) enhance both bond strength and
durability, as shown by the elastic-plastic-cohesive model introduced in this
paper (Fiber-Reinforced Concrete).
It is true that in many cases the crack number is dictated by the geometry of
the cover and by the arrangement of the reinforceinent, but it is equally true that
in well-embedded bars (isotropic cover) the crack number should be dictated by
the static and kinematic consistency of the crack pattern, and by the maximum
exploitation of bond (maximum interface pressure = "minimum" crack
number). However, further effort is needed to identify the actual solution
("true" crack number), in terms of stable equilibrium, but to do that an energy-
based approach should be adopted.
ACKNOWLEDGEMENT
Financial support by the Italian Ministry of Higher Education (MURST), as a

part of the National Project "Innovative Techniques and Modelling in RC/PC
Structures" (1994-97), is gratefully acknowledged.
This paper reflects closely the stimulating discussions and innovative
thinking, that are taking place within the Task Group TG 2/5 "Bond Models"
of CEB-Comité Eurointernational du Béton.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542634 830 W
REFERENCES
1. CEB Bulletin No. 151, "Bond action and bond behaviour of reinforcement: State-of-the
Art Report", CEB Task Group 611, Reporter R. Tepfers. April 1982, 153 pp.
2. Tepfers R., "Cracking of concrete cover along anchored deformed reinforcing bars",
Magazine of Concrete Research. V.31, No. 106. 1979, pp. 3-12.
3. Vandewalle L., 'Theoretical prediction of the ultimate bond strength between a
minforcement bar and concrete", Proc. Int. Conf. "Bond in ConCretc from Researcb
to Practice", CEB (Comité Eurointernational du Béton) RTU (Riga Technical -
University), Riga (htvia), October 1992, V.1, pp. 1.1-1.8.
4. Reinhardt H.W. and van der Veea C., "Splitting failure of a strain softening material
due to bond stresses", Application of Fracture Mechanics to Reinforced Concrete, Ed.
by A. Carpinîeri, Elsevier Applied Science, 1990, pp. 333-346.
5. Rots J.G., "Simulation of bond and anchorage: usefulness of Softening Fracture
Mechanics",Applications of Fracture Mechanics io Reinforced Concrete, Ed. by A.
Carpinten, Elsevier Applied Science, 1990, pp. 285-306.
6. Rosati G.P. and Schumm C.E., "Modeling of local bar-to-concrete bond in RC
beams", Proc. Int. Conf. "Bond in Concrete: from Research to Practice", CEB-RTU,
Riga (Latvia), October 1992, V.3, pp. 12.34-12.43.
7. Noghabai K., "Splitting of concrete covers - A fracture mechanics approach", Proc.
2nd Int. Conf. on Fracture Mechanics of Concrete Structures, Ed. by F.H. Wittmann,
Aedificatio Publ., Zurich (CH), July 1995, V.11, pp. 1575-1584.
8. Gambarova P.G., Rosati G.P. and Schumm C.E., "An elasto-cohesive model for steel-
concrete bond", Proc. Europe-US Workshop on Fracture and Damage in Quasibrittle
Structuras. Ed. by Z.P. Bazant, Z. Bittnar, M. Jirasek and J. Maars, Prague (Czech
Republic), September 1994. pp. 557-566.
9. den üijl I.A. and Bigaj A.J., "A bond model for ribbed bars based on concrete
confinement", Heron, V.41, No.3, 1996 (in press).
10. Malvar L.J.,"Bond of reinforcement under controlld confinement", AC1 Materials
Journal, V.89, No. 6, 1992, pp. 593-601.
11. Modena C.. Coltm T. and Rossaro G.P., "Investigation on skel-concrete bond after
concrete splitting: tests at constant confinement" (in Italian), Studi e Ricerche, School
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
for the Design of RC Structures, Milan University of Technology, V.10188, Milan
(Italy), 1989, pp. 179-218.
12. Gambarova P.G. and Rosati G.P., "Bond and splitting in bar pull-out: behavioural laws
and concrete cover role", Magazine of Concrete Research, V.49, No. 179, pp. 99-1 10.
13. Ma& M., Otani S. and Aoyama H., "Effect of Confinement on Bond Splitting
- -
Behavior in Reinforced Concrete Beams", Structural Ennineenna International. No. 3,
1995. pp. 166-171.
14. Cairns J. and Jones K., "An Evaluation of Bond-SplittingAction of Ribbed Bars", AC1
Materials Journal. V.93. No. 1, 1996, DD. 10-19.
15. Plizzari G.A., Deldossi M.A. and Masiimo S., "Experimental study on anchored bars
in RC elements with transverse reinforcement", Materials and Structuras-RILEM,
V.29, 1996, pp. 534-542.
16. Ya0 B. and Murray D.W., "Study of Concrete Cracking and Bond Using a Distributed
Discrete Crack Finite Element (FE) Model", AC1 Materials Journal, V.92, No. 1,
1994, pp, 93-104.
17. Cox J.V. and Hernnaiui L.R., "Confinement-stress dependent bond behavior: a two-
degree of M o m plasticity model", Proc. Int. Conf. "Bond in Concrete: from
Research toPractice", CEB-RTU, Riga(Latvia), October 1992, V.3, pp. 11.11-11.20.
18. Nagatomo K. and Kaku T.,"Bond behavior of deformed ham under lateral compressive
and tensile stress", Proc. Int. Conf. "Bond in Concrete: from Research to practice",
CEB-RTU, Riga (Latvia), October 1992, V.1, pp. 1.69-1.78.
19. Cappellini L., "On the ultimate capacity of short anchoragw embedded in elastic-
cohesive cement composites" (in Italian), MS Thesis, Advisors: G.P. Rosati and C.E.
Schumm, Dept. of Structural Engineering, Milan University of Technology, Milan
(Italy), December 1996, 171 pp.
20. Giunani E. and Rosati G.P., "Behaviour of concrete elements under tension after
cracking", Studi e Ricerche, School for the Design of RC Structures, Milan University
of Technology, V.8/85, Milan (Italy), 1986, pp. 65-82.
Ob62949 0542655 777 W
SP 180-3
Contribution of Concrete between Cracks

at Inelastic Steel Strains and Conclusions
for the Optimization of Bond
by R. Eligehausen, J. Otbolt and U. Mayer
Svnopsis: The bond between reinforcement and concrete should

ensure high structural stiffness and small cracks in the serviceability
limit state, generate small splitting forces and allow full utilization of
the reinforcement ductility in the ultimate limit state. While bond
behavior at service load and splitting behavior has been investigated
intensively, bond behavior at large inelastic steel strains is not known
very well. Therefore, in this paper the contribution of concrete
between cracks at inelastic steel strains is investigated numerically
based on a rational mechanical model and using realistic constitutive
material laws. The model predictions agree rather well with a large
number of test results. According t o the results of the parametric
study, after steel yielding the ratio of mean steel strain t o the steel
strain at the crack is mainly influenced by the reinforcement
percentage and the shape of the steel stress-strain curve. It is much
lower than at service load. Due t o this lower ratio of mean steel strain
t o steel strain at the crack, the rotation capacity of plastic hinges and
thus the structural ductility is reduced significantly and may be very
low if reinforcement with low ductility is used. Therefore an
optimization of bond seems t o be necessary. Corresponding extensive
numerical and experimental studies are under way in Germany.
Keywords: Bond (concrete to reinforcement); concrete; cracks; reinforcement; rib

pattern; rotation capacity; tension stiffening; yielding
45
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542616 603
46 Eligehausen, Ozbolt and Mayer
AC1 member Rolf Eligehausen is a Professor and Department Head a t

the Institute for Building Materials, University of Stuttgart. He studied
at the University of Braunschweig and obtained his Doctor from the
University of Stuttgart in 1979. After 2 years of research work at the
University of California at Berkeley, he returned t o the University of
Stuttgart and was appointed t o the rank of Professor in 1984. He is a
member of several international committees on reinforced concrete
and fastening technique, and has published several papers in these
fields. He is member of AC1 Committees 355 Anchorage t o Concrete,
408 Bond and 446 Fracture Mechanics.
Assoc. Prof. Dr.-lng. Josko Ozbolt graduated in 1978 at the

University of Zagreb (Croatia). As a postgraduate student he spent
one year (1980) at the TNO Institute in Delft. He got his PhD from
the University of Zagreb in 1982. As a postdoctoral student he
worked for one year at the Northwestern University, USA. Since
1989 he is working at the Institute for Building Materials, University
of Stuttgart. In 1995 he finished his postdoctoral thesis ,,Size effect
and ductility of concrete and RC structures". In this field he is author
or coauthor of a number of publications.
Utz Mayer got his diploma degree in civil engineering from the
University of Stuttgart in 1991. After that he worked as a structural
engineer in a design office dealing with the design of bridges. Since
1995 he is working as a research engineer in the field of bond at the
Institute for Building Materials, University of Stuttgart.
1 INTRODUCTION
The behavior of reinforced concrete structures is significantly

influenced by the interaction (bond) between concrete and
reinforcement. The bond should fulfill the following requirements:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0hh2949 05426317 5 4 T - W
a) In the serviceabilty limit state (SLS)the width of cracks in the

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
concrete and the deformation of RC members should be smaller than
allowable values. This requires small crack spacings and a high
contribution of concrete between cracks or a low ratio between the
mean steel strain , ,E and the steel strain E,, at the crack (bare bar
strain) respectively, which can be obtained by a high bond stiffness
(high bond stress with corresponding small slip values) and a
sufficiently high bond strength.
b) In the ultimate limit state (ULS) at inelastic steel strains the ratio
between the average steel strain and the steel strain at the crack
should be rather large to ensure ductile behavior of RC members with
large rotation capacity at plastic hinges. Therefore the bond
resistance should be reduced after passing the yield strain.
c l In the area of anchorages and lap splices of reinforcement high

bond strength should be accompanied by small splitting forces to
ensure short anchorage and lap lengths.
These requirements must be fulfilled for both top and bottom bars.
Top bars might show reduced bond strength due t o setting of the
concrete below the bar. These requirements partly contradict each
other and cannot totally be fulfilled simultaneously. Therefore a
compromise between the different requirements is necessary.
The bond behavior of deformed bars depends significantly on the

related rib area aR which is defined as the ratio between the rib
bearing area and the rib shearing area (11. According to the European
Pre-Standard ENV 10080 (2) the related rib area of bars with
diameter d, 2 11 mm must be aR 2 0.056. ASTM Standards require
only minimum rib heights and spacings for deformed US Billet (A
615/A 615M-96a, (19)) and Low Alloy (A 706/A 706M-96b, (20))
steel. These minimum values lead to related rib areas (not defined in
the US) aR = 0,057 to 0,073. The related rib area for deformed US-
steel wire (A 496-95a, (21)) and welded wire fabric (A 497-95, (22))
amounts aR = 0,016 t o 0,133 depending on the bar diameter and the
rib spacing (aR increases with increasing bar diameter). The rib
patterns of deformed bars used currently in Europe and in the US
mainly take into account requirements a) and cl. Therefore they
ensure good bond behavior in the serviceabilty limit state and high
capacity of anchorages and lap splices. Furthermore the notch effect
of the ribs has been reduced by rounding the rib base resulting in high
fatigue strength of the bars. However, at inelastic steel strains the
Obb2949 0542b18 4ôb
contribution of concrete between cracks may be too high, thus

reducing the rotation capacity of plastic hinges.
In this paper the stiffness of tension members up t o rupture of the

reinforcement is investigated by physical tests and numerical
parametric studies. The results are valid also for the tension chord of
beams and slabs. They show, that an optimization of bond seems t o
be necessary t o take account of the ULS requirements better than
they have been todate.
2 NUMERICAL MODEL AND ASSUMED MATERIAL PARAMETERS
2.1 Numerical Model
In numerical studies, the behavior of a reinforced concrete tension

member under increasing load up t o rupture of the reinforcement is
investigated using the model proposed in (3). The numerical model
follows the crack formation process from the first crack up t o the first
stabilized crack pattern by taking into account the statistical
distribution of the concrete tensile strength and the crack distance.
The first and last crack are assumed t o occur when the tensile stress
in the concrete reaches the 5%- and 95%-fractile of the concrete
tensile strength. Further cracks may be formed at inelastic steel
strains and corresponding large slip values between reinforcement and
concrete when the force transferred into the concrete between t w o
existing cracks by bond stresses is larger than the concrete cracking
force calculated with fct,95%.At each load step the distribution of the
reinforcement strains between the existing cracks is calculated by
iteratively solving the differential equation of bond. The corresponding
numerical procedure is described in (3).The average steel strain E,, of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the tensile member is calculated as a function of the crack formation

factor 5 by Equation 12-11 (for more details see (3)).
with 5 = factor describing the degree of cracking,

O no cracking, 1 stabilized cracking
E S k ) = average steel strain in the cracked part of the
member
m Ob62449 0542bLï 332 m
&s> = steel strain in the uncracked part of the member
In the model the following assumptions for the material behavior are
used.
2iL Assumed Material ProDerties
Behavior of Concrete in Tension--For concrete in tension linear elastic

behavior up t o the concrete tensile strength is assumed. The modulus
of elasticity is taken from (5). The average axial tensile strength ,f
may be estimated according t o (61 from the compressive cube
strength using Equation 12-21.
f c t = c, .fZ3 íMPa1 [2-21
with C
, = 0,24
Assuming a coefficient of variation of the concrete tensile strength of

1 5 % fct,5% = 0.75 ,f and fct,95%= 1.25 .,f The descending
branch of the load-deformation behavior of concrete in tension is
neglected.
Stress-strain Relationship of Reinforcement--The stress-strain relation-

ship of reinforcement is desribed by a number of data points for
stress c and corresponding strain E. In this way any stress-strain
diagram of hot rolled, cold worked or heat treated steel can be
modelled (7). Scatter of the mechanical properties of reinforcement
along the bar is neglected.
Bond Stress-SliD Relationshio--The crack formation process, the crack

width and the deformation behavior of reinforced concrete structures
are significantly influenced by the bond properties of the
reinforcement. Therefore realistic bond stress-slip relationships are
needed t o describe accurately the behavior of reinforced concrete
members.
The assumed bond stress-slip relationships are shown in Figure 1.

Figure l a shows the principal form of the bond stress-slip relationship
and Figure 1b shows their variation as a function of the ratio between
distance t o crack and bar diameter. The assumed bond stress-slip
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ûbb2949 0542b20 034
relationships take into account decreasing bond strength close t o

cracks. The characteristic points defining curves 1 t o 6 are given in
Tables 1 and 2.
The above desribed bond model was first proposed in ( 8 ) . The bond
stress-slip relationships in (8) were modified in (3) based on results of
special pull-out tests (9)t o take into account the influence of splitting
cracks on the bond stress-slip relationships in case of a small
concrete cover. The model is valid also for high slip values, which
occur at inelastic steel strains.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
3 COMPARISON OF NUMERICAL RESULTS WITH TEST RESULTS
The results of numerical analysis are compared with test results. In all
tests deformed reinforcing bars with a related rib area meeting or
exceeding the values required by ENV 10080 (2) were used.
In (10) reinforced concrete specimens (f, = 26.4 MPa; p = AJA,=

0.27%) with pre-formed cracks were tested in axial tension. The steel
strains along the bar were measured at different load steps below
yielding with strain gauges which were attached in a groove inside
the bar. In Figure 2 the measured and calculated distribution of the
bar forces along the specimen (deduced from the steel strains) are
plotted for different load levels. In the calculation the crack distance
as measured in the tests was assumed. According t o Figure 2 the
numerical results agree well with the measured values.
In ( 1 1) long specimens made out of light weight concrete were tested

in tension and the average strain was measured with extensometers
as a function of the applied load. In Figure 3 the steel stress at the
crack is plotted as a function of the measured and calculated average
strain. For comparison the strain of the bare bar is shown as well.
The agreement between numerical and test results is satisfactory.
In the previous examples the specimens were loaded up t o the yield

force only. In the next examples tension specimens were loaded up t o
steel strains well beyond yield or t o rupture of the reinforcement. The
results are of special interest for the problem a t hand.
In (12) tension members reinforced with one bar were tested. In

Figure 4 the applied loads are plotted as a function of the average
strains measured over the total specimen length and the calculated
D Obbiï4ï 0542b21i T70
values. For comparison the load-strain curve of the reinforcing bar is

shown as well. The reinforcement percentages were small (JA =
0.26% and p = 0.50%). The figure shows that calculated and
measured average strains agree well and that for forces larger than
the yield force the average strain is much smaller than the strain of
the reinforcing bar outside the specimen.
A t ETH Zurich reinforced concrete walls were tested in axial tension

(13). The specimen dimensions and the reinforcement layout is
shown in Figure 5a, the test Set-up in Figure 5b. The geometrical
reinforcement ratio and the type of steel were varied. The stress-
strain relationships of the reinforcement are given in Figure 5c. The
ductility of steel type 1 is small compared with type 2. The results of
t w o tests (p = 0.98%, f, = 50 MPa) are shown in Figures 6 and 7.
Plotted are the steel stresses calculated from the applied forces as a
function of the measured average strains for strains slightly larger
than the yield strain (Figures 6a and 7a) and up t o peak load (Figures
6b and 7b). Furthermore the ctress-strain curve of the reinforcement
is given as weil. It can be seen, that in the elastic state after
stabilization of cracking the average strain of the tension member was
not much smaller than the strain of the reinforcement at the crack.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
However, after yielding, the ratio E ~ ~ was

/ E low.
~ ~
In Figure 8 the measured and calculated ratios E,,,,/E~~ are plotted as a

function of the strain es,. es. is the strain of the reinforcement at the
crack at the given load. Figure 8a shows results for the specimen
reinforced with steel type 1 (low ductility) and Figure 8b shows steel
type 2 (high ductility). Numericat and measured results agree
reasonably well. A t failure the ratio average strain to strain at the
crack of the specimen 24 reinforced with low ductile steel is very
small (E,,/E~~ iz 0.2). For specimen Z1 reinforced with high ductile
steel the ratio E,,/E,, drops down to about 0.2 after yielding, but
increases t o about 0.6 for bar strains in the strain hardening range.
Further comparisons of numerical and test results are given in (3) and
( 7 ) . All comparisons show that the real behavior of reinforced
concrete specimen loaded in tension or bending can be predicted with
good accuracy with the numerical model.
S 0662949 0542622 907
4 PARAMETRIC STUDIES ON THE CONTRIBUTION OF CONCRETE

BETWEEN CRACKS AT INELASTIC STEEL STRAINS
4.1 Varied Pararneters
The numerical studies are done for monotonic loading up t o failure.

The following parameters are varied:
(a) Stress-strain relationship of reinforcement
Three types of steel according t o CEB-FIP MC 1990 ( 5 )

defining different ductility classes are used. They are
characterized by the ratio of tensile strength f, t o yield strength
f, and the strain E,, at maximum load.
class B, ft/fv = 1.05; E, = 2.5%

class A, ft/fv = 1.08; E,, = 5.0%
class S , ft/fv = 1 .I5; E,, = 6.0%
The classes B and A represent reinforcement with normal and

high ductility respectively. Reinforcement according t o class S
should be used in structures subjected to seismic excitations
(see (5)). The values given for the ratio ft/fv and for E,, are
defined as 5%-fractiles. The assumed stress-strain relationships
of the different steel classes are plotted in Figure 9 for
f, = 500 MPa. They represent the typical stress-strain behavior
of cold worked steel (class A and B) or heat treated steel
without yield plateau (class S ) .
(b) Ductility parameters of reinforcement (fv = 500 MPa)
E,, = 5.0%, f,/fv = 1.001, 1.02, 1 .O5 and 1 . I O

ft/fv = 1 .05, E,, = 1.5%, 2.5% and 5.0%
(c) Reinforcement ratio
The geometrical reinforcement ratio p = A,/A, (A, =

reinforcement area, A, = concrete area) was varied between p
= 0.3% and 1.5%.
(d) Concrete strength: ,f = 25 MPa t o 45 MPa
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 û542b23 8 4 3 U
(el Yield strength: f, = 500 MPa t o 600 MPa
Reinforcing bars used in the US are mainly hot rolled. Their ductility
will reach or exceed the ductility of steel class S . If these bars are
delivered in coils and straightened at the workshop their ductility will
decrease and will agree approximately with the ductility of steel class
A. In the US as in Europe welded wire mesh is mainly produced from
cold formed wires. The ductility of this reinforcement may be rather
low and may agree with the ductility of steel class B.
4.2 Numerical Results
General Behavior--Figure 1O shows a typical calculated steel stress

average strain diagram for a tension member with reinforcement ratio
p = 1.5%. For comparison the steel stress average strain relationship
given by the CEB-FIP Model Code 1990 (5) and the stress-strain
relationship of the reinforcement are plotted as well. Figure 10a
shows results up t o yielding and Figure 10b up to rupture of the
reinforcement. In the analysis steel class A with f, = 500 MPa and
,f = 25 MPa are assumed.
The behavior of a tension member under monotonic loading is

characterized by a high stiffness in the uncracked state (line O-A)
followed by a decreasing stiffness during the crack formation process
(region A-BI. After stabilization of cracking up t o yielding of the bars
the (J,-E,, relationship is almost parallel to the o,-E, relationship of the
reinforcement (line B-C). The distance between these t w o lines is
mainly influenced by the reinforcement ratio. It increases with
decreasing reinforcement ratio. After yielding the average strain of the
tension member becomes much smaller than the strain E$, of the bars
at the crack. Therefore the ratio E,,/E, is much lower than before
yielding. This can be explained as follows. Due t o the low stiffness of
the reinforcement after yielding only relatively small bond forces are
needed t o significantly reduce the force in the reinforcement and thus
the reinforcement strain (compare also Figure 14). Therefore inelastic
steel strains are limited to a small region beside the crack which
increases in length with increasing steel strain at the crack.
The calculated behavior is modeled relatively well by the simplified O,-

,E
, relationship defined in CEB-FIP Model Code 1990 (5). The
procedure for the inelastic range is based on results of numerical
investigations for steel class A (3).This steel has been used in the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Obbi2949 0542b24 78T
example shown in Figure I O . For inelastic steel strains of steel class B

or S the agreement between numerical results and the prediction by
CEB-FIP Model Code 1990 (5) is not as good as for steel class A (see
Figure 11). This is also true for steel with a yield plateau, because the
very low ratio E,,/E, at steel strains before strain hardening (compare
Figure 8b) is not taken into account in the equation given in (5).
..
Influence of SteeI Classes and Ductilitv Parameters--In Figure 11 the
calculated ratios E,,/E, are plotted as a function of the steel strain E,,
for reinforcement according t o three ductility classes. The figure
shows results for a reinforcement ratio p = 1.5%. In the analysis
f, = 500 MPa and ,f = 25 MPa are assumed. For steel strains less
than the yield strain the stress strain relationships of the different
steel classes are identical. Therefore in the elastic range the ratio
&,,/E,, is not influenced by the steel ductility class. After yielding of
the reinforcement the ratios &,,/E,, decrease considerably until they
reach an almost constant value. This plateau is higher for the more
ductile steel class S than for the rather brittle steel class B.
The three different ductility classes are defined by increasing values

of the ratio ft/fy and the ultimate steel strain E,,. To separate the
influence of these t w o parameters on tension stiffening, studies are
done for reinforcement with a constant value E, = 5.0% and varying
ratios f,/f, and a constant ratio f,/f, = 1 .O5 and varying values for E,
respectively. Reinforcement ratio (p = 1.5%), yield strength
(f, = 5 0 0 MPa) and concrete compressive strength (fcc = 25 MPa)
are kept constant. The results are plotted in Figure 12b and Figure
13b. Figures 12a and 13a show the assumed stress-strain
relationships of the reinforcement. While the influence of the ultimate
steel strain E,, on the ratio E,,/E,, is small (Figure 12b), the tension
stiffening is significantly influenced by the ratio f,/f, (Figure 13b). The
ratio E,,/E, decreases t o almost 0.1 for bars with a plateau after
yielding and reaches about 0.7 for bars with fJf, = 1. I O . The
explanation for this behavior is given in Figure 14. For an assumed
constant reduction of the steel stress caused by bond the steel
strains are reduced much more for bars with a low ratio f,/f, than for
reinforcement with a high ratio f,/f,.
For reinforcement with a yield plateau followed by strain hardening,

the ratio E,,/E, will drop down t o very small values after yielding and
increase again in the region of strain hardening (see Figure 8b). This
behavior can be explained by Figure 13b.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2’‘’ 0542b25 bLb
Influence o f Reinforcement RatiQ--Figure 15 shows the ratio E,/E, as

a function of the bare bar strains for a specimen with a reinforcement
ratio p = 0.5%. Parameter is the steel class. By comparing with
Figure 11, it is evident that the contribution of concrete between
cracks is significantly influenced by the reinforcement ratio. A t large
inelastic steel strains the ratio E,~/E,, is about 0.4 for p = 0.5% while
it reaches about 0.7 for JA = 1.5%.
Jnfluence of Concrete Comoressive Strenativ-Because the concrete

tensile strength increases with increasing concrete compressive
strength, under otherwise constant conditions the steel stress at first
cracking increases with increasing concrete compressive strength.
This influences the specimen stiffness during the crack formation
process (Figure 16a). However, after stabilization of cracking and at
inelastic steel strains, the influence of the concrete compressive
strength on the ratio E,/E, is small (Figure 16b).
Influence of Yield Strength--According t o the results of the numerical

study (1 4) the influence of the yield strength in the investigated range
(500 MPa If, I600 MPa) on the ratio E,~/E,, is negligible.
More details of the parametric study are given in (14).
5 CONCLUSIONS FOR THE OPTIMIZATION OF BOND
The results of tests and numerical investigations clearly demonstrate,

that for deformed bars with current rib pattern the contribution of
concrete between cracks at inelastic steel strains is significant
influenced b y the following parameters:
Reinforcement ratio. The ratio between average steel strain ,E,

and steel strain E,, at the crack (bare bar strain) decreases
significantly with decreasing reinforcement ratio.
0 Shape of the stress-strain relationship of the reinforcement.

The ratio E , ~ / E , ~ increases significantly with increasing ratio
f,/f,. The influence of the strain E,, at peak load on the ratio
E,,/E,, is small for strain values ,E, S 5.0%. For hot rolled and
heat treated steel with a pronounced yield plateau before strain
hardening, the ratio E ~ / decreases
E ~ ~ t o very small values after
yielding, but increases again when strain hardening is reached.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542b2b 552 9
0 Concrete compressive strength and steel yield strength have

only a small influence on the tension stiffening at inelastic steel
strains.
For low reinforcement percentages the ratio EJE,, at inelastic steel

strains may be as little as 0.2 t o 0.6 depending on the shape of the
stress-strain relationship of the reinforcement due t o the high bond
action of deformed bars with current rib patterns. This low ratio
E,,/E,, significantly reduces the plastic rotation capacity of plastic
hinges in reinforced concrete beams or slabs or the ductility of the
structure respectively. Reduction of plastic rotation capacity can be
seen from Figure 17, which shows the moment-rotation relationship
of t w o slabs (p = A,/(bd) = 0.24%) reinforced with welded wire
mesh produced from cold worked smooth and deformed bars. The
stress-strain relationship of the reinforcement used in the t w o tests
was almost identical and complied with the minimum requirements of
CEB-FIP Model Code 1990 (5) for steel class B. The significant
influence of bond on the rotation capacity of plastic hinges has also
be demonstrated numerically in ( 7 ) (see Figure 18).
The ductility and the rotation capacity of plastic hinges is small when
deformed bars or deformed welded wire meshes with current rib
patterns and a stress-strain relationship complying with the
requirements of steel class B according t o CEB-FIP Model Code 1990
(5) are used (3,7). Very little redistribution of moments after yielding
is obtained (16). Therefore, according to CEB-FIP Model Code 1990
(5) and Eurocode 2 (17) this reinforcement may only be used if the
static analysis of the internal forces is done according t o the theory of
elasticity or if the assumed amount of moment redistribution is very
small ( S 10%).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
According t o AC1 31 8 (23) only a linear analysis with a redistribution

of moments up t o maximum 20% depending on the amount of
flexural reinforcement is allowed. For this design, the necessary
plastic rotation capacity is rather small and is ensured when using
current US-reinforcement and employing the detailing rules of AC1
318 (23).
However, from the point of view of a designer a nonlinear or plastic

analysis is very beneficial, because it allows an increase in quality of
the structure by shifting moments from highly stressed t o less
stressed areas thus reducing the amount of necessary reinforcement
in highly stressed areas and improving the conditions for casting the
concrete. A plastic analysis is allowed by European Standards only for
m Obb2949 0542627 499 Tribute to Peter Gergely 57
very ductile structures and is not allowed by AC1 318 (23). To ensure
ductility of the structure, the steel ductility must be high when using
deformed reinforcement with current rib patterns. However, the
ductility of welded wire mesh produced from cold formed wires and
of cold worked bars transported in coils and straightened before use
complies just with the requirements of CEB-FIP Model Code 1990 (5)
for steel ductility class B. Therefore, t o guarantee the required
structural ductility in case of a plastic analysis when using cold
formed reinforcement, the following possibilities are available: (a)
steel production method can be improved t o provide the necessary
steel ductility, (b) rib pattern can be changed t o increase the ratio
E,/E, after yielding, and (c) a combination of both methods.
According t o steel producers, method (a) is rather expensive.
Therefore, in Germany intensive research work is currently under way
with the aim t o optimize the rib pattern of deformed welded wire
mesh t o fulfill the requirements in the serviceability and ultimate limit
state. Extensive non-linear FEM studies based on the non-local
microplane model (1 8 ) are performed t o investigate the influence of
major parameters on the bond stress-slip relationship and the splitting
behavior. The results of the numerical studies will be verified by
testing pull-out specimens, tension specimens, slabs with overlapped
splices and continuous beams. It is believed that the related rib area
of reinforcement with an optimized rib pattern may be smaller than
the values required todate.
6 SUMMARY
The behavior of reinforced concrete structures is significantly

influenced by bond between concrete and reinforcement. Bond should
(a) ensure a high structural stiffness and small cracks in the
serviceability limit state, (b) generate only small splitting forces and
(cl allow a full utilization of the reinforcement ductility in the ultimate
limit state by a reduction of the contribution of concrete between
cracks after steel yielding. These requirements partly contradict each
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
other and cannot totally be fulfilled simultaneously, especially when

taking into account the top bar effect. Therefore a compromise
between the different requirements is necessary. The rib patterns of
deformed bars used currently are mainly optimized in respect of the
requirements (a) and (b) above.
In this paper the contribution of concrete between cracks is

investigated numerically based on a rational mechanical model using
m Ob62949 0542b28 325 =
realistic constitutive material laws. The model predictions agree well
with the results of tests.
The studies show, that after yielding of the reinforcement the ratio
between average steel strain E , ~and steel strain at the crack E,, is
mainly influenced by the reinforcement ratio and the shape of the
steel stress-strain relationship. It may be as small as 0.2 t o 0.6 for
small reinforcement percentages (p 2 0 . 5 %) and a low ratio of steel
tensile strength t o yield strength. This low ratio E,,/&,, significantly
reduces the rotation capacity of plastic hinges and thus the structural
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ductility, which might be small when reinforcement with low ductility

(which may be found in practice) is employed. To increase the
structural ductility either the steel ductility should be improved by
better production methods or the bond strength at large steel strains
should be reduced. In Germany extensive numerical and experimental
investigations are under way to check whether the rib pattern can be
optimized such that the contribution of concrete between cracks a t
inelastic steel strains is reduced without changing the bond behavior
a t service load and the splitting forces.
7 REFERENCES
( 1) Rehm, G . : Über die Grundlagen des Verbundes zwischen Stahl und

Beton (On the basic behavior of bond between steel and concrete).
Deutscher Ausschuß für Stahlbeton, Heft 138, Verlag W. Ernst &
Sohn, Berlin, 1961
(2) ENV 10080: Steel for the reinforcement of concrete; weldable

ribbed reinforcing steel B 500 - Technical delivery conditions for bars,
coils and welded fabric, (final draft May 1994)
(3) Kreller, H.: Zum nichtlinearen Verformungsverhalten von

Stahlbeton-tragwerken unter Last- und Zwangeinwirkung (On the non-
linear behavior of reinforced concrete members under load and
imposed deformations). Dissertation Universität Stuttgart,
Mitteilungen No. 1989/4, Institut für Werkstoffe im Bauwesen,
Universität Stuttgart, 1989
W Obb2949 0542629 2bL W
(4) Ciampi, V.; Eiigehausen, R.; Bertero, V.V.; Popov, E.P.;: Analytical
Model For Concrete Anchorages Of Reinforcing Bars Under
Generalized Excitations. Report No. UCB/EERC - 82/23,Earthquake
Engineering Research Center, College of Engineering, University of
California, Berkeley California, November 1982
(5) CEB Bulletin d’Information No. 213/214: CEB-FIP Model Code

1990. London: Thomas Telford Ltd., May 1993
(6) Heilmann, H. G.: Beziehung zwischen Zug- und Druckfestigkeit

des Betons (Relationship between concrete tensile and compressive
strength). beton, Heft 2/1969, pp. 68-70
(7) Langer, P.: Verdrehfähigkeit plastizierter Tragwerksbereiche im

Stahlbetonbau (Rotation capacity of plastic hinges in reinforced
concrete). Dissertation Universität Stuttgart, Mitteilungen No.
1987/1,Institut für Werkstoffe im Bauwesen, Universität Stuttgart,
1987
( 8 ) Eiigehausen, R.; Popov, E. P.; Bertero V. V.: Local bond stress slip
relationships of deformed bars under generalized excitations.
Earthquake Engineering Research Center, University of California,
Berkeley: Report No. UCB/EERC 83/23,1983
(9) Eiigehausen, R.; Kreller, H.; Langer, P.: Untersuchungen zum

Verbund-verhalten gerippter Bewehrungsctäbe mit praxisüblicher
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Betondeckung (Investigations on the bond behavior of deformed

reinforcing bars with normal concrete cover). Mitteilungen No.
1989/5,Institut für Werkstoffe im Bauwesen, Universität Stuttgart,
1989
(1O) Krips, M.: Rifibreitenbeschränkung im Stahlbeton und

Spannbeton (Crack control in reinforced and prestressed concrete
structures). Dissertation TH Darmstadt, 1984
D 0662949 0542b30 T83 =
(11) Rostásy, F.; Koch, R.; Leonhardt, F.: Zur Mindestbewehrung für
Zwang von Außenwänden aus Stahlleichtbeton (On the minimum
reinforcement of external light weight concrete walls subjected t o
imposed deformations). Deutscher Ausschuß für Stahlbeton Heft
267, Verlag W. Ernst & Sohn, Berlin, 1976
(12) Martin, H.; Schießl, P.; Kutsch, H.: Einfluß der Form der
Spannungs-Dehnungslinie von geschweißten Betonstahlmatten auf
das Tragverhalten von Stahlbetonkonstruktionen (Influence of the
stress-strain relationship of welded wire mesh on the behavior of
reinforced concrete structures). Institut für Betonstahl und Stahlbeton
e. V., München 1980
(13) Alvarez, M.; Marti, P.: Versuche zum Verbundverhalten von

Bewehrungsstahl bei plastischen Verformungen (Tests on the bond
behavior of reinforcing steel at plastic deformations). IBK-Bericht Nr.
222, Institut für Baustatik und Konstruktionen, ETH Zürich,
September 1996, 135 pp, Birkhäuser Verlag, Basel
(14) Eligehausen, R.; Mayer, U.: Parameterstudie zur Mitwirkung des

Betons unter Kurzzeitbelastung (Parametric study on the contribution
of concrete between cracks under short-term loading). Research
report, Institut für Werkstoffe im Bauwesen, Universität Stuttgart,
1997
(15) Eibl, J.; Bühler, A.: Untersuchung des Einflusses verschiedener

Stahlparameter auf die mögliche plastische Rotation bei
Stahlbetonplatten (Investigation on the influence of different steel
parameters on the plastic rotation capacity of reinforced concrete
slabs). Test report, Institut für Massivbau und Baustofftechnologie,
Universität Karlsruhe, 1991
(16) Eligehausen, R.; Fabritius, E.: Tests on continuous slabs

reinforced with welded wire mesh. In: CEB Bulletin d’Information No.
21 8: Ductility Reinforcement, Lausanne, August 1993, pp. 133-148
(17 ) ENV 1992-1-1 : Eurocode 2: Design of concrete structures: Part

1
1 General rules and rules for buildings. June 1992
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542b31, 91T
(18) Ozbolt, J.: Microplane model for quasibrittle Materials - Part 1:

Theory, Part 2: Verification and numerical examples for concrete.
Submitted t o International Journal for Solid and Structures, 1997
(19) ASTM - Standard specification for Deformed and Plain Billet-

Steel Bars for Concrete Reinforcement. Designation A 61 5/A 61 5 M -
96a.
(20) ASTM - Standard specification for Low-Alloy Steel Deformed and

Plain Bars for Concrete Reinforcement. Designation A 706/A 706M -
96b.
(21) ASTM - Standard specification for Steel Wire, Deformed, for

Concrete Reinforcement. Designation A 496 - 95a.
(22)ASTM - Standard specification for Steel Welded Wire Fabric,

Deformed, for Concrete Reinforcement. Designation A 497 - 95.
(23) AC1 Standard 31 8-89, Building Code Requirements for
Reinforced Concrete, 1989 Edition
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2747 0542b32 85b
TABLE 1-PARAMETERS DEFINING THE 7c -s-curve 6 AFTER (3)

Parameter good bond area acc. all other bars (top remarks
t o CEB-FIP Model bar effect)
Code 1990
f, iMPa1
usual
concrete
cover and
no stirrups
c narrowly
a = 0.36-+2.00
ds spaced
stirrups
rR
-
-1 0 . 1 5 .max
~
c1 =I 0.0030.f,. + OJ 625
“L.
f, iMPal
s, = 0.25 mm 0.40 mm
s2 = 0.35 mm 0.60 m m
- 1 .O0 rnm 2.50 mm usual
s3 -
concrete
cover and
no stirrup!
s3 = 2.00 mm narrowly
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
spaced
stirrups
TABLE 2-PARAMETERS DEFINING THE Es-curves 1 -!i’),

after (3)
bond good bond area acc. t o CEB- all other bars (top bar effect)
law no. FIP Model Code 1990
Obb2949 0542b33 792
bond stress T
52 s3 slip s
a)
Fig. la-General form of the bond stress-slip relationship after (8)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 1b-Bond stress-slip relationships as a function of the ratio distance from

crack to bar diameter after (8)
m Ob62949 0542634 b29 m
60
50
40
30
20
10
x [cml
Fig. 2-Distribution of steel strains along the specimen length according to
numerical analysis and test, test results after (1O)
steel stress os[Mpa]
O 0.001 0.002 0.003

mean steel strain E~,,, [-I
Fig. 3-Steel stress as a function of strain according to numerical analysis and

test, test results after (11)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542635 565
c
u,
al
c.
rn-
CI
u,
c
al
U
æ
m
.-urn,
-
r
m
E
-mm
n .-u
L
al
5e
O
c.
CI)
.-I=
-E
O
u
u
m
.-c
L
c.
rn
-I-
O
E
.-
w
O
u
e
w-3
Cu
rn
Co
al
E
O
.c
L
m
m
I
h d
(11
&
iz
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0662947 0542636 4 T l W
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
c
al
L
o
c
O
O
U
al
t
w-O
.-c
L
E
Q O
3 wa
.c. c
al rA
a
u)
c
4-a
u) c
c
01 .-O
VY
h c
L
I al
I-
n
l
In
.c.
3
O
x
E
E
-m
c
C
O
(v 2
01 t
II
J=
$
O
Yi
.-c
?!
h
m
= Obb2949 0542637 338 Tribute to Peter Gergely 67
700
600 y /+-I--
-
i
- - i - - - - !
I
/ I
500 - i d
! #
I l
400
i
l
I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
I
300 -
-steel 1, fJfy=l.05,~,,=0.032 (24) ¡
Es [-I
c) stress-strain relationship of reinforcement
Fig. 5c-Tension tests on reinforced concrete walls,

after (13)
68 Eligehausen, Ozboit and Mayer
W O b b 2 î Y ï 0542638 27Y =
0 s [MPal
700
600
500
400
300
200
IO0
O
0,000 0,001 0,002 0,003 0,004
Es [-I
a)
0 s Wal
700
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
600
500
400
300
200
1O0
O
0,000 0,Ol o 0,020 0,030 0,040
1
Es [-I
Fig. 6-Steel stress as a function of strain for specimen 24 reinforced with low
ductile steel, after (13). (a) strains 50.004, b) total strain range
0bb2949 0542639 LOO W
0s [MPaI
700
600
500
400
300
200
1O0
O
O 0,001 0,002 0,003 0,004
700
600
500
400
300
200
1O0
O
0,000 0,030 0,060 0,090 0,120
Es
Fig. 7-Steel stress as a function of strain for specimen 21 reinforced with high
[-I
ductile steel, after (1 3). (a) Strains I 0.004, b) Total strain range
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Obb2949 0542b40
&sm/Esr [-I
1,o0
- i
0,90
0,80
0,70
u 1
0 - test24
-numerical
T
model
fc = 50 MPa
0,60
0,50 u
O ,40 I I L! \,o
"i.
-.I_
1 n
0,30
0,20
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0,lO
0,oo
O 0,01 0,02 0,03 0,04
Esr [-I
1,o0 -
0,90
0,80
C' - - o - testZ1 ---
f, = 50 MPa
l
0,70
0,60
0,50
0,40
0,30
0,20
0,lO
0,OO
O 0,02 0,04 0,06 0,08 0,l 0,12
Esr [-I
Fig. 8-Ratio E,,IE,, as a function of steel strain E,, at the crack according to
numerical model and tests by (13): (a) low ductile steel (b) high ductile steel
m 0662949 0542641 8b9 m Tribute to Peter Gergely 71
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
600
500
400
300
200
I O0
O
0,OO 0,010 0,020 0,030 0,040 0,050 0,060
steel strain E [-]
Fig. 9-Stress-strain relationships of reinforcement assumed in the parametric

study. The reinforcement complies with the minimum requirements for ductility
Classes A, B and S (51, respectively
= 0hb2949 0542b42 7T5 m
steel stress os [MPa]
500
400
300 -_____-
200
1O0
O ~ a)
0,0000 0,0015 0,0030 0,0045
strain E [-I
steel stress G~ [MPa]
600
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
500
400
300
b)
200
-CEB-FIP MC 90
1O0
O
0,oo 0,005 0,010 0,015 0,020 0,025
, strain E [-I
Fig. 1O-Steel stress-strain relationship according to numerical model and CEB-FIP
Model Code 1990 (5). Steel Class A, p= 1.5 percent. a) strains 50.0045, b) total
strain range
0662949 0542643 631 9
0,oo 0,Ol 0,02 0,03 0,04 0,05 0,06

Esr [-]
Fig. 1 1-Ratio E,,IE,,as a function of the steel strain E,, at the crack according
to numerical model and CEB-FIP Model Code 1990 (5)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W O662949 0542644 578 =
steel stress CJ [MPa]
600
500
400
300
200
1O0
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O
0,OO 0,Ol 0,02 0,03 0,04 0,05
steel strain E [-I
1,o0
0,90
0,80
0,70
0,60
0,50
0,40
0,30
0,20
0,l o
0,oo
0,OO 0,Ol 0,02 0,03 0,04 0,05
Esr [-]
Fig. 12-Influence of ultimate steel strain on the contribution of concrete between
cracks a) assumed stress-strain relationships of reinforcement b) ratio E,, ,IE,~ as a
function of steel strain E, at the crack; parameter is E,,
Obb2949 0542b45 1104 m
steel stress G [MPa]
600
500
400
300
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
200
1O0
O
0,OO 0,Ol 0,02 0,03 0,04 0,05
steel strain E [-I
a)
1,o0
0,90
0,80
0,70
0,60
0,50
0,40
0,30
0,20
0,lO
0,oo
6
0,OO
= 5.00%
fy = 500 MPa
‘p = 1.50 %
0,Ol 0,02 0,03 0,04 0,05

Esr 1-1
Fig. 13-Influence of ratio f,lfy on the contribution of concrete between cracks
a) assumed stress-strain relationships of reinforcement b) ratio as a
function of steel strain csrat the crack; parameter is the ratio f,lfy
m Obb2949 0542b4b 340 =
steel 1
A L
/
>
E
Fig. 14-Reduction of steel strain by a constant bond force for reinforcement with
different ratios f,lf,, schematic
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Obb2949 0542b47 287 Tribute to Peter Gergely 77
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0,oo 0,Ol 0,02 0,03 0,04 0,05 0,06

Esr [-]
Fig. 15-Ratio E,,I&,, as a function of the steel strain E, at the crack for a
tension specimen with a reinforcement percentage=I 0.5 percent
Ob62949 0542648 113 W
600 I I I I l
I.' = 1.5%
500 E, = 0.025 ~
!
400
300
200
-fW=35MPa I
1O0 t
d- fW = 45 MPa
I/ I i l -
O
0,oo 0,001 0,002 0,003 0,004 0,005
Es [-I
a)
0,oo 0,005 0,010 0,015 0,020 0,025

Esr [-I
Fig. 16-Influence of concrete strength on the contribution of concrete between
cracks, Steel Class B a) strains I 0.0045, b) total strain range
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2949 0542649 05T = Tribute to Peter Gergely 79
moment M FNm]
40
32 -. 8
I
1-
24 t
i:
- deformed bars
16
1 I
I,' IP -
fc = 25.5 28.1 MPa 6 0 8 mm
O
I= 2000
4 i I !
I
O 0.01 0,02 0,03 0,04 0,05 0.06 ' 0.07

total rotation O [rad]
Fig. 17-Influence of bond of reinforcement on the rotation behavior of a slab

reinforced with Steel Class B, after (15)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
plastic rotation O,,

[mrad]
35
30
25
20
15
10
O
O 02 094 0,6 03 1 1’2 1,4
reinforcement ratio [%I
Fig. 18-Influence of bond on the plastic rotation capacity of plastic hinges,

after (7)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
W 0662949 0542651 708
SP 180-4
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
A Dilatational-Interface Model for Bond
by J. V. Cox
Synopsis: A bond model is presented that uses an interface idealization of the

bond phenomena and incorporates dilation to characterize the wedging effect of
the ribs. Since this type of model can potentially predict both pull-out and
splitting failures, it may provide an approach for characterizing the observed
experimental response of bond specimens in a form that can be used to better
understand the progressive failure of complicated structural components. Using
the mathematical framework of plasticity theory, the model is defined to
characterize the effects of damage in the region near the bar. The form of the
model is based upon experimental results that include a variation of the
confinement stress; thus, the model fully couples the tangent and normal response
so that it: (1) exhibits a sensitivity to the confinement stress and (2) produces
longitudinal cracking in models of bond specimens. Validation problems based
on experiments from several research groups are considered to highlight the
strengths and weaknesses of the model. The model reproduces the experimental
data with acceptable accuracy using a single calibration, but the results also
suggest that the limitations of the interface idealization merit further investigation.
Keywords: Bond model; dilatation; interface; mechanics; phenomenological;

plasticity; pull-out; splitting (cracking); validation
81
82 Cox
= 0bbZï49 054Zb52 b44
At the other "extreme," the member-scale, either a structural component or complete
structure is modeled; this requires a careful selection of the discretization and
element types. The reinforcing is usually treated with a discrete, embedded or
smeared model (12), but other novel approaches to modeling components have also
been developed (see e.g., Filippou (13)). Typically at the member-scale the
reinforcement is treated as a one-dimensional element, and bond laws have been
limited to single-stress models (i.e., relating bond shear stress to slip).
The bar-scale is an intermediate scale where the bar and concrete are both treated as
continuums, and the mechanical interaction of the ribs is homogenized and often
further idealized as an interface phenomenon. Phenomenological models at this
scale can not accurately reproduce the transverse cracking in the bond zone,
however they can have the following benefits over one-dimensional
characterizations: (1) ability to predict splitting failures, and ( 2 ) greater sensitivityto
the concrete stress state -i.e., a greater measure of generality.
The objective of this on-going research is to develop and validate a model that will
qualitatively reflect the mechanics of bond while having greater potential for
application than rib-scale models (e.g., to model structural components). This
objective motivates the use of a "bar-scale" model which will have a
phenomenological nature. The model should require only one calibration (for
standard steel bars and regular concrete) and be capable of reproducing (with an
accuracy comparable to experimental scatter) bond tests results from several
independent research groups.
Some bar-scale models have treated the bond zone completely as a continuum (i.e.,
without explicitly including a strong discontinuity between the bar and concrete);
see e.g., Bresler and Bertero (14), and Pijaudier-Cabot et al. (15). Other models
directly characterize the wedging effect of the ribs relating the normal and
longitudinal components of the interface tractions to their conjugate relative
displacements (dilation and slip). These "two-dimensional" models are the subject
of this study. Among the existing two-dimensional models are those of De Groot et
al. (16), Morita and Fuji (17), Zhiming et al.(l8), and Mainz et a1.(19). While
these models demonstrate the potential of two-dimensional, bar-scale models, they
are not defined to reproduce results from several experimental studies using a single
calibration.
Bond behavior typified by the classic experiments of Eligehausen et al. (6)
motivated the use of elastoplasticity as a mathematical framework for the model
developed in this study. While the terminology of classical plasticity theory is
adopted to describe the model (e.g., yield surface, flow rule, isotropic and
kinematic hardeningísoftening), the approach is distinctly different than applying
plasticity theory to model the constitutive behavior of a material. For a bond model,
the potential slip plane is singular, predefined, and macroscopic.
Mathematical Model
The key components of a plasticity model are definitions of the generalized stresses
and strains, elastic moduli, yield criterion, and flow rule. Complete descriptions of
the model for monotonic and cyclic applications are presented in Cox (20) and
Herrmann and Cox (21), respectively. This paper emphasizes the strengths and
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m O b b Z î Y ï 054-2653 SB0
James V. Cox is an assistant professor of civil engineering at Johns Hopkins

University, Baltimore, MD. His research interests are in modeling of reinforced
concrete, constitutive modeling, and finite element methads.
Introduction
By design, bars are used to reinforce concrete (a brittle matrix) to both prevent and
bridge cracks. As such, bond behavior is important in determining the nature of
localized failures and the amount of energy dissipated by reinforced concrete
components. Structural models that totally disregard the effects of bond, can fail to
predict localized cracking which can have global ramifications. The corresponding
stiffness degradation increases the period of vibration, decreases the energy
dissipation capacity, and can result in a significant redistribution of internal forces
(see e.g., Bertero (1)).
For plain bars (without ribs), "adhesion" and fiction are the principal mechanisms
of bond. The apparent qladhesion''has contributions from chemical bonding and the
effect of shrinkage stresses that develop during curing. The failure of this
"interaction" is characterized by the initiation and propagation of an interfacial
crack. (Except for this case, the use of the term "bond" is an accepted misnomer
that is commonly used to describe the total interaction of the reinforcement with the
concrete.) For ribbed-bars, these mechanisms are secondary (by design) to the
mechanical interaction of the ribs with the surrounding concrete. Adhesion breaks
down relatively early in the bond response, and subsequently the bond force is
transferred by friction and the mechanical interaction of the ribs with the adjacent
concrete. With increasing bar force, the mechanical interaction dominates the
transfer of force which is now concentrated near the rib faces. At this point, "bond
stress" usually refers to an average force per unit area. Increased loading will begin
to fail the concrete near the ribs in two ways: crushing of concrete adjacent to the
contact area and transverse cracking (typically cone-shaped cracks - also called
secondary or bond cracks) that initiate at the ribs. The extent of these cracks
conceptually defines the size of the bond zone. More detailed explanations of the
mechanics of bond are available in the literature (see e.g., R e h (2), Lutz and
Gergely (3), Goto (4), Hungspreug (9, and Eligehausen et al. (6)).
The cumulative effect of the "bond mechanisms" has been directly modeled at
several scales. Three descriptive scale names are adopted here to classi@ bond
models. Rib-scale analyses, usually include an explicit discretization of the ribs on
the bars. Among the difficulties in modeling bond at this scale is that the scale is
significantly smaller than the "unit cell for concrete;" that is, much of the aggregate
is larger than the rib so the homogeneous concrete assumption is idealistic. In
addition, rib-scale bond models are no better than the concrete model's ability to
model hcture and crushing failures. While analyses at this scale have not provided
a "general analysis capability," they have provided much insight to the mechanics of
bond - especially in the early response. Interesting examples of rib-scale analyses
include the studies of Hungspreug (9, hgraffea et al. (7), Reinhardt et al. (8),
Rots (9),Ozbolt and Eligehausen (lo), and Brown et al. (1 1).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
84 Cox
ûbb2949 0 5 4 2 b 5 4 417 =
weaknesses of a particular two-dimensional, bar-scale model. Thus for brevity, the
following description only presents a qualitative overview of the model for
monotonic loading that is sufficient for interpreting validation results. While the
model may be applicable to nonstandard bars (Le., with experimental rib patterns or
epoxy coatings), this study only addresses standard steel bars.
Generalized Stresses and Strains
The conceptual region called the "bond zone" bounds the crushing and transverse
cracks produced by the mechanical interaction but does not contain: (1) all
longitudinal cracks, some of which can split a test specimen or the cover of a
flexural member, or (2) transverse cracks that produce a local failure such as a pull-
out cone. In the modeling approach taken here, these large scale cracks must be
modeled by the global representation of the structural component. The growth of
transverse cracks during a progressive failure of bond suggests that the size of the
bond zone evolves, so most definitions of the bond zone are for the maximum
radius of the region. Researchers have suggested different radii for the bond zone.
For example, Gambarova et al. (22) experimentally determined that the
microcracking extended the thickness of the maximum aggregate size.
Bar-scale bond models are phenomenological characterizations that represent "all"
or "pari of' the behavior of the bond zone. In the context of finite element
modeling, for a bond model to represent all of the behavior of the bond zone it
would have to be applied over the complete volume of the bond zone. It is more
common to characterize the bond phenomena as if they are constrained to an
interface. Since the thickness of the bond zone is significant compared to the bar's
diameter this is clearly a modeling idealization. Gerstle and Ingraffea (23)
emphasize this by suggesting that, "bond slip may not exist in the sense in which it
has been previously described in the literature." This is also suggested by many
rib-scale analyses that reflect the importance of transverse cracking in bond (see
e.g., Reinhardt et al. (8), Hungspreug (5), and Rots (9)). Interface descriptions of
bond are a pragmatic idealization of a very complicated problem, but this
idealization is potentially very useful in larger scale analyses.
In addition to the interface idealization, the version of the model presented here
assumes that the bond zone response can be approximated as axisymmetric. This
does not suggest that the response outside of the bond zone is necessarily
approximately axisymmetric,but it does preclude application of the model to some
problems (e.g., where significant doweling forces exist). This assumption also
implies that the detailed effects of (1) nonaxisymmetric rib patterns (Le., all rib
patterns except those that have ribs normal to the axis of the bar), (2)
nonaxisymmetric cracking of the concrete (e.g., where a small number of dominant
longitudinal cracks occur), and (3) material heterogeneity are implicitly averaged in
defining the generalized stresses and strains for the model.
The interface idealization of bond can be described in terms of two simplifications.
The first, common to all bar-scale models, is the homogenization of the interface
traction distribution and the simplification of the interface geometry. Consider a
I state where no adhesion remains between the bar and the concrete (i.e., mechanical
I interaction dominates the bond behavior). Figure la presents a schematic of the
horizontal component of the interface traction (shown on the concrete) between the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
' II Ob62949 0542b55 353 W
bar and concrete for a unit surface element. The unit surface element consists of the
complete bar surface (due to the axisymmetry assumption) for one cycle of the rib
geometry. In this case, the ribs are assumed to be perpendicular to the bar's axis,
so the length of the unit surface element is the rib spacing (sr). (A cylindrical
coordinate system is assumed with the z-axis corresponding to the axis of the bar.)
Figure 1b presents a schematic of the horizontal component of the interface traction
between the bar and concrete for a bar-scale idealization. The uniform distribution
of z shown in Figure l b ("macroscopically homogeneous") would occur when the
actual traction distribution for adjacent unit surface elements is identical to that
shown in Figure la. In conjunction with the simplification of the traction
distribution, the actual surface geometry is also idealized as a cylindrical surface.
Though not depicted in Figure 1, the actual interface geometry changes with
concrete crushing and cracking. The homogenized stress is deñned so that the
average of each traction component in a Cylindrical coordinate system is the same;
for example
?
where A denotes the area of the unit surface area, and T, denotes the z-component
of the actual interface traction.
The second simplification, unique to interface idealizations, addresses the
kinematics of a unit cell of the bond zone. Figures IC and Id give schematics of the
deformation of the actual unit cell versus the deformation of the same unit cell in the
bar-scale model. For the two unit cells, note the difference in the distribution of the
horizontal displacement (e.g., along the lefi side) relative to the top of the cell.
Eliminating the ribs of the bar and the corresponding traction concentrations- the
first simplification -produces a different response in the concrete matrix, even for a
"perfect concrete model." While transverse cracking can still occur in the bar-scale
model, the omission of the ribs is likely to produce less cracking and a more
uniformly distributed cracking pattern. Defining the bar-scale model so that it will
produce similar bar displacements requires a concentration of relative displacement
at the interface. That is, reduced "shear deformation" of the concrete in the bar-
scale model is supplemented by an increase in the strong discontinuity of
displacement at the bar-concrete interface.
The generalized strains for the bar-scale model are defined to be the tangent and
normal displacements (4and 4, respectively) of the concrete relative to that of the
bar interface nondimensionalized by the bar diameter (Ob).The genedized stresses
and strains are thus defined as
where the tangent and normal directions correspond to the z- and r-directions,
respectively, and the 6s are defined for points on the interface that are coincident in
the undeformed state. (Figure Id depicts a deformation where &<O and &>O.)
The generalized stresses and strains are work conjugate within a multiplicative
Constant, If&.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
86 Cox
= 0662949 0542b5b 29T
Though this is not a unique strain definition, this particular form has effectively
incorporated the diameter of the bar for a variety of test results. Morita an Fujii (1 7)
normalized their definition of slip in this manner for their uniaxial model and
demonstrated its appropriateness for their test results. The strain measure is the
first of two definitions that includes a characteristic length related to the rib
geometry. Unlike the case of smooth bars where the propagation of an interfacial
crack is principally a function of the material properties, for ribbed bars the bond
response is dominated by the mechanical interaction that the rib geometry affects.
Characteristic lengths for the rib pattern in the tangent and normal directions are the
rib spacing and height, respectively. For standard steel bars, both of these lengths
are nearly proportional to the bar diameter, and thus a single parameter of bar
diameter was adopted for this study. Other rib geometry parameters (e.g., rib angle
-the angle of the rib with respect to the bar's axis) have also been considered (2 1).
An additive decomposition of the generalized strains (and thus of the relative
displacements)is adopted such that
y + QP
4= c (3)
where CJ~ and gP denote the elastic and plastic strain components, respectively. (The
same notation is adopted for the elastic and inelastic contributions to the relative
displacement components - slip and dilation.)
A thorough discussion of the potential limitations of bar-scale models is beyond the
scope of this paper, however a few of the limitations that relate to the deformation
of the bond zone should be noted. Comparisons between models and experiments
are usually based upon global responses such as bar force and end slips - partly
because it is difficult to measure local quantities. Any slip measurement that is
defined as a relative displacement between the surface of the bar and a reference
point within the bond zone may be predicted inaccurately by a bar-scale model. The
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
potential inaccuracy is due to the difference in the deformation of the actual bond
zone versus that of the bar-scale model (reference Figure 1). However, the same
model can accurately predict slip based upon a reference point outside of the bond
zone. The influence of the deformation of the concrete matrix might require the
bond model to be recalibrated for different concrete models; this is currently being
investigated.
Since a bar-scale characterization inherently ignores the details of the transverse
cracking, some observed behaviors are not easily reproduced by such models. For
example, detailed rib-scale behaviors that produce the arrest and propagation of
different transverse cracks (see e.g., Rots (9)) is beyond the scope of bar-scale
models; this suggest that the "conceptual unit-cell" might need to consist of several
rib spacings. Other non-local effects related to the interaction of transverse cracks
with one another and with boundaries (Le., phenomena that might be considered
nonlocal in the context of the bar-scale idealization) are also potentially difficult to
capture. Examples might include: (1) the interaction of transverse cracks produced
by adjacent bars, and (2) the interaction of transverse cracks with a free boundary.
The former might produce a change in bond response with bar spacing, and the
latter might produce a change in bond response with the distance from a free surface
or ultimately a cone pull-out failure. The extent to which a concrete constitutive
model can reproduce some of the effects of transverse cracking with a bar-scale
bond model has not been fully established. A non-local, bar-scale model might be
able to characterize some of these effects (e.g., the interaction of transverse cracks
- - U b b i S 4 9 0542b57 226
produced by adjacent bars), but this study is limited to models that relate
generalized stresses and strains at a point - a local description.
Elastic Moduli
The "elastic response" deviates from elastic behavior since it includes the effects of
the underlying contact problem between the ribs and adjacent concrete after the
breakdown of adhesion. Analysis of the experimental data of Malvar (24,25)
suggests very little elastic coupling occurs between the tangential and normal
response (thus the small off-diagonal terms given below). This is consistent with
previous experimental observations (e.g., Lutz and Gergely (3)) that indicated
significant wedging action of the ribs occurs only after preliminary crushing near
the rib face. Experimental results also indicate the presence of elastoplastic
coupling which is not accounted for in the current model.
For the current model(20), the elastic response under monotonic loading is
characterized by the following incremental moduli
o. 1
-0.0012 s e ( & )
-0.0012 Sgn(6J
0.04 1
where sgn(O)=O, and E, is the elastic modulus of concrete. The sgn(af)factor
reflects the contact character of the interface between the inclined rib surface and
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the adjacent concrete, accounting for: the slip direction and the contact symmetry at
zero slip (Le., an incremental decrease in 6, does not produce a generalized shear
stress when the slip is zero). The implementation accounts for finite precision and
treats 6,as zero if IS&, where E denotes a very small tolerance.
Magnitudes of the individual moduli were obtained principally fiom the

experimental data of Malvar (24,25), Gambarova et al. (22), and Eligehausen et al.
(6). As noted previously for the total response, the elastic response in the model is
the difference between the elastic response of the actual unit cell and the response of
the concrete in the bar-scale unit cell (Figure I). Thus, the response is attributed
principally to the local interaction of the ribs and adjacent concrete, which is
consistent with the observed variation of the elastic response with rib geometry.
The elastic modulus, E,, provides a simple measure of the effect of the concrete's
elastic properties upon the elastic response of the model.
Yield Surface
The form of the yield surface and it's evolution was determined from the
experimental data of Maivar (24,25) by making some simplisring assumptions.
The first assumption was that a single "damage measure", given by d =
min(~/sr,l),could be used to characterize the evolution of the yield surface. The
maximum value of 1 reflects the observation (see e.g., the Eligehausen et al. (6))
that when SFSS, the bond is completely failed and only fictional resistance remains.
Maivar's test series consisted of 5 small cylindrical specimens, each with different
0662949 0 5 4 2 b 5 8 Ob2 H
88 Cox
normal tractions applied to their outer cylindrical surfaces. Malvar identified each
test within a series by the value of -cr (the confinement stress) that would occur if
the specimen did not carry hoop stress; this value is only representative of -owhen
several splitting cracks are open, but it was assumed to adequately reflect the value
of cr for the first phase of the model development. The last simplification was that
the effects of differing rib geometries for standard steel bars were neglected.
With these assumptions, data for the yield-surface and its evolution was extracted.
The following yield criterion was proposed to represent the experimental data
where ft is a measure of the concrete tensile strength, and M is a calibration

constant. The evolution of the yield surface is characterized by three functions: C
and 6 are the isotropic and kinematic hardeningísoftening functions, respectively;
and We is the weighting function for the exponential component of the yield surface
that characterizes the "friction-dominated response." Consistent with the
assumptions made in extracting the yield surface data, for the monotonic model C ,
&, and We are only functions of d.
Figure 2 shows the graphs of the evolution functions, and Figure 3 illustrates the
corresponding evolution of the yield surface from its initial state to the purely
frictional state. The yield function initially has a power form (Le., W,=O) and then
with increasing damage transitions to an exponential form. Figures 2 and 3 show
that isotropic hardening (increasing C) and kinematic softening (decreasing 8)occur
simultaneously. The former can be attributed to the stable propagation of transverse
cracks andior concrete crushing near the rib-face. The latter can be attributed to the
geometric effect of a rib wedging under the concrete and reducing the effective
contact angle. The validation results presented in the next section will show that the
simultaneous combination of hardening and softening allows a large variation in the
"ductility" of the bond response to be reproduced. The mechanisms associated with
the subsequent isotropic softening include crushing, fracture and friction. The small
change in C for &0.7 reflects that friction is the dominant mechanism "in the end."
Flow Rule
The wedging action of the ribs at the rib-scale is kinematically accounted for in the
bar-scale model through the flow rule, which produces radial dilation and
contraction of the interface. The experimental data of Malvar (24,25) and
Gambarova et al. (22) were examined in developing the flow rule, but little data
were available so the flow rule's current form must be regarded as tentative.
Analysis of these data, with a homogenized view of the bond response, indicated
that a nonassociative flow rule is necessary. That is, as with frictional materials, a
g p normal to the yield surface would over-predict the dilatational response. Since
friction contributes to bond behavior, the need for a nonassociative flow rule was
expected. Other important experimental observations are: (1) the rate of radial
dilation decreases with confinement stress (-o), ( 2 ) most of the dilation occurs near
the ultimate strength, (3) radial contraction occurs after the radial dilation followed
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ObbZ9!47_0592$5_9T T 7
by a purely friction response, and (4) radial response changes with rib-geometry.
As for the yield surface, the effects of differing rib geometries are neglected in the
initial development. The other experimental observations should be reproduced by
the mathematical model. In particdar, observations (i) and (2) are important for
predicting longitudinal cracking.
The following form was adopted to simpliQ the flow rule description:
where denotes the consistency parameter, which k.positive during plastic loading
and zero otherwise. Since q f has a magnitude of il,g denotes the rate of plastic
dilation with respect to plastic slip. To obtain an approximation for g, limited data
presenting radial dilation versus slip (24,25) were analyzed. Consistent with
assumptions made in obtaining the yield surface, we assumed that: (1) g is only
dependent upon -crand d; (2) the -0value reported by Malvar (24,25) is the normal
stress at the bar; and (3) the radial elastic contraction due to the applied confinement
stress is not significant. With these assumptions, approximations to g were
determined from derivative estimates based upon the experimental data of Malvar.
Figure 4 presents experimental and model approximations of g for fixed values of
cr. The model is defined over four stages: (1) radial dilation, (2) constant radius,
(3) radial contraction, and (4) sliding. As reflected in the model's response, the
shape of the graph of g is similar for each value of -0. For the model, the limits
and description of each stage are, generally, confinement stress dependent.
While the yield surface and flow rule are independently defined, their combination
must characterize the mechanics of the bond behavior, for the model to have a
measure of "generality." The combined behavior is important for most bond
specimens. Typically radial dilation is constrained by concrete cover or secondary
reinforcement, so the radial dilation increases the normal stress which affects the
bond stress. This is significantly different than a model which merely scales the
bond stress as a function of the normal stress; with radial dilation incorporated in
the flow rule, the model "actively participates" in determining the local normal
stress state. As such, the model can potentially predict splitting as well as pull-out
modes of failure. The cost of this benefit, is that the model needs accurate normal
stress data to give the correct response.
CALIBRATION AND VALIDATION
This section presents selected calibration and validation results. Since "validation"
in an inherently incomplete process, the pimipie objectives are to establish some of
the capabilities and limitations of the model. First, some results that were presented
previously (20,26) are shown to establish the scope of the validation study; these
results are not discussed in detail. The remaining results focus on some of the
apparent strengths and weaknesses of the model with an emphasis on the "radial
response."
An inherent difficulty to examining bond behavior is that it can not be

experimentaily isolated, it is integrally tied to bond specimens. Thus, to simulate
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
90 cox
the response of a given test requires a model of the bond specimen. While two-
dimensional finite element analyses are used for specimens with large embedment
lengths, some specimens with short embedment length were modeled using
"strength of matenal descriptions" of the specimen behavior. Because these models
do not have a spacial discretization associated with them, for brevity they are
referred to as point models. The point models were implemented as special
"structural models" for nodal link elements. All of the specimens modeled as two-
dimensional are treated as being axisymmetric except for the specimen of
Gambarova et al. (22).
A simple linear elastic brittle material model was used to model the concrete of the
axisymmetric specimens. The model allows longitudinal cracks to initiate based
upon a maximum tensile strength criterion, but strain softening is not included in
the formulation. The specimen of Gambarova et al. (22) is modeled as plane stress
and linear elastic. Admittedly these are simple specimen models which presume
that most of the nonlinearity is captured by the bond model. Because "bar-scale''
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
models eliminate some of the complicated behavior that actually occurs in the bond
zone, it is not clear that sophisticated concrete models are needed for pull-out
specimens. While these simple specimen models may be sufficient for the pull-out
specimens considered in this study, more sophisticated concrete models are being
considered in our current research so that: (1) tensile specimens can be considered
in future studies, and (2) the dependence of the bond specimen model's behavior
upon the concrete constitutive model can be examined.
Figures 5 presents calibration results for Malvar specimens with three different
levels of confinement stress. Consistent with the model development, a point
model is used that does not account for the specimen response. The model is able
to reproduce two important trends with the increase in confinement stress: (1) an
increase in the bond strength, and (2) an increase in the "ductility" of the response.
Figures 6 through 9 present validation results for several different experimental
studies. The only model parameters changed for these results were Db, s,.?fr,and
E,. The values ofA, (6) and E, were usually obtained from the formulas (in units
of MPa)
fi=O.3lfCpi and Ec=41000(l-e-0.075~) (70)
The range of specimen characteristics for the results shown are: (1) bar diameter -
14 to 30.7 mm, (2) compressive strength - 18.5 to 40.2 MPa, (3) specimen
"diameter" - 7.62 to 50 mm, and (4) embedment length - 3 to 20 Db. All of the
results are within an acceptable accuracy for the first generation of the model. Only
the specimen of Eligehausen et al. (6) was modeled with a point model.
The largest difference between the experiment and model occurs with the 20Db-
embedment specimen shown in Figure 9a. It is not clear whether the principal
source of the difference is due to the bond model, specimen model, or experimental
uncertainties. As shown, the difference is not significant until the bar force exceeds
the original yield strength of the bar (i.e., prior to the authors' heat treatment). The
bond model exhibits a linear relationship between pull-out force and embedment
length. This behavior follows from the almost constant bond stress distribution
predicted by the model that occurs near the maximum bar force. Figure 1O shows
the model predictions for the evolution of the bond stress (z) and radial dilation (6,)
0662949 0542663 0 7 m
Tribute to Pater Gergely 91
distributions. The contraction of the interface at the free end is not physically
correct. This inconsistency results from the current definition of De;in particular,
O& appears to be too small at the free end. While many bond experiments give an
i d r e c t measure of the magnitude of Di (the tangent elastic stifhess), there is little
data available to base the n o d elastic response upon. intuitively, one would
expect O$ to increase with an increase in the contact area between the bar and the
concrete &us preventing the contraction exhibited by the model. A preliminary
attempt to incorporate a variation of Dg with contact into the model exhibited
improved results (29), but the approaci has not been fully developed. If the
contraction were prevented at the free end of the bar, the normal stress and thus
bond stress would be higher in this region - a trend exhibited in Shima et als' data.
Ail of the validation results reflect on both the tangential and normal response of the
model, since they are fully coupled in the formulation; however, let's consider
some results that indicate the potential of this type of model to produce longitudinal
cracking in the adjacent concrete. Probably the most important observation is that
each specimen model exhibited the same failure mode as the experiment -pull-out
or splitting.
First let's consider the behavior of an axisymmetric model of the Malvar specimen.
Figure 11 gives a schematic representation of the axisymmetric model. F i p 12
shows the bond stress versus slip response for the case of "-0=17.2 N/mm (2500
psi)." All of Malvar's specimens were loaded until a splitting failure occurred prior
to performing the bond test. Simulating the splitting phase of the loading (with
-0=3.45 N/mm2- 500 psi) gave a residual end slip of 0.22 mm that is near the high
end of Malvar's experimental scatter r0.05 mm, 0.24 mm]. The radial dilation was
also under-predicted in this analysis which was part of the motivation for
readdressing the radial elastic response, as discussed above.
The tests of R e h and Eligehausen (27) and Shima et al. (28) did not fail in
splitting. Figure 13 shows the model's predictions of the maximum extent of the
longitudinal cracking in two of these specimens.
Tepfers and Olsson (30) developed a cylindrical specimen cast within a steel tube to
measure the splitting tendency of different rib shapes. The thin tube had strain
gages centered on opposite sides to determine the hoop strain in the tube. The hoop
strain measurements were assumed to provide an indirect measure of the average u
when the splitting cracks were open. As is common with short-embedment length
specimens, the average bond stress (z) was obtained from the bar force.
With estimates of both components of the interface traction, the authors presented
plots of the angle of the interface traction with respect to the axis of the bar - the
traction angle. (A zero traction angle corresponds to a tangential traction - bond
stress only.) Figure 14a shows comparisons of measured traction angles with
those obtained from the models. Two analyses are conducted for each experiment:
one withf, equal to the cube splitting strength (the larger values), and one withf,
calculated fiom equation (7a). With increasing values of end slip, the model and
experimental results show better agreement. The model results for test 3, which
has measurements for larger slip values, shows excellent agreement with the
measured values at larger slips.
The models' traction values at smali slips significantly over estimate the measured
values. The source of the difference can be partially attributed to the simplicity of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
92 Cox
= 0hb2949 0542662 593
the specimen model. Let's consider three possible contributions: (1) the
axisymmetry approximation, ( 2 ) strain softening in the concrete, and (3) bond of
the steel ring to the concrete. Axisymmetry only provides an approximate
description of the state in any cylindrical specimen, but this approximation could be
very significant towards interpreting the hoop strain in the ring. As the number of
longitudinal cracks increases the axisymmetry approximation improves, but the
amount of hoop strain necessary to soften the cracks increases too. In either case,
the models' reduced hoop strain estimate in the concrete increases the strain in the
tube, and thus increases the traction angle. The axisymmetry assumption also
implies that the bond between the steel tube and the concrete does not affect the
results. If the number of longitudinal cracks that breach the edge of the actual
specimen is small and the strain gages lie between these cracks, the strain
measurement would under estimate the traction angle until the steel is free to slip
relative to the concrete. Thus the significant differences that occur at small slips,
may simply reflect the limitations of the specimen model rather than inaccuracies in
the bond model.
Figure 14b presents the last comparison of the model and experimental results.
Tepfers and Olsson's (30) test series 4 consisted of 5 "identical tests." The model
over estimates the average bond strength. For5 based on the equation (7a) and the
measured cube splitting strength, the mean of the tests is over estimated by about 14
and 11 percent, respectively. The most significant difference might be the apparent
excessive stiffness of the model for small slips. This validation test may reflect
some of the limitations of the bar scale model. This specimen is among the smallest
considered, and it does not have an unbonded segment of concrete on the loaded
end (unlike all the other specimens). Though the reaction force due to the teflon-
covered bearing ring provides some confinement, it may produce a significantly
different stress state than a specimen with an unbonded segment of concrete. Thus,
the reduced stiffness of the bond response may be partially due to the interaction of
the bond mechanisms (e.g., transverse cracks) with the boundary. This type of
effect would be more pronounced with a free boundary (e.g., at a fully-developed
primary-crack), and thus the model is unlikely to accurately reflect these types of
local variations in bond response which are inconsistent with the "unit cell
characterization" of the problem. To what extent a more sophisticated concrete
model would overcome this problem is unclear, since the details of the mechanical
interaction are not addressed at the bar scale.
CONCLUSIONS
Two-dimensional bar-scale (or dilatational-interface) models for bond reflect the

disparity between the scale at which the basic mechanics of bond occur and the
scale of reinforced concrete structures. The common interface idealization
inherently (1) simplifies the interface geometry, (2) smoothes the actual interface
tractions, and (3) concentrates some of the kinematic response at the interface.
Eliminating the details of the mechanical interaction of bond has several potential
I limitations including an inability to predict the detailed response near the bar, which
can be affected by bar spacing and boundaries (e.g., the face of a primary crack).
The key strength of this type of model is that they can be defined such that they (1)
are sensitive to and ( 2 ) actively "participate" in determining the stress state of the
adjacent concrete (via D and z). This provides the potential to predict both pull-out
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2q4q 42T '- Tribute to Peter Gergely 93
and splitting failures at a scale that could allow structural component behavior to be
examined.
While the models of the bond specimens were simple, the models reproduced the
experimental results from several independent studies with acceptable accuracy.
Direct measures of the radial response of bond specimens are limited, thus the
related components of the model are difficult to characterize and are considered to
be in preliminary forms. The radial contraction of the interface for a long
embedment length specimen demonstrated the need for refining the radial elastic
response. Nonetheless, most the radiai responses and corresponding splitting
response of the concrete showed acceptable agreement with the available data -
predicting both puíl-out and splitting fäilures consistent with the actual tests.
REFERENCES
1. Berîero, V.V. Seismic Behavior of Structural Concrete Linear Elements (Beam

and Columns) and their Connections, Comité Euro-International due Béton
(CEB), Bulletin D'Information No. 131,Paris,France, 1979.
2. Rehm, G., The Basic Principle of Bond between Steel and Concrete, Deutscher
Ausschuss fur Stahlbeton, no. 138, Wilhelm Ernest and Sohn, Berlin, (C &
CA Library Translation no. 134, 1968), 1961.
3. Lutz, L.A., and Gergely, P. , "Mechanics of Bond and Slip of Deformed Bars
in Concrete," AC1 JOURNAL, Proceedings V 64,1967, pp. 71 1-721.
4. Goto, Y ., "Cracks Formed in Concrete around Tension Bars," AC1 JOURNAL,
Proceedings V 68, No. 4, 1971, pp. 244-251.
5. Hungspreug, S., Local Bond Between a Steel Bar and Concrete Under High
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Intensity Cyclic Loading, Ph.D. dissertation, Cornel1 University, Ithaca, NY,

1981.
6. Eligehausen, R., Popov, E.P., and Bertero, V.V., Local Bond Stress-Slip
Relations of Deformed Bars Under Generalized Excitations, Report
UCB/EERC-83/23, University of California, Berkeley, CA, 1983.
7. Ingraffea, A.R., Gerstle, W.H.,Gergely, P., and Saouma, V., "Fracture
Mechanics of Bond in Reinforced Concrete," Journal of Structural Engineering,
ASCE, V. 110, NO. 4, 1984, pp. 871-890.
8. Reinhardt, H.W., Blaauwendraad, J., and Vos, E., "Prediction of Bond
between Steel and Concrete by Numerical Analysis," Material and Structures,
V. 17, NO. 100, 1984, p ~ 31. 1-320.
9. Rots, J.G.Computational Modeling of Concrete Fracture, P1i.D. dissertation,
DelR University of Technology, Delft, 1988.
10. Ozbolt, J., and Eligehausen, R., "Numerical Simulation of Cycling Bond-Slip
Behavior," Bond in Concrete, Proceedings of the International Conference,
CEB, 1992, pp. 12.27-12.33.
~
W Ohia2949 0542bb4 3bb W

94 cox
11. Brown, C.J., Darwin, D., and McCabe, S.L., Finite Element Fracture
Analysis of Steel-Concrete Bond, SM Report No. 36, Department of Civil
Engineering, University of Kansas, Lawrence, KS., 1993.
12. ASCE Task Committee on Finite Element Analysis of Reinforced Concrete
Structures, State-ofthe-Art Report on Finite Element Analysis of Reinforced
Concrete, ASCE special publications, 1982.
13. Filippou, F.C., "A Simple Model for Reinforcing Bar Anchorages under Cyclic
Excitations," Journal of Structural Engineering, ASCE, V. 112, No. 7, 1986,
pp. 1639-1659.
14. Bresler, B., and Bertero, V.V. "Behavior of Reinforced Concrete Under
Repeated Load," Journal of the Structures Division, ASCE, V. 94, No. ST6,
1968, pp. 1567-1590.
15. Pijaudier-Cabot, G., Mazars, J., and Pulikowski, J., "Steel-Concrete Bond
Analysis with Nonlocal Continuous Damage," Journal of Structural
Engineering, ASCE, V. 117, No. 3, 1991, pp. 862-882.
16. De Groot, A.K., Kusters, G.M.A., and Monnier, T., Numerical Modeling of
Bond-Slip Behavior, Heron, V. 26-1 b, I.B.B.C. Institute TNO, Delft,
Netherlands, 1981.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
17. Morita, S., and Fujii, S., "Bond-Slip Models in Finite Element Analysis,"
Proceedings of the Japan-US Seminar on Finite Element Analysis of Reinforced
Concrete Structures, ASCE, 1985, pp. 348-363.
18. Zhiming, T., Hueizhong, L., and Jinping, Z., "A New Bond Model for Finite
Element Analysis of Structures," Bond in Concrete, Proceedings of the
International Conference, CEB, 1992, pp. 12.9-12.16.
19. Maim, J., Stöckl, S., and Kupfer, H., "FE-Calculations Concerning the Bond
Behavior of Deformed Bars in Concrete," Bond in Concrete, in Proceedings of
the International Conference, CEB, 1992, pp. 12.17-12.26.
20. Cox, J.V., Development of a Plasticity Bond Model for Reinforced Concrete -
Theory and Validationfor Monotonic Applications, Technical Report TR-2036-
SHR, Naval Facilities Engineering Service Center, Port Hueneme, CA, 1994.
21. Hermann, L.R., and Cox, J.V., Development of a Plasticity Bond Model for
Reinforced Concrete - Preliminary Calibration and Cyclic Applications,
Contract Report CR 94.001 -SHR, Naval Facilities Engineering Service Center,
Port Hueneme, CA, 1994
22. Gambarova, P.G., Rosati, G.P., and Zasso, B., "Steel-to-Concrete Bond after
Concrete Splitting: Test Results," Materials and Structures, V. 22, No. 127,
I
1989, pp. 35-47.
23. Gerstle, W.H., and A.R. Ingraffea, "Does Bond-Slip Exist?" Micromechanics
of Failure of Quasi-Brittle Materials, Proceedings of the International
Conference, Albuquerque, NM, 407-416 (1990)
obb2q4q 0542bb5 'TZ
24. Malvar, L.J. Bond of Reinforcement under Controlled Conjînement, Technical

Note 1833, Naval Civil Engineering Laboratory, Port Hueneme, CA, 1991.
25. Malvar, L.J., "Bond of Reinforcement under Controlled Confinement," ACZ
Materials Journal, V. 89, NO. 6, 1992, pp. 593-601.
26. Cox, J.V., and Hermann, L.R., "Confinement Stress Dependent Bond
Behavior, Part II: A Two Degree of Freedom Plasticity Model," Bond in
Concrete, Proceedings of the International Conference, CEB, 1992, pp. 1 1 . 1 k-
1 1.20.
27. R e h , G., and Eligehausen, R., "Bond of ribbed bars under high cycle
repeated loads," AC1 Journal, Proceedings V. 76, No. 2, 1979, pp. 297-309.
28. Shima, H., Chou, L., and Okamura, H., "Micro and Macro Models for Bond
in Reinforced Concrete," Journal of the Faculty of Engineering, í?te University
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O f TO@O(B), V. 39, NO.2. 1987, pp. 133-194.
29. Cox, J.V., "Elastic Moduli of a Bond Model for Reinforced Concrete,"
Proceedings of the 1Ith ASCE Engineering Mechanics Specialty Conference,
1996, pp. 84-87.
30. Tepfers, R., and Olsson, P., "Ring Test for Evaluation of Bond Properties of
Reinforcing Bars," Bond in Concrete, Proceedings of the International
Conference, CEB, 1992, pp. 1.89-1.99.
W Obb2949 0542bb6 139 m
96 Cox
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 1 -Model idealizations: Idealized distribution of the horizontal traction
component over the unit surface element: (a) actual, and (b) bar-scale model.
idealized deformation of the bond zone: (c) actual, and (d) bar-scale model
7..: ...:. ' ' ' ' ' ' ' ' ' ' a ' ' a '
:.
'
..
6i -0..
5- ..... .... w
a-.
...
4- -...
.e... ...
3- .'.a..
-..___
.. .
.-.___
Fig. 2-C, 6 and We versus d
= 0bb25’49 0542bb7 0.75
-25 O 2.5 - ~ l f 5, 7.5 10
Fig. 3-Yield surface evolution
Fig. 4-Model’s flow rule, g, compared with experimental data
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~ ~
W Obb2949 0542bb8 TOI
'0 2 4 6 8 10 12
End Slip (mm) Average Dibtion (mm)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 5a-Calibration results for 2 confinement stresses of the Malvar (24,25)
tests. (a) for- o = 17.2 Nlmm2 (2500 psi)
A v e m Dilation (mm)
Fig. 5b-Calibration results for 2 confinement stresses of the Malvar (24,25)

tests. (b) for- o = 31.O Nlmm2 (4500 psi)
= 0662949 0542bb9 948 D Tribute to Peter Gergely 99
1 EndSlip(mm) 3 4
Fig. 6-Model versus experimental results of Gambarova et al. (22)
Fig. 7-Model versus experimental results of Eligehausen et al. (6)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662747 0 5 4 2 b ï û bbT I
100 cox
Fig. 8-Model versus experimental results of Rehm and Eligehausen et al. (27)
f ..(b) i
...................................... :.. ..........
t /---7-.*m :
............. j .................... ...................
................
P 4
....................
Fig. 9-Model versus experimental results of Shima et al. (28): (a) bar force versus
end slip and (b) n versus "slip" for a point 24, from the free end (bonded
length = 60,)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62947 0542b7L 5 T b . m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 1 O-Predicted distributions of n and 6"for Shima et al. (28) Specimen 4
I prescribed specimen confinement stress I
Fig. 1 1-Axisymmetric FEM model of the Malvar (24,251 specimen
W Obb2949 0542672 432 9
102 cox
4 1 : i 4
i............... i ...............i ............... < ............
............ j ____....._.....
Fig. 12-Axisymmetric specimen model versus experimental results of Malvar

(2425)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
i s l
Fig. 13-Predicted extent of longitudinal cracking in the specimens of: (a) Rehm
and Eligehausen (27) and (b) Shima et al. (28)
m 0662749 0542673 379 ,m Tribute to Peter Gergely 103
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 14-Model uersus experimental results of Tepfers and Olsson (30): (a)
traction angle versus end slip, and (b) zaveversus end slip for Test Series 4
Obb2949 0542674 205 m
SP 180-5
Tension Stiffening and Cracking Behavior in

High-Strength Concrete
by 6. Creazza, R. Di Marco, S. Russo and E. Siviero
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Svnopsis: This paper describes a theoretical and experimental analysis designed to
characterize the initial branch of the bond-stressíslipping curves (T-s) for normal-
strength and high-strength concretes. The theoretical analysis is used to interpret
the results of experimental trials on reinforced concrete ties prepared with class 50
and 100 iMPa concrete mixtures and submitted to tensile forces without inducing
aiìj. yield in the bar. 'lhe purpose of the investigation was to study any changes in
bond behavior (over the limited range of slipping values considei.cd) due to the
better mechanical features of the 100 MPa coiicrete, and thus contribute to a better
understanding of how high-strength reinforced concrete elements behave in a ser-
viceability state.
Keywords: Bond stress; crack width and spacing; modulus; slippage; stress;
tensile strength
105
106 Creazza e t ai.
m Ob62949 0542b75 141 m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Giuseppe Creazza, born in Venezia in 1927, is Full Professor of Theory of Ela-

sticity, head of Laboratory of Strength of Materials in Venice Institute University.
At present he is carrying out research on non-linear analysis of RC structures, en
crack control with H.S.C., on assessement of masonry building, on bond and duc-
tility of r.c. elements.
Roberto Di Marco, born in Roma 1948, is Associate Professor of Civil Enginee-

ring.
At present he is carrying out research on non-linear analysis of R.C. structures, on
steel-concrete composite structures and on the use of new materials in civil engi-
neering.
Salvatore Russo, born in Bologna in 1962, is Technical Researcher in Laboratory

of Strength of Materials in Venice Institute University
At present he is carrying out research on tensile strength of H.S.C., on structural
behavior of FRP beams, on ductility of reinforcement steels and structures.
Erizo Siviero, born in Padova in 1945, is Full Professor of Civil Engineering, head
or the Department of 'Construction of Architecture'.
At present he is carrying out research on High Strength Concrete, on Assessement
of Hystorical Bulding and in bridge design, using advanced materials.
INTRODUC'lXON
It is common knowledge that the bond between the steel bar and the concrete si-
gnificantly influences the behavior of reinforced concrete strtxtures, both during
the formation of cracks and in the subsequent tension stiffening phase.
For normal-strength concretes, a number of experimental trials have been perfor-
I Y ~ mainly
, using pull-out tests, and have led to the formulation of bond-
stress/slipping curves to illustrate the phenomenon right up to the failure of the
concrete bond (1j.
The characteristic bond properties of high-strength concretes, on the other hand,
are still being investigated, both in terms of the relationship between the ultimate
tangential strength and the compressive strength, and for the definition of the
whole bond-stresses/slipping curve.
The results of experimental research by Konig et al., De Larrard et al., and of ot-
her reports in the CEB Bulletin dedicated to the use of high-strength concretes,
(2), (3), (4), seem to demonstrate that the ultimate bond strength deuends on the
square root of the compressive strength, e.g..
r,,, = f(E)
It1 rhe book edited by Yves Malier, (5j, Lorrain showed how high-strength cuiicre
= 0662949 0542bïb O B B Tribute to Peter Gergeiy 107
tes behave differently from normal mixtures: as regards the initial tangent of the
curves for the two types of concrete, pull-out test findings showed that the high-
strength mixtures have a significantly higher stifiess value than the lower-class
concretes.
This feature is also demonstrated in Zhiming's graphs ( 6 ) and in results of tests
performed by Galeota et al. (7), Noghabai et al. (8), Lahoud & Allaire (9), Modena
et al. (IO), and Bigaj (1 i), who used a highly-sophisticated and complicated expe-
rimental method to define the whole 5-s curve, even in the case of yielding of the
steel bar.
Tiie recent CEB document (12) concerning the raccomanded extensions of the
Model Code 90 to high-strength concretes proposes adopting a linear relationship
between the compressive strength and the ultimate bond strength for these mate-
rials, and considering a bond curve of the same kind as for ordinary concretes.
Moreover, as mentioned in Farra's thesis (13), an analysis of cracking problems
only calls for a knowledge of the ascending branch of the T-s curve, which corre-
sponds to small slipping values.
So to show the differences in behavior between high and normal strength reinfor-
ced concrete elements in a serviceability state, it seems important to establish the
significant parameters of the initial branch of the relevant bond-stresslslipping cur-
ves.
To achieve this goal an experimental method was adopted, that is easy to imple-
ment, using tensile tests on cylindrical ties instead of the usual pull-out tests.
A model derived from the one presented by Gupta (14), and assuming a linear
bond behavior for each loading step, is used to interpret the experimenta]findings,
which consist in the distances measured between the cracks under progressively in-
creasing tensile loads.
The theoretical and experimental analysis described in this paper proposes a nume-
rical evaluation of the mean slipping modulus, which characterizes the nitial branch
of the bond-stresdslipping curve, for two classes of concrete (50 and 100 m a ) .
In the authors' opinion, this is a topic of considerable interest because it can lead
up to a definition in qualitative terms of the difference in the serviceability state be-
havior and in the tension stiffening effects between normal and high-strength con-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
cretes.
THEORETICAL ANALYSIS
The study considers a reinforced concrete tie, as illustrated in Fig. 1, where "a" in-
dicates the semi-distance between two cracks.
The first step is to establish the state of equilibrium for the steel and concrete in a
small section of the tie, dx in lenyth.
The following equilibrium equations of the forces in the x direction are obtained,
where As and Ac are the areas of the steel and concrete, p is the perimeter of the
reinforcement bar and ~ ( x is) the bond stress (Figs. 2a and 2b):
0 b b 2 9 4 9 O542677 TL4 I
108 Creazza et al.
- for the steel:

o , ( x ) A , = ( ~ , ( x ) + d o , ( x ) ) A ,- r(x)pdx (1)
- for the concrete:
o , ( x ) A , =(o,(x)+do,(x))A, +r(x)pdx ( 2 )
It is then assumed that the bond stress varies with the slipping function s(x), ac-
cording to the linear law given below:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(where Eb is the slipping modulus).

Having inserted (3) in ( I ) and (2) and simplified the equations, the result is:
dsc(x) - E6Psjx)
dX Ac
The slipping function s(x) thus equates to:
where us(x) and uc(x) are the displacements of the steel and concrete, respectively,
in the x direction.
According to Hooke's law, if the moduli of elasticity for the steel and concrete are
called Es and E,, then:
d.S,(X)
-- -Es- d"u,(x) (7)
U'X dX
do,(x) - E,- d2li,(X)

-- (8)
dX dx2
If (4) and (5) are compared with (7) and (8), then:
m Obb2949 0542b78 950
In brief, returning to (6),we shall have:
Assuming that E, = &,and that p = AJA+ then:
which can be briefly expressed as:
dx
Thus:
s(x) = C, sinh Zx + C> cosh Zx ( 1 4)
Since s(x) = O for x = O, then CZ= O, so:
s(x) = CIsinh Zx (15)
Replacing the s(x) function in (4) and integrating twice leads to:
u,(x) = as(a)- 3C,(coshZn - coshZx)

ASZ
( 16)
(in calculating the integration constants, it is assumed that if x = O, then us(x) =: O

and if x = a, then o,(x) = os(a))
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542679 877
110 Creazza e t al.
Now, replacing the s(x) function in (5) and integrating twice,
EP
sC(x)= &.C1(coshZn-coshZx) (18)
C
(in calculating the integration constants, it is assumed that if x = a, then o,(x) = O,

and if x = O, then u,(x) = O).
Assuming that x = a in (1 5 ) and also in (17) and (1 9), then:
The end result is:
In the above formula, Cl can be deduced on an experimental level because the rela- --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
tive value of "a" coincides with a known value of o,(a), correspoiiding to the
load applied to the reinforcement at the time of a crack opening, (the Z parameter
is established as illustrated below).
Between the end sections x = O and x = a, there will be an equilibrium between the
forces acting on the steel and concrete, which must mean that:
Taking the value of (16) to replace CI, then:
Since oc(0)- ac(a) = f,,, Le., the tensile strength ofthe concrete, then from (23):
Obb2949 0542680 509
For equilibrium, we have:
which means that:
If we replace this value in (25) and assume that r = os(a) / o,(equil.) > 1, then:
zn=/n [ -
r-1
+/FI --
r-ly
(29)
Since, by experiment, we can know the ratio "ri' and measure the distance "a" for
every crack that opens, equation (29) can be used to calculate the 2 parameter and
(12) can be used to establish the slipping modulus Eb.
We can consequently also know the slipping function which, by means of (22) and
of (17) and ( 1 9), can be expressed as:
(Zxcosh Zu - .~idiZx)
Z cosh Za
The mean slipping can be expressed as:
This slipping corresponds to the mean bond stress.
5(x)n/(wn = E6s~x)r1wut1 (32)

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
112 Creazza et al.
Going on now to implement the above on the basis of experimental findings, it is

assumed that the analytical results apply to each loading step.
Disregarding in such a way the effects of bond reversal, as the craching phenome-
non evolves, the superposition principle applies.
Therefore, to experimentally illustrate the evolution of the bond-stresdslipping
curve at the time of the opening of the "n"th crack, we shall have:
i i
1 I
aid the whole trend orthe curve can be traced by means of straight line segments.
In conclusion, if "oS(a)" and "a" are known from experiment, the above-explained
steps can be used to obtain both quantitative and qualitative information on the law
governing the concrete bond for the initial, and most significant, branch of the cur-
ve that defines it.
EXPERIMENTAL PROCEDURE
The samples used in the experiment were cylindrical reinforced concrete ties con-
taining a single steel bar lying along their cross-sectional axis 2nd partially embed-
ded in the concrete (Fig. 3).
The tensile stress was applied directly to the bar and was consequently transferred
tc tne concrete through the bond, ¡.e., by induced deformation.
The bars were 1300 mm long and either 10 or 12 mm in diameter, thus slightly di-
versiGing the geometric percentage of the reinforcement.
The concrete cylinder was 900 mm long and 78 mm in diameter and the distance
between the test machine clamps was 1000 min.
The concrete element was long enough to ensure an adequate evaluation of the
overall cracking phenomenon, disregarding any failures induced by local bond
stress phenomena, such as those occurring near the ends of the samples.
The tests were performed under controlled loading conditions, at a rate of 0.002
m d s , using an all-purpose 1200 kN machine.
For each tie, the test was stopped when a load equating to about 80% of the yield
strength of the reinforcement was reached, since this loading gradient allowed for a
suitable development of the cracking phenomenon when the behavior of the bar
was still in the elastic-!inear phase.
Two precision deflectometers with a stroke of 5 mm were attached to the concrete
cylinder, 180" apart, to check the alignment of the bar with the axis of the cylindri-
cal cross section of concrete; this instrumentation enabled the actual centroid of the
reinforcement to be kept under control during the test by measuring the deforma-
tions in the concrete under load.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
U b b è ï ~ ï05426192 381 D
The steel bars belonged to the Italian class Feb44k Tempcore (i.e., they liad under-
gone air cooling) and corresponded to the international class B500 in the ductility
category S (CEB).
.4s for the concrete, two strength classes were used, ¡.e., 50 and 100 m a , for
which the mechanical features 28 days afler casting (mean values expressed in
m a ) are given in Table 1, where Rc is the compressive strength, fctm is the mean
tensile strength deduced from the compressive strength, fct(sp) is the tensile
strength resulting from splitting tests and fct" is the direct tensile strength deduced
from fct(sp) using a conversion factor of 0.85.The 50 MPa concrete was made
with Portland cement 42.5 in proportions of 355 kg/m3, with a water-cement ratio
of 0.41 and addins 2% of Rheobuild (2000) admixture, for a slump of 10 cm.The
100 MPa was prepared with Portland 52.5 cement in proportions of 433 kg/m3,
with 2.5% of Rheobuild (5000) admixture, a water-cement ratio of 0.31, a slump
of 22 cm and the addition of Silica Fume (MSBl0)40.The samples for the two dif-
ferent strength classes were all prepared at the same time and with the same con-
crete as was used to obtain 4 cubes for compressive tests and 12 prisms for split-
ting tests in order to establish the corresponding strength characteristics.
The specifications were only formally complied with, simply attributing the lower-
strength concretes to the nominal class 50 MPa and the higher-strength samples to
the nominal class 100 Mpa.
As already mentioned, Table 1 shows the mean strengths emerging from the tests.
On the whole, the experiments involved a total of 20 ties, divided as follows ac-
cording to their concrete class and geometric proportion of reinforcement: 5 ties
with 50 MPa concrete and dia. 10 inm bars;5 ties with 50 MPa concrete and dia.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
12 mm bars;5 ties with 100 MPa concrete and dia. 10 mm bars;5 ties with 100
MPa concrete and dia. 12 mm bars.
CRACK PATTERN OF SPECIMEN
A metric reference scale was prepared to enable a more accurate check on the
cracking situation along the surface of the concrete (¡.e., the location and number
of the cracks, and the distance between them).The figures n.4 and n. 5 shows the
effective cracking evolution during the test for compressive concrete strength class
equal to 50 MPa and diameter bar of reinforcement steel equal to 10 and 12
mm.The figures n.6 and n.7 shows in the same way the effective cracking evolution
for concrete class equal to 100 MPa and diameter bar of reinforcement steel equal
to 10 and 12 mm.
RESULTS
For the purposes of the present study, the amplitude "a" and the corresponding
stress 'bs(a)" being applied to the reinforcement was measured during the test as
W ûbb2949 0542683 2 L B
114 Creazza et al.
each crack opened in each of the ties being tested. These results as been showed
in Tables 2 and 3 .
At this point. the equation (29) was used to calculate the Z parameter and equation
(12) was used to establish the slipping modulus Eb.
The mean slipping s(x) was then obtained by means of (3 I ) and the mean ~ ( x was )
established using (32). Having performed the calculations on the whole cracking
cycle for each tie, the mean XT and the mean Cs were calculated and the results are
summarized in Tables 2 and 3, distinguishing between the two classes of concrete.
These tables also illustrate the difyerent loading steps in each test, for which corre-
spmding pairs of information (7-s, measured in MPa and mm, respectively) were
measured, also distinguishing between the two different reinfòrceniint bar diame-
ters.
The number of loading steps considered was not always the same from one test to
another, since the most significant step was analyzed in each case
The rsults referred to each tie would have been represented by a set og straight line
segments which slopes equal the slipping modulus El, at each step of the loading
process.
To a better understanding of the intluence of concrete class on the bond behavior,
the results of the tests concerning a given class of concrete were considered as a
whole.
The experimental coordinates (T-s) and the corresponding regression straight lines
are illustiated in Fig. 8 for the 50 MPa class of concrete and in Fig. 9 for the 100
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M’a class of concrete.
As a result of statistical processing of the test data based on the hypathesis of a li-
near relationship between the bond stresse5 and the slipping table 4 shows the
estimated values of the bond modulus and the corresponding values of the estima-
ted variance and of the standard error.
CONCLUSIONS
As illustrated in Table 2 , the bond modulus was about 18.7 MI->a/mnifor the 50
MPa concretes and 39.3 MPa/nim for the 100 MPa concretes. confirming the two
materials’ substantially different behavior
It is common knowledge that a linear behakior for the law governing the concrete
I bwd in the two classcs of concrete is only a rough hypothesis. but for a conipari-
son of the perforinance of the two classes of concrete in the serviceability state, it
is of absolutely no consequence whether the regression lines are straight or curved.
since our only concern is to veri6 wlierlier the tension stiffening ei’fect is substan-
tially different.
It was estimated, therefore, that the straight regression lines were the most signifi-
cant because they provide slipping moduli that generally remain valid over the
whole of the initial branch considered in the present analysis.
In our specific case, it emerged that the high-strength concretes had a bond mo-
dulus (Eb) that was virtually twice the value found for the normal-strength concre
- tes.
0bbiq49 ' Tribute to Peter Gergely 115
While more in-depth expenmental trials are underway, the findings of the present
s t d y could, in our opinion, prove a valid basis for calculating structural effects in
service for high-strength concretes.
ACKNOWLEDGMENTS
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The authors are grateful to the MAC plant of Treviso for providing the concrete to
make the ties.
REFERENCES
1- Model-Cade 90, Bulletin d'hforination n. 2 i 3\21?,Thomas Telford Edi-

tor,May1993
2 - Konig G . , Fehling E.,( 1988), Zur Rissbreitenbeschrankung im Stahlbetonbau.

Beton vnd Stahlbetonbaii 83.
3 - De Larrard F., Malier Y.(1989) Propriétés constructives des bétons à très

hautes per formances, Arimles de L 'I. TH. W .
,Paris
4 - High Strength Concrete, (1990), State of the Art repart, Bdlettin

d 7nformíriioti II. 197, CEB-FIP
5 - MalierY., (1992), Les Bétons à hautes performances, Presses de l'école

~mtio~iale des Po?i/sei ccrh-iisskes.
6 - Zhiming. T, (1992), Bond beiiaviour of deformed bars in high strength con-

crete, ¡triernationcri Conj2rrnc~Hotiú iii C'oticreie I.i..orii I<e.warcbto Practice,
vol.2 4- I I N - I9
7 - Galeota D, Giammatteo M.M, Marino R. (1991) Mechanical Properties of

High Strength Concrete, Proceediiigs of Ciornale Aiccrp 9I , (in itnlinn)
8 - Noghabai K. (1 993) Bond Properties of high strength concrete. High Strength
û b b 2 9 4 9 0542685 O90 I
116 Creazza et al.
9 - Lahoud A. E. , Ailaire, G. J , (1991) Etude expérimentale de !’adhérence

béton-armature dans les bétons à hautes performances. Depar/kmetil de
gétiiti civil, ütiiversiie de Sherbrooke, Ihpporí No. 06-9 I .
10- Modena C . , Berto P. (1995) Indagirii coiiiparative sull’interazione con I’acciaio

e Sulla fessurazione di calcestruzzi normali ed ad alta resistenza, I’roceeditgs of
Giorticrie Aiccrp 95, (¡ti itcilicoi)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
11 - Bigaj, A.J., (1995), Bond behaviour of deformed bars in NSC and HSC,
DeYi Utiiversiw of lectiology, Reporl 25.5.95-11
12 - High Performance Concrete Recommended Extensions tc the Model Code90

C l 3 I I . 228.
Research Needs”, ( 1995) Bii/lei/in J ’lt~f¿)i.-iiiciíioti
13 - Farra.B ( 1 995), Influence de la Résistance du beton et de son dlierence avec

l’armature sur la fissuration, l*,’cole I’olyíechiiiqiie bëc,‘L.r.íileLIC.I,iiiixitiiie, These n.
1359.
14 - Gupta A.K., Maestrini R, (1989) Tension stiftening Model for Reinforced

Concrete Bars. Joitr-?icrlASCE vol. I I 7, tio. 3 Mcwch.
m 0662944 0542686 T27 m
TABLE 1-MEASURED CONCRETE MEAN STRENGTH
I Concrete I Rcm 1 fctm I fctkol I fct* I

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
TABLE 2-TEST RESULTS OF 50 MPa CONCRETE CLASS SAMPLES
W Obb2949 0542b87 î b 3 M
118 Creazza et ai.
TABLE 3-TEST RESULTS OF 100 MPa CONCRETE CLASS SAMPLES
TABLE 4-ESTIMATED BOND MODULUS AND STATISTICS OF EXPERIMENTAL

DATA
Rck Estimated Estimated Standard

MPa Eb Variance Error
MPdmm
50 I 18.7290 I 0.440838 I 0.81635
100 I 39.2592 1 5.41986 I 2.75798
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542b88 8 T T R
T T
1 B
a a
steel
I dx I
Fig. Pa-Forces acting on steel reinforcements
concrete
OC
+
=a-
I dx I
Fig. 2b-Forces acting on concrete
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ûbb2949 0542b89 73b
120 Creazza et al.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 3-Cylindrical test samples and arrangement of measuring apparatus
2xr
..
I
-
a
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 4-Effective cracking evolution for specimen with 50 MPa concrete class and
bar diameter 1 O mm
Fig. 5-Effective cracking evolution for specimen with 50 MPa concrete class and
bar diameter 12 mm
~
M 0 b b 2 9 4 9 0542b9L 394
122 Creazza et al.
--
---
--
- -< i
Fig. 6-Effective cracking evolution for specimen with 100 MPa concrete class
and bar diameter 1O mm
2nr
..<
- .. ~ ..- . i
I ' I p__T-I
,
-\
I I-'-. -1 .
'
;-i
I
---i I -
c-
I
<
'
i 1 I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 7-Effective cracking evolution for specimen with 100 MPa concrete class
and bar diameter 12 mm
0662949 0542692 220 0
-f(MPii)
6.0
5.0
4.0
3.0
2.0
I .o
O 0.1 0.2 0.3 s(mm)

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 8-Test results and estimated linear (n-s)relationship for 50 MPa class of
concrete
11.0
10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.o
O 0.1 0.2 0.3 s(mni)
Fig. 9-Test results and estimated linear (ws)relationship for 100 MPa class of
concrete
0662949 0 5 4 2 b î 3 Lb7
124 Creazza et al.
T (hlPn1
1 I .o
10.0
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9.0
8.0
7.0
6.0
5.0
-1.0
3.0
2.0
1 .o
o o I u 2 o < qllllil i
Fig. 1 O-Comparison between linear (x-s)relationships for C50 and C 1 O0
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
SP 180-6
Bond under Repeated Loading
by 6. L. Balázs
Synopsis: Charact-c results of more than 100 cyclic pd-out tests are
presented including various load histories (simuiahg reahtic load spectra) like
random loading or variable amplitude loading with increasing or decreasing
tendencies in addition to the constant amplitude loadiug with different amplitudes.
Siip measurements are compared to acoustic emission measurements.
Repeated loading produces a progressive deterioration of bond caused by the

propagation of micro-cracks and progress of micro-cnishmg m concrete.
Deterioration of bond may be observed by measuring Sig, or acoustic emission
events. It is quantitatively shown that the actual ci1p is signiñcantly influenced by
the load history: maximum and mmimUm levels of the repeated load, type of
amplitude (constant or variable), frequency, sequence of amplitudes and number
of load cycles, respectively.
Keywords: Bond (concrete to reinforcement); cyclic creep; cyclic loads;

deformation; fatigue (materials); fatigue tests; reinforcing bars; slippage
125
IObb2949 0542695 T 3 T
126 Balázs
György L. Balá= is an assoc. prof. in the Dept. of Reinforced Concrete

Structures at the Technical University of Budapest (head of laboratory). His main
research interests are bond behavior of reinforcing bars, cracking, fiber reinforced
concrete and fiber reinforced polymers. He is chairman of CEB Task Group on
Serviceabiltty Models and member of several other CEB and FIP Commissions.
INTRODUCTION
Fatigue behavior of concrete structures as well as that of bond between

embedded reinforcements and concrete is gainmg increasing interest owhg to
structural as well as experimental reasons. On one hand the number of load cycles
e.g. on bridges exceed los repetitions (1) which is far more than the traditional
technical limit of 2.106 repetitions. Many structural elements have become more
slender due to the application of concretes with improved strength leading to a
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
decreased dead load to live load ratio.
On the other hand, the development of experimental devices and methods

opened new possibilities m fatigue testing. Any kind of variable or random
amplitude cyclic loadmg (being closer to real load histories) can be produced
under computer control.
Intention of this contnbution is to summarize the test results since the last
AC1 state-of-the-art report m this field (2). Present results were obtained mainly in
an extensive experimentai study at the Otto-Graf-Institute Stuttgart, Germany on
the cyclic bond behavior with various load histories (3) under the auspices of the
German Research Society (DFG).
ACOUSTIC EMISSION MONITORING OF BOND FATIGUE
Repeated loading produces a progressive deterioration of bond.

Accumulation of bond damage is supposed to be caused by the propagation of
micro-cracks (4) ( 5 ) and progressive micro-crushing of concrete ( 6 ) m fiont of
the lugs. Their effect is often observed in increasing slip (7)...(13). Instead of
measuring sip, the registration of acoustic emission (AE) signals and their
accumuíation may give very valuable information on the progress of local
damages over the steel-concrete interfaces.
In an experimental study (14) simultaneous registration of the AE signals

and the bond-slip behavior was carried out. A typical test result is shown in Fig. 1.
The upper part of Fig.1 indicates the registered AE amplitudes as bar graphs and
the load history (two blocks of constant amplitude cycles). The amplitudes are
given in relative values of the maximum peak-to-peak (PA) values of the signals
Obbt949 0542696 976
and are summed in 15 second intervais (denoted as PA sum per time unit). The
bottom pcut of Fig.l indicates the acoumulated damage as the sum of the AE
- Tribute to Peter Gergely 127
amplitudes till the mvesîigated time with another series of bar graphs as a function
of the time (called cumulative PA) together with the simultaneously measured
siip versus time (number of load cycles) relationship.
Frequency of repeated loading was 4 Hz. Load levels are given in

percentage values, ie., the maximum vaiue of repeated load (Tb,-) related to
the monotonic bond strength (Tbu) measured on an average of 15.4 N/mm2. The
minimum value of repeated load was 10 % of the maximum value. The bond
length (denoted as l b m Figs. 1 and 2) was twice the bar diameter being
approximately twice the rib spacing. The mean compressive strength of concrete
at testing measured on cubes of 150 mm sides was 34.9 N/mm2.
Most of the AE signals were registered by applying the initial monotonic

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
load up to the mean value of cyclic load, then at the beginning of the cyclic load
or at an increase m the cyclic load level and preceding the pd-out fahre.
However, signals were also detected during a block of constant amplitude cyclic
load (see second block).
The comparison of the two independent measurements in Fig. 1.b indicate

that the internal damage accumulation, defined as the cumulative AE peak
amplitudes versus time relationship (bar graph) gives a very similar tendency to
the slip versus time relationship (continuous h e ) .
The same conclusion was drawn fkom another test presented m Fig.2 *
applying four blocks of constant amplitude loadings mstead of two blocks and
detecting both the siip development and the AE signals.
Nevertheless, AE technique seems to be a very promisiug method for

analysis of internal damages over the steel-concrete interfaces and should be
developed to be able to give indication on the whole fatigue life even for
practical cases &ridges, etc.). Further details on above AE measurements may be
obtained in Refs. (14) and (ZO), respectively.
Above test results mdicated our first efforts for using the AE technique on
analyshg the bond damage process under cyclic loading. In the foliowing chapters
ñuther cyclic bond test results are presented appiyig the traditional way of Siip
measurement by LVDT, however, considering a wide range load histories.
DEVELOPMENT OF SLIP UNDER REPEATED LOADING
One of the most typical representation of cyclic bond test results is the siip
versus number of load cycles diagram indicating the development of slip (s) as a
function of the load cycles (n). Most important loading parameters are: maximum
0662949 0542697 8 0 2
128 Balázs
and minimum levels of the repeated load, type of amplitude (constant or variable),
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
fiequency and sequence of amplitudes and number of load cycles.
In the following, characteristic results of more 100 cyclic pull-out tests are
presented. First, the cyclic pull-out test results will presented by the most general
load history (random loading), then variable amplitude loading where the
amplitude changes according to a mathematical formula based on the load spectra
of bridges and crane constructions. Finally, the simplification of all: the constant
amplitude loading. The test results were obtained on pull-out specimens with five
bar diameter bond length, 16 mm bars, 500 N / m z characteristicyield strength of
the bar with 0.0065 relaive rib area. Mean compressive strength of concrete at
testing measured on cubes of 150 mm sides was 30 N/mm2. The sinusoidal cyclic
load was applied with a frequency of 4 Hz.
Development of Slip under Random Cvclic Loading
Test results in Fig.4 are obtained by applying a series of random cyclic

load levels on pull-out specimens (15). One block of the load history indicates
10000 load cycles with the same amplitude (i.e. 200 blocks to 2 miilion load
cycles). Random values were generated between O and 1, where 1 meant the
highest value of cyclic load ( r b , W ) for one specific test taken as percentage of
the monotonic pull-out strength (rbu). Random values smaller than 1 meant cyclic
load levels 7b,maw< rb,MAX.
Fig.4 presents three test results on high (rb,-/7bu=0.6, 0.7 and 0.8) load
levels. Poriions of Fig.3 up to 250 O00 then 750 O00 and 1 250 O00 cycles are
enlarged in Fig.5. The increasing random values (hence increasing cyclic load
levels) induced increasing shp rate in the first three blocks. In the third block, the
random vahie of 0.903 was reached. Subsequent blocks of cyclic loads did not
produce remarkable slip increase until the 29th block which was the first block
with a bigher random value of 0.989 than that of the third block (0.903).
I
1 Consideration of the whole slip development clearly indicates that only
load levels approaching or exceeding the previous highest level induce
considerable slip increase. The contribution of low load levels is almost
negligible. Close to pull-out failure somewhat lower levels than the previous
highest level may be decisive for pull-out failure as weli.
Development of Slip under Variable Amplitude Loading
In another series (16), variable amplitude cyclic loading was applied on

pull-out specimens. The maximum value of cyclic load had linearly,
parabolically or logarithmically increasing or decreasing tendencies. The load
histories and the observed slip versus number of load cycles relationships are
indicated in Fig.6. The length of one block in these tests was 1 million cycles. One
block consisted of 10 subblocks of constant amplitude cycles.
0662949 0542b98 749
With increasing segueme of amplirudes (Fig.ó.a,c and e), the first block
of loading produced a pronounced Sig, mcrease with increasing shp rate. Cycling
at lower load levels m the next block, howevery did not conmiutte to the Sig,
mcrease. With decreasing sequence of amplitudes (Fig.ó.b,d and f) sigaiñcant Siip
mcrease was produced only by the highest cyclic load levels both m the ñrsí and
the later blocks. Lower levels had no remarkable mfhience on the slip increase.
The same trends of Sig, development were observed when furuner blocks of
loadings were applied.
Fig.7 compares slip increments reached after 106 load cycles produced m
100 blocks of variable amplitude loads with a length of 104 load cycles as a
function of the cyclic load history. The abscissa of Fig.6 indicates the load history
expressed as a unitless value smaller or equal to 1 (called intensity factor of cyclic
load). It is calculated as the integral of the upper level curve of the load history
related to the integral of constant amplitude loadiug m e . The cyclic load
mt&ty factor is equal to 1 for constant amplitude loading and <I for all kinds of
variable amplitude loadings having the same maximum vahie as q,max of the
constant amplitude loading.
Fig.7 indicates that the higher the cyclic load intensity factor, the higher
the Sig, mcrement. Sltp values after lo6 load cycles produced by increasing (left
side of the diagram) or decreasing (right side of the diagram) tendencies of the
amplitudes do not differ considerably. This is empasized by the quasi-qmmetry
of the diagram as weü. It means that these tests did not mdicate a sigaiñcant
sequence effect of the various amplitudes.
Development of Slip under Constant Amditude Loa-
When considering the whole bond fatigue process under constant

amplitude cyclic loading, it was observed (17) that the slip rate tends to decrease
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
initially, then remains constant, and increases rapidly before pull-out failure.
The third failure phase starts by reaching the shp at the monotonic bond strength.
If this Sig, value has not been reached, the faihire p h s e is avoided. Above
behavior was found to be independent of the load level or number of cycles to
faihire, however, the slq, rate during ali the three phases is a function of the load
level Recent test results ( 18) that are presented m Fig. 8 supported this hypothesis.
In Fig8 cyclic pulí-out test results are presented m h e a r scale with various load
levels and number of load cycles to fahre.
According to earlier pull-out test results by Rehm and Eligehausa (9) with
constant amplitude cyclic loading, the slip versus number of load cycles
relationship was found to be approximately linear in double logarithmic scale if
the fàtigue failure of bond has not been reached. This approach was adopted m the
CEB-FIP Model Code 90 (19) for the prediction of cycles dependent Sig> (s,,) as a
9 Obb2949 0 5 4 2 b î î b â 5 W
130 Balázs
function of the initial slip before cyclic loading (so)and the number of load cycles
(n):
%=SO.( 1+kn) where kn=( 1tn)b-l and b=O. 107 (1)
Recent test results (18) are presented in Fig.9. Every curve indicates the
average of 5 measurements with constant amplitude cyclic loading in double
logarithmic scale. Straight lines show the prediction by Eq.(l). Relevant
measurements and predictions were fit to start Com the same slip at 10 load
cycles for all load levels. Comparing the measurements (dotted lines) and the
predictions (continuos lines) differences may be observed both in the course of
diagrams and the inchation of them (power b in Eq. (1)). In these tests (1 8)
slightly higher power b values were obtained than given by Eq.( 1).
The amplitude of cyclic loading was also found to be an important

parameter concerning the development of slip (3) which is almost obvious but
quantitative results are not available in the literature. Fig. 10 gives a comparison of
cyclic pull-out test results with dgerent amplitudes (3). The y-axis of Fig.9
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
indicates the slip after 2 million load cycles and the x-axis indicates the amplitude
of cyclic load (ATb) related to the maximum value of cychc load: A T ~ / T ~ . ~ ~ ~ = O ,
0.3, 0.6 and 0.9. Zero amplitude (ATb/Tb,max=O) meant long term tests over the
same testing time. There were three series of cyclic tests carried out on load
levels: ~ ~ , ~ ~ ~ 0.5 / ~ and
~ ~0.7=(where
0 . 3Tbu, means the monotonic bond strength)
with always three kinds of amplitudes, respectively. The broken lines in Fig.10
indicate the average of the three measurements. The slip versus number of load
cycles relationships are presented in Fig. 11.
Figs.10 and 11 indicate that the higher the amplitude the higher the slip
increase to the same cyclic load level. The slip increase is a non-linear function of
the amplitude. For amplitudes between ATb/Tb,max=O and 0.3 the innuence of the
amplitude is, however, almost neglectable.
CONCLUSIONS
Following conclusions can be drawn from cyclic pull-out tests with various
load histories:
1. Acoustic emission (AE)technique seems to be a very promising technique

to follow the bond fatigue process. Internal damage accumulation can be
d e h e d as the cumulative AE peak amplitudes versus time relationship
which was found to give very similar tendencies to the &p measurement
versus time relationship. This technique should be further developed to be
able to give indications on the fatigue life of actual structures.
2. Randomly generated cyclic load levels indicated that only load levels
approaching or exceeding the previous highest level induce considerable
0662949 0542790 i127
sip, mcrease. Contribution of low load levels was almost negligible. With
mcreasing sequence of amplitudes, the first block of loading produced a
pronounced sip, mcrease with mcreasiug shp rate. Cychg at lower load
levels m the next block, however? did not contriite to the slip increase.
With dmeasiug sequence of amplitudes, significant siip mcrease was
produced oniy by the highest cyclic load levels both m the first and the
later blocks. Lower levels had no remarkable infiuence on the slip mcrease.
3. The dip development under constant amplitude cycles gives the upper
bound of test results with any kind of Variable amplitude loadings. The
higher the integral of the load history curve the higher the slip increment
considerhg the same lower value for the cyclic load.
4. The a m p h d e of cyclic loading was also found to be an important

parameter considering the development of &p. The higher the amplitude
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the higher the slip increase, however? amplitudes smaller than 30% of the
maximum level of the cyclic load did not mdicate Merences. The slip
mcrease above 30 % amplitude was a non-linear function of the amplitude.
NOTATIONS
power of bond stress vs. siip relationship -

cyclic creep factor for shp -
bond length mm
relative displacement (shp) mm
slip under cyclic load mm
initial sly by applying cyclic loads mm
bond stress N/m2
bond strength obtained by monotonic load N/llUl12
maximum value of cyclic load N/mll12
highest maximum value of cyclic load m a block of N / m 2
variable amplitude loading
minimum value of cyclic load N/mll12
relative load level of cyclic load %
relative amplitude of cyclic load YQ
ACKNOWLEDGEMENTS
The ñuacial support from the German Science Community (DFG) and the
supervision of the research project by Dr.Koch is greatfdly acknowledged.
REFERENCES
1. König, G., Gerardt, H.-Ch., "Nachweis bei nicht vorwiegend d e n d e r

Belastung für mit HLV-Spanngiiedeni vorgespannte Tragwerke,"
Mitteilungen I f l t 611992,pp. 191-195.
W üb62949 O542701 063 W
132 Balázs
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
2. AC1 408.2.R-92, "State-of-the-Art Report on Bond Under Cyclic Loads,"
ACZ Com. 408, Nov. 1992
3. Baiázs, G.L, Koch, R, "Betonbauteile unter Betriebsbedingungen (Concrete

members in Service Conditions)," Research Report, Otto-Graf-Institute
Stattgar&,Germany, Nov. 1996
4. Broms, B.B., "Technique for Investigation of Internal Cracks in Reinforced

Concrete Members," ACI Journal, Vo1.62, No.1, Jan. 1965, pp. 35-43.
5 . Goto, Y., Otsuka, K., "Studies on Internal Cracks Formed in Concrete

Around Deformed Tension Bars," ACI Journal, VoL68. No.4, A p d 1971,
pp. 244-251.
6. Gambarova, P. and &ani, E., "Fracture Mechanics of Bond in Reinforced

Concrete," Discussion, Journal of Stnicturai Engineering ASCE, Vol. 111.
No.5, May 1985, pp.1161-1164.
7. Bresler, B., Bertero, V. "Behaviour of Reinfoced Concrete Under Repeated

Load," ASCE Journal of Structural Division, Vo1.94. ST6 June, 1968,
pp. 1567-1590.
8. Edwards, A.D., Yannopoulos, P.J., "Local bond stress-slip relationship under

repeated loading," Magazine of Concrete Research, Vo1.30, No. 103, June
1978, pp.62-72
9. Rehm, G., Eligehausen, R, "Bond of Ribbed Bars Under High-Cycle

Repeated Loads;"ACI Journal, Proceedings, Vo1.76, No.2, Feb. 1979,
pp.297-309.
10. Tassios, T.P., "Properties of Bond Between Concrete and Steel Under Load
Cycles Idealising Seismic Actions,'' AICAP-CEB Symposium: Structural
Concrete Under Seismic Action, Rome, CEB Bulletin d' Information
No.131, Vol.1, 1979
11. Ciampi, V., Eligehausen, R, Bertero, V.V., Popov, E.P., "Analytical Model
for Concrete Anchorages of Reinforcing Bars Under Generalized
Excitations," Report No. EERC82-83, Earthquake Engineering Research
Center, University of Califonia, Berkeley, California, Dec. 1982, 111 p.
12. CEB, "Bond Action and Bond Behaviour of Reinforcement," Chapter 7.4.2:
Sustained and repeated loadings and Chapter 11: Loading history effect$
CEB Bulletin d'Information No. 151, A p d 1982
13. Eligehausen, R, Popov, E.P., Bertero, V.V., "Local Bond Stress-Slip

Relationships of Deformed Bars Under Generalized Excitations," Report No
IDbb2949 0542702 TTT
UCBEERC 82-23, Earthquake Engineering Research Center, University of

California, Berkeley, Caiifornia, Oct., 1983.
14. Balázs, G.L., Grosse, U.C.,Koch, R, "Deterioration m Fatigue Detected by

Acoustic Emission Technique," Proceedings, IABSE Symposium San
Francisco 1995, pp.1289-1294.
15. Balázs, G.L., Koch, R, "Bond Behaviour under Random Cyclic Loading,"
Otto-Graf-Journal, Vo1.5,1994, pp. 52-67.
16. Balázs G.L., Koch, R, "Infiumce of Load History on Bond Behaviour,"

Proceedings, "Bond in Concrete" Codaence Riga 1992, pp.7-1.-7-10.
17. Balázs G.L., "Fatigue of Bond", ACI Materials JournaI, Vo1.88, N0.6,Nov-
Dec. 1991, pp.620-629.
18. Koch, R, Balázs G.L., "Siip Increase Under Cyclic And Long Term Loads,"
Otto-Graf-Joownal,Voi.4, 1993, pp. 160-191.
- -
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
19. CEB, "CEB-FP Model Code 1990 Design Code", Comité Euro
international du Béton, Thomas Telford, London, 1993
20. Balázs, G.L., Grosse, C.U., Koch, R, Reinhardt, H.W., "Damage

accumulation on deformed steel to concrete interaction detected by acoustic
emission technique", Magazin of Concrete Research, Vol. 48, No. 177, Dec.
1996, pp. 3 11-320.
~
0662949 0542703 936 I

134 Balázs
Number of load cycles

-500 O 500 1000 1500 2000 2500
n a> 25
A
h
(o
23
20 -3
g
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
2
W 2,o
._
Y 15
1 1s
0
E 10 2
w
*s l,o Y
rn
8 U
;
2
0s 5 3
2 090 O
O 100 200 300 400 500 600 700 800
Time [ s ]
2 90
,Mregistration
1 9 5
n
E
E
1,o Y
.d
a
s
095
090
O 100 200 300 400 500 600 700 800
Time [ s ]
Fig. 1-Comparison of bond-slip behavior and acoustic emission signals in a cyclic

pull-out test for the whole fatigue life in two blocks of constant amplitude loading;
(14)-(a) peak amplitudes summed in 15 s intervals and the load history
(q, vs. time.) (b) cumulative peak amplitudes vs. time and slip vs. time relationships
Obb2949 0542704 872

O 5000 loo00 15000 2oooO 25000 3oooO
24
22
2
O
Time [ s ]
5000 10000 15000 2oooO 25000 3oooO
b)
80
O
, I
I I
1 I
I I
I I
I ' 1 1,25
1900
0,75
E
U
0,50 .-
a
ùì
0,25
0,oo
Time C S ]
Fig. 2-Comparison of bond-slip behavior and acoustic emission signals in a cyclic

pull-out test in four blocks of constant amplitude loading;(î4)-(a) peak
amplitudes summed in 50 s intervals and the load history (5vs. time);
(b) cumulative peak amplitudes YS. time and slip vs. time relationships
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
136 Balázs
= Ob62949 0542705 709
PC
Control board
of pulsator
L
4
I Hydraulic pulsator
Fig. 3-Test setup to cyclic pull-out tests
1 i:
Y
2.50
;?
m 2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
Fig. 4-Development of slip under random cyclic loading; (1 5)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2949 054270b b 4 c
a)
-
a
1.00
80.80
0.60
0.40
0.20
0.00

- 3.00
c)
-
92.25
2.75
2.50
i
3 2.00
1.75
1.50
Numbex of load cycles
Fig. 5-Enlarged portions of Fig. 3; (15)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
IObb2949 0542707 581 W
138 Balázs
0.5 1.0 1.5 3.0 0.5 1.0 1.5 2.0

Number of load cycles, Million Number of load cycles, Nllion
Number of load cycles, Million Number of load cycles. iviillion
slip 'b,max
Number of load cycles. Ullion --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---

Number of load cycles, MilliOn
Fig. 6-Slip increase as a function of load history; load level:

Tb,MAXhbu=0.4; (16)
a),b) parabolically variable 7b,max
c), d) linear variable Tb,max
e), f) logarithmically variable Tb,max
OLL2949 0542'708 418 W
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 7-Slip values at lo6 load cycles (¡.e., 100 cyclic blocks) with three different
cyclic load histories all with increasing and decreasing tendencies, load levels:
Zb,m&bu'O.4,0.5 and 0.6, one block Of loading: IO4; (3)
~
M 0662949 0542709 354 M

140 Balázs
I ! I I l l
100 300 250 500 750 lo00
Number of?oad cycles Number of load cycles
a3
3
2
O0 1000 2000 Moo 4ooo 5ooo

Number of load cycles Number of load cycles
P4
U
&3
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
!a
I
li
2
,
.
-
e
.*
1
I ---rb rh-0.70
Number of load cycles Number of load cycles
Fig. 8-Slip versus number of load cycles relationships for the whole bond fatigue
life under constant amplitude cyclic loading, load levels are indicated on the
diagrams (as ~ ~ ,values);
~ ~ (1 8)~ l ~ t , ~
ûbb29119 0542730 076
Fig. 9-Comparison of measured (18) slip developments under constant amplitude

cyclic loads and the predictions by the CEB-FIP Model Code (19) formula
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ûbbZ949 0542711 TO2 9
142 Balázs
2.0
B
Y
(li
1.5
1.0
0.5
0.0
azb/zb,max [%]
Fig. 1O-Slip at 2 .1 O6 load cycles as a function of the amplitude

of the cyclic load (zero amplitude AT&,,,,^^=^) means long term tests
over the same testing time; (3)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542712 949 a
n
Fig. 1 1-Slip versus load cycles measurements on three different cyclic load levels
(tb,,,&bu) and three amplitudes &&,max). The amplitudes are given in the
diagrams; (3)
a) cyclic load levels: ~b,m&bu'O.7
b) Cyclic load levels: Zb,mx/Zbu=0.5
c) cyclic load levels: Zb,m&bu= 0.3
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662749 0542713 885
SP 180-7
Bond Modelling of Prestressing Strand
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
by J. A. den Uijl
Synopsis: From pull-out and push-in tests on specimens with short embedment
length an empirical relation has been derived, which describes the local bond
stress as a function of the local slip and steel stress change. With the help of this
bond model the transfer length and the bi-linear relationship for the development
length of a pretensioned strand (ACI Building Code 1989, CEB-FIP Model Code
1990) is simulated. It is also used to indicate the influence of strand yielding on
the development length.
For the estimation of the concrete cover and strand spacing required to
prevent the occurrence of visible bond splitting cracks the response of the
concrete to the radial displacement of the strand-to-concrete interface is analyzed
by means of a so-called thick-walled-cylinder model. The radial interface
displacement consists of transverse deformation of the strand coupled with steel
stress change (Poisson effect) and wedging action caused by the shape of the
strand (lack-of-fit effect) and surface roughness. Besides the section geometry,
this model takes into account the softening behaviour of concrete loaded in
tension. It is used to explain the influence of various parameters such as concrete
cover, strand spacing, strand diameter and concrete strength on the bond
properties of strand.
Keywords: Bond model; bond splitting; strand (prestressing); strand (spacing);

transfer length
145
m 0bb2949 0542734 733 m
146 den Uiji
Joop A. den Uijl - associate professor at the Faculty of Civil Engineering, Delft
University of Technology, member of CEB Task Group 2/5 "Bond Models",
convenor on part "Bond of Prestressing Steel", member of Rilem TC 147-FMB
"Fracture Mechanics to Anchorage and Bond".
INTRODUCTION
In practice, design aspects related to bond of pretensioned tendons are

solved without taking into account the actual bond stress distribution. However,
for a better understanding of the influence of various parameters on transfer and
development lengths and for stress analysis of the transmission zone of pretens-
ioned members a realistic image of the bond stress distribution is desirable.
For pretensioned tendons two bond situations can be distinguished. Along

the transfer length the steel stress is reduced at prestress transfer, which causes
an expansion of the tendon and, hence, contributes to the bond resistance (push-in
situation). On the other hand, under flexural loading the steel stress in a cracked
section is increased, which results in a contraction of the tendon and a lower
bond resistance (pull-out situation). These two bond situations are discerned in
AC1 318-89 and CEB-FIP Model Code 1990 (MC90) by defining a lower bond
strength along the flexural bond length than along the transfer length.
The bond capacity of a strand is directly related to concrete confinement.

1 Wedging action of the strand generates circumferential tensile stresses in the
concrete that may initiate radial cracks. When such a crack reaches the concrete
surface bond resistance is considerably reduced. Hence, adequate concrete cover
and strand spacing shall be provided to prevent bond splitting failure.
In this paper a local bond model is described that takes into account the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
two bond situations, as well as a thick-walled cylinder model with which the
bond splitting phenomenon can be analyzed.
MODELLING
Bond Mechanism
In case of plain wire bond is attributed to adhesion and friction. The

designation adhesion stands for the contribution to elastic bond, which refers to
the deformation of the cementitious layer around the tendon. It consists of both
m Obb2949 0542715 b 5 8 m Tribute to Peter Gergely 147
chemical and physical adhesion and interlocking between cement stone and the
microscopically rough steel surface. Failure of adhesive bond occurs at a very
small relative displacement and therefore adhesion will only play a minor part
in practice. After breakage of the adhesive bond (debonding), force transfer is
provided by dry friction. The radial compressive stress connected to friction is
caused by shrinkage, roughness of the sliding plane and wedging of cement stone
particles. Due to wear and compaction of the sliding plane with increasing slip
this radial compressive stress diminishes and, as a consequence, the frictional
resistance is reduced.
At prestress transfer the diameter of a pretensioned tendon increases,

which generates radial compressive stresses and, hence, contributes to the
frictional bond resistance. This Poisson effect is proportional to the change in the
longitudinal steel strain. As a result of this effect the bond resistance of a plain
wire is strongly reduced when the steel stress is increased.
The bond stress versus slip relation of plain seven-wire strand strongly
differs from that of plain wire. After debonding the bond stress of strand remains
constant or slightly increases with slip, whereas it drastically drops in case of
wire. The transfer of forces between strand and concrete is ascribed to adhesion
and friction as well, but the frictional component is enhanced by some kind of
mechanical action. The typical shape of a strand, characterized by helical outer
wires around a straight centre wire, is held responsible for this mechanical action,
but different opinions exist about its nature.
When a free strand is loaded its pitch will change, but this change is
restrained when the strand is embedded. Due to this restrained deformation a
torsional moment develops in the strand that is equilibrated by contact stresses
along the perimeter. According to Stocker and Sozen (1) the contribution of these
stresses to the frictional bond resistance is small because of the low torsional
stiffness of the strand. They considered the wedging effect caused by irregular-
ities in the shape of a strand as the most important reason for the different bond
behaviour of plain wire and plain seven-wire strand. Variations in the pitch of the
outer wires and non-similarity of subsequent cross-sections due to variations in
the distribution of the space left between the outer wires because of the greater
diameter of the centre wire are the causes of this lack-of-fit effect.
This explanation is contradicted by Russell and Burns (2), who advocate

the existence of a strong interaction between restrained twist and mechanical
interlocking, arguing that Stocker and Sozen (I) came to the opposite conclusion
because of the specific boundary conditions in their pull-out tests. Due to the
relatively long distance between the entrance plane and the strand section where
the twist was restrained, the rotation of the strand corresponding to the slip at the
entrance plane was smeared out over a much larger distance than what would
happen near a bending crack, where this distance equals half the crack width.
Thus, in these pull-out tests the torsional moment caused by restrained twist was
much smaller than in a reguiar bending test. Unfortunately, the existence of such
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
148 den Uiji
W Ob62949 05427Lb 594 m
a restrained twist effect was only qualitatively indicated.

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Besides the above-mentioned Poisson effect, which is proportional to the

change in local steel strain, Den Uijl (3) took into account a term proportional
to the absolute value of the change in the local steel stress, which he denoted as
pitch effect. By introducing this term, an empirical bond law was obtained, with
which different bond situations could be simulated (next Section). Although no
conclusive physical explanation of the pitch effect could be given, it was found
helpful to draw a parallel with the frictional resistance that is experienced by a
tendon in a curved duct. When loaded in tension or compression normal stresses
between tendon and duct occur at the inner or outer side of the bend, respective-
ly, which contribute to the frictional resistance.
Local Bond Model
Existing models -- Local bond models that have been used to describe the
bond stress distribution along the transfer length of a pretensioned strand, can
roughly be subdivided into two categories: bond stress as a function of slip and
bond stress as a function of radial tendon expansion. A good example of the first
category has been given by Balks (4), who derived an analytical solution for the
second order differential bond equation by assuming a power function for the
bond stress-slip relation. Models in the second category relate the local bond
stress to the radial compressive stress acting at the steel-to-concrete interface as
a response of the concrete to the radial expansion of the tendon at prestress
release. To quantify this confining action the concrete around the tendon is often
conceived as a partly cracked thick-walled cylinder taking into account the
1 softening behaviour of concrete loaded in tension, e.g. Weerasekera and Loov (5).
A bond model developed by Akesson ( 6 ) for use in FEM analyses gives the local
bond stress as a function of slip and concrete confinement, thus combining both
categories. An important feature of the model is that shear deformation generates
a normal stress according to a so-called expansion curve, which represents the
wedging action of a strand and the subsequent wear of the sliding plane.
An important draw-back of these models is that they only give the bond
stresses along the transfer length and not along the flexural bond length, because
they do not properly take into account the different bond situations. In the first
category of models the assumption of an invariable bond stress-slip relation
denies a possible influence of strand transverse deformation on bond stress. The
models of the second category are explicitly based on expansion of the tendon
(push-in situation), but when the tendon contracts no bond is developed (pull-out
situation). The latter was also found to occur with Akesson’s model.
Bond Model for Pull-out and Push-in -- With the objective to develop a
local bond model that covers both bond situations Den Uiji (3) carried out pull-
W Obbi-949 0542317 420 D Tribute to Peter Gergely 149
out and push-in tests with short embedment length (9.3 mm diameter strand,
55 MPa cube strength, 50 and 88 mm embedded length, 46 mm clear cover). The
following empirical relationship characterizes the experimental results (Fig. 1):
% = 3+0.46+2.510-3Acrp+1.510"1Acrp1 [MPa] (1)
with $ bond stress

6 slip
Aop change in steel stress
7 2
Note: PS,cross-sectional area of strand ( -xBP)
36
perimeter
7~0~ of strand used to calculate bond stress
0p diameter of prestressing strand
The first two terms at the right hand side of this equation combine adhesion and
the lack-of-fit effect, the third term reflects the Poisson effect and the pitch effect
is given by the last term. The relative displacement associated with adhesion has
been neglected. This is justified if the concrete reference point is taken at the
interface. However, in general this reference point is taken at some distance from
the tendon surface, which means that before debonding the relative displacement
consists of the elastic deformation of the concrete layer between steel surface and
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
reference point. Under the assumption of a linear relation between slip and bond
stress until the adhesion strength is reached, this aspect can be taken into account
by multiplying the bond stress with a factor q, for which holds:
q=-6 for O < S S ,

'a
q = 1 for 6>óa
with q reduction factor for bond stress before debonding

6, relative displacement at debonding
The total bond force associated with this pre-debonding behaviour is small and
it can not be deduced from the above-mentioned bond tests. The practical
meaning of this phenomenon is discussed further on.
Model Limitations -- The approximate nature of (1) should be kept a.

in mind. First, the given constants will depend on variables that are known to
influence the transfer length such as concrete strength, strand position during
casting, rate of prestress transfer and steel surface roughness. Second, the shape
of the experimental curves is not always similar. For example, tests with
lightweight concrete showed a higher adhesion than found with gravel concrete
of the same compressive strength and at debonding a clear drop of the bond
stress was observed (7). This behaviour may have resulted from a higher
shrinkage and lower elastic modulus of the concrete due to the lower stiffness of
W ûbb2949 0542718 3bï 9
150 den Uiji
the coarse aggregate. Third, the contribution of each parameter has been lineariz-
ed. This puts limits to the applicability of Eq. (1). For example, the average bond
stress in a pull-out test with short embedment length only increases until a slip
of about 5 mm is reached. The range of linear concrete response to tendon
expansion depends upon the cover, and the linear response of the steel holds up
to the yield strength.
Influence of Strand Yielding on Bond Resistance -- With increasing

stress a strand contracts, which reduces its wedging action and thus the bond
resistance. Beyond the yield stress the longitudinal strain increases rapidly and
so does the strand contraction (Poisson effect). At a certain moment the strand
contraction annuls the wedging action caused by steel surface roughness and
lack-of-fit, leaving only the pitch effect to provide bond resistance. The strong
reduction of the bond strength due to yielding will result in a considerable
increase of the development length, as has been discerned by Buckner (8).
Although bond of a yielding strand has not been studied explicitly, an

attempt is made to simulate it with the above-mentioned bond model. To that end
the third term in Eq. (I), which represents the Poisson effect, has been rewritten
as a function of strain instead of stress:
T~= 3 + 0.46 + ~OOAE,+ 1.5 1Aop 1 [MPal (3)
with Acp change in longitudinal steel strain
Eq. (3) is followed until the sum of the first three terms at the right hand
side becomes zero. Then, for negative values of this sum its contribution is
neglected and only the term representing the pitch effect is considered. Simula-
tion results with this approach are discussed further on.
Thick-walled Cylinder Model
The force transfer between a reinforcing tendon and concrete is connected

with radially directed compressive stresses, irrespective of the transfer mechan-
ism. These stresses are equilibrated by circumferential tensile stresses that may
initiate radial cracks starting at the steel-to-concrete interface. The confining
capacity of the concrete, that is the maximum radial compressive stress that can
be resisted, determines the bond resistance in case of bond splitting failure. For
the evaluation of this confining capacity, the concrete around the tendon is often
conceived as a partly cracked thick-walled cylinder. Analytical solutions describ-
ing the internal pressure as a function of the crack penetration depth have been
obtained by assuming a certain distribution of the circumferential stresses over
the cracked part of the cylinder wall, and more recently by taking into account
the softening behaviour of concrete loaded in tension. A survey of both analytical
and numerical solution methods has been given by Noghabai (9).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Obb2949 054271i9 2T3 Tribute to Peter Gergely 151
In many cases the interest in the response of the thick-walled cylinder

concerns only the maximum value of the internal pressure, which is related to the
maximum bond stress by choosing the angle between the resultant of the radial
stress and the bond stress and the tendon axis. However, to include for example
the Poisson effect in a bond model based on confining capacity, the relation
between the radial displacement of the interface and the radial compressive stress
at the interface has to be known. To that end the stresses and deformations in a
thick-walled cylinder under internal pressure are estimated taking into account
concrete softening by means of the fictitious crack model assuming a bi-linear
softening law. The resulting confinement model has been used to analyze the
bond behaviour of ribbed bars included the influence of Poisson contraction after
yielding of the bar (IO). It has also been used to explain the influence of concrete
cover on transfer length and to analyze the bond splitting behaviour of strand in
NSC and HSC (1 1).
BOND STRESS DISTRIBUTION FROM LOCAL BOND MODEL
With the above-mentioned local bond model the bond stress along the
transfer length and development length has been described using the finite
difference method. This method considers equilibrium of forces and compatibility
of strains and displacements in subsequent elements of finite length, along which
a uniform bond stress distribution is assumed, taking into account the constitutive
laws for concrete, steel and bond (Fig. 2). The stress-strain behaviour of concrete
is assumed to be linear, whereas for steel an elasto-plastic behaviour is taken into
account. The bond stress follows from Eq. (3) and the multiplier defined in
Eq. (2) is applied. The boundary conditions are indicated in Fig. 3.
Transfer Length
The bond stress along the transfer length is found starting at the end of
the transfer length, where the concrete strain equals the elastic shortening of the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
strand that occurs at prestress transfer. The calculation procedure starts with the
assumption of a small initial slip mm) and it ends when the steel stress
becomes zero. Fig. 4 shows the resulting bond and steel stresses along the
transfer length for a concentrically embedded single strand. The results for the
flexural bond length are discussed further on. In these simulations the relative
displacement at debonding (6,) has been varied. It is clearly shown that the
transfer length increases with 6,. However, the distance over which for example
95% of the prestress is being built up, is not influenced by the value of 6,.
Bond Stress Distribution -- Another criterion to judge the need for

considering the pre-debonding behaviour in the bond model may be adopted from
the distribution of the longitudinal strains at the concrete surface, which is often
152 den Uiji
= Obb2949 0542720 TL5 m
supposed to be similar to the distribution of the steel strain. 2D-FEM-analyses
reflected, however, that this strain distribution is dispersed (12). This is
demonstrated in Fig. 5, where the concrete surface strain is given for a uniform
bond stress distribution. By comparing the actual bond stress distribution to that
obtained by differentiating the concrete strains, it is apparent that concrete strain
measures do not provide reliable information about the pre-debonding behaviour.
Anticipating the discussion on development length, it is noticed that the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
inclusion of the pre-debonding behaviour yields a too conservative value of the
development length when the latter is taken as the sum of the transfer length and
the flexural bond length. On the other hand, neglecting this pre-debonding
behaviour will not result in an unconservative value of the development length.
The peak value of the bond stress is found near the end face, where the
expansion of the strand at prestress transfer is maximum. It is noticed that this
peak value can only occur if sufficient confinement is delivered by the concrete,
otherwise the maximum bond stress is limited until a level that depends on the
available confinement. In the latter case the distribution of the bond stress will
be flattened near the end face. This aspect will be discussed further on.
Draw-in and Transfer Length -- The relation between the transfer length
and the slip of the strand at the end face (6,) is defined by the distribution of the
steel stress, which can be expressed as:
E
1, = a P 6, (4)
O,
with a shape factor for steel stress distribution. For linear ascending,
uniform and linear descending bond stress holds a = 1.5, 2.0 and
3.0, respectively.
E, Young’s modulus of strand
op0 steel stress before prestress transfer
6, draw-in
A characteristic upper bound value of a = 2.86 has been reported in ( 1 3),

based on a statistical evaluation of many laboratory and field observations. Tests
on gravel concrete have given average values of a ranging from 2.3 to 2.6 with
a standard deviation of 0.3 (12)(14). It is noticed that the accuracy of a strongly
depends on the accuracy with which the draw-in and the transfer length are
estimated. In this respect the estimation of the transfer length is the most
questionable, considering the scatter in concrete strain measurements and the
above-mentioned dissimilarity between concrete surface strain and steel strain.
In the simulations a lies between 2.4 and 2.7 for 6,= O to 0.01 mm, which is
in agreement with the test values.
0bb2q4q 0542721, q53 Tribute to Peter Gergely 153
Transfer Length -- The above-mentioned statistical evaluation of measur-

ing results (13) has yielded the following expression for the transfer length:
with \v = 7, 10 and 13 for the characteristic lower bound, average and

characteristic upper bound value, respectively.
0,, diameter of prestressing strand
opi steel stress just after prestress transfer
fWi cube compressive strength at prestress transfer
The simulation results are in good agreement with Eq.(5) showing values of yf
between 9.0 and 10.4 for S,=O to 0.01 mm.
Development Length
To find the development length of a pretensioned strand in a beam loaded

in flexure, the tensile zone of the beam is conceived as a concrete prism from
which the concentrically embedded pretensioned strand is pulled out. In the
simulation the flexural bond length is calculated and added to the transfer length,
which results in the development length. The procedure to calculate the flexural
bond length is similar to that of the transfer length, starting at the end of the
flexural bond length with an initial slip, but the boundary conditions differ as
shown in Fig. 3.
Bond Stress Distribution -- Fig. 4 demonstrates that the inclusion of the

pre-debonding behaviour in the bond model creates a valley in the bond stress
distribution, which grows with Sa. As a consequence there is a corresponding
increase in the transfer and the flexural bond lengths. However, the bond stress
valley will gradually be filled up when the pull-out force is raised, and general
slip of the tendon will occur as soon as the strand has been debonded over the
entire length of the valley. Therefore the development length can be calculated
to the neglect of the pre-debonding behaviour (Le. for 6, equal to zero).
Strand Yielding -- Until now the stress at the loaded side of the strand has
been below the yield stress. The ability to simulate the development length in
case of a yielding strand using the above-mentioned approach is demonstrated by
comparing the calculated steel strain distribution with the observed one in an
experiment reported by Mitchell et al. (15). The cylinder strength in the experi-
ment amounted to 2 1 and 3 1 MPa at prestress transfer and bending test, respect-
ively. The cube strength of the concrete used for the estimation of the coeffi-
cients in Eq. (1) was 55 MPa. This difference is taken into account by multiply-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 O662949 0542722 898 M
154 den Uijl
ing the calculated bond stress with the square root of the ratio between both
concrete strengths. However, in the simulation no distinction has been made
between the strength at prestress transfer and at the bending test and an average
cube strength of 32 MPa has been used. The calculated transfer length of 848
mm agrees well fairly with the average value of 804 mm found in the test.
A stress of 1686 MPa applied to the loaded end of the strand yields a
steel strain of 0.019, which equals the maximum strain recorded just before
failure. The simulated steel strain distribution (Fig. 7) compares well with the
measured one (Fig. 6). Two bends (points A and B, Fig. 7) in the bond stress
curve correspond to those in the stress-strain diagram of the steel. The third bend
(point C) marks the point where strand contraction due to the Poisson effect
becomes greater than wedging caused by roughness of the sliding plane and lack-
of-fit. Beyond that point the contribution by wedging is fixed at zero leaving
nothing but the pitch effect to provide bond.
Failure Behaviour -- In the test a small slip had been observed prior to
flexural failure, which indicates that the flexural bond wave had reached the end
of the transfer length. Hence, the 1865 mm embedment length was close to the
development length if first free end slip were chosen as failure criterion. Under
this condition the calculated development length amounts to 1777 mm, which
again agrees fairly well with the experiment.
In the case of a constant embedment length and increased pull-out loading

the length over which the strand yields becomes longer (Fig. 8). This yield
penetration is coupled with a strong reduction of the bond stress over the flexural
1 bond length. On the other hand, the slip along the entire embedment length
strongly increases, which is attended with higher bond stresses over the transfer
length. To what extent this increase compensates the above-mentioned loss will
depend on the confining capacity of the concrete and the slip beyond which lack-
of-fit does not further increases the bond resistance (about 5 mm).
Design Rules for Development Length -- The development length

formula for plain seven-wire strand included in AC1 318 since 1963 reads:
o f -o
1, = z O p+ L!210p
21 7
with op- prestress including all losses

fpd tendon tensile strength
and constants in MPa
However, this expression has appeared to be unconservative, among other

things due to the increase of effective prestress enabled by the higher strength
and lower relaxation of modern strands. Buckner (8) proposes a more conser-
vative expression by replacing the effective steel stress in the first term of Eq.
(6) by the steel stress just after prestress transfer and by introducing a multiplier
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Ob62949 0542723 724
h in the second term, that takes into consideration the reduction of the bond
stress over the flexural bond length caused by yielding of the strand
o. f -0
1 , = 2 0 , + ) t Pd p - 0 , (7)
21 7
with l.OS[h =0.6+40ePd]S2.0

e# tendon strain that corresponds to fpd considering strain com-
patibility over the cross-section at compression zone failure
The first term at the right hand side of this equation is considered to be
an upper bound limit of the transfer length. However, for verification of the
member shear resistance an average value suffices since in AC1 318 a strength
reduction factor of 0.85 is applied to the shear resistance. Therefore Buckner (8)
proposes to use 1, = 600, instead of the current 1, = 500,.
The MC90 clauses on transfer and development lengths are summarized

in Annex A. In Fig. 9 the development length according to AC1 318,Buckner
and MC90 are compared with an upper bound of the required embedment length
that follows from the above-mentioned bond model. The latter has been derived
taking into account a 30% lower bond strength and a reduction of the steel stress
over the transfer length that is proportional to the ratio between the effective
prestress and the steel stress just after prestress transfer.
Let’s compare the differences in the formulations in AC1 318 and MC90.
In MC90 the bond strength is proportional to the tensile strength. On the
contrary, in AC1 318 the concrete quality is not included. Further, MC90
distinguishes between gradual and sudden release by defining a 25% longer trans-
fer length in the latter case (Annex A, factor a&,whereas AC1 318 does not
discriminate in this respect.
MC90, which has been verified against 150 test results (3), requires a
longer embedment length for the development of a certain steel stress than AC1
318.Compared to MC90, Buckner’s proposal yields a shorter embedment length
for high reinforcement ratios, which involve a small steel strain at flexural failure
(X+1 ), but a longer embedment length for low reinforcement ratios ( b 2 ) and
steel stresses above the prestress level. The calculated upper bound limit of the
required embedment length follows Buckner’s proposal for kl until the yield
stress (e.g. 0.1% proof stress 0.9fp) is reached, thereafter it rapidly increases
with the steel stress, or rather with the steel strain, and approaches the proposal
for k2.This emphasizes the importance of including the influence of yielding
on the bond resistance of strand. However, instead of reducing the bond strength
over the entire flexural bond length as suggested by Buckner it seems more
rational to apply such’a reduction only over the part of the flexural bond length
where yielding may occur. This would result in a tri-linear envelop curve where
the third branch starts for example at the 0.1% proof stress of the strand having
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
156 den Uiji
an inclination of one fifth of that of the second branch.

- Obb2949 0542724 660
CONFINEMENT FROM THICK-WALLED CYLINDER MODEL
Influence of Concrete Confinement on Transfer Length
The transfer length versus concrete cover found in experiments on HSC

is shown in Fig. 10. The tests were carried out on rectangular prisms centrically
prestressed with one or two pretensioned strands (0,=9.3 and 12.5 mm, fcci=40
and 70 MPa, o =1570 MPa) (12) and on beams eccentrically prestressed with
three strands ($p=i2.5 mm, fcci=65MPa, ow=1340 MPa) (14). For reasons of
comparison the transfer length has been normalized using Eq. ( 5 ) and an effective
cover is introduced to take into account the contribution of concrete in the cover
and between adjacent strands. To that end it is supposed that the concrete
between the strands is more effective in carrying the circumferential tensile
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stresses than the concrete in the cover because of a more uniform tensile stress
distribution. This is taken into account by an effectiveness factor 6, which has
arbitrarily been fixed to 1.5:
2c +(m - i)E,s
Ceff =
2m
with c clear concrete cover
m number of strands in the considered row
5 effectiveness factor for concrete between strands
s clear strand spacing
Fig. 10 leads to the following observations. First, with increasing effective

cover the occurrence of visible bond splitting cracks becomes less and beyond
a certain point no splitting cracks are observed. This is discussed in the next
section, Second, with increasing effective cover the transfer length decreases until
it stabilizes at an effective cover of 3 to 4 times the strand diameter. This can be
explained by the non-linear concrete response to the wedging action of the strand
as is discussed below. Third, the value of y! tends to become clearly smaller than
the above-mentioned average of 10 for normal strength gravei concrete. This may
partly be attributed to the relatively high autogenous shrinkage of HSC, which
causes an initial normal compressive stress at the strand-to-concrete interface, and
to an increased density of the interface layer due to the use of silica fume.
Results from the thick-walled cylinder model for various concrete covers
and two types of concrete are given in Fig. 11. NSC and HSC of 40 MPa cube
compressive strength have been simulated by a different tensile strength, viz. 3
and 4 MPa, respectively. Thus, the same bi-linear softening diagram has been
used, which is justified as long as cracks follow the mortar-to-aggregate interface.
At a higher concrete strength cracks will run through the aggregate, which results
= 0bb2949 05r12725 5T7 = Tribute to Peter Gergely 157
in a more brittle behaviour and, hence, in a steeper first softening branch. The
influence of this increased brittleness has also been observed in the tests (1 1).
To understand the influence of the cover on the transfer length it should

be kept in mind that the radial interface displacement has an upper limit given
by the slip related wedging and the Poisson expansion. Further, the relative
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
displacement of the interface due to Poisson expansion diminishes from about

2%0 near the end face of the concrete member to about zero at the end of the
transfer length. Thus, near the end face the top in the response curve has just
been past in the case of NSC and a cover of 30p, but along the remaining part
of the transfer length the top has not been reached. Therefore, an increase of the
cover will not result in a substantial reduction of the transfer length. A smaller
cover, on the other hand, involves a response beyond the top for a greater part
of the transfer length and, hence, a longer transfer length because of the lower
frictional resistance coupled with the lower radial compressive stress acting at the
interface. It is obvious that NSC requires a greater cover than HSC to provide
equal confinement.
Influence of Concrete Confinement on Bond Splitting
In Fig. 12 the occurrence of bond splitting cracks in the above-mentioned

prism tests on HSC (12) and similar tests on NSC (3) is shown as a function of
strand spacing and concrete cover. It can be seen that HSC is clearly less
sensitive to bond splitting than NSC. The no-splitting conditions that correspond
to the shaded areas in Fig.12 are:
Ceff C
for NSC: -22.25 and -22.5
0P 0P (9)
C
for HSC: 2 2 1.5 and -2 1.5
*P 0P
This different behaviour can be explained from the fact that for equal
compressive strength during hardening HSC has a higher tensile strength than
NSC. As a result, the strain at reaching the tensile strength is also higher in HSC
than in NSC. Hence, for equal cover the interface displacement at which the
cover cracks is greater in HSC than in NSC. This is also illustrated in Fig. 1I by
the influence of the type of concrete on the confinement curve keeping in mind
that the radial cracks reach the concrete surface just after the top, but they will
only become visible at a certain crack width, thus somewhere along the
descending branch. In NSC that point is reached at a smaller radial displacement
than in HSC, as follows from the comparison of the corresponding curves for
equal concrete cover.
m 0bb2949 O542726 433
158 den Uijl
CONCLUDING REMARKS
Local bond model -- A bond model for prestressing strand has been
proposed considering the various parameters that play a part in force transfer.
The bond stress is expressed as a function of slip and change of local steel stress
and strain. With this model realistic distributions of the bond stress along the
transfer and flexural bond lengths are found, thus showing the applicability to
these two distinct bond situations. Although not experimentally verified for that
case, the model has also been found to give a realistic bond stress distribution in
the case of strand yielding.
The constants in the model have been found from pull-out and push-in
tests with short embedment lengths. With those tests the influence of various
parameters can easily be estimated. However, attention has to be paid to the
support conditions of the specimen in view of the bursting stresses in the
concrete, especially when the concrete cover is too small to prevent the
occurrence of bond splitting cracks.
Thick-walledcylinder model -- For the evaluation of concrete confinement

on the bond behaviour of prestressing strand and the occurrence of bond splitting
cracks, the relationship between the radial displacement of the interface and the
radial stress acting at the interface was derived considering the post-peak
behaviour of concrete loaded in tension. To that end the strand cross-section is
conceived as circular and the concrete around the strand is conceived as a thick-
walled cylinder, the wall thickness being a weighted average of available cover
and spacing. The influence of transverse reinforcement is not yet included. With
the help of this model the experimentally observed influence of type of concrete
and concrete cover and spacing on transfer length and occurrence of bond
splitting cracks has been explained. It provides a more rational basis for the
development of design rules on concrete cover and strand spacing.
REFEENCES
1. Stocker, M.F., Sozen, M.A.,"Investigation of Prestressed Concrete for

Highway Bridges; Part IV: Bond Characteristics of Prestressing Strand",
University of Illinois. Structural Research Series No. 344, June 1969.
2. Russell, B.W., Bums, N.H., "Design Guidelines for Transfer, Development

and Debonding of Large Diameter Seven Wire Strands in Pretensioned
Concrete Girders", Center for Transportation Research, The University of
Texas at Austin, Research Report 1210-5F, January 1993.
3. Den Uijl, J.A., "Bond and Splitting Action of Prestressing Strand", Intern.
Conf. BOND IN CONCRETE, Riga, October 15-17, 1992, pp. 2/10-2/19.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0bb2949 0542727 37T W
4. Balks, G.L., "Transfer Control of Prestressing Strands", PCI Journal,

November-December 1992, pp. 60-71.
5. Weerasekera, R., Loov, R.E., "Prestress Transfer Length Predicted by

Cohesive Crack Model", Micromechanics of Failure Òf Quasi-Brittle
Materiais, Elsevier Applied Science, 1990, pp. 417-426.
6. Akesson, M., "Fracture Mechanics Analysis of the Transmission Zone in

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Prestressed Hollow Core Slabs", Chalmers University of Technology,

Division of Concrete Structures, Publication 935, 1993.
7. Den Uijl, J.A., "Background of the CEB-FW Model Code 1990 Clauses on
Anchorage and Transverse Tensile Actions in the Anchorage Zone of
Prestressed Concrete Members", CEB Bulletin No. 212, 1992, pp. 71-94.
8. Buckner, C.D., "An Analysis of Transfer and Development Lengths for

Pretensioned Concrete Structures'', Federal Highway Administration,
Publication No. FHWA-RD-94-049, December 1994.
9. Noghabai, K.,"Splitting of Concrete in the Anchoring Zone of Deformed

Bars", Luleå University of Technology, Division of Structural Engineering,
Publication 1995:26L.
10. Den Uijl, J.A., Bigaj, A.J., "A Bond Model for Ribbed Bars Based on
Concrete Confinement", HERON, Vol. 41, NO. 3, 1996, pp. 201-226.
11. Den Uijl, J.A., "Bond of Strands in High Performance Concrete", 4th
International Symposium on the Utilization of High-StrengWigh-Perform-
ance Concrete, May 29-31, 1996, Paris, pp. 1175-1184.
12. Den Uiji, J.A., "High Performance Concrete in Prefab Industry; Part 3:
Transfer length of Prestressing Strand", Delft University of Technology,
Report 25.5-95-3, January 1995 (in Dutch).
13. OleSniewicz, A., "Statistical Evaluation of Transmission Length of Strand",

Research and Design Centre for Industrial Building BISTYP, Warsaw, 1975.
14. Den Uijl, J.A., "High Performance Concrete in Prefab Industry; Part 5:
Anchorage Capacity of Prestressing Strand", Delft University of Technology,
Report 25.5-95-5, January 1996 (in Dutch).
15. Mitchell, D., Cook, W.D., Khan, A.A., "Influence of High Strength Concrete
on Transfer and Development Length of Pretensioned Strand", PCI Journal,
May-June 1993, pp. 52-66.
0662749 0542728 206
160 den Uiji
ANNEX A: BOND OF PRETENSIONED TENDONS IN MC90
This annex summarizes the CEB-FIP Model Code 1990 clauses on

prestress transfer and development length using the CEB-terminology with ACI-
terminology in italics. A rationale of these clauses can be found in (7). In MC90
the treatment of bond in reinforced concrete and in prestressed concrete has been
harmonized. Fig. 13 shows the various items involved in the calculation of the
transmission length (transfer length) and anchorage length (development length).
The basic anchorage length, lb,, which can be considered as the

embedment length required to anchor a non-pretensioned tendon, follows from:
1bP =-- fPd

xOp fbpd
1 2 7 2
with Ap cross-sectional area: wire -nap and strand -x0
4 36
fpd design strength of tendon
fbpd design bond strength
x 0ptendon perimeter irrespective type of tendon
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The design bond strength is proportional to the design tensile strength and
takes into account type of tendon and position during casting:
fbpd = q p i q p 2 fctd (A2)

1 with qPl type of tendon: indented and crimped wire 1.4, plain strand 1.2
%2
casting position: inclined and at top 0.7, at bottom 1.0
fctd design value of concrete tensile strength
fCtk
fc td =-
"I,
~~~
fctk characteristic lower bound value of axial tensile strength

'
I, material factor (1.5 for tensile strength)
It is remarked that in MC90 the relation between tensile strength and

compressive strength is taken as:
fctk= q fy
with q=0.2 for characteristic lower bound value

q=0.3 for characteristic upper bound value
Thus, implicitly the bond stress is taken proportional to f,"." as well.
The transmission length (transfer length) is calculated considering way of
ûhh2949 054272'1 142
release, effect to be verified and bond situation. An upper and a lower bound
value are estimated for moment and shear capacity verification and for analysis
of tensile stresses in the anchorage zone, respectively. The bond situation refers
to the fact that along the transmission length bond is enhanced by transverse
deformation of the tendon at prestress release (Poisson effect):
with c ~ g graduai release 1.0, sudden release 1.25

019 moment and shear verification 1.0, analysis of anchorage zone 0.5
alo Poisson effect: strand 0.5, indented wire and crimped wire 0.7
opi steel stress just after release
(Note: a,to ol, are used with ribbed bars to quantify influence of bar shape,
welded transverse bars, confinement by concrete, transverse reinforcement and
transverse pressure, laps in bars and welded wire mesh, respectively)
The design anchorage length (development length) is found assuming that

along the transmission length the effective prestress (after losses) can be
anchored and that beyond the transmission length the design bond stress is
available:
with lbpt characteristic upper bound value of transmission length

opd design steel stress
op prestress after all losses
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The cover on a tendon and the clear spacing of tendons in members

without confining reinforcement should comply with the following:
if clear spacing 230, cover shall be 2 3 0 ,
if clear spacing <30, cover shall be 240,
W Obb2949 0542730 964
162 den Uijl
-- bond stress T~ N a ]
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
1P
a: puU-out free
P
b pull-out fixed
IP,-P+P,
c: push-in
steel stress change slip s [mm]

PupI [mal
Fig. 1-Bond stress as a function of slip and stress change
X dx x+dx
Fig. 2-Principles of finite difference method
9 Obhë9g9 0542731 &TO = Tribute to Peter Gergely 163
I development l a @ =
transfer length + flexural bond length 1
strand
X=l x=o x=o x=lfl

x=o Sx-0 x=o S,-O
upx=upi opx=Qpi
x=i, u,,-o x = & apx=fp
Fig. 3-Boundary conditions for transfer and flexural bond length calculation with
finite difference method
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
10 T ~u4200
, [MPa], ~ 4 [mm/m]
5
O 0.5 1.0 1.5

distance to end face [m]
Fig. 4-Transfer and development length simulation with local bond model
~
9 Obb2947 0542732 737 m

164 den Uiji
1.00 ~cxi&c,max
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0.75 1.5
1 .o
0.50
os
0.25
O
O
O O5 1.o O 0.5 1.o
dit : 3
Fig. 5-Concrete surface strain for uniform bond stress (left) and bond stress
derived from concrete strain (right)
steel strain E, [mm/ml

24
20 1
16
12
O 1
O 016 1:2 1.8 2.4 3.0 6

position along beam [m]
Fig. 6-Steel strain along beam in bending showing yield of strand;

after Mitchell et al. (15)
9 0662949 0542733 673
7b [Mpal, qJ200 FIPaI, S/2 [-I, cd5 ~Wml

10 ~
8-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O 0.5 1.0 1.5 2.0
distance to end face [m]
Fig. 7-Simulated steel strain along development length with strand yielding
Fig. 8-Simulated yield penetration for a constant embedment length and

increasing puli-out force
~~
W Obb2949 0542734 SOT 9

166 den Uijl
O 60 120 180
240
Fig. 9-Development length according to AC1 318, MC90 and as proposed by

Buckner (8)
20 -symbol - @p fm
* o
v
9.3 40
9.3 70
A 12.5 40
15 - U 12.5 70
O 12.5 64
t :
o uncracked
m cracked
10 -
5-
average of 26 tests
Fig. 10-Relative transfer length as a function of effective concrete cover
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542735 446
24
20
16
O
12
8
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
o
relative strand expansion A 04 [mm/m]
Fig. 1 1-Concrete response to strand expansion as a function of cover and type of

concrete
W Obb2949 054273b 382
168 den Uiji
clear spacing s / 0 "

4.0 24.0
3.0 3.0
2.0 2.0
1.o 1 .o
0.0 0.0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
C/0,4.O C/0I0
24.0
I param. top left topright b~ttomieftl
conc.lype NSC NSC HSC

0,, [mml 12.5 9.38~12.5 9.38~12.
strandnr. 3 1U 1&2
fd IMPal 44.5 33&44 41&71
f,, [Wal 2.6 2.3U.6 4.08~5.2
Op0 Wal 1400 1400 1570
1.o
......
0.0
0.0 1.0 2.0 3.0 4.0
clear cover c/0*
Fig. 12-Occurrence of bond splitting cracks in NSC and HSC (ilj means i out of i
tests presented cracks; shaded area corresponds to no cracking condition)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542737 2lî m
O 50 100
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 13-Definitions used in CEB-FIP Model Code 1990 (MC901
Obb2949 0542738 155 W
SP 180-8
Confinement Role in Anchorage Capacity
by E. Giuriani and 6. A. Plizzari
Synopsis: In the present paper, the role of the confining forces both on bond
capacity and on splitting crack opening under service loads is shown. In par-
ticular, theoretical relationships between the transverse reinforcement area,
the bond strength, the splitting crack opening and the stirrup stress are pre-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
sented. The theoretical predictions are contrasted with some experimental
results and a discussion on the values provided by building codes is presented.
The results show the strong influence of transverse reinforcement whose
confining force is expressed by the stirrup index of confinement, which gov-
erns bond behavior and is suitable for design. The comparison with several
experimental results showed a good agreement between theory and tests.
Adequate values of fl are also required to control splitting crack opening
under service loads. For common amounts of transverse reinforcement, the
splitting crack opening can be larger than one half of the flexural crack, which
could be unacceptable for structural durability.
Keywords: Anchorage; bond strength; bond stress; cracking (fracturing); stirrups
171
W Obb2949 0542739 O91
172 Giuriani and Plizzari
Ezio Giuriani is professor of structures in the Dept. of Civil Engineering of

the University of Brescia. His research interests include bond and cracking
in R.C. structures, steel-concrete composite structures and structural reha-
bilitation of historical buildings. His professional experiences include R.C.
structures, bridges and structural rehabilitation.
Giovanni Angelo Plizzari is an assistant professor in the Department of Civil

Engineering of the University of Brescia. His research interests include steel-
to-concrete bond, fracture and fatigue of concrete.
INTRODUCTION
In anchorages and lap splices, the radial component of the rib force loads
the surrounding concrete like an internal pressure and induces tensile hoop
stresses which cause splitting cracks propagating along the anchored bar [i]
(Fig. i). These cracks impair good bond performance and make bond very
sensitive to confinement [ 2 ] . After splitting, the confining action around the
anchored bars can be produced by transverse reinforcement [i, 3, 41, by ten-
sile strength of the uncracked surrounding concrete and by cohesive stresses
between the crack faces in the split concrete [5, 61 (Fig. 21,
Anchorage capacity has been investigated by several researchers who pro-
posed empirical relationships between bond strength, concrete cover and
transverse reinforcement [4,7,8]. A theoretical model for the local bond stress
- slip relationship after splitting was proposed by Giuriani and co-workers
[9]; this model shows that bond stiffness and strength depends strongly on
the amount of transverse reinforcement. Experimental results also show the
importance of stirrups in limiting splitting crack opening and increasing an-
chorage capacity [i,4, 10, 11, 12, 131. Experiments [14] clearly demonstrate
the well-known bond strength mechanisms. When the confining actions are
insufficient, a “splitting-type” failure occurs, while a “pull-out-type” failure
occurs otherwise because of the shear failure of the concrete corbels between
the bar ribs. Experiments [15] also have shown that stirrups can yield before
reaching the bond strength prescribed by code.
In beams with closely spaced deformed bars and usual concrete cover,
transverse reinforcement provides the most relevant confining action so that
the evaluation of the required amount of stirrups becomes important for de-
sign purposes. In fact, for usual concrete cover and closely spaced anchored
bars, concrete tends to split in the plane containing the bars and the confining
action of the cohesive stresses is negligible [13](Fig. 2).
The evaluation of the minimum stirrup area is very important when epoxy-
coated bars are used since the loss of friction between the bar surface and the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542740 803
surrounding concrete requires more confinement [16, 171.

When splitting of concrete cover occurs, another important aspect con-
cerns structural durability. In experimental test [is],on a R.C.beam, split-
ting cracks formed under service loads starting from flexural cracks. More-
over, after severe loading, the splitting crack did not close and a significant
residual crack width remained. Building codes do not give indications about
the acceptable values for splitting crack opening. It seems reasonable that
splitting crack control should be more severe than for flexural cracks. In fact,
while the latter exposes a very limited bar length to the environment, splitting
cracks develop longitudinally along the bar so that a considerable length of
the reinforcement can be exposed to aggressive agents. As mentioned above,
transverse reinforcement limits splitting crack opening and therefore improves
structural durability.
The problem of splitting-crack control was studied by the authors on the
basis of the theoretical model proposed in [9]. A relationship between the
stirrup area, the splitting crack and the flexural crack opening was obtained,
making the problem of the evaluation of splitting crack opening possible 1191.
In fact, splitting crack opening was determined as a function of flexural crack
opening which is more easily determinable.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
For design purposes, it is important to provide relationships giving the

area of transverse reinforcement both to obtain the design bond strength
and/or to control splitting crack opening. The theortical approach showed in
[9] underlines the importance of two governing parameters, namely the “stir-
rup index of confinement” (0) and the “concrete index of confinement” (B),
whose formulations are based on a rational approach. In the present paper,
on the basis of this theoretical model, relationships between the transverse
reinforcement area, the bond strength, the splitting crack opening and the
stirrup stress are determined. These relationships are particularly significant
for design since they allow designers to determine the amount of transverse
reinforcement both to control splitting crack opening and to reach the design
bond capacity. The theoretical predictions are contrasted with some exper-
imental results and a discussion on the values provided by building codes
[20, 211 is presented.
THEORETICAL CONSIDERATIONS
The “splitting-type” bond failure occurs when the confining forces can no
longer equilibrate the wedge forces provoked by the ribs. As mentioned above,
these forces depend on the splitting crack opening [2]. The “pull-out-type”
failure can also depend on splitting crack opening since, for an increasing crack
width, the wedge force also increases and requires more confinement. The
theoretical model proposed in [9] was developed by imposing the equilibrium
between the wedge forces and the confining forces in the influence zone Az of
one stirrup, where the splitting crack is assumed as completely propagated.
9 Obb2949 054274% 7 4 T 9
For the sake of clarity, the main relationships governing the model are
here reported.
From the experimental results obtained by Gambarova and Co-workers
[22, 231, a relationship between bond stress T , bar-to-concrete slip s and
splitting crack opening w, was determined:
where 4, is the principal bar diameter and rm,wis the bond strength for a
constant splitting crack opening w,. The latter is given by:
From the same experimental results, the following relationship between bond
stress r , wedge pressure u, and w, was proposed:
with the limitation r = T ~when , r~ > Tm,w-

Parameters r m , o , 71, yz, pl, TO, TI, IC1 and IC2 depend on mechanical
properties of concrete and geometrical characteristics of the ribs [24]. Their
values, proposed in [19],were obtained by the best fit to experimental results.
For the confining force of stirrups, the model proposed by Giuriani [25]
was adopted:
where: q5st is the stirrup diameter, E, is the Young's modulus for steel, us
is a factor characterizing the shape of transverse reinforcement [24] (Fig. 3),
a2 = 4r1z4,t/ES, a1 = Sroz/Es, ao = ( a 1 2 r ~ ~ / r ~ l ) / ( 4 a-~1))
( r iand
~ r ri ~
o2,m
and r12are coefficients of the ideal bilateral local bond stress-slip law.
For the confining force given by the cohesive stresses of the split concrete,
the softening law for cracked concrete proposed by Giuriani and Rosati [26]
was adopted:
jct0
= (5)
+1
where feto and K are material constants and q5a is the maximum size of the
aggregate.
For equilibrium, the global radial force produced by the np anchored bars
in Az, has to be equal to the confining force given by nSt stirrup legs and by
the cohesive stresses transmitted by the split concrete; this yields:
un = nust + Bu,, (6)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D 0bb2949 0542742 böb
is the stirrup index of colafinement, Ast is the cross-sectional area of one

stirrup and A; is the section area of one principal bar in the split plane which
is proportional to the radial force (Fig. 4), and
B = ( b - nP4PPz - ( b - n,#p,)Az (8)

nP4&
is the concrete index of confinement, where b is the specimen width.
From Eqs. 3 and 6, bond stress r as a function of ast,u,, and w,can be
Bond Capacity and Transverse Reinforcement
Equations given by theory can be arranged in order to obtain a more suitable

relationship between stirrup index of confinement R and bond capacity ru.
This relationship also depends on splitting crack opening at rupture (wS,+).
At the ultimate limit state, maximum bond stress ru provokes a radial
pressure un,+that can be obtained from Eq. 3:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Stirrup stress osti, and concrete cohesive stresses u,,, can be determined
from Eqs. 4,5 for ws,,.By substituting these values, Eq. 6 yields the following
relationship between R and ru:
with the limitation ast,u5 f,,.

For a given splitting crack opening w,,,,maximum bond stress, stirrup
stress and concrete cohesive stress can be obtained from Eqs. 2, 4 and 5
respectively; the stirrup index of confinement is then determined from Eq. 11.
The rough approximation that no redistribution of bond stress occurs along
the bar is assumed here, so that the local collapse of bond corresponds to the
collapse of the anchorage. When the stirrups yield (ust,,= fs,), a “splitting-
type” failure of bond occurs. A relationship between ust,uand R, which
evidences the range of the “splitting-type” and the “pull-out-type” failures of
bond, was also proposed in [27].
W ûbb2749 0542743 5 1 2 =
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The equations of the theoretical model involve many parameters and co-
efficients; however, only for some of them the variability of their values play
a significant role [24]. The comparison with experimental results allowed to
determine the best fitting value of these coefficients. In [19], a discussion on
these coefficients was presented and average values were proposed for normal
strength concrete; these values are shown in Table 1.
Splitting Crack Opening and Transverse Reinforcement
When the influence zone AZ of the stirrup closest to the flexural crack is
considered (Fig. i), the theoretical model allows one to determine the rela-
tionship between the maximum splitting crack opening and the flexural crack
width at the reinforcement level, which is related to the maximum slip s of
the anchored bar. In fact, s is a fraction of flexural crack opening wf and can
be expressed as
s = - Wf
"f
where af is a coefficient that depends on beam restraints and loading (cxf =
2 in the case of symmetry and af= 1 when the slip ideally occurs only from
one side of the crack, Fig. 5).
Substituting the value of s given by Eq. 12, Eq. 1 can be rewritten as:
The relationship between wf and w, is obtained by substituting Ost (Eq. 4)

and oTc(Eq. 5) in Eq. 9, and then substituting T in Eq. 13. This relationship
I can be solved for wf and written in a dimensionless form as:
The splitting crack opening (w,) as a function of flexural crack opening can
be indirectly determined from Eq. 14. This relationship is valid not only for
anchorages, for which the present theory was developed, but also for splitting
cracks close to flexural cracks in ordinary beams.
Splitting crack control requires the evaluation of w,. In [19], a simplified
expression for w, as a function of wf was determined by means of a Taylor's
series development of Equation 14.
By assuming only the terms of the series which are linearly dependent on
w s and R, and the values of the governing parameters proposed in Table 1,
Eq. 15 can be written as:
W Ob62949 0542744 459
1%- 0.0022 + 0.0034R

_
WS - af 4P
-
h 59.78Q + 0.6703
In this way, the splitting crack opening (w,) can be easily evaluated owing
to its dependence on flexural crack opening (wf)which depends on the loads
through well-known relationships.
CODE PROVISIONS
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Some remarks about code provisions are here presented to make some com-
parisons and a discussion possible.
Eurocode 2 (EC2) [ai] gives provisions for anchorages and splices only
requiring a minimum amount of transverse reinforcement. The AC1 building
code (AC1 318-95, [ao]) recently introduced coefficients related to the concrete
cover and the transverse reinforcement, recognizing the importance of the
confining forces. These provisions regard the anchorage capacity but never
refer to the issues related to durability.
According to Eurocode 2, the anchorage length can be determined by as-
suming a constant bond stress along the bar (fbd,EGZ), which can be expressed
as :
where ^lC (=1.5) is the safety factor for concrete and fbd,EC2 as well as fck
are expressed in MPa. It can be noted that fbd,EC2 is independent of the
transverse reinforcement present along the bar; EC2 only requires a minimum
amount of stirrups.
The AC1 318-95 code [20] prescribes a development length that, for normal
concrete and good bond conditions, is given by:
where y = 0.8 for No. 6 (A 5 19 mm) and smaller bars and y = 1.0
for all other bars, f, is the specified yield strength of reinforcement, fi is
the specified cylinder compressive strength of concrete, e is the smaller of
either the distance from the center of the bar to the nearest concrete surface
or one half the center-to-center spacing of the bars. Ki,. is the transverse
reinforcement index which is given by:
IL -
_ - nstAstfyt
i$p 1500q5pAznp
where fyt is the specified yield strength of transverse reinforcement. In Eqs. 18
and 19, the stresses are expressed in psi.
m 0662949 0542745 395
The term in the denominator of Eq. 18 that takes into account the con-
fining force provided by concrete cover and transverse reinforcement has the
limit at ion:
c ICtr
(- -) 5 2.5 + (20)
4P 4,
Note that ICtr is a coefficient similar to the stirrup index of confinement R
(Eq. 7); in fact, it can be written as:
By assuming a uniform bond stress distribution along the bar and expressing
the stresses in MPa, the bond strength according to AC1 318-95 becomes [15]:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
and the limitation of Eq. 20 can be written as:
It is worth noting that the bond strength according to AC1 is linearly depen-
dent on the concrete cover (or on the clear spacing between the bars) [7, 81,
and on the stirrup index of confinement, up to a value flrnax,~c1
beyond which
a constant bond strength should be adopted.
COMPARISONS AND DISCUSSION
Some theoretical results are here presented to show the role of the confining
reinforcement both on bond capacity and on splitting crack opening. Com-
parisons with experimental results and code provisions are also shown and
discussed. The theoretical results are determined by using the values of the
coefficients proposed for normal concrete in Table 1.
The diagrams shown in Fig. 6a are obtained from Eq. 11 and exhibit bond
capacity ru as a function of the amount of confining reinforcement which is
expressed by the stirrup index of confinement R. The four curves correspond
to different stirrup diameters through the ratio as4st/q5p, and to a constant
value of the concrete index of confinement (B=2). The circles indicate the
beginning of stirrup yielding, as also shown in Fig. 6b, by assuming a stir-
rup yield strength fsy = 400 MPa. The curve connecting the circles is the
“splitting-type” failure curve.
Figure 6a shows that the smaller the stirrup index of confinement is, the
smaller the bond capacity becomes. Furthermore, for a chosen value of 0, the
smaller the stirrup diameter, the higher the bond strength is. In fact, small
diameter stirrups have a stiffer behavior which limits splitting crack opening.
Obb2949 054274b 22%
Figures 7a and b show the negligible influence of the concrete index of

confinement on bond strength ru and on stirrup stress Ost,%. In fact, this
parameter is related to the cohesive stresses between the split concrete faces
that produce a small confining force for usual concrete covers and clear spac-
ing between the bars. The three curves correspond to different values of B
(=0,2,4) and to a constant value of parameter c~,&/q5~(=0.6).
The same diagrams of Fig. 6 are shown in Fig. 8 in a more suitable form for
practical applications. Here, the bond strength is plotted versus the stirrup
spacing AZ for the case of a beam with three 24 mm diameter anchored bars
and three different (twc-legged) stirrup diameters (Fig. 8d). The diagrams
show that bond capacity dramatically decreases for increasing values of the
stirrup distance. Circles indicate the stirrup yielding onset ( f s , = 400 MPa),
which occurs for very small values of Az, as shown in Fig. 8b. Figure 8c
exhibits the splitting crack opening at failure (waU)as a function of R; it can
be noticed that w,, reaches high values.
Figure 9 exhibits a comparison between experimental and theoretical re-
sults. Each experimental point of Fig. Sa represents the average value of the
bond strength obtained by several pull-out tests performed by Plizzari and
Marchina [28]. They simulated short anchorages (Az=150 mm) with two q$,
24 mm anchored bars and one two-legged stirrup having different diameters.
Notice that the theoretical points obtained by Es. 11 match the experimental
results quite well. The experimental results from specimens having small stir-
rup diameters show a higher bond strength. However, Plizzari and Marchina
indicated that the stirrups yielded so that the higher strength could be justi-
fied by the stirrup steel hardening, not considered by the present model.
The building code provisions are also indicated in Fig. 9 for a comparison,
and refer to concrete having a compressive cylindrical strength of 25 MPa.
It should be observed that the codes refer to the design value of the bond
strength which includes the safety factor. It can be noted that the bond
strength prescribed by AC1 follows the general trend of the experimental
results owing to the linear increase with R. On the contrary, Eurocode 2
increasingly underestimates the experimental results when large stirrup areas
are adopted. For small values of R (< 0.006), the theory gives lower bond
strength values than that proposed by the building codes.
Figure 9b exhibits the experimental results obtained by Eligehausen and
co-workers [3], by performing pull-out tests on short anchorages. The authors
found a negligible influence of stirrups on bond capacity. However, it is worth
mentioning that they tested anchorages characterized by very high values of
stirrup index of confinement (0 > 0.05), for which no significant differences
in bond capacity occurs. The theoretical predictions obtained by using the
same coefficients adopted and discussed in [9, 281, are also shown in Fig. 9.
As can be noted, the theoretical predictions match the experimental results
very well.
Figures 10-14 refer to the problem of splitting crack control under service
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D 0bb2949 0542747 Ib8
conditions. Some tests concerning R.C. beams showed that splitting cracks
form under service loads starting from the flexural cracks [18, 29, 301.
Figure 10 refers to a beam with a lap splice well confined by 10 mm
diameter stirrups (Az=lOO mm); outside the lap length, the stirrups were
placed at a double distance (Az=200 mm) 1181. The first splitting crack
developed starting from a flexural crack formed at the splice end outside
the lap zone, where the confinement was lower. When the steel stress in
the principal bars reached a value of 200 MPa, which is common for service
loads, the flexural crack opening was about 0.1 mm and the splitting crack
width was about 0.06 mm (point A in Fig. il). After unloading, a significant
residual splitting crack opening (w,=0.046 mm) was still present (point B).
In Fig. 12, the maximum splitting crack opening as experimentally de-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
termined by Plizzari and co-workers [15] is plotted versus the flexural crack
opening. The authors performed several pull-out tests on anchorages with two
20 mm diameter anchored bars and stirrups having different diameters; the
splitting crack opening was measured near the stirrup closest to the loaded
end of the bar. In the same figure, the theoretical curves obtained by Eq. 14
and by the approximate relationship (16) are also shown. A good agreement
between the experimental and theoretical results can be noted.
Figure 13 exhibts theoretical dimensionless curves of splitting crack open-
ing versus flexural crack opening, for different values of R. These curves were
determined by the coefficients of Tab. 1 and assuming (Y,&/&, = 0.75 and
a j = 1.5. The good agreement between the theoretical and the approximate
curves can be observed. Similar curves are plotted in Fig. 14 in a dimensional
form for different values of the stirrup distance Az. They refer to the same
beam of Fig. 8d with three anchored bars (& = 24 mm) and two-legged stir-
rups (q5st = 8 mm). Notice that splitting crack opening can reach remarkable
values (up to 0.2 mm), when the flexural crack opening is about 0.3 m, which
is significant for service loads. To limit the splitting crack opening, a large
amount of transverse reinforcement should be adopted. For instance, a stir-
rup distance of 50 mm, which is quite unusual for ordinary beams, would be
necessary to limit splitting crack opening below 0.1 mm.
CONCLUDING REMARKS
In the present paper, theoretical results are presented showing the role of the
confining forces both on bond capacity and on splitting crack opening under
service loads, which is important for structure durability.
In particular, the results show the strong influence of transverse reinforce-
ment. Its confining force is expressed by the stirrup index of confinement R
(Eq. 7), which governs bond behavior and is suitable for design. For common
concrete covers, the confining force provided by the the split concrete cohesive
stresses plays a minor role.
Ob62949 05427Y8 OT4
The comparison with several experimental results showed a good agree-

ment between theory and tests (Fig. 9). A discussion on code provisions, again
underlines the importance of transverse reinforcement on bond strength. For
low values of stirrup index of confinement Q, some code provisions give a
safety factor lower than for higher values of 0.
Adequate values of R are also required to control splitting crack open-
ing under service loads. For common amounts of transverse reinforcement,
the splitting crack opening can be larger than one half of the flexural crack
(Fig. 14), which could be unacceptable for durability. In fact, according to
the authors’ opinion, the risk of bar corrosion due to splitting cracks should
be higher than that due to flexural cracks. Therefore, the allowed maximum
splitting crack opening should be remarkably lower than the maximum crack
width allowed for flexural cracks. The lack of experimental results on the
splitting crack related durability as well as the lack of code provisions on this
topic should be remarked.
REFERENCES
[i] R. Tepfers. A Theory of Bond applied to Overlapped Tensile Reinforce-

ment Splices for Deformed Bars. PhD thesis, Chalmers University of
Technology, Goteborg, 1973.
[2] G.P. Gambarova and E. Giuriani. Discussion of ‘Study of the transfer of
Tensile Forces by Bond.’ by D. H. Jiang, S. P. Shah, and A. T. Andonian.
Proceedings, American Concrete Institute .Journal, 82(3):381-383, 1985.
[3] R. Eligehausen, V.V. Bertero, and E. P. Popov. Local Bond Stress-Slip
Relationships of Deformed Bars Under Generalized Excitations: Tests
and Analytical Model. Technical Report UCB/EERC-83, Earthquake
Engineering Research Center, University of California, Berkeley, Califor-
nia, 1983.
[4]D. Darwin, 2. Zuo, M.L. Tholen, and E X . Idun. Development Length
Criteria for Conventional and High Relative Rib Area Reinforcing Bars.
ACI Structural Journal, 93(3):347-359, 1996.
[5] G.A. Plizzari, C. Schumm, and E. Giuriani. The Effect of Residual
Tensile Strength of Cracked Concrete on the Local Bond-Slip Law after
Splitting. Studi e Ricerche, School for the Design of R.C. Structures,
Malan University of Technology, 9:129-155, 1987.
[6] P. Balaguru, P.G. Gambarova, G.P. Rosati, and Schumm C.E. Bond of
Reinforcing Bars and Prestressing Tendons in HPFRCC Matrices. In
A.E. Naaman and H.W. Reinhardt, editors, Second International Work-
shop: High-Performance Fiber Reinforced Composities - Volume 2, pages
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 05Li2749 T30
325-363, AnnArbor, Michigan, June 11-14 1995. University of Michigan

- Dept. of Civil and Environmental Engineering.
[7] C.O. Orangun, J.O. Jirsa, and J.E. Breen. A reevaluation of Test Data
on Development Length and Splices. ACI Journal, 74(11):114-122, 1977.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
[8] D. Darwin, S.L. McCabe, E.K. Idun, and S.P. Schoenekase. Development
Length Criteria: Bars Not Confined by Transverse Reinforcement. ACI
Structural Journal, 89:709-720, 1992.
[9] E. Giuriani, G.A. Plizzari, and C. Schumm. Role of Stirrups and Resid-
ual Tensile Strength of Cracked Concrete on Bond. ASCE, Journal of
Structural Engineering, 117:1-18, 1991.
[lo] S.Morita and S.Fujii. Bond Capacity of Deformed Bars due to Splitting
of Surrounding Concrete. In P. Bartos, editor, Bond in Concrete, pages
331-352. Applied Science Publishers, London, England, 1982.
[il] T. Kaku, M. Yamada, M. Shibata, and H. Takahashi. Bond Splitting
Strength of Simply Supported Reinforced Concrete Beams. Transactions
of the Japan Concrete Institute, 1988.
[12] M. Maeda, S. Otani, and H. Aoyama. Effect of Confinement on Bond

Splitting Behavior in Reinforced Concrete Beams. IABSE, Structural
Engineering International, 5(3):166-171, 1995.
[13] G.A. Plizzari, A.M. Deldossi, and S.Massimo. Experimental Study on
Anchored Bars in R.C. Elements with Transverse Reinforcement. RILEM
Materials and Structures, 29:534-542, 1996.
[I41 T.P. Tassios. Properties of Bond between Concrete ans Steel under
Load Cycles Idealizing Seismic Actions. In Italian Concrete Associa-
tion, editor, Symposium on Stuctural Concrete under Seismic Actions,
CEB Bullettin N0.131,pages 67-122, Rome, Italy, 1979. CEB - Comite
Eurointernational du Beton.
[15] G.A. Plizzari, A.M. Deldossi, and S.Massimo. Transverse Reinforcement
Effects on Anchored Deformed Bars. Magazine of Concrete Research,
1996. Accepted for publication.
[is] C.J. Hester, S. Salamizavaregh, D. Darwin, and S.L. McCabe. Bond of
Epoxy-Coated Reinforcement: Splices. AC1 Structural Journal, 90:89-
102, 1993.
[17] J. Cairns and R. Abdullah. Foundamental Tests on the Effect of Epoxy
Coating on Bond Strength. AC1 Materials Journal, 91(4):331-338, 1994.
Obb2949 0542750 752 = Tribute to Peter Gergely 183
[18] G.A. Plizzari, E. Marchina, and E. Giuriani. Experimental Study of

Splitting and Flexural Cracks in a RC Beam with Overlapped Splices.
RILEM Materials and Structures, 29:19-27, 1996.
[19] E. Giuriani and G.A. Plizzari. Interrelation of Splitting and Flexural
Cracks in Reinforced Concrete Beams. ASCE, Journal of Structural
Engineering, 1997. Accepted for publication.
[20] AC1 Committee 318. Building Code Requirements for Reinforced Con-
crete, (ACI 318-95). American Concrete Institute, 1995.
[21] Commission of the European Communities. EUROCODE N0.2: Design
of Concrete Structures - Part 1: General Rules and Rules for Buildings,
1989.
[22] P. G. Gambarova and C. Karakoç. Shear-Confinement interaction at the
Bar-to-Concrete Interface. In P. Bartos, editor, Bond in Concrete, pages
82-98. Applied Science Publishers, London, England, 1982.
[23] P.G. Gambarova, G.P. Rosati, and B. Zasso. Steel-to-Concrete Bond

after Splitting: Test results. RILEM Materials and Structures, 22:3547,
1989.
[24] E. Giuriani and G.A. Plizzari. Local Bond-Slip Law after Splitting of
Concrete. Studi e Ricerche, School f o r the Design of R.C. Structures,
Milan University of Technology, 7:57-118, 1985. in Italian.
[25] E. Giuriani. On the Effective Axial Stiffnes of a Bar in Cracked Concrete.
In P. Bartos, editor, Bond in Concrete, pages 107-126. Applied Science
Publishers, London, England, 1982.
[26] E. Giuriani and G. P. Rosati. Behaviour of Concrete Elements under

Tension After Cracking. Studi e Ricerche, School for the Design of R. C.
Structures, Milan University of Technology, 8:65-82, 1986. (in Italian).
[27] G.A. Plizzari and C. Schumm. On Bond Collapse due to Bar Pull-out
and Concrete Splitting. Studi e Ricerche, School for the Design of R. C.
Structures, Milan University of Technology, 12:81-116, 1990. in Italian.
[28] G.A. Plizzari and E. Marchina. Studio del collasso dell’aderenza in pre-
senza di armatura trasversale. Technical Report 53/96, Department of
Civil Engineering, University of Brescia, Brescia, Italy, 1996. in Italian.
[29] E. Giuriani and P. Ronca. The grid moiré method applied to study a
reinforced concrete beam. In Atti del VII Convegno Nazionale A.I.A.S.,
pages 655-668, Cagliari (Italy), September 26-29 1979. in Italian.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m 0662949 05i42751 b y 9 =
[30] E. Giuriani and G.A. Plizzari. Propagation and distance of cracks in r.c.
beams with a bending moment gradient. Studi e Ricerche, School for the
Design of R. C. Structures, Milan University of Technology, 11%-106,
1989.
TABLE 1-PROPOSED VALUES OF THE GOVERNING PARAMETERS FOR

NORMAL STRENGTH CONCRETE
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Oh62949 0542752 525
cplittin flexural crack
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 1-Flexural and splitting cracks in R.C. elements [19]
= Obb2949 0542791 275 111
Qst Qp
1w partially developed
\ \transverse bar oction

tensile stress of
uncracked concrete
Fig. 2-Splitting crack and confining forces around a ribbed bar [91
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
a S =I a, =2
Fig. 3-Constrain conditions of the transverse bars i241
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ast
+-
Fig. 4-Geometrical characteristics of principal and transverse bars [91
* a r =2
L 1
v I
i
9
1
Wf =s, Wf = 2 s , =2s,
Fig. 5-Constrain conditions of the principal bars 1191
188 Giuriani and Plizzari m 0662949 05112793 O48 m
500
I I
400
300
200
0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Stirrup Index of Confinement !2 Stirrup Index of Confinement Q
Fig. 6-Influence of parameter ctsOstjOp in the diagrams of bond strength (a) and
maximum stirup stress (bl versus the stirrup index of confinement
P'
10
4
300
400
.200
;_'
i
..-----i
--
.
I
B=O
B=2
8=4
\
,
o' ' I ' I ' ! ' I ' 01 " ' I ' I ' I "
0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05
Stirrup Index of Confinement Q Stirrup Index of Confinement s2
Fig. 7-Influence of parameter B in the diagrams of bond strength (a) and

maximum stirrup stress (b) versus the stirrup index of confinement
Obb2949 0542794 TI!!?-=
500
c
I l
fS"
400
300 ........... $*= 6 rnm

-$=, 8 mrn
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
200
O mm
$,=l
- 0
O 50 100 150 200 O 50 100 150 200
Stirrup distance & [mm] Stirrup distance Az [mm]
np= 3
-E
0.3
nst=2
u
E 0.2 8 , =24mm
b=150mm
0.1 -$=, 8 mm
0.0' " ' I ' I ' I

L
1
b 1
L (d)
O 50 100 150 200
Stirrup distance Az [mm]
Fig. 8-Influence of the stirrup diameter in the diagrams of bond strength (a),
maximum stirrup stress (b) and maximum splitting crack opening (c) versus the
stirrup distance
= 0662949 0542795 910 =
0.25 1 O Tests
Model (Eq.11)
[28]
15
o t
B=6.5
Ij;
..........
8 o Tests [3]-No Stirrups
P
w-
O 0
_ -_ _ - - - z o Tests [3]-Os,=6.37mm
o Tests [3]-$,,=12.7 mm
-2 0.1o I?' 5
A Tests [3]-@,=25.4m m
0.05 1 B=l.75 (4
0.00
0.00
' 0.01
I I
0.02
"
0.0 0.2 0.4 0.6 0.8
Stirrup Index of Confinement R Stirrup Index of Confinement R
Fig. 9-Bond capacity z, versus stirrup index of confinement, as obtained in tests

[28](a), in tests [31 (b) and by theoretical predictions
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0 5 4 2 7 ï b 857
20 10 10 10 10 I O 10 20
t t t t t t t t t
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 10-A splitting crack and flexural crack in the R.C. beam tested in [18]
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542798 b2T m
0.02 I I
I- ........... Tests [151$&nm
Theoty(Eq.14) I
Ø
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0.00
0.000 0.005 0.010 0.015 0.020
Wf 4 p
Fig. 12-Splitting crack opening versus flexural crack opening, as obtained in

tests [15] and by the theoretical predictions
.- &2=0.015Eq.14
-ík0.030
----
Eq.14
Sk0.002Eq.16
,/
/ / /
0.010 // 00°'
& = 50mm
Az=lüûrnm
& = 2 û û m x
-
E
0.2 .......... & = 300 mm ........
E
Y
as=1.5 ,,,.
.. /
Fig. 14-Influence of the stirrup distance on the splitting crack-flexural crack

opening relationship
m Obb2949 0542799 5bb m
SP 180.9
Bond of Ribbed Bars Modeled

through Concrete Confinement
by A. J. Bigaj, J. A. den Uijl and J. C. Walraven
Synopsis: Bond properties are usually described by empirical relations that are
based on pull-out tests of bars with short embedment lengths cast in low to
medium strength concrete. The limited validity of these formulations is
recognized and their applicability in structural analyses is reconsidered.
Results of two test series with various confining conditions and concrete
strengths provide the basis for the derivation of a new bond model for ribbed
bars. Pull-out bond failure in confined concrete and splitting bond failure in
unconfined concrete have been studied. Steel yielding is found to have a
considerable influence on bond strength. Significant differences in bond of NSC
and HSC are confirmed.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
An analytical bond model for ribbed bars is developed. It is based on the

confining capacity of the concrete surrounding the bar. This is evaluated with a
thick-walled-cylinder model, from which the relation between the radial
displacement and the radial compressive stress at the steel-to-concrete interface
is derived. The radial displacement at the interface is linked to the slip of the
ribbed bar, distinguishing between the two modes of bond failure: pull-out and
cover splitting. The model takes into account concrete toughness and bar
contraction, also after yielding. Verification of the model against selected
experimenta1 results reflects the potential of the model to be used in a broad
range of applications.
In its present form the model is used for the analysis of isolated bond
problems using a one-dimensional finite difference approach, but its application
is also considered to count for three-dimensional deformations in FE codes that
treat the bar as a one-dimesional element.
Keywords: Bonding; bond splitting; bond stress; concrete durability; confined

concrete
195
Obb2949 0542600 O06
196 Bigaj, den Uiji and Walraven
Agnieszka J. Bigaj - research associate at the Division of Mechanics and

Structures, Delft University of Technology (till April 1997) and at the Institut für
Werkstoffe im Bauwesen, Stuttgart University (from April 19971, member of
CEB Task Group 2.2 "Ductility" and CEB Task Group 2.5 "Bond Models"
Joop A. den Uijl - associate professor at the Faculty of Civil Engineering, Delft
University of Technology, member of CEB Task Group 2.5 "Bond Models",
reporter on part "Bond of Prestressing Steel", member of Rilem TC 147-FMB
"Fracture Mechanics to Anchorage and Bond"
Joost C . Walraven - professor of Concrete Structures, Delft University of

Technology, vice-chairman of CEB Committee "Material and Behavioral
Modeling", member of CEB Commission I1 and CEB Task Group 2.2 "Ductility"
PROBLEM DEFINITION
Bond of reinforcing bars to concrete influences the behavior of reinforced

concrete structures in many respects. It affects not only the anchorage of bars and
the strength of lap splices, but also the deformation capacity of the members.
Bond plays an important role in the cracking behavior (crack widths and crack
spacing) and the tension stiffening. Also, the length of a plastic hinge, which is
of importance for the rotation capacity of reinforced concrete members, depends
among other things on the force transfer from reinforcement to concrete between
subsequent bending cracks (1). For a refined analysis of these phenomena, a bond
stress-slip relation is required that accurately takes into account the response of
concrete surrounding the bar to the slip and the transverse deformation of the bar,
This behavior is particularly important as current design codes do not consider
the influence of bar stress on bond stress.
Until now bond models have mostly been based on pull-out tests with
short embedment length. For normal concrete grades and normal type of
reinforcing steel, yielding of the steel will not occur in such a test. Hence, the
empirical bond stress-slip relations derived from such tests will reflect only
softening of the concrete surrounding the bar, but not the effect associated with
'softening' of the bar itself. However, experiments have shown that the local
bond stress-slip relationship is considerably influenced by yielding of the bar
(2)(3). It has been concluded that yield penetration and plastic deformations are
strongly underestimated when calculations are based on the bond model provided
by the CEB-FIP Model Code 1990 (MC90). Modified bond stress-slip
relationships with a reduced bond strength for bar yielding have been proposed
(3)(4), but the mechanism of the steel-concrete interaction in the post-yield
regime has yet to be explained.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D 0662949 0.542801 T 4 4
Furthermore, since most of the tests have been carried out using only one
bar size or varying it in a very limited range, the corresponding empirical bond
models do not adequately account for the bar diameter. Yet, internal coneshaped
cracking, which contributes significantly to the local slip, is a strongly localized
phenomenon that is directly related to the bar diameter and concrete brittleness.
Therefore, a bar size effect is plausible.
Considering that most empirical bond models are based on experimental

data from tests carried out on bars embedded in low to medium strength concrete,
the implications of the use of HSC are of great concern. Despite the large
differences in fracture characteristics between NSC and HSC, bond properties are
often extrapolated from NSC to HSC. This results in unreliable bond properties
for HSC, which may cause design errors or at least lead to an essential loss of
accuracy in analyses.
With these concerns in mind, a systematic experimental study was carried

out to develop a basis for the formulating of a bond model for ribbed bars in
concrete. The new bond model should take into account the strength and fracture
characteristics of concrete, the state of deformation of the bar (bar contraction is
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
significant after yielding), the degree of confinement and the corresponding mode
of bond failure. In the following section, an outline of the experimental
investigation on bond of ribbed bars in concrete is given and some characteristic
results are presented. Next, the steps taken in formulating the new bond model
are discussed. Finally, the verification of the model is reported and some diverse
experiments are simulated to show the range of applicability of the model.
EXPERIMENTAL INVESTIGATION
Scow of Test Series
In practical cases, two different bond failure mechanisms can take place,
i.e. pull-out and splitting. Two test series were designed in such a way that both
types of bond failure could be studied. The first series was carried out on bars
with long embedment lengths cast into massive cylinders. In this series the bond
behavior for good bond conditions in confined concrete, defined according to
MCW, was examined. The bars were loaded up to yielding and failure was
caused by pull-out of the bar, i.e. shearing of the concrete between the ribs. The
second series of tests was carried out on beam-type elements with usual concrete
cover. The test Set-up complied with that of good bond conditions in unconfined
concrete according to MCW. In this case, splitting bond failure was prevailing.
In total eight pull-out tests and four (double) bond bending tests were performed.
The scope of the test series was limited to the three prime variables, i.e.
confining condition (confined and unconfined concrete), concrete type (NSC and
Obb2949 0542802 980 9
HSC) and bar diameter. The Young’s modulus E,, cube compressive strength f,,
and tensile splitting strength f,, were directly measured using specimens that
were cured together with the bond test specimens. The uniaxial compressive
strength f, and the uniaxial tensile strength f,, were computed with the
expressions given in MC90.The average values obtained for both test series are
f,=28.5 and 96.9 MPa andf,,=2.3 and 4.1 MPa for NSC and HSC, respectively.
Hot rolled reinforcing steel FeB 500 HWL was used and two bar diameters ds
were chosen, namely 16 and 20 mm. The mechanical characteristics of both the
original and the prepared bars (see below) were determined in tensile tests
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(measuring length of lOd,). The average values obtained for the prepared bars
were used in the analysis of the test data, i.e. yield stressfy = 533 MPa, tensile
strength& = 618 MPa and ultimate strain e,, = 11.3%.
Test Assembly
Bond of deformed bars in well confined concrete was studied using a

500 mm diameter massive cylinder with a concentrically placed bar. The length
of the specimen equaled 60ds, with an embedded length of 50d, and a bond-free
length at the loaded end of lOd,. Fig. 1 shows the details of the specimen
geometry. A similar Set-up was used by Shima et al. (2)(4). The concrete was
cast in vertical direction parallel to the bar (the pull-out force is acting in the
opposite direction). A steel cone, which was removed after demoulding, provided
the bond free length at the loaded end. By connecting the loading device to the
bar with the help of a taper-threaded coupler, yielding could be ensured to take
place at the beginning of the embedded bar.
To monitor the steel strain development, machined and instrumented bars

were used. The longitudinal ribs were removed and grooves were milled on both
sides all along the bar; see Fig. 2. In these grooves strain gauges and electrical
wires were placed. To assure a proper insulation of the strain gauges, the grooves
were filled with an epoxy resin afterwards. The location of the strain gauges is
shown in Fig. 1. Post-yield strains could be measured up to 5%.
A series of bending bond tests on beam-halves was used to examine

concrete cover splitting and bond behavior of bars in unconfined concrete for
large steel deformations (Fig. 3). The specimens had a rectangular cross-section
of 100x212 and 160x220 mm for the 16 and 20 mm diameter bars, respectively,
resulting in a constant reinforcement ratio. The bottom cover on the bars equaled
2.5ds. The beam-halves were reinforced with a single reinforcing bar and no
transverse reinforcement was applied. The reinforcing bars were prepared in the
same way as the bars used in the pull-out tests and provided with 2x14 strain
gauges equally spaced at 40 mm. At mid length the bars were connected by
means of a threaded coupling. A knife-edge hinge was placed in the compression
zone of the beam to create a well-known internal lever arm, with which the force
0662949 0542803 817
in the bars could be estimated taking into account the applied force and the 2000
mm span of the coupled beam-halves. The casting position was perpendicular to
the bar, with the bar located at the bottom side of the specimen.
The local steel strains, the applied load, the relative displacement at both
ends of the bar (in puli-out tests), the deflection, the crack pattern development
and the overall deformation of the tensile cord (in bending bond tests) were
recorded. On the basis of the local steel strain measurements, the bond behavior
was reconstructed for both types of bond failure. The test results were used in the
validation of the current MC90 design bond model. Moreover, the experimental
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
data could be used for the calibration and validation of a new bond model that
would incorporate concrete confinement. Some test results are presented in the
following section. Additional details are provided in references (5) and (6).
Discussion of Test Results
From the local steel strain measurements of each pull-out test a unique
steel strain distribution curve could be attained by parallel shifting the sets of
data points (steel strain versus location) found in subsequent load steps.
According to Shima et al. (4)this is the case as long as the boundary condition
at the free end of the bar consists of zero strain and zero slip. The slip and bond
stress distributions along the bar were derived from the unique steel strain
distribution curve. Neglecting the concrete deformation parallel to the bar, the
local slip at each location along the bar could be computed by integration of the
steel strains from the free end of the bar to the point considered. The slip values
obtained in this way were verified by comparison with the measured loaded end
displacements after correction for the slip in the threadeä coupling. The latter was
estimated in separate tests on couplings. The bond stress distribution followed
from the first derivate of the strain distribution. In these analyses the
characteristics of the prepared bars were used.
From the slip and bond stress distribution curves a bond stress-slip
relation was derived for each pull-out test. This relation is independent of the
location along the bar, Fig. 4.For small slip values in the elastic range of steel
strains, bond stress increases with slip. The sudden transition from the ascending
to the descending branch of the bond stress-slip curve corresponds to yielding of
the bar. After this point a substantial reduction of the bond stress with increasing
slip is found. Although similar tendencies are observed in all tests, some major
differences appear between NSC and HSC. In the initial stage the bond stiffness
is higher for HSC than for NSC. Also, peak bond stress values are much higher
for HSC. However, when the bond stress is normalized the values for
HSC are about 25% lower than for NSC. The degree of the bond stress reduction
after yielding differs, and a more sudden drop is noticed in case of HSC. The
effect of the bar diameter, which for the investigated range of diameters is not
Dronounced. could be eliminated bv normalizing the sliu 6 to the bar diameter.
Dbb2949 0542804 753
thus by using the non-dimensional slip S=6/d,; see also (4).
In Fig. 4 the experimental bond stress-slip relations for confined concrete

are compared with the MC90 bond model. The latter distinguishes between good
and all other bond conditions, related to the position of the bar during casting,
and between unconfined and confined concrete, for which the bond failure is
concrete splitting and shearing of the concrete keys between the ribs,
respectively. In the first branch the bond stress is an exponential function of the
slip, the second branch consists of the maximum bond stress that is proportional
to the square root of the compressive strength, and in the third branch the bond
stress linearly descends to a value that remains constant in the fourth branch. The
slip values that mark the transitions between subsequent branches, depend on the
bond failure type and the clear rib spacing. The shaded area in Fig. 4 reflects the
influence of the variation in concrete strength on the bond stress.
It can be seen that for small slip values the experimental bond stiffness
is considerably higher than the model prediction, especially for HSC. This can
partly be explained by differences in confinement and bond conditions between
the present tests and the experiments on which the MC90 bond model has been
based. MC90 does not in any way reflect the change in bond characteristics due
to steel yielding. For both concrete types the descending branch and the residual
bond stress plateau of MC90 fall far above the experimental values. On the other
hand, a much better agreement is found between the test results and a relation
proposed by Engström (3) for NSC; see Fig. 4 (left). This relation considers the
influence of yielding on bond resistance. It shows a lower peak value, a steeper
slope of the falling branch and a lower residual bond strength than MC90.
Data obtained from the bending bond tests have been used to validate a
new bond model (see next section) for the case of splitting bond failure in
unconfined concrete. In (7) this model is used for the simulation of tension
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stiffening, taking into account realistic values for concrete cover and bar spacing.
Fundamental disparity between the MC90 bond model and the present test
results concerning the effect of large steel strains, and inaccuracy in formulation
of bond in HSC called for a thorough analysis of the bond failure mechanisms
and for the formulation of a more general model that would consider the actual
confinement degree, the material brittleness and the stress in the bar.
MODELING OF BOND BEHAVIOR
General considerations
To describe bond behavior in an analytical way two important aspects
0662949 0542605 69T
have to be known, Le. the magnitude of the force transmitted by the ribbed bar
to the surrounding concrete and the capacity of concrete to resist this force. In
analyzing the fist aspect, attention is paid to the transfer mechanism and its
connection with the bond failure mode and the strains in the bar. The second
aspect is evaluated by estimating the concrete confining capacity.
After breaking of the initial bond, which consists of adhesion and

interlocking of the cementitious matrix with the steel surface, force transfer is
mainly governed by bearing of the ribs against the concrete. The concentrated
bearing forces in front of the ribs cause the formation of cone-shaped cracks
starting at the crest of the ribs. The resulting concrete keys between the ribs
transfer the bearing forces into the surrounding concrete. In this stage the
displacement of the bar with respect to the concrete (slip) consists of bending of
the keys and crushing of the concrete in front of the ribs (8); see Fig. 5 left. The
bearing forces, which are inclined with respect to the bar axis, can be
decomposed into the directions parallel and perpendicular to the bar axis. The
sum of the parallel components equals the bond force, whereas the radial
components induce circumferential tensile stresses in the surrounding concrete,
which may result in radial cracks.
Now two failure modes can be considered. If the radial cracks propagate
through the entire cover bond splitting failure is decisive. In that case the
maximum bond stress follows from the maximum concrete confinement, which
is reached when the radial cracks have penetrated the cover for about 70% (12).
Further crack propagation results in a decrease of the confining stresses. At
reaching the outer surface these stresses are strongly reduced, which results in a
sudden drop of the bond stress. Yet, the load bearing mechanism remains the
same as in the previous stages. When the confinement is sufficient to prevent
splitting of the concrete cover, bond failure is caused by pull-out of the bar. In
that case the concrete keys are sheared off and a sliding plane around the bar is
created. Thus, the force transfer mechanism changes from rib bearing to friction
(Fig. 5 right). The shear resistance of the keys can be considered as a criterion
for this transition. It is attended by a considerable reduction of the bond stress.
Under continued loading the sliding surface is smoothed due to wear and
compaction, which will result in a further decrease of the bond stress.
In general, the stress in the bar is considered to have a negligible

influence on the bond resistance of ribbed bars. As long as rib bearing is the
force transferring mechanism (splitting bond failure) this can be justified,
considering that the transverse deformation coupled with the local steel stress
change (Poisson effect) is small in comparison with the rib height. However,
when the force transfer mechanism changes from rib bearing to friction (pull-out
failure) the local transverse deformation of the bar cannot be disregarded, since
the roughness of the sliding plane is of much smaller size than the rib height.
This transverse deformation may considerably influence the development of the
radial compressive stress and, hence, the frictional bond stress. The contraction
of the bar coupled with an increasing steel stress will result in a reduction of the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D Obb2949 0542öûb 52b
202 Eigaj, den Uiji and Walraven
bond stress, which may become pronounced when the bar starts to yield. These
aspects must be taken into account in formulating the 'action' part of the bond
phenomenon.
Similar attention is given to the 'reaction' part of bond. The importance

of the concrete confining capacity for the bond resistance of a ribbed bar follows
directly from the analysis of the force transfer between the bar and the
surrounding concrete. The radial components of the concentrated forces, which
radiate from the bar into the concrete, are equilibrated by circumferential tensile
forces in the concrete and by any additional confinement, such as transverse
reinforcement and external forces. It has been recognized in earlier studies that
the concrete confining capacity not only affects the magnitude of the ultimate
bond stress but also the mode of bond failure (9)( lo)( 11). This is reflected in
MC90 as well, where the bond stress-slip relation is defined as a function of the
bond failure mode and the bond conditions. However, it is not given as a
continuous function of the degree of confinement, but only for the upper and
lower limits. Such a restriction has consequences for the accuracy of the
calculated bond strength in unconfined concrete, considering the significant
influence of the cover thickness on the occurrence of splitting cracks and, hence,
on the resulting bond capacity. Therefore, it was thought necessary to include a
more refined estimate of the confining capacity in the new bond model.
Confinement Model
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
To describe the resistance of the concrete cover against splitting due to
bond a thick-walled-cylinder model is used. This approach, previously introduced
by Tepfers (11) and extended by Van der Veen (12) with the inclusion of the
softening behavior of concrete in tension, is further developed here. Whereas
former studies aimed at estimating the maximum confinement to establish a
criterion for bond splitting failure, now the radial compressive stress 6, at the
bar-to-concrete interface is defined as a function of the radial displacement of
this interface. By dividing the radial displacement by the bar radius a
dimensionless quantity is obtained, that is denoted as the radial strain at the
interface
In the thick-walled-cylinder model the radial components of the bond

action are conceived as an internal radial pressure. The response of the cylinder
to this pressure is subdivided into three stages, i.e. uncracked (I), partly cracked
(II) and entirely cracked (III). During stage I the tensile stresses across the
cylinder wall remain below the tensile strength and a linear elastic behavior of
the material is assumed, for which analytical expressions are available (1 3). The
transition into stage II is marked by the initiation of radial (micro)cracks that
start at the interface when the circumferential stress reaches the tensile strength.
The cylinder is partly cracked and the response of the concrete to the internal
Obb2949 0542807 4b2
pressure becomes nonlinear. In the numerical analysis the penetration depth of

the radial cracks is the control parameter in this stage. The maximum radial stress
is reached after penetration of about 70% of the cylinder wall. Further crack
penetration results in a decrease of the radial stress, which is accelerated as soon
as the entire cylinder wall is cracked. Then stage LIi begins, which is controlled
by the radial crack opening at the outer perimeter of the cylinder.
For the determination of the radial stress-strain relation at the interface for
the (partly) cracked cylinder, the softening behavior of concrete loaded in tension
is taken into account by applying the fictitious crack model (14), which describes
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the tensile stress transferred by the crack as a function of the crack width. A
bilinear softening relation has been chosen (15). The characteristics of the thick-
wailed-cylinder model and an example of the resulting radial stress-strain relation
are given in Fig. 6. In the partly cracked stage (stage Ii) the radial displacement
follows from the sum of the radial displacement of the fibre at the crack front
and the change in wall thickness of the cracked part. In the entirely cracked stage
(stage III) the former contribution is replaced by the radial displacement of the
outer surface, which follows from the crack width and the elastic strain of the
concrete between the cracks.
The confinement model has been verified against test results (16). In these
tests a conical steel bar was pulled through a concrete disc and the force versus
bar displacement was measured. Under the assumption of a uniform bond stress
distribution and a constant coefficient of friction this relation is proportional to
the radiai stress-strain relation found with the confinement model. Test
simulations have shown a considerable influence of the number of radial cracks
assumed in the model on the descending branch of confining capacity curve.
Good agreement was found by fixing the number of cracks to three, which is the
value used to simulate concrete confinement in the present bond model.
Bond Model
In the formulation of the bond model the following steps are taken
(Fig. 7). A boundary layer is assumed in which locally imposed forces and
displacements induced by rib bearing are spread (for practical reasons the
thickness of the boundary layer is neglected). Owing to this simplification the
stress-strain state at the interface is considered continuous. The wedging effect
that occurs when the bar is displaced with respect to the concrete, is modeled
conceiving the interface between the boundary layer and the surrounding concrete
as conical. By assuming dry friction at the interface (0 denotes the friction angle)
the bond stress %b is directly proportional to the radiai compressive stress or:
T~ = O p t ( e ) (1)
The radial compressive stress is the response of the surrounding concrete
W û b b 2 9 4 9 0542808 3 T î m
to the radial displacement of the interface, which is evaluated with the above-
mentioned thick-walled-cylinder model.
The radial displacement of the interface is a function of the slip and the
cone angle q, which is attributed to the wedging effect. In reality the wedging
action becomes active after the transition from adhesion to rib bearing, however,
in this model the initial stage of adhesion and interlocking is neglected. The cone
angle is influenced by the failure mechanism. With bond splitting, rib bearing
continues to be the force transferring mechanism and in that case the cone angle
remains the same. On the contrary, the development of the cylindrical sliding
plane around the bar in the case of puli-out failure is coupled with a reduction
of the cone angle. Moreover, as pull-out failure develops, progressive smoothing
of the sliding plane will cause a further reduction of the cone angle. It is assumed
that this process is a function of the slip, and that the rate with which it occurs
decreases with slip. This holds also for the radial displacement of the interface,
which reduces as the sliding plane becomes smoother.
These considerations result in different relationships between slip and

radial strain for splitting and pull-out failure. With increasing slip the entire
confinement capacity curve is followed in the case of splitting failure, whereas
in the case of puli-out failure only a part of the ascending branch of this curve
is followed, first upwards and then downwards (Fig. 8, the subscript s denotes
the longitudinal steel strain and, hence, the Poisson effect).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
In the mathematical formulation of the bond model a clear distinction is
made between both bond failure modes. For the splitting mode the most
important influence is the wedging effect coupled with the rib bearing
mechanism. As argued above, the influence of the Poisson effect is small in this
case and therefore it is neglected. Wear and compaction do not play a role with
this failure mode. Hence, for the splitting mode the radial displacement of the
interface is expressed by:
= 6 Wcp)
Cr,r,rs (2)
For the puli-out mode of bond failure it has been reasoned that after the
occurrence of the cylindrical sliding plane the cone angle diminishes as the slip
increases, that the Poisson effect cannot be disregarded and that wear and
compaction of the sliding plane shall be taken into account. Thus, in this case the
radial displacement of the interface is influenced by a number of factors that can
be connected with the slip and the local steel strain:
with 'ps slip dependent cone angle

a,, coefficient representing the effectiveness of the radiai strain release at
the interface due to bar contraction
E, longitudinal steel strain in the bar
Obb2949 0542809 235,-
v, Poisson constant of steel

F, slip dependent function representing the radial strain release due to the
wear and compaction of the sliding plane
In absence of the information needed to quantify precisely all these

influences, an expression is used that represents the combined influences of
wedging, Poisson effect and wear of the sliding plane, F1(8,&,), which is a
function of local slip and steel stress. The criterion for the transition from the
splitting failure mode into the pull-out failure mode is a critical bond stress T ~ ~ ,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
which is taken proportionally to the tensile strength of the concrete. To calibrate

the bond model, simulations of experiments have been performed for various
loading and confinement conditions, and for different material characteristics of
both steel and concrete (16).
POTENTIAL APPLICATIONS OF BOND MODEL
Depending on the available concrete Confinement a distinction is made

between the two types of bond failure, i.e. bond splitting, when the confinement
is not sufficient to prevent the radial crack from penetrating the entire cover, and
pull-out, when enough confinement is available to have a sliding plane originated
before the radial cracks have penetrated the entire cover. The ability of the model
to distinguish between these two failure modes is of great importance for the
analysis of reinforced concrete members, considering the different bond resistance
in both cases. Fig. 9 (left) shows the bond stress as a function of slip and steel
strain for pull-out failure. Before yielding the influence of the steel strain on the
bond stress appears to be small, but after yielding a strong influence is observed.
In the case of splitting failure (Fig. 9, right) the bond stress is hardly influenced
by the steel strain, as argued above. The bond characteristics, however, are
strongly related to the cover thickness. Since the model takes into account the
softening behavior of concrete in tension, the influence of concrete toughness can
be evaluated (examples in Fig. 9 for NSC and HSC withf, = 35 and 107 MPa,
respectively, and 20 mm bar diameter).
To arrive at the actual bond stress-slip relation for member analyses, the
following steps must be taken. For the use of the thick-walled-cylinder model,
the actual concrete! fracture properties must be defined. The choice of the wall
thickness must take into account the cross-sectional geometry. The concrete
confinement is evaluated by taking the wall thickness as a weighted average of
the cover around the bar. Since the concrete confinement is the result of the
radial displacement of the interface, it is essential to take into account the actual
stress-strain relationship of the steel. The effect of a so-called cone pull-out near
a free edge must be considered. This is achieved by relating the activated
concrete area (and thus the wall thickness of the thick-walled-cylinder) to the
distance from the bar entrance plane and to the degree of boundary restraint.
More details can be found in (18). Considering that the present confinement
0662749 0542830 T57
206 Bigaj, den Uijl and Walraven
model solely deals with the clamping action of the concrete surrounding the bar,
only members without confining reinforcement are analyzed in the following.
To demonstrate the model’s potential to simulate the bond resistance for

a wide range of steel deformations, the above-mentioned pull-out tests are
simulated. An equally good agreement is found in the case of NSC and HSC for
the steel stress and slip distributions along the transmission length and for the
bond stress-slip relation; see Fig. 10. It is important to realize that, whereas the
bond stress-steel strain-slip relations are unique for a given set of material
properties and a well-defined confinement condition, the resulting bond stress-slip
relation depends on the actual loading and boundary conditions.
Results of tensile tests on short NSC prisms are used to validate the
present bond model for the case of a short embedment length, combined with
large steel strains in the post-yield range. In this case failure is defined as rupture
of the steel (17). The slip and steel strain distributions along the bar at failure are
simulated well. This can be seen in Fig. 11, where measured and simulated
values are given for a crack spacing of 100 mm in a specimen with one
concentrical bar.
The potential of the model to take into account the actual stress-strain
property of the steel is shown by simulating pull-out tests with long embedment
lengths, in which this aspect has explicitly been studied (2). Three kinds of hot
rolled reinforcing steel bars were used, which differ in yield strength and
hardening behavior. The model predictions compare well with the test results, and
the influence of steel properties on the bond strength characteristics is evident,
in particular with respect to the slip at which the bond stress drops down and the
range of slip with the highest bond stress (Fig. 12). For large steel strains in the
post-yield range a strong reduction of bond strength is found. With further slip
the bond strength remains almost constant. The ability of the bond model to
reproduce these effects is a basic requirement for the analysis of plastic
deformations in reinforced concrete and for the study of the influences of
material properties on related phenomena.
Finally, attention is given to the influence of the bar diameter on bond

behavior. The proposed model can represent this effect, owing to the fracture
mechanics based description of the radial crack propagation. There is little
experimental data available on this topic. However, some research confirms that
the bar size effect on the bond behavior exists (18). A clear tendency is also
found with the concrete confinement based bond model (Fig. 13). A change in
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the rate of bond strength decrease with an increase of the bar diameter is
predicted for a varying concrete cover. When pull-out bond failure takes place
the ultimate bond stress does not depend any more on the bar diameter.
= Obb2949 0542811 993 U Tribute to Peter Gergely 207
CONCLUSIONS
A bond model for ribbed bars in concrete has been derived. In the model
formulation the concrete confining capacity is used as a starting point. This is
estimated using a thick-walled-cylinder model. Depending on the confinement
conditions it distinguishes between two types of bond failure, namely bar pull-out
and cover splitting, which are thoroughly described. An important extension with
respect to most currently used bond models is that bar contraction at a strain
increment is taken into account, thus including the bond behavior of yielding
steel. By the virtue of the implementation of fracture mechanics characteristics
of concrete, also the bond strength of HSC can be properly estimated. This bond
model has been verified against various tests, thus showing its potential
applications.
Considering the analysis of reinforced concrete members, the involvement

of the actual geometry of a member in determining the degree of confinement,
and thus the mode of bond failure, is an important improvement, compared to
other modeling approaches at this scale. The ability of the model to predict the
actual bond stress-slip relation for a wide range of steel deformations and to take
into account the actual stress-strain characteristics of reinforcing steel, is of major
importance, in particular for the study of the relation between the ductility of the
reinforcement and the deformation capacity of the members. Facing the necessity
of establishing design rules for newly developed consmction materials (HSC,
FRC, FRF'), the flexibility of the model in considering fracture properties of the
matrix and deformation characteristics of reinforcing units is an important feature.
Acknowledgment: The financial support provided by the Dutch Technology

Foundation (STW grant DCT 22.2774) is gratefully acknowledged.
REFERENCES
1. Bigaj A.J., Walraven J.C., (1996), Modelling of size efJect on rotation

capacio of plastic hinges, Onderzoekschool Bouw, pp 67-78.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
2. Shima H., Chou L.-L., Okamura H., (1987), Bond Characteristics in Post-
Yield Runge of Deformed Bars, Concrete Lib. of JSCE No. 10, pp 113-124.
3. Engström B., (1992), Ductiliry in tie connections in precast structures, PhD

Thesis, Chalmers University of Technology, Göteborg.
4. Shima H., Chou L.-L., Okamura H., (1987), Bond-Slip-Strain Relationship

of Deformed Bars in Massive Concrete; see (2), pp 79-94.
A Obb2949 0542812 82T
5. Bigaj A.J., (1995), Bond behavior of deformed bars in NSC and HSC, Stevin
Report 25.5-95-11. TU Delft.
6. Bigaj A.J., (to be published in 1997), Bond behavior of deformed bars in

NSC and HSC -part II, Stevin Report, TU Delft.
7. Bigaj A.J., Den Uiji J.A., (1997), Tension stiffening simulation with
confinement based bond model, Progress in Concrete Research, Vol. 5, TU
Delft.
8. Goto Y., (1970), Cracks formed in concrete around deformed tension bars,
Journal of the ACI, Proc. Vol.68, No. 4, pp 241-251.
9. Malvar J.L., (1992), Confinement stress dependent bond behavior, Proc. of

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the Int. Conf. on Bond in Concrete, Riga, pp 1.79-1.88.
10. Cairns J., Jones K., (1993, The splitting forces generated by bond, Magazine
of Concrete Research, Vo1.47, No.171, pp 153-165.
11. Tepfers R., (1979), Cracking of concrete cover along anchored deformed
reinforcing bars, Magazine of Concrete Research Vol. 31, No. 106, pp 3-12.
12. Van der Veen C., (1990), Theoretical and experimental determination of
crack width in RC at very low temperatures, Heron Vol. 35, No. 2.
13. Timoshenko S., (1976), Strength of materials, Part I I , Advanced Theory and
Problems, Van Nostrand Reinhold, pp 205-210.
14 Hillerborg A., (1983), Analysis of One Single Crack, in Fracture Mechanics

of Concrete, Elsevier, pp 223-249.
15. Roelfstra P.E., Wittmann F.H., (1986), Numerical Method to Link Strain
Softening with Failure of Concrete, Fracture Toughness and Fracture Energy
of Concrete, Elsevier, pp 163-175.
16. Den Uiji J.A., Bigaj A.J., (1996), A bond modelfor ribbed bars based on
concrete confinement, Heron, Vol. 41, No. 3, pp 221-226.
17. Manfredi G., Pecce M., (1996). Bond behavior between concrete and steel
in large post-yieldfield, Journal of Mechanics and Structures, pp 21.
18. Morita S., Fuji S . , Kondo G., (19941, Experimental study of size effect in
concrete structures, JCI Conference on Size Effect in Concrete Structures,
pp 27-46.
D Obb2949 0542813 7 b b 209
4.5 YFLA-2 strain gauges
Fig. P-Test bar geometry
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb2949 0542834 b T 2 =
strain gauges
Crack Pattern
Fig. 3-Test Set-up for bending bond tests
bond stress [MPa] _. bond stress [MPa]

30
20 20
10 10
O - . . . . _ O
o 4 8 S =s/d;100
Fig. 4-Test results and MC90 formulation of bond stress-slip relation for NSC,
including Engström's model (31, (left) and for HSC (right)
Fig. 5-Deformations around bar for splitting bond failure after Goto (8) (left) and
pull-out bond failure (right)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542835 539
Tribute to Peter Gergely 21 1
r
c
IL.,Bb,
1
(Y
r
u-stage I
%
c
stage Ill
Fig. 6-Partly cracked thick-walled cylinder
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 7-Modeling steps, ¡.e., interface model and wedging action
er
Fig. 8-Radial stress o,versus radial strain E, and slip 6 for splitting bond failure
(left) and pull-out bond failure (right)
~
W Ohh2949 054281b 475

30 30
bond stress bond stress
iMh1 iMb1
O O
steel stmin 60
[dm cover [mm
O.
O 1
30 , 30 I
bond stress bond stress

iMk1 [mal
O O
steel sirah 60
[m/m cover [mm
O.
O slip [miil 1 O slip [mm] I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 9-Bond stress-slip relation for NSC (top) and HSC (bottom) for splitting bond
failure (left) and pull-out bond failure (right)
bond stress [MPa] bond stress [MPa]

30 30
20
--- I simulation
20
--- simuiation
10 10
O O
o 1 2 3 slip[mm] o 1 2 3 slip [mm
Fig. 10-Bond stress-slip relation of pull-out tests with long embedment length
from experiments and test simulations for NSC (left) and HSC (right)
,. .- steel strain [wm]
V.15,
sliu
. *Imml -
9
6 1-- simulation
0.05 - test resuits

--- simulation
3
Fig. 11-Steel strain (left) and slip (right) distributions from experiments and
simulations of tensile tests with short embedment length
0bb2949 0542817 3 0 s
Fig. 12-Influence of reinforcing steel properties on bond strsss-slip relation; test

results after Shima et al. (2) at top, simulations at bottom
Cover : Bond failure mode :

3.0 ds pdl-out
2.5 ds splitting
splitting
1.5 d, splitting
splitting
I I I I
14 16 24
O
O 10 20 30 ds [ml
Fig. 13-Simulated influence of bar diameter on bond strength
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~ ~~
0662949 0542818 248
SP 180-10
Pull-Out Bond Behavior of Ribbed Bars

in Normal and High-Strength Concrete
with Various Confinements
by B. Engström, J. Magnusson and 2. Huang
-When ribbed bars are anchored in linear structural members, the bond-
slip behavior and the anchorage capacity is strongly influenced by splitting
cracks. Many factors influence the formation of the splitting cracks, among others
the anchorage length, the concrete cover, the bar spacing and arrangements,
confinement from stirrups, flexural and shear cracks in the vicinity of the
anchorage region, transverse pressure from support bearings, etc. These
parameters often interact in a complex manner, and common design methods for
anchorage regions are derived from empirical evaluations of test data and are
often strongly simplified. The present study was carried out with the aim of
studying the anchorage behavior of ribbed bars in structural members of high
strength concrete and to check the applicability of some common design methods
to these new materials. The influence of concrete type, normal or high-strength
concrete, and various detailing of the node regions was examined. The tensile
force in the active end of the anchorage zones was evaluated from steel strain
measurements and was compared with predictions by means of strut and tie
models. These models were found to consider the effect of inclined cracks in an
appropriate and consistent way. The observed anchorage capacity was compared
to some common design methods. It was found that the methods, to a considerable
degree, were unable to reflect the real behavior. Further improvement and
development of design and analytical tools is required.
Keywords: Anchorage; bond; confinement; cover; high-strength concrete; pull-out

tests; ribbed bar; slip; splitting
215
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0 b b 2 9 4 9 0542839 184
216 Engström et al.
Björn Engström: Ph.D., Professor, Chalmers University of Technology, Göteborg

Sweden
Jonas Magnusson: M.Sc., Research Assistant, Chalmers University of Technology,
Göteborg, Sweden
Zhongyi Huang: B.Sc., Engineer, Ministry of Coal Industry, Shenyang Design &
Research Institute, Shenyang, China. Research student at Chalmers University of
Technology 1994 - 1995.
1. ANCHORAGE OF RIBBED BARS IN HIGH STRENGTH CONCRETE

- A RESEARCH TOPIC OF CURRENT CONCERN
During the last years, the possible advantages with high strength concrete have
been recognized, and high strength concrete and its applications have been of
major concern for research activities. This is also true in Sweden where a large, 6-
years, National research program on high strength concrete and high strength con-
crete structures is now in its final stage. As part of this program, the bond and an-
chorage of ribbed bars in high strength concrete are studied at Chalmers University
of Technology. The need for fiirther research in this field has been pointed out,
among others in a recent CEB Bulletin on Proposed Extension of the Model Code
(1)-
Many parameters influence in a complex way the bond behavior, the bond strength,
and the anchorage capacity. Of this reason the present project on anchorage has
been divided in several separate studies on various aspects of bond, using different
types of test specimens. The local bond stress-slip relationship was studied in pull-
out tests with a short embedment length. The results from such studies can be used
to predict analyticallythe global response of an anchored bar, and the distributions
along the anchorage length of the steel stress, the bond stress and the local slip.
The global behavior of an anchored reinforcement bar is revealed by the relation-
ship between the anchored tensile force and the end-slip of the bar, and was exam-
ined in pull-out tests with larger embedment lengths. To study the anchorage of a
group of bars in various types of nodes in reinforced beams, full-scale tests on sim-
ply supported beams with short shear span were carried out. In this case the force
to be anchored and the behavior of the anchorage regions depend on the successive
development of flexural and shear cracks, which makes the tests hard to evaluate
and mutually compare. Of this reason, simplified beam-like specimens, as well as
ordinary reinforced beams were used.
It is generally assumed that the bond behavior of ribbed bars anchored in concrete
is considerably influenced by the confinement along the transfer length. Confine-
ment can be provided by the concrete cover, or by a transverse pressure across the
anchorage regions. Transverse reinforcement can provide a passive confinement,
which means that a certain development of the bond mechanisms is required to
strain the transverse steel and thereby activate the confining action. When the con-
finement is sufficient, longitudinal splitting cracks are prevented, at least from an
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb294î 0542820 î T b D
uncontrolled development and it is possible to obtain a bond failure of puil-out

type, where the concrete is sheared off along the tips of the ribs. In this case the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
bond resistance will not be essentially affected by a further increase of the conñne-
ment. However, in normal applicationsthe conñnement is rather small, because of
the small concrete covers normally used. In that case the bond resistance is
strongly affected by a further increase of the confinement.
This paper presents a separate study on how the anchorage behavior of ribbed bars
in high strength concrete is innuenced by the concrete cover and transverse rein-
forcement. The aim of the study was to compare the effect of confinement in high
strength concrete with that in normal strength concrete. One aim was also to study
how the end-slip response and the steel stress distribution were affected by a con-
siderable reduction of the confinement and to compare the experimental results
with numerical predictions based on measured or assumed local bond stress-slip
relations.
2. EXPERIMENTAL STUDIES
2.1 Test Specimens
The influence of the confinement was studied in pull-out tests with a rather long
anchorage length. The main parameters that varied were the confinement and the
concrete type, high strength or normal strength concrete. One series concerned an-
chorage in well confined concrete and two series were used to study how the an-
chorage behavior was affected by a reduction of the confinement. The anchored
bar was then placed either in the mid position of an edge region, or in a comer
region. The type and dimension of the reinforcement bar were kept constant
through all the test series, a 16 mm ribbed hot-rolled bar of type K500.
The test specimens had a simple cube-like geometry. Two main types of specimens
were used. For each test in a high strength concrete specimen, a similar test in
normal strength concrete was made as a reference. The high strength concrete had
a compressive strength of about 105 MPa (cylinder strength according to ISO-
Standard), while the compressive strength of the ordinary concrete was about 30
MPa.
The first type of specimen, type I, was used to study the anchorage behavior in
well-confined anchorage regions. The specimen was cubical, with dimensions 400x
400x400 mm and with the anchored bar placed along the centroidal axis, see Fig-
ure 1. In this case the concrete cover was 192 mm or 12 times the bar diameter.
No stirrups were provided. Thé anchorage length varied between 90 and 250 mm
in specimens of high strength concrete, and between 150 and 500 mm in normal
strength concrete. In the test with the longest anchorage length, the depth of the
specimen was increased to 500 nun.
The second type of specimen was used to study the anchorage of bars with a smali
concrete cover, either in a mid position of a fiee edge region (type Ii), or in a cor-
ner region (type III), see Figure 2.Here the anchorage length was constant and
equal to one of the anchorage lengths from the first series, but the concrete cover,
and the number of stirrups enclosing the anchored bar varied. The cross-section of
the test specimen was the same as before, 400x400 mm. However, this type of
m 0662949 0542823 832 m
2 18 Engström et al.
specimen was provided with a nose-like extended part which was used to support
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the specimen in the testing machine. Thereby the eccentricity of the applied tensile
force could be balanced and the force transferred from the bar to the concrete
could be linked off in a natural way in the specimen. The state of stresses was then
similar to that in an anchorage region of a reinforced beam. Without this extended
part, the comer or edge with the anchored bar would be pulled off during the test.
In most of the specimens of type II and III, 2 or 4 stirrups $6 K500 (ribbed bars)
were distributed along the embedment length, enclosing the main bar. The actual
location of stirrups appears from Figure 3. However, in some of the tests the main
bar was anchored in plain concrete. Two different concrete covers was used in the
plain concrete specimens of type II.
The embedment length was chosen to be just sufficient to anchor the yield capacity
of the anchored bar in case of well coniined conditions, but insufficient to anchor
the tensile capacity at steel rupture. Hence, one embedment length, 170 mm, was
used for specimens of high strength concrete and another embedment length, 290
mm, was used for the tests with normal strength concrete.
For all types of specimens the embedded part of the reinforcement bar was placed
immediately inside the front face of the specimen without any debonded part. The
test program is presented in Table 1.
2.2 Materials
The compositions of the concrete mixes appear fi-om Table 2. From each batch of
concrete, a set of specimens for standard material tests was cast as well. Among
others, the compressive strength was evaluated from a series of three tests on 41 50
x300 mm cylinders according to the ISO-Standard at the same day as the main
test. The results are presented in Table 1. The fracture energy was evaluated from
beam tests according to the REEM Standard and was in average 165 N/m for high
strength concrete and 110 N/m for normal strength concrete. Full information
about the material tests and results is presented in a technical report (2).
The reinforcement steel was of the Swedish type KSOO,with a characteristic yield
stress of 500 MPa and a specified relative rib area of 0.056. The dimension of the
main bar was 16 mm and the stirrups had a dimension of 6 mm. The stress-strain
relationships were determined by tensile tests in the laboratory. The actual yield
stress of the 16 mm bars was evaluated to 569 MPa and the tensile strength to
648 MPa. For the 6 mm bars the same properties were evaluated to 570 MPa and
642 MPa respectively.
2.3 Test Procedure
The tests were camed out at an age of about 35 - 40 days after casting. For tests
specimens of type I, the load was applied by a hydraulic jack that was supported
via a load distribution beam on two bearings directly on the end face of the speci-
men. The applied tensile force was measured by a load gauge. The specimens of
types II and III were tested in an universal hydraulic material testing machine of
0bb2q4q D542822 77q
Tribute toPeter Gergely 219
eHUS
T
o
1OO. In all the tests the loading was controlled by the end displacement
the bar at the active end, with a rate of about 0.10 d m i n . When the maximum
load was passed, the deformation rate was increased in several steps.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Instruments were provided to measure the displacement of the bar, at the active
end as well as on the passive en4 relative to the stable faces of the specimens. The
steel strain was measured at several points along the embedded pari of the main
bars. The strain gauges had a gauge length of 1 mm and were mounted on the sur-
face between the ribs of the reinforcement bar, without any damage to the ribs.
3. TEST RESULTS AND EVALUATION

3.1 Anchorage in Well Confined Concrete
In the test specimens of type I, with the reinforcement bar well confined in the
center, three type of failures were obtained. For the tests with small anchorage
lengths, smaller than about 150 mm in high strength concrete and about 250 mm in
normal strength concrete, a pull-out failure occurred when the steel bar was still in
the elastic range. When the anchorage length was more than about 200 mm in high
strength concrete and about 300 mm in normal strength concrete the anchorage
capacity exceeded the capacity of the reinforcement bar and the steel finally rup-
tured.For the intermediate anchorage lengths, 170 mm in high strength concrete
and 290 mm in normal strength concrete, the anchorage could resist the yield
capacity of the bar, but during strain hardening of the steel a pull-out failure
occurred &er considerable plastic deformations and a simultaneous yield pene-
tration at the end of the embedded bar.
In case of pull-out failure, either before or after yielding of the steel, the bar
seemed to be sliding along a circular crack plane along the tips of the ribs. Con-
crete was still lefi in between the ribs, when the bar was pulled out from the speci-
men. During the tests no visible splittin cracks were observed on the exterior
f
faces of the test specimen. However, a ocal failure cone was formed in the con-
crete around the active end of the bar. This concrete cone remained bonded to the
bar, also in the final state, and moved with the bar, see Figure 4.
The depth of the failure cone was measured after the test. Of a total of seven pull-
out tests in normal strength concrete the depth of this failure cone varied between
18 and 35 mm with an average value of 26.4 mm, which corresponds to 1.65 times
the bar dimension. For seven tests in high strength concrete specimens the depth of
the failure cone varied between 16 and 30 mm with an average value of 27.0 mm
or 1.68 times the bar diameter. Hence, the failure cone seemed to develop in the
same way in high strength and n o d strength concrete.
It was found that the anchorage capacity in the pull-out type of failures increased
almost linearly with an increased anchorage length. For the same embedded length
the anchorage capacity was about twice as large in high strength concrete as in
normal strength concrete.
The tensile force development in the anchored bars was evaluated from the steel
strain measurements. Typical distributions of the tensile force along the bar at
maximum load are presented in Figure 5. It appears that the tensile force varied
almost linearly along the elastic part of the bar, except for the tests with the longest
anchorage lengths. Here the steel to concrete interface at the passive end of the bar
was not fully strained at maximum load. It is also possible to recognize a certain
boundary effect near the active end of the bar, indicating that the bond was not
fully effective near the free end face of the specimen. In the region where the yield
stress was exceeded, the tensile force varied with a much smaller rate than in the
region which was still in the elastic range. Hence, it is obvious that the yielding of
the steel results in a drastic decrease of the bond stress, see also (3).
Due to the local conic failure the effective transfer length decreased during the test.
For the tests with pull-out failures the average bond stress at the maximum load
along the final reduced transfer length is presented in Table 3 and related to the
concrete compressive strength.
As appears from Table 3, the average bond strength for the actual embedment
lengths seems to be proportional to the square root of the concrete compressive
strength. This is especially true for the anchorage lengths that was used in series II
and III, 170 mm for high strength concrete and 290 mm for normal strength con-
crete. This assumption will be used in the following to normalize the test results
when tests with various concrete strength's are compared.
3.2 Anchorage at Mid Position with Small Concrete Cover
For the tests in specimens of type II, with one bar placed in the mid-region of one
free edge, a longitudinal splitting crack always appeared along the bar through the
rather small concrete cover. The splitting crack developed successively, and in a
controlled way, when the load increased. Also transverse cracks appeared one after
the other during the test. The transverse cracks were inclined and developed in
direction towards the loaded end. A typical crack pattern is presented in Figure 6.
The cracking behavior was approximately the same in high strength concrete as in
normal strength concrete. In the specimens with stirrups, it was observed that the
location and the development of the transverse crack were influenced by the stir-
rups.
In spite of the rather small concrete cover and the longitudinal splitting crack that
developed along the bar, the anchorage capacity in the test specimens of type II
was almost the same as in the corresponding tests in well-confined concrete
(type I). However, at the maximum load an abrupt anchorage failure occurred
resulting in a sudden drop of the tensile force. Of this reason the post-peak behav-
ior was quite different compared to the corresponding specimens with the bar in
the center. The test with a small concrete cover had a more brittle behavior and a
very small residual capacity in the post-peak stage, while the specimens with the
bar in the center had a rather ductile behavior during the pull-out fictional mode,
with a residual capacity that decreased successively. The different behaviors appear
from the relationships between tensile force and end-slip presented in Figure 7a - b.
The normalized anchorage capacities of the tests of type II are summarized in
Table 4. The presented results are compared to the tests in well-contined concrete
with the same anchorage length.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Ohb2îb7 0542824 543
The tests showed that the reduced confinement had no essential uifluence on the
anchorage capacity in the specimens of normal strength concrete. When the con-
crete cover on one side of the bar was oniy 16 mm (one bar dimension) without
stirrups or with two stirrups 96 mm along the anchorage length, the anchorage
capacity was r e d u d by about 1û?hcompared to the average capacity in well con-
fined concrete. When four Stirnips 96 mm were provided in a 16 mm concrete
cover, or the concrete cover was twice as large but without stinups, the anchorage
capacity was about the same as in well conñned concrete.
The high strength concrete specimens seemed to be more sensitive to a reduction
of the conñnement. Here a reduction of the concrete cover to 16 mm resulted in a
-
reduction of the maximum load of about 25 30%. When two stirrups were pro-
vided, there was no increased capacity in one test, but a considerable increase in
the twin test. However, four stinups & mm along the transfer length, or an
increased concrete cover to 32 nun without stirrups, resulted in about the same
anchorage capacity as in the reference test.
Hence, compared to the case with smallest confinement, the effect of either an
increased concrete cover or an increased amount of transverse reinforcement
seemed to be similar in high strength as in normal strength concrete.
3.3 Anchorage in Corner Reaions with Small Concrete Cover
In the test specimens of type III, with two main bars placed in the corner regions,
the crack development was similar to that of the type ïI specimens. Longitudinal
splitting cracks develo ed successively along the anchored bars and transverse
cracks appeared one s er the other. The transverse cracks were inclined towards
the active end of the bar, but in this case on both sides of the loaded corner region.
In a late state the transverse cracks fiom the two adjacent corner regions joined
each other to a combined systems of cracks, see Figure 8. The crack pattern was
similar in specimens of high strength and normal strength concrete.
The anchorage capacity was substantially reduced when the bar was anchored in a
comer region compared with anchorage in well-conñned concrete (type I), as well
as in a mid position with small concrete cover (type II). At maximum load there
was an abrupt anchorage failure, normally associated with a sudden drop of the
tensile force, and the corner regions fell apart successively when the bars were
forced to slip further. However, in specimens with stirrups the residual capacity in
the post-peak stage was much higher than in the comparable tests of type II. This
was especially true in case of the higher amount of transverse reinforcement. The
Merent behaviors in the post-peak stage appear fiom the relationships between
tensile force and end-slip presented in Figure 9a-b.
Hence, the transverse reinforcement seemed to have a more positive influence on
the anchorage in a corner region than in a mid position. One explanation could be
that the confinement fiom the stirrup is greater in the corner of the stirrup where
two legs meet.
The normalized anchorage capacities of the tests of type II are summarized in
Table 5. The presented results are compared to the tests in well-conñned concrete
with the same anchorage length. For each test it was possible to determine the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
E Ob62749 0542825 488 U
222 Engström et ai.
anchorage capacity individually of two bars, one in each comer. These are indi-
cated by the notations (a) and (b) in Table 5.
The results showed that the anchorage capacity was considerably reduced when
the bar was anchored in a comer region compared to anchorage in confined con-
crete. For mecimens without transverse reinforcement. the reduction was about
55% in norhal strength concrete and even larger in high strength concrete, about
60 - 75%.
For each specimen it was possible to determine the anchorage capacity of each
comer bar individually. It is noticeable that the difference between the two anchor-
age capacities in the plain concrete specimens was small in normal strength con-
crete, but fairly large in high strength concrete. This indicates that the possibility
for force redistribution is smaller in high strength concrete. This has also been
observed by Ghassemi (4). When 2 or 4 stirrups were provided, the two anchorage
capacities in one specimen were always close to each other in normal strength con-
crete as well as in high strength concrete.
The stirrups had a positive influence on the anchorage capacity, which increased
with an increased amount of transverse reinforcement. However, especially for
normal strength concrete specimens the effect was moderate. With 4 transverse
bars $6 mm, the anchorage capacity in normal strength concrete was still about 35
- 40 % smaller than in the reference test in well confined concrete. In high strength
concrete, the same amount of transverse reinforcement resulted in an anchorage
capacity that was not more than 10 - 15 % smaller than in the reference test. In the
high strength concrete specimens the anchorage length was smaller and hence, the
4 stirrups were more closely spaced. This indicate that not only the number of
transverse bars is of importance, but also the spacing.
4. EVALUATION AND ANALYTICAL PREDICTIONS
4.1 Predictions Based on Local Bond Stress-Slip Relationships

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
An attempt was made to predict the end-slip response of the pull-out tests analyti-
cally. The analysis was based on assumed schematic relationships between the local
bond stress and slip. A numerical approach was used to find the solution of the
equilibrium equations along the transfer length with due concern of the boundary
conditions, following the intention of the CEBLFIP Model Code 90 (MC90) (5).
However, as the bars were anchored in specimens of substantial width, the con-
crete strain was always ignored. Furthermore, a bond free length of 2 times the bar
diameter was always assumed to consider the local conic failure at the active end
of the bar, see Figure 4.
For an anchorage region distinction is made in MC90 between confined concrete

and unconfined concrete. Confined concrete results in a pull-out failure where the
concrete between the ribs is sheared off, while splitting of the concrete occurs
when the concrete is unconfined. According to MC90 the anchorage zone can be
considered as well confined when the concrete cover is not smaller than 5$,, the
clear spacing between the anchored bars is not smaller than lo$,, or closely spaced
transverse reinforcement A,, > nA, is enclosing the anchored bar, or a high trans-
Obb2949 û54282b 314
verse pressure is present. When the concrete cover has a minimum depth of c
and the anchored bars are enclosed by a minimum amount A,,,, = 0.25A, of trans-
+,,
verse reinfoment, the anchorage region is considered as unconfined.
Schematic bond stress-slip relationships are defined in MC90 for unconfined and
confined normal strength concrete, and in each class the bond conditions can be
considered as "good" or "other". In absence of more experimental results, it has
been recommended in (6) to adopt the same model also for high strength concrete.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
According to the definitions in MC90 the test specimens of type I (bar in the cen-
ter) provided confined concrete. The specimens of type II and III without stirrups
did not fiiiñll the minimum requirements concerning transverse reinforcement and
then the schematic bond model for unconfined concrete is not applicable, at least
not for the residual state after the maximum bond stress is reached. However, it
could be anticipated that the model would give reasonable estimates for the
ascending branch. In specimens of type II and III with stirrups the actual conñne-
ment vaned according to the definitions from unconfined to confined concrete.
Specimens of type II with two stirrups had a ratio AJA, = 0.56, and with four stir-
nips a ratio AJA, = 1.13, In specimens of type III two main bars were anchored
and consequently the corresponding ratios were AJA, = 0.28 and AJA, = 0.56.
According to MC90 AJA, = 0.25 corresponds to unconñned concrete, and AJA. =
1 to confined. For intermediate values linear interpolation is recommended.
4.2 Behavior in Well Confined Concrete
For anchorage in well confined concrete, specimens of type I, it was found that the
model in MC90 gave results in reasonable good agreement with the observations
for specimens in normal strength concrete but the anchorage resistance in high
strength concrete was considerably underestimated. It was also found essential to
consider the drastic decrease of the bond resistance in the areas where the
anchored bar reached yielding.
The local relationship between bond stress and slip for the 16 mm ribbed bar was
studied in a separate series of pull-out tests with a short embedment length of 2.5
times the bar dimension. The diameter of the circular concrete specimens was 300
mm, which was large enough to give well confined conditions and prevent splitting
cracks, and the detaiïmg was made in order to prevent boundary effects. The
details of the tests and the results are presented in (7).
The test results were compared to the schematic bond stress-slip relationship given
in MC90. For normal strength concrete it was found that the model in the Code
was in good agreement with the observed behavior except for the residual stage.
Here the Model Code assumes a constant frictionai resistance, while in the tests the
resistance gradually decreased with increasing slip. For high strength concrete
specimens the bond resistance was substantially underestimated by the Model Code
and there was also a small overestimation of the slip at maximum load.
In the Code it is assumed that the maximum bond stress is related to the concrete
compressive strength by the expression T, = 2.5(f in case of good bond con-
ditions. However, from the test results it was foun that the maximum bond stress
under weü confined conditions was proportional to the concrete compressive
strength by the expression T,, = 0,45f,. This relationship was found to be valid
Obb2949 0542827 250
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
both for normal strength concrete and for high strength concrete. Based on the
experiences fi-om the tests, a modified model for the bond stress-slip relationship
was proposed, see Figure 10. The parameters that govern the various branches are
given in (7).
The proposed model for the local behavior also considers the response of the steel.
When the bar remains in the elastic range, the bond stress is expressed by the rela-
tionship (I) in the figure. However, if the bar begin to yield, an alternative branch
(II) should be assumed. It should be noted that the inserting point of branch (II)
depends on the actual global response of the anchorage region. Hence, the values
(T,,, ),s, of the inserting point should be determined for the actual section of the bar
by equilibrium and compatibility equations and can not be predetermined.
The model in Figure 10 for the local behavior was used to predict numerically the
global behavior of the tests with long embedment lengths in well confined concrete
(specimens type 1). The global behavior is then understood as the relationship
between the tensile force and the end-slip at the active end. Predictions were made
for two alternative bond conditions, either good bond conditions or other bond
conditions. For good bond conditions the maximum bond stress was assumed to be
T~~ = 0.45fcm, and for other bond conditions half this value was assumed.
It was generally a good agreement between the numerical predictions and the
observed behaviors, both with regard to the failure mode and with regard to the
end-slip response. The scatter in the experimental results was normally well in
between the estimates for "good" and "other" bond conditions but the response
was normally closer to the predicted one for the case "other bond conditions".
Typical results of the comparison between numerically predicted and observed
behaviors are presented in Figure 11.
Hence, the proposed local bond stress-slip relationship seemed to be valid for pre-
dictions of the global response in well-confined concrete.
4.3 Anchorage in Regions with Small Concrete Covers

It was found that the bond model in MC90 gave calculation results in reasonable
good agreement with the observation from specimens of type II and III, when
unconfined concrete was presumed. The model seemed to be applicable also to
high strength concrete, which means that the maximum bond strength was calcu-
lated as T~~ = 2.5(fJli2, in contradiction to the results found for anchorage in well
confined concrete.
Comparisons between predicted and observed end-slip relationships for specimens
of type II and III are presented in Figures 12-13. It appears that the observed
responses in general are well between the predictions for "good" and "other" bond
conditions for unconfined concrete. A large amount of transverse reinforcement
did not change the behavior to a typical one for confined concrete, as anticipated
by the Code. However, for specimens of type III with the anchored bars in comer
regions, the residual bond resistance increased with a large amount of transverse
reinforcement. The model does not consider the placement of the bar, in an interior
position or in a corner region, only the depth of the concrete cover. Also the effect
of the transverse reinforcement is treated in a too rough way by the model. From
the actual pull-out tests it was found that not only the number of transverse bars
m Obb2999 0542828 197 9
but also the spacing infiuenced the anchorage behavior, and more obvious, the
effect of the transverse reinforcement was stronger when the anchored bar was
placed in the comer of a stirrup than at an intermediite position along a leg.
Further development of the bond model for anchorage in concrete with small con-
ñnement is needed.
5. CONCLUSIONS
The bond strength in well confined concrete and under good bond conditions was
evaluated fiom puli-out tests with short embedment lengths, and was found to be
proportional to the concrete compressive strength by,,T = 0.45f,, for normal
strength concrete as weil as high strength concrete.
In pull-out tests with long embedment lengths in well-confined concrete the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
anchorage capacity increased with increasing anchorage length and increased con-
crete strength. When the concrete compressive strength increased by a factor 4,
from normal to high strength concrete, and the embedment length was the same,
the anchorage capacity was approximately doubled. For moderate embedment
lengths the tende force varied almost linearly along the anchored bar at maximum
load.
When the yield load was reached, the bond resistance decreased substantially near
the active end of the bar.
When the anchored bar was placed with a small concrete cover, in a mid-region or
in a comer region, the anchorage behavior was more brittle, normally with a drastic
drop of the resistance at maximum load. The anchorage capacity decreased with a
decreased concrete confinement, but more pronounced in high strength concrete.
In plain concrete with the same concrete cover, the anchorage capacity in a corner
region was about half of that in the mid region.
In the mid region, with a concrete cover c = 4, the anchorage capacity was only
slightly smaller than in weil confined concrete, in spite of the fact that a longitudi-
nal splitting crack had appeared along the bar.
For anchorage in a comer region, the influence of transverse stirrups was much
stronger in high strength concrete. One reason could be that the spacing between
the stirrups was smaller in the high strength concrete specimens.
Based on the experimental results from puli-out tests with short embedment
lengths, schematic bond slip relationships, similar to the one in the CEB/FIP Model
Code 90 were proposed, for well confined normal strength and high strength con-
crete. From these schematic local bond stress-slip relationships, it was possible to
predict analytically the end-slip response of puli-out tests with long anchorage
lengths in well-confined concrete.
For anchorage in concrete with small confinement, the end dip response could be
predicted in a reasonable way using the model in MCW for unconfined concrete,
assuming rnm = 2.5 &,)la. Some of the tests with small confinement were close to
confined conditions accordmg to the dehition in the Code. However, the observed
D 0662949 0542829 O23 =
end-slip response were in all cases closer to the prediction assuming unconfined
conditions.
ACKNOWLEDGMENTS
This study is part of a Swedish national research program concerning high strength
concrete and its applications. The authors wish to express their gratitude to the
Swedish Council for Building Research, Nutek, Strängbetong AB, NCC, Skanska,
Cementa, Elkem Materiais and Euroc Betong for financing this study. The financial
assistance from the Swedish Institute, the Swedish Council for Building Research,
and the China State Education Commission for Mr. Huang Zhongyi’s studies in
Sweden is highly appreciated.
REFERENCES
CEB: High Performance Concrete, Recommended Extensions to the Model

Code 90. Research Needs. CEB Bulletin d’informationNo. 228, Lausanne
1995.
Huang Z., Engström B., Magnusson J.: Experimental investigation of the
bond and anchorage behavior of deformed bars in high strength concrete.
Chalmers University of Technology, Division of Concrete Structures,
Report 95:4, February 1996.
Engström B: Ductility of tie connections in precast structures. Chalmers
University of Technology, division of concrete Structures, Publication 92: 1,
Göteborg 1992.
Ghassemi N: Influence of support pressure on the anchorage of ribbed bars in
high strength concrete. Chalmers University of Technology, Division of Con-
crete Structures, Master‘s thesis 96:3. Göteborg 1997. (In Swedish).
CEB/FIP Model Code 1990. CEB Bulletin d’information No. 203, Lausanne
1991.
FIPíCEB: High strength Concrete. State of the Art Report. CEB Bulletin
d’informationNo. 197, London August 1990.
Huang Z., Engström B., Magnusson J.: Experimental and analytical studies of
the bond behavior of deformed bars in high strength concrete. 4th Interna-
tional Symposium of High-strengthhligh-performanceconcrete, Paris, 1996,
pp 1115 -1 124.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
U Obb2949 0542830 845
TABLE 1-TEST PROGRAM
Bar location Number Concrete Concrete Number

ofbars compressive cover of stir-
strength')
f,FIpal c[-I
centroidal 1 101.5 192
-
No
centroidal 1 103.3 192 No
H130a centroidal 1 101.5 192 No
H130b centroidal 1 103.3 192 No
centroidal 1 101.5 192 No
H210 centroidal 101.5 No
H250 centroidai 103.3 No
H170m- 16-0 II mid edge 110.9 No
H170m-32-0 I I I mid edge 110.9 No
H170m- 16-2a II mid edge 105.6 246
H170m-16-2b
H17Om-16-4
H170~-16-0
II
II
III
mid edge
mid edge
corner
105.6
25
No
III
H170~-16-2
H 1700 16-4
N150a
III
I
comer
comer
centroidal
2 112.2 16
4:No:
N150b I centroidal 29.6 192 No
N220 I centroidal 26.8 192 No
N290a I centroidal 26.8 192 No
N290b I centroidal 30.6 192 No
N360 I centroidal 30.6 192 No
N500 I I centroidal 27.6 192 No
N29Om-16-0 I II mid edge 28.6 16 No
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~
N29Om-32-0
N29Om- 16-2a
N29Om-16-2b I
I:: mid edge
mid edge
1 28.6
28.3
32
16
No
246
N29Om-16-4 I
N290016-O
II
II
I III
mid edge
mid edge
comer ~ 2
1
1
28.3
28.6
29.4
16
16
16
2:
No
corner 2 29.4 16
corner 2 29.4 16
gth evaiuated rom a series of 3 tests on cylinders
ISO-StandW
m O662949 0542833 783 m
TABLE 2-COMPOSITION OF CONCRETE MIXES
Material Normal strength High strength

component concrete concrete
[kg/m31 [kg/m3]
Cement 280 500
Water 191 106
Sand 1219 768
Stone 677 954
Silica 1O0
W/C-ratio 0.68 0.33
TABLE 3-MAXIMUM BOND STRESS IN TEST SPECIMENS OF TYPE I, AVERAGE

VALUE ALONG THE FINAL TRANSFER LENGTH
k
Test Concrete Final Maximum Average , , , ,z T,,#,
specimen strength transfer
length
load bond
stress
JE fcc
fcc 4red Nmm 7m.mar

[Wal tml [W [ma]
N150a 26.8 125 47.5 7.55 1.46 0.282
N15Ob 29.6 125 35.9 5.71 1.,o5 0.193
N220 26.8 185 66.7 7.17 1.39 0.268
N290a 26.8 260 102.8 7.87 1.52 0.294
N290b 30.6 260 126.7 9.69 1.75 0.317
average 1.43 0.271
H90a 101.5 70 75.8 21.5 2.13 0.212
H90b 103.3 60 75.6 25,l 2.47 0.243
H130a 101.5 100 94.2 18.7 1.86 0.184
H130b 103.3 1o5 78.5 14.9 1.47 O. 144
H170 101.5 145 116.4 16.0 1.59 0.158
average 1.90 0.188
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~
-~ ~
W Obb29i19 0542632 b38_ D

Tribute to Peter Gerply 229
* n
Ly
a
u
æ
O
o
LL
O
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Ob62949 0542833 554
va
?
U
m
Y
O
W
I
I-
u-
O
Ioocnb-inAOo
Oqqm-\qrq\q
-000000 I
k
I-
U
O
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obbëï49 0542834 490
Displacement
8 transducer
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
j, 4s Displacement
bmsducer
Fig. 1-Test specimens of Type I with the bar anchored along the centroidal axis
under well confined conditions
JSfxwL Series ïïI

Fig. 2-Test specimen of Type II with the bar anchored in mid position near a free
edge, and Type 111 with bars anchored in corner regions
Obb2949 0542835 327
-
I, = 290
€=
Fig. 3-Placement of stirrups along the embedment length in most tests of Type II
and 111
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 054283b 2b3
Yi
O
.-c
a
9
a2
r,
.c
O
U
e
al
al
cc
m
.-u
e
O
u
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542837 L T T
Tensile force Influence of embedment length

[Wl
-
- N220
-
-0-
N290a
N500
H90b
-*- H170
- * - H210
- * - H250
O 1O0 200 300 400 soo

Position x [mm]
Fig. 5-Typical examples of the tensile force development along the anchored bar
at a maximum load. Anchorage in well confined concrete, specimens of Type I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 6-Typical crack pattern in test specimens of Type II. N290m-16-4
Obb2949 0542838 03b
a) normal strength concrete

Tensileforce N290
[IrNl
140
O 1 2 3 4 5 6 7 8 9101112131415
Active end-slip [mm]
b) high-strength concrete
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m-16-0
m32-~
.-....m-16-2;
--- m-16-21
-m-16-4
-h170
O 1 2 3 4 5 6 7 8 9101112131415
Fig 7a and 7b-Relationships between tensile force and end-slip from tests of
Type II in comparison with that in well-confined concrete (Type I)
m 0662947 0542839 T72 D
al N290c-16-2
b) H 1 7 0 ~ - 1 6 - 2
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 8a and 8b-Typical crack pattern in test specimens of Type 111
0bb2949 0542840 794 W
a) normal strength concrete

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
b) high-strength concrete
Tensile force E170
[W
140
.. .- .....- ...-.- I I I - - - - - - - - Tensi!$ CaPECity. ..
I
-C- 16-4
O 1 2 3 4 5 6 7 8 9101112131415
Fig. 9a and 9b-Relationships between tensile force and end-slip from anchorage
in corner regions (Type Ill)in comparison with that in well-confined concrete
(Type 1)
m 0662949 0542843 620
Bond stress, z
SY SI $2 SY.f s3 SS s4 Slip, s
Fig. 1O-Proposed local bond-stress-slip relationship for ribbed reinforcement bar

in confined concrete, according to (7)
Tensile force N290

IW
Analysis - Good bond conditions
Analysis - Other bond conditions

- 0
O 2 4 6 8 10 12 14 16 18 20
Fig. 1 l-Typical test results from anchorage in well confined concrete (specimen
Type I) in comparison with numerical predictions
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542845 567 m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O 1 2 3 4 5 6 7 8 9101112131415
a) test specimens of Type II
Fig. 12a-Relationships between tensile force and end slip in normal strength
concrete with small confinement in comparison with predicted relationships.
CG = model for confined concrete, good bond conditions
CO-model for confined concrete, other bond conditions
UG =model for unconfined concrete, good bond conditions
U0 =model for unconfined concrete, other bond conditions
0662947 0542843 4 T 3 =
Tensile force N290

[W
- Tt?nsi!e caficitu- - -
O 1 2 3 4 5 6 7 8 9 101112131415
b) test specimens of Type 111
Fig. 12b-Relationships between tensile force and end slip in normal strength
CO =model for confined concrete, other bond conditions
UG = model for unconfined concrete, good bond conditions
UO= model for unconfined concrete, other bond conditions
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542844 33T
Tribute to Peter Ger~ely241
Tensüe forœ H170

rw
................... m-16-0
-.-. m32-0
...... m-16-2a
--- m-16-2b
-m-16-4
-CG
---CO
--UG
--u0
O 1 2 3 4 5 6 7 8 9101112131415
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Active end-slip (mm]
a) test specimens of Type II
Fig. 13a-Relationships between tensile force and end slip in high strength
CO = model for confined concrete, other bond conditions
UG =model for unconfined concrete, good bond conditions
U0 -model for unconfined concrete, other bond conditions
= Obb2949 0542845 276
.....
--- C-16-2
-C-16-4
---CO
O 1 2 3 4 5 6 7 8 9101112131415
b) test specimens of Type 111
Fig. 13b-Relationships between tensile force and end slip in high strength
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
CO =model for confined concrete, other bond conditions
UG = model for unconfined concrete, good bond conditions
U0 =model for unconfined concrete, other bond conditions
Ob62949 054284b L O 2 =
SP 180-11
Preventing Brittle Failure of Tension Splices

in High-Strength Concrete
by A. Azizinamini
Svnopsis: Safety concerns and a lack of test data on bond capacity of deformed
reinforcing bars embedded in high-strength concrete are among the reasons for the
AC1 318 building code imposing an arbitrary limitation of 10,000 psi (69 MPa)
when cdcuiaîing tension development and splice lengths. This limitation was first
introduced in the 1989 revision of the AC1 318 building code.
In an attempt to evaluate the impact of this limitation and develop provisions for its
removal, an investigation was carried out at the Universiîy of Nebraska-Lincoln,
partial result of which will be presented in this paper. Results of the investigation
are used to discuss the differences that exist between normal and high strength
concrete, develop hypotheses to explain these observed differences, and suggest
alternatives for removal of the current concrete compressive limitations existing in
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the AC1 318 building code for calculating tension development and splice lengths.
In this paper high strength concrete is defined as concrete with compressive strength
exceeding 10,000psi (69 MPa).
Keywords: Bond (concrete to reinforcement); building codes; deformed

reinforcement; high-strength concrete; lap connections; reinforcing steel; splicing;
structural engineering
243
W O662949 0542847 049 9
244 Azizinamini
Dr.Atorod Azizinamini is an associate professor in civil engineering department at

university of Nebraska-Lincoln. He was employed at Construction technology
Laboratories at Skokie Illinois for four years before joining the University of
Nebraska in 1989. His research interest include design issues related to high strength
concrete and seismic performance of composite structures. He is secretary of AC1
408 committee and active member of number of other AC1 committees. He is a
registered professional engineer in state of Nebraska.
INTRODUCTION
Due to a lack of test data, the AC1 318-95 building code requirements include an
arbitrary limitation of 10,000 psi (69 MPa) on specified compressive strength of
concrete, f that may be used in calculating tension development and tension splice
IC,
lengths. This limitation is stated in Section 12.1.2 of the AC1 318-95 building code.
As a result, an investigation was conducted to evaluate the bond performance of
reinforcing bars embedded in high strength concrete. Results of the investigation
provided a basis to develop a behavioral model in the form of a failure hypothesis
to explain the observed differences between reinforcing bars embedded in normal
and high strength concrete. The major conclusion from this study is that
increasing the splice length is not an efficient approach for improving the
bond performance of reinforcing bars embedded in high strength concrete and
that placing some minimum amount of transverse reinforcement over the
splice regions is an efficient and safe approach.
Experimental Work
A typical splice test specimen used is shown in Figurel. Each specimen

consisted of 2 or 3 reinforcing bars spliced at midspan. The Iongitudinal
reinforcement was either ASTM A615 Grade 60 (specified minimum yield
strength of 414 MPa) No. 11 (36 mm diameter) or No. 8 (25.4 m m diameter)
steel reinforcing. The transverse reinforcement used in some of the specimens
over splice regions consisted of ASTM A615 Grade 60, No. 3 (9.5 mm
diameter) steel reinforcing. The thickness of the side or top concrete cover
was equal to one or two times the bar diameter. The clear spacing between
reinforcing bars at midspan of each specimen was approximately two times
the side cover. Compressive strength of the concrete at the time of testing
were approximately 15,000 psi (103 Mpa). Strength data was obtained by
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
U Qbb294q- 05r12444 -5
testing 4 by 8-in.(102 by 204 mm) diameter cylinders cast at the same time
as the specimen and cured alongside the specimens. Two different
deformation types were used fix Iongitudinal reinforcement. Figure 2 shows
a photo of sample of reinforcing bars with bamboo and diamond shape
deformations. Majority of the test specimens were bottom cast.
The high strength concrete mix used in construction of test specimens included Type
I cement, Class C fly ash, silica fume and superplasticizer. The maximum aggregate
size was % in. (12.7 mm). Water-cementatiousmaterial ratios ranged from 0.21 to
0.27.
Concrete was provided by local ready-mix suppliers. Construction of each specimen

consisted of pouring the concrete at one end of the wood form and proceeding to the
other end. The concrete slump for the high strength concrete test specimens was
about 9 in. (229 mm). Little effort was required to finish the top surface of the
beams. Immediately following casting, the specimens were covered with a plastic
sheet. After approximately five days the forms were removed and the specimens
were covered with wet burlap and plastic sheets, generally until two days before
testing. The casting of each specimen lasted approximately 30 minutes.
Test Set-ur, and Testing Procedures--The test set-up and loading arrangements for
each test are shown schematically in Fig. 3. The test set-up consisted of beam
specimens placed on two roller type supports and loaded equally at each end using
two hydraulic rams and spreader beams. The applied load and resulting deflections
at each bean end and midspan and strains fÌom longitudinai bars and stirrups were
monitored and the data stored in a computer.
The test was begun by applying equal loads at each end of the beam. The load at
each end was applied in increments ranging fiom 0.5 (2.22 KN) to 2 (8.89) kips,
depending on the estimated strength of the beam specimen. Displacement control
was used, for specimens with stirrups, following yielding of the longitud bars. The
load was held constant for approximately five minutes after each load or
displacement increment, during which time cracks were mapped and test
observations recorded. Load or displacement increments continued until the
specimen failed.
Failure HvPothe&-A behavioral model in the form of failure hypothesis was

developed using the results of the test. A detailed discussion and analyses of data
culminating in development of this behayioral model is provided in Ref 1. Using
this model, an attempt was made to describe the observed differences between bond
performance of deformed reinforcing bars embedded in normal and high strength
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
concrete. A brief description of the above-noted behaviorai model is given below.
For the sake of simplicity and to briefly outline the general concept, Fig. 4 shows
a segment of a d e f m e d bar embedded in concrete and subjected to diffèrent levels
of axial tensile forces. This figureshows the free body diagram of the reinforcing
bar at several load stages. At low axial load levels (see Fig. 4a), the outermost lugs
M û b b 2 1 4 9 0542849 9 L L
246 Azizinamini
(i.e. those closest to the loading point) come in contact with concrete.
Consequently, these lugs exert a bearing force on the concrete. The horizontal
component of this force produces bond stress. (The horizontal component of the
friction force is not shown in this figure but also adds to the bond strength.) Figure
4a also shows the corresponding bond stress distribution. As load increases, this
bearing force causes crushing of concrete in the vicinity of the lug. This action
allows the next adjacent lug to come in contact with concrete and participate in
resistng the applied axial tension (See Fig. 4b. The AC1 building code assumes that
at ultimate the bond stress distribution is uniform, which implies that all the lugs
bear against concrete at the ultimate stage (Fig. 4c) and help in resisting the applied
axial force. This is a reasonable assumption to make for NSC and has been shown
to be valid by experimental testing. In the investigation reported in Ref 1,
experimental evidence did not indicate the same behavior for HSC. This
observation could be explained as follows.
Referring to Fig. 4a, when the first lug comes in contact with concrete, a bearing
force acting against the lug is created. The horizontal component of this bearing
force results in what is referred to as bond-stress. The vertical component of the
bearing force creates a radial force which is responsible for splitting the surrounding
concrete. Note also that the be 'ng capacity of concrete is related to f whereas
7
the tensile capacity is related to f IC. Therefore, as an example, assuming that the
bearing capacity and tensile capacity of concrete are given by 0.85 f and 5JfIC,
IC,
IC
respectively, the ratio of the bearing capacity of 15,000 psi (104 MPa) concrete over
that of 5,000 psi (35 MPa) concrete would be 3, whereas the tensile capacity of
15,000 psi (104 Mpa) concrete over that of 5,000 psi (35 MPa) concrete would be
1.73. in other words, increasing the compressive strength from 5,000 psi to 15,000
psi would result in a bearing capacity three times as large, while the tensile capacity
increases only 1.73 times.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
In the case of HSC, the higher bearing capacity of the concrete will prevent crushing
of the concrete in the vicinity of each lug to the extent that would otherwise take
place in normal strength concrete. This implies that, at ultimate, all lugs may not
participate in resisting applied axial forces, and demands that the first few lugs
contribute the most. With the first few lugs being more active, and considering the
fact that in HSC tensile capacity does not increase at the same rate as bearing
capacity, it could be concluded that in the case of HSC failure could be by splitting
of concrete prior to achieving uniform load distribution.
The experimental evidence and use of the above behavioral model (1) led to the
following major conclusions:
(a) In the case of HSC, especially in the presence of small cover, increasing the
splice length is not an efficient approach for increasing bond capacity and, in
fact, may be a waste of material. A mechanism which could delay splitting
of the concrete over the development or splice length would be more
effective in increasing the bond capacity of deformed reinforcing bars
0bb2q4q 0542850 b33 Tribute to Peter Gergely 247
embedded in HSC. This mechanism could be provided by requiring some

minimum amount of stirnips over development or splice lengîhs.
(b) in the case of HSC, the assumption of uniform bond stress over the
development length may not be valid
(c) In the case of HSC, top cast bars exhibit slightly higher bond strength than
bottom cast bars, which is the opposite of thaî generally reported for the case
of NSC. Using the behavioral model described above, Ref. 1 provides a
possible explanation for this observation.
(d) In the past, researchers have used the ratio of bond stress obtained from tests,
UTEST, over bond stress implied by AC1 codes, , as an index when
investigating the bond capacity of reinforcing bars embedded in NSC.
,
U is defined by the following equation:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
where f, is the maximum bar stress in the reinforcement obtained from test,
4, is the diameter of the reinforcement, and 1, is the splice length.
When the value of this index exceeds unity it is assumed that the bond
capacity is adequate. However, in the case of HSC, it was concluded that
resorting oniy to the UmSflAa ratio exceeding one in assessing the bond
capacity of reinforcing bars is a criterion that is necessary but not sufficient.
It was shown thaîthis criterion does not guarantee that members with tension
splices will fail in a ductile manner.
Selection of an Index to assess Bond Data-Results of this investigation indicate that

use of U m a A a ratio as a criterion to study the safety of reinforcing bars embedded
in high strengîh concrete is not adequate. For instance, for two tests carried out in
this investigation, which we will refer them, hereafter, as tests 29 and 30,the splice
lengths were 80 in. (2.03 m)and 57.5 in. (1.46m), respectively. The splice lengths
for Tests 29 and 30 were calculated based on AC1 318-89 (ignoring the f 'c
limitation) and AC1 318-83 building code requirements. The AC1 318-95 code
requirement for the same condition (No.11 reinforcing bars, concrete cover of one
times the bar diameter and f of 103Mpa)would require approximately 45 in. (1.14
m) splice length (ignoring the f 'c limitalion). The U m a A a ratio for specimens 29
and 30 were 1.17 and 1.63,respectively, which is greater than 1. However, both
specimens failed in a very brittle and violent memer without exhibiting ductility.
Figure 5 gives end load versus îhe midspandisplacement for these two tests. As will
be shown later, providing some minimurn amount of transverse reinforcement
results in a significant increase in ductility. In this study, the displacement ductility
ratio, as an index, was selected to assess the test results. Figure 6 gives the
definition of the displacement ductility ratio. For most tests, when incorporating
transverse reinforcement over the splice region, the plot of applied end load versus
248 Azizinamini
W Obb2949 0542851 57T =
midspan displacement exhibited a relationship of the type shown in Fig. 6. The
displacement ductility ratio is defined as the ratio of the maximum midspan
displacement over the first yield. The first yield displacement is defined as the
intersection of the two dashed lines, approximating the load displacement curve, as
indicated in Fig. 6. A displacement ductility ratio of greater than 1 signifies firstly,
that longitudinal bars are capable of developing at least their actual yield stress and,
secondly, specimens will exhibit some level of ductility. Therefore, the use of the
displacement ductility ratio ensures the dual criteria of strength and ductility,
whereas the bond stress ratio is only a strength criteria.
Brief Discussion of Effect of Stirrum Over Splice Region--The observed effect of

transversereinforcement on bond performance of reinforcing bars for test specimens
with #8 or #11 bars and concrete cover thicknesses of one or two times the cover
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
concrete were similar, in general. In this paper, discussion of the effect of stirrups
is provided only for specimens utilizing #11 bars and having one times the bar
diameter as concrete cover thickness.
As discussed earlier, specimens 29 and 30 had 77% and 27% more splice length than
rquired by AC1 3 18-95 provisions, if the f limitation were ignored. However, both
IC
specimens failed in a very brittle and violent manner. Figure 7 shows the midspan
displacement versus total applied end load curves for tests specimens with same
splice length as that of test 30, however, with different amounts of transverse
reinforcement over the splice region. Figures 7a, 7b, 7c and 7d give load
displacement response of specimens having 57.5 inch. (1.46 m) splice length and
varying amounts of stirrups over the splice. Figures sa, 8b and 8c give load
displacement response of specimens having the same bar sizes and concrete cover
thicknesses (#i 1 bars and one times the bar diameter concrete cover thickness) as
that of test 30, except that splice lengths for these specimens were 45 inches (11.4
m), rather than 57.5 inches (1.46 m). Figure 8d gives the load displacement response
of specimen which was similar to specimen 30, except that it had a splice length of
40 inches (1.02 m). Figures 7 and 8 indicate that incorporating some transverse
reinforcement over the splice region results in a significant increase in ductility.
Behavior of Specimens Without Stirrups--A summary of results obtained from all

specimens without stirrups are given in Fig. 9 in terms of splice length versus
displacement ductility. Shown in Fig. 9 are horizontal dashed lines indicating target
displacement ductility ratios needed to ensure adequate behavior for different
combinations of bar sizes and cover thickness. In a previous study, this level of
ductility was judged to provide an adequate level of performance in terms of
displacement ductility. Figure 9 shows a series of vertical lines corresponding to the
required splice lengths for different bar sizes and cover concrete thickness,
calculatedbased on AC1 3 18-95 provision (ignoring the f limitation). Therefore, IC,
for a particular condition, test points that fall to the right of the vertical lines and
above the corresponding horizontal dashed line would represent a safe design
condition. Further, the intersection of sloped lines, produced by connecting test data
with vertical lines, would represent a level of displacement ductility that could be
Obb2949 0542852 406
anticipated if design was based on AC1 318-95 requirements (neglecting f IC

limitations). If this intersection point falls above the horizontal dashed limes, hen
the conclusion could be made that for those particular bar sizes and cover concrete
thickness the AC1 318-95 provisions (with f 'c limitations eliminated) are adequate.
As indicated in Fig. 9, none of the cases presented (#i 1 bars with either one or two
times the bar diameter thickness of cover concrete or #8 bars with either one or two
times the bar diameter thickness of cover concrete) satisfies this scenario.
From Fig. 9 it can also be concluded that the problem with lack of ductility is more
pronounced with #l 1 bars having one times the bar diameter as concrete cover.
CONCLUSIONS
By reviewing the information generated during the investigation, following main

conclusions were drawn:
(a) In the case of HSC, especially in the presence of small cover, increasing the
splice length is not an efficient approach for increasing bond capacity. A
mechanism which could delay splitting of the concrete over tension
development or tension splice length would be more effective in increasing
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the bond capacity of deformed reinforcing bars embedded in HSC. This
mechanism could be provided by providing some minimum amount of
stirrups over development or splice lengths.
(b) In the past, researchers have used the ratio of bond stress obtained fiom tests,
UTEST, over bond stress implied by AC1 codes& , as an index when
investigating the bond capacity of reinforcing bars embedded in NSC. When
the value of this index exceeds unity, it is assumed that the bond capacity is
adequate. However, in the case of HSC, it was concluded that resorting only
to the UmTíü Ac, ratio exceeding one in assessing the bond capacity of
reinforcing bars is a criterion that is necessary but not sufficient. It was
shown that this criterion does not guarantee that members with tension
splices will fail in a ductile manner.
(c) By studying the behavior of specimens having #8 and #11 reinforcing bars
and varying amounts of concrete cover, it was concluded that even in the case
of larger concrete cover thicknesses, one will have to provide some minimum
amount of stirrups over the tension development or tension splice lengths.
However, as concrete cover thicknesses increases, the minimum stirrup
requirements should be decreasing.
ACKNOWLEDGEMENTS
This project was carried out in two phases. Phase II of the project, which was the
major part of the investigation, was supported by the National Science Foundation.
W 0662749 0542853 342
250 Azizinamini
The authors are very grateful for this support and to Dr. Ken Chong, program
director at NSF. Phase I of the project was funded by Portland Cement Foundation.
This support is gratefully acknowledged. Partial support for this project was also
provided by the Center for Infrastructure Research at the University of Nebraska-
Lincoln, for which the authors are greatly appreciative.
The contents of this paper reflect the views of the author, who is responsible for the
facts and accuracy of the data presented, and do not necessarily represent the views
of the sponsors.
REFERENCES
(1) Azizinamini, A., M. Stark, J.J. Roller, and S.K. Ghosh, “Bond performance
of reinforcing bars embedded in high strength concrete”, AC1 Structural
Journal, Sept.-Oct. 1993, pp. 554-561.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb29Y9 0542854 28q
Tribute to Peter Gergeiy 251
Splice
r"i I
Elevation
Plan
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
-
Section A A
Fig. 1-Test specimen
252 Azizinamini
= Obb2949 0542655 115 =
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 2-Different bar sizes used in experimental program
m 0662949 054285b 0 5 1 m Tribute to Peter Gergely 253
Hydraulic L m s
Fig. 3-Test setup
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~
254 Azizinamini
= Obb2949 0542857 T98
Forces on Bond S k e s s
Reinforcing Bar Distrlbutlon
N o r m a l Strength
Concrete
\High Strength
Concrete
Fig. 4-Freebody diagram of a reinforcing bar embedded in concrete and subjected

to tension
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542858 924 W
Specimen 29
,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
6
Mid-span displacement, in.
(a)
80-
Specimen 30
9
O
7 40-
B
o
Fig. 5-Load displacement response of Specimens 29 and 30
Ob62949 0 5 4 2 8 5 9 BbO 9
256 Azizinamini
Fig. 6-Definition of displacement ductility ratio
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
obb2q4q 05428b0 582 Tribute to Peter Gergely 257
v)
P
.-c2.
L
v)
.-
c
I
v)
E
.-E
u
aa
n
v)
w.-
O
a2
v)
c
O
u
v)
L
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0 5 4 2 ô b 1 419 m
258 Azizinamini
I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M
n
2
.-c
L
IA
.
..-
e
m
vl U
m
C O
e
'II
I
8
v>
n
O
W
3
9 Obb2949 û5428bë 355
IO 20 30 40 50 60 70 80
Splice Length, In.
-+--NO.8, Idb +NO.8, 2db - NO.11, Idb - NO. Il,2db
Fig. 9-Behavior of specimens without stirrups
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
06629119 05112863 29s m
SP 180-12
Bond of Epoxy-Coated Reinforcement

in Normal and High-Strength Concrete
by T. Grundhoffer, P. A. Mendis, G. W. French and R. Leon
Synopsis: Epoxy-coated reinforcement and high-performance concrete are

commonly used materials in exposed structures located in cold regions and marine
environments of the United States. Their popularity is due to their resistance to
corrosion in areas where chlorides are used as deicers in roads and bridges. This
paper summarizes an experimental investigation regarding the difference in bond
behavior of epoxy-coated and uncoated reinforcement in normal and high-
strength concrete. The objectives were to investigate the effect of bar surface
(epoxy. uncoated), bar size (No. 6. No. 8. No. 1 I), concrete strength (6. IO, 12.
14 ksi) and the addition of micro-silica to concrete. Ninety-four inverted half-
beam specimens were tested. All of the specimens were designed to fail in bond
by splitting of the concrete. The reinforcement in four of the specimens (two
uncoated and two epoxy-coated reinforcement) was instrumented with internally
embedded strain gages to measure the distribution of strain along the embedment
length. The tests showed clear differences in the strain distribution at service level
between coated and uncoated reinforcement. A comprehensive review of the
effect of epoxy-coating on bond strength was conducted using the results of this
study and 151 test results from seven other research studies in the USA. The
experimental results were compared to values of design bond strengrh calculated
using AC1 318-89 ( I ) and AC1 318-95 (2) equations.
Keywords: Bond (concrete to reinforcement); bridge specifications; building codes;

deformed reinforcement; development; lap connections; reinforcing steels; relative
rib area; reliability; splicing; structural engineering; variability
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
261
Obb2949 05Li2864 128
262 Grundhoffer et al.
Tim Grundhoffer is presently with the St. Paul Army Corps of Engineers. Tim
completed his M.S. Thesis on ?Bond Behaviour of Uncoated and Epoxy-coated
Reinforcement in Concrete? at the University of Minnesota in March 1992.
Priyan Mendis is a Senior Lecturer at the University of Melbourne, Australia. In

1996, he worked as a visiting Associate Professor at the University of Minnesota.
His primary research interests include High-strengthhigh-performance concrete,
seismic design of concrete structures and tall buildings. He is the chairinan of the
international sub-committee of AC1 363 commiltee on high-strength concrere.
Catherine French is an Associate Professor at the University of Minnesota, USA.

Her primary research interests include the behaviour of reinforced and prestressed
concrete structures, high-performance concrete, durability and seismic issues. She
is a member of AC1 committees 318,352, 368,423 and 442. She is also the
Secretary of AC1 Committee 445 on Shear and Torsion.
Roberto T. Leon is a Professor of Structural Engineering and Engineering

Mechanics in the School of Civil and Environmental Engineering at the Georgia
Institute of Technology. He received his Ph.D. from the University of Texas in
1983, and is currently chairman of AC1 Committee 408 - Bond and Development
of Reinforcement and a member of AC1 352,335 and ITG-2.
INTRODUCTION
In cold regions of the United States, extensive use of deicing salts on

roadways causes severe durability concerns for bridge decks and parking garage
structures. To increase the life of these structures, epoxy-coated reinforcement
has been commonly used to inhibit reinforcement corrosion in these severe
environments. It has recently become commonplace to use high-performance
concrete in combination with epoxy-coated rebar to further inhibit the corrosion
process by increasing the impermeability of the concrete. The increased concrete
impermeability inhibits the ingress of chlorides into the concrete. This two-
pronged approach is an effective way to at least delay the corrosion problems
associated with aggressive environments.
A disadvantage of epoxy-coated reinforcement is that longer anchorage

lengths are required to fully develop the reinforcement. Current AC1 and
AASHTO codes (2,3) recognize the decreased ultimate bond strength of epoxy-
coated reinforcement by specifying amplified development lengths for epoxy-
coated reinforcement. The amplified development lengths are 20 to 50 percent
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 06b2949 05428b5 Ob4
greater than those of uncoated reinforcement (4,5). These requirements. based on

limited sets of data, were included in design codes as an effort to prevent
problems while more research clarified the mechanistic basis for this increase (6-
9). It is generally assumed that the decreased bond capacity of epoxy-coated
reinforcement is due to the different friction characteristics and the lack of initial
chemical adhesion between the bars and the concrete. This paper reviews this
assumption and provides quantitative results that expand the understanding of
decreased bond capacity of epoxy-coated reinforcement.
The main objective of the study reported in this paper was to investigate
the bond strength of epoxy-coated reinforcing bars cast in concrete with
compressive strengths ranging from 6 to 14 ksi. Specimens were also cast using
concrete of the same strength with and without micro-silica to investigate the
effect of micro-silica on bond behavior. Additionally, the effect of epoxy coating
on the rebar strain distribution along the development length was investigated.
The results of seven other research studies conducted in the USA were used to
support the findings of this study. The bond length equation given in AC1 318-89
has been modified to a more “user-friendly” format in AC1 318-95. It is shown
from the results of this study and other studies that in some cases, the AC1 318-
95 predicts unconservative values for bond strength of epoxy-coated bars.
BOND MECHANISM AND FAILURE MODES
Bond stresses modify the steel stresses along the length of the bar by transferring
load between the bar and the surrounding concrete. The following expression IMY be
derived from equilibrium of the concrete and bar forces:
where A h and tfb are the area and diameter of the reinforcing bar, id is the bond
length of bar,& is the stress developed in the bar, and u is the average bond stress.
The average bond stress can be related to the bar diameter, bar stress, and bond
length
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
This formula is used to determine the average bond stress developed between the
reinforcing bar and concrete.
The bond of deformed reinforcenient @ concrete is a complicated mechanism

which is mostly understood in a qualitative nature. It is essential that the bar force is
transferred to the concrete to maintain structural integrity. The bar force is transferred
= Ob62949 0542866 TTO
to the concrete by adhesion, friction and mechanical bearing between the deformation
and concrete. Figure 1 illustrates schematically the three mechanisms of bond. Upon
initial loading the forces are transferred by adhesion created through chemical bonding
between the steel bar surface and concrete. The adhesion is not a sustained resistance.
At low bar stresses the adhesion is lost. After adhesion is lost the bar slips relative to
the concrete which enables development of the friction and mechanical bearing
mechanisms. Due to the rib face angle (Figure I ) the forces are transferred to the
concrete by bearing perpendicular to the rib face and friction between the rib face and
the adjacent concrete. The resultant force of the bearing and friction forces produces
radial tension in the concrete surrounding the bar.
Two types of bond failures exist: pullout failure and splitting failure. If adequate
confinement exists in the form of transverse steel, large cover, or a combination
thereof, a pullout failure occurs. A pullout failure is a direct shear failure of the
concrete key at the level of the outer edge of the deformation. The confinement allows
the tensile stresses in the concrete to be resisted. This allows the bearing pressures
between the rib face and the concrete to increase with incrcasing bar load and the
frictional component becomes less significant. The high bearing pressures result in
failure of the concrete keys in shear.
If sufficient concrete cover and/or transverse confuiement are not provided to

resist the radial tension stress in the concrete. a splitting failure occurs. Once the
concrete cracks. the deformations push the concrete away from the bar by wedge
action. As the concrete starts to ride up the rib face the component of friction
between the rib face and the concrete becomes more significant.
Splitting bond failures are more likely to occur in slabs and other structural
members without transverse reinforcement or large concrete cover. The study,
conducted at the University of Minnesota, focussed on splitting bond failures.
Treece and Jirsa (5) stated that the primary reason for the reduction in bond
strength appears to be the loss of adhesion between the concrete and epoxy-coated
bars which destroys most or all of the friction capacity. Their hypothesis is supported
by the results of this investigation. Uncoated bars have good adhesion to concrete.
The reduced friction of epoxy-coated bars increases the radial pressure component
which sets up radial tension in the concrete cover (Figure 2 ) . Therefore the bond
strength at initial cracking is controlled by the magnitude of the radial pressure that the
concrete cover can resist. In addition, the researchers in this study believe that the
epoxy coating reduces the effective bearing area of the reinforcement by reducing the
rib height of the deformations (Figure 3), which has been shown to reduce bond
stiffness and capacity. As seen from Figure 3. the rib height is reduced by twice the
coating thickness.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0662949 05428b7 937 m Tribute to Peter Gergely 265
Some researchers have suggested that the presence of silica fume may actualiy
alter the concrete transition zone at the bar surface (10,ll). However, how the surface
is altered and the direct effects of silica fume on bond appear inconclusive, a$ it has
been found to both increase and decrease the bond capacity of concrete. Hwang et ai.
(1 1) suggested that the presence of silica fume decreases the adhesion at the surface of
the bar and concrete, invariably decreasing the developed bond stress. Gjorv et ai.
(10) suggested that the presence of silica fume increases the adhesion between the
concrete and bar surface, invariably increasing the bond stress due to increased
adhesion and fiction.
RESEARCH SIGNIFICANCE
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Epoxy-coated reinforcement is commonly used in exposed structures in cold

regions and marine environments of the USA. Epoxy coating and micro-silica
fume are used in combination to reduce both the corrosion potential of the
reinforcing bars and the permeability of the concrete. To better understand the
effects of epoxy coating on the reinforcement bond to the concrete, an
experimental investigation was conducted at the University of Minnesota. Tests
were performed on 96 beam-end specimens. The variables included the
reinforcement bar diameter (No. 6.8 and 11). concrete strengths (6 to 14 ksi) and
use of micro-silica in the mix. The effect of micro-silica on the bond strengths of
epoxy-coated bars had not been studied before. In this study, load-slip behavior and
the rebar strain distribution along the development length were measured
experimentally to investigate the differences in load distribution and transfer between
the uncoated and epoxy-coated rebar. A comprehensive review of 151 test results
of this study and other research studies was performed to confirm the findings of
the research at the University of Minnesota.
LITERATURE REVIEW
Clifton et al. (1974) (12) evaluated the effect of epoxy coating on bond strength
by testing puiiout type specimens. Twenty three specimens were fabricated with
epoxy-coated No. 6 rebar,and five had uncoated rebar. The bond performance of the
epoxy-coatedrebar was compared to the uncoated rebar using a slip criteria. The test
resuits showed that the epoxy-coated rebar with coating thicknesses between 1 to 11
mils had a critical bond stress 6 percent lower than the uncoated rebat specimens. It
was recommended that the coating thickness shouid not exceed 10 mils.
Johnston and Zia (1982) (4) conducted a study to compare the bond
characteristics of epoxy-coated rebar specimens to uncoated rebar specimens under
rn 0662949 0542868 873 rn
static and fatigue loadings. Forty inverted haif-beam specimens were tested under
static and fatigue loadings to investigate the ultimate bond strength and rebar slip
relative to the concrete. The test variables in the inverted half-beam tests included the
rebar surface condition, bar size, and development length. The performance of the
epoxy-coated rebar was compared to that of the uncoated rebar using a slip criteria
and ultimate bond strengths. From the test results it was recommended for design that
the basic development lengths for epoxy-coated rebar be multiplied by I . 15.
Treece and Jirsa (1987) ( 5 ) tested 21 splice-type beam specimens to compare the
ultimate bond strength of epoxy-coated rebar with that of uncoated rebar. The test
variables included bar size, coating thickness, concrete strength, and casting position.
The specimens included No. 6 and No. 11 spliced rebar with coating thicknesses
varying from 5 to 12 mils. The concrete strengths were 4, 8, and 12 ksi. Seventeen
specimens were cast with the spliced bars in the top position. The remainder of the
specimens were cast with the spliced bars in the bottom position. AU specimens
exhibited a splitting failure. The theoretical bond strength was determined using the
following equation developed by Orangun. Jirsa. and Breen (13).
where UOJB = ultimate bond stress, psi

fc = concrete compressive strength, psi
C = lesser of the side cover and bottom cover, in.
dh = nominal rebar diameter, in.
1s = splice or development length, in.
The above equation will be referred to as the OJB equation for the remainder of
this paper. It was shown that the mean bond strength ratio of uncoated to coated
rebar was 1.33. The report stated that the reduction was independent of the bar size,
coating thickness ( 5 to 12 mils), and concrete strength. It was recommended that
epoxy-coated rebar with a cover less than 3d1,or a clear spacing less than 6 4 have its
required basic development length multiplied by a factor of 1.5. The development
length for all other epoxy-coated rebar should be multiplied by 1.15 as found by
Johnston and Zia.
Choi et al. (1990) (7) reported the results of the fist phase of a continuing study
at the University of Kansas Center for Research. The study considered the effects of
bar size, coating thickness, deformation pattern, concrete cover, and casting position.
Inverted haif-beam specimens and splice type specimens were used. Bar sizes of No.
5,6,8,and 11 were used with coating thicknesses varying from 3 to 17 mils. Three
deformation patterns were used and designated by the S, C, and N shapes. None of
these deformation patterns were the same as the diamond shape deformation patterns
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb29ir9 05428bï 70T
used by Johnston et al. and T r w x et al., two of the studies described above. Clifton
et ai. (12) also used a barrel shape similar but not identical to the S shape. Top and
bottom casting positions along with bar covers of 1,2 and 3 bar diameters were also
studied. A total of 394 inverted half-beam tests and 4 splice type tests were
conducted. Au specimens exhibited a splitting failm. The inverted half-beam
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
specimens showed that the ratio of uncoated to coated rebar bond strength was 1.18
after averaging results from ali of the tests. The splice type specimens showed that
this ratio ranged from 1.06 to1.40 with a mean value of 1.15. The reduction in bond
strength varied with the different deformation patterns. It was stated that deformation
patterns with iarger rib bearing areas were affected less by the epoxy coating.
Increased concrete cover did not produce a significant difference in bond reduction
caused by the epoxy coating. However, the trend was that the reduction decreased as
the concrete cover increased. It was concluded that for No. 6 and larger bars, coating
thicknesses in the range of 5 to 12 mils did not affect the bond strength. However,
results fkom No. 5 bars indicated that bond strength decreased as coating thickness
increased.
Devries et ai. (1991) (8) investigated the effect of concrete strength, casting
position and anti-bleed agents on the bond strength of uncoated and epoxy-coated
rebar. Concrete strengths ranged from 8ooO to 15ooO psi. All of the mixes contained
micro-silica ranging in content from 7 to 12 percent. Micro-silica was not an intended
variable in the study. It was varied to o h i n higher concrete strength. The test bars
were cast in both the top (at least 12 inches of concrete below the test bar) and
bottom position. The presence of an anti-bleed agent (Kelco Gum) was also studied.
Thirty-six splice type specimens similar to those of Treece and Jirsa(l989) were
tested. Two bar sizes. No. 6 and No. 9 bars, were tested with no replications of the
specimens. The cover was 1.125 inches for all specimens. The ratio of uncoated to
coated rebar bond strengths ranged from 1.30 to 1.38 for the No. 6 bottom cast test
bars and 1.10 to 1.40 for the No. 9 bottom cast test bars. It was stated that the AC1
modification factor of 1.5 was conservative for ali concrete strengths tested and that
no limit should be put on the square root offe They stated measured bond strengths,
normaked with the OJB equation, showed a decreasing trend in normalized bond
strength asfc increased. No recommendation was made to adjust the modification
according to varyingf,. The top cast bars were found to give conservative results
reiative to the AcI318-89 equation using a modification factor of 1.3. Top cast bars
with the presence of an anti-bleeding agent showed a 5 percent increase in bond
strength compared to those of other top cast bars without an anti-bleeding agent.
Cleary and Ramkz (1991) (14) conducted a study of the bond of epoxy-coated
reinforcement in &type members. The slabs were 13 fi. (approx. 4 m) long, 2 ft
(0.6 m) wide,and 8 in. (20.3 cm) deep. Reinforcement consisted of three No. 6 bars
spliced at midspan. Splice lengths varied from 16 in. (40.6 cm) to 10 in. (25.4 cm).
The cover-to-bar diameter ratio was 2.67. They conciuded that due to the increase of
rib bearing forces with epoxy-coated reinforcemnt, it is likely that splitting could
Ob62949 0542870 42%
occur with epoxy-coated bars at cover-to-bar diameter ratios greater than the limit of
3 suggested in AC1 318-89 (and AC1 318-95) for transition between splitting type
failure and puii-out type failure.
Hester et al. (1991) (15) studied the effects of epoxy coating and transverse
reinforcement on the splice strength of reinforcing bars in concrete. A total of 65
beam and slab splice specimens containing No. 6 and No. 8 bars were tested. Three
deformation patterns (S,C and N) were used. The results of their study were analyzed
along with the results of 48 specimens from earlier studies, and used to suggest
improved modification factors for use with epoxy-coated bars. They concluded that
the percentage decrease caused by epoxy-coatings is independent of the degree of
confining reinforcement, which provides approximately the same percentage increase
in the strength of splices for both coated and uncoated bars. They suggested a
maximum development length modification factor of 1.35 for design with epoxy-
coated reinforcement.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Hadje-Ghaffari et al. (1991) (7) conducted a study which was a continuation of
the study done by Choi et al. summarized earlier. In the continuation study, the added
effect of concrete slump, consolidation, transverse reinforcement, and concrete
strength were investigated. Slumps varied from 2 to 8 inches and concrete
consolidation was varied by either vibrating the concrete or not vibrating the concrete.
The concrete compressive strengths were nominally 5000, 6000 and 13000 psi. Only
12 of the 645 specimens evaluated had concrete strengths of 13000 psi. Higher
slumps and vibration improved the bond strength of both epoxy-coated and uncoated
bars. The epoxy-coated bars with transverse reinforcement showed bond strengths
near those of the uncoated bars without transverse reinforcement. The reduction of
bond strength due to epoxy coating decreased about 6 percent with transverse
reinforcement present. A somewhat surprising result was that more than doubling the
concrete strength had a little or no effect on the bond strength. Specimens with
13000 psi concrete compressive strength showed a little or no increase in bond
strength compared to those with a concrete compressive strength of 6000 psi. It was
suggested that the modification factor for epoxy-coated bars smaiier than No. 7
should be 1.25. For larger bars, a modification factor of 1.35 was suggested. The
study concurred with the AC1 top-bar modification factor of I .3 for uncoated rebar.
They suggested that the epoxy-coated top bars should have a combined niodification
factor of 1.5.
Cusens and Yu (1993) (16) studied the behavior of concrete beams with fusion-
bonded epoxy-coated reinforcing bars (Unisteel bars manufactured in U.K.) under
static, repeated and sustained loads. A total of 12 specimens were tested with
different embedment lengths in static tests. In static load tests, higher bond stresses
(with lower values of loaded-end slip) were observed in both epoxy-coated and
uncoated bar specimens with short lengths of embedment. In beams with similar
embedment lengths, the coated bar specimens gave more slip at both ends and lower
Obb2949 054t87L 3 b 8 ~ m
values of bond stress. The effect of epoxy coating on the bond slip resistance was
observed to be more significant with smaller diameter bars.
Hamaà and Jrsa (1993) (17) investigated the effect of epoxy-coated transverse
reinforcement on the strength of epoxy-coated bar splices. Twelve beams were tested
in negative bending with multiple splices in a constant moment region at the center of
the bem The main variable was the amount of transverse reinforcement crossing the
spiitting plane in the splice region. Other variabies were bar size and bar spacing. Test
results indicated that the presence of transverse reinforcement in the splice region
increased deformation capacity of the beams and improved anchorage strength of
epoxy-coated bar splices relative to uncoated bar splices. The improvement was
independent of the number of splices, bar size, or bar spacing.
Cairns and Alxìullah (1994) (18) tested six half-scale beams in pairs, each pair
containing epoxy-coated bars and uncoated bars. The bars were produced in U.K.
with a characteristic yield strength of 67 ksi. Two bar sizes equivalent to No. 5 and
No. 6 and three lap lengths equal to 204,28.5d6 and 37dh were used. Each beam was
tested in four-point bending. The minimum number of stirrups determined in
accordance with BS 8 1 10 (19) was provided throughout the constant moment region.
The bond ratios at failure (uncoated/coated)for lap lengths of 2Od6,28.566 and 37db
were 1.19, 1.15 and 1.11 respectively. They also compared the results of 88 test
results from six other research studies and concluded that the present rules suggested
for the design of uncoated reinforcement appear to provide an adequate safety factor
against the bond îäüure of coated bars in most practical circumstances.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
EXPERIMENTAL PROGRAM
A total of 96 inverted half-beam specimens were used for the study. This
type of specimen allows the test bar to be in an area of flexural tension under a
moment gradient and constant shear. The gross cross section of the specimens
was 12 in. wide and 18 in. deep. Specimen length varied with bar size to reduce
shear stresses and produce a more realistic strain field. The total length of the No.
6, 8 and 1 1 specimens were 36,42 and 47 in. long, respectively. A typical cross
section of the test specimen is shown in Figure 4. The testing frame is shown in
Figure 5. The loads were obtained using load cells attached to the jacks and strain
gages applied to the reinforcement in the loaded end region. The loaded and
unloaded end siips were monitored with high-precision linear variable differential
transformers (LVDT) capable of detecting movements as small as 0.01 mm.
Auxiliary reinforcement in the form of U-shaped stirrups was provided to prevent
any shear failures.
W Obb2949 0542872 2 T 4 W
Nominal concrete cover was 2dh (1.5 in.) for the No. 6 bars. For the No. 8
bars, the nominal cover was increased to 2.5& (2.5 in.) to accommodate even
spacing of the internal strain gages over the embedded length. Nominal cover for
the No. 11 bars was decreased to 1.8dh (2.5 in.) to enable increasing the
embedded length. Embedded lengths were controlled by shielding portions of the
reinforcement with PVC pipe. The lead length, distance from the loaded-end
concrete face to the beginning of the embedment length, was taken as 0.5 in. for
all specimens. When considering the specimen width and comparing test results of
Cho¡ et al. (7) with the OJB equation (Eqn. 3), this lead length value gave the best
correlation.
Primary variables included rebar surface condition (epoxy-coated, uncoated),

bar size (Nos. 6.8 and 111, nominal concrete strength (6 to 14 ksi), and concrete
type (with and without micro-silica). The No. 6 and No. I 1 bars were used for
most of the studies to obtain maximum differences in behavior. The No. 8 bars
were used for the detailed bond distribution studies utilizing embedded strain
gages.
All bars were Grade 60 ASTM A615 bars (60 ksi, minimum elongation at
failure of 8%) (20) and were obtained from a local producer. Special care was
taken to ensure that both the uncoated and epoxy-coated reinforcement of the
same size came from the saine heat of steel, The actual yield strengths were 62 ksi
for the No. 6 and I 1 bars. and 69.2 ksi for the No. 8 bars. Ultimate strengths
were 98, 101 and 109 ksi, for the No. 6, 8 and 1 i bars, respectively. The bar
deformation pattern was the "N-type" which consists of diagonal ribs, inclined at
70 degrees to the longitudinal axis of the bar, joined by two longitudinal ribs
along opposite sides. The deformation pattern resulted in average relative rib
areas of 0.084, 0.080 and 0.087 for the No. 6, 8 and 11 bars, respectively. The
relative rib area is defined as the ratio of projected rib area normal to bar axis to
the product of the nominal bar perimeter and the center-to-center rib spacing.
The epoxy coating was fusion bonded and conformed to ASTM A775 (21).
Actual measurements in selected bars indicated average thicknesses ranging from
5 to 8 mils for the No. 6 bars, and 10 to 15 mils foi- the No. I l bars. All uncoated
rebar specimens had identical epoxy-coated companion specimens.
The concrete was obtained from a local ready-mix plant. Specimens were
divided into eight groups depending on the concrete characteristics. Table 1
shows the variables used for each study. Note that groups 3 and 5 showed
significant differences from the specified strengths. The bonded embedment
lengths of the reinforcement in each specimen are also listed in Table I .
The normal mix design in Groups 1 and 8 (6 ksi) was a mix specified for
structurai bridge decks in the State of Minnesota. The 6 ksi micro-silica mix
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662749 0542873 i130 m
(Group 3) was a standard parking ramp deck mix. The 10 ksi to 14 ksi micro-
silica mixes were used typically in columns and core walls of multistory buildings.
Because research on bond strength is used to gage the required development

lengths for design codes, and because the relationship between bond stress and
development length is not linear, it is desirable to test specimens that develop bar
stresses close to the desired design code bar stresses. The approach used in this
study was to select embedment lengths that would produce bond failures at
stresses near the nominal yield stress without exceeding it. To estimate these
embedment lengths, use was made of the relationship derived by Orangun, Jirsa,
and Breen (Eqn. (3)).
The No. 8 bars (Group 8) were instrumented with internal strain gages to
investigate the distribution of bond stresses along the bar length. The bar was split
in half, two grooves were machined into the flat surfaces, gages were installed,
and the bar was then glued back together so that the concrete rebar interface was
not disturbed (Figure 6). This method was first used by Mains (22) and provides
the determination of undisturbed strain distributions for reinforcement embedded
in concrete. The authors are not aware of any research of this nature conducted
with epoxy-coated rebar.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
RESULTS AND OBSERVATIONS
Typical load-slip curves for both loaded and unloaded end slips for Group 4
No. 6 and No. 11 bars are shown in Figs. 7(a) and 7(b) respectively. All
specimens failed in bond, exhibiting a splitting type failure. The initial slope
indicates the relative slip among the bars for the same change in load. The initial slope
of the free-end slip curve was obtained by siilpile linear regression analyses of the first
20 data points. This corresponded with loads of approximately 8, 14, and 20 kips for
the No. 6, No. 8, and No. 11 bars, respectively. It was found that generally the initiai
slopes obtained for the uncoated bar specimens ranged from one to two times the
initial slope of the epoxy-coated bar specimens (6).
The reduction in initial slope supports the hypothesis that the reduced "effective
relative rib area" due to the coating thickness is reducing bond stifhess. The coating
at the top of the deformation cannot provide much bearing resistance and thus cannot
be inchded in the "effective" relative n i area. The end result is that relative rib area is
decreased. Work done by Martin and Noakowski (23) showed that bond strength and
stiffness decrease with decreasing relative rib area. Better adhesion exhibited by the
uncoated bars would also tend to give iarger bond st&ss. Another possibie reason
for the initial decrease in bond stiffness is the epoxy coating could have compressed
between the deformation and the concrete.
Obb2949 0542874 077
A solution to the reduced relative rib area with epoxy-coated bars would be to
increaie the height of the deformation to offset the coating thickness so that the
epoxy-coated and uncoated rebar had the same "effective" relative rib area.
The reduced friction between the epoxy coating and concrete and the reduced
effective bearing area cannot resist the wedging action of the bar. The friction
developed by the uncoated bar and larger effective relative rib area are enough to
resist wedge action and allow significant bearing between the deforination and
concrete after initiai bond cracks have occurred. In addition to increasing the
deformation height ofepoxy-coated bars, grit might be added to the epoxy to increase
the friction between the concrete and the rebar surface (as done for epoxy-coated
prestressing cables).
The results of this study show that the companion epoxy-coated and uncoated
test bar specimens exhibit initiai splitting of the concrete cover at different loads. The
epoxy-coated specimens exhibit initiai craciung of the concrete cover at loads that are
average 91 percent of the uncoated speciniens initiai cracking loads (6). The bar load
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
corresponding to the splitting of the concrete cover is directly related to the radial
pressure exerted by the bar on the surrounding concrete. The lower splitting loads for
the epoxy-coated bars support the hypothesis that the reduction in bond strength is
due to the loss of adhesion between the concrete and epoxy-coated bars which
destroys most or all of the friction capacity. A further extension of the hypothesis is
what happens after splitting of the concrete cover. After initiai cracking, the epoxy-
coated bars showed an increased rate in free-end slip while the rate of increase in free-
end slip for the uncoated bars remained constant. The same behavior is seen on the
loaded-end slip curve. After initial cracking, the slope of the curve decreases more for
the epoxy-coated bar than for the uncoated bars.
As mentioned before, overall adhesion between the concrete and steel is lost at
low bar stresses, but the adhesion which improves bond performance is that of the
concrete particles which remain attached to the bar surface. Reported by this study
and other studies, after failure of the bond specimen localized concrete particles were
observed to stdl adhere to the uncoated bar. The interlock of these particles with the
adjacent concrete develops friction. The friction developed at the rib face. is most
signifcant in resisting wedge action (or slip). On areas of the bar without local
adhesion, friction was developed due to the roughness of the rebar surface produced
when hot rolied. Bbst clean bars have shown bond strengths in between those of
epoxy-coated and uncoated b a s (Johnston and Zia (4)). The roughness of the blast
clean bars is most likely between that of epoxy-coated and uncoated bars. The lack of
localized concrete particles adhering to the epoxy and the smoothness of the coating
decreased the ability to develop signilicant friction forces especiaiìy after initiai bond
cracks had occurred.
Obb2"' 05'2875
Figure 8 shows a comparison of the initial and final strain distributions for
the No. 8 bars. At low stress level (below 10 ksi) the epoxy-coated rebar showed
rebar strains throughout the entire embedment length. The uncoated bars showed
noticeable amounts of strain only in the first 8 to 10 in. The strain distribution
results at low stress level imply the uncoated bars transferred load to the concrete
over a shorter length than the epoxy-coated bars. For the uncoated bar specimens,
strain was not observed throughout the entire embedment length until the bar was
loaded to 20 to 30 ksi stress (medium stress levels). On average, as the stress
level was increased the uncoated bar specimens develop an initial bond crack at
30 ksi, while the epoxy-coated bar specimens developed the crack at 28 ksi. This
also coincided with the loads at which the strain distribution between the coated
and uncoated bar specimens were almost the same.
As higher stress levels were reached (above 30 ksi) the bond crack
propagated farther along the embedment length. This occurred as the wedge
action of the deformation pushed the concrete away and thus decreased the
transfer of load to the concrete. The deterioration of the bond produced large
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
increases in strain in the rebar especially near the loaded end where the greatest
extent of concrete deterioration had occurred.
At about 45 ksi the uncoated bars exhibited the same or larger strains than
the epoxy-coated bars. The increased rate of slip for the epoxy-coated bars was
relieving the load and not allowing bond to increase the load at a significant rate.
This would be similar to a mobilized friction situation. The uncoated bars were
able to resist slip (wedge action) which allowed strain in the deteriorated regions
to increase and more efficient transfer of load to the concrete in area where the
concrete was deteriorated. At failure, the uncoated bars showed an almost
uniform strain in the first 3 to 4 in. of the embedment length. This was most likely
due to a loss of bond transfer in that area and the ability of the rest of the
embedment length to restrain load without significant slip. At failure, the epoxy-
coated bars showed a trend to produce uniform strain in only the first 1.5 in. of
the embedment length. However the straifi in the epoxy-coated bar dropped off
more rapidly than those of the uncoated bar. The drop off is probably due to the
excessive slip in the area of the embedment length that is still transferring load. As
the bar slips at a faster rate. the resistance to load is relieved due to the lack of
friction and between the epoxy and the concrete and also the reduced effective
bearing area.
Effect of Epoxv Coating
Table 2 lists the ultimate bond strengths and the ultimate bond strength ratios for
aii the test groups. The results are illustrated in Figure 9. The ultimate bond strength
of epoxy-coated and uncoated No. 6 bars increased with concrete strength (6-14 ksi).
= Ob62949 0542876 9 4 T
The rate at which the bond strength increased with concrete strength was nearly the
same for uncoated and epoxy-coated No. 6 bars. The bond strength of epoxy-coated
and uncoated No. I l bars increased with concrete strength up to 12 ksi. Above 12 ksi
concrete strength, the uncoated No. 11 bars exhibited no significant change in bond
strength while the epoxy-coated bars exhibit a decrease in bond strength. The rate of
increase in bond strength with concrete strength was lower for epoxy-coated No. I 1
bars as compared to uncoated No. 1 I bars. The lower bound for ultirnute bond
strength of the epoxy-coated bars is represented by the highest bond ratio. The
highest bond ratio for all of the tests was 1.33. thus for this case the uncoated bars had
a bond strength I .33 times higher than the epoxy-coated bars. This occurred for test
Group 2, loo00 psi normal concrete with No. 11 bars.
The ultimate bond strength ratios ranged from 1.01 to 1.08 and 1.20 to 1.33 for
the No. 6 and No. i l bars, respectively. The average of the ultimate bond strength
ratios for all No. 6 bars was 1.04 and the average for ali No. i 1 bars was 1.26. The
bond ratios varied with concrete strength which will be discussed later.
Johnson and Zia (4) obtained bond ratios of 1.17 for both No. 6 and No. I 1
bars. However. their specimens were limited in number and contained transverse
reinforcement which produced a combined splitting and pullout failure. Treece and
Jirsa ( 5 ) obtained ratios ranging from I . 15 to 1.S2 with an overall average of 1.5 and
stated there was no variation for coating thickness, bar size. and concrete strength.
The diamond shape deformation pattern was used which tends to amphfy the effect of
epoxy coating thus giving larger ultimate bond strength ratios (poorer performance of
epoxy-coated bars). Hadje-Ghaffari et d.(9) obtained bond ratios ranging from I .O I
to I .25 and 1.01 to 1.37 for No. 6 and No. 11 bars, respectively. The overall average
was 1.12 for all No. 6 bars and 1.20 for all No. 11 bars. The N shape deformation
pattern, the same as used in this study, gave bond ratios of 1 .O8 for the No. 6 bars and
1.35 for the No. 1 I bars. These results are similar to those obtained in the current
study despite large differences in embedment lengths and concrete strengths.
Effect of Concrete Strength
The results of this study showed that the bond strength reduction due to epoxy
coating varied with concrete strength in some cases. Figure 10 shows the relationship
between the ultimate bond strength ratio and concrete strength for the No. 6 and No.
I l bars, respectively. The ultimate bond strength ratio, generally decreased with
increasing concrete strength for the No. 6 bars. The slight decreasing trend for the No.
6 bars was judged as insignificant. The No. 1 1 bars showed a more significant trend of
increased ultimate bond strength ratios with increasing concrete strength as shown in
Figure 10.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2q4q '¿lb Tribute to Peter Gergely 275
Effect of Bar Size
The most signiticant effect was on the reduction of bond strength due to the
epoxy coating. The average ultimate bond strength ratio for No. 11 bars was 21
percent greater than that of the No. 6 bars. In other words epoxy coating has a more
detrimental e k t on the bond performance of larger epoxy-coated bars relative to
uncoated bars. The limited data for the No. 8 bars without strain gages showed that
the ultimate bond strength ratio for the No. 8 bars fell between the ultimate bond
strength ratios of the No. 6 and No. 1I bars. The reason for the differences due to bar
size is not apparent. A possible explanation is the transverse strain relative to the bar
size. The effect of transverse strain is dependent on bar size and would tend to be
detrimental to the bond peiformance. For the same bar stress transverse strains will
increase with the diameter of the bar. Though small, the effect of transverse strain is
cumulative with the effect of epoxy coating. Theoretical transverse strains at the
nominal yield point reduce the nominal radius of the bar by 0.8, 1.0 and 1.5 mil for the
No. 6, No. 8, and No. 11 bars respectively. These reductions in radius are equivalent
to 6 to 30 percent of the allowable coating thicknesses for epoxy-coated bars (5-12
mil).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Effect of Micro-silica on the Ultimate Bond Strength Ratio
Group 6 specimens fabricated with 12000 psi concrete had only 5 percent micro-
silica and did not show any reduction in ultimate bond strength ratio compared with
the loo00 and 14ooo psi concrete with 10 percent micro-silica. The No. 1 I bars
showed a significant decrease in the ultimate bond strength ratio (uncoatedepoxy-
coated) when comparing micro-silica concrete to normal concrete. Figure 11 shows
the best fit lines through the data for No. 6 bars and data for No. 1 1 bars respectively,
for the ultimate bond strength ratios versus percentage rhicro-silica relationship. The
ultimate bond strength ratios were normalized with the OJB equation given in Eqn.
(3) to eliminate the effect of concrete cover. Overall, the limited data suggests that the
micro-silica does not affect the bond ratio significantly.
AC1 3 18 PROVISIONS
The AC1 318-89 equation is derived from the following equation for bond
stress given in AC1 318-63.
= Obb2949 0542878 712
The bar stress was conservatively assumed to achieve 125% of the yield stress
(fs=1.25f,,)and this expression for bond stress was substituted into Eqn. ( 2 ) to form
the basic development length equation given in AC1 3 18-89 (1 3):
The simplification of the basic development length equation to 0.04 in the AC1
318 code reduces the expected bar stress from 125% to 119% of yield when the
bonded length of the bar equals kdb.
For smail bar diameters, a puli-out failure rather than a splitting failure may
govern. AC1 3 18-89 specifies a minimum factored development length given in Eqn.
(6) to prevent a pull-out failure.
After re-examining this equation a more “user friendly” format has been
suggested for AC1 3 18-95.
-_ - 3fyaßr
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(7)
(7) 22.5
The ACI-95 implied bond stress can be derived using Eqn. (2) by substituting
for ld from the equation given above.
o! =1.3 (Top bar factor) for 12 in. concrete cast below the rebar
ß = 1.5 for epoxy-coated bars with less than 3 dbor clear spacing less than
6 d/,.
= 1.2 for other epoxy-coated bars
y = 0.8 (Reinforcement size factor) for No. 6 and smaller bars
c = Smaller of either the cover to the bar measured from center of the bar or
half the center-to-center spacing of the bars being developed.
A,,.= Cross-sectional area of transverse reinforcement within the spacing

s (in’)
s = Maximum spacing of transverse reinforcement within i,, (in.)
I I = number of bars being developed
Th; splice factor for normal splices (Class B) = 1.3
The comparisons of test results with AC1 318-89 and AC1 318-95 implied
bond stresses are given in Table 3. The 10 ksi limit for concrete strength was
ignored in the analysis. The modification factors a,ß and y (Eqn. 7) were applied.
The bar stresses were assumed to be equal to l.19fv. As mentioned earlier, the
simplification of the basic development length equation given in AC1 318-89 code
reduces the expected bar stress fiom 125%to 119%of yield. When the ratios given
in Table 3 are multiplied by 1.19, AC1 implied bond stresses to achieve yield stress in
the bars are obtained.
Observations from comparisons with AC1 formulae

1. For No. 6 bars, the AC1 318-89 and AC1 318-95 were conservative in
predicting the bond strength of both uncoated and epoxy-coated bars for the concrete
strengths tested (6-14 ksi).
2. For No. 11 bars, the AC1 3 18-89 was conservative for uncoated and epoxy-
coated bars for the concrete strengths tested (6-14 ksi),. but AC1 318-95 was
unconservative for some cases. The AC1 318-95 was unconservative for uncoated
bars Groups (2, 4-7). The AC1 318-95 was also unconservative for some epoxy-
coated No. 1 1 bars (Groups 2, 7) corresponding to concrete strengths of 10.47 and
13.98 ksi.
3. AC1 3 18-89was more conservative than AC1 3 18-95 for all of the groups.
Limited Concrete Strengths for Desip
As seen from the results, AC1 3 18-95 gives unconservative development lengths
for both uncoated and coated bars for concrete strengths greater than IO ksi. The
same factors for epoxy-coated bars should be maintained for concrete strengths
greater than 10 ksi, until further research is conducted. The applicability of AC1
equations to uncoated bars in high-strength concrete is covered in a companion paper.
COMPARISON WITH OTHER STUDIES
A total of 182 test results from the work at University of Minnesota and
seven other research studies conducted by Treece and Jirsa (3,Choi et al. (7),
Devries et al. (8), Hadje-GMari et al. (9), Cieary and Ramirez (14), Hester et al.
(15). Hamad and Jirsa (17) were compared to values of design bond strength
calculated using AC1 318-89 and AC1 318-95. In all cases, tests were conducted
on lapped joints or beam-end specimens of coated and uncoated bars to assess the
reduction in bond due to coating. A brief description of the research studies was
given earlier. A considerable variety of bar types and diameters, concrete
strengths, concrete covers and bars cast top and bottom are represented in the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m 0662949 0542880 370
various studies. These details are presented elsewhere (24). The 10 ksi limit for
concrete strength was ignored in the analysis. The modification factors a, ß and y
(Eqn. 7) were applied.
Figure 12, shows the ratio of the bond stresses measured from the tests relative
to those predicted by Eqns. ( 5 ) and (7) plotted with respect to concrete strength for
all the test results for coated bars. The mean and standard deviations are shown in
Table 4. As seen from Figure 12, AC1 318-95 is unconservative for some coated bars.
Generally AC1 318-89 was more conservative than AC1 318-95 for all the test results.
The AC1318 implied bond stress equations derived from Eqns. (5) and (7) are
unaffected by the embedment length or the extent of yielding, but, as given in Eqn.
( 2 ) , the experimental bond stress is a function of both. Although it is convenient for
design purposes to assume that the bond strength is proportional to development
length, several researchers (e.g. Orangun et al. (13), Eqn. (3)) have shown that the
average bond strength reduces with increasing development length. The experimental
bond lengths are selected to produce bond failures at stresses near the yield stress
without exceeding it. Therefore the extent of yielding in relation to the experimental
development length provides greater insight to how the bar actually behaved (25). It is
important. when evaluating the results, to investigate the relationship between the
level of stress in the bar (f:df\.) to the relative einbedment length (Ile5SJ/~c/.~~~).
The
specimens with a bar stress closer to yield give better predictions of the relative
conservatism with respect to AC1 3 18.
From Eqn. (2):
(Iresr I1AC1318-9 5 ) l4AC13I S -Y5
The v&)
versus (lres,,&-13~~) graph for coated bars is shown in Figure 13. The
data points lying above the line show that the AC1 318-95 development length
provides an adequate embedment length for developing the reinforcement, assuming
that the relationship between the bar stress and bond length is h e a r (Eqn. (8)). As
seen a number of points fall below the diagonal line for coated bars, indicating the
unconservative nature of the ACI318-95 equation. This was also found for the
uncoated bars. The AC1 318-95 implied bond stresses for uncoated bars in high-
strength concrete are compared in a companion paper submitted to AC1 Structural
Journal.
The uncoatedkoated bond ratios are plotted with respect to concrete

I
strength in Figure 14. The means and standard deviations for both bottom cast
and top cast bars were approximately 1.25 and 0.17. Although the AC1
specifications recognize the lesser effect of epoxy coating on top cast bars by limiting
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542öBL 207
the rnodikation factor for top cast bars to 1.7, it was not observed in these results.
The best-fit lines for both bottom cast and top cast bars showed a negative trend.
This c o n f i i the trend observed in the investigation at the University of
Minnesota for bottom cast bars. i.e. bond performance ratio remains constant in
some cases (No. 6 bars) and reduces with concrete strengths for some other cases
(No. 11 bars).
The bond ratio (uncoatedkoated) versus bar size graph is shown in Figure
15. For clarity the points indicating the results for top bars are shifted slightly to
the right. The mean and standard deviations are given in Table 5. These results
support the observations of this study. i.e. The bond ratio (uncoatedcoated) increases
with the bar size for bottom cast bars. Top cast bars did not show this effect.
Epoxv-Coated Modification Factor
Only the AC1 318 design recommendations for bottom cast bars can be
addressed by this investigation. In this investigation only specimens with concrete
cover less than 3dh were tested and thus oniy recommendations are made for that
case. For bar sizes No. 3 to No. I I , the current AC1 318 mod8cation factor for
epoxy-coated bars is 1.5 for concrete cover less than 3 bar diameters (3dh) or il clear
spacing between bars less than 6 bar diameters.
This investigation found lower bound modification factors for epoxy-coated bars
of 1.08, 1.10, and 1.33 for No. 6, No. 8, and No. 11 bars, respectively. From the
results of this study the modification factor for epoxy-coated bars could be reduced to
at least 1.33 for No. 9, No. IO, and No. 11 bars. No. 6. No. 7, and No. 8 bars could
have a modification factor of 1.10. The data on the No. 8 bars were limited and to
simp@ the design code a modification factor of 1.33 could be used for bar sizes No.
6 t h No. I 1. No recommendations could be made on other bar sizes outside the
sizes tested in this investigation. However results from other observations c o n f m d
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the conservativeness of the AC1 modification factors. The bars used in this study had a
relative rib area of approximately 0.09.The typical range for current production in the
USA is 0.06-0.09(26). A caution is stated that the modification factors recommended
above can oniy be used with bars having high relative rib areas ciose to 0.09.
However it must be noted that the 1.5 modification factor in the current codes
was derived from tests (Treece and Jirsa ( 5 ) ) done on bars having a diamond shaped
deformation pattern. Hadje-Ghafíari et al. (9) investigated other deformation patterns
(N, C and S ) which did not include the diamond shape deformation pattern. Hadje-
Ghaffari et ai. proposed a development length modification factor of 1.25 for No. 6
bars and smaller and a factor of 1.35 for No. 7 bars and larger. The moditication
factors in the current AC1 3 18 codes should remain unchanged as long as the diamond
= 0662947 0542882 143
deformation pattern is used and it is still common place not to speclSi the type of
deformation pattern in contract specifications. The modification factors proposed by
Hadje-Ghaffari et al. (9) and endorsed by this study could be used for the N, C and S
deformation patterns if specified in contract specilications.
CONCLUSIONS
The following sections state the conclusions ‘based on the evaluation of

experimental results. comparisons with design codes and results from other research
studies.
1. Epoxy coatings reduce bond performance because of decreased friction and
the reduced relative rib area of the bar. The bond performance prior to initial concrete
splitting is primarily affected by the reduced relative rib area which reduces bond
stiffness. After initiai concrete splitting, bond perfoimance is mainly affected by
decreased friction which cannot resist the wedge action of the bar, which reduces
bond strength and stiffness.
2. Epoxy-coated test bar specimens exhibit initiai bond cracks at lower loads
than uncoated test bar specimens. Reduced friction at the rib face with epoxy-coated
bars increases the radial tension stresses in the concrete around the bar causing the
bond cracks to form at lower bar loads.
3. Epoxy-coated bars have different straiii distributions than uncoated bars at
low bar stress and near bond failure. At low bar stresses the epoxy-coated bars
developed the same load as uncoated bars but over a much longer length. Near failure
the epoxy-coated bars showed less strain in the bar than the uncoated bars. The
epoxy-coated bars were unable to resist slip which caused the strain in the epoxy-
coated bar to be relieved.
4. Epoxy coating reduces the ultimate bond strength as compared to uncoated
bars for the concrete strengths tested (6 to 14 ksi).
5. The reduction in bond strength due to epoxy coating increased with bar size.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
On average. the ultimate bond strength ratios (uncoatedepoxy-coated) for all No. 6,
No. 8, and No. 11 bars were 1.04. I . IO. and 1.26, respectively.
6. For No. 6 bars. the ultimate bond strength ratio (uncoatedepoxy-coated) was
unaffected by concrete strength. For No. 1 1 bars. the ultunate bond strength ratio
increased as concrete strength increased.
7. Epoxy-coated bars slip more than uncoated bars. For epoxy-coated bars. the
load-slip behavior is affected mostly from the splitting of the concrete around the bar.
Load-slip behavior for uncoated bars seemed unaffected by initial bond cracks.
8. AC1 318-95 development length equation is unconservative for some coated
bars. Generally AC1 3 18-89 was inore conservative than AC1 3 18-95 for all the test
results.
m 0662947 0542883 08T
CONVERSION FACTORS
1 in. = 2 5 4 m
1 ksi = 6.895MF’a
REFERENCES
1. AC1 Committee 3 18: “Building Code Requirements for Reinforced

Concrete”, Detroit. American Concrete Institute, 1989.
2. AC1 Committee 3 18: “Building Code Requirements for Reinforced

Concrete”, Detroit. American Concrete Institute, 1995.
3. Standard Specifications for Highway Bridges. 14th Edition. Washington

D.C. AASHTO. 1989.
4. Johnston, D.W. and Zia, P., “Bond Characteristics of Epoxy-coated

Reinforcing Bars,” Washington, D.C. Federal Highway Administration,
Report No. FHWA-NC-82-002, 1982.
5. Treece, R. A. and Jirsa, J. O., “Bond Strength of Epoxy-Coated

Reinforcing Bars”, AC1 Mat. Journal. Vol. 86. No. 2. March-April 1989.
6. Grundhoffer, T, “Bond Behavior of Uncoated and Epoxy-Coated --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Reinforcement in Concrete”, M.S. Thesis, The University of Minnesota,

February 1992.
7. Choi, O.C., Hadje-Ghaffari, H., Darwin. D., and McCabe, S.L.: Bond
Epoxy-Coated Reinforcement to Concrete: Bar Parameters. SL Report 90-
1. University of Kansas Center for Research. January 1990.
8. DeVries, R.A., Moehle, J.P., and Hester, W.: Lap Splice Strength of Plain
and Epoxy-Coated Reinforcements - An Experimental Study Considering
Concrete Strength, Casting Position, and Anti-Bleeding Additives.
Berkeley. University of California. Report UCB/SEMM-9 1/02. January
1991.
9. Hadje-Ghaffari, H., Darwin, D.. and McCabe, S.L.: Effects of Epoxy-

Coating on the Bond of Reinforcing Steel to Concrete. Lawrence. The
University of Kansas Center for Research. SM Report No. 28. July 1991.
0662949 0542884 Tlib I
10. Gjorv, O.E.. Monteiro, P.J.M., Mehta, P.K.: Effect of Condensed Silica
Fume on the Steel-Concrete Bond. Trondheim. Norwegian Institute of
Technology. BML Report 86.201, May 1986.
11. Hwang, S., Lee, Y., and Lee, C., “Effect of Silica Fume on the Splice
Strength of Deformed Bars in High-performance Concrete”, AC1 Structural
Journal, Vol. 91, No. 3, 1994, pp. 294-302.
12. Clifton, J.R.. Beeghly. H.F.. and Mathey, R.G., “Nonmetallic Coatings for
Concrete Reinforcing Bars”. Building Science Series, No. 65, National
Bureau of Standards, Washington, 1975.
13. Orangun, C. O., Jirsa, J. O., and Breen, J. E., “A Reevaluation of Test Data
on Development Length and Splices”, Journal of the American Concrete
Institute. Vol. 74. No. 3. March 1977.
14. Cleary. D.B., and Ramirez. J.A., “Bond Strength of Epoxy-coated

Reinforcement”, AC1 Materials Journal, Vol. 88, No.2, 1991, pp. 146-149.
15. Hester, C.J., Salamizavaregh, S. Darwin. D., and McCabe, S.L., “Bond of
Epoxy-coated Reinforcement to Concrete: Splices”, SL Report 9 1- I.
University of Kansas Center for Research, 1991.
16. Cusens, A.R., and Yu, Z., ‘‘ Bond Strength and Flexural Behaviour of RC
beams with Epoxy-coated Reinforcing Bars”, The Structural Engineer, Vo.
71, No. 7, April 1993, pp. 117-124.
17. Hamad, B.S.. and Jirsa, J.O., “Strength of Epoxy-coated Reinforcing Bar
Splices Confined with Transverse Reinforcement”, AC1 Structures Journal.
Vol. 90, No. 1. January 1993. pp. 77-88.
18. Cairns, J., and Abdullah, R., “Ultimate Strength of Lapped Joints of Epoxy-
Coated Reinforcement”, Proceedings Institution of Civil Engineers,
Structures and Buildings, Vol. 104, Nov. 1994, pp. 391-400.
19. BS 81 10, British Standards Institution, “Structural Use of Concrete”, 1985.
20. ASTM: Standard Specification for Deformed and Plain Biller-Steel Bars
for Concrete ReinforcementíASTM A 6 15-87a). Philadelphia. American
I
Society for Testing and Materials. 1989.
2i . ASTM: Standard Specification for Epoxy-Coated Reinforcing Steel Bars

(ASTM A 775/A 775M-883). Philadelphia. American Society for Testing
and Materials. 1989.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542885 952
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
22. Mains R.M., “Measuremnt of the Distribution of Tensile and Bond Stresses
along Reinforcing Bars,” ACI Journal, Proceedings V. 48, No. 9, Nov:1951,
pp. 225-252.
23. Martm H. and Noakowski, P.: Verbundverhalten von Betonstählen,

Untersuchungen auf der Grundiage von Ausziehversuchen, Shriftenrehe auf der
Deutscher Ausschuss f-r Shahibeton, Heft 319, p. 239,1981.
24. Mendis. P.A., and French, C.W., “Bond Strength of Epoxy-Coated Bars”,
Research Report, October, 1996, Dept. of Civil & Environmental
Engineering, University of Melbourne.
25. Fasching, C., “Flexural Behaviour and Bond Strength of High-strength

Concrete Members”, M.S. Thesis. The University of Minnesota, 1995.
26. Tan, C., Darwin, D., Tholen, M. and Zuo, J., “Splice-strength of Epoxy-
coated High relative rib area Bars”, SL Report 96-2, The University of
Kansas Center for Research. 1996.
ACKNOWLEDGMENTS
The experimental investigation reported in this paper funded by grants form

the University of Minnesota Center for Transportation Studies and the National
Science Foundation of USA. The views expressed herein do not necessarily
reflect the views of the sponsor.
ü b 6 2 9 4 9 054288b 899
TABLE 1-TEST VARIABLES
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Qbb2949 0542887 725 m
TABLE 2-EXPERIMENTAL RESULTS
Group Bar size Concrete Bond Bond Bond ratio
lli,
Strength Strength Strength
(ksi) (Psi) (psi)
6.65 1208 1147 1.O5
6.36 822 664 1.24
9.7 1 1354 1256 1.O5
10.47 745 559 1.33
i<
9.29 1499 1412 1.O6
8.40 930 1 774 I 1.20
9.29 1358 I 1258 I 1.08
9.29 756 I641 I 1.24
11 9.16 1501 I 1439 I 1.04
9.16 804 656 1.22
12.03 1597 1555 1.O3
yg
12.03 849 1642 I 1.26
13.98 1625 I 1602 I 1.01
13.98 847 642 1.32
5.63 977 887 1.10
5.63 922 900 1.o2
imen)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ûbb2747 0542888 b b l
TABLE 3-COMPARISONS WITH AC1 BOND STRENGTHS
5 6 9.16 1.58 2.27 1.26 1.82

II 9.16 1.25 1.53 0.94 1.16
6 6 12.03 1.47 2. i4 1.17 1.72
11 12.03 1.15 1.38 0.86 1 .O3
7 6 13.98 1.39 2.05 I .40 1.64

II 13.98 1.07 1.21 0.80 0.92
8 8 5.63 1.35 1.83 1.31 1.79
8 5.63 1.27 I .86 1.24 1.81
Mean 6 1.47 2.1 1 1.22 1.69
11 1.31 IS O 0.95 1.14
Std.Dev 6 0.08 0.1 1 o. I O 0.09
II 0.30 0.28 0.18 0.22
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0542889 5T8
TABLE 4-MEANS AND STANDARD DEVIATIONS
AC1 3 18-95 1 AC1 318-89

Uncoated I Coated I Uncoated 1 Coated
Mean I 2.04 1 2.4 I 2.49 2.89
TABLE 5-MEANS AND STANDARD DEVIATIONS
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Bond Mechanisms: Adhesion, Friction, Bearing
Fig. 1-Bond mechanisms for deformed bars in concrete
0662949 0542890 21T
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Bearing torce
Friciion
Splitting
force
Ikliid
Fig. 2-Forces on bar lung, with and without friction (after (Il))
RIB HEIGHT EFFECTIVE

RIB HEIGHT
\EPOXY COATING
(THICKNESS 5-12 m i l )
--t
Fig. 3-Reduced rib height with epoxy coatings
m Obb2949 0542893 15b m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542892 O92
r - HYDRAULIC ACTUATOR
II II II II
II Il II II
Y W Y
STRONG CONCRETE REACTION

FLOOR BLOCK
Fig. 5-Testing frame (side view)

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
5 MIL O.2 "
EPOXY
o . 190
Fig. 6-Cross-section of glued strain gage bar
W 0662949 0542893 T29 W
30 ,- I
n
.-
Y
Q
W
n
a
9
LOADED END SLIP (inch)
30
25
20
.-a
n
-
Y
n
<
15
2 10
5 -UNCOATED
0.000 0.002 0.004 0.006 0.008

FREE END SLIP (inch)
Fig. 7a-Load-slip curves for Group 4, No. 6 bars
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
Obb2949 0542894 ï b 5
1 O0
80
- a
.-
Y
60
v
n
Q 40
O
-1
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
20
-
n
0.000 0.005 0.010 0.015 0.020 0.025 0.030
LOADED END SLIP ( i n c h )
1 O0
80
-_.-
60
40
20 UNCOATED
l , , # l # 8 ,
O
0.000 0.005 0.010 0.015 0.020
FREE END SLIP ( i n c h )
Fig. 7b-Load-slip curves for Group 4, No. 11 bars
Obb2949 0542895 8T.L U
450
500 h
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
POSITION (inch)
Fig. 8b-Strain distributions-at failure

2400
2200 r I
O SPECIMEN A
V SPECIMEN B
_---- EPOXY
UNCOATED
z2
I
1400
+
1200
O
E
2 1000
"i
I
200
01 I I I I I I I I I I I I
w -...
d
O 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4
POSITION (inch)
Fig la-Strain distributions-stress level= 1O ksi
W Obb29Y9 0542896 738 =
1 2 3 4 5 6 7 8
Group No.
Fig. 9-Experimental results
O 5 10 15
Concrete Strength (ksi)
Fig. 1 O-Bond ratio versus concrete strength
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542897 b74 D
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
.-O
CI
2 1.4
i
ir.
- .
- .
s
U
8
P 1.3
U
8
a
o
s!
Q
1.2
1.1 1. - No. 11 Bars
-+-
_ . _ _ _ _ _ . . _ . _ -. - .- .- -
I
0 1
æ
O 5 lo
%Micro Silica
Fig. 1 1-Bond ratio versus percentage micro-silica
4.5
3.5
3
4 A C I 318-95 Coated
2.5
1.5
1
0.5
O
0.0 5.0 10.0 15.0 20.0
Fig. 12-Testlpredicted bond ratio vs. concrete strength
~~
W Obb2949 0542898 500 W

OY
O 0.2 0.4 0.6 0.8 1
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Bond lengthlAC1318-95 bond length
Fig. 13-Extent of yield vs. bond length ratio for coated bars
O 5 10 15 20
Fig. 14-Bond performance ratios for top and bottom cast bars
m 0662949 0542899 447
2
A 1
o Top cast
O 5 10 15
Bar Size (#)
Fig. 15-Bond performance ratios for different bar sizes
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542900 T99
SP 180-13
Overview of Research to Improve the

Development Characteristics of Reinforcing Bars
by D. Darwin, J. Zu0 and M. L. Tholen
-
: The results of the first major reevaluation of reinforcing bar geometry in
the United States in nearly 50 years is described. The study involves experhental
and analytical efforts designed to broaden the understanding of factors that control
bond strength, improve the development characteristics of reinforcing bars, and
develop Practical design expressions that more accurately represent development
and splice strength than existing expressions. The research has established that
deformation pattern has little effect on the bond strength of uncoated bars that are
not confined by transverse reinforcement. Deformation pattern, however, as
represented by the relative rib area, does have a major effect on the bond strength of
bars that are confined by transverse reinforcement. increases in relative rib area,
obtained with either higher ribs, closer ribs, or a combination thereof, result in an
improved bond strength for confined bars. The study has also established limits on
how closely ribs can be placed without resulting in a pullout failm. High relative
rib area bars provide a reduction in developmentísplice length of 20 percent for all
coated bars, independent of the presence or absence of confining transverse rein-
forcement. Based on the experimental work, expressions are developed that
accurately characterize developmentísplice strength. in the development of the
expressions, FC1/4is shown to be superior to fC1/2 for characterizing the contribu-
tion of concrete strength to bond. The resulting design expressions are accurate for
compressive strengths between 2500 and 16,000 psi (17 and 110 m a ) . The most
accurate representation of the effect of transverse reinforcement on bond strength
includes parameters that account for the number of transverse reinfoKing bars that
cross the developedlspliced bar, the area of the transverse reinforcement, the
number of bars developed or spliced at one location and the relative rib area, and
size of the developedíspliced bar. The yield strength of the transverse reinforce-
ment does not play a measurable role. Practical comparisons illustrate reductions in
splice lengths of 20 percent for conventional bars and 30 percent for high relative
rib area bars compared to current requirements in AC1 318-95.
Keywords: 8ond strength; cracking (fracturing); deformation; reinforcing steels;

tensile strength; yield strength
299
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
II Obb29L}9 0542901 9 2 5 m
300 Darwin, Zu0 and Tholen
David Darwin, FACI, is the Deane E. Ackers Professor of Civil Engineering and
Director of the Structural Engineering and Materials Laboratory at the University of
Kansas. He is chairman of the TAC Technology Transfer Committee and a mem-
ber of AC1 Committees 224, Cracking 408,Bond and Development of Reinforce-
ment, 446, Fracture Mechanics, and ACI-ASCE Committees 445, Shear and
Torsion, and 447, Finite Element Analysis of Reinforced Concrete Structures.
AC1 member Jun Zu0 is a graduate research assistant in civil engineering at the
University of Kansas. He obtained his BS in architectural engineering from Tongji
University in China and his MS in civil engineering from the University of Kansas.
He is pursuing the Ph.D. degree.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
AC1 member Michael L. Tholen is a structural engineer with Burns &

McDonneii International in Kansas City, Missouri. He holds a BS in architectural
engineering and an MS and Ph.D. in civil engineering from the University of
Kansas.
INTRODUCTION
To say that, historically, structural designers and reinforcing bar manufac-

turers have approached changes in reinforcing bar geometry in a conservative
manner would be the height of understatement. In 1877,Thaddeus Hyatt demon-
strated the superior performance provided by deformed reinforcement (1). in 1884,
when Ernest Ransome introduced his square twisted bar, the first deformed bar
patented in the U.S., he had problems convincing the technical community of the
value of the product (2).Not only did the twisted bars provide improved bond, but
they exhibited higher tensile strength than smooth bars. Accused of substituting
higher strength steel, Ransome overcame his critics by submitting samples in which
, only one-half of each bar was twisted.
After the turn of the century, there was impetus to increase the strength of
the steel used for reinforcing bacs. This ran into resistance in some camps, particu-
larly in France, because of concern that the use of higher strength steel [yield
strength = 50 ksi (345MPa) versus 30 ksi (207 MPa) ] would correspondingly
result in higher tensile strains in the concrete, which would result in the formation
of cruch that would expose the steel to corrosion (3). Concern with the potential
for splitting cracks resulted in a strong preference by European engineers for
smooth rather than deformed reinforcement (4).
By the end of the first decade of this century, many different deformation
patterns had been developed, few of which are still in use today. In the first
systematic study of bond in the US., Duff Abrams compared smooth and de-
formed bars using pullout and beam specimens (5). Abrams observed that the ratio
of the bearing area of the projections (measured perpendicular to the bar axis) to the
full exterior surface area of a bar over the same length could be used as a criterion
for evaluating the bond resistance of deformed bacs. He recommended that the ratio
not be less than 0.2, which would have resulted in much closer spacings and/or
higher ribs than in use at the time or today, for that matter. In spite of his recom-
mendations, few changes were made, and the early patterns, most of them propri-
W Obb2949 05112702 8 b L - m
etary. continued in use in the U.S. until the late 1940% when current ASTM
requirements for reinforcing bars were estabiished based on the combined efforts of
Arthur P. Clark (6, 7), the American Iron and Steel Institute Committee on Rein-
forced Concrete Construction, and American Concrete Institute Committee 208,
Bond Stresses (8).
The new requirements, first embodied in the “Tentative Specifications for
Minimum Requirements for Deformations of Deformed Steel Bars for Concrete
Reinfomment,” ASTM A 305-47T (9), are now incorporaîed in all ASTM specifi-
cations covering reinforcing bars. The r e q k m e n t s include a minimum height and
maximum spacing for deformations, along with limitations on the included angle of
the deformation with the longitudinal axis of the bar and limitations on the gap
between ends of the deformations.
At the time of his research, Clark not only provided criteria for rib spacing
and rib height, he also recommended thaî the ratio of the shearing area (bar perime-
ter times length between ribs) to rib-bearing area be limited to a maximum of 10
and, if possible, 5 or 6. Today this criterion is usually described in terms of the
inverse ratio, that is, the ratio of the bearing area to the shearing area, which is
known as the relative rib area, R,. Clark‘s recommendations then become a mini-
mum relative rib area, R,, of 0.10, with desirable values of 0.20 or 0.17 (not so
different from Abrams’ recommendations). These values can be compared to
typical values for bars manufactured in the U.S.,which range from 0.057 to 0.084
(10). Clark’s latter recommendations were not adopted because of a desire on the
part of the reinforcing bar producers to remove only the weakest patterns, rather
than establish the best possible anchorage to concrete (1 1).
Since the completion of Clark’s work, much effort has been expended to
gain an understanding of bond and development in reinforced concrete strucms.
However, the vast majority of that effort has involved the use of standard reinforc-
ing bars. This paper provides an overview of the first large-scale study to improve
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the development characteristics of reinforcing bars in the United States since 1949.
The goals of the study, under way since 1991, have been to broaden the under-
standing of the factors that control bond strength, improve the development charac-
teristics of the reinforcing bars themselves, and develop practical design expres-
sions that more accurately represent development and splice strength than current
expressions.
SCOPE OF CURRENT STUDY
The study has been broad ranging and has involved substantial experimental
and anaiytical effort, carried out in a number of overlapping phases. The experi-
mental studies began with the development of a beam-end specimen in to what is
now the ASTM standard specimen for evaluating the relative bond strength for
reinforcing bars [ASTM A 944 (12)]. To compare the effects of rib height, rib
spacing, and relative rib area on the load-slip behavior and bond strength of rein-
forcing bars, tests were run using both machined and conventionai reinforcing bars
with and without laterai confining reinforcement (13,14). The effects of bar size
and clear spacing between the ribs (15) as well as face angle and epoxy coating
were also evaluaîed (16). In conjunction with an evaluation of the combined effects
of epoxy mating and deformation pattern, tests were run to determine the coeffi-
cient of fiction between both uncoated and epoxycoated minfoxcement and con-
m 0662749 0542903 7Tö
Crete (16). The results of these tests proved important in interpreting the experi-
mental results.
The major experimental effort involved the construction and testing of 229
full-scale splice specimens. Production runs of high relative rib area bars, designed
based on the guidelines obtained from the beam-end tests, provided bars with R,
ranging from 0.10 to 0.14. The splice tests included both conventional and high
relative rib area No. 5, No. 8, and No. 11 bars, with and without transverse
reinforcement and with and without epoxy coating. Coarse aggregate type and
quantity were varied to determine the effect on bond strength. The negative impact
of epoxy coating on bond was evaluated based on relative rib area, rib face angle,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
degree of confinement provided by concrete and transverse steel, and concrete
strength. High relative rib area bars were evaluated for the top-bar effect, while the
impact of unsymmetrical splice placement within a cross section was studied using
both high relative rib area and conventional bars. Ultra-high relative rib area bars,
R, = 0.28, were tested for bond with, and without confining transverse reinforce-
ment. Concrete strengths used in the tests ranged from 3800 to 14,400 psi.
Additional areas of study include the crack density and crack width obtained with
conventional and high relative rib area bars, the strength of splices in regions
subjected to shear, and the strength of splices when placed in multiple layers. in
response to concern that improved bond strength might limit flexural ductility at
ultimate, moment-rotation tests were carried out to compare the flexural properties
of beams cast with conventional and high relative rib area bars. The low slip
exhibited in the beam-end tests suggests improved bond behavior under cyclic
loading. Such testing is now underway.
The analytical work has consisted of a detailed analysis of bond strength as
a function of developmenthplice length, bar size, concrete strength, beam geome-
try, transverse reinforcement, and epoxy coating (17, 18). A reliability-based
strength reduction (@)factor was developed in the process of formulating develop-
menthplice design expressions (19). Designs using the new expressions for both
conventional and high relative rib area bars have been compared with past and
current AC1 and AASHTO design criteria (U), and savings in splice length and
total percent of steel have been evaluated for specific design projects. As part of
1 ongoing work to improve the understanding of bond and development, work has
also continued on the development of realistic fracture-based finite element bond
models with the goal of eventually developing a rational, rather than experimental,
basis for development and splice design (15,20,21).
This paper provides a summary of the key findings of the study as of
January 1997. Work is ongoing.
EXPERIMENTAL WORK
Beam-End Tests
Tests carried out using beam-end specimens were used to gain a basic
understanding of the effects of rib geometry on bond strength under varying condi-
tions of confinement and bar surface condition. The specimen used is a modified
version of beam-end specimens used in prior work at the University of Kansas (22-
24) and elsewhere (25). The specimen, shown in Fig. 1, was adopted in 1995 as
- 0bb2947 0542704 b34 9
the ASTM standard for evaluating the relative bond strength of reinforcement
[ASTM A 944 (is)]. The specimen was adopted because it provides a state of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stress in which both the steel and the sum>unding concrete ate in tension at the time
of bond failure, in contrast to puilout specimens, which not only place the conmte
in compression while the steel is in tension, but place the surface of the bar in
compression during the test.
The initial experimental efforts in this study were aimed at determining the
effects of rib height and spacing on bond strength with and without confining
transverse reinforcement (13, 14). Using values of R, between 0.05 and 0.20, the
study established that the bond force-slip relationship of reinforcement is a function
of the relative rib area, independent of the specific values of rib height or rib spac-
ing. The initial stiffriess of the bond force-slip curves increased with an increase in
relative rib atea, independent of the d e m of transverse reinforcement. As shown
in Fig, 2, under conditions of low confinement (no transverse reinforcement), bond
strength is independent of deformation pattern. With increasing confínement
provided by transverse Ginforcement, however, not only does bond strength
increase (as expected), but the magnitude of the increase depends on the value of Ri
(Fig. 2). Additional work, reported by Tholen and Darwin (15) using bars with
relative rib areas up to 0.28, shows that the contribution of transverse reinforcement
to bond strength continues for even higher relative rib areas. The results of the
study shown in Fig. 2 provided the basis for developing guidelines for the produc-
tion bars with relative rib areas between 0.10 and 0.14 that were used in the full-
scale splice tests.
As the splice tests got underway, additional studies using the beam-end
specimens were carried out to develop additional guidelines on deformation proper-
ties (15). As illustrated in Fig. 2, it is the relative rib area,not the precise combina-
tion of rib height or rib spacing, that governs the contribution to bond strength
provided by irarisverse reinforcement. Thus, higher bond strengths can be obtained
using higher ribs or closer ribs or a combination thereof. The diffculty with using
ribs that are closely spaced is that the concrete between the ribs becomes progres-
sively narrower and may, for very close spacings, result in a concrete shear (puil-
out) failure around the periphery of the bar at bar forces below that which would
cause a typical longitudinal splitting failure. To develop guidelines, beam-end
specimens containing bars with relative rib areas of O. 10 and 0.20 were tested both
with and without transversereinforcement for rib widWrib center-tocenterspacing
ratios ranging from 0.20 to 0.75. The results of those tests indicate a sharp drop-
off in bond strength for rib widWcenter-to-center spacing ratios 2 0.60 for bars not
confined by transverse reinforcement and 2 0.45 for bars confined by transverse
reinforcement. These findings lead to a recommendation that the rib WiWcenter-
úxenter spacing ratio should not exceed 0.45. Splice tests using production high
relative rib area bars with different rib widtldcenter-tocenter spacing ratios support
this conclusion. In the current study, the bars in both the beam-end tests,which
were machined, and the bars in the splice tests, which were rolled, have flat upper
surfaces on the ribs, and thus rib width is taken as the width of the upper surface.
For bars with rounded ribs, a realistic but conservative representation for rib width
WillbereqM.
To evaluate the effect of rib face angie, the angle between the longitudinal
axis of the bar and the face of a deformation, on the strength of epoxycoated bars,
matched sets of coated and uncoated reinforcing bars were tested with rib face
angies of 30, 40, 45,60, and 90degrees (16). This group of specimens included
both machineù and d i e d bars. In support of this study, tests were made to deter-
Ob62949 0542905 570
mine the coefficient of friction between the surface of rolled reinforcing bars and
concrete in both the epoxy-coated and uncoated conditions. Average coefficients of
friction of 0.49 and 0.56 were obtained for the epoxy-coated and uncoated bars,
respectively. The results of the tests are illustrated in Fig. 3, along with a theoreti-
cal relationship developed by Hadje-Ghaff~et al. (26),which includes the results
of studies by Rehm (27,28) and Lutz and Gergely (29) who observed that, for face
angles greater than 40 degrees, the concrete in front of bars crushes at bond failure,
producing ribs with effective face angles between 30 and 40 degrees. In Fig. 3, a
value of 40" is adopted. Since it is difficult to roll bars with face angles steeper than
45 degrees, it is worth noting that most practical cases are represented by the data
between 30 and 45 degrees, a region in which high variability in the effect of epoxy
coating on bond strength is expected.
SDlice Tests
In the course of the study, 229 splice specimens have been cast and tested.
A schematic of the test setup is shown in Fig. 4. Specimens containing either
uncoated or epoxy-coated reinforcement were tested both with and without confin-
ing transverse reinforcement. In typical specimens, the bars were spliced in the
central, constant moment region, of the beam, as shown in Fig. 4. In some speci-
mens, the loading arrangement was changed slightly and the bars were tested
outside of the constant moment region to evaluate the bond strength of bars in
regions subjected to shear and, therefore, changing tensile force in the bars.
Concrete strengths ranged from 3,800 to 14,400 psi, and concretes were manufac-
tured with various amounts of limestone and basalt coarse aggregates to evaluate the
effect of concrete properties on bond strength. In most specimens, the splices were
bottom cast; however, in some specimens, top-cast reinforcement was used. The
number and placement of the splices within the cross section were varied to obtain
data on beams with different values of bottom cover, side cover, and bar clear
spacing. In addition to rolled bars with relative rib areas ranging from 0.065 to
O . 14, threaded high-strength prestressing bars with a relative rib area of 0.28 were
evaluated. Specimens with laterally, unsymmetrically placed splices were tested to
determine the effect of bar misplacement or misalignment during construction, often
required because beam bars do not easily pass through column bar positions. One
set of specimens with multiple layers of reinforcement was also tested. A large
number of the specimens were instrumented to measure the crack width at the end
of the splices, and crack density and crack widths during the progression of the
tests were also monitored.
The results for symmetrically placed bottom-cast splices, reported in refer-
ences 15, 16, 30, and 31, reinforce the observations in the beam-end tests and
provide realistic data for application in design. The tests show that, for bars not
confined by transverse reinforcement, splice strength is independent of deformation
pattern. However, when the splices are confined, the increase in splice strength
resulting from a given amount of transverse reinforcement increases as both the
relative rib area and the bar diameter increase. Increases in aggregate strength also
result in an increase in the bond strength of bars confined by transverse reinforce-
ment. An analysis of the test results indicates that the use of high relative rib area
bars (R, 2 O. 12) could result in as much as a 26 percent decrease in splice length.
The savings are highest for low values of cover and clear spacing and high values
of confining steel. The savings decrease with increasing values of cover and
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D ObbZ949 0542906 407 D
decreasing values of confining steel. For high total confinement provided by

concrete and/or transverse steel, the nature of bond failure, normally longitudinal
splitting, wiü switch to puiiout, negating any additional strength provided by added
confimement. This behavior is accountedfor in the proposed design expressions.
For epoxycoated bars (16,30,31), although many of the tests are clustered
in the region of high sensitivity to face angle, as shown in Fig. 3, the overail trend
shows that bond strength increases as the relative rib area of the bar is increased, as
demonstrated in Fig. 5. An analysis of the data shows that the use of bars with
relative rib areasof 0.10 and above will allow a reduction in the maximumdevelop-
ment length modifcation factor for epoxycoated bars from 1S, as requiredby ACI
318-95 (32) and the 1996 AASHTO Bridge Specifications (33). to 1.2, a 20
percent savings. This reduction is independent of the presence or absence of
transverse reinforcement.
To summarize the other principal observations obtained from the splice
tests: The top-bar effect for high relative rib area bars is approximately the same as
that obtained for conventional bars. For ultrahigh relative rib area bars, R, = 0.28,
an extrapolation of the improved bond strength provided for confiied bars with R,
up to 0.14 provides a good estirnate of the added strength of the ultrahigh R, bars.
The threaded bars with R, = 0.28 have an added advantage; if the ribs are allowed
to interlock, they provide an even greater bond strength. The strength of specimens
in which the splices are placed unsymmetrically across the beam width is nearly
identical to that of specimens in which the splices are placed symmetrically, indicat-
ing that the average clear spacing may safely be used in place of the minimumclear
spacing for splice design.
The strength of splices placed in regions subjected to shear is uniformiy
greater than or equal to the strength of splices in regions of constant moment.
When bars are placed in multiple layers, strength is nearly identical, whether both
layers are spliced or only a single layer is spliced. The strength exhibited in this
pair of tests, in which the splices were widely spaced across the width but had 1 in.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
clear between the layers, exceeded the predicted strength based on the minimum
clear spacing but was below the predicted strength based on the spacing across the
beam width. Crack density and crack width are nearly identical for conventional
and high relative rib area bars, although specimens containing epoxycoated bars
exhibit fewer and wider cracks than obtained for specimens containing uncoated
bars. In the most recent tests, specimens containing high-strength concrete appear
to benefit more from the presence of transverse reinforcement than do n o d -
strength concrete specimens.
Moment-Rotation Tests
Based on concern that an improved bond strength would reduce the length
of flexural reinfomment that would yield at a plastic hinge and thus reduce the
,,phSitic rotation capacity of the comte (34), two pairs of beams were evaluated to
determine the effect of relative rib area on member ductiiity in bending (15). Each
pair consisted of matching specimens, differing qnly in the relative rib area of the
bars. Reinforcement ratios equal to 43 and 68 percent of the balanced reinforce-
ment ratio were used. The tests resulted in nearly identical load-deflection and
moment-rotation curves for the matched beams and nearly identical crack patterns,
indicating that reduced slip obtained by high relative rib area bars does not detri-
mentally affect the flexural ductility of beams.
ANALYTICAL STUDIES
Splice Streneth and Desim Eauations
The first 83 splice tests in the current study were combined with 216 splice
and development tests from earlier studies to characterize the developmenthplice
strength of bars both with and without transverse reinforcement as affected by
developmentkplicelength, bar size, placement geometry, and amount of transverse
reinforcement. The resulting equation (18) is
-
[63 ld(cm + 0.5 db) + 2130 Ab]
Tb Abfs
=
f’l/4 f’l/4
C C
(1)
+ 2226 t,td -
n
+ 66
in which Tb = force in bar at development or splice failure, in lb

Ab = bar area, in in2
db = nominal bar diameter, in in.
fS = steel stress at failure, in psi
fc = concrete compressive strength, in psi; fC1/4in psi
b = development or splice length, in in.
cm,CM = minimum or maximum value of c, or cb (CM/C,I 3.5), in in.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
CS = min (c,i + 0.25 in., eso), in in.

csi = one-half of clear spacing between bars, in in.
c,,, Cb = side cover or bottom cover of reinforcing bars, in in.
N = number of transverse reinforcing bars (stirrup or ties) crossing
h
Am = area of transverse reinforcement crossing the potential plane of
splitting adjacent to the reinforcement being developed, in in2
n = number of bars being developed or spliced along the plane of
splitting
tr = 9.6 R, + 0.28
= 0.72 db + 0.28
R, = ratio of projected rib area normal to bar axis to the product of
the nominal bar perimeter and the center-to-center rib spacing
The final term in Eq. 1, 66, is used only if the member has confining
transverse reinforcement.
Eq. 1 was combined with a reliability-based strength reduction factor of 0.9
(which can be embedded in the equations) to obtain design equations with various
levels of complexity (18, 19).
m 0 b b 2 î q ï 0542908 28T m
I
(3)
72 (-
+dbKir)
in which C = (Cm + 0.5 db)(o.l cM/cm + 0.9) for Eq. (2) and = (Cm + 0.5 db) for Eq.
(3), & = 35.3 t&&/sn, and (C + &)/db 4.0. The upper limit on (C + &)/db is
needed to prevent a pullout failure (18).
Key observations made in the development of Eq. (1) include the following:
bond strength is linear with an inmase in development length but not proportional
to development length [matching earlier studies( 17,35)]; thus, a given percentage
increase in bond strength requires a greater percentage increase in development
/splice length. f c l Q does not provide an accurate representation of the Contribution
of concrete strength to bond strength. Bond strength is underestimated for low
strength concrete and overestimated for high strength concrete. f$4 provides a
good representation for concrete strengths between 2,500 and 16,000 psi. The
latter two points are illustrated in Figs. 6 and 7. Fig. 6 shows a comparison be-
tween test results and predicted development/splice strengths for 133 bars without
confining transverse reinforcement in which the contribution of the concrete
strength is characterized based on f C W The figure shows an evaluation based on
categories of concrete strength. For these categories, from below 3000 psi (21
MPa) to over 10,ooO psi (69 MPa), the intercepts on the vertical axis decrease as
the concrete strength increases. The line representing concrete with compressive
strengths above 10,ooO psi (69 m a ) is significantiy below the rest of the data.
f'$2, however, does give a good representation for concrete strengths between
4500 and 7500 psi (31 and 52 MPa). Outside of this range, PJ/2 does not give a
good representation. Fig. 7 shows a similar comparison based on Eq. (i), in
which the concrete contribution is characterized by f'$4. The best-fit line for ail
categories of concrete strength nearly coincide, with the exception of specimens
with concrete strengths in excess of 10,ooO psi (69 m a ) , largely the result of two
relatively low splice strengths that dominate the results for this category. if these
two tests are removed, all strength categories produce nearly coincident best-fit
lines (18).
The most accurate of several representations (18) for the contribution of
transverse reinfomement to bond strength [used in Eq. (111 includes the number of
transverse bars crossing the main bar, the area of the transverse bars, the number of
bars spliced or developed at one location, and the relative rib area and bar diameter
of the spliced bar. The yield sîrength of the transverse reinforcement does not play
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 054ZîOî 116
a role (18). Based on strength criteria, which include the reliability-based $-factor,
a single value may be used for both for development and splice length. The obser-
vation that clear spacing between bars is effectively greater in comparison to equiva-
lent amounts of side cover or bottom cover (17, 18) is included in the design
expression by adding an additional 0.25 in. to csj, as indicated following Eq. (1).
Work is continuing based on additional test results both in the current study and
elsewhere (36). An improved version of Eq. (1) is now being formulated.
Practical Comparisons
To obtain a measure of the savings in splice length, reduction in congestion,

and savings in total steel obtainable with both high relative rib area bars and the new
design expressions [Eq. (2) and (3)], comparisons were made for a bridge struc-
ture, containing both coated and uncoated reinforcement, and a 12-story office
building containing uncoated reinforcement (15). For the bridge, all of the
uncoated reinforcement was either not confined by transverse reinforcement or so
heavily confined that no savings were obtained from using high relative rib area
bars. However, the application of Eq. (2) and (3) resulted in significantly shorter
splice lengths, averaging 67 and 69 percent, respectively, of the values obtained
with AC1 3 18-95 (32). For the coated reinforcement, the splice lengths for conven-
tional bars averaged 78 percent of the values required by AC1 318-95, and the
splice lengths of high relative rib area bars averaged 65 percent of the AC1 values,
or about 82 percent of splice lengths attained with conventional bars using Eq. (2)
or (3). Total savings on steel amounted to 0.93 percent for conventional and 1.35
percent for high RI. bars, compared to steel obtained using the AC1 Code.
For the building, beam splices averaged 80 percent and 70 percent of the
splice lengths calculated using AC1 318-95 for conventional and new high R, bars,
respectively. For columns, the splice lengths averaged 80 percent of the AC1 splice
I lengths, with no advantage for the high relative rib area bars because of the high
confinement. A total steel saving of 1.3 percent was obtained, compared to AC1
3 18-95 based values.
The reduction in steel congestion obtained through the use of shorter splices
is a principal advantage of high R, bars. When congestion is caused by close bar
spacing and large amounts of transverse reinforcement, the benefits of using high
relative rib area bars is the greatest. This means that the new bars will have the
largest effect on reducing congestion in the most highly congested areas of struc-
tures. Under all conditions, the splice lengths of epoxy-coated bars are significant-
ly reduced when high relative rib area bars are used in preference to conventional
bars.
SUMMARY
Historically, structural designers and reinforcing bar manufacturers have

approached changes in reinforcing bar geometry in a highly conservative manner,
and deformation criteria have not changed in nearly 50 years. The current study
represents the first large-scale effort to improve the development characteristics of
reinforcing bars in the United States since 1949. The study is aimed at broadening
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D Obb2949 0542910 9.3e
the undersîanáing of factors that control bond strength, improving the development
characteristics of the reinforcing bars themselves, and developing practical design
expressions that more accurately represent development and splice strength than
provided by current expressions. The results of this study, summarized in this
paper through January 1997, illustrate that increasing the relative rib area of rein-
forcing bars can significantìy improve the bond strength of bars confined by trans-
verse reinforcement and coated bars with and without confining reinfomment. The
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
study also includes the development of improved expressions to characterize bond

strength and design expressions that include a reliability-based strength reduction
factor and provide SisnifiCantsavings in the splice lengths required by AC1 318-95
for bars with and without transverse reinforcement and for conventional as well as
high relative rib area bars.
ACKNOWLEDGEMENTS
Support for this research has been provided by the National Science Foun-
dation under NSF Grants No. MSS-9021066and CMS-9402563,the U.S.Depart-
ment of Transportation - Federal Highway Administration, the Civil Engineering
Research Foundation under CERF Contract No. 91-N6002,the Lester T.
Sunderland Foundation, ABC Coating, Inc., Amensteel (formerly Florida Steel
Corporation), Birmingham Steel Corporation, Chaparral Steel, Fletcher Coating
Company, Herbem-O’Brien Inc., Morton Powder Coatings, Inc., North Star Steel
Company, and 3M Corporation. Support has also been provided by Geiger Ready-
Mix, Iron Mountain Trap Rock Company, and Richmond Screw Anchor Company.
REFERENCES
1. Hyatt, Thaddeus, An Account of some Experiments with Portland-Cement-
Concrete Combined with Iron, as a Building Material, Chiswick Press, London,
1877, 45 pp.
2. Ransome, Ernest L. and Saurbrey, Alexis, Reinforced Concrete Buildings,
Mcûraw-Hill Book Co., New York, 1912,235 pp.
3. Colby, Albert L., Reinforced Concrete in Europe, The Chemical Publishing Co.,
Easton, PA, 1909.260 pp.
4. Brooks, John P., Reinforced Concrete, McGraw-Hill Book Co., New York, 1911,
220, pp.
5. AbramS. Duff A.. “Tests of Bond between Concrete and Steel.” Bulletin. No.
71, Engineering .Experiment Station, University of Illinois, Ukana, IL, Dec.
1913, 238 pp.
6. Clark, A. P.,“Comparative Bond Efficiency of Deformed Concrete Reinforcing
Bars,” AC1 Journal, Proceedings V. 43, No. 4, Dec. 1946, pp. 381-400.
7. Clark, A. P., “Bond of Concrete Reinforcing Bars,” AC1 Joumal, Proceedings
V. 46, NO.3, NOV1949, pp. 161-184.
8. AC1 Committee 208. “Pro~osedTest Procedure to Determine Relative Bond
Value of Reinforcing‘Bars,’; AC1 Joumal, Proceedings V. 41, No. 6, Feb. 1945,
pp. 273-292.
31O Darwin, Zu0 and Tholen
= 0hh2949 0542933 874 m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9. ASTM A 305-47T, “Tentative Specifications for Minimum Requirements for

Deformations of Deformed Steel Bars for Concrete Reinforcement,” American
Society for Testing and Materials, Philadelphia, PA, 1947.
10. Choi, Oan Chul; Hadje-Ghaffari, Hossain; Darwin, David; and McCabe, Steven
L., “Bond of Epoxy-Coated Reinforcement to Concrete: Bar Parameters,” SL
Report 90-1, Unh. of Kansas Center for Research, Lawrence, Kansas, Jan. 1990,
43 PP.
1 1 . Wildt, Roger. (1991). Personal communication at Annual Meeting of Rein-
forced Concrete Research Council, Skokie, IL.
12. ASTM A 944-95, “Standard Test Method for Comparing Bond Strength of
Steel Reinforcing Bars to Concrete Using Beam-End Specimens,” American
Society for Testing and Materials, West Conshohocken, Pennsylvania, 1996, 4
PP.
13. Darwin, D. and Graham, E. K. “Effect of Deformation Height and Spacing on
Bond Strength of Reinforcing Bars,” SL Report 93-1, University of Kansas
Center for Research, Lawrence, Kansas, Jan. 1993,68 pp.
14. Darwin, D. and Graham, E. K. “Effect of Deformation Height and Spacing on
Bond Strength of Reinforcing Bars,” AC1 Structural Journal. V. 90, No. 6,
Nov.-Dec. 1993, pp. 646-657.
1 5 . Tholen, M. L. and Darwin, D., “Effects of Deformation Properties on the Bond
of Reinforcing Bars,” SM Report No. 42, University of Kansas Center for
Research, Lawrence, Kansas, Oct. 1996, 370 pp.
16. Idun, Emmanuel K. and Darwin, D., “Improving the Development Characteris-
tics of Steel Reinforcing Bars,” SM Report No. 41, University of Kansas Center
for Research, Lawrence, Kansas, Aug. 1995, 267 pp.
I
17. Darwin, D., McCabe, S . L., Idun, E. K., and Schoenekase, S. P., “Development
Length Criteria: Bars Not Confined by Transverse Reinforcement,” A CI
Structural Journal, V. 89, No. 6, Nov.-Dec. 1992, pp. 709-720.
18. Darwin, D., Zuo, J., Tholen, M. L., and Idun, E. K., “Development Length
Criteria for Conventional and High Relative Rib Area Reinforcing Bars,” AC1
Structural Journal, V. 93, No. 3, May-June 1996, pp. 347-359.
19. Darwin, D., Idun, E. K., Zuo, J., Tholen, M. L. “Reliability-Based Strength
Reduction Factor for Bond,” SL Report 95-5, University of Kansas Center for
Research, Lawrence, Kansas, May 1995,47 pp.
20. Brown, C. J., Darwin, D., and McCabe, S . L., “Finite Element Fracture Analysis
of Steel-Concrete Bond,” SM Report No. 36, University of Kansas Center for
Research, Lawrence, Kansas, Nov. 1993, 100 pp.
21. Darwin, D., McCabe, S.L., Brown, C.J., and Tholen, M.L., “Fracture Analysis of
Steel- Concrete Bond,” Fracture and Damage in Quasibrittle Structures:
Experiment, Modelling, and Computer Analysis, Z.P. Bazant, Z . Bittnar, M.
Jirasek and J. Maars, Eds., E & FN Spon, London, 1994, pp. 549-556.
22. Donahey, Rex C. and Darwin, David, “Bond of Top-Cast Bars in Bridge
Decks,” AC1 Journal, Proceedings V. 82, No. 1, Jan.-Feb. 1985, pp. 57-66.
Obb2949 0542912 700 W
23. Bretbnann, Barrie B; Darwin, David; and Donahey, Rex C., “Bond of Reiiforce-
ment to Superplasticized Concrete,” AC1 Journal, Proceedings V. 83. No. 1,
Jan.-Feb. 1986, pp. 98-107.
24. Choi. Om Chul; Hadje-Ghaffari, Hossain; Darwin, David; and McCaòe, Steven
L., “Bond of Epoxy-Coated Reinforcement: Bar Parameters,” AC1 Materials
Journal, V. 88. No. 2, Mar.-Apr. 1991, pp, 207-217.
25. Johnston, David W., and Zia, Paul, “Bond Characteristics of Epoxy Coated
Reinforcing Bars,” Report No. F”A-NC-82-002, Federal Highway Adminis-
tration, Washington, DC,1982, 163 pp.
26. Hadje-Ghaffari, Ho&; Darwin, David; and McCabe, Steven L., “Effects of
Epoxy-Coating on the Bond of Reinforcing Steel to Concrete,” SM Report No.
28, University of Kansas Center for Research, Lawrence, Kansas, July 1991, 288
PP.
27. Rehm, G; “The Fundamental Law of Bond,” Proceedings Symposium on
Bond and Crack Formation in Reinforced Concrete (Stockholm 1957), RLEM,
Paris, (published by Tekniska Hogskolans Rotaprinttrychkeri, Stockholm,
1958).
28. Rehm, G., “Uber die Grunlagen des Verbundes Zwischen Stahl und Beton.”
Deutscher Ausschussjùr Stahlbeton. NO. 1381, 1961, 59 pp., (C & CA Library
Translation No, 134, 1968. T h e Basic Principle of the Bond between Steel
and Concrete.”).
29. Lutz, L. A., and Gergely, P., “Mechanics of Bond and Slip of Deformed Bars in
ConcTete,” AC1 Journal, Proceedings V. 64, No. 11, Nov. 1967, pp. 711-721.
30. Darwin.D., Tholen, M. L.. Idun, E. K..and Zuo. J.. “Splice Strength of High
Relative Rib Area Reinforcing Bars,” AC1 SfrucfurulJournal, V. 93, No. 1, Jan.-
Feb. 1996. pp. 95-107.
31. Tan, C., Danvin, D., Tholen, M. L., Zuo, J., “Splice Strength of Epoxy-Coated
High Relative Rib Area Bars,” SL Report 96-2, University of Kansas Center for
Research, Lawrence, Kansas, May 1996,69 pp.
32. AC1 Committee 318, Buikìing Code Requirements for Structural Concrete (ACI
318-95) and Commentary (AC1 318R-95), American Concrete Institute,
Farmington Hills, MI, 1995, 369 pp.
33. AASHTO Highway Sub-committee on Bridges and Structures, Standard
Specifications for Highway Bridges, 16th Edition, American Association of
State Highway and Transportation Officials, Washington, DC,1996.
34. Cairns,J. “Disc. 90465,” AC1 Srrucrural Journal, V. 91, No. 5, Sep.-Oct.
1994, pp. 626-629. (Discussion of Reference 14).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
35. Orangun, C. O.; Jirsa, J. O.; and Breen, J. E., “Reevaluation of Test Data on
Development Length and Splices,” AC1 Journal, Proceedings, V. 74, No, 3,
MU. 1977, pp. 114-122.
36. Hatfield, Eileen; Azizinamini, Atorod; and Pavel, Rob, “Minimum Stirrup
Requirements for.Tension Splices in High Strength Concrete,” Proceedings, 4th
Intl. Symp. on Utilization of High-StrengWigh-Performance Concrete, Paris,
1996, 10 pp.
- 0662749 0542913 647 m
U
aa
.-
c
w-
c
O
o
rn
.-c
o
U
c
O
n
c
Y)
p1
c
U
c
P)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542914 583 m
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
.-& .-E .-t rn

orno O
O b 0
00- 0
OOÔ
II II II
rrrLToZ
, I
I I I I I I I
m Ob62949 0542915 4 1 T
3 14 Darwin, Zu0 and Tholen
\ 1.1
o I l I I I I I 1 I n I I I I I 1 I n I I I l I I I l I I I I I I I I I 1 I I I 1
- - E
5
a 1 .o
X
u
c f
L - O W FI -
0.9
v; O F2
A CI
U
c 0.8 - O V M1 -
O
m
M45.3 -
X M45.4 1
a, 0.7 - U M60.3 1
.-> O M60.4 1
U
0.6
i
''' ' ' I ' I I ''' ' I ' ''I''''I ' I ' '' I ' I I ''''I'' 'I ' ' I'
O 10 20 30 40 50 60 70 80 90 100
Rib Face Angle, y (deg.)
Fig. 3-Relative bond strength of epoxy-coated bars to uncoated bars (C/U) versus
rib face angle, y, for beam-end tests (16)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Obb2149 0542937 292 W
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542918 129 U
3000
2500
N.
C
*- 2000
n
CI
v)
Q)
t 1500
5
<2
\
1000
500
O
O 500 1000 1500 2000 2500 3000
A,,f,/fe'''* (Prediction) =
[8.76b(~,,,+0.54)+ 187~+,](0.14c,,,/~+0.86), in?
Fig. 6-Experimental bond force, T,=Abf,., normalized with respect to f C 'I2

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
versus predicted bond force Abf,/f C'12Ø as a function of concrete compressive

strength for bars without confining reinforcement (1 8)
= Ob62949 0542939 Ob5 m
20005 1 I I
r:
< 3000
(psi) -
O--.-
.----
3000-3500
A----
v - . -.-e.
3500-4000
4000-4500
15000 -o-- -- - 4500-5000
pI.
.-
c x- ..-..- 5000-5500
5500-6000
c
m
v)
a,
L 1oooc
\
2 500í
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O
O 5000 10000 15000 20000
Abf,/f =‘ ”* (Prediction) =
[63~(c,+0.5db)+21 30Ab](0.1 c&,,-tO.g), in!
Fig. 7-Experimental bond force, T,= Abfs, normalized with respect to f tc114 versus
predicted bond force Abf,/f ‘c114, as a function of concrete compressive strength
for bars without confining reinforcement (181
H Ob6_2’ii4~.-0542920 887
SP 180-14
Bond Slip of Bridge Column Reinforcement

Anchored in Cap Beams
by S. Sritharan, J. M. Ingham, M. J. N. Priestley and F. Seible
SYNOPSIS :
Until recently, bridge design specifications in California permitted longitudinal
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
column reinforcement to be terminated in monolithic cap b e d c o l u m n joints
with straight bar development not extending fully up the joint height. Since the
development length could be construed not to comply with AC1 3 18 requirements,
it was suspected that the anchorage length provided for the column bars was
insufficient to develop the full moment capacity of the column at the joint
interface. Considering that this detailing was widely used in bridges in California,
an experimental investigation was initiated at the University of California, San
Diego, where response of a bridge knee joint incorporating as-built reinforcement
details was examined under simulated seismic loading. Following inadequate
performance of the test unit, the behavior of the knee joint was investigated based
on the experimental readings, giving consideration to bond slip of the longitudinal
column reinforcement. The response of the test unit indicated that the bond
conditions developed when anchoring the longitudinal reinforcement of circular
bridge columns is unlike that developed along the beam reinforcement anchored
into building joints.
Keywords: Bonding; bond slip; bridge anchorages; bridge structure; caps

(supportsl; cyclic loading; seismic loading; slippage
319
0662949 0542923 733
320 Sritharan et al.
Sri Sritharan obtained his BSc degree in civil engineering from the University of
Peradeniya, Sri Lanka, and his ME degree from the University of Auckland, New
Zealand. After working as an Engineering Seismologist at IGNS in New Zealand,
he is pursuing his PhD at the Universi0 of California, San Diego. His research
interests include seismic design of structures and earthquake engineering.
Jason M. Ingham is a lecturer in the Department of Civil and Resource

Engineering at the University of Auckland, New Zealand. He received his BE and
ME degrees from the University of Auckland and his PhD from the University of
California at San Diego, La Jolla, California. His research interests include
earthquake-resistant design of concrete and masonry structures.
M. J. Nigel Priestley, FACI, is a professor of structural engineering at the

University of California at San Diego, La Jolla, California. He received ACI’s
Raymond C.Reese Award in I984 and 1989, and ACI’s Wason and Anderson
Awards in 1997. His research interests include reinforced and prestressed
concrete structures and earthquake-resistant design.
AC1 member Frieder Seible is a professor of structural engineering at the

University of California at San Diego, La Jolla, California. He received his PhD
from the University of California, Berkeley. His research combines large-scale
experimental testing and nonlinear analytical modeling of structural systems. He
is a member of ACI Committee 341, Earthquake-Resistant Concrete Bridges.
INTRODUCTION
in current practice, structures are designed to respond in a ductile manner with

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
little or no strength degradation when subjected to severe earthquake loading. If

the response of the structure is associated with significant lateral strength
degradation, the structure may no longer be able to support its self weight and
catastrophic collapse may be inevitable. The lateral load-resisting ability of a
bridge structure can be affected by several design deficiencies such as: insufficient
shear reinforcement in beam or column members, inadequate confinement in the
hinge regions, insufficient lap splice of reinforcement, buckling of longitudinal
reinforcement, poor detailing of connections, and bond slipping of longitudinal
bars due to inadequate anchorage length into the connecting member. All of these
structural defects have been identified as inherent in old concrete bridges in
California [7]. Collapse of bridge structures as a consequence of such design
deficiencies was evident in the 1971 San Fernando, 1989 Loma Prieta and 1994
D 0662947 05429221 65T
Northridge earthquakes.
Retrofitting some of the structural defects in existing bridge structures is essential

and also relatively inexpensive. inadequate shear resistance of cap beam or
column members, and lap splice and confinement related problems fall into this
category. However, retrofitting of poorly detailed b e d c o l u m n and
colundfooting connections and bond-slip related problems are both complicated
and expensive, and not necessarily essential. It is considered that if the lateral
load resisting ability of a structure is not severely affected by structural
deficiencies inherent in the structure, acceptable performance can be anticipated
when subjected to severe seismic loading. in order to establish that the overall
response of the structure would be satisfactory under the maximum credible
earthquake, a push-over type analysis may be performed considering the possible
strength degradation resulting from poor detailing inherent in the structure. in this
paper, bond slip of bridge column reinforcement anchored in the cap beam is
explored.
Bridge design specifications of the California Department of Transportation [2]

permitted the longitudinal column reinforcement to be terminated part way
(typically at a distance of three quarter of the beam depth) up the cap beam with
straight bar ends. This detailing, which had been commonly adopted until the
1989 Loma Prieta earthquake, was suspected to be inadequate for anchorage of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the column reinforcement into the cap bedcolumn joint. Experimental evidence
and analytical results presented in this paper for a bridge knee joint system enable
the significance of imperfect anchorage of longitudinal column reinforcement in
cap beams to be clearly understood. As a consequence, rational decisions about
the necessity for complex column/cap beam joint retrofits to ensure adequate
seismic response can be made using models for joint force transfer.
LARGE-SCALE TEST ON A BRIDGE KNEE JOINT
A large-scale bridge knee joint system incorporating recently completed as-built

details was tested under simulated seismic loading at the University of California
at San Diego (LJCSD) [5]. The test unit was adapted from a one third scale model
of outrigger bent #31 of the E105 elevated freeway in Los Angeles, which was
designed in 1987. In Figure 1, the key reinforcement details of the test unit are
shown while the overail dimensions of the laboratory model may be obtained
0hh2949 0542923 5 ï b
from Figure 4. The detailing of the joint region is fairly satisfactory and shows a
considerable improvement when compared to a typical bridge joint from pre- 1971
structures. However, it is noted that the column bars terminate well below the cap
beam top reinforcement and have a development length of 29db1, where dbl is the
longitudinal bar diameter. This is 25% less than the anchorage length required by
recent AC1 3 18 provision [ 11 for the prototype structure, given measured material
strength o f f , = 30.7 MPa and fy = 470 MPa, and was justified assuming that the
column flexural strength is fully developed adjacent to the joint interface. When
the anchorage length of 29dbl provided for the column bars is compared to the
AC1 requirement, it can be concluded that the maximum stress that can be induced
in the column tension bars at the joint interface would be the yield strength.
Therefore, if the column bars are subjected to tensile stresses exceeding the yield
strength, bond slip would be inevitable.
An average bond stress corresponding to the AC1 anchorage length requirement

can be interpreted as 0.76&, where fi is the concrete compressive strength in
MPa. However, bond stress exceeding 25& was repeatedly obtained for
column reinforcement anchored in bridge joints in large-scale laboratory tests [6-
81, implying that the column bar strength can be developed over a length
significantly less than that indicated by the AC1 provision. When joint force
transfer is considered in a rationalized manner as outlined below, it is clear that
the anchorage length resulting from a bond stress of 2.5& is the effective
development length, and this can only be relied upon if the straight bar extensions
of the column reinforcement are extended as close to the top longitudinal beam
reinforcement as possible.
Probable Joint Mechanism

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The bond slip of bridge column reinforcement anchored in cap beams can be
examined conceptually using force transfer mechanisms. in Figure 2, two bridge
knee joints are shown with probable force transfer mechanisms across the joints,
where the total column tension force, T,, is represented by two ties with equal
demands. In both cases, the cap beam is assumed to have reserve capacity so that
external struts can be developed to the right of the joint face [7]. In the knee joint
shown in Figure 2a, the column reinforcement is considered to be extended close
to the top beam reinforcement. As a result, good clamping action is expected to
develop for all the column tension reinforcement with effective bond stresses as
high as 25&. In such circumstances, the bar strength is developed along an
anchorage length less than that required by the AC1 318 provision [ 11 and bond
slip of the column reinforcement is avoided. This has been experimentally
verified in laboratory tests on bridge joints [8,9].
Ob62949 0542924 422
When considering the force transfer of a knee joint, where insufficient anchorage
length is provided for the column bars, as shown in Figure 2b, it is seen that the
column tension reinforcement, particularly those bars located close to the
centroidal axis of the column, are not effectively clamped into the joint strut.
Consequently, the.se bars are expected to have poor bond condition and are likely
to slip when subjected to large tensile stresses. Because of good clamping action,
a better bond condition is expected for the extreme tension reinforcement of the
joint in Figure 2b. Depending on the capacities of the struts providing the
clamping action and the ties supporting the siruts, these bars can also experience
bond slipping when subjected to large tensile forces. Consequently, the total
column tension force is reduced, providing a reduced flexural moment resistance
at the top of the column,
When the reinforcement details of the knee joint test unit were considered, a
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
mechanism similar to that shown in Figure 2b was expected to develop. Hence,

slipping of the column reinforcement in the test unit was likely to occur as
predicted based on current AC1 provisions for anchorage length requirement.
Theoretical Bond Slip Model
The bond slip of the column reinforcement embedded in the joint shown in Figure
2b is discussed in conjunction with a theoretical bond-slip model. The bond-slip
model presented in Figure 3 may be considered to be adequate for the #6 (dai= 19
mm) column bars of the test unit when subjected to monotonic loading. This
model was derived using the average bond model proposed by Eligehausen et al.
[4] and measured concrete properties. In Figure 3, the maximum bond stress was
taken as z, = 25A = 13.9 Mpa for a concrete strength fc = 30.7Mpa. This
assumption was consistent with the peak bond stresses obtained in the test
program of Eligehausen et al. The component zf in Figure 3 was chosen such
that the ratio of 2.7 between zm and ? T ~of the original model could be retained
while slip values SI, s2 and s3 were taken as identical to the average values
suggested by Eligehausen et al. When the reinforcement is subjected to cyclic
loading, a reduced bond stress envelope should be considered [4].
Based on the possible force transfer model shown in Figure 2b for the test unit
with inadequate anchorage of longitudinal column reinforcement, it was
concluded that a better bond condition was expected for the extreme tension bars
than for those located close to the column centroidal axis. A peak bond stress as
high as 13.9 MPa was unlikely to hold for the latter and a reduced bond stress
with increased slip ( ~ 2 was
) expected for these bars consistent with the model in
Figure 3. However, the maximum bond stress 2, , which corresponds to slip of
0 b b 2 9 4 9 0542925 369 =
up to 3 mm, can be developed along the effectively clamped portion of the
extreme tension reinforcement. Therefore, in a circular column, it is possible that
the extreme tension reinforcement may experience less bond slip than the column
bars placed closer to the centroidal axis of the column section.
When significant slip occurs, bond resistance is provided by the frictional

component zf.The value zf= 5.1 MPa and anchorage length 1, = 559 mm suggest
that a bar strength of 599 MPa can be developed in the tension reinforcement of
the test unit at the joint interface. Nonetheless, a reduction in zf is expected with
cyclic loading and, due to cracking, the total embedment length can not be
considered as effective for developing tension in the reinforcement.
Consequently, a much reduced tensile stress would be developed at the joint
interface.
Although the bond stress model of Eligehausen et al. [4] was considered in the
above discussion, it is noted that this model was developed based on simple pull-
out tests in conditions similar to that typically experienced by the beam
reinforcement in building joints. All of the beam tension reinforcement anchored
in a building joint, except perhaps those bars located in the corners, may be
considered to be exposed to identical bond condition. However, the force transfer
models in Figure 2 have indicated that a similar situation is not expected when
anchorage of the longitudinal reinforcement of a circular column into a bridge
joint is considered. Thus, applicability of the model to bridge column bars
embedded in cap beams needs to be validated.
EXPERIMENTALOBSERVATIONS
Test Set UP
The test unit was subjected to a reverse cyclic loading using the test set-up shown
in Figure 4. The actuator simulating the seismic forces was inclined so that the
variation of column and cap beam axial and shear forces due to seismic response
could be appropriately modeled. In the push direction loading, the joint was
subjected to opening (positive) moments, which induced an axial tension force in
the column. Similarly, when closing moments were applied to the joint, an axial
compression force was developed in the column as would be expected in the
exterior column of a multi-column bridge configuration [7]. More details of the
test Set-up are given in reference [ 5 ] .
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bbt949_D54i926 Z T 5
Visible Damage
The behavior of the test unit under simulated seismic loading was reported by
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ingham et al. [5]. Test observations relevant to the current study are summarized
in this section.
When subjected to joint-opening moments, it was noted that incipient spalling

developed on the outside face of the column at ductility 1.5, but no further
spalling of column cover concrete occurred during the remainder of the test, as
can be seen in Figure 5. This suggests that the concrete compressive strain did not
significantly exceed the threshold value for spalling of about 0.006 in the column.
In the absence of further visible damage beyond displacement ductility 2, it was
suspected that intemal joint rotation markedly contributed to the vertical cap beam
end displacement measured at this and larger ductility levels. A wide crack
originating at the inside comer of the joint (see Figure 5 ) supported this
hypothesis as it presumably allowed the cap beam to rotate with respect to the
column. At ductility 2.8, bulging of the joint was seen. Based on these
observations, it was concluded that the column longitudinal reinforcement
experienced bond slipping at large ductilities when the joint was subjected to
opening moments. As shown in Figure 6, the post-test investigation of the test
unit, which was conducted once the cap beam was returned to its original position
by removal of the joint cover concrete, confirmed that at least some of the
longitudinal column reinforcement slipped with respect to the joint core concrete
during the test.
When the joint was subjected to closing moments, a full distribution of fan
cracking developed on both joint faces (Figure 5 ) with spalling of cover concrete
occurring in the plastic hinge region of the column starting at ductility 1.5. Joint
damage continued to increase during the remainder of the test. No bond slipping
of column reinforcement was suspected due to joint-closing moments. Therefore,
the bond slip study reported in this paper is limited to the response of the knee
joint under opening moments.
Force-DisulacementReswnse
The hysteresis loops obtained for the test unit, plotting the actuator force against
the vertical displacement measured at the intersection point between the cap beam
and actuator, are r e p d u d in Figure 7. Although the energy dissipation capacity
of the test unit appears to be satisfactory, it is noted that the ideai flexural strength
(pi) of the test unit was not fully developed for either joint-opening or joint-
closing moments. In this context. ideal flexural strength is the actuator force
corresponding to development of column flexural strength based on measured
m Ob62949 0542927 131
material properties, the appropriate level of axial force, and a critical section
curvature of 5 times the nominal yield curvature, using a bilinear approximation
to the theoretical moment-curvature curve. When a column frames into a well-
designed joint, flexural capacity should exceed ideal strength at high ductility
levels as a result of strain-hardening of flexural reinforcement. The hysteresis
loops shown in Figure 7 indicate pinching in the region corresponding to zero
displacement, when compared with the behavior of other, well-designed knee
joints [5] with a similar configuration. Also observed in Figure 7 is that
noticeable strength deterioration of the system occurred at ductility 2.8 during
cycling under opening moments.
It appears that the response of the test unit under negative moments was also
affected by bond slip of the column bars since the ideal moment capacity was not
reached under closing moments whereas a fully competent joint should exceed the
ideal capacity as a consequence of strain hardening of the longitudinal
reinforcement. However, no significant deterioration in the joint-closing force
resistance of the system took place during cycling at p = 3. The displacement of
the test unit was restricted to moderate ductilities because a repair of the unit in
the joint region was planned as a part of the research program and hence excessive
joint damage was considered undesirable.
Strain Gauge Readings
A large number of strain gauges recorded reinforcing steel strains in the joint
region. Two strain gauge histories and corresponding stress histories obtained for
a longitudinal column bar located at a distance equal to 0.82 time the column
diameter from the extreme column compression fiber under opening moment, are
discussed in this section. The strain measurements shown in Figure 8
corresponded to a gauge located in the column 203 mm below the joint interface
while the strain history obtained at the interface is shown in Figure 9. Stress
estimates corresponding to recorded strains in this investigation were established
using the constitutive model developed by Dodd and Restrepo-Posada [3] with a
softening factor of 0.7 as used elsewhere [6].
The variation of strain and stress quantities presented in Figure 8 for the gauge
located in the column, shows typically expected response for the column
reinforcement, There was no sudden reduction in the stress history encountered,
suggesting that slipping of this reinforcement, if any, was gradual. Strain gauges
which were placed below and above the joint interface on several other
longitudinal column bars exhibited behavior similar to that shown in Figure 8.
Almost all of the gauges placed within the joint on the column tension
reinforcement failed as the target displacement was approached during the first
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542928 078
Tribute to Peter Gergety 327
joint-opening cycle at ductility 2.8, presumably due to large slip Occ&ngto the
column reinforcement.
Many strain gauges located on the column reinforcement at the joint interface
exhibited apparently anomalous behavior. An example of this is shown in Figure
9, where stress histories obtained at ductilities 2 and 2.8 are also presented as a
function of the column bending moment at the joint interface. The stress and
strain histones of this gauge (Figures 9a and 9b) appear to be similar to that
shown in Figure 8 for cap beam displacements over a range from -92 mm to 126
mm. When larger displacements were imposed at the end of the cap beam,
significant stress drop in the reinforcing bar was obtained in the proximity of the
peak displacement, particularly for joint-opening moments. The rate of loading is
believed to have influenced this behavior. The test was controlled manually and
hence the rate of loading was not constant during the test. As the peak
displacement was approached, a reduced loading rate was used and also the load
in the actuator was held for a considerable time at the peak displacements to
observe and record the damage that developed in the test unit. In order to validate
this argument, consider the stress history shown in Figure 9c. At displacement
ductility 2 under opening moments, higher stress drop was obtained in the first
and third cycles than in the second cycle. The displaced shape of the test unit was
indeed maintained at the peak displacement for a considerable time in the first and
third cycles, but the actuator load was quickly reversed when the target
displacement was reached during the second cycle as visual observations were not
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
made in this loading cycle. At ductility 2.8 (seeFigure 9d). the test was paused at
the peak displacement during each cycle to observe the damage.
It is believed that the significant stress drop seen in Figure 9 was not induced as a
resuit of sudden bond slip occurring in the reinforcement. This statement is
supported by considering: (a) the stress drop encountered at the joint interface was
not reflected in the stresdstrain histories obtained within the joint or in the
column, (b) no corresponding change was introduced to the overail force-
displacement response (Figure 7), (c) estimated forces in the longitudinal column
reinforcement based on measured .strains satisfied moment equilibrium at the top
of the column before stress drop occurred, but not immediately afterwards, and (d)
no strength deterioration was observed during cycling at ductility 2. It is,
therefore, concluded that the recorded strains shown in Figure 9 were not
representative of the actual strains developed in the column reinforcement at the
interface at large ductilities (p 2 2). The strain reduction recorded at the peak
displacements might have been due to slipping of the gauges with respect to the
reinforcement.
Obb2949 0542929 TO4
Summarv
The bond slipping of the column reinforcement was not clearly identified in
strainktress histories presented in Figures 8 and 9. The visible damage which
occurred to the test unit, the force-displacement response of the system together
with curvature measurements taken in the beam and column regions adjacent to
the joint indicated that gradual slipping of column reinforcement took place
starting from at least p = 2. This is further investigated in the following section.
It is also believed that significant slipping to column reinforcement occurred when
the joint-opening displacement corresponding to p = 2.8 was first approached.
When significant reinforcement slippage developed, the bond stress deteriorated
rapidly. As a result, the force resistance of the system reduced with an increasing
number of loading cycles, as observed for the response of the test unit at ductility
2.8 under opening moments (Figure 7).
ANALYSIS OF THE KNEE JOINT
Displacement Components
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
As a first step towards understanding the response of the test unit, the
displacement components were investigated. Flexural response of the bent cap,
deformation of the joint, and flexural rotation at the top of the column with
respect to the base mainly contributed to the vertical displacement that was
measured at the end of the cap beam. If bond slip of the column reinforcement
had occurred, this would have also influenced the beam end displacement. The
1 joint panel region and the column were sufficiently instrumented and therefore
their displacement contributions could be readily obtained. Because the cap beam
was expected to respond in an elastic manner, only a segment of the beam
adjacent to the joint was instrumented with curvature devices. By extrapolating
the recorded curvature to the uninstrumented portion of the beam, the deflection
contribution of the cap beam was also calculated [9].
For a given actuator force, member end moments, shears and axial forces can be
obtained from equilibrium conditions. Therefore, theoretical estimates of the
displacement components can also be established as reported elsewhere [8,9]. In
determining the column and cap beam theoretical displacements, only flexural
deformation was considered, taking the influence of strain penetration term into
account wherever appropriate. When damage occurs to the joint, the displacement
component due to joint deformation is often grossly underestimated in the
theoretical prediction. Hence, in order to minimize the error due to joint
Ob.b2949 0542930 72b
deformation, the theoretical displacement component of the joint was taken equal
to that established from experimental measurements.
In Table 1, the experimentally measured and theoretically calculated displacement

components are shown at the peak joint-opening displacements at selected
ductilities. Good agreement between the accumulated displacement from
experimental measurements and the total measured displacement is seen, except
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
for the third cycle at p = 2.8. At this stage of testing, it was possible that the joint
deformation component was not accurately measured. Calculated column and cap
beam displacement contributions were significantly underestimated when
compared to the values obtained from experimental readings. It is believed that
bond slip of the column reinforcement and corresponding joint mechanism was
largely responsible for the unaccounted displacement [Error (cal)] of up to 53%,
when comparing the accumulated displacement from theoretical estimates to the
total measured displacement.
When bond slip occurs to the column reinforcement, it introduces a fixed-end (or
bond-slip) rotation at the top of the column, which increases the cap beam end
displacement proportionally. Although it is a secondary effect, bond slip of the
column reinforcement also significantly influences the cap beam displacement
component. When column reinforcement, particularly the most extreme tension
bars, are not adequately anchored within the joint, external struts are developed, as
shown in the force transfer models of Figures 2 & 11, in order to provide
additional clamping. This increases the curvature in the beam region adjacent to
the joint, resulting in a larger beam displacement contribution than would be
expected from simple section analysis [6,9].
By comparing the experimental and Calculated column and cap beam

displacement contributions, an estimate of the displacement component
intraduced by bond slip of the longitudinal column reinforcement can be obtained.
It is seen in Table 1 that the total measured cap beam end displacement was
influenced by the bond slip of the column bars, even when the actuator load
corresponding to the theoretical yield strength (Fy)was applied. Bond slip
rotation at the top of the column of up to 30.1% of the total measured
displacement was recorded while the additional curvature introduced to the cap
beam provided displacement contribution of up to 12.4%. In total, the bond slip
effect contributed to 24.2% of the total displacement at the theoretical yield force
and 43% at ductility 2.8.
üb62749 0542931 bb2 =
Bond-Slip Rotation
Based on the displacement component [Col. (exp) - Col. (cal)] in Table 1, the
bond slip rotation which occurred at the top of the column was established. In
Figure 10, the bending moment, which was induced at the top of the column, is
plotted against the bond-slip rotation. It is seen that bond slip of the column
reinforcement introduced a fixed end rotation having an almost elastic-perfectly
plastic relation to the applied joint-opening bending moment. Given that
significant strength deterioration occurred during cyclic loading at ductility 2.8
(Figure 7), a negative stiffness in the moment-rotation relation may be expected at
displacements ductilities greater than 2.8. This should be taken into account when
assessing the seismic performance of the prototype structure.
Bond-Slip Estimates
Bond slip of the column reinforcing bars was not directly measured during the
test. However, an estimate of the slip was obtained based on the bond-slip
rotation at the top of the column (Figure 10). If €4, is the bond slip rotation at the
top of the column, then the bond slip of individual tension reinforcement at the
joint interface can be approximated to:
A, = eb*dnb
where As is the bond slip of the reinforcement and dnb is the distance to the bar
from the neutral axis of the column section. Considering zero slip at the column
neutral axis, it is assumed in Eq, 1 that the bond slip of the tension reinforcement
increases linearly with the distance from the neutral axis. Estimated bond slips at
the extreme tension reinforcement and at the centroid of the column tension force
are given at selected ductilities in Table 2. Considering that the bond slip
estimated at the centroid of the column tension force may be representative for all
tension reinforcement, it is seen that the column reinforcement slip estimated at p
I1.5 was less than the maximum slip (s2) corresponding to the peak bond stress
in the theoretical model (Figure 3). Larger bond slip was obtained for p > 1.5. At
ductility 2.8, the bending moment at the top of the column was reduced by 14%
from cycle 1 to cycle 3. However, almost identical slip was obtained for both
loading cycles, presumably due to deterioration in bond stress.
At ductility 2 , the bond slip at the centroid of the column tension force was 3.8
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542932 5T9 W
mm, which was higher than the value corresponding to s2 (= 3 mm) in the
theoretical model (Figure 3). The experimental investigation by Eligehausen et al.
[4] indicated a considerable deterioration in bond stress when the reinforcement
was subjected to slip of the order of s p . Aithough there may be an error associated
with the predicted slip in Table 2, it can be concluded based on the estimated
bond slip that some deterioration to bond stress should be expected at this
ductility. However, this was not supported by the force-displacement response of
the test unit and the tensile stress that was measured at 152 mm from the bar end.
In both cases, no significant deterioration was observed with an increasing
number of loading cycles, suggesting a similar bond stress was repeatedly
developed in the tension reinforcement at ductility 2.
In Eq. 1, it was assumed that reinforcement slip was proportional to the distance
to the column neutral axis. On the contrary, the mechanism shown in Figure 2b
implies that a larger slip is likely to occur for the column bars placed close to the
centroidal axis than those bars located close to the joint interface. As a result, a
larger slip than that given in Table 2 should be expected for the column bars close
to the joint faces. The residual slip of about 12 mm estimated for the inner
column reinforcement from Figure 6 is higher than that calculated at the centroid
of column tension force at p = 2.8 in Table 2.
Force Transfer Mechanism
Based on the anchorage detail of the column longituâinal reinforcement of the test
unit, a force transfer model similar to that shown in Figure 2b was expected to
develop in the joint region. in this section, the joint mechanism is briefly
examined using the recorded response of the test unit. A good set of strain gauge
data was obtained during the test until the displacement corresponding to p = 2.8
was first imposed. Therefore, the force transfer of the joint region was
investigated using the data that was obtained at the peak displacement in the first
opening cycle at p = 2.
Including the mechanism shown in Figure 2% a ;umber of force transfer models,

which are considered favorable for opening of the knee joint, were first examined
using strut-and-tie modeling procedure outlined in references 17,101. However,
none of them resulted in a stable mechanism. Considering the following
information, the mechanism shown in Figure 11 was finaiiy derived as the joint-
opening mechanism developed at the peak displacement in the first cycle at p = 2.
0 under opening moments arch shaped cracks developed close to the reentrant
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542933 435
comer of the joint at the beginning and then moved towards the bottom
outside corner as the test progressed (Figure 5). There was a clear indication
that a part of the diagonal strut was anchored in the bottom outside comer of
the joint at final stages of testing.
the effective anchorage length for the column bars was 152 mrn,
corresponding to a uniform bond stress of about 2 . 5 J f ; o and
f: = 30.7 MPa . This is consistent with the observation that stress close to yield
strength was developed at 152 mm from the bar end in the column tension
reinforcement at p = 2. There was no significant increase in the tensile stress
was measured below this point in the anchorage portion of the reinforcement.
For the estimated bond slip in Table 2, the theoretical bond-slip model of
Figure 3 would predict a lower bond stress value.
measured tensile stress in the bottom beam reinforcement at the joint interface
indicated that the tension demand exceeded the expected level by 270 kN,
where the expected tension force was calculated based on a moment-curvature
analysis of the section.
the vertical component of each diagonal strut clamping the column tension
force was OST,. Considering different T,ratios for the vertical components of
diagonal struts resulted in unstable mechanisms.
The total column tension force in the force transfer model is represented by a
single tie (Figure I I ) as it was recognized that the column tension bars could not
be effectively clamped in joint struts VZ and TZ because of insufficient
embedment length. Furthermore, in the strut-and-tie model, the tension force T, =
1295 kN was mainly provided by the beam bottom reinforcement while joint
vertical reinforcement and horizontal spirals contributed to tension forces 238 kN
at node T and 276 kN at node W respectively. To develop tension forces TW,
WX and UV the joint reinforcement must be subjected to stresses beyond yield
strength, which was observed during the test. More complete details of derivation
of the force transfer model and experimental evidence supporting the proposed
mechanism are given by Sritharan [9].
Moment-CurvatureAnalysis
In Figure 12, the column moment-curvature response measured at the joint
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M 0662949 0542734 37%
interface is shown. As would be expecteù from the hysteretic response shown in

Figure 7,the ideal moment capacity of the column was not developed during the
test. The envelope of the measured response indicated an almost elastic-perfectly
plastic behavior with moment capacity 19% higher than the theoretical yield
moment.
Considering the mechanism shown in Figure 11, a simple moment-curvature

analysis was performed for the column section corresponding to joint-opening
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
moments. Based on the AC1 318 provision, it was concluded that the maximum
tensile stress that could be developed in the column tension bar at the interface
was fy, the reinforcement yield strength. Therefore, in the moment-curvature
analysis, it was assumed that the maximum tensile stress developed in the most
extreme tension bars was fy while the peak stress was restricted 0.59fy in the
remaining tension reinforcement. This was decided based on the total column
tension force T, = 1566 kN F igure 11) and the assumption that the most extreme
tension bars were responsible for providing OST,. Zero strain hardening was
considered for all the tension reinforcement to allow for bond slip following
development of the maximum expected bar stress. In Figure 12, it is seen that by
limiting the tensile stress in the reinforcement the predicted moment-curvature
response satisfactorily captured the recorded response envelope. As noted
previously, a reduced moment resistance at the top of the column should be
expected for response beyond ductility 2.8 because of significant bond
deterioration. The predicted response was not expected to capture this behavior
since reduction to the maximum bar stress at large displacements due to bond
deterioration was not considered in the analysis.
DISCUSSION A N D CONCLUSIONS
The seismic performance of a bridge knee joint incorporating reinforcement

details typical of a recently completed outrigger bent is presented in this paper.
The most inefficient detailing of the joint region was that the longitudinal column
reinforcement was prematurely terminated within the joint. The AC1 318
provision and probable joint force transfer mechanism indicated that the
anchorage length of the column reinforcement into the joint was insufficient and
that bond slip of the column bars was inevitable. When the test unit was
subjected to simulated seismic loading, the ideal flexural capacity of the system
was not fully developed for joint-opening or joint-closing moments. With a we&
designed joint between the column and cap beam, flexural capacity should exceed
predicted ideal strength due to strain hardening of the flexural reinforcement.
Based on this observation, visible damage of the test unit, and pinching of the
hysteresis force-displacement response, it was suspected that the bond slip of the
column reinforcement occurred during the test particularly for joint-opening
m Obb2949 0542935 208 m
moments. This was confirmed in the post-test investigation of the joint region.
Based on the experimental evidence and analysis of the test data, the following
conclusions have been drawn for joint response under opening moments:
The seismic performance of the test unit was affected by slipping of the
column reinforcement starting at early stages of testing. The displacement
component calculation confirmed that bond slip of the column bars occurred
when the actuator load corresponding to the theoretical yield strength was
applied. However, the bond slip was not significant until the displacement
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
corresponding to p = 2.8 was imposed. As a result, gradual strength
deterioration to the force resistance was first encountered with an increasing
number of load cycles at this ductility.
The bond slip of column reinforcement markedly influenced the vertical

displacement measured at the end of the cap beam. The displacement
contribution due to bond slip of the column bars was estimated to be 24% of
the total displacement at Fyand 43% at p = 2.8.
Over the maximum displacement range applied to the test unit, the bending
moment and bond-slip rotation at the top of the column exhibited an almost
elastic-perfectly plastic response characteristic.
Considering the significant strength deterioration measured at ductility 2.8, it

is suggested that the moment resistance at the top of the column would be
further reduced at p > 2.8, introducing negative stiffness to the response
envelope of the system.
Estimated bond slip of the column tension reinforcement indicated slip values
as high as 5 mm and 9 mm in the extreme tension reinforcement at ductilities
2 and 2.8 respectively.
0 Force transfer across the joint region under opening moments was adequately
represented by a strut-and-tie model. This is similar to the anticipated
mechanism of a bridge knee joint with inadequate anchorage of the column
reinforcement.
Obb29Y9 054293b I144
0 Following consideration of the force transfer model, it was suggested that,

unlike for the beam reinforcement anchored in building joints, not all the
tension reinforcement of a circular bridge column may be exposed to similar
bond conditions. This should not be ignored when examining the bond slip
behavior of bridge column reinforcement based on theoretical models.
Measured moment-curvature response of the column adjacent to the joint

exhibited an almost elastic-perfectly plastic response envelope under opening
moments, with moment capacity 19% higher than the theoretical yield
capacity. Consistent with the AC1 3 18 provision for reinforcement anchorage
and based on the established joint force transfer model, a simple moment-
curvature analysis was performed by limiting the maximum tensile stress in
the reinforcement to account for bond slip. This accurately captured the
measured response envelope.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial support provided by the

California Department of Transportation. Conclusions in the paper are those of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
the authors and should not be construed to have endorsement from the financial
sponsor.
REFERENCES
American Concrete Institute, Building code Requirements for Structural

Concrete (AC1318-95) and Commentary (AC1318R-95), Michigan, 1995.
Caltrans, Bridge Design Specifications, California Department of

Transportation, State of California, Sacramento, California, 1993.
Dodd, L. L. and Restrepo-Posada, J. I., ‘‘Model for Predicting Cyclic

Behavior of Reinforcing Steel,” J o u m l of Structural Engineering, ASCE,
Vol. 121, No. 3, March 1995,pp. 433-445.
Eligehausen, R.,Popov, E. P. and Bertero, V. V., Local Bond Stress-Slip

Relationships of Deformed Bars under Generalized Excitation Earthquake
Engineering Research Center, Report No. UCB/EERC-83/23, University of
Ohh29Y9 0542937 O80
California at Berkeley, California, October 1983.
(5) Ingham, J. M., Priestley M. J. N. and Seible, F., Seismic Performance of

Bridge Knee Joints - Volume II, Structurai Systems Research Project,
Report No. SSRP 94/17, University of California at San Diego, California,
October 1994.
(6) Ingham, J. M., Seismic Performance of Bridge Knee Joints, Doctoral

Dissertation, Division of Structural Engineering, University of California at
San Diego, California, 1996.
(7) Priestley, M. J. N., Seible, F. and Calvi, M., Seismic Design and Retrofit of
Bridges, John Wiley & Sons, New York, 1996.
(8) Sritharan, S., Priestley M. J. N. and Seible, F., Response of ColumdCup

Beam Tee Connections with Cup Beam Prestressing, Structural Systems
Research Project, Report No. SSRP 96/09, University of California at San
Diego, California, December 1996.
(9) Sritharan, S . , Analysis of Concrete Bridge Joints Subjected to Seismic

Excitation, Doctoral Dissertation, Division of Structural Engineering,
University of California at San Diego, california, (in preparation).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(10) Schlaich, J., Schafer, K. and Jennewein, M., “Toward a Consistent Design
of Structural Concrete”, PCI Journal, Vol. 32, No. 3, May-June 1987, pp
75- 149.
ObbEiï49 0542938 Ti17
o =
‘o“
=
Y -
=e
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
- ObbS1949 0542939 953 =
TABLE 2-ESTIMATED BOND SLIP OF THE COLUMN TENSION

REINFORCEMENT BASED ON BOND-SLIP ROTATION
at extreme tension bar at centroid of tension force

(mn-4 (mm)
F,, cycle 1 2.2 1.7
p = 1, cycle 1 2.2 1.7
p = 1.5, cycle 1 3.2 2.4
p = 2, cycle 1 5.O 3.8
p = 2.8, cycle 1 9.2 7.0
u = 2.8. cvcle 3 9.0 7.0
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Ob62949 0542940 b75 D
4" (102 mm)
/ U
l.33"
(33.9 mm) 24 #6 Grode'60
I 24" 28" (711 mm)
is
Fig. 1-Reinforcement details of a knee joint test model (model scale 1:3)
u T, = Total column
tension force
Fig. 2-Probable force transfer models for bridge knee joint-opening mechanism:
(a) a desirable mechanism; (b) an undesirable mechanism
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0542943 503 m
T,= 13.9
o
œ
s,=l s,=3 s3=10.5

Bond slip (mm)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 3-Theoretical bond-slip model for the #6 (dbl= 19mm) column longitudinal
reinforcement
2’-8”
simulation
1880
r
6’L2”
I
rnm
.
ir
T
Fig. 4-Overall dimensions and test set-up of the knee joint model
= 0662949 0542942 448 m Tribute to Peter Gergely 341
Fig. 5-Damage to joint region at final stages of testing

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 6-Slipping of column reinforcement as evidence in the post-test investiga,tion
0 b b 2 9 4 9 0542943 384
Displacement (mm)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
-250 -200 - 1 5 0 -100 -50 O 50 100 150 200 250
-10 -8 -5 -3 O 3 5 8 10
Displacement (in.)
Fig. 7-Force-displacement hysteresis response of the test unit
D kplacem en t (m m ) Dlsplacement (In.)

I -200 -100 o 100 200 -200 -100 o 100 200
ao
500
60 400
300
40
200 -
n
20
100 e
5.
3 0 0 :
L
-100 g
-20 III
-200
-40 -300
-60 -400
-500
-80
. I O -8 -6 .4 -2 O 1 4 6 8 10 -10 -a -6 -4 -2 O 2 4 6 8 IO
Displacement (in.) Dlsplacemcnt (In.)
Fig. 8-Measured strain and calculated stress histories of a longitudinal column

reinforcement gauge located below the joint interface: (a) strain history; (b) stress
history
Obb2949 0542944 210
DiapIicenent (mm) DLplieemeit (In.)

-200 -100 o 100 200 -200 -100 o 100 200
80I
soo
60 400
c. 20
40 300
200
100
-
x
5
::o 0 : I
2
3 -20 -100 f
VY
-200
-40
-300
-60 -400
-500
-PO
-10 -8 -6 -4 -2 O 2 4 6 8 10 -10 -8 -6 -4 -2 O 2 4 6 8 IO
D b p l i c e m e i i (in.) Displacement (in.)
(a) Co)
Bending moment (kNm) Bending moment (kNm)
-750 -500 -250 O 250 500 750 -750 -500 -250 O 250 500 7SO
80 80
500 500
60 400 60 400
40 300
200 -
40 300
200 -
5., 20
100 x s *O Io0 9:
P
h O o %l : o Y 0 :
3 -20 -100 ;5 -20 -100 r’
VY
.zoo -200
-40
-300 -300
-60 -400 -60 -400
-500 -500
-80 .so
-600 -400 -200 O 200 400 600
Bending moment (Kips-ft) Bending momeni (Kips-lt)
Fig. 9-Measured strain and calculated stress histories of a longitudinal column

reinforcement gauge located at the joint interface: (a) strain history; (b) stress
history; (c) stress history at p-2 (3 cycles); Id) stress history at p-2.8 (4 cycles)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Ob62949 0542945 157
6000
.
1 I I
0
-.-
e
5000
p=2 p = 2.8- 600
- 500
",
c2
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
5 4000
- - 400
- 300 oE
c
3000
o
E E
00
M
.-e 2000 - zoo .-
;
.o
C
Y en
1000
O I I I I I
1 loo
IO
0.0 0.1 0.2 0.3 O4 O5 0.6
ñond.slip rotation (rad.)
Fig. 10-Variation of bending moment against bond-slip rotation at the top of

column
881 k 2
ij
t
4I
2
I 1566 kN 1
1566 k N 45mm
168
m m
Fig. 11-Force-transfer model for the joint region of the test unit at p=2 under
opening moment
W Obb2949 054294b 093
Corvature (Vin.)
Fig. 12-Moment-curvature response in the column curvature cell located adjacent

to the joint interface
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m ObbS949 0542947 T2T D
SP 180-15
Pullout of Horizontal Reinforcement Embedded in

Masonry
by E. Vinttileou and N. Psilla
Synopsis: This paper presents the results of an experimental investigation on the

anchorage of deformed steel bars embedded in bed joints of masonry. The effect
of bar diameter, anchorage length and vertical load on the mortar joint is
investigated. Although the horizontal bars in masonry are under unfavourable
conditions, such as low strength of mortar, small cover values, lack of bond at
places where the bars pass over large holes of the masonry unit, tests have proved
that embedment lengths as low as 30 to 40 times the bar diameter seem to be
sufficient for the bars to develop their yield strength.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Keywords: Anchorage; bonding; deformations (reinforcing steels); masonry;

reinforcing steels
347
348 Vintzileou and Psilla
Elizabeth Vintzileou is an Assistant Professor at the Laboratory of Reinforced

Concrete, National Technical University of Athens. She is a member of CEB
Commission 2 (Material and Behaviour Modelling) and of several CEB Working
Groups. She has authored papers on shear transfer mechanisms, behaviour of
masonry infilled R.C. frames, seismic behaviour of reinforced concrete and
reinforced masonry structures.
Nikoletta Psilla is a post-graduate student at the Laboratory of Reinforced

Concrete, National Technical University of Athens. She is preparing her Doctor
Thesis on the behaviour of reinforced masonry.
INTRODUCTION
Reinforced masonry is a quite rapidly developing system in Europe,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
especially for use under seismic conditions. Normally, vertically perforated
I masonry units are used, laid in lime-cement mortar. Diffised vertical
reinforcement is placed in vertical holes (which are subsequently filled with
cement grout), whereas horizontal reinforcement is accomodated in bed joints.
The effectiveness of this reinforcement is greatly influenced by the anchorage
conditions which are rather adverse, especially in case of horizontal
reinforcement. In fact, horizontal bars are embedded in small thickness mortar
joints, which may not ensure sufficient cover to the bars. In addition, at places
where the bars pass over the vertical perforations of the unit, no cover at all is
provided to them. Thus, the available anchorage length is reduced. In addition,
the oblique (shear) cracks appearing in reinforced masonry walls due to seismic
actions create unfavourable anchorage conditions to the (horizontal and vertical)
bars placed close to the edges of the wall (Fig. l), since their available anchorage
length may not be sufficient, thus not allowing the wall to develop its full bearing
capacity.
Within a BRITEEURAM Project financed by the European Union.' the

pullout of reinforcement embedded in masonry was investigated. The complete
program of investigation includes both horizontal and vertical reinforcement
under monotonic or cyclic actions. This paper presents the part of the
I
The project entitled ((Industrial development of reinforced masonry buildings)) is
jointly carried out by brick producing firms (Philippou S.A., Greece and
Consorzio Poroton, Italy), by Contractors firms (Helliniki Technodomiki,
Greece and Vittorio Vigni, Italy), as well as by Universities (National Technical
University of Athens, Greece, University of Padova and University of Pavia,
Italy and Technical University of Darmstadt, Germany).
Obb2949 0542949 8T2 D
investigation which regards the horizontal reinforcement subjected to

monotonically increasing slips.
RESEARCH PROGRAM AND TEST SET-UP
Fig. 2 shows the geometry and dimensions of specimens used for the
investigation of pullout of horizontal reinforcement. For the construction of
specimens, the clay brick developed within the project in Greece was used. This
brick (Fig. 2) exhibits a mean compressive strength parallel to the holes equal to
8,5 N/mm2.Its compressive strengths perpendicular to the holes are equal to 2,20
N/mm2 and 1,15 N/mm2 (directions 1 and 2 respectively, see Fig. 2). It should be
noted that compressive strengths were measured over overall area of block. The
mix proportions of the mortar used throughout this part of the project are the
following: Cement: Lime: Sand = 1: 114: 4 (by volume). Its mean compressive
strength (measured on conventional cubic specimens 4Omm x 40mm x 4Omm)
was equal to 7,86 N/mm2, its mean flexural tensile strength being equal to 2,15
N/mm2. In Table 1, which summarises the research program, the respective
strengths per individual specimen are given.
The bed joints in all specimens were 15 mm thick (as required by Eurocode
6 for structural masonry). Thus, the top and bottom mortar covers provided to 6
mm bars were equal to 4 3 mm. For 10 mm bars, the mortar cover values were
reduced to 2,s nun. All bars had large side cover.
Fig. 3 shows schematicaily the experimental Set-up. The testing procedure

was the following: The specimen (2) is placed within the testing fiame (I) and
kept in its testing position by means of lateral screws (10). Vertical load
(corresponding to predetermined values of the vertical stress on the wail) is
applied by means of a mechanical jack (8). The vertical load is measured by
means of a load cell (9). It is corrected, when necessary, and kept constant during
testing. The reinforcing bar is anchored by means of a steel cone (4). All tests are
displacement controlled. Slips are imposed to the bar by means of a screw (5).
The imposed front slip, as well as the back slip of the bar are continuously
recorded by means of four electrid transducers (6), whereas a load *li (7)
allows for recordmg of the axial force of the bar. Strain gauges are positioned on
the bar, close to the front face of the specimen, in order to measure the strains of
the free part of the tensioned bar.
The parameters which were investigated are the following:
1. The normal stress applied to the mortar joint. To simulate real conditions, in
which there is a vertical load imposed on bed joints and in order to
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obh2949 0542950 514
investigate the favourable effect of this load on pullout behaviour, three

normal stress values were selected, namely: 0,05 N/mm2, 0,15 N / m 2 and
0,40 N/mm2.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
2. The bar diameter: 6mm and I O m m deformed bars were tested. Typical ribbed
reinforcing bars were used. Their mean o-E curves, as defined
experimentally, are shown in Fig. 4. The mean value of the yield strength of 6
mm bars was equal to 608 N/mm2, whereas their mean tensile strength was
equal to 672 N/mm2. The respective values for 10 mm bars were 492 N / m 2
and 61 5 N/mm2.
3. Anchorage length: Two lengths were considered, namely: one brick length
(320 nun) or two brick lengths (655 mm). Taking into account that at places
where the bar passes over the large brick holes (65mm x 65mm) there is no
bond (see Fig. 5), the embedment length of the bars was equal to 190 mm in
the short specimens and 39Smm in the long ones.
I
Table 1 summarises the testing program. Three specimens per combination
of parameters were tested.
TEST KESULI’S
Failure Modes
One has to distinguish between 320 mm and 655 mm long specimens: In the
short specimens, failure was due to the pulling out of the bar. After the
completion of tests and the removal of the upper brick, it was observed (Fig. 5a)
that a horizontal splitting crack had opened along the bar, whereas a ((triangular))
part of the mortar had separated and pulled out together with the bar. No damage
was observed at the back end of the specimens. In the long specimens, yielding of
the bars took place. At larger imposed slips, fracture of the 6 mm bars occurred
whereas, in specimens with 10 mm bars, a vertical crack appeared in the first
vertical mortar joint (Fig. 5b), i.e. at a distance of approximately 160 mm fiom
the front face of the specimen. In one case (shown on Fig. Sb), vertical cracking
of bricks has also occurred. In case of long specimens too, splitting cracks were
observed along the bars after completion of tests.
Axial Stress vs. Slip Curves
Fig. 6 shows some typical axial stress vs. slip curves obtained for 6mm and
1Omm bars. The axial stress at the front face of the specimen is plotted against the
slip measured both at front and at back face of the specimen. The following can
be observed on the diagrams of Fig. 6:
6 mm bars. 320 mm long suecimens- The axial stress of the bar at the fiont
face of the specimen is linearly increasing with the imposed slip until the bar
reaches its yield strength, while the back slip is equal to zero, indicating that
sufficient anchorage length is provided to the bar. Further increase of the imposed
front slip causes a further increase in the axial stress in the bar (by approximately
lo%, see Table 2). This increase of the axial force of the bar is associated with
penetration of yield along the bar and with the hardening of the steel in the
already yielded portion of the bar. At this stage, slip starts to be recorded also at
the back face of the specimen. After the maximum steel stress is reached, a very
steep falling branch is recorded, associated with an increase of the back slip
indicating that pulling out of the bar is taking place. A residual axial stress of
approximately 70 to 100 N / m z is recorded.
10 mm bars. 320 mm long specimens- Here again, there is a practically

linear initial part of the OSOvs. s curve (associated with zero value of the back
slip), up to a steel stress value depending on the external normal stress applied to
the bed joint, A less steep ascending branch follows (associated with the
occurrence of slip at the back face of the specimen) up to a maximum stress
value. The ratio between this maximum stress value and the stress corresponding
to the end of the initial linear part of the curve varies between 1,lO and 1,50 (see
Table 2). The higher values of this ratio occur for the lower normal stresses
imposed to the bed joint (Fig. 7). M e r the attainment of the maximum stress,
increasing back slips are recorded, whereas the axial stress is decreasing to a
residual value which again varies between 70 and 100 N/mm2.
6 mm bars. 655 mm lone; specimens-- In this case, the osovs. s curves are
similar to those described for 6 mm bars embedded in short specimens.
Nevertheless, since the anchorage length is large, even after the bar reaches its
yield strength, no slip occurs between the bar and the back face of the specimen
(see, for example, specimen E23 in Fig. 6, for which zero back slip values were
recorded throughout testing). Thus, the axial stress at the front face continues to
increase until it reaches the tensile strength of the bar which fractures.
10 mm bars. 655 mm long specimens-- The curves are very similar to those
of 6 mm bars in 655 mm long specimens. Nevertheless, there is a difference in the
failure mode. In case of 10 mm bars, failure is not due to the fracture of the bars
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ob62949 0 5 4 2 9 5 2 397 m
but to the opening of a crack along the first vertical mortar joint of the specimen
(see Fig. 5b). which causes an abrupt decrease of the axial stress in the bar. After
the appearance of the vertical crack within the mortar joint, the front
displacement values indicate the opening of this vertical crack rather than the slip
between bar and mortar along the bed joint. Thus, testing was not continued to
displacement values larger than 18 m.
DISCUSSION OF TEST RESULTS
A summary of the main experimental findings is presented in Table 2.

Column 2 in Table 2 gives the steel stress values corresponding to the end of the
linear part of the c -s curve for all specimens, and Column 3 gives the respective
front slip values. It should be noted that the back slip value which corresponds to
oso,luicar is equal to zero. In Column 4, the values of initial stiffness of the
mechanism are given (i.e. the values of Column 2 divided by those of Column 3).
In Column 5 the values of the maximum steel stress are listed, and in the last
I column of the Table, the (calculated) mean and maximum local bond stresses
developed along the bars are presented. Although the scatter of experimental
results is rather large, the following can be observed.
1 The experimental results indicate that the 6 mm bars were able to develop
their yield strength even for the small embedded length of 190 mm (Le. 32 times
the bar diameter approximately). This was proved also by strain gauge
measurements on the bars close to the front face of the specimens (Fig. 8). Taking
into account the adverse anchorage conditions of the bars embedded in a bed
joint, one would expect such short embedment lengths not to be sufficient.
Nevertheless, looking at the interface between brick and mortar joint after the
completion of tests, one may observe that during the construction of masonry,
mortar is introduced to the small vertical perforations of the bricks (see Fig. 5).
Thus, small mortar keys are formed. Those keys are confined by the walls of the
vertical holes and they provide large top and bottom mortar cover to the bars.
This may be an explanation for the unexpectedly satisfactory behaviour of the
bars.
2. In case of 10 mm bars, obviously the embedded length of 190 mm was

insufficient and, thus, the maximum axial stress developed in the bars of
specimens E4 to E6 and E10 to El 5 was smaller than their yield strength. On the
other hand, as expected, the maximum axial stress of the bars is influenced by the
magnitude of the normal stress imposed to the mortar joint. As shown in Fig. 9,
there is a substantial improvement of anchorage conditions of the 10 mm bars for
normal stress increasing from 0,15 N/mm2 to 0,40 N/mm2. No improvement is
observed, however, when the normal stress increases from 0,05 N/nim* to O,] 5
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Ubbi9Li9 05Y2953 223
N/mm2. The same holds true also for the steel stress which corresponds to the end
of the linear stage (See Table 2). It is also observed that the front slip value which
corresponds to CJJO,I-~ increases with increasing normal stress value (Fig. IO).
Thus, the initial stifiess of the mechanism seems to be practically independent of
the normal stress imposed to the bed joint (Fig. 11). On the contrary, the ratio
between the maximum axial stress and that corrseponding to the end of linear
stage seems to decrease with increasing normal stress (Fig. 7).
3. By comparing the behaviour of short specimens with 6 mm bars with that

of long specimens with 6 mm bars, one may observe that larger slips are required
in case of long specimens to reach their yield strength (see Table 2). An
explanation for this would be that the diagrams of axial and bond stresses along
the bar being smoother in case of larger anchorage lengths, the integration of axial
deformations of the bar along the specimen resultsto larger slip values.
4. On the basis of the maximum axial force carried by each specimen, the
mean and the maximum local bond stress was calculated (last column in Table 2).
The mean local bond stress was calculated as following:
where, A, is the area of the bar, db is the diameter of the bar and CI denotes
the embedment length. The values of maximum local bond stress T were
calculated assuming that z ,-2/3r
For 6mm bars and embedment length equal to 190mm, the calculated
maximum local bond stress values seem to be very high. Nevertheless, they are
comparable to those measured experimentally in (1) and (2). Modena and
Cecchinato have tested the local bond mechanism between deformed bars and
mortar in specimens in which the embedment length between bar and mortar was
approximately equal to 5 times the bar diameter. The lime-cement mortar they
used had a mean compressive strength equal to 6,O N/mm2 and a mean tensile
strength of 1,6 N/mm2. Although the mechanical chamteristics of the mortar they
used were lower than those of the mortar used in this program (compare with
Columns 5 and 6 of Table l), they measured local bond stresses as high as 7,O to
8,O N/mm2 for 6 mm deformed bars (and for zero normal stress).
Lower local bond stress values were calculated for 6 mm bars with
anchorage length of 395 mm. This is obviously due to the fact that, since the
available embedment length was larger than needed (as indicated also by the
fracture of the bars and the zero back slip values), the local bond capacities were
oniy partly mobilised.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542954 LbT
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
In case of 10 mm bars with an anchorage length of 395 mm, the calculated

maximum local bond stresses varied between 5,48 and 6,15 N/mm2. Again, these
values are in accordance with the experimental results by Modena and
Cecchinato, who measured local bond stresses as high as 4,70 to 6,40 N/mm2 for
10 mm deformed bars, under zero normal stress. The fact that there is a decrease
in the maximum mobilised bond stress for increasing bar diameter can be
attributed to the well known tendency of larger bars to induce larger splitting
stresses to the surrounding material (concrete or mortar). This may result in
required anchorage lengths increasing non-proportionally with the bar diameter.
On the other hand, the top and bottom mortar cover provided to 10 mm bars was
smaller than for 6 mm bars, since the mortar joint thickness was the same for both
diameters. This led to premature splitting ofthe mortar around the 10 mm bars.
Finally, substantially lower local bond stresses were calculated for I O mm bars
with insufficient anchorage length (1 90 mm). obviously because early longitudinal
splitting of the mortar caused premature failure of the mechanism of bond.
CONCLUSIONS
The anchorage conditions of horizontal bars in masonry are considered to be

very unfavourable due to the small mortar cover provided to bars embedded in
small thickness bed joints, as well as due to the reduction of anchorage lengths
because of the large vertical holes of masonry units. Nevertheless, the results of
this program have proved that for bar diameters normally used in reinforced
masonry construction (6mm and IOmm), lengths of 30 to 40 times the bar
diameter seem to be sufficient for full anchorage of the bars. This may be
attributed to the favourable erect of mortar keys formed by the mortar which
enters the small vertical holes of the bricks, thus providing large cover to the bars
along a large part of the anchorage length.
The tentative conclusion can also be drawn that for increasing bar diameter,
also the ratio between the required anchorage length and the bar diameter is also
increasing. This is probably due to the fact that the splitting stresses induced to
the mortar by the bar are larger for larger diameter bars, as well as to smaller
mortar cover provided to larger diameter bars.
REFERENCES
1 . Modena C . , Cecchinato P., ((Researches on the interaction mechanisms

between steel bars and hollow clay units masonry)), 4th North American Masonry
Conference, Los Angeles, 1987.
W 0662949 05g2955 OTb
2. Modena C., Cecchinato P., ((Bond of plain and deformed bars in block
masonry mortarjoints», 4th Canadian Masonry Symposium, i 986.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obh2949 0542456 " 3 2
TABLE 1-TESTING PROGRAM
E22 6 655 0,15 9,04 2,39

E23 6 65 5 0,15 7,09 1,68
E24 6 655 0,15 7,09 1,68
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
TABLE 2-SUMMARY OF EXPERIMENTAL RESULTS
m 0hb2949 0542958 805 =
Fig.1 -Adverse anchorage conditions of reinforcing bars in reinforced masonry
Idirection 1
320
Fig. 2-Geometry and dimensions of specimens
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Ob62949 0542959 74L
LAYOUT
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 3-Experimental set-up (schematic)
m 0bb2949 0542960 463 D
6mmbars
'O0 I I '
J '
600 11
Il 1 O mm bars
0.00 o. 10 0.20 0.3

Strain
Fig. 4-Stress-strain diagrams for reinforcing bars used in the program
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Obb2949 0 5 4 2 î b 1 3 T T Tribute to Peter Gergely 361
Fig. Sa-Failure mode of specimens: Longitudinal splitting crack and pulling out of
mortar at the front face
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. Sb-Failure mode of specimens: Vertical crack in the mortar joint and in the
brick
= íJhb2949 05429h2 23b =
,'-,.
\
i
\
',I
6 mm bar
\
I=32Omm 1.
a=O,15 N l m d E231
6 mm bar
1=655mm
o=O.15Nlmd
E4 IEllj
10 mm bar
\.
I O mm bar
\ 1=320mm 1=320 mm
2
o=O,OSN/md o=O, 15Nlmm
\,
~ .',. ,-
1-1 L I 1 I I LI L I , I
O 2 4 6 8 1 0 1 2 0 2 4 6 8 101214161
Slip (mm) Slip (m)
LG.1 rE20/\
\
' I O mm bar I
I O mm bar
1=655mm l\
-g 200 o=O,1 5N/mm2 \
"IL
D
0 2 4 6 8 10 12 14 i6 18 O 2 4 6 8 10 12 14 16 18
Slip (m) slip (mm)
Fig. 6-Typical steel stress vs. slip curves

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb2949 0542963 I172 D
o
o
0.8 -
I
--
0.00 0.20 0.40 0.60
Normal stress on the bed joint o\rm
l mz
)
Fig. 7-The effect of normal stress on the ratio between maximum and linear steel
stress (10 mm bars)
700
600 --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
6 m m bar
~ 0 ,SN/mm
l
100
O 1
0.000 0.020 0.040 0.060 0.080 0.100 0.12
Steel siraio
Fig. 8-Stress-strain diagrams for 6 mm bars as obtained from strain gauge
measurements
W Ob62949 0542964 O09 m
500
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
LI
i I
0.000 0,100 0.200 0.300 0.400 0.50
Normal stress on the bed joint (N/mmz )
Fig. 9-The effect of normal stress on the maximum steel stress
0.70 \
\ c,
0.60 i ii
I
0.50 t
Fig. 10-The effect of normal stress on the front slip
0bb2q4q 05429b5 T 4 5 Tribute to Peter Gergely 365
300.0 I 6 I I , J I
0.00 0.10 0.20 0.30 0.40 0.5

Normal stress (N/&)
Fig. 1 1-The effect of normal stress on the stiffness of the mechanism

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
SP 180-16
Effects of Transverse Reinforcement and Bonded

Length on the Side-Blowout Capacity of Headed
Reinforcement
by R. A. DeVries, J. O. Jirsa and T. Bashandy
SynoDsis: pullout tests were conducted on deepiy embedded headed

reinforcement to d e t e k the effect of transverse reinforcement and boded
length on the side-blowout capacity and load-slip behavior of the anchorage.
It was found that transverse ties or stirrups in the anchorage zone had little
effect on the ultmiate capacity. Increases in anchorage capacity were oniy
observed when the head was positiveiy anchored in contact behind a iarge
crossing bar. Transverse reinforcement also had little efkct on the load-slip
behavior before faihue. However, when large mounts of transverse
reinforcement were piaced near the head, the amount of load maintahed &er the
blowout failure occurred was increased. --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Additional bonded length of a deformed reinforcing bar increased the

anchorage capacity and reduced the head siip for a given load. The amount of
increase in capacity can be predicted using current AC1 provisions for
development length.
Design produrestaking into account the effects of transverse reinforcement

and bonded length were developed.
Keywords: Anchorage; bonding; reinforced concrete; slip; transverse reinforcement
367
Obb2949 0 5 4 2 9 b ï 818
368 DeVries, Jirsa and Bashandy
Richard A. DeVries, an AC1 Member, is an Assistant Professor of Civil and

Environmental Engineering at Oklahoma State University. His research interests
are in the areas of reinforced concrete design, repair and rehabilitation of
structures and hybrid construction.
James O. J i m , FACI, holds the Janet S. Cockreìi Centenniai Chah in

Engineering at the University of Texas at Austin and is chairman of the
Department of Civil Engineering. He is a member of AC1 408, Bond and
Development of Reinforcement, AC1 318, Standard Budding Code, and chair of
AC1 3 18-E, Shear and Torsion. He has served on the Board of Directors and as
chah of the Technical Activities Committee.
Tamk Bashandy, received his Ph.D. fiom the University of Texas at Austin in
December 1996. while at the University of Texas he conducted research on the
applications of headed reinforcement in concrete members. He is currently a
project engineer with RAST in Cairo, Egypt.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
INTRODUCTION
Recent advances in welding technology have made it cost efficient to weld

steel plates to ends of smooth or deformed bars creating a headed reinforcing bar
(Figure 1). It is proposed that headed reinforcement can replace hooked bars in
situations where there is insufficient development length for a straight bar
anchorage. Anchorage of headed reinforcement is provided by the bearing of the
head on the surrounding concrete. Using headed reinforcement may offer many
advantages over straight or hooked bar anchorages such as improved conñnement
of joints, the a b w to anchor larger bars than currently possible with hooked bars,
lower slip of reinforcement, improved d e t u , and improved response to cyclic
loading. In some applications using headed reinforcement may reduce the
congestion of the reinforcing cage by replacing tail extensions of hooked bars with
compact heads, which could reduce erection costs.
In spical applications of headed reinforcing bars in concrete members it is

expected that the primary mode of failure would be a side-blowout failure. A side-
blowout failure is characterized by spalling of a portion of the concrete cover over
the head and occurs when the ratio of embedment depth to side cover is large.
For example, a headed reinforcing bar used to provide anchorage for the
longitudinal reinforcement in a beam or column with minimumclear cover over the
bar might experience a side-blowout failure when placed in tension.
Obb29119 05429b8 754
Tribute to Peter Gargely 369
Based on the resuits of over 90 pullout tests on headed bars [i], side-
blowout îàiiures occur when the ratio of embedment depth to concrete cover is
greater than 5. when a deeply embecíded headed reinforcing bar is placed in
tension, the bearing of the head on the surrounding concrete creates a wedge of
crushed concrete on the head. The bearing of this wedge creates a lateral force
which ultimately blows out the side cover (Figure 2). The blowout capacity of
deeply embedded smooth headed reinforcing bars, as shown m Reference 1, can
be predicted by:
where PU is the ultimate anchorage capacity in kN, CI the minimum edge

distance to the center of the bar in mm, An the net beating area of the head in
mm2, and f, the concrete compressive strength in MPa. The term AdA-
accounts for reductions in the available faim surface area due to close spacing
of multiple bars or comer placement and is less than or equal to 1. A b is the
available failure area based on the dimension of 1.5C1 and A- is the nominai
failure surface area for a single bar placed away h m corners or other bars:
A,, = 3 K i
In addition to the reduced availabíe Mure surface area fiom corner placement,
the k t o r Y is another comer placement factor taking into account disturbances
to the assunied radiai stress pattern for a singie bar placed nem oniy one edge:
0.3- c* < 1.0

Y = 0.7 i- (3)
3c, -
where CZis the minimum edge distance perpendicular to CI. Equation 1 is a

best-fit equation based on test data. To change the equation to a design equation
that conservatively predicts the capacity, the constant of 58 in Equation 1 is
cbanged to 80.
It is assumed in Equations 1 through 3 that the head has suflicient stifñ~ssto

prevent yielding of the head m bending before yielding of the bar in tension. The
heads on all headed bars reported here had adequate thickness to satis@ this
assumption.
Equations 1 through 3 are based on the resuits of pullout tests on Wed

bars without considering bond along the bar or effects of transverse
reinforcement in the anchorage zone. The effects of these two parametm on the
uitimate blowout capacity are reviewed in this paper.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Obb2949 0 5 4 2 9 b ï b90
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Test SetuD and Variables
In the basic test setup for the deep-embedment puiiout test, a headed bar was
placed near the side or comer of a block of concrete with a sheath over the bar
to prevent any bond along the deformed bar. When the bar was placed in
tension, anchorage capacity was developed entirely by the head. Ratios of
embedment depth to clear cover greater than five were used. The center-hole
loading ram bore directly on the concrete where the bar protruded from the
concrete block. With deep embedments, the confinement effects of the bearing
reaction around the bar fiom the puliout force were found in an earlier phase of
the research to have little effect on the anchorage strength [i].
The embedment depth defines the size of a possible pullout cone. The
embedment depth, hd, is defined as the distance fiom the top, or loaded, surface
of the test specimen to the top of the head. In concrete members, the
embedment depth for headed reinforcement in tension would be deñned as the
distance, parallel with the bar, measured from the head, to a critical section (a
point of maximum stress in the bar). For aii tests reported here the embedment
depth was 762mm.
The bonded length, It,,for headed reinforcement is defined as the length from
the head along a deformed bar bonded to concrete. The bonded length of a
smooth reinforcing bar or bolt attached to a head is zero since there are no
deformations to bear on the surrounding concrete. The bonded length was
controlled in the test program by sheathing the bars with rigid PVC tubes. The
ends of the tubes were sealed with silicon caulk to prevent any paste from
bonding with the bar inside the sheath. The bonded length was always at least
15ûmm less than the embedment length so that the bearing reaction of the test
setup would not interact with the development of the headed bar along the
bonded length.
The minimum edge distance, CI, is defined as the distance, perpendicular to

the length of the bar, fiom the closest surface of concrete to the center of the
bar. The distance perpendicular to Cl from the center of the bar to the second
closest orthogonal surface is C2 which will always be greater than or equal to Cl.
AU dimensions measured paraiiel with CI are denoted with the subscript 1, and
all dimensions parallel with C2 are denoted with the subscript 2. The head
dimensions are denoted as hl and h2. In all tabulations and notations used
herein, the order of the head dimensions represents the head orientation with the
hl dimension being parallel with Cl and listed first followed by the h2 dimension,
parallel with C l , and the third figure representing the head thickness, t.
Obh2949 0542970 302 m Tribute to Peter Gergely 371
The effect of transverse reinforcement on the anchorage capacity of W e d

bars was studied using five mërent types of transverse tie codigwations. The
four types of ties tested for edge bars and the type used for corner ban are
shown in Figure 3. The conñguratkns of transverse reinforcement were chosen
to model typicai armgexnents of transverse reinforcement in concrete members.
The number of legs which provide a component (paralie1 to Ci for edge bars or
to CIor Ca for corner bars) of tensiie resistance across the assumed fidure piane
are included in Figure 3. AU tests with transverse ties had 305mm of bonded
length and ail ties were located along this bonded length. The number of ties, tie
diameter, da, distance h m the head measured along the bar to the first tie, SPI,
and spacing of the remaining ties, SP, were varied m addition to the tie
configuration The total area of transverse reinforcement along the development
length, A,,, is calculated by multiplying the cross-sectional area of the tie bar by
the number of legs per tie providing tensiie resistance and the number of ties.
The effect of anchoring a headed bar behind a crossing bar was also tested.
Anchorage behind a crossing bar occm when the head is in contact with another
bar so that when the headed bar is loaded in tension, the head wiii bear drectiy
against the crossing bar. Typical arrangements of bars in a b e a m - c o b joint
may result in headed reinforcement pkicsd behind a column bar which then acts
as a crossing bar. The diameter of the crossing bar is denoted as dbr.
Test Specimens
Concrete blocks with 4 to 12 headed bars anchored in the block were

constructed for the puilout tests. Each concrete block was 92- deep with the
bars cast around the perimeter. The plan dimensions of the blocks were either
92Omm by 9 2 k or 122onim by 122Omm. The bars were spaced to preclude
overlap of anticipated M ure surfaces. The blocks were cast in bolted plywood
forms and a series of spacers between the bars and the form were used to
maintain the desired cover over the headed bars and location of transverse
reinforcement during construction.
Materials
Concrete - The concrete was dehered by a local ready-mix company.

Nominal strengths ranged between 21 and 28 MPa. No admixtures were used
and all concrete mixes had 19mm (3/4 in.) maximum aggregate size. Ail mixes
were ordered with a 102mm slump, but the measured slump at the time of
castingvaried.
Control cyiinders were tesîed on the day puiiout tests we^ conducted to
determine the concrete compressive strengjh and these d u e s are reported as f,
mthispaper.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M û b b 2 9 4 9 0542971 249
Steel -The headed reinforcing steel was supplied by the sponsor. Three bar
s i s were used for the headed reinforcement: 20, 25 and 35mm diameter. The
majority of the bars had standard head sizes as designated by the
manufacturer/sponsor. In addition to the bars with standard head sizes, some
20mm and 25mm bars had 9Ox9Ox2Omm or lOOx55x25mm heads. These larger
heads increased the options for changing the head area anaor aspect ratio by
saw cutting the head for some of the tests. Nominal yield strengths of the
reinforcing bars and heads supplied by the sponsor was 500 m a . Transverse
ties were fabricated using Grade 60 #3, #4 and #6 (10, 12 and 20mmj bars.
Crossing bars were taken fiom the stock of headed bars supplied by the sponsor.
The properties of the reinforcing steel are listed in Table 1.
Testing Procedure
Each bar was tested monotonically in tension to Mure. The number of load
stages for each bar depended on the expected capacity of the bar but at least 8
load stages were imposed. Load, deflection, strains and crack widths were
measured at each load stage. Photographs were taken to record crack and
failure patterns. Loose (spailedj concrete cover was removed before the failure
surface was photographed. For some tests, additional loading after failure was
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
required to loosen the failed cover. For several tests with transverse
reinforcement, loading was continued past the peak load until the transverse
reinforcement fiactured.
Equipment and Measurements
Load was measured with a load cell and checked with a pressure gauge.
Head deflection was measured relative to the concrete block. A small hole was
driiied into the bottom of the heads and a rod was threaded into this hole. The
rod protruded through the bottom of the block. A copper tube which supported
the headed bar at the proper depth prevented bonding of the rod with the
concrete. A dial gauge or linear transducer was attached to the rod to measure
head deflections. Strains along the bar and on transverse reinforcement were
monitored using strain gauges attached to the bar surface and protected with
waterprootìng.
TEST RESULTS AND EFFECTS OF VARIABLES
Transverse Reinforcement
In most applications of headed reinforcement in concrete members, it is likely

there will be transverse reinforcement near the head. Since a side-blowout
failure involves the tensile capacity of the concrete cover, transverse
Obb2949 O542472 365 W Tribute to Peter Gergely 373
reinforcement could provide additional tensile strength and increase the capacity
of the anchorage. H a s s e M e r [2] made a preiimimy conciusion on îhe b i s
of two tests that transverse rehforcemenî did increase îhe side-blowout
anchorage strength of anchor bits with washers. Cotes and Wallace 131 also
concluded thaî transverse reinforcement would incl.ease the side-blowout
anchorage capacity of headed reinfolcement based on the results of knee-joint
tests.
A total of 22 bars (14 edge bars and 8 corner bars) were tested with
transverse reinforcement in the anchorage zone. The parameters and resuits for
these tests are summanzed in Table 2. For each test in Table 2, Pmi represents
the capacity normaüzed for 27 MPa concrete using a square root function to
reflect tensile strength of the concrete (a prime variable for blowout faiiures):
PUNI =.(E) (4)
where f is the measured concrete strengthon the day of testing.
In Figure 4, the normalized results h m 11 edge bar tests on 35mm bars with
55xlOox25mm heads are compared The oniy variable for these 11 tests was the
amount and location of transverse reinforcement. Transverse reinforcement had
no effect on the d t i i t e capacity in these tests. Results fiom the other groups of
tests with transverse rehfbrcement showed simiiar results. Neither the types nor
amount of transverse reinforcement significantly affected the ultimate capacity.
The main infiuence of transverse reinforcement was on the residual strengîh

after fruhire. For bars with no transverse reinforcement, the capacity of the
anchorage dropped to less than 30% of the ultimate capacity after blowout
failure occurred and continued to decline with increased displacement. Similar
behavior was observed for bars with transverse reinforcement placed away itom
the head. For bars with a iarge amount of transverse reinforcement piaced near
the head, the strength of the anchorage did not decrease as precipiiousiy at
faihire and dropped only after large head displacements were reached. A few of
the tests retained a residual strength of 80% of the ultimate capacity until the ties
fractured.
In Figure 5, the measured head slips for three tests are compared. The tests
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
were identical except for the amount and location of transverse reinforcement. A
large amount of transverse reinforcemeni near the head made little merence in
the ultimate capacity but substantiaüy improved post-peakperformance.
Strain gauges were placed on several transverse ties. In Figure 6, the tie
force cdcuiated h m the measured strains using a modulus of 200,000 MPa is
0hb2949 0542973 OLL
shown along with the measured head deflection h m one test. During the initiai
phase of loading both the tie force and head deflection were essentially zero,
suggesting that the compression strut and tension strut were parallei., making the
perpendicular tie force zero. As the loading increased both the tie force and head
deflection began to increase. Based on this observation, as the wedge of
concrete formed on the head, cracks may have begun to form in the cover
around the head and the head moved outward due to the bearing of the wedge
on the interior concrete, mobilizing the tie. As the load neared u l t i i t e , both the
head deflection and tie force increased rapidly. Also at ultimate, the cover
spalied off and the head moved outward with oniy the tie restraining it,
mob- the tie further. M e r failure the tie force held the head and concrete
wedge against the remaining concrete, ailowing the wedge to bear on the
remaining concrete providing anchorage untii the tie hctured.
In this group of tests, transverse reinforcement did not contribute to the

ultimate capacity of the anchorage. The ductility (maintaining load &er failure
through large deflections) of the anchorage was increased by placing a sufficient
amount of transverse reinforcement near the head (Atrof a single tie greater than
50% of the headed bar area and SPI less than 5Omm).
Anchorage Behind Crossing Bars
To test the effect of anchoring the head behind a crossing bar, a series of 8
tests were conducted on edge bars. Three of these tests had heads that were not
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
anchored behind crossing bars and five tests had headed bars anchored behind
25mm or 35mm bars. In addition to the presence of a crossmg bar, the head size
was also varied. For ail tests with crossing bars, the larger head dimension was
placed behind the crossing bar (bl > b2 and the length of the crossing bar was
parae1 with b). The parameters and results of these tests are iisted in Table 3
and in Figure 7 the results are compared. The addition of a crossing bar
increased the anchorage capacity by 7% to 28%. The increase in anchorage
capacity was higher for iarger crossing bars and larger head dmensions. In
Figure 8 the measured head slips for two tests with 9Ox35x2Omm heads are
compared. The presence of a 25mm crossing bar increased the ultimate capacity
by 23% and improved the s î f i e s s of the anchorage, especially at higher loads.
In Figure 9, the measured head slip for two tests with S5x55x25mm heads are
compared. The presence of a 25mm crossing bar increased the ultimate capacity
by 11% but had little effect on the head siip.
The crossing bars increase the anchorage capacity by mobilizing bearing

stress over a larger area which includes some portion of the crossing bar. This
increases the effective net bearing area of the head (A. in Equation 1).
Increasing the diameter of the crossing bar increases its surface area and stifiness
and the amount of bearing stress that can be transferred fiom the head through
Obb2949 0542974 T58 9
the crossing bar into the concrete. Increashg the siz~of the head improves the
transfer of stress to the crossing bar.
If the head extends to or beyond the center of the crossing bar (hi-db > dk),
bearing is direct@trans- to the crossing bar as shown in Figure loa. If not,
the effectiveness of transkrring bearing stress into the crossing bar is reduced as
shown in Figure lob. of course, if the crossing bar is not in contact with the
head, there is no direct transfer of bearing stresses as shown in Fw 1Oc.
Another mechanism through which crossing bars increase the anchorage
capacity is by providing resistance to the spaiiing of concrete cover. A crossing
bar is similar to transverse reinforcement type TE-2 (Figure 3) in the previous
group of tests. However, the crossing bars in this series which were intended to
simulate column bars, have much higher s ~ i f h s s e sand strengths than those
tested to simulate stirrups. The crossing bar can resist the spalling of the cover
through a shear or dowel action, provided it has suEcient anchorage on each
side of the headed bar beyond the Illnits of the blowout failure. in the tests
where the head did not bear directly (hl - d b < dk) on the crossing bar, the
anchorage capacity was stül increased due to dowel action but with less
effectiveness than direct bearing.
The increase in capacity due to a crossing bar is a function of severai

variables such as head size, crossing bar size and anchorage of the crossing bar.
However, due to the small number of tests conducted, an equation cannot be
formulated takhg into account d of these variables; but a simple factor
increasllig the capacity predicted by Equation 1 is proposed. Based on the test
setup and results, the anchorage capacity of a headed bar positively anchored
behind and bearing directly on a crossing bar can be increased by a &tor of
1.25. Positive anchorage is defíned by three criteria:
1. The head is in contaci with the crossing bar and the clear distance of the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
head behind the crossing bar is greater than one-halfthe diameter of the
crossing bar. This insures transfer of bearing stress fiom the head to the
crossing bar with no laterai component.
2. The crossing bar has a diameter greater than three-fourths of the diameter
of the heaâed bar b e i anchored. This insUres the crossing bar is stif€
and strong enough to transfer the bearing stress fiom the head to the
surrounding concrete.
3. The crossing bar extends a minimum distance of 6oOmm on each side of
the headed bar being anchored or other anchorage of the crossing bar
into the core concrete is provided beyond the limits of the blowout
failure. This insures the crossing bar had d c i e n t anchorage to resist
the iaterai thrust associated with blowout fidures.
m Obb2949 0542975 994
Development Along. Bonded Length
The behavior of deeply embedded headed bars is similar to that of deeply

embedded anchor bolts. One major difference between headed reinforcement and
anchor bolts is the presence of deformations along the length of the headed
reinforcing bar. Bearing of the lugs on the surrounding concrete along the
available bonded length may provide additional anchorage capacity. A total of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
eleven tests resulting in blowout failures were conducted with varying bonded
lengths to study the interaction of development (bond) and head bearing. In
addition, ali tests with transverse reinforcement, though none with crossing bars,
had bonded lengths greater than zero.
The eleven bars, six edge bars and five corner bars, were divided into five
groups with only the bonded length varied within a group. The parameters for
these tests are summarized in Table 4 and the measured capacities are compared
in Figure 11. In ali cases, but one, an increase in capacity was noted with
increased bonded length.
Additional bonded length changed the cracking behavior of headed bars.

When no bonded length was present, cracking was noted just before ultimate
load was reached, at which point cracks near the head appeared. For bars with
bonded length, a single crack formed at the beginning of the bonded length and
propagated along the bar down to the head as load increased. This crack did not
seem to affect the appearance of the final blowout failure. When the spalled
cover was removed, it was observed that concrete was sheared at the lugs along
the bonded length, indicating that stress was being transferred at the lugs.
Six tests with bonded length and transverse reinforcement were instrumented
with strain gauges located on the bar near the head. The bar force determined
fiom strain measurements is plotted against the applied load in Figure 12. Since
the gauge is located near the head, the vertical axis represents the portion of the
total load carried by the head. The diagonal lhe in Figure 12 represents the
force carried by the head ifthere were no bonded length, that is, the total load on
the bar. Under initial loading, the force carried by the head is very smaii and
nearly ail the bar force is being transferred to the concrete along the bonded
length. As the load increases the head carries an increasing portion of the load.
The cracking behavior observed indicates that radial stresses develop along the
bonded length. As the load increased and the crack along the bar opened, the
bond along the bar degraded and the portion of bar force carried by the head
increased. At failure the bonded length carried approximately 33% of the total
load based on data fiom six tests with strain gauges near the head.
The current AC1 318-95 [4] provisions for development length take into
account the effects of cover and transverse reinforcement on the bond strength of
m Obbè949 0542976 820 m Tribute to Peter Gergely 377
a defonned bar. Section 12.2.3 contains Equation 12-1 with aii terms in
customary us units:
where b is the development length required to reach the yield force of the bar, db
the bar diameter, f, the yield stress of the reinforcement, f, the concrete
strength, and c the smaiiest clear cover. The terms a,ß, y and h are factors for
reinforcement location, bEu coating, reinforcement size and lightweight aggregate
concrete, respectively. For this study of headed reinforcement, a,ß and .A are
1.O. The factor y is 0.80 for 2Omm bars and 1.O for 25mm and 35mm bars. Kh
is a transverse reinforcement index:
where Ah is the area of transverse reinforcement resisting the spiitting failure, ffi
the yield stress of the transverse reinforcement, s the spacing of the transverse
reinforcement, and n the n&r of bars being developed. For the tests on
headed reinforcement s was taken as the development length divided by the
number ofties and n was 1.
Assuming a linear relationship, the bond force for the bonded length of a
headed reinforcing bear can be caiculated by:
i
F -LF, (7)
Id -1,
where b is the bonded length provided, b the development length fiom Equation
5, and FYthe yield force of the bar based on the yield stress f,, either nominai or
actual,used in Equation 5.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
For two tests which are identical except for the amount of bonded length or
transverse reinforcement, adding the difference between the predicted bond
forces h m Equation 7 to the capacity of the first test in the pair should predict
the capacity of the second test. The tests on headed bars with bonded length can
be grouped into 35 pairs of tests where the oniy merence between tests in a pair
is the bond force predicted by Equation 7. On average the ratio of the measured
capacity of the second test divided by the measured capacity of the first test plus
0662949 0542977 767 W
the difference in predicted bond force was 1.02 for the 35 pairs with a standard
deviation of 0.08. The maximum ratio was 1.19 and the minimum was 0.85.
The AC1 318 provisions reflect the increase in capacity from bonded length
reasonably weil. The analysis indicates that Equation 7 is accurate even when
the provided bonded length is less than 300mm (the minimum development
length ailowed in AC1 318). It is conservative, however, to ignore the
contributions of bonded lengths less than 30ûmm.
In Figure 13, the load-slip responses of three bars are compared. The
variables for these tests are identical except for the bonded length. From the
figure it can be seen that with an increase in bonded length, there was an increase
in the initial stiffness of the anchorage.
DESIGN RECOMMENDATIONS
The foliowing design equation can be used to estimate the side-blowout

capacity of deeply embedded headed bars:
where h equals 1.25 if positive anchorage behind a crossing bar is provided and
1.0 otherwise. In the second term of Equation 8, lb is the bonded length of a
deformed bar provided, b the development length from Equation 12-1 of AC1
3 18-95, and FYthe yield force of the headed bar. The bonded length, ib, must be
greater than 300mm and the ratio of Idld is ümited to 1 in Equation 8. The
measured capacities of the puilout tests reported here are compared with the
predicted capacities from Equation 8 in Figure 14. In can be seen that Equation
8 is conservative and allows the designer to consider the beneficial effects of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
anchorage behind crossing reinforcement and bonded length. The average value
of the measured to predicted capacity based on Equation 8 is 1.55 with a
standard deviation of 0.24 compared with 1.45 and 0.23 for the test data upon
which Equation 1 is based (the portion of Equation 8 in brackets).
Obb2e149 05Y2978 bT3 Tribute t o Peter Gergely 379
CONCLUSIONS
Based on the resuits of over 40 puiiout tests on deeply embedded headed

reinforcmg bars,the fobwing conclusions can be made:
1. Transverse reinforcement did not affect the blowout anchorage capacity.

Large amounts of transvem reinforcement placed near the head
increased the amount of load that was maintained &er the blowout
faihue occurred.
2. Anchoring the head behind a crossing bar increased the side-blowout
anchorage capacity 25%, provided the head was bearing directìy on the
crossing bar. In order to obtain this increase, the diameter of the crossing
bar had to be grater than three-fourths of the diameter of the headed bar
and the crossing bar had to be anchored in the core concrete beyond the
limits of the blowout Mure.
3. Bonded length of a deformed bar increased the side-blowout anchorage
capacity of deeply embedded headed bars and decreased the head slip.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
The increase in capacity could be predicted using current AC1 provisions
for development length.
For the design of anchorage of headed bars it is conservative to ignore the

conmiutions of bonded length and crossing bars on the anchorage capacity.
However, adjustments to the basic design equation permit consideration of the
benefits of these parameters.
ACKNOWLEDGMENTS
This research was conducted under the sponsorship of the Headed

Remforcement Corporation The research could not have been completed
without the assistance of Kjell Dahl and David Mitcheii of HRC and the staff of
the Ferguson Engineering Laboratory at the University of Texas.
REFERENCES
[i] DeVries, R A., “Anchorage of Headed Reinfircement in Concrete,”

PhD. dissertation,University of Texas at Austin, 1996.
[2] HasseIwander, G. B., Jirsa, J. O., Breen, J. E., and Lo, K, “Strength and
Behavior of Anchor B o b Embedded Near Edges of Concrete Piers,’,
Research k r t 29-2F,University of Texas at Austin, May 1977.
m Ob62949 0 5 4 2 9 7 9 53T E
[3] Cotes, P., and Wallace, J. W., “A Study of Reinforced Concrete Knee-
Joints Subjected to Cyclic Lateral Loading,” Report No. CU/CEE-94/1l,
Ciarkson University, Potsdam, Febniary 1993.
[4] American Concrete Institute (ACI 3 18-95), Building Code Reauirements

for Structural Concrete and Commenta, American Concrete Institute,
Detroit, 1995.
TABLE 1-PROPERTIES OF REINFORCING STEEL
r, 514 I 515 508

f,, ( M W 669 I 660 660
.Bar Diameter (mm) 10(#3)

I 12(#4) I 20(#6)
2, (MPa) 503 I 434 I 420
I f,(MPa) I 72 1 I 709 I 682
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 05r12980 2 5 1 m
c
Y
Y
o
œ
Y
z
W
a
W
VY
a
W
>
VY
æ
4
a
+
I
k
3
VJ
I-
Y
I-
Y
O
VY
! I
a
CI)
W
œ
n
æ
e
VJ
œ
w
L
2
a
3
N
I
W
A
m
U
I-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542983 398
TABLE 3-PARAMETERS AND RESULTS OF TESTS WITH CROSSING BARS
TABLE 4-PARAMETERS AND RESULTS OF TESTS WITH BONDED LENGTH

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
-~
m Obb2949 0542982 024 Tribute to Peter Gergely 383
Fig. 1 -Headed reinforcing bar
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= 0662949 0542963 Tbû
Fig. 2-Side blow-out failure
- -
l
I
l
I
-
I
I
I
I l
I I
-.
I
I
I-
I
-1
I
-1 I
I I
:---I
I I
I
Type: TE-1 TE-la TE-2 TE-3 TC-1

#ofLegs: 1 2 O 2 2
Fig. 3-Configurations of transverse ties
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542984 9T3
35mm Bars with 55xlOûx25mm Heads
900
4 (m2)
157 157 314 314 O O O O 226 628 628
600
W
E
n" 300
Test ID
Fig. 4-Comparison of capacities of tests with transverse reinforcement
Load vs. Head Slip
0.00 2.00 4.00 6.00 8.00

Head SUP (mm)
n
RC5B2
No Ties
102mm-f+
C9B1
2 -1ûmm Ties
C13B2
1 - 2 h Tie
Fig. 5-Comparison of head slip of tests with transverse reinforcement
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Dbb2949 0542985 8 3 3 m
Tie Force
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Test ID: C13B1
90 2.00
1.50 .s
C
1.00 3
*E
0.50 8
U
O 0.00
O 200 400 600
Load (kN)
Tie Force = O Tie Force > O Tie Force >> O
Fig. 6-Comparison of tie force and head deflection
Effect of Crossing Bar

W No Crossing Bar O 25mm Crossing Bar O 35mm Crossing Bar
90x35~20 70~44x20 55X55X25

Head Dimension (mm)
90x35~20 70x44~20 55X55x25
Fig. 7-Effect of crossing bar on capacity
œ 0662949 0542986 ïïT œ Tribute to Peter Gergely 387
Load vs. Head Slip
0.00 1 .OO 2.00

Head Slip (mm)
Fig. 8-Effect of crossing bar on head for bar with 90 x 3 5 x 20mm head
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
.---------_--------....---.-------------
Bar Yield Load
g
n
400 --
_ _ . . _ _ _ _----. -. - - - - - - - - ..__._,
-2 5 m Crossing Bar - P14

NO Crossing Bar - P4
--....
I
, +?&
P hl hl I
Positive Anchorage Not Pwitive Anchorage Not Positive Anchorage
I
dhdk
I+hi*l I+
+
;':?
%
hdb dbc h 4 b < dbc No Contact '

(a) (b) (c)
Fig. 1O-Conditions for positive anchorage behind crossing bar
0bb2949 0542987 bob
Bonded Length
900
' 6oo
300
Fig. 11-Effect of bonded length on capacity
Bar Force at Head

Test ID: C9B3
900
Fi
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
i600
U
c)
a
i 300
i
i o
O 300 600 900
Total Applied Load (kN)
Fig. 12-Comparison of head force and total load
0bb2949 0542988 542 Tribute to Peter Gergelv 389
Load vs. Head Slip
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 1 %Effect of bonded length on head slip
Measured Pu vs. Predicted Pu

-
Equation 8 Comprehensive Design Equation
900 -
No Crossing Bar
o Crossing Bar
O 300 600 900

Predicted P" (kN)
Fig. 14-Comparison of predicted and measured capacities
~
W Obb2949 0542989 489
SP 180-17
Performance of Concrete Bridge Decks and Slabs

Reinforced with Epoxy-Coated Steel under
Repeated Loading
by H. O. Hasan, D. B. Cleary and J. A. Ramirez
Svnovsis:
This experimental study addresses the performance of reinforced concrete

slabs containing epoxy-coated reinforcement subjected to high-cyclellow stress
range repeated loading typical of those encountered in bridge decks. The behavior
under repeated load indicated that epoxy-coated reinforcement does not
significantly increase deflectionsdespite the larger bar slip associated with wider
cracks. The wider cracks do increase the potential for increased amount of
corrosive agent at the level of the top mat of reinforcement in bridge decks. The
average bond strength ratios of coated to uncoated specimens support a proposed
single modification factor of 1.35 for specimens with low cover. --```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Keywords: Bond (concrete to reinforcement); crack width; coatings; concrete

width; cyclic load; deflections; deformed reinforcement; epoxy resins; fatigue
(materials); lap connections; splicing; structural design
391
m 0662949 0542990 LTO
392 Hasan, Cleary and Ramirez
AC1 Student Member, Hendy O. Hasan currently serves as the coordinator of

Private Universities, Region II, Ministry of Education and Culture, in Palembang,
Indonesia. He received his Civil Engineering degree from Catholic Parahyangan
University in Indonesia in 1980, his MSCE from Purdue University in 1990, and
his Ph.D. from Purdue in 1994.
AC1 Member Douglas B. Cleary is a structurai and coastal engineer with Black
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
& Veatch. He received a BSCE in 1987, a MSCE in 1988, and Ph.D. in 1992
from Purdue University. He was a visiting assistant professor at Purdue from
1992 to 1994.
AC1 Member Julio A. Ramirez is a professor of structural engineering at Purdue

University. He received a BSCE degree from Universidad Automoma de Mexico
(ITESO) in 1977, an MS from the University of Texas at El Paso in 1979, and a
Ph.D. from the University of Texas at Austin in 1983. Dr. Ramirez is chairman
of the Joint ACI-ASCE Committee 445, Shear and Torsion; and a member of
Joint ACI-ASCE Committee 423, Prestressed Concrete; and AC1 Committee 408,
Bond and Development of Reinforcement.
INTRODUCTION
Changes in bond properties due to the presence of epoxy coating have long
been recognized. These changes include reduction of the ultimate bond strength
and differences in the service load behavior of members with coated bars.
Numerous studies have shown reductions in the ultimate bond strength from 5 to
35 percent caused by epoxy-coated reinforcement, especially in splitting-type
bond failures (1-9). These past studies have included beam-end and splice
specimens with several sizes of reinforcement, with and without confining
reinforcement, subjected to static and/or repeated loading. The reduction in bond
strength, due to a loss of friction between the steel and concrete, particularly along
the front face of the ribs, is accounted for in the design specifications for
development length in the AC1 Committee 318-95 Building Code (IO). Also,
fewer but wider flexural cracks have been associated with the use of epoxy-coated
reinforcement,
Presently, these paucity data in the case of bridge deck type members
under repeated loading. This experimental study addresses the structurai
performance of members containing epoxy-coated reinforcement subjected to
high-cycle/low stress range repeated loadings common to concrete bridge decks
and slabs. The behavior of companion specimens with coated and black steel is
ObbZq4’ 0542’q3 Tribute to Peter Gergely 393
compared under repeated loading and at failure to assess the impact of the epoxy-
coating. Low-cyclehigh-stress range repeated loading was not part of this study.
The laboratory experimental program was conducted at the Karl H.

Kettelhut Structural Engineering Laboratory in Purdue University. It consisted
of a series of 34 reinforced concrete slab specimens that model a bridge deck
section. Two types of specimens were used (see Figure 1). The Type A specimen
was 13 feet (3.96 m) long with a cross section of 24 inches (610 mm) wide and
8 inches (203 mm) deep. Reinforcement consisted of 3 #7 (22 mm) bars spaced
at 6.5 inches (165 mm) with a i2 inch (305 mm) splice at midspan. The Type B
specimen was also 13 feet ( 3.96 m) long, but with a cross section of 28 inches
(711 mm) wide and 12 inches (305 mm) in depth. Type B specimen
reinforcement consisted of 3 #11 bars (35 mm) spaced at 8 inches (203 mm) with
a 28 inch (71 1 mm) splice at midspan. Type A and B specimens had 2.5 inches
(63.5 mm) of clear concrete cover and transverse reinforcement of #3 (10 mm)
bars spaced at 6 inches (152 mm). The concrete cover resulted in cover-to-bar-
diameter ratios (C/db) of 2.86 and 1.82 for specimens Type A and Type B
respectively. These ratios are toward the midrange of previous studies. A ratio
of 3.0 was used by Cleary (4) whereas a ratio near I .O was used by Treece (2) in
earlier work. Both of the c/db ratios used in this study are less than 3.0 and
therefore require a 1.5 development length factor in accordance with current code
(10). In addition a factor of 1.3 is required for a Class B splice. The development
length may be reduced for the Type A specimens by 0.8 because of the cover.
The splice was designed to fail in splitting mode prior to yielding of the
longitudinal steel. This is standard practice in evaluating bond strength
experimentally. The specimens were loaded 6 inches from the ends and had a 4
foot constant moment region at the midspan. Loads were applied with a hydraulic
actuator. Of the 34 slab specimens, 30 slabs were tested under repeated loading
and four were tested with a single monotonic load cycle.
i n the test program, the specimen dimensions, including the beam length,
width, depth, bar size, concrete cover, and splice length were held constant for a
given type of specimen. The concrete strength, peak stress and stress range for
repeated loading, bar deformation pattern and coating thickness varied. The
combination of variables used for the 34 tests is shown in Table 1. In this table
the first letter of the specimen code indicates epoxy-coated or uncoated steel. The
11 or 7 that follows is the bar size and the next two digits are the peak stress, in
ksi, used for the repeated loading. The last digit is the repetition number for the
listed variable combination. For the extra coating thickness specimens, the D
indicates a diamond deformation pattern, S indicates a spiral deformation pattern,
and X indicates extra coating thickness. The variable combinations were selected
to cover a range of load cases. Although not commonly coated today, and not
permitted in some areas, the diamond pattern was used for a few tests. The
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0542992 T33
diamond pattern was not an intended variable however, the bars were provided by
the supplier and could potentially be sent to a job site in Indiana.
The maximum and minimum stress values used in the repeated loading
were selected to fall in the upper limits of what are considered service loads for
fully developed bars. In allowable stress design, 0.6 f,, is used as an upper stress
limit. This corresponds to 36 ksi ( 250 MPa ) for Grade 60 steel and to 24 ksi
(165 MPa) for Grade 40 steel, thus upper stress limits of 36, 30, and 24 ksi (250,
210, 165 MPa) were selected for the repeated loading.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Each test specimen was initially cracked by application of 2 or 3 static

load cycles, up to the load corresponding to the peak stress used in the repeated
load test. After the initial cracking, the specimens were loaded at a rate of 260
cycles per minute, limited by the testing machine and specimen stiffness. The
repeated loads were applied in blocks of approximately 100,000 cycles for the
first 300,000 cycles after which blocks of approximately 150,000 cycles were
used. Between loading blocks, reinforcement strains, end and centerline
deflections, and flexural crack widths were measured. After one million load
cycles, if failure had not occurred, the specimens were then loaded monotonically
to failure.
BEHAVIOR UNDER REPEATED LOADING
The behavior under repeated loading observed in this study was similar to
that seen in Cleary’s study (4). This same behavior was also observed in previous
static load tests (2,3). It was found that the total deflection, defined as the sum of
the end and centerline deflections, of beams with epoxy-coated reinforcement
were only slightly larger than for beams with black steel. In this study beam
deflections were 5.4 percent larger initially and approximately 3.5 percent larger
after one million cycles of load. These results occurred for both Type A and Type
B specimens.
As with past studies (2,3,4), flexural cracks were fewer but wider with
epoxy-coated reinforcement. For both Type A and Type B specimens the average
width of a flexural crack in the constant moment region was 30 percent larger,
0.023 in. (.58 mm) for Type A and 0.0129 in. (0.33 mm) for Type B, during the
early load cycles; this difference reduced to 20 percent after one million cycles,
0.025 in. (0.64 mm) for Type A and 0.014 in. (0.36 mm) for Type B. The total
crack width is defined as the sum of the widths of ali flexural cracks in the
constant moment region. For Type A specimens there was no difference in the
total crack width between specimens with coated or black steel. This was not true
with Type B tests. For the larger bar size it was found that the total crack width
was 25 percent greater with coated steel in the first cycles and still 12 percent
greater after one million cycles.
0bb2q4q 'OT Tribute to Peter Gergely 395
The thickness of epoxy coating or the use of a diamond deformation

pattern had no effect on cracking loads or initial flexural crack widths. The effect
of the coating thickness on deflections was inconclusive. For Type A specimens,
a thicker coating increased specimen deflections by approximately 7 percent. The
Type B specimens with a thicker epoxy coating had deflections approximately 5
percent lower than similar specimens with a standard coating thickness.
Four specimens with coated steel failed during the repeated loading phase
of the tests. Specimen E7362 failed after 600,000 cycles, E736SX failed after
710,000 cycles, El 1303 failed after 336,000 cycles, and E1130DX failed after
1,700 cycles. All of the specimens which failed during this phase had epoxy-
coated reinforcement. In the two Type A specimens, the peak repeated stress used
was only slightly below the stress which caused failure of the majority of the other
slabs with coated steel. In addition, specimen E726SX had an epoxy coating
greater than the standard. Thus it is not surprising that these two Type A
specimens would fail during the repeated loading portion of the test. In the Type
B specimens however, the failure stress was significantly lower than the typical
failure stress for the remaining beams. Specimen EI 130DX had both a diamond
deformation pattern and a considerable extra thick epoxy coating with an average
value of 26.1 mils, both factors shown to significantly reduce bond strength (5).
The early failure of this specimen indicates the bond problems introduced when
a diamond deformation pattern and extra thick epoxy coating are combined.
Specimen El 1303 on the other hand was subjected to both a large peak stress and
a large stress range. The failure stress is somewhat lower than found for the other
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Type B specimens. It should be noted, however, that no other Type B specimen
was subjected to as severe a combination of stress range and peak stress.
BEHAVIOR AT FAILURE AND BOND RATIO
Tables 2 and 3 contain failure loads and failure reinforcement stress for the
specimens tested in this study. The reinforcement stress was calculated assuming
a Hognestad stress distribution for unconfined concrete with a peak stress value
of 0.85 ff in the concrete compression zone and no contribution from the
concrete in the tension zone. The specimens were designed for one million
cycles by limiting the stress range in the reinforcement to less than 20 ksi (13)
(1 38 MPa) and also by keeping the stress range less than 40 percent of the steel
yield strength (14). As mentioned above, four specimens did fail during the
repeated loading phase. Ail slabs had a splitting bond failure over the bar splices
except for U1 130SX which failed in shear.
The bond ratio for the Type A tests can be found in column 5 of Table 2,
and for the Type B in column 5 of Table 3. For the Type A specimens the average
bond ratio (defined as the ratio of reinforcement stress in the specimen with
coated steel to that of a companion specimen with uncoated steel) was 0.78. The
range of ratios was 0.64 to 0.95. If the two slabs which failed during fatigue
0662949 0542994 846
396 Hasan, Cleary and Rarnirez
loading are removed, the average bond ratio for Type A specimens was 0.83 with
a range from 0.73 to 0.95. For the Type B specimens the average bond ratio was
0.75. The range of ratios was 0.58 to 0.86. If the two slabs which failed during
fatigue loading are removed the average bond ratio for Type B specimens was
0.79 with a range from 0.75 to 0.86. The bond ratio results, with and without the
removal of values from the specimens failing during the repeated loading phase,
fall in the range found in the previous studies.
The tests conducted in this study indicate a reduction of bond ratio with
increasing concrete compressive strength for both Type A and Type B specimens.
For Type A specimens, neglecting the beams that failed in fatigue, the average
bond ratio for the beams with concrete compressive strength of 3000 psi (21 MPa)
was 0.88. For the 4000 psi (28 MPa) test, the strength ratio was 0.84. For the
4700 psi (32 MPa), test the ratio was 0.82. And for the 5200 psi (36 MPa) test
(series 7361), the ratio was 0.73. For the Type B specimens, the average bond
ratio for the specimens with concrete compressive strength of 3000 psi (21 MPa)
was 0.84. For the 4700 psi (32 MPa) test, the ratio was 0.79 and for the 5200 psi
(36 MPa) to 5300 psi (37 MPa) tests, the ratio was 0.76. In these comparisons not
all of the test variables have been held constant. In the cases in which all
variables except concrete compressive strength are constant the same trend occurs.
In comparing series 7241 and 7242, the bond ratio was 0.95 with 3000 psi (21
MPa) concrete and 0.82 with 4700 psi (32 MPa) concrete. For series 1 1241 and
11242, the bond ratio was 0.86 with 3000 psi (21 MPa) concrete and 0.79 with
4700 psi (32 MPa) concrete. Between series 7363 and 7361 only the stress range
and concrete strength varied, the peak stresses were the same in each series. In
series 7363 with 4000 psi (28 MPa) concrete, the bond ratio was 0.84. In series
7361 with 5200 psi (36 MPa) concrete, the bond ratio was 0.73. The influence of
concrete strength found in the current study is consistent with results of previous
studies (3,7,11).
BOND EFFICIENCY
The basic development length factor according to Section 12.2.2 of the

AC1 3 18-89 Building Code is given by:
0.04Abf ,
,1 = * (rnod@ution factors).
fi (1)
Substituting 1, for l,,, f, for f,, and solving for f, yields,

1 .#
f' = 0.04Ab (modification factors)
For both types of specimens, the applicable modification factors are 1.3
for a class B splice according to section 12.15, and 1.5 for epoxy-coated
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662749 0542995 782 m
reinforcement according to section 12.2.3.4 of the AC1 318-95 Building Code.

The modification factor for bar spacing, cover, and transverse reinforcement from
section 12.2.3.4 is 0.8 for the Type A specimens and 1.0 for Type B specimens.
Equation (2) with four combinations of modification factors is used to calculate
the reinforcement stress following AC1 for the beams tested in this study. Results
of these calculations are given in columns 6 through 9 of Table 2 for Type A
specimens, and Table 3 for Type B specimens. The first case is the bond without
the splice or epoxy coating factors (Col. 6), the second case reflects the
application of the splice factor only (Col. 7). The third case (Col. 8) uses the
epoxy factor only, and the last case (Col. 9) includes both the both the splice and
epoxy factors. The first two cases apply to the specimens with black steel and the
last two cases are intended for coated bars. The bond efficiency, defined as the
ratio of the measured to AC1 calculated failure stress, is given in columns 10
through 13 of Tables 2 and 3, for each of the four cases respectively.
The main conclusion from these comparisons is that the simultaneous

application of the splice and epoxy coating factor results in more conservative
results for beams with coated reinforcement than those obtained for the specimens
with black reinforcement. Column 13 of Table 2 shows the average bond
efficiency for the Type A coated specimens was 1.87 with the coating and splice
factors while Column 11 shows that the average bond efficiency with black steel
was 1.60 with a splice factor. Similar results are found in Table 3 for the Type B
specimens. The bond efficiency was 2.47 for the specimens with coated bars and
2.21 with black steel. Only in the four tests that ended during repeated loading
was the bond efficiency with coated bars less than or equal to that with uncoated
reinforcement. It must be remarked that two of the four failures during the
repeated loading portion of the test were associated with extra coating thickness
over the current limit of 12 mils. Of the other two, one can be attributed to peak
stresses near ultimate values and the other to a severe combination of peak stress
and stress range.
CONCLUSIONS
The behavior under repeated loading found in this study is similar to the
behavior observed in past works. It appears that epoxy-coated reinforcement does
not significantly increase deflections despite the increased bar slip that is indicated
by larger flexural crack widths. The presence of fewer cracks negates any
increased rotation and thus deflection due to additional initiai slip.
The increased width of cracks with coated reinforcement can be seen as

a potential cause of concern as these cracks could result in larger quantities of
corrosive agents at the level of the reinforcement. As long as the epoxy coating
remains intact, corrosion would be prevented or at least delayed.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0 b b 2 9 4 î 0 5 4 2 î î b b1iî
In this study average bond strength ratios of 0.78 for Type A specimens
and 0.75 for Type B specimens were found. The reciprocal of these values 1.28
or 1.33 indicates that the proposed single modification factor for spliced epoxy-
coated bars of I .35 for specimens with low cover proposed by Hester et. al. (6)
would also be appropriate for the work presented in this paper. The factor was
shown applicable up to the #11 bar size used in this study. The more conservative
bond efficiencies found in this study for coated bars, when both the splice and
coating factors are used, also supports Hester’s (6) recommendation of a single
modification factor.
Of the variables considered in this study three seem to be important. It

was observed that the diamond deformation pattern for reinforcement, when
combined with extra coating thickness, has poor bond qualities (5). The failure
of the beam with the extra thickness of coating and diamond deformation pattern
after 1,700 cycles of loading demonstrates this finding. In addition one of the
Type A specimen failures during the repeated load phase had an extra thick epoxy
coating as well. The loss of rib area due to the extra coating can be critical during
repeated load application if stresses are to be taken close to the upper bond limit
as was done in this study. Finally, increased concrete compressive strength
lowered coated bar to uncoated bar bond ratios.
ACKNOWLEDGMENT
The work described formed part of the doctoral dissertation of the first
author, at Purdue University, with the collaboration of the second author, and
under the direction of the third author. The research was funded by the Federal
Highway Administration and the Indiana Department of Transportation. Any
opinions, findings, and conclusions expressed in this paper are those of the
authors, and do not necessarily represent the views of the sponsors.
CONVERSION FACTORS
1 in. = 25.4 mm
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
1 kip = 4.448 kN
1 Ksi = 6.895 MPa
REFERENCES
1. Johnston, David W. and Zia, Paul, “Bond Characteristics of Epoxy Coated

Reinforcing Bars,” Report No. FHWALVC/82-002, Department of Civil
Engineering, North Carolina State University, Raleigh, Aug. 1982, 163
PP.
O b b 2 ï r i ï 0542997 555
2. Treece, R. A. and Jirsa, J. O., “Bond Strength of Epoxy-Coated

Reinforcing Bars,” ACIMaterials Journal, V. 86, No. 2, Mar.-Apr. 1989,
pp. 167-174.
3. Cleary, D. B. and Ramirez, J. A., “Bond Strength of Epoxy-Coated

Reinforcement,” ACI Materials Journal, V. 88, No. 2, Mar.-Apr. 1991,
pp. 146-149.
4. Cleary, D. B. and Ramirez, J. A., “Epoxy-Coated Reinforcement Under

Repeated Loading,” AC1 Structural Journal, V. 90, No. 4, Jul.-Aug. 1993,
pp. 451-458.
5. Choi, Oan Chui; Hadje-Ghaffari, Hossain; Darwin, David; and McCabe,

Steven L., “Bond of Epoxy-Coated Reinforcement: Bar Parameters,” ACZ
Materials Journal, V.88, No. 2, Mar.-Apr. 1991, pp. 207-217.
6. Hester, Cynthia J.; Salamizavaregh, Shahn; Darwin, David; and McCabe,

Steven L., “Bond of Epoxy-Coated Reinforcement to Concrete: Splices,”
ACI Structural Journal, V. 90, No. 1, Jan.-Feb. 1993, pp. 89-102.
7. Hamad, Bilaal S . and Jirsa, James O., “Influence of Epoxy Coating on

Stress Transfer From Steel to Concrete,” Proceedings, First ASCE
Materials Engineering Congress, Denver, Aug. 1990, pp. 125-134.
8. Hamad, Bilaal S.; Jirsa, James O.; D’Abrea de Paulo, Natalie I.,
“Anchorage Strength of Epoxy-Coated Hooked Bars,” ACZ Structural
Journal, V. 90, No. 2, Mar.-Apr. 1993, pp. 210-217.
9. Hamad, Bilaal S . and Jirsa, James O., “Strength of Epoxy-Coated

Reinforcing Bar Splices Confined with Transverse Reinforcement,” ACI
SfmcturalJournaZ,V. 90, No. 1, Jan.-Feb. 1993, pp. 77-88.
10. AC1 Committee 318, Building Code Requirements for Reinforced

Concrete (AC1 3 18-89) and Commentary (AC1 3 18R-95), American
Concrete Institute, Detroit, 1995,353 pp.
11. Grundhoffer, T.; French, Lorentz C.; and Leon, R., “Bond Strength and
Durability of Coated and Uncoated Rebar,” Proceedings, Third NSF
Workshop on Bridge Engineering, Research in Progress, La Jolla, Calif.,
NOV.1992, pp. 3 1-34.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0542998 491
12. Hasan, Hendy O., “Structural Performance of Concrete Bridge Decks

Reinforced with Epoxy-Coated Steel Under Fatigue Loading,” a thesis
submitted to the faculty of Purdue University in partial fulfillment of the
requirements for the degree of Doctor of Philosophy, Purdue University,
West Lafayette, Indiana, 1994, 300 pp.
13. AC1 Committee 2 15, “Considerations For Design of Concrete Structures

Subjected to Fatigue Loading,” ACZJournal, Vol. 71, No. 3, March 1974,
pp. 97-121.
14. AC1 Committee 408, “State of the Art Report on Bond Under Cyclic
Loading,” AC1 408.2R-92, American Concrete Institute, Detroit, 1992, 32
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
PP.
TABLE 1-TEST PROGRAM
Specimen Size
Bar
U7241 I #7
I
%%+-
IE7242 #7
U7361
E736 1
U7362 I #7
E7363 I #7
U736SX I #7
E7STAT I #7
U11241 I #11
E11242 I #11
U11243 I #11
E11301 I #11
U11302 I #11
E11303 I # l l
UllSTA I #11
311STAT #11
üïï%i&K
El 130SX I #11
3
D
i 1 1 1 1 ;I 1
9.2
15.5
2.5
2.5
2.5
2.5
28
28
28
3.8
5.2
5.2
-
8.74
8.74
1
l,OOO,OOO
l,OOO,OOO
JI(zoDxI#II D
3113ODX #11 26.1 2.5
28
28
4.3
4.3
30
30
8.74
8.74
l,OOO,OOO
1,700
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0 b b 2 9 4 9 0543000 748 m
I-
w
I-
l
N
I-
L
-
.-
LmL
W
.-u
OD
Lr.
I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0543001i b84
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
-H
404 Hasan, Cleary and Ramirez rn 0543002 510
6"
Plan View
Cross Section
P P
Moment
Shear
Specimen Bar Size a b C d s timp

(in.) (in.) (in.) (in.)
#7 24 8 2.5 6.5 #3 @ 6in.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
TypeA
TypeB #11 28 12 2.5 8 #3 @ 6 in.
Fig. 1-Test specimen
Obb29Y9 0543003 Y57
SP 180-18
Design Recommendations for Epoxycoated

Reinforcement
by J. Cairns, J. O. Jirsa and S. L. McCabe
S y n 0 - e Fusion bonded epoxy coated reinforcement (FBECR) has been developed

to help combat problems of corrosion in reinforced concrete structures. The surface
texture of the coating is smoother than the normal mill scale surface of reinforcing
bars and alters bond characteristicsof the bar. Although FBECR has now been in use
for more than 30 years and production Standards have been established, rules for
design using the material are not well developed. CEB Task Group 2/5 is currently
reviewing data on bond and structurai performance of elements reinforced with
FBECR with the aim of deriving recommendations for design practice which will
enable structures reinforced with FBECR to achieve equivalent performance to that
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
of structures reinforced with millscale surface ribbed bars. This paper presents
proposals for amendments to the CEB-FIP Model Code 1990 for design of
anchorages and splices of coated bars, and briefly reviews other aspects of structurai
performance influenced by the different bond characteristics of FBECR.
Keywords: Anchorage; bond; deflections; epoxy coating; lapped splice;

reinforcement; tension stiffening
405
Ob62949 0543004 393
406 Cairns, Jirsa and McCabe
John Cairns is a Senior Lecturer at Heriot-Watt University, Edinburgh, Scotland. He
has been a member of CEB Task Group 2/5 'Bond Models' since its inception in
1993, and convenes its Working Parties on epoxy coated and on corroding
reinforcement. He is also active in research into assessment, repair and strengthening
of deteriorated concrete structures.
James O. Jirsa holds the Janet S. Cockrell Centennial Chair in Engineering at the
University of Texas. He is a member of AC1 Committees 3 18, Standard Building
Code; 408, Bond and Development of Reinforcement; and 369, Seismic Repair and
Rehabilitation. He is also a member of the CEB Bond Models Task Group.
Steven L. McCabe is Associate Professor of Civil Engineering at the University of

Kansas He is an active member of CRSI, ASCE, CEB and AC1 among others.
Currently, McCabe is the chair of AC1 Committee 439, Steel Reinforcement, a
member of Building Code subcommittee 3 18-B, Development and Reinforcement,
and a member of the CEB Bond Models Task Group.
INTRODUCTION
Epoxy coated reinforcing bars were developed to improve durability of concrete

structures. The epoxy coating is intended to resist corrosion of reinforcement by
providing a physical barrier to the oxygen and moisture necessary for the corrosion
reaction, and by interrupting electrical continuity of the corrosion cell. The surface
texture of an epoxy coating is much smoother than the normal millscale surface of a
hot rolled reinforcing bar, and af€ectsthe interaction, or bond, between reinforcement
and concrete.
Bond characteristics of reinforcement influences several aspects of the structural

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
performance of reinforced concrete. A review of the design rules in the CEB-FIP

Model Code shows that at the serviceability limit state, bond influences crack widths
and deflection. At the ultimate limit state, strength of laps and anchorages depends
on bond. In addition, bond characteristics of flexural reinforcement will influence
rotation capacity of plastic hinges.
There is general agreement that bond strength and stiffness of coated bars is lower
than that of nominally identical 'black' bars. The strength reduction is generally
estimated at between 10% and 25% in research studies. However, bond strength
ratios (ratio of bond strength of fusion bonded epoxy coated reinforcement (FBECR)
to that of non-coated or 'black' bar) of 0.54 to 1.1 have been reported. The diversity
of bond strength ratios reported in research literature is reflected in the range of splice
length factors proposed in national Codes of Practice for design, with increases in
splice length of between zero and 50% required in different Codes. At the
serviceability limit state, national Codes either take epoxy coatings to have no effect
on performance or ignore potential effects.
0bb2949 0543005 22T
The Comite Euro-International du Beton (CEB) is a major European and world

research group working on pre-standardisation development studies. Its aim is "to
synthesise research findings and technical information and translate them into
practice". Task Group 2/5, Bond Models, was established in April 1995 to further the
work of CEB in this particular area. From the foregoing it was evident that practical
guidance on design using epoxy coated bars was capable of improvement. The
Authors, all members of the Task Group, agreed to undertake development of
supplementary design rules to enable the CEB-FIP Model Code 90( i) to be applied
to design of structures reinforced with epoxy coated bars. This paper reports on the
part of this work concerned with development and splice length of epoxy coated bars.
The reader is referred to a recent CEB Bulletin on Protective Systems for
Reinforcement (2) for information on manufacturing processes, handling procedures
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
and durability matters.
Aim
The aim of this paper is to review the influence of an epoxy coating on bond and
anchorage strength of reinforcement, and to develop recommendations for revisions
to Model Code 90 to enable structures to be designed using epoxy coated bars to a
margin of safety consistent with that for ordinary non-coated reinforcement.
STANDARDS FOR FBECR
Coated reinforcing bars are manufactured to ASTM775(3) in the USA. in the UK,
BS7295(4) for coated reinforcement was published in 1990, and in Japan coated
reinforcement is produced in conformity with Specifications of the Japan Society of
Civil Engineers(5). More recently, ASTM Standards have been introduced for
prefabricated bars(6) and for welded wire fabnc(7). German and Norwegian
Standards also are reported to be in preparation. Although epoxy powder may be
applied by electrostatic or tribostatic spray or by fluidised bed dipping techniques,
Standards apply only to spray methods of application to individual straight lengths
of bar.
A minimum coating thickness of 0.2mm has been generally accepted as meeting both
durability and structural performance requirements, and this is reflected in most
production Standards and materials specifications. Results reviewed later in this
paper are based on coating thicknesses at this value and applied by spray methods.
REVIEW OF PARAMETERS INFLUENCING BOND STRENGTH OF

FBECR
This section of the report reviews the findings of research studies into the factors
which affect anchorage bond performance of fusion bonded epoxy coated
m Obb2949 054300b Lbb m
reinforcement. The form adopted is to compare performance of coated bars against
that of equivalent non-coated bars under identical conditions. The objective is to
identifj those parameters which affect the relative bond strength of coated and non-
coated bars. a brief review of the bond strength rules in the 1990 CEB-FIP Model
Code(1) is first presented to help identifj the important parameters. Relevant Clause
numbers in the CEB-FIP Model Code 90 are given in square [ ] brackets
The basic design bond strength,fbd, is defined by Equation 1, Equation [6.9-41 in

MC90.
fbd = vi.qZ.v3&d Equation 1
-L
Bar diameter
I l
1 - Casting position
Reinforcement type
where -ql -
a coefficient for the type of reinforcement, given as 2.25 for ordinary
ribbed bars.
v2 - a coefficient reflecting the reduced strength of bars cast in slipform
construction or near the top of a pour.
q, - a coefficient which considers the reduced bond strength of bars over
32mm in diameter.
fcid - design concrete tensile strength.
The basic anchorage length is then given by Equation 2, Equation [6.9-51 from
MC90.
Ib = @#rfyd/fbd Equation 2
where 4 is the bar diameter, fydis the design yield strength of the bar, and & is the
design bond strength fiom Equation 1.
The design anchorage or lap (splice) length is given by Equation 3, Equation [6.9-6/71
from MC90.
Equation 3
Proportion of reinforcement lapped

Transverse pressure
Transverse reinforcement
Cover
Welded transverse bars
Straighthookedhent
where the several a coefficientsrepresent various factors influencingthe design bond

length as indicated.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m 0662949 0543007 OT2 m Tribute to Peter Gergely 409
Test data on epoxy coated bars are now reviewed to examine the innuence exerted
by the various parameters identified above on the relative bond strength of coated and
non-coated bars.
[3.3] Bar Type - Coating and Rib Geometry (q,)
No evidence is reported which suggests that the proprietary brand of epoxy powder
used to coat the bars has any effect on bond behaviour (8),(9). Several studies using
different types of bond test specimen have shown that variations in thickness of spray
coatings within the range 0.13mm-0.3Omm,the range permitted by ASTM A775, do
not influence bond strength significantly'(lO)(ll). Choi et al(12) also agree that
variations in coating thickness within the allowable range have an insignificant
influence on bond strength ratio for No. 6 (19mm) and larger bars, but reported that
bond strength ratio dropped by around 0.08 with No. 5 (16mm) bars as coating
thickness increased from 0.13mm to 0.30mm. Even with the thickest allowable
coating of 0.30mm, however, the bond strength ratio for the No. 5 bars did not drop
significantly below that for the larger bars. Hamad and Jirsa(13) also concluded that
variations in coating thickness within the range permitted by ASTM A775 had no
influence on bond strength raîio for No.6 and No. 11 bars. In a short series of pullout
tests, Kobayashi and Takewaka(l4) report an average reduction of 4%, half that
measured by Choi et al, for equivalent parameters. However, tests by Abrishami et
al on beams( 15) and on slabs( 16) indicate that coating thickness has a significant
infiuence on crack widths and possibly on tension stiffening.
The uniformity of coating thickness, and consequent modification to rib geometry,

may vary according to the degree of control attainable at the coating plant. Because
of difficulties in obtaining a holiday free film, the diamond rib pattern employed in
the US is no longer supplied coated. Some studies report the shape of a rib to be
altered by coating, that the rib becomes rounded over, and that the effective height of
the rib is reduced. Effects of non-uniformity of coating thickness will be greater on
smaller diameter bars where the coating forms a greater proportion of bar diameter.
Given the relatively small influence measured by most investigators, the effect of
coating thickness may be of little practical importance given the strength advantage
conferred by high cover ratios where small bars are used.
Cairns and Abdullah (8) report a series of pullout tests on pairs of coated and non-
coated machined bars of constant relative rib area and varying rib face angie, Figure
1. Rib face angle clearly influenced the bond stress ratio of coated to non-coated bars
at low values of free end slip, but at high slips (of 1.Omm and beyond) the infiuence
of coating was minimal. Maximum bond strengths in these specimens were high
( > l o m a ) , and failure generally occurred in a pull-through mode,Figure 2(a). The
dependence of the bond stress ratio on the slip at which it is measured probably
BS7295 imposes a tighter specification of 0.1 5mm-0.25mm.

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
- 0662947 0543008 T39
reflects a difference in the location of the slip surface. At low slip, baríconcrete
movement will occur at the bar concrete surface, as shown in Figure 2(b) or (c). At
high slips, barhoncrete movement will occur on a surface running across the tops of
the ribs, as indicated in Figure 2(a), and it is evident that the surface condition of the
bar would then have a minimal influence on bond performance. Hamad and Jirsa (13)
also tested manufactured bars with various rib face angles, using an eccenîric pullout
test specimen. The 30" bars exhibited the lowest bond strength ratio, as would be
expected from earlier studies on rib geometry by Gergely and Lutz( 17), but the 45"
bars exhibited a higher bond strength ratio than the 60" bars. Bond strengths were
relatively high (>lOMPa) and it is reported that the steeper rib face angles tended
towards a pull-through failure mode, whereas the shallower ribs tended to fail by the
ribs riding over the concrete keys, in a manner akin to a splitting failure.
In tests on specimens representing lapped splices Cairns and Abdullah showed that
the strongest epoxy coated bars could. develop nearly the same bond strength as
weaker black bars (9). Their statistical analysis of test results suggested that the
weaker rib patterns, with lower relative rib areas, generally exhibited smaller bond
strength reductions, Figure 3. In a later study of lapped splices, Darwin et al(l8)
reached the opposite conclusion, namely that epoxy coating had a less detrimental
effect on splice strength for more highly ribbed bars. In a series of beam end tests,
Choi et al (12) also reported that bond strength of bars with a higher relative rib area
was less affected by coating. However, splice tests conducted as part of the same
study do not show the same dependence on rib pattern. The bar pattern exhibiting the
greatest reduction in the beam end tests exhibited the smallest reduction in the splice
tests. It is worth noting that the bond strengths attained in the beam end tests were
around four times greater than those attained in the splice tests, and may have
affected the failure mechanism. The difference trends may therefore be associated
with differences in confinement. The opposing conclusions of these various studies
have still to be accounted for. However, it would appear to be prudent to introduce
limitations on the rib geometry of bars to be coated within Standards for FBECR.
Both Cairns and Abdullah and Darwin and Idun( 19) suggest that their may be a rib
face angle beyond which coated and non-coated bars would develop the same bond
strength. This may be attainable if the rib face angle of coated bars was steep enough
to ensure that splitting failures took place by shearing of the concrete below the
bearing face of the rib, Mode 2, Figure 2(b), and not by Mode 3. Although there is
some experimental evidence to support this suggestion, it appears to be based on
machined bars with ribs perpendicular to the bar axis. In modern rolled bar profiles,
ribs are generally inclined to the bar axis to give a uniform cross sectional area along
the length of the bar.
[6.9.3] Design Bond Strength

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Casting Position [q2]-- Compaction of concrete tends to be poorer close to the top
of a lift, and settlement of fluid concrete may lead to formation of voids below bars
cast near the top of a lift. Bond strength is dependent on the quality and compaction
m 0662949 0543009 975 m
Tribute to Peter Gergely 41 1
of the surrounding concrete, and bars cast in locations near the top of a lift tend to be
weaker in bond than those cast near the bottom. Hadje-GhafYari et al(20) used beam
end test specimens up to 1.Om deep in an extensive study to examine the interaction
of slump, casting position and compaction. They found bond strength of coated bars
to be affected less by ‘poor’ casting position thanblack bars,with the ‘top cast’effect
being negligible for coated bars cast in high slump concrete. Taking other bond
strength factors into account, they suggested the top bar factor for coated bar should
be around 14% higher than for black bar. Tests conducted at BCA(21) produced a
similar conclusion.
- --
Bar Diameter [fi3] None of the studies conducted to &te at BCA(21), Kansas
(12) or Texas (1O),( 13) in which the ratio of bar diameter to cover thickness was
maintained constant indicate a significant dependence of bond ratio on bar diameter.
Concrete Strength - Ifd] -- Jirsa and Treece (10) tested lapped splices in beams
constructed using concrete cylinder strengths over a range of 27MPa to 88MPa, but
did not detect any influence of concrete strength on the ratio of bond strengths for
coated to that for non-coated bars. Cleary and Ramirez(22) found no significant
difference in bond strength ratio between specimens manufactured using concrete
strength of 27.6MPa and 48.3MPa. Hamad and Jirsa (13) reported concrete strength
in the range 35Mpa to 65MPa to have only a minor influence on bond strength ratio.
In contrast, Hasan et al report average increases in bond stress ratio from 0.88 to 0.73
as concrete strength increased from 21N/mm2to 36N/mm2in ‘Type A’ specimens,
and from 0.84 to 0.76 as concrete strength increased from 21N/mm2to 36N/mm2in
‘Type B’ specimens(23). As will be described later, an analysis of splice strength
results fiom a number of studies could not discern a significant influence of concrete
strength on bond strength ratio.
Anchorage and Lap Length -- Little reliable test data is available on bond strengîh
of coated bars over a range of anchorage or lap lengths. Many test series on lapped
splices either contain only a very narrow range of lap length ratios (lap length
expressed as a multiple of bar diameter) (12),( 13) or simultaneously vary lap length
ratio and other important factors (10),(22),(24).
Three studies contain tests in which lap length has been varied independently
(21),(25)(26), and results are summarised in Figure 4. Note that the various studies
all employed different lap details and quantities of confining reinforcement, and
results are therefore not directly comparable. Bond strength was clearly reduced by
îhe epoxy coating. The reduction varied from 30% for one of the shortest to 12% for
the longest lap tested. In all cases average bond strength of non-coated bars was
found to reduce with lap length, but lap length had littie influence on average bond
strengîh of coated bars. The difference between coated and non-coated bars therefore
tended to reduce with increasing lap length. All three studies indicate that, unlike
ordinary black bars, bond strength of epoxy coated laps is substantially independent
of lap length. Beam end tests conducted by Cusens and Yu(27) provide further
supporting evidence, based on comparisons of bond stress at critical values of slip,
that the difference in average bond strength between coated and non-coated bars
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
reduces with increases in bonded length.
Results from puliout tests have demonstrated that an epoxy coating reduces bondlslip
stiffnessof reinforcement. Using a RILEM pullout type specimen, Cairns & Abdullah
(8) showed bond stifhess to be reduced by an epoxy coating by around 20% at slips
of 0.25mm, the slip at which bond failure of a lapped splice has been observed to
occur. It can readily be demonstrated by techniques such as Modulus of Displacement
Theory, as developed by Tepfers for analysis of bond stresses along a lapped
splice(28), that mean bond stress at failure will decrease with increasing lap length.
The reduction in mean bond stress with increasing lap length has been found to be
less marked for bars with more flexible bondhlip characteristics, however.
Design Anchorage and Lap Lengîh i6.9.5 & 6.9.6)
Hooks and bends [a,]-- Hamad et al report that anchorage capacity of hooks and
bends are reduced by coating, although the differential between coated and non-
coated bars is slightly less than for straight bar anchorages(29).
Welded Fabric [a,] -- Tests by Bartoletti and Jirsa have indicated that
epoxy-coated welded-wire fabric has essentially the same development and splice
strengths as uncoated fabric since the cross wires provide the primary anchorage for
the wire (30). Rehm gives no indication of any problems in anchorage of epoxy
coated welded fabrico 1). King reports adequate anchorage strength with welded
cages coated by the fluidised bed method (1 l), where the thicker coating tends to
smooth over bar deformations.
Confinement [ a ,ad,a,] -- Bond action of ribbed bars generates bursting stresses

which tends to split the surrounding concrete. Bond strength in a splitting failure
mode is considered to be limited by the splitting resistance of a concrete hoop around
the bar, together with additional resistance from any confining reinforcement crossing
the failure plane. Tensile capacity of the concrete hoop is assumed to be dependant
on the lesser of minimum cover thickness or half the clear spacing between adjacent
lapped splices. Semi-empirical regression analyses of test data demonstrate that bond
strength of non-coated bars increases with confinement to the bars from increased
cover, transverse reinforcement, and lateral pressure.
In tests using beam end type specimens, Choi et al (12) demonstrated that the
absolute increase in bond strength with concrete cover is similar for both coated and
non-coated bars, and hence that the difference in bond strength between coated and
uncoated bars (expressed as a ratio) reduces with increasing cover. The change in
bond strength ratio depended on the precise basis on which the comparison was
made, but averaged 0.88 to 0.93 as cover increased from 2 to 3 bar diameters. The
change in bond ratio reported by Hamad et al (29) for an increase in cover from 1 to
2 bar diameters was a more modest 0.72 to 0.74.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2949 0543011 523 m
Hamad et al tested special eccentric pullout tests in which one half of the bar was
embedded in concrete, and a confining force could be applied directly to the other
face. No very clear effect of confining force on bond strength emerged fiom the
study. This result may be attributable to the altered mode of failure as the 'top load'
was increased. The confining force applied was probably somewhat greater than
would be expected in a lapped splice, for example.
Hamad et al and Hester et al(32) have both examined the influence of confining
stirrups on bond strength of coated reinforcement. Both demonstrate that bond
strength of coated bars and of non-coated bars is increased by the provision of
stirrups. in a series of 12 tests on beam laps, Hamad et al report average increases in
bond strength ratio fkom 0.74 where splices were unconfined by stirrups to 0.81 and
0.84 where increasing amounts of confining reinforcement were provided for No. 11
bars, both coated and non-coated. For No. 6 bars, bond strength ratio increased from
0.67 to 0.74 with the addition of stirrups. However, when Hester et al analysed results
fiom their own 65 test specimens along with the 12 of Hamad et al, they concluded
there was no significant variation in bond strength ratio for nominally identical
specimens with varying amounts of confining reinforcement. Differences between the
test specimens in these studies may have influenced measured bond ratios and their
analysis. For example, the range of confining reinforcement investigated by Hester
et al is greater than that used by Hamad et al, and extends beyond the limit, suggested
by Orangun et ai(33), to which additional quantities of confining reinforcement
increase bond strength. Nonetheless, the effect of increasing confinement on bond
strength ratio remains rather unclear. Both increased cover and quantities of confining
reinforcement would be expected to exert a similar effect. However, results reported
to date suggest that bond strength ratio increases with increasing cover, but may not
be increased by confining reinforcement.
Sustained and Cyclic Loading -- Clifton and Mathey report that creep of epoxy
coated bars is of the same order as that of black bars at service stresses(34). Similarly,
Johnston and Zia(35) concluded that behaviour of coated bars under cyclic loading
was essentiallysimilar to that of black bars, although a tendency for the differential
to diminish as the number of cycles increased was noted. In a series of splice tests,
Cleary and Ramirezalso found repeated loading to be less detrimental to coated bars
(24).
Fire resistance -- Fusion bonded epoxies are thermosetting polymers that cannot
be turned to a molten state by application of heat, although they may pyrolyse at
higher temperatures. The material does soften around the glass transition temperature
of approximately 110°C, however. pullout tests conducted on coated and black
reinforcement cast into a reinforced concrete slab(36) under fire conditions showed
that all test bars were able to attain their yield strength in the test, and that the
reduction in bond strength attributable to coating was not significantly affected by
elevated temperature. Coated bars displayed markediy greater free end slips than
companion black bars at temperatures of 110°C and above. However, scatter in
measured bar slips was considerable. Although it is evident that the coating material
softened at temperatures above 110°C and that bond behaviour was affected, it should
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
414 Cairns, Jima and McCabe
= 0665747 0543032 4bT =
not be concluded that fire resistance is lessened by use of epoxy coated bars.
However, it is necessary to bear in mind reservations concerning the relevance of
pullout type tests to practical situations set out above.
Summary
From the foregoing review it can be seen that :
- there is no evidence that variations in coating type and thic..ness (within the
range permitted in standards for coated bars, or bar diameter (for a given
coverhar diameter ratio) need be given individual consideration when
examining bond reduction factors for design.
- an assessment of bond performance based on performance of straight bars at

ambient temperatures under monotonic short term loading should tend to
give bond ratios which are reasonable or conservative for other
circumstances.
- different bond strength factors are likely to be appropriate for top and bottom
cast positions.
- evidence on the influence of concrete strength, rib geometry and

confinement, whether from concrete cover, stirrups or lateral pressure, on
bond strength ratio is in conflict to some degree.
- anchorage of welded wire fabric will not be affected by epoxy coating.
DERIVATION OF PROPOSED MODIFICATIONS TO BOND STRENGTH
The objective of this section is to determine the value of bond coefficients for use in
MC90 to enable structures to be designed using fusion bonded epoxy coated
reinforcement to the same margin of safety as that currently used for ordinary (black)
reinforcing bars. Consideration of the safety margins to be assigned to bond failure
lies outwith the scope of this paper. Only bars produced in accordance with a
recognised standard for fusion bonded epoxy coated reinforcement such as ASTM
A775 (3) or BS7295 (4) are considered.
Although there is widespread agreement on which parameters exert an influence on

bond strength, the form of the various expressions that have been proposed to relate
bond strength to these parameters show considerable variation. It is not the purpose
of this review to consider the merits of the various expressions. The form of bond
strength expressions used in MC90 has been retained. The value of the coefficients
used for black bars in these expressions have, however, been temporarily adjusted in
some instances to provide a better fit to experimental data.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0543033 3Tb
The approach adopted in derivation of FBECR bond strength d e s is to ñrsî analyse
individual results fiom 7 studies comprising around 200 tests on lapped splices.
These studies have largely been carried out on paired sets of nominally identical
specimens, one specimen of each pair reinforced with ordinary black bars, the other
reinforced with FBECR. Bond strengths measured on specimens reinforced with
black bars have first been analysed to determine the margin of safeîy currently used
in MC90 and the statistical precision of the MC90 design rules. Results from
specimens reinforced with coated bars are then analysed, and coefficients in the
MC90 expressions adjusted so that :
a) the ratio of measured to MC90 design strength is independent of the

various parameters identified as influencing relative strength of coated
and black bars.
b) a similar overall ratio of measured to MC90 design strength is
obtained for coated and for black bars.
Proposals derived from the analysis of the strength of lapped splices are then further
validated against conclusions from anchorage tests.
MC90 rules for strength of lapped splices of black bars -- Jirsa and Treece (1O),
Choi et al (12), Cleary and Ramirez (22),(24), Hamad and Jirsa (13), Hester et al (32)
and Cairns and Abdullah (25) have all reported experimental investigations in which
the strength of lapped splices of epoxy coated bars has been compared with that of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
nominally identical lapped splices of black bars. Details of the specimens tested and
the measured bond strengths cannot be fully reported here, but will appear in a CEB
Bulletin. It should be noted that several test specimens did not comply with detailing
requirements of MC90 for minimumcover or for minimum transverse reinforcement.
The format of the MC90 bond strength rules has been set out above. The ratio of the
measured strength of lapped splices of ordinary black ribbed bars to the design
strength calculated from these rules was obtained, and the influence of the various
parameters listed in the preceding sections on this ratio was analysed where possible
to verify that the safety margin embodied in the MC90 niles was consistent across the
range of test parameters. Concrete strength, bar diameter, and the quantity of
confining reinforcement did not exert a significant innuence on the measureddesign
strength ratio within the range of available data.The general form of Equation 1 thus
appears to be satisfactory. Similarly, there is no evidence that changes are required
to expressions for q3or for u4,u, or u6.
MC90 overestimated the adverse effect of poor casting position, and underestimated
the beneficial influence of extra cover on splice strength, Figure 5. in order to
improve the fit to experimental dah, the casting position factor q2 was taken as 0.9,
and not as 0.7 as specified in MC90. Cover factor 0 1 ~is derived fiom Equation 4,
taken from [Table 6.9. i], using a value of h=O.15. To provide a better fit, MC90
design strength of black bars has instead (for the time being) been based on a value
of h 4 . 2 1 .
9 Obb2949 0543034 232
~ 1 , = 1- ka*(Cd- $)/a Equation. 4.
The mean and the coefficient of variation of the ratio of measured to MC90 design
bond strength so obtained were 2.47 and O. 17 respectively.
Strength of lapped splices of epoxy coated bars

The ratio of the bond strength of a lapped splice of epoxy coated bars to that of a
companion specimen reinforced with black bars in the 7 studies ranged from 0.95 to
0.54, with an average of 0.77. The simplest approach to modifjing MC90 rules for
coated bars would be to reduce design bond stresses by 23%. The factor q, would
then be reduced from a value of 2.25 for black ribbed bars to 1.75 for coated bars.
However, such an approach ignores the influence of several factors including cover,
lap length, and casting position on the lap strength differential between coated and
black bars.
Bond strength of epoxy coated bars has been found to be less severely affected by
casting position than bond strength of black ribbed bars, as discussed earlier. Casting
position was not a variable within any of the test programmes, however. Casting
position would be classified as ‘not good’ in 2 of the 7 studies examined here, and as
‘good’ in the remaining 5 studies. (Although tension reinforcement was cast in the
top face of the specimens in two of these five studies, the depth of concrete below the
lapped bars fell within the 250mm limit specified in MC90 for bond strength to be
classified as ‘good’). A reasonable fit to the lap data was again obtained using a top
cast factor q, of 0.9, as for the black bar tests. Within the analysis of lap test results,
the value of q2has therefore been taken as 0.9 for both types of bar. The value of the
coefficient q2for use with coated bars will be further considered later in the analysis
of anchorage tests.
The ratio of minimum concrete cover to bar diameter encompassed within the seven
studies ranged from 0.35 to 3.16, with an average of 2.0. The best fit to experimental
data was obtained using a value of 0.27 for coefficient &, Equation 4. This results in
an increase in the value of cover factor clj for coated bars over that for black bars for
concrete covers above the minimum of one bar diameter. The magnitude of the
increase varies with coverhar diameter ratio, greater differences being found at
higher covers. The net effect is a reduction in the difference in bond strength between
coated and black bars with increasing cover. Trends observed for the grouped test
results are thus consistent with those from individual studies, reviewed earlier.
Coefficient ka was determined as 0.21 for black bars in the analysis reported here.
The corresponding value for coated bars was around 25% greater, at 0.27. The value
given in MC90 for black bars is O. 15. To maintain equivalence between coated and
black bars, it is proposed that the value of coefficient ka for coated bars also be set
around 25% higher, at 0.20.
Splice lengths varied between 12.8 and 37.0 bar diameters in tests, with an average
of 17.8. in comparison, a full strength lap splice would typically be in the range 30-50
bar diameters. As reviewed earlier, average bond strength reduces more rapidly with
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0543015 179
increasing lap length for black bars than for coated bars. Relatively short laps are
common in experimental studies in which it is desired to ensure bond failure, but
results tend to overestimate the average bond strength that can be obtained in a full
strength iap. Cairns and Abdullah’s (25) results suggest that the average bond stress
in a full strength black bar splice is around 92% of that of a haif length splice,
whereas for coated bars the proportion was 98%. BCA (21) and Makhlouf‘s (26)
results indicate greater differences, Figure 4. Taking a conservative average for the
three studies suggests that a haif-length coated bar lap need develop only 90% of the
bond strength of an equivalent black bar splice in order to develop the full strength
of a bar in a full length splice.
The average ratio of MC90 design lap strength to measured strength in tests on black
ribbed bars was 2.47, using best fit coefficients. To obtain an equivalent margin of
safety for fùll strength laps of coated bars the corresponding ratio need only be 2.47
x 0.9 = 2.22. Assigning a value of 1.75 to factor q, for coated bars, and using ‘best
fit’ coefficients derived earlier, gives an average ratio of MC90 design lap strength
to measured strength of 2.20 with a coefficient of variation of 0.20. The proposals
thus result in consistent safety margins for black and coated bars. Figure 6 shows the
fiequency distributions of the measured to MC90 design bond strength for both
coated and black bars. (Measured black bar lap strengths have been multiplied by a
factor of 0.90 to account for the extra bond strength obtained when short splice
lengths are tested). The fiequency distributions are similar for both types of bar.
Average results from each of the seven investigations analysed here are plotted in
Figure 7. Results reported by Hester et al (32) have been divided into beam and slab
specimens, giving a total of eight comparisons. The average ratio of measured to
MC90 design strength lies within the relatively narrow band between 1.95 and 2.20
in nearly 70% of cases. Minimum margin of safety between measured and design
bond strength in any series is 1.96 for the coated bars, compared to 1.98 for black
bars. In only one set of resuits does the difference between m e a s d d e s i g n strength
ratio for coated and black bars exceed 8%, and there the margin of safety is higher for
the coated bars.
Despite the wide variations in bond strength ratio reported in the several studies, the
rninimum margin of safety (assessed against the provisions of MC90) is remarkably
consistent. In particular, Figure 7 suggests that in the study by Treece and Jirsa (1O)
on which the AC1 factor of 1.5 for the ‘low cover’ condition is based the splice detail
employed has accentuated differences between coated and black bars. It could also
be mentioned that the diamond rib pattern bar used in these tests is no longer
considered suitable for coating.
Anchorage tes&
Having derived proposals for bond coefficients on the basis of lap test results, it is
now necessary to veri@ that the proposed coefficients are also satisfactory for
anchorages.
Results of lapped splice tests anaiysed here did not indicate any difference in the top
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
Ob62949 0543016 005

cast bond strength reduction between coated and black bars, and the casting position
bond strength reduction was not as severe as allowed for in MC90. As discussed
earlier, however, casting position was not a controlled parameter in any of the lap
studies. More Controlled studies suggest that the casting position reduction factor
should be increased by 14%. in MC90, a casting position factor q2of 0.7 is currently
specified for black bars. On the basis of the anchorage results, it is reasonable to
increase factor q2to 0.8 for coated bars.
Hadje Ghaffari et al (20) investigated the influence of coating, cover and casting
position on anchorage bond strength. Results presented in Table 2 of their study show
an average reduction in bond strength on coating of 11%. The average reduction in
design bond stress obtained using the above proposals for amended values of 7 q2 ,,
and k, is also 11%. Measured and calculated values of bond strength ratio follow
similar trends, Figure 8.
Further comparisons with test data on anchorages, to be reported in detail in a CEB

Bulletin, also indicate that the present proposals derived from tests on lapped splices
are reasonable or slightly conservative for anchorages. From the earlier discussion of
the effects of repeated and sustained loading it was concluded that recommendations
based on monotonic loading will be conservative under other circumstances.
Similarly, recommendations based on straight bar anchorages will be conservative if
applied to anchorages by hooks or bends. Insufficient evidence has been found to
justifj changes to remaining bond factors in MC90.
Effect of proposed modifications on lap and anchorage length -- The net effect
of the proposed factors on the increase in lap length required for coated reinforcement
over that required for black bars depends on cover and casting position. The lap
length factor for coated bars, i.e the ratio of the required lap length for a coated bar
to that required for a similar black bar, is plotted as a function of cover for ‘good’ and
‘other’ casting positions in Figure 9.
The greatest increase required by the present proposals, of 29%, is required for ‘good’
casting positions in which a minimum cover of 1 bar diameter is provided. (MC90
in particular and national Codes in general spec@ a minimum cover of one bar
diameter for siructural reasons). For a top cast bar with a minimum cover of 2.7 bar
diameters or more, no increase is required. Given the generally acknowledged bond
strength reduction where coated bars are used, it may appear generous not to require
an increase in anchorage or splice length in all circumstances. However, remember
that MC90 tends to underestimate the effect of cover ratio on bond strength, and thus
tends to be conservative at higher covers in any case.
Reduction factors for bond strength at anchorages and lapped splices were first
introduced by the AC1 Building Code(37)(38) in 1988 (Clause 12.2.4.3). The rules
specify a 50% increase in lap length for coated reinforcement over requirements for
non-coated bars where cover is less than 3 bar diameters and clear spacing between
laps is less than 6 bar diameters, and an increase of 20% in lap length over
requirements for non-coated bar in all other circumstances. These modification
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 Ob62949 0543017 T4L
fáctors are derived from two studies, one of tension lapped splices conducted by Jirsa
and Treece (lo), the other of beam end specimens conducted by Johnston and
Zia(39). The new proposals therefore suggest that the AC1 provisions for anchorage
of coated bars are unduly conservative.
FBECR will not generally be used with cover to bar diameter ratios as low as 1.O.
FBECR is to be considered to provide an additional measure of corrosion protection
for reinforcement. Specification of coated reinforcement is intended to supplement
and not to not supplant existing cover requirements for durability2. As the extra cost
of FBECR is generallyjustified only where there is risk of significant reinforcement
corrosion, its use will be justified primarily in harsh environments, and it will not
generally be specified for exposures less than Exposure Class 3, as defined in [Table
1.5.11. Concrete cover to reinforcement in structures subjected to such environments
will be relatively high. By way of illustration, compare minimum concrete cover
requirements for concrete of Grade 40. Grade 40 is commonly specified for
reinforced concrete structures such as bridge decks in chloride contaminated
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
environments. Nominal cover to reinforcement in Class 3 conditions is 50mm,as

opposed to 20mm for a ?Mild? exposure. The cover ratio (minimum cover expressed
as a multiple of bar diameter) for a 32mm bar in a slab in a Class 3 environment is
thus 1.56. in a beam element, minimum cover and requirements for stirrups around
main reinforcement means that cover ratio to lapped bars will be at least 1.45, even
for 40mm diameter bars. in comparison, cover ratio for bars of size 20 or greater in
a mild exposure need only be 1.O. The increase in lap or anchorage lengths required
for a cover equal to 1.5 times bar diameter is 25% and 10% respectively for ?good?
and ?poor?casting positions. The increase in lap or anchorage lengths required for a
cover equal to 1.5 times bar over that required for a black bar with nominal minimum
cover of 1 bar diameter are 16% and 1%. Thus the net increases in lap length required
for coated bars are less than might be thought at first sight.
-
Anchorage of stirrups Results from McGormley et al(40) have demonstrated that
coated No.3 (9.5mm) stirrup bars have adequate anchorage for shear, but caution
against extrapolation to larger diameters. The BCA study also confirmed that shear
capacity was not impaired by use of coated stirrups.
Summary of proposais
[6.9.3] Design bond stressfor reinforcing bars. A h ??q,= 1.4 for indented
bars?, inseri ?q, = 1.75 for fusion bonded epoxy coated
reinforcement? Note also that q, = 2.25 may continue to be used for
welded wire fabric.
After ?q2= 0.7 ......built with slip forms?, insert ?except that q2= 0.8
Personal communication from J Hartley, Allied Bar Coaters, UK.
for epoxy coated bars in such locations”
[6.9.5] Design anchorage length. In Table 6.9.1, add “for coated bars, ct3 =
1- 0.20 (cd - $)I$ (0.7s ct) < 1.0)
Other Aspects of Structural Performance
While anchorage and lap strength generally command the greatest attention in
structural design, an epoxy coating can influence other aspects of structural
performance. CEB Task Group 215 will address all aspects of performance in the final
report. A brief review is presented here for completeness.
Plastic Hinge Rotation -- The ability of a structure to redistribute elastic moments

is governed by the rotation capacity of the plastic hinges which form in regions of
peak bending moment values. The CEB-FIP Model Code 90 provides a model for
plastic hinge rotation capacity which takes account of bond Characteristics of
reinforcement. The model predicts that a reduction in bond stiffness will lead to an
increase in hinge rotation capacity.
In what appears to be the only published study to report on ductility of beams

containing epoxy coated bars, Abrishami et al (15) reported a reduction in rotation
capacity with epoxy coated bars. Beams contained 0.92% tension reinforcement.
However, in an as yet unpublished report, Cairns and Carbaugh(41) measured plastic
rotation capacity of paired beam specimens reinforced with black and with coated
reinforcement. Rectangular beam sections with 0.43% and 0.77% reinforcement
ratios were tested over three different spans. Results are summxised in Figure 10.
Rotation capacity of the short span is less than that of the longer spans, as would be
expected, but the difference in rotation capacity between specimens reinforced with
coated and black bars is not significant.
Conclusions from these two studies thus conflict to some degree. However, in the
absence of sufficient data to show that rotation capacity is affected to a significant
degree by use of coated reinforcement, and as the lesser bond stiffness of coated bars
would be expected to have the beneficial effect of increasing rather than reducing
rotation capacity, it is proposed that for the time being the coefficients for black
ribbed bars in [Clause 3.2.31 be retained for coated bars.
Tension stiffening and deflection -- The lesser bond-slip stiffness of coated

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
reinforcement would be expected to result in a reduction in tension stiffening. The

effect of the reduction in bond stiffness will depend on the section detail under
consideration, however, and deformations of more heavily reinforced sections and ‘T’
sections will hardly be affected by a change in bond properties. Thus, although
several investigators report little change in deflection where coated bars are used, the
test specimens were often relatively highly reinforced and would be insensitive to
loss of tension stiffening. The presence of a lapped splice may also have masked loss
of tension stiffening in a number of studies.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
in a series of tests on seven pairs of nominally identical beams of various cross
sections, Cairns measured an average increase of around 20% in average tensile
strains of coated bars, Figure 11. Deflections for individual pairs increased by
between 5% and 35%(42). Abrishami et al conducted a short series of tests on flat
slabs(l6) of normal and high strength concretes in which surface strains on the
tension face of the slab were increased by an average of over 20% by coated bars.
Hasan et al(23) report increases in total crack width, consistent with a reduction in
tension stiffening, of 25% on initiai loading and 12%after 1 miliion cycles in one set
of specimens. Cairns has suggested that tension stiffening is effectively halved by
coated reinforcement. Work continues within CEB Task Group 2/5 to develop
detailed amendments to MC90 rules.
Control of crack width -- Several Authors report increased crack spacing and crack
widths when FBECR is used in place of black bars in nominally identical beams. A
wide range of Increases of up to 100%have been reported. As with tension stiffening,
the diversity of increases reported must be due at least in part to the greater sensitivity
of some sections to changes in bond characteristics. Cairns tested 7 nominally
identical pairs of beam specimens, one of each pair reinforced with coated and the
other with non-coated bars (41). He used the crack width prediction model that
formed the basis of the CEB-FIP Model Code 1978 and of EC2 to deduce that bond
stiílhess of epoxy coated bars was half that of black bars, Figure 12. However, íùrther
work remains to be undertaken within CEB Task Group 2/5 to develop detailed
amendments to MC90.
CONCLUSIONS
The work reported in this study has led to development of proposais for amendments
to enable the CEB-FIP Model Code 1990 to be used for design of splices and
anchorages of coated reinforcement. The net effect of the proposals is for bond
lengths of coated reinforcement to be increased by up to 25% (good casting position)
and 10% (other casting position), with lesser increases for covers exceeding 1.5 times
bar diameter. No increase in bond length is required for welded wire fabric.
A review of literature also demonstrates that tension stiffening may be reduced and
deflections increased where coated reinforcement is used, at least for more lightly
reinforced beams. Greater crack widths may also be found. There is conflicting
evidence on the infiuence of coating on plastic hinge rotation capacity.
The infiuence of rib pattern on the reduction in bond strength on coating remains
unclear. It may be advisable to place restrictions on the rib geometry of bars for
coating.
REFERENCES
1. CEB-FIP Model Code 1990. Comite-Euro International du Betoflederation internationale

de la Precontrainte. Thomas Telford, London. 1993.437 pp.
m Obb2949 0543020 5 3 b =
2. Protection systems for reinforcement. CEB Bulletin No. 21 1. CEB. Lausane. Feb 1992.
3. Standard specification for epoxy coated reinforcing steel bars. ASTM A775/A775M-88a,
1989. Annual Book for ASTM Standards, Vo1.1.04, Philadelphia, USA, pp548-552.
4.- British Standard BS7295:1990. Fusion bonded epoxy coated carbon steel bars for the
reinforcement of concrete. British Standards Institution, London, 1 9 9 0 . 7 ~ ~ .
5. Standard Specificationand test methods for epoxy-coated reinforcing steel bars. JSCE EPlO-
EP17. Japan Society of Civil Engineers. Tokyo. 1986.
6. Standard specification for epoxy-coated prefabricated steel reinforcing bars. ASTM

A934IA934M-95.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
7. Standard specification for epoxy-coated steel wire and welded wire fabric for reinforcement.
ASTM A884.
8. Cairns, J., & Abdullah, R. Fundamental tests on the effect of an epoxy coating on bond
strength. American Concrete Institute Materials Journal. Vol. 91, No. 4. JulyíAugust 1994.
pp 331-338.
9. Cairns, J., & Abdullah, R. Influence of rib geometry on strength of epoxy coated
reinforcement splices. American Concrete Institute Structural Journal. Vol 92, No. 1. Jan-Feb
1995. ~ ~ 2 3 - 2 7 .
1o. Jirsa, J.O. & Treece, R.A. Bond strength of epoxy-coated reinforcing bars. AC1 Materials
Journal, Vol 86, No. 2. March-April 1989. pp167-174.
11. King, E. Post-fabrication coating systems. Paper presented at Seminar on Durable

Reinforcement for Aggressive Environments. Luton. 1990.
12. Choi, O.C., Hossain, H-G., Darwin, D., & McCabe, S.L. Bond of epoxy coated
reinforcement : bar parameters. American Concrete Institute, Materials Journal. Vo1.88, No,
2. Mar-Apr 1991. pp 207-2 17.
13. Hamad B.S. & Jirsa J.O. Strength of epoxy-coated reinforcing bar splices confined with
transverse reinforcement. AC1 Structural Journal Vol 90, No. 1 Jan/Feb 1993. pp77-88.
14. Kobayashi, K. & Takewaka, K. Experimental studies on epoxy coated reinforcing steel for
corrosion protection. International Journal of Cement Composites and Lightweight Concrete,
Vol 6, No. 2. May 1984. pp99-116.
15. Abrishami, H H, Cook, W D, and Mitchell, D. Influence of epoxy coated reinforcement on

response of normal and high strength concrete beams. AC1 Structural Journal, Vol 92, No.
2. March April 1995., pp157-166
16. Abrishami, H H, Cook, W D, and Mitchell, D. The effect of epoxy-coated reinforcement and
concrete quality on cracking of flat plate slab-column connections. AC1 Materials Journal,
Vol 93, No. 2. March-April 1996. pp 121-128.
9 Obb2949 0543021 472 9
Tribute to Peter Gernelv 423
17. Lutz, L.A. & Gergely, P. Mechanics of bond and slip of deformed bars in concrete.
Proceedings of the Amencan Concrete Institute, Vol. 64,No. 11, Nov. 1967. pp 71 1-721
18. Darwin D, Tholen M L, Idun E Kyand Zu0 J. Splice strength of high relative rib area
reinforcing bars. AC1 Structurai Journal, Vol 93, No.1. Jan/Feb 1996. pp 95- 107.
19. D& D, and Idun E K.Discussion of Reference (9). AC1 Structurai Journal, Vol 92, No.6.
Nov-Dec 1995. pp 782-785.
20. Hadje-Ghaffari H, Choi O C, Darwin D & McCabe, S L. Bond of epoxy coated

reinforcement: cover, casting position slump and consolidation. AC1 Structural Journal, Vol
91, NO.1, Jan 1994. ~ ~ 5 9 - 6 8 .
21. Clarke, J L. Tests on fusion-bonded epoxy-coated steel. BCA Project Report RN1.025.01.
Slough, UK. Nov 1 9 9 0 . 2 3 ~ ~ .
22. Cleary D.B. & Ramirez, J.A. Bond strength of epoxy-coated reinforcement. American
Concrete Institute Materiais Journal, Vo1.88 No.2, Mar-Apr 1991. pp 146-149.
23. Hasan H O, Cleary D B, and Ramirez J A. Performance of concrete bridge decks and slabs
reinforced with epoxy-coated steel under repeated loading. AC1 Structurai Journal, July-Aug
1996. pp. 397-403.
24. Cleary D B and Ramirez, J A. Epoxy coated reinforcement under repeated loading. AC1
Structural Jounial, Vol 90, NO.4, July-AUg 1993. ~ ~ 4 5 1 - 4 5 8 .
25. Cairns, J., t Abdullah, R. Ultimate strength of lapped joints of epoxy coated reinforcement.
Proceedings of the Institution of Civil Engineers, Structures & Buildings. Vol 104. Nov.
1994. ~ ~ 3 9 1 - 4 0 0
26. Makhlouf H My Al-Tamimi, A, Akbari, A H.Bond Carrying capacity of epoxy coated

reinforcement. Proc. Conf. On Corrosion and Corrosion Protection of Steel in Concrete.
Sheffield Academic Press. Sheffield. 1994. pp1218-1230.
27. Cusens A R,and Yu, Z. Bond strength and flexural behaviour of beams with epoxy coated
reinforcement. The Structural Engineer, Vol 71, No. 7, April 1993. pp 117- 124.
28. Tepfers R. A theory of bond applied to overlapped tensile reinforcement splices for deformed
bars. Publication 73:2, Division of Concrete Structures, Chalmers University of Technology.
Göteborg, Sweden. 1973.
29. Hamad, B S, Jma,J O & d'Abreu d Paolo N. Effect of epoxy coating on bond and anchorage
of reinforcement in concrete structures. Report 1181-1F. Center for Transportation Research,
University of Texas at Austin. Dec 1990.
30. Bartoletti, S . J. and Jirsa, J.O., "Effects of Epoxy-Coating on Anchorage and Splices of
Welded Wire Fabric." AC1 Structural Journal, Nov.-Dec. 1995, pp. 757-764.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
31.
IObb2949 0543022 309
Rehm, G. Plastic coated reinforcement in concrete construction. Betonwerk + Fertigteil-
technic. HEFT 1, 1987.
-
32. Hester, C.J., Salamizavaregh, S., Darwin, D. and McCabe, S.L. Bond of epoxy-coated
reinforcement: Splices. AC1 Structural Journal, Vol. 90, No. 1, Jan-Feb 1993. pp89-102.
33. Orangun C.O., Jirsa J.O., & Breen J.E. The strength of anchored bars ;a re-evaluation of test
data on development length and splices. Research Report 154-3F, Center for Highway
Research, The University of Texas at Austin. Austin. January 1975. 78pp.
34. Clifton, J.R. & Mathey, R.G. Bond and Creep Characteristics of Coated Reinforcing Bars in
Concrete. American Concrete institute Journal. Vol 80, No.4. JulyíAugust 1983.pp288-293.
35. Johnston, D.W. & Zia, P. Bond fatigue of epoxy coated reinforcing bars. Materiais &
Structures : Research & Testing. No. 97. J d e b 1984. pp30-34.
36. Lin, T.D. et al. Pullout tests of epoxy coated bars at high temperatures. AC1 Materials
Journal. Nov-Dec 1988. pp544-550.
37. AC1 Committee 3 18. Revisions to Building Code Requirements for Reinforced Concrete.
Vol. 85, No. 6. Nov/Dec 1988. pp645-674.
38. AC1 Committee 3 18. Building Code Requirements for Reinforced Concrete and Commentary
(AC1 318-89/ACI 318 R-89). American Concrete Institute. Detroit. 1989. 353pp.
39. Johnston D W, and Zia P. Bond characteristics of epoxy coated reinforcing bars. Report No.
FHWA-NC-82-002, Center for Transportation Engineering Studies, Civil Engineering
Department, North Carolina State University, Raleigh. 163 pp.
40. McGormley, J C, Cleary, D B, and Ramirez, J A. The performance of epoxy-coated shear

reinforcement. AC1 Structurai Journal, Vol. 93, No. 5. SepîíOct 1996. pp. 531-537.
41. Cairns J & Carbaugh C. Deformability of beams reinforced with epoxy coated bars. To be
published.
42. Cairns, J. Performance of epoxy coated reinforcement at the serviceability limit state.
Proceedings of the institution of Civil Engineers, Structures& Buildings. Vol 104. Feb 1994.
~~61-73.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m 0662749 0543023 245 m
1.2 -
-
.-O
2
u)
1 - -
. .~ ...
-
. .
-.
t
----.-.-----~
__.--
__.---
0.4
20 30 40 50 60 70 80
Rib Face Angle (Degrees)
Slip (mm)
. -~~ ~~~
~i
r----o.01
L
~ ~~
-
0.10 1.00 . -
._~~~___._--p-_l -- Ult '
Fig. 1 -Influence of rib angle on bond stress ratio, measured in RILEM Pullout Test
(Ref. 8)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
8 Ib
Mode 1 Hode 2 Mode 3
Pullout Failure Splitting Failure Splitting Failure
Fig. 2-Bond failure modes
0662949 0543024 LBL W
rn
a
2
3.0
5
cn
c
s!
-u 2.0
c
O
m
1.o
-
MO-1 MO-2 B20 C20 D20 E20
Bar Reference
Black FBECR
Fig. 3-Measured bond strengths for various bar patterns, both non-coated and
coated (Ref. 9)
Ref 25
300
I Y
,JO
31 I
Ref 21
'O0O
10 20 30 40
Lap Length Ratio
I -c Black FBECR I
Fig. 4-Influence of splice length on stress developed in bar. Derived from
Refs. (21), (25) and (26)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb2ïLiï 0543025 018
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
I I
eI
I
w- P
O 1 2 3
Cover ratio
Fig. 5-Influence of cover ratio on ratio of measured to MC90 design bond
strength
O 1 2 3 4
Ratio Measured to Design Bond Strength
FBECR
--A- - Black
Fig. 6-Frequency distribution of measured to MC90 design bond strength: Non-

coated and coated bars
m ûbb2949 0543026 T 5 4 W
L
ci
p 3
i?!
2.8
K
'i
n
2.6
0 2.4
m
2 2.2
8
% 2
O
.
m
s1.8
0.65 0.7 0.75 0.8 0.85 0.9
Rat io Coa ted/B Iack
Fig. 7-Comparison between ratio of measured to MC90 design bond strengtn and
bond strength reduction-by test series
cn 1.1 - 1
5 I
I
I---
CI)
t 1 -__
I
ci --
I I
U
0.9 ..
-1
c I
I -'i
c
n i
'O 0.6-.
.- ~- -
* --* of Equality
Line , ..
I
0.5 7 ,
0.5 0.6 0.7 0.8 0.9 1

Measured Ratio Coated/Black
Fig. 8-Influence of concrete cover on bond strength ratio: Comparison of results

from Hadje Ghaffari et al. (20)with proposed design rules
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Ob62949 0543027 990 m Tribute to Peter Gergely 429
1.5
B
2 1.3
lm4
5 1.2
5P 1.1
3 1
0.9
1 1.5 2 2.5 3 3.5 4
Cover ratio
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
- Good Casting Position Other
Fig. 9-Lap length factor for FBECR.net effect of proposals
0.05
40.04
n
!z
c 0.03
O
F2i 0.02
o
m
ñ
0.01
O 1 1
I.Om 2.0m 3.0m

Span
Black a FBECR p=O.43%

Black FBECR p=O.77%
Fig. 10-Plastic rotation capacity (ref. 41)
Obb2949 0543028 827
3500
.i
h
3000
c
L.
8 2500
t
5 2000
v
5 1500
!
1000
S
(TI
2 500
n-
KWI1 KW12 DWll DW12 DWl3 DW14 DW15
Specimen Reference
1 Black
~
FBECR
II
Fig. 11-Effect of epoxy coating on mean strains in tension reinforcement

(Ref. 42)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
.-c
O
Ò? 1.5
e3
2 4 1
o
o
e
c
rn 0.5
r"
o KW/I KW/2 DW/I DW/2 DW/3 DW14 DW/5
Specimen Reference
Fig. 12-Comparisons of measured and calculated differences in crack width

between epoxy coated and black bar specimens
= ûbb2949 0543029 ïb3
SP 180-19
Fundamental Analysis of RC lap Splices
by T. Ichinose, T. Hayashi and W. Lin
Synopsis: An analytical model is presented to evaluate the local bond-slip

relationship in RC lap splices considering the influence of splitting cracks,
friction between the concrete and rib face, microscopic fracture of concrete in
front of rib face, flexural deformation of the cover concrete, and stirrups. The
local bond-slip relationship is integrated along the splice length considering the
strain in the concrete in the longitudinal direction. It is shown that the lack of
friction at a rib face due to epoxy coating reduces the ductility of the local bond
slip relationship without stirrups, resulting in lower splice strengths. The
contribution of stirrups to the splice strength is dominated by the flexural
strength of the cover concrete with stirrups, and is the function of concrete
strength rather than the yield strength of the stirrup.
Keywords: Analysis; bond (concrete to reinforcement); cracking (fracturing); epoxy

coating; fracture; friction; stirrup; strength
43 1
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
432 Ichinose, Hayashi and Lin
111 0662949 0543030 485 =
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
T. Ichinose is professor of Structural Engineering in the Department of
Architecture at Nagoya Institute of Technology. He received his doctorate from
the University of Tokyo. His major interest is in seismic design of reinforced
concrete buildings.
T. Hayashi was a M.S. student at Nagoya Institute of Technology. He joined

Dai Nippon Construction in 1997.
W. Lin is a researcher in the Department of Architecture at Nagoya Institute of

Technology. He received his M.S. from the Xi’an University of Architecture
and Technology of China. His research interests include concrete seismic
construction, and design and behavior of concrete structures.
INTRODUCTION
There have been a large number of experimental studies on RC splices,

resulting in numerous empirical equations to calculate the strength of a splice.
As far as the authors know, all the equations assume that the force carried by a
splice (TJconsists of the contributions of concrete and stirrup as follows:
T, =T, +Ts
where Tc and Ts are the concrete and steel contributions, respectively. A couple
of related questions arise concerning this equation:
0 Do the concrete and the stirrups contribute to bond force simultaneously?
[Question #I = QI]
Doesn’t the contribution of concrete vanish when splitting crack widens and
the stirrup starts to contribute?
Moreover, the contribution of concrete, Tc , is reported to be largely impaired
by epoxy coating due to the lack of friction at the rib face (Treece and Jirsa
1989). Therefore, the following additional question arises in this case:
How does the lack of friction affect the local bond-slip relationship? [QZ]
The answer to this question can also provoke another question:
0 Does the notion ‘local bond-slip’ have a physical meaning at all? [Q3]
Concerning the contribution of stirrup to bond force, Ts, Darwin et al.

(1996) proposed a new empirical equation as follows:
where,
Obb2949 0543031 311
f,? = concrete compressive strength (MPa),

db = bar diameter (mrn),
N = number of stirrups crossing potential splitting plane,
A, = sectional area of a stirnip (mm?), and
n = number of bars spliced along potential splitting plane.
The equation is quite different from the equation below proposed by Orangun et
al. (1977) which is the basis of the current AC1 318-95:
where,
x,, = yield strength of stirrup (MPa),
= length of lap splice (mm).
1,
If one assumes that the more modern expression (Eq. (2)) is better than Eq. (3)
at predicting strength of splices, we may have the following questions:
Why does the bar diameter ( db) affect T, ? [@I
O Why does the yield strength of the stirrups not affect T,? Experimental
observations show that stirrups do not yield at bond failure (e.g. Sakurada et
al. 1993), though wide splitting cracks are sometimes observed; why don?t
the stirrups yield?) [QS]
0 Why is there no upper limit of T, in Eq. (2)? [Q6]
The objective of this study is to present an analytical model which may

give answers to these questions, i.e. Ql through 46, and establish quantitative
limits to the empirical equations that have been proposed. For splices without
stirrups, the analysis consists of three steps:
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
1. Obtain the relationship between a uniform radial stress normal to the bar
and the crack width at the bar; this is done using discrete elements;
2. Derive the local bond-slip relationship considering the microscopic fracture
of the concrete in front of ribs and the friction at the interface between
concrete and rib face; and
3. Integrate the local bond-slip relationship to calculate the splice strength
considering both (a) the equilibrium between the bar axial stress and bond
stress resultants and, (b) the compatibility of strains and slip.
For splices with stirrups, the flexural resistance of the cover concrete
including stirrups will also be considered when deriving the local bond-slip
relationship.
Ob62949 0543032 258 =
LOCAL BOND-SLIP RELATIONSHIP WITHOUT STIRRUPS
Sdittina of Cover Concrete
A section of a splice in a beam, including typical side splitting cracks,

is shown in Fig. 1. The shaded zone in Fig. 1 is discretized for analysis as
shown in Fig. 2 , where capis the average of cc and c,, defined in Fig. 1. A
uniform radial stress g is assumed at the interface between the bar and the
concrete. The top surface of the discretized zone in Fig. 2 is free. The two sides
and the bottom of the zone are assumed to be supported by rollers. The analysis
is conducted using the rigid-body-spring model shown in Fig. 3, as proposed by
Kawai (1990). The stiffness and strength of the springs are determined based on
the dimensions of the rigid bodies and the mechanical model for the concrete,
as explained below.
Mechanical Model for Concrete
The modulus of elasticity for concrete is calculated by the following

equation proposed in the New RC project in Japan (Tomosawa et al. 1994):
where y is concrete specific gravity ( y =2.4) a n d i ' is concrete compressive

strength.
The concrete is assumed as an elasto-softening material having the

failure surface shown in Fig. 4 with a von Mises's flow rule (the normality
condition). The line XY shows the Mohr-Coulomb failure with dilatation. The
angle of friction 4 is assumed as 37 degrees. The cohesion c is assumed to
vanish when the relative shear displacement at a failure plane,V,, reaches w'
(defined later) as shown in Fig. 5. The initial value of the cohesion c is
assumed as one half o f f , that is twice of the value for the normal Mohr-
Coulomb model since the analysis deals with a very small scale (this is a form
of size effect).
It is also assumed that concrete may fail in shear without dilatation at a

strength of zmXas shown by the line Y 2 in Fig. 4. The initial value of qwr is
assumed as a function of f,' as shown in Fig. 6, based on the pullout tests shown
in Fig. 7 (a) (Akashi et al. 1991). In these tests the steel pipe is thick enough to
be considered as rigid and the bar was pulled out in a direct shear mode as
shown in Fig. 7 (b). Based on the same experiment, T,,,,~ is assumed to vanish
when VAreaches 20 mm as shown in Fig. 8. In effect, however, this degradation
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
is negligibly small, since splices normally reach their strengths when local slip
reaches around O. 1 mm as will be shown later.
The tensile strength of concrete,f , should also be evaluated considering

size effects. The size of aggregate is normally similar to the bar diameter, and
the aggregates between the bars would block the splitting cracks shown in Fig.
1. Thus, the tensile strength of concrete is assumed to be twice of the value
calculated by the CEB-FIP Model Code (19W), as shown by the dashed line in
Fig. 6. The softening branch after the maximum tensile strength is reached is
assumed as a function of the crack width as shown in Fig. 9. The fracture
toughness of the concrete, G,, is assumed to be four times of the value
calculated by the CEB-FIP Model Code (1990). This, again, considering the
blocking effect by aggregates. Figure 10 shows the relationship between
compressive strengthL,’,maximum aggregate diameter da , and the crack width
at which tensile stress vanishes, w,.
Fracture Modes and Equilibrium at a Rib
The side splitting failure shown in Fig. 1 is the dominant failure mode in
most cases. Main bars are normally placed so that the transverse ribs are
located at the upper and lower sides of the bars as shown in Fig. 11 (a). Thus, in
the following analysis, a bar section is modelled as a rectangle, as shown in Fig.
11 (b), considering equilibrium in the vertical direction only.
Since the assumed failure surface of Fig. 4 has two modes, XY and YZ,
and the inelastic deformation should occur perpendicular to the failure surface,
there should be at least two failure patterns as shown in Figs. 12 (a) and (b).
The horizontal displacement y (accumulation of AVs of Fig. 12) gives ‘slip’,
whereas the vertical displacement Vc gives one half of the vertical opening at
the bar. Another failure pattern is friction at the rib surface shown in Fig. 12 (c).
In the case of epoxy-coated bars, the hction coefficient p is assumed as O. In
case of uncoated bars, the friction coefficient p is assumed as 0.5. The failure
pattern is selected so that the local bond stress z (defined later) be minimized.
The failure pattern may vary as the slip increases.
In order to evaluate the local bond stress 7 , equilibrium at a rib is

considered as shown in Fig. 13, where we have:
QP = N s i n a + Q c o s a (5)
Qv =Ncosa-Qsina (6)
where a is the angle of the failure plane and is determined later; N and Q are the
forces normal and tangential to the failure plane, respectively; and Qp and Q,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 ObbZî49 0543034 O20 W
are the horizontal and vertical forces carried by the rib. Q, is given by the
following equation:
where q is the radial stress as a function of crack width as given by the analysis
of Fig. 2; and Rsis the interval between the ribs. The tangential force Q is given
by one of the following equations depending on the failure patterns in Fig. 12;
Pattern 1 (Mohr-Coulomb): Q = CS+ N tan I$ (8)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Pattern 2 (Direct shear): Q =~maxS (9)

Pattern 3 (Friction at rib face): Q = p V (p=Oor0.5) (10)
where S is the area of the failure plane.
Eliminating N and Q from Eq. 5 using Eqs. 6 through 10, we obtain Q,, as a
function of a,which is determined so that Q, is minimized. Thus, Q, is obtained
from q in Eq. 7. On the other hand, unis a function of Vc, whose increment is
determined from AVs as follows:
Pattern i AV, = AV, tm(a + 4) (1 1)
Pattern 2 AV, = AVs t a n a (12)
Pattern 3 AV, = AV, tanß (13)
Thus, Q,, is the function of slip Vy . The local bond stress T i s defined as an
average shear stress around the bar as follows.
We can then obtain the local bond-slip (z-VJ relationship since Q,?is a function
of V+as just described.
Examde
The specimens of Nester et al. (1993) were analyzed following the

technique described above. The bar dimensions and the cross section of one of
the specimens are shown in Fig. 14. The calculated relationship between the
vertical force Q, of Fig. 13 and the opening at the bar is shown in Fig. 15, where
it is assumed that Q, decreases linearly from the peak point to the point of w c,
m 0bb2949 0543035 Tb7 W
since the analysis of Fig. 2 was controlled by an incremental load approach. The
calculated relationship between the local bond stress and the slip is shown in
Fig. 16. Epoxy coating does not impair the local bond strength in Fig. 16 but
reduces the ascending and descending stiffnesses. In the case of the specimens
with uncoated bars, the failure mode is always Pattern 2 (direct shear) with a
between 30 and 40 degrees. In the case of the specimens with coated bars, the
failure mode is always Pattern 3 (sliding at the rib face), which accounts for the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stiffness reduction. The difference of the failure pattern agrees with the
experimental observation by Treece and Jirsa (1989) that the ribs of coated bars
were very clean with no concrete residue whereas concrete deposit was
observed at the ribs of uncoated bars.
LOCAL BOND-SLIP RELATIONSHIP WITH STIRRUPS
Trianaular Beam Hypothesis
As splitting cracks propagate, stirrups start to carry tensile forces,

resisting the crack opening as shown in Figs. 17 (a) and (b). This tensile force
will induce a triangular-shaped failure surface, as shown in Fig. 17 (c), or a
surface made of triangular beams as shown in Fig. 17 (d). The actual section of
the beam might be as shown in Fig. 18 (a), but it is simplified in this analysis to
that shown in Fig. I8 (b) with equal sectional areas of concrete and stirrups.
The moment-curvature relationship for the simplified section is modelled in
Fig. 19, where,
EI = the initial rigidity when the beam is bent with tension on top, and is
calculated based on the elastic moduli of concrete and steel;
q5m = the curvature when the compressive strain at the bottom edge of the beam
reaches 0.003;
Mcu= the ultimate moment calculated as per AC1 318-95;
M , and q5d, = the moment and the curvature when the compressive strain at the
bottom edge of the stirrup reaches E, which is defined below;
E, = the strain at which concrete loses its compressive stress completely, and is
assumed as in Fig. 20;
M, and q5, = the moment and the curvature with tensioned bottom when the
strains at the top and bottom edge of the stirrup reach the yield strain; and
M, = the moment at 4," = 54,, representing large curvature after yielding.
The deformation of a triangular beam is shown in Fig. 21, where P,

represents the uplifting force from the main bars at the comers. The
intermediate main bars may also contribute some smaller uplifting forces which
are neglected in this study. The spring at each end represents the pullout of the
stirrup from the concrete below the splitting crack. The stiffness of the springs
is calculated assuming that the strain is distributed uniformly along a length
equal to ten times the diameter of the stirrup. (This assumption, however, does
IOb62949 0543036 9T3 =
not affect much the analytical results reported in this paper, since the elongation
of the springs is always much smaller than that of flexural deformation of the
triangular beams.)
We will assume that the triangular beam has uniform rigidity and its two
ends are fixed against rotation. Thus, the curvature at the supports and the
loaded points, qìl and qì2, are expressed as follow:
where V$ is the contribution of the flexural deflection to the crack opening at

the outer bar. Applying the calculated curvatures to Fig. 19, we obtain the
bending moments at the supports and the loaded points, which lead to the
uplifting force PIas a function of VLf.The total opening at the outer bar Vc is
given as the sum of V', and the elongation of the spring at the support.
Triangular beams with inner stirrups are modelled similarly as shown in Fig.
22.
This model gives answers to [Q4] and [QSI:
[A41 The diameter of the spliced bar dbin Fig. 18 (a) increases the depth of the
triangular beam, and therefore contributes to T, in Eq. (1).
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
[A51 The flexural strength of the triangular beam (and therefore splice strength)
is more sensitive to the concrete strength than to the stirrup strength. Stirrups at
the splitting crack (represented by the springs in Figs. 21 and 22) do not yield in
tension because the triangular beam fails in flexure.
Eauilibrium at Rib and Local Bond-Slip Relationship
In the case of a splice with stirrups, the vertical force on a rib, Q, in Fig.
13, may be decomposed as follows:
Qu = + Q,, (17)
where Q, is the contribution of cover concrete given by Eq. (7) and Q,, is the
contribution of the stirrup. Q, should depend on which region in Fig. 23 a rib
belongs to. Ribs in the cover concrete provide no contribution. Ribs inside
triangular beam take the following contribution:
0662949 0543037 83"
&=P/m,
-
(18)
where P is the uplifting force in Fig. 21 or 22 ( P,or P, ), and m,is the number
of ribs inside a triangular beam and given as follows (see Fig. 23).
1
m, = -[(l +cos3ûo)d, +2Hd cos30"
R,
The ribs in the core concrete are assumed to provide the following contribution,
where the denominator s /R3 represents the number of ribs in a spacing of
stirrup, s:
P
=-
s / R,
Q,,
Equilibrium gives the horizontal force Q, at every rib, whose weighted average
gives local bond stress as a function of slip Vs.
ExamDle
As an example, one of the specimens of Sakurada et al. (1993) with

shear reinforcement ratio A, lb.s = 1.2%was analyzed. The cross section of the
specimen is shown in Fig. 24. The calculated relationships between the vertical
force in the stirrups and the opening at the bar are shown in Fig. 25. The
vertical force from the inner stirrup exhibits larger stiffness than that of outer
one, but it is still much smaller than that of the cover concrete shown in Fig. 15.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Figure 26 shows the local bond-slip relationships. The local bond stresses of the
intermediate and corner bars have the same peak strength of 6 MPa and
decrease until z = 2 MPa and V, = 0.04 mm. Until this point both results are the
same as those without stirrups. After this point, the stirrups start to contribute.
This is due to the large difference of stiffness of confinement in Figs. 15 and 25.
DISTRIBUTION OF STRESS, STRAIN AND SLIP
Definition. Assumption. and Equilibrium
Slip is defined as the displacement of the bar relative to concrete. Thus,

we have:
m 0bb2549 0543038 776
dVsR = EsR -Ec

dx
where V , and V,, are the slips of the bars extending to the left and the right; E,
and &SR are the strains of the bars; and is the strain of the concrete.
If the spliced bars remain elastic, equilibrium between bar stress and
bond stress gives:
where E is elastic modulus of steel and rL is the bond stress along the bar
extending to the left. Differentiating Eq. (21) and substituting Eq. (23), we
have:
Similarly,
We assume that local bond stress of a bar is transferred directly to

another bar at any section as shown in Fig. 27. In other words,
rL= - T ~ (26)
is assumed. This means that,
v,, = -v, (27)
Substituting Eq. (27) into Eq. (21) and adding Eq. (22), we have:
&SL -&SR
i
E, =
2
meaning that the concrete strain is the average of the strains of the spliced bars.
Substituting Eqs. (26) and (27) into Eq. (24) and adding Eq. (25), we have:
ds,= O
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
dx
meaning that concrete strain is uniform along the splice. In reality, a uniform
cohcrete strain is impossible, since tensile cracks always occur discretely.
9 Ob62949 0543037 602 = Tribute to Peter Gergely 441
crack2
Thus, Eq. (26) should be replaced with Jcrackl (TL + ZR )dx = o, where
cracks 1 and 2 represent neighboring tensile cracks. Consideration of discrete
tensile cracks, however, requires more sophisticated analysis, and is left for
future research.
Equations (27) through (29) mean that slip occurs as shown in Fig. 28,
where (a) shows the stage without slip and strain; (b) shows the stage when slip
occurs only in the end regions hatched in one direction, whereas in the
crosshatched central region, slip does not occur and the steel strain is equal to
the concrete strain; and (c) shows the stage when slip occurs even at the center
of the splice. The distribution of slip in Fig. 28 is quite different from that in
pullout tests where concrete strain is negligible and thus slip distributes as
shown in Fig. 29. This difference might affect the sensitivity of embedment
length to bond strength.
The specimen of Sakurada et al. (1993) with shear reinforcement ratio Av

/b.s = 1.2%was analyzed. The calculated distribution of bond stress is shown in
Fig. 30, where x = O and 285 mm (= 15 d,) indicate the center and the end of
the splice, respectively. This gives an answer to [Ql]:
[Al] The tensile strength of concrete resists splitting and contributes to the
splice strength near the center of splice length, whereas the flexural strength of
the triangular beams with stirrup contribute near the edges. Thus, these two
actions can occur simultaneously.
The third question [Q3] on the notion of ‘local bond-slip’ is more

difficult to answer. The first point is whether we may apply the relationship
between radial stress and crack width such as Fig.15 to a minimal length dx or
not. Since the region dominated by tensile stress of concrete is about twice 140
mm in Fig. 30 due to the symmetry about x = O, and is much longer than the
clear spacing between the spliced bars, we may consider the first point
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
acceptable. The second point is whether we may average the horizontal forces
on the ribs in the three regions shown in Fig. 23. This may be acceptable if the
region dominated by triangular beams is much longer than the stirrup spacing.
In the case of Fig. 30, the stimp spacing was 80 mm, which may not be
considered acceptably small. However, the analysis could be valid as a rough
estimate.
EFFECT OF EPOXY COATING AND STIRRUPS
Analytical results of the specimens of Hester et al. (1993) without

stirrups were compared with the experimental results in terms of average bond
Obb2949 0543040 324
stress as shown in Fig. 31 . The main variables were the presence/absence of

epoxy coating, splice length and deformation pattern. Connected points
represent a pair of specimens with or without epoxy coating. The analytical
results agreed with the experimental ones irrespective of the variable settings.
Recall that epoxy coating did not impair the local bond strength in Fig. 16 but
only deteriorated the ductility. This deterioration, however, reduced the
average bond stress as shown in Fig. 3 1.
The analytical results for the specimens of Sakurada et al. (1993) are
shown by the solid line in Fig. 32, and are compared with the experimental
results represented by the black dots. The parameters were stirrup spacing and
the presence/absence of an inner stirrup. The contribution of the stirrup to the
analytical bond strength is 20% larger when inner stirrups are provided, which
agrees with the test. The predictions of the equations of Orangun et al. (1977)
and Darwin et al. (1996) are also plotted in the figure. In the latter case, the
effect of the relative rib area was also considered. The contribution of the
stirrup was observed beyond the upper limit of Orangun's equation. Darwin's
equation predicted well the contribution of the stirrups but not that of the
concrete.
The solid line in Fig. 33 or 34 shows the analytical results of fictitious

specimens where the main bar diameter is doubled to 38 mm or the stirrup
diameter is reduced to 6 mm, keeping the other parameters such as clear spacing
etc. the same as those of Fig. 32 (b). The vertical arrows in Figs. 33 and 34
show the points where s/d, = 1, around where the analytical results cease to
increase. At this point, the section of triangular beams becomes as shown in
Fig. 35 so that the sum of the strength of triangular beams ceases to increase. In
usual design where such small spacing is impossible, however, we need not
consider an upper limit on the contribution of stirrups.
The solid line in Fig. 36 shows the analytical results of fictitious

specimens where only the yield strength of stirrup is varied, keeping the other
parameters the same as the case of Au/b.s = 1.2% in Fig. 32 (b). The effect of
the yield strength is small enough in the practical range ( > 300 MPa) as was
assumed by Darwin et al. (1996). This is because flexural strength of triangular
beam is more affected by concrete strength than by stirrup strength.
CONCLUSIONS
The analytical results agreed well with the experimental ones. If the
assumptions of this study are correct, the following conclusions are obtained.
(1) The tensile resistance of the concrete against splitting contributes to the
splice strength near the center of splice length, whereas the flexural strength of
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 0543043 2b0 = Tribute to Peter Gergely 443
the triangular beams with stirrups contribute near the edges. Thus, these two
actions can occur simultaneously.
(2) Lack of friction at rib face due to epoxy coating deteriorates the ductility of
local bond-slip relationship without stirrups, which results in smaller splice
strength.
(3) The notion of local bond-slip relationship in this study is valid when the
splice length is much longer than both the clear spacing of the bars and the
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stirrup spacing.
(4)The diameter of spliced bar increases the strength of the triangular beam,
and therefore may contribute to Tsin Eq. (1).
(5) Stirrups do not yield in tension at splitting cracks because the triangular
beam fails in flexure first. The flexural strength of the triangular beam is more
affected by concrete strength than by stirrup strength. The effect of stirrup
strength on splice strength may be neglected in the practical range.
(6) An upper limit of the contribution of stirrups to splice strength need not be
considered in the practical range.
ACKNOWLWEDGEMENTS
Critical reading and valuable comments by Dr. John Bolander Jr. of the
University of California, Davis, is gratefully acknowledged.
REFERENCES
Akashi, J., S. Fujii, and S. Morita (1991): Effects of Concrete Strength and
Deformation Shape on Bond Strength, Proceedings of the Japan Concrete
Institute, V. 18, No. 2, pp. 127-132, (in Japanese)
CEB-FE’ Model Code (1990). Bulletin d’Information No. 190a, 190b, Comite
Euro-International du Beton (CEB), Lausanne.
Darwin, D., M. L. Tholen, E. K.Idun, and J. Zou (1996): Splice Strength of

High Relative Area Reinforcing Bars, AC1 Structural Journal, V. 93, No. 1,
Jan.-Feb., pp. 95-107
Hester, C. J., S . Salamizavaregh, D. Darwin, and S . L.McCabe (1993): Bond of

Epoxy-Coated Reinforcement Splices, AC1 Structurai Journal, V. 90, No. 1,
Jan.-Feb., pp. 89-102
Kawai T. (1990). “Rigid-body-spring model”, Baihukan Publishing Co., Tokyo,

Japan, 147 pp. (in Japanese)
m 0662949 0543042 LT?
Orangun, C . O., J. E. Jirsa, and J. E. Breen (1977): A Reevaluation of Test Data

on Development Length and Splices, AC1 Journal, V. 74, pp. 114-122
Sakurada, T., N. Morohashi, and R. Tanaka (1993): Effect of Transverse

Reinforcement on Bond Splitting Strength of Lap Splices, Transactions of the
Japan Concrete Institute, V. 15, pp. 573-580
Tomosawa, F., M. Abe, and Y. Masuda (1994): Development of High Strength

Concrete, Concrete Journal, Japan Concrete institute, V. 32, No. 10, pp. 11-19,
(in Japanese)
Treece, R. A. and J. O. Jirsa (1989): Bond Strength of Epoxy-Coated

Reinforcing Bars, AC1 Materials Joumal,VSó, No. 2,March-April,pp. 167-174
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0662949 0543043 033 W
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
cc 2cs 242, 2c, cc

H H H H H
Fig. 1-Typical section of a beam with splice and side splitting cracks
- db em=-
Cc+CS
Fig. 2-Discretization of concrete around bar

2
Normal Spring ri^$ Body
Shear Spring
Fig. 3-Rigid-bodyspring model
6 Obb2949 0543044 T ï T I
Shear stress
Fig. 4-Assumed failure surface of concrete
Fig. 5-Cohesion c as a function of shear displacement VA
Fig. 6-Characteristic strength of concrete
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
D Obb2949 0543045 906 M
Unbonded
U (a) Specimen
(b) Failure mode
Fig. 7-Pullout test with steel pipe
Fig. 8-Direct shear strength ,T as a function of shear displacement VA
Crack width
Fig. 9-Softening rule of concrete in tension
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Obb2949 0 5 4 3 0 4 b 8 4 2 m
~
o 12
du=20 m m
__-_*---
o 02
O 20 40 60 80
fc'(MPa)
Fig. 10-Relationship between compressive strength and the crack width at which
tensile stress vanishes, w,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
(a) Original section (b) Modelled section
Fig. 11-Model of main bar
Concrete
Main Bir
(a) Pattern 1 (Mohr-Coulomb)
Concrete
& MainBar
(b) Pattern 2 (Direct shear)
Cimcrete
t
Main Bhr
(c) Pattern 3 (Friction at rib face)
Fig. 12-Failure patterns in front of ribs
= Obb2949 05q3047 789 Tribute to Peter Gergely 449
Fig. 13-Forces on potential failure plane in front of a rib
(a) Bar dirpension
63.5h-R
? ?
463.5
(b) Cross,section
$:Ki
Fig. 14-Specimens 8C3-16-0 of Hester et al. (1993)
g 2000
1500
2
r 1000 * -
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
9 500
O 002 004 006 O08 O10

Opening at the bar V, (mm)
Fig. 15-Relationship between vertical force and opening at the bar
I
I Ob62949 0543048 bL5
450 ichinose, Hayashi and Lin
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
I.., . . . . . I ‘ 1 . ... , ,..,I

O 0.025 0.050 0.i 75
Slip V , (mm)
Fig. 16-Local bond-slip relationship of Hester’s specimens
Cover concrete
Core concrete
(a) Section (b) Side view
Cover concrete
(c) Detail
(d) Isometric view
Fig. 17-Triangular beam
054304q
Stirrup
(b) Simplified section
Fig. 18-Simplified of triangular section

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 19-Moment-curvature relationship of triangular beam
Fig. 20-Assumed strain at which concrete loses its compressive stress
6 0662949 0543050 273
Fig. 21 -Deformation of triangular beam without inner stirrup
. . . . . . . . . . . . . . . .
Fig. 22-Deformation of triangular beam with inner stirrup
.. ... . ... . ...:,.. ~slimp'.

.......... . . . . . . . . .
4'
' '
......-. . Rib in Cover Concrete . :
'
Fig. 23-Failure of concrete at ribs in three regions
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0bb2q4q 0543053 'OT
. .
15.4
(a) Bar dimension
(b) Cross section
Fig. 24-Specimens of Sakurada et al.
Of . ' o4 ' 0:tl 1'2 1'6

Opening at the bar V, (mm)
Fig. 25-Relationship between vertical force and opening at the bar

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Intermediate bar
Comer bar
O 0.05 0.1 0.15 0.2

Slip V,(mm)
Fig. 26-Local bond-slip relationship of Sakurada's specimens
111 Obb2949 0543052 046 M
A 6 . A A
Fig. 27-Transfer of bond stress in a section
C.L. Concrete
(a) Before slip
A
I
I
, I
I
(b) Slip in end regions only
I Y
(c) Slip in wholc span
Fig. 28-Slip in splice
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Obb2949 0543053 Ti92 Tribute to Peter Gergely 455
Concrete
1 I
SV
1 Before slip
,n
S i b distribution
After slip
Fig. 29-Slip in pullout specimen
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
o)
2 2
O
140
Distance from the centery (mm)
Fig. 30-Calculated distribution of bond stress
2 3 4 5 6
Analytical results (MPa)
Fig. 31-Splice strength of specimens of Hester et al. (1993)
= 0662949 0543054 Y19
........
5
F 3
L
Expenment
O 0 2 0.4 0 6 0.8 1 1.2
A,/(h s) ('70)
(a) Without inner stirrup
Analyst
Darwin
O 0 4 O 8 1 2 16 2 24
A,/(b s) ('70)
(b) With inner stirrup
Fig. 32-Splice strength of specimens of Sakurada et al. (1993)
45; ' ' ' ' '
0. ,:
L 3 Analysis
d d h =1
(A
O 0 4 O 8 12 16 2 4
A,/(h s) (70)
Fig. 33-Splice strength of D3812 bar
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Obb2949 0543055 855 m
AvKb ' S) (%)
Fig. 34-Splice strength when stirrup is D6 (#2)
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
S
Fig. 35-Triangular beams when s/db= 1
O 200 400 600 800 loo0

Yield strength, cry (MPa)
Fig. 36-Effect of yeild strength of stirrup on splice strength

(A,l(b.s)= 1.2 percent
ObbZ949 054305b 791
SP 180-20
Bond Properties of High-Strength Fiber

Reinforced Concrete
by B. Aarup and B. C. Jensen
Synopsis:
CRC (Compact Reinforced Composite) is a special concept for high

performance concretes, where ductility is achieved through incorporation of
a large content of short, stiff and strong steel fibres (6 vol.%). This ductility
combined with high strength (150-400 MPa) - and the ability of the small fi-
bres to provide an effective reinforcement against even small cracks, makes
it possible to obtain exceptional bond properties for deformed reinforcing
bars.
Results show that full anchorage is achieved with an embedment

length of only 5-10 diameters for ribbed bars, which has led to applications
in buildings, where CRC is used for in-situ cast joints between pre-cast
decks -joints which can transfer full moments with a width of 100 mm.
This type of joint, which was used for a university building, has been
extensively tested for different loading situations and for behaviour in a
standard fire. As the fibre reinforced matrix provides a strong, ductile joint
which responds well to dynamic loads it is expected to perform well under
seismic loads.
Keywords: Applications; bond; fiber concrete; fire resistance; high performance;

high-strength; joints; silica fume; steel fiber
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
459
9 Obb2949 0543057 628 m
460 Aarup and Jensen
Bendt Aarup (M.Sc., Civil Engineer) has been involved in cement and
concrete research since 1988 at the Cement & Concrete Laboratory of Aal-
borg Portland A / S , PO Box 165, DK-9100 Aalborg. He has mainly worked
with High Performance Concrete and Fiber Reinforced Concrete and is now
manager of the CRC Marketing Department.
Bjarne Chr. Jensen (Professor, Ph.D., Civil Engineer) is a manager of

research works in the consulting engineering company Carl Bro A/S (Ver-
mehrensvej 14, DK-5230 Odense). He is personally active within concrete
research work on structural design. He is active in Danish and European
works on Structural Design Codes.
INTRODUCTION
Extensive research has been carried out to determine the bond proper-
ties of reinforcing bars in conventional concrete and a number of models
have been proposed. However, the behavior in special concretes is some-
what less well described. Some investigations have been carried out with re-
gard to the effect of using High Performance Concrete ( 1,2), a few have
concentrated on the effect of using condensed silica fume ( 3 ) , while another
few have investigated the effect of using different types of fibers to rein-
force the concrete (4).Finally, investigations regarding bond in SIFCON
(Slurry Infiltrated Fiber CONcrete) have been carried out (5).
Generally, bond is improved with improved quality of the concrete,

but the pull-out failure in HSC is usually also more brittle than for conven-
tional concrete. Condensed silica fume provides a moderate improvement of
the bond properties, while the use of steel fiber reinforcement improves
ductility rather than ultimate bond stresses, unless fiber contents are very
high - as in SIFCON - in which case also ultimate stresses are improved. In
a special type of concrete called CRC these factors - high performance, sil-
ica fume and fiber reinforcement - are combined, and a number of pull-out
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
tests have been carried out to investigate the effect on bond properties of de-
formed reinforcing bars using this type of concrete.
m Obb2949 0543058 5b4 9
COMPACT REINFORCED COMPOSITE (CRC)
CRC - developed in 1986 at Aalborg Portland - is a special concept

for high performance concretes, where ductility is achieved through incor-
poration of a large content of short, stiff and strong steel fibers (typically
3-6 vol.%). This ductility combined with high compressive strength
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
( 150-400 MPa) and exceptional durability makes it possible to utilize a
large amount of reinforcement, thus giving new structural possibilities com-
pared to conventional concrete.
The mechanical properties of CRC are indicated as a range, as the

properties depend on type of aggregate, fiber type and content and type of
main reinforcement. E.g. for compressive strength a range of 150-400 MPa
is given, as a compressive strength of 150 MPa measured on cylinders is
achieved with quartz aggregates, but with calcined bauxite aggregates and
heat curing a compressive strength of more than 400 MPa can be achieved.
The typical mix composition - what could be called the standard com-
position - of CRC, however, includes a binder with a large content of micro
silica, a waterhinder ration of O. 16, quartz sand up to 4 mm and 6% by vol-
ume of steel fibers with a length of 12.5 mm and a diameter of 0.4 mm. This
typical composition will give a compressive strength of 150 MPa and a
bending strength of the matrix of 25 MPa. When nothing else is mentioned
in the text, the standard composition has been used. This composition can
be mixed and placed with standard equipment, whereas some of the more
exotic compositions - such as mortars with 12% by volume of small steel fi-
bers - would require special equipment.
CRC has shown very impressive behavior in bending, where the

bending strength of a reinforced CRC beam may be higher than 320 MPa,
approaching the strength of steel, but with only half the density. The possi-
bility of having another alternative to steel and concrete provide designers
and builders with more freedom in design, giving better and more cost ef-
fective solutions (6).
OVERVIEW OF PULL-OUT TESTS
A number of tests have been carried out regarding bond properties of

reinforcing bars in CRC over the past ten years. Some of the results have ai-
ready been reported in detail, and in the following will be given an over-
0662947 0543059 4TO =
view of the tests carried out, describing the parameters investigated and the
relevant results.
Some of the first tests carried out investigated the effect of varying
the cover to diameter ratio (7). The two types of test specimens used - types
FA and FAK - are shown in fig. 1. Compressive strength of the concrete
was i60 MPa and 6 % of brass coated steel fibers with dimensions 6 ~ 0 . 1 5
mm were used. For reasons of comparison, tests were also carried out on
concretes without fibers and in this case compressive strength was 96 MPa.
The bars that were pulled out were 8 mm deformed bars with a yield
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
strength of 644 MPa and an ultimate strength of 765 MPa. Results are pre-
sented in fig. 2 where ultimate bond stress as weil as pull-out force is
shown. With the very short lengths of embedment used, the results are obvi-
ously influenced by boundary conditions.
These tests were later expanded on in connection with a EUREKA-

project with partners from Denmark and England (8). The specimens tested
in this project are shown in fig. 3, and as can be seen they are a modified
version of the flat FA specimens. The specimen in fig. 3 is shown with ap-
plication of a lateral pressure, as this was one of the parameters investi-
gated. Other specimens included transverse reinforcing bars. Compressive
strength of the concrete was 165 MPa, 6 % by volume of 6 ~ 0 . 1 5mm steel
fibers were used, and yield strength of the 8 mm deformed reinforcing bars
was 577 MPa.
To achieve yielding i n the reinforcing bar with diameter 8 mm before

pull-out, an embedment length of 50 mm is necessary, but if confinement
pressure is applied or transverse reinforcing bars are used, an embedment
length of only 30 mm is adequate for achieving full anchorage. A lateral
compressive stress of only 5% of the compressive strength of the concrete
resulted in an increase in the bond strength of more than 60%.
Fatigue was also investigated using pull-out specimens, and results

show that if embedment length is sufficient to achieve full anchorage for
static load, fatigue failure will always occur in the reinforcing bar (9). This
coincides well with other investigations on fatigue of CRC, where redistri-
bution of stresses and ductility of the material influences fatigue behavior,
so that CRC shows better behavior in fatigue than conventional concrete.
As most of the CRC applications are performed with a slightly larger

steel fiber - a fiber with dimensions 12.5x0.4 mm - a comparative test series
M Obb2949 05430b0 112
has also been carried out, where the type and content of fibers has been var-
ied. The results are shown in table 1 . The better results are obtained with a
large content of the smaller fiber, as the size of the fiber relates to the size
of the cracks against which the fiber becomes effective.
A recent investigation (lo), carried out in connection with the MINIS-

TRUCT project, a project under the Brite/EuRam programme with support
from the EU Commission, and with partners from Denmark (Aalborg Port-
land and Carl Bro), France (Bouygues) and Spain (Instituto Eduardo Tor-
roja), used the test specimens shown in fig. 4. The size of the deformed
reinforcing bar and the ratio of cover to diameter of reinforcing bar was var-
ied.
The concrete had a compressive strength of 150 MPa and contained

6% by volume of the 12.5x0.4 mm steel fibers. Ultimate strength of the de-
formed bars was approximately 700 MPa.
Based on a curve fit of the test results a model was developed for esti-
mation of the bond of deformed steel bars in CRC.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
-
TU
=OS + 17 $,+0.7 2
Jf.
where T,= ultimate bond stress in MPa
f, = CRC compressive strength in MPa
c = concrete cover
d = reinforcing bar diameter
L = embedded length of reinforcing bar
$, = -
nA >,
dL
A,,,= cross-sectional area of a transverse reinforcement bar
n = number of transverse bars
The first part of the model consists of a constant Contribution, which

relates to the roughness of the bar. For prestressing strands which are rather
smooth and which have been used as passive reinforcement in CRC beams,
there would be no adhesion. A second part of the model is the contribution
from the presence of transverse reinforcing bars, which prevent splitting
cracks from developing. This contribution is again dependent on the rough-
ness of the reinforcing bar. Finally, a contribution is added depending on the
cover thickness.
Obb2949 05430bL O59
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Not included in the model is the contribution from transverse stresses

induced in the anchorage zone, i.e. when placed over supports.
In fig. 5 is shown a comparison between calculated anchorage length

in a 75 MPa concrete - from (1 1) - and in CRC calculated according to the
proposed model. As can be seen the effect of improved bond is most pro-
nounced for small covers. The comparison has been made for a case with no
transverse reinforcing bars.
APPLICATION
The superior bond properties of CRC have been utilized in a new,

flexible building system which has recently been used for new buildings at
the University in Aalborg, Denmark. A column/beam/slab system has previ-
ously been used for university buildings, but in order to have more flexibil-
ity with regard to changes in the interior of the building the Ministry of
Education wanted a column-slab system, with fewer restrictions on changes
in the facade and the layout of the rooms in the building. This could be ac-
complished with an in-situ cast slab, but the builder also wanted the speed
of construction, the quality and the degree of quality control which can be
achieved with a precast system.
A research project sponsored by the Ministry of Education and led by

Carl Bro A/S was initiated and it soon turned out, that the objectives of the
project - to design and implement a flexible building system in a 6x6 m grid
- could be achieved with conventional columns and 3x6 m slabs - in con-
nection with the use of CRC as a jointing material. The slabs - with a depth
of 200 mm - were cast with 80 mm of protruding deformed 8 mm reinforc-
ing bars and placed at a distance of 100 mm from each other as shown in
fig. 6. The joint - with a width of 100 mm - was poured with the standard
CRC mortar and compacted with a poker vibrator to ensure that the system
would have the strength and stiffness of an in-situ cast slab. That way the
architect would not have to consider the placement of load-bearing walls or
beams in designing the lay-out of the building.
The system is shown in fig. 7, and as can be observed the joints are
placed in the position where the largest moment is anticipated. As this sys-
tem had not previously been used in buildings and could not be validated
based on existing standards, a rather comprehensive set of tests was carried
out ( 1 2). The investigations included pull-out tests, tests on column-slab
connections, beam-beam connections and fire resistance tests. All tests
= O662949 0543062 T95 Tribute to Peter Gergely 465
showed that failure would occur as yielding in the reinforcement outside the
joints - except for the case when a connection was loaded to failure after
one hour of exposure to a standard fire ( 1 3). In this case the surface of the
connection was sufficiently weakened by the fire to cause a pull-out failure
at 75% of the load sustained on un-damaged specimens. This was, however,
well above the service load so the system could be classified as BS60 (fire
resistant for at least 60 minutes according to the Danish standard). If fire ex-
posure was continued for 97 minutes under service load, failure took place
as yielding in the reinforcement.
The tests also included an investigation on the robustness of the sys-

tem - the sensitivity to placing of transverse reinforcing bars, length of em-
bedment and changes in geometry - and it was concluded that a high degree
of safety was available in the system.
The building was finished in 1996 and the cost of the new system was
10-15% less than what would have been the cost of the "old" system with
columns, beams and slabs. Also, the job went smoothly on site, and the con-
tractor could use conventional equipment. The system demands a high de-
gree of precision, but this was achieved without problems by the contractor
and the producer of the precast slabs.
A new building is scheduled for erection in 1997 and the Ministry of

Education has requested yet another development project, where the aim is
to extend the effectiveness of the joints further, by making a stiff joint be-
tween the columns and the slabs. This possibility has already been demon-
strated in the MINISTRUCT project mentioned earlier, and the current
project will aim at optimizing the joint further and performing the necessary
tests to have the system accepted by the authorities in Aalborg.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
CONCLUSIONS
A number of pull-out tests have been performed on a special high

strength steel fiber reinforced concrete called CRC. These tests have been
compared to results obtained on conventional and high strength concrete,
showing a considerable improvement of bond of deformed reinforcing bars
in CRC and a simple model based on a frictional and an adhesional contri-
bution is proposed to describe behavior of bond of deformed bars in CRC.
The material has been used for a building system, where the bond
properties of CRC were utilized beyond what is commonly accepted in stan-
9 Ob62949 0543063 923 =
dards, but it was demonstrated in tests that the carrying capacity of the
joints was adequate to ensure that the system acted monolitically.
This system is to be developed further, and it is expected that due to

the ductility of CRC it would be especially well suited for buildings in seis-
mic areas where joints are often a problem in precast buildings. This as-
sumption will hopefully be tested in a new project which has recently been
proposed under the Brite/EuRam program with a number of European part-
ners. The project - if accepted by the European Commission - is specifically
aimed at improving design in structures exposed to seismic action by using
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
high performance concrete.
REFERENCES
1. Noghabai, K. & Ohlsson, U. & Olofsson, T., "Bond Properties of
High-Strength Concrete", Proceedings from the Third Int. Symposium
on Utilization of High Strength Concrete, June 1993, Lillehammer,
Norway, pp. 1169-1176.
2. Larrard, F. & Schaller, I & Fuchs, J., "Effect of Bar Diameter on the
Bond Strength of Passive Reinforcement in High-Performance Con-
crete", ACI Materials Journal, July-August 1993, pp. 333-339.
3. G j ~ r vO.E.
, & Monteiro, P & Mehta, P.K., "Effect of Condensed
Silica Fume on the Steel-Concrete Bond", AC1 Materials Journal,
November-December, 1990, pp. 573-580.
4. Harajli, M.H., "Development/Splice Strength of Reinforcing Bars

Embedded in Plain and Fiber Reinforced Concrete", ACI Structural
Journal, September-October 1994, pp. 5 11-520.
5. Hamza, Ali M. & Naaman, Antoine E., "Bond Characteristics of De-

formed Reinforcing Steel Bars Embedded in STFCON", ACZ Materials
Journal. November-December 1996.
6. Aarup, Bendt, "Ultra High-Strength Concrete", 22nd Annual Conven-

tion of the Institute of Concrete Technology, 28-30 March 1994,
Swindon, England.
œ Obb2949 0543064 868 œ
7. Heshe, G., "Experimental Research on Compact Reinforced Compos-

ite (CRC) Beams", Bygningsstatiske Meddelelser, vol. 59, no. 1,
1988,83 pp. Dansk Selskab for Bygningsstatik, KØbenhavn, 1988.
Also available as CBL Reprint no. 19.
8. Aarup, B. & Nepper-Christensen, P., "Ultra High-Strength Con-

crete", XIV Nordisk Betonkongres & Nordisk Betonindustrimode,
Reykjavik 6-8 August 1992. Also available as CBL Reprint no. 24.
9. Nielsen, C.V., "Presentation of Cyclic Load Tests of Rebars An-

chored in Steel Fiber Reinforced High-Strength Composite", Fatigue
of Concrete Structures (ed. L.P. Hansen), Department of Building
Technology and Structural Engineering, Aalborg University, Aalborg,
Denmark, 1993, pp. 65-71.
10. Nielsen, C.V. & Olesen, J.F. & Aarup, B, "Effect of fibers on the
bond strength of high strength concrete". BHP96 Fourth International
Symposium on Utilization of High-S trength/High-Performance Con-
crete, 29-3 1 May, 1996, Paris, France.
11. Al-Jahdali, F.A. & Wafa, F.F. & Shihata, S.A., "Development
length for Straight Deformed Bars in High-Strength Concrete". High
Performance Concrete, 507-521 (ed. V.M. Malhotra). SP- 149, AC1
1994, Detroit, USA.
12. Jensen, B.C. & Jensen, L.R. & Hansen, L.P. & Hansen, F.T., "Con-
nections in Precast Buildings Using Ultra High-Strength Fiber Rein-
forced Concrete", Proceedings: Nordic Symposium on Modern Design
of Concrete Structures, Aalborg University, Denmark, May 3-5, 1995.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
13. Jensen, B.C. & Aarup, B, "Fire Resistance of Fiber Reinforced Silica
Fume Based Concrete". BHP96 Fourth International Symposium on
Utilization of High-StrengtMHigh-PerformanceConcrete, 29-3 1 May,
1996, Paris, France.
0662949 0543065 7 T 4
TABLE 1-EFFECT OF USING DIFFERENT TYPES OF FIBERS. THE 8 MM

DEFORMED BARS YIELD AT A LOAD OF 29 kN AND INDICATES THAT AT
LEAST ONE OF THE BARS IN A SERIES OF THREE HAVE YIELDED BEFORE PULL-
OUT
Fiber type Fiber content Compressive Splitting Embedment Ultimate force

strength strength length
[vol.%] iMPa1 WPal [mml ikN1
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W Obb2’449 05430bb b30
measuring length I
1 A 7 I
4
4
f66
I I
I I,¡ 1
I
5 5 0 - 2 x 1, , Li
1 110 I 550 ,
1
110
FA A A
B-l measuring length I

- I-t
, I*, ,
I
4 7 6 - 2 x I, , Id ,
125 I
I 476 I 125 ,
FAK BI. l
d = 8mm measurements in mm
,9, 2 4 , 24 ,9
ISF8 jqo
10
66 I 10,15,1510
t
1 50 I
Fig. 1-Test specimen FA and FAK for pull-out tests
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
9 Obh29Liï 0 5 4 3 0 6 7 5 7 7 9
embedment length
Fig. 2-Pull-out force P, and ultimate bond stress t, as functions of the

embedment length for anchorage prisms FA and FAK
tp
-deformed rein-
+P forcement bars
a a
u _ . .0 . , '
.* .. ..'I.' '0 . :. 0.
u
lateml pressure
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
.i0 25
Fig. 3-Test principal for pull-out of reinforcing bars
m Obb2949 05430bô 403 m
Embedded bar
Through bars
Fig. 4-Concrete prisms used for investigations of anchorage
Fig. 5-Ratio between anchorage length on 75 MPa concrete (1 1) and CRC
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~ ~
IOhh29Y9 0543069 3 4 T
Fig. 6-View from above of part of the slab connection. All measures in mm.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Facing wall
-
.u- .-.....
Fig. 7-The building system used at Aalborg University
9 Obb2949 0543070 Oh1
SP 180-21
Bond Studies of Reinforcing Bars

in Silica Fume Concrete
by B. S. Hamad and M. S. Itani
SvnoDsis: This paper reports on research in progress conducted at the

American University of Beirut to evaluate the effect of silica fume on bond
and anchorage of reinforcement in high performance concrete (HPC)
structures. The program includes testing the effect of a wide range of
variables on the bond strength of beam bar splices and bars anchored in
pullout specimens.
Results of the first phase of the research program have been analyzed.
Ten beam specimens were tested. Each beam was designed to include two
bars in tension, spliced at the center of the span. The splice length was
selected so that bars would fail in bond, splitting the concrete cover in the
splice region, before reaching the yield point. The beams were loaded in
positive bending with the splice in a constant moment region. The variables
used were the percentage replacement of cement by silica fume and the
casting position. Test results indicated that replacement of 5 to 20 percent of
the cement by an equal weight of silica fume resulted in an average 8 percent
reduction in bond strength regardless of casting position.
Keywords: Bond (concrete to reinforcement); development length; high-strength

concrete; high performance concrete; reinforced concrete; silica fume
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
473
0662949 0543073 T T B
474 Hamad and Itani
Bila1 S. Hamad is currently an Associate Professor of Civil Engineering at

the American University of Beirut, Lebanon. He received his BE degree from
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
AUB and MS and PhD degrees from The University of Texas at Austin. He is
member of ACI, ASCE, and IABSE. His research interests include design and
behavior of reinforced concrete structures and concrete technology. He
received the AC1 Structural Research Award in 1995.
Maher S. Itani is a structural engineer. He received his BE and ME from The

American University of Beirut, Lebanon.
INTRODUCTION
Silica Fume Concrete
Production of HPC depends not only on the use of very low water-to-
cementitious material ratio with the aid of high-range water-reducers or
superplasticizers, but also on other factors. These factors include
superplasticizer-cement compatibility, coarse aggregate quality, mixing
equipment, and a careful selection and dosage of a mineral pozzolanic
admixture such as silica fume. Silica fume is a by-product resulting from the
reduction of high purity quartz with coal in electric arc furnaces in the
manufacture of ferro-silicon and silicon metal. The fume which contains
between 85 and 98 percent silicon dioxide (Siof), and consists of extremely
fine spherical glassy particles, is collected by filtering the gases escaping
from the furnaces. The average particle size is O. 1 mm or about two orders of
magnitude finer than cement particles. The specific area is in the order of 20
m2/g as measured by the nitrogen absorption method, and the relative density
is 2.2.
Research reported in the literature on the effect of silica fume on fresh

concrete and on the strength and durability characteristics of hardened
concrete, has indicated that silica fume increases the strength of the cement
paste by acting as a filler for the space between cement grains. Because of its
fine size, silica fume results in greater cohesiveness, reduced segregation, and
reduced bleed water products underneath the aggregate particles and
reinforcing bars. Also, the pozzolanic action of the amorphous silica-rich
fume reduces the number and size of calcium hydroxide crystals in the
cement paste-aggregate transition zone. This results in thinner and less porous
transition zone. Most of the improvement in HPC is attributed to this
interfacial modification and to the improved bond between the cement paste
and aggregates.
= 0662949 0543072 934 Tribute to Peter Gergely 475
Bond of Reinforcing Bars in HiPh Streneth Concrete
On the basis of the reported research (1-5) on bond performance of

reinforcing bars in HSC, four important observations can be made:
1. Inspection of reinforcing bar surfaces in splice regions of HSC test
specimens after bond splitting failure indicated much less concrete crushing
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
in the vicinity of the bar lugs as compared with NSC test specimens.
2. The normalized bond strength ('isI )
lower for HSC than for
4E
NSC.
3. Top cast bars in HSC resulted in slightly higher bond strength when
compared with bottom cast bars.
4. HSC specimens with splices designed based on the 1983 and 1989
AC1 building codes (AC1 3 18-83 and AC1 3 18-89) (6,7), without applying the
upper limit of 70 MPa (1 0,000 psi) imposed on f by Section 12.1.2 of the
code for determining the development length and without transverse
reinforcement in the splice region, failed in a very brittle and undesirable
splitting mode in the splice region without exhibiting any level of ductility.
Previous Research on the Effect of Silica Fume on Bond Strength
Very little research has been reported on the effect of silica fume on
bond and anchorage of reinforcement in concrete structures.
In 1990, Gjorv, Monteiro, and Mehta reported on a series of pullout

tests, carried out according to ASTM C 234, to study the effect of condensed
silica fume (CSF) on the mechanical behavior of the steel-concrete bond (8).
The pullout strength of # 6 (20 mm) Grade 60 deformed steel bars was
investigated at four levels of concrete compressive strength: 3000, 6000,
9000, and 12000psi (35,42,63, and 84 MPa). For these strength levels, three
levels of CSF were used O, 8, and 16 percent by weight of cement. According
to the test results, Gjorv, et al., concluded that the pullout strength of
deformed bars increased with increasing compressive strength. Increasing
additions of CSF up to 16 percent by weight of cement showed an improving
effect on pullout strength up to concrete strength of around 11,000 psi (76
MPa).
In 1994, Hwang, Lee Y., and Lee C., reported on an experimental

study designed to examine the effect of silica fume on the splitting bond
strength of deformed bars of high performance concrete (9). Eight beams in
four pairs were tested. Each pair included a specimen with plain portland
cement concrete and one with concrete in which 10 percent of the portland
0662749 0543073 870
476 Hamad and Itani
cement was replaced by equal weight of silica fume. Concrete compressive

strength values ranged from 62.1 to 84 MPa. All beams were tested with lap-
spliced #9 deformed bars centered on mid span in a region of constant
negative bending. The splice length of 300 mm (12 in.) was selected so that
bars would fail by bond splitting before reaching yield. In all specimens,
failure was characterized by face-and-side splitting of the concrete cover.
Based on the data reported, Hwang, et al. concluded that silica fume
decreased bond strength of deformed bars in high performance concrete with
water-to-cementitious material ratios in the range of 0.28 to 0.33. Bond
degradation was attributed to the loss of adhesion between concrete and steel
at the ribs due to the presence of silica fume.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ten beams were tested in positive bending. The loading system was
designed to produce a constant moment region in the middle of the beam
specimen. Reinforcement on the tension side consisted of two 25-mm (#8)
Grade 60 reinforcing bars spliced at the center of the span. No transverse
reinforcement was provided in the splice region. The variables used were the
percentage replacement by weight of portland cement by silica fume (O, 5, 10,
15, and 20 percent) and casting position (top or bottom). The test specimens
are identified in Table 1. A three part notation system was used to indicate the
variables of each beam. The first part of the notation indicates whether the
beam specimen contained silica fume (SC), or not (PC). The second part is
the percentage of silica fume used as a replacement for portland cement: O, 5,
10, 15, or 20 percent. The third is the casting position of the specimen; i.e.,
whether it is bottom cast (B) or top cast (T).
The splice length of the deformed bars was set at 305 mm (12 in.) in
all beam specimens. This value was selected to develop a steel stress less than
yield to ensure splitting mode failure in all beam specimens. A concrete cover
of 38 mm (1.5 in.) to the reinforcing bars in the splice region was chosen as a
typical side and bottom or top cover. The clear spacing between the splices
was 76 mm (3 in.) or twice the cover, allowing identical concrete confinement
for both splices. With 38-mm cover and 76-mm clear spacing between the
splices, the beam width was 256 mm (10.1 in.). The depth was 305 mm (12
in.) for all beams. The length of the beam was chosen to be 2000 mm (78.75
in.) with a distance of 1800 mm (70.90 in.) between the supports. The
distance between the two applied concentrated loads was 650 mm (25.60 in.).
Longitudinal and cross-section details of the beam specimens are shown in
Figure 1.
Obb2qYq 0543074 707 m
Bars used were fiom the same heat of steel and had the same parallel
(bamboo) deformation pattern. The bars met ASTM specifications and were
Grade 60. Transverse reinforcement used in the shear spans to avoid shear
failure were 10-mm (#9) Grade 60 deformed bars. The average yield stress
was 426 MPa (61.8 ksi) for the 25-mm bars and 525 MPa (76.1 ksi) for the
10-mm bars. Non air-entrained concrete mixes were designed to provide
minimal concrete compressive strength of 70 MPa (10,000 psi). Water-to-
cementitious materials ratio ranged from 0.33 to 0.37. The silica fume used
contained 96.7 percent of Si02. In all ten beams, the superplasticizer dosage
was 2 liters per 100 kg of cementitious material. The superplastcizer had 40
percent solids and had a specific weight of 1.2. Concrete was cast in two
layers in each beam specimen. The slump varied from 40 to 215 mm.
The specimens were tested using the M.T.S. (Materials Testing

System) closed loop servo-hydraulic machine with a 1000-kN capacity
dynamic actuator. Since all beams were tested in positive bending, specimens
cast in a top cast position were turned upside down before being placed in the
test frame. Load was applied incrementally until failure occurred. At each
load stage, deflection readings were taken at the center of the beam using a
dial gauge and flexural cracks were marked and measured using a crack-width
compactor.
MODE OF FAILURE
Failure of the beams occurred just after longitudinal splitting cracks

formed in the bottom cover directly below the splice region or between the
lapped bars and in the side cover adjacent to the bars. All beams failed in a
sudden face-and-side splitting mode of failure. The failure was brittle and
noisy. The observed cracking patterns on the bottom tension face and on the
side of all beam specimens were similar regardless of the percentage
replacement of cement by silica fume or casting position. After failure, the
bottom (tension face) and side concrete covers of few beams were removed to
reveal the failure plane in the splice region. Better adherence between the
reinforcing bars and the surrounding concrete was noted in beams with no
silica fume as compared with beams with silica fume replacing part of the
cement. This observation suggests that adhesion between the steel and
concrete with silica fume is relatively poor.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
478 Hamad and Itani
= 0662949 0543075 643 E
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
TESTRESULTS
To allow direct comparison of all beam specimens, the corresponding

load-deflection data and bond strengths were normalized at a common
concrete strength of 70 MPa (10,000 psi). The adjustment was made by
112
multiplying the load at each deflection or crack width by (7% ,where f
is the concrete strength in MPa of the beam specimen under consideration at
the day of testing. The stress in the steel, fs, was calculated based on elastic
cracked section analysis. To evaluate the bond stress, 4, the total force
developed in the bar, Abf,, was divided by the surface area of the bar over the
splice length, pd&:
(Abfs)
Ut =
(ndbls)
(fsdb)
Ut = Eq. 1
41s
Results of the ten beam specimens tested in the research program are
presented in Table 1. The listed data includes the parameters of each beam,
the ultimate load at failure (Pmm) normalized at f = 70 MPa (10,000 psi),
and the corresponding: deflection at the center of the beam (d), steel stress
(fs), bond stress (ut), and bond ratio. The bond ratio is the bond stress of the
beam with silica fume divided by the bond stress of beam with no silica fume
in each group of beams.
ANALYSIS OF TEST RESULTS
Beam Stiffness
Five beams were cast in a bottom position with O, 5 , 10, 15 and 20

percent replacement silica fume. The beams had almost identical load-
deflection stiffness up to around 65 percent of the ultimate load. At higher
load levels, the beam without silica fume had greater stiffness (greater load
for a given deflection) than all other beams with different percentages
replacement silica fume. Load-deflection curves of the latter beams remained
to be very similar. The second set of five beams were top cast. Load-
deflection curves of these beams, shown in Figure 2, indicate very similar
stiffness up to the cracking load around 40 kN (9 Kips). After cracking, load-
deflection stiffness decreased (more deflection for the same load) as the
m 0hh2949 054307b 587'
amount of silica fume increased from O to I5 percent. The specimen with 20

percent silica fume had a greater stiffness than the specimen with 15 percent
silica fume. It could be concluded that regardless of casting position, the
replacement of part of portland cement by an equal weight of silica fume
resulted in a reduction of the load-deflection stiffness above the cracking load
level.
Load-deflection curves of top and bottom cast beams with no silica

fume were almost identical for all stages of loading. On the other hand, when
part of cement was replaced by silica fume, a slight decrease in load-
deflection stiffness of top cast beams relative to bottom cast beams was noted.
The difference between load-deflection curves of top and bottom cast beams
was most evident with 20 percent replacement silica fume (see Figure 3).
Bond Stren&h
Bond stresses corresponding to the ultimate loads of the ten beam

specimens, normalized at a common concrete compressive strength of 70
MPa (10,000 psi) and listed in Table 1, are plotted in Figure 4. Clearly, it
could be concluded that regardless of casting position, bottom or top, the
replacement of part of cement by an equal weight of silica fume resulted in
slight reduction in bond strength. The average bond ratio (bond strength of
beam with silica fume to beam with no silica fume) was 0.96 for bottom cast
beams and 0.88 for top cast beams.
The reduction in bond strength is attributed to the loss of adhesion

between the concrete and steel at the deformations due to presence of silica
fume. The importance of the friction component of the bond was recognized
by Lutz, Gergely, and Winter (lo), and by Treece and Jirsa (il). After
presenting the results of tests on beam specimens designed with epoxy-coated
bar splices, Treece and Jirsa presented a failure hypothesis of coated bars. The
hypothesis is illustrated in Figure 5. When the friction between the concrete
and the steel is lost, the only component of bond is the bearing force
perpendicular to the face of the rib. The vertical component of the resultant
bond force is the radial pressure which is controlled by the resistance of the
concrete cover to splitting. The horizontal component is the effective bond
strength. If the resistance to splitting is the same for either case, with friction
and without fiction, then the bar with no friction will have a smaller bond
capacity.
It should be noted that the reduction in bond strength of deformed bars

in silica fume concrete (8 percent on the average) is smaller than the reduction
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
480 Hamad and Itani
= Obh2949 0543077 416
of bond strength of epoxy-coated bars (around 30 to 35 percent). This is

because in the case of silica fume although the friction component is lost,
however, the radial pressure component of the bond is greater than in the case
of concrete with no silica fume. Silica fume increases the tensile strength of
concrete and hence the resistance of the cover to splitting due to the positive
effect it has on the strength of the cement paste and the bond between the
cement paste and the aggregate particles. This will provide a partial
compensation for the reduction in bond strength due to the loss of friction.
For beam specimens with no silica fume, top cast bar splices
developed slightly greater bond strength than companion bottom cast splices
(refer to Figure 4). To explain these results, reference is made to the failure
hypothesis of deformed bars in HSC which was presented by Azizinamini, et
al. (5). In HSC, the tensile capacity of concrete does not increase at the same
rate as the bearing capacity. The higher bearing capacity will prevent
participation of all bar lugs in resisting applied axial forces. With the first lugs
being more active, the bond stress distribution along the splice length is not
uniform. In the case of top bar splices, bleeding and segregation would result
in a layer of inferior concrete with relatively higher water-to-cement ratio
right below the reinforcing bars. This relatively weaker concrete would limit
the bearing capacity that each lug can provide and would prevent the creation
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
of a few highly stressed lugs. This would lead to the participation of more
lugs in the splice region and hence higher bond strength for top cast bar
splices than bottom cast bar splices in HSC.
For beam specimens with 5, 10, 15, or 20 percent placement silica

fume, top cast bar splices developed slightly lower bond strength than bottom
cast bar splices (refer to Figure 4). The average reduction was 7 percent. The
presence of silica fume increases concrete cohesiveness and therefore reduces
segregation and bleed water pockets under reinforcing bars and under coarse
aggregate particles. Hence, top cast bars would not benefit from the bleeding
phenomenon to develop more uniform stress transfer between steel and
concrete as compared with bottom cast bars. The slight reduction in bond
strength of top cast bars relative to bottom cast bars in silica fume concrete
could be explained by a slight reduction in the strength of the cement paste
and the splitting tensile strength of the concrete cover.
COMPARISON WITH AC1 318-95
The measured splice bond strength of each beam, corresponding to the

actual concrete compressive strength, was compared with the theoretical
D Ob62949 0543078 352
value computed using AC1 3 18-95 (12) bond specifications using 1, = 1.3 l d
and ignoring the limit on A imposed by the code. Taking into
consideration bar spacing and cover, then according to Section 12.2.2 of the
code:
Id fflßh
k 1 . 3 ld, -=-
db 20Jfc
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Combining the above equations with undbls= Abfy, and setting the
appropriate value of 1.0 to the reinforcement location factor (a), the coating
factor (b), and the lightweight aggregate concrete factor (1):
U = 3.85& Eq.2
The predicted bond stresses computed using Eq. 2 are listed in Table 2. The
measured bond stress for each specimen was divided by the predicted value to
obtain the bond efficiencies listed in Table 2. The mean bond efficiency for
ail bar splices is 2.74 with a standard deviation of 0.26. The results imply that
for the bar splices tested in the study, the current 1995 AC1 code provisions
are overly conservative.
ASSESSMENT OF THE AC1 LIMIT OF 10,000 psi ON fc
The current AC1 Building Code (AC1 3 18-95) places an upper limit of
70 MPa or 10,000 psi on the value of the concrete compressive strength that
may be utilized in calculating the required development length. In the
commentary to this requirement, AC1 3 18-89states that the limit is based on
the insufficiency of research data on development of bars in high strength
concrete.
The concrete compressive strength f C was greater than 70 MPa

(10,000 psi) in 7 out of the 10 beams tested in this research program. The
trend of the bond efficiencies relative to Eq. 2 (AC1 318-95 Section 12.2.2),
where no limit on f was used, shows no change that could be attributed to
whether f c was greater or smaller than 70 MPa (10,000 psi). Similar
observations were made by Azizinamini, et al. (5), and by Hwang, et al. (9).
Based on the above, the current AC1 limit on f c in computing

anchorage length seems to be unnecessary and unwarranted. However, before
9 Obb2949 0543079 299
482 Hamad and Itani
recommending the removal of the limit from Chapter 12 of the AC1 Code,
two points should be taken into consideration:
1. The mode of failure of bar splices in HSC is very brittle.
2. Only few lugs over the splice region contribute to the stress transfer
between the deformed bars and the surrounding HSC. Failure occurs before
all lugs participate in resisting the applied axial forces.
Such behavior could indicate that the removal of the current f c

limitation from the AC1 building code in designing tension splice lengths,
without some ductility requirement, could lead to structural components with
tension splices failing in a brittle mode of failure. This would be contrary to
ACI’s general design philosophy of precluding any brittle mode of failure
prior to achieving a certain level of ductility. One suggestion is to require
some minimum transverse reinforcement over spliced regions to allow all bar
lugs to be utilized in the stress transfer mechanism leading to higher bond
strength.
CONCLUSIONS
Based on the analysis and comparison of load-deflection curves and

bond strengths, the following conclusions were made:
1. Load-deflection curves of top and bottom cast beams with no silica
fume were almost identical for all stages of loading.
2. At load levels above the cracking load, the replacement of part of
portland cement by an equal weight of silica fume resulted in a reduction of
the load-deflection stiffness regardless of casting position. However, at
relatively low load levels, prior to cracking, the presence of silica fume did
not reduce the load-deflection stiffness of beams with the same casting
position. This proves that silica fume does not affect the flexural cracking
level of structural members.
3. The replacement of part of cement ( 5 to 20 percent) by an equal
weight of silica fume resulted in an average 8 percent reduction in bond
strength regardless of casting position.
4. For beam specimens with no silica fume, top cast bar splices
developed slightly greater bond strength than companion bottom cast splices.
This trend was reversed when part of the cement (5 to 20 percent) was
replaced by silica fume.
5 . When the measured bond stresses of bar splices were compared to
predicted values computed using AC1 3 18-95, it was found that the 1995 AC1
code provisions are overly conservative.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
= Ob62949 0543080 T O O D Tribute to Peter Gergely 483
6. Ratios of measured bond stresses of all 1O tension lap splices tested,

relative to predicted values computed according to AC1 318-95, showed no
change that could be attributed to whether f was greater or smaller than
10,000 psi or 70 MPa.
NOTATION
Ab = area of one reinforcing bar being spliced

db = diameter of reinforcing bar
fC = compressive strength of concrete
fS = stress in reinforcing steel
IS = length of lap splice
ld = development length
a = reinforcement location factor
ß = coating factor
h = lightweight aggregate concrete factor
ACKNOWLEDGMENT
The authors gratefully acknowledge the support of the University

Research Board at the American University of Beirut for supporting this
program. Also, the assistance of Mr. Hilmi Khatib, Supervisor of the
Materials Testing Laboratory at AUB is greatly appreciated.
REFERENCES
1. DeVries, R. A.; Moehle, J. P.; and Hester, W., “Lap-Splice Strength of
Plain and Epoxy-Coated Reinforcements: An Experimental Study
Considering Concrete Strength, Casting Position, and Anti-Bleeding
Additives,” Report No. CB/SEMM-91/02. Department of Civil Engineering,
University of Califomia-Berkeley, Berkeley, CA, January 1991.
2. Hadje-Ghaffari, H.; Darwin, D.; and McCabe, S., “Effects of Epoxy

Coating on the Bond of Reinforcing Steel to Concrete,” SA4 Report No. 28,
University of Kansas Center for Research, Lawrence, KS, July 1991.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
484 Hamad and Itani
= Ob62949 0543081 947 m
3. Kaku, T.; L i d a , S,; Yamada, M.; and Zhang, J., “A Proposal of Bond
Strength Equation for R.C. Members including High Strength Concrete
Level,” Proceedings of the International Conference on Bond in Concrete,
Riga, Latvia, October 1992.
4. Grundhoffer, T. M.; French, C.W.; and Leon, R.T., “Bond Behavior of

Uncoated and Epoxy-Coated Reinforcement in Concrete,” Report No. 92-04,
University of Minnesota, Minneapolis, MN, May 1992.
5 . Azizinamini, A.; Stark, M.; Roller, J. J.; and Ghosh, S.K., “Bond
Performance of Reinforcing Bars Embedded in High Strength Concrete,” AC1
Structural Journal, Vol. 90, No. 5 , Sept.-Oct. 1993, pp. 554-561.
6. AC1 Committee 3 18, “Building Code Requirements for Reinforced

Concrete and Commentary (ACI-3 18-83/ACI-318R-83),” American Concrete
Institute. Detroit, MI, 1989.

Concrete and Commentary (ACI-3 18-89/ACI-318R-89),” American Concrete
8. Gjorv, O. E.; Monteiro, P. J. M.; and Mehta, P. K., “Effect of Condensed

Silica Fume on the Steel-Concrete Bond,” AC1 Materials Journal, Vol. 87,
No. 6,Nov.-Dec. 1990, pp. 573-580.
9. Hwang, S.; Lee Y.; and Lee, C . , “Effect of Silica Fume on the Splice
Strength of Deformed Bars of High-Performance Concrete,” AC1 Structural
Journal, Vol. 91, No. 3, May-June 1994, pp. 294-302.
10. Lutz, L. A.; Gergely, P.; and Winter, G., “Mechanics of Bond and Slip of
Deformed Reinforcing Bars in Concrete,” Report no. 324, Department of
Structural Engineering, School of Civil Engineering, Cornell University,
Ithaca, Aug. 1966, pp. 299.
11. Treece, R. A.; and Jirsa, J. O., “Bond Strength of Epoxy-Coated

Reinforcing Bars,” AC1 Materials Journaf, Vol. 86, No. 2, Mar.-Apr. 1989,
pp. 167-174.

Concrete and Commentary (ACI-3 18-95íACI-318R-951,’’ American Concrete
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M Obb2949 0543082 883
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~~
W 0662747 0543083 7 1 T
486 Hamad and Itani
*
v)
E
0
o
W
L
VI
5
Y
m
w
I
l-
U
O
v)
u I
E
U
U
Y
n
æ
O
m
O
æ
a
z
v)
v)
w
er
G
O
æ
O
m
hl
I
w
2
m
e
I-
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m 0662949 0543084 b5b m
N
i-
-0-
n
m
h
-I
7
a3
m
L
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
m Ob62949 0543085 592
488 Hamad and Itani
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0 b b 2 9 4 9 0 5 4 3 0 8 b 429 = Tribute to Peter Gergely 489
0 0 0 0 0 0 0 0 0
W W ~ O a J W b C u
Y 7 Y
NM 'd
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
~
I0bb2949 0543309 856 I

490 Hamad and Itani
-a- Bottom cast

9.00 ~~~
+Top cast
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
O
m
6.00 - T
I I I
O 5 10 15 20
Yo Replacement silica fume
Fig. 4-Effect of percentage replacement silica fume on bond strength;

1 ksi = 6.9 MPa
Obb2949 0543110 578
CI
c
.-
c
O
.-u
L
*
m
.-
2
o,
.3
~
7
U
m L
I=
m m
O
œ
L
o al
u
n al
-
.-o f
U
o>
c
2
I-
O
o
F
b
u
O
O
u
2
m
$Ji
I
E
j,
x c
E?
M
O
.-c m n
w
U
E
O
n
ô - 2 r
œ o O
m Y al
c
E?
o
c
O
o
-
al
a¶
c
M
L c c
O al
.-
c
O
c
c
s
al
n
.-u L
rc
CI
L
O .c
n O
al
c cc
Y-
O w
O I
U m
c
3 .-m
U
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
SP 180-22
Bond of FRP Reinforcement in Concrete:

A State-of-the-Artin Preparation
by R. Tepfers
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Svnowis: The Comité Euro-International du Béton (CEB) has set up under its
Commission 2 "Material and Behavior Modelling" a Task Group TG 2/5 "Bond
Models" with terms of reference to write a State-of-the-Art report. Chapter 8
in the report will deal with bond of non-metallic reinforcement, FFW. The work
has been started using the Japanese State-of-the-Art knowledge presented in [ 5 ] .
Now research experience is added. The bond concept elaborated for steel rebars
is used to interpret the action of the FRP rods/bars. The different bond
influencing factors are discussed also for FRF' roddbars. Comparison is
performed with steel rebars. The bond of FRP rod/bar depend on more
parameters than bond of steel. Variables of interest are form of rod/bar section,
type of surface deformations and treatment, modulus of elasticity, Poissons ratio
etc. Therefore it is appropriate to use the known bond action of deformed steel
rebars in its different stages as a reference, when investigating the bond
performance of F W rodslbars.
Keywords: Bond; concrete; high-strength concrete; modulus of elasticity;

Poisson's ratio; reinforcing steel
493
494 Tepfers
= 0662947 0543112 340 =
BIOGRAPHY
Ralejs Tepfers is born in Rezekne, Latvia 1933. Since 1944 in Sweden. Civ Eng
(MSc) 1958, T e h lic degree 1966, Associate professor 1969, Tekn dr 1973 and
professor 1995 in Building Technology at Chalmers University of Technology,
Göteborg. Member of the ACI. Active in Nordic Concrete Federation, CEB and
RILEM. Convener of CEB, TG 2/5 "Bond Models".
INTRODUCTION
The Comité Euro-Intemational du Béton (CEB) has produced and issued in 1993
the CEB/FIP Model Code 1990, [i]. The bond clauses in the Model Code builds
on knowledge presented in the CEB Bulletins No 151, [2] and No 181, [3]. The
Model Code clauses do not treat FRP reinforcement, because at the information
inflow closure to the code work in 1986 the FRF' reinforcement was a very rare
type of reinforcement. After finishing the work with CEB/FIP Model Code 1990
the CEB reorganized its operative structure. The targets of the work are now
directed towards identifying in Model Code not covered techniques and items.
These should be treated by preparing State-of-the Art reports. The basic
understanding should be provided, and means elaborated for prediction of the
physical behavior of the materials concrete and reinforcement and their
interaction and hence the means for prediction of element behavior. The basic
thinking should be promoted about material and element behavior which should
lie behind design rules rather than their direct development. This does not imply
that results of the work will necessarily be unusable in design but that a
generality of approach will be aimed for which will usually require further
refinement or simplification for use in specific situations.
One important field of knowledge, where a considerable development has taken

place since the work with Model Code 1990, is that of bond between
reinforcement and concrete. The Riga conference on "Bond in Concrete" in
1992, [4], was a starting point for the new information flow. Therefore CEB has
set up under its Commission 2 "Material and Behavior Modelling" a Task Group
TG 2/5 "Bond Models" with terms of reference to write a State-of-the-Art
report, with the following chapters:
1
- Innovative developments in the light of bond philosophy
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
- Evaluation of bond models

- Bond of high strength concrete
- Bond under repeated loading
- Bond of corroded reinforcement
- Bond of epoxy coated bars
- Bond of prestressed reinforcement
- Bond of non-metallic reinforcement, Working Party 8.
0662945 0543L13 287
The bond of reinforcement in concrete has become today a very broad and
complex topic. It is no longer possible that a few researchers are able to have
all the necessary knowledge. Therefore the work in the Task Group 2/5 "Bond
Models" has been subdivided in Working Parties. Party 8 has the task to
evaluate the knowledge in the fields of bond of FRP reinforcement to concrete
and this paper treats the item of the Working Party 8. The work has been started
using the Japanese State-of-the-Art knowledge presented in [ 5 ] . All work within
the Task Group is done using E-mail. The fast electronic connections with
written texts transmitted direct into the computers of the group members enables
to work in bigger groups and makes it easy to engage especially young
researchers, who are interested and are willing to contribute.
The time schedule for the work of CEB, TG 2/5 "Bond Models" is to present
the State-of-the-Art Report at FIP Congress in May 1998, which will be held
together with CEB, just before the expected merger of the two international
concrete organizations.
The work with a State-of-the-Art report creates new thinking and gives rise to
new research. Therefore an intention of the Working Party 8 is to arrange an
one day Symposium on "Bond Between FRP Reinforcement and Concrete" in
connection to the regularly held "Tenth International Conference on Mechanics
of Composite Materials" (MCM) in Riga, Latvia, April 21-24, 1998.
FRP RESEARCH
The research within the field of FRP reinforcement in concrete is very

comprehensive and complex, and is for the time being characterized by small
separate investigations, where tests are done and reported at frequent
conferences. A systematic approach is lacking at least concerning the reported
research. Attempts are made to systematize research on an international base
using CO-FORCE International Consultants, (IRACC), as center for information
available on Internet with address: < http://www.iper.net/co-force/iracc.htm> ,
[6] and [7]. National research programs are also started. However the results
have not yet been reported in greater extent. Comprehensive investigations are
performed by companies and universities, but kept confidential because of
pending patent rights. In a few years it is expected that these investigations will
become open and a lot of new knowledge will be introduced, hopefully
systematically.
BOND OF STEEL AND FRP REINFORCEMENT IN CONCRETE
Bond between steel and also FRP reinforcement and concrete is governed by the
strength and modulus of elasticity of the constituent materials and the interface
properties. The modulus of elasticity and the Poisson ratio is about the same for
all types of steel. This is not the case for FRP reinforcement. The different
types of resin and fibers have different modulus of elasticity and Poisson ratio,
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0562949 0543114 113 I
496 Tepfers
and mixtures of them and also braiding of fibers give rise to even more
variations. The bond between FRP reinforcement' and concrete is for strong
concretes governed by the strength of resin lugs only. For rods made of braided
fibers the bond is determined of resin strength in combination with fibers. The
strength of concrete governs if its strength is lower than that of the surface
deformations of FRP rods. Combined failures exist also. The number of possible
combinations of bond failure of FRP reinforcement are greater than that of steel
reinforcement.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
For time being three main types of fiber are used for FRP rods. These are
aramid, glass and carbon fibers, which can be impregnated by different kinds
of resins usually epoxy and vinyl ester type of resin. The mentioned types of
fiber and resin are subdivided in fibers and resins with different strengths,
modulus and environmental resistance. The strength of fibers is also influenced
by the way of arranging the filaments - straight, winded or braided continuous
fibers. The strength of resin in the rod is not necessarily that of clean and
compact resin, because the impregnation method influences the resin structure,
adhesion to the fibers and matrix porosity. When estimating material properties
using mixing laws the possible disturbing factors must be kept in mind.
A moratorium has been suggested concerning the number of for time being
existing types of fiber and resin. Research should be performed only using them
and the necessary codes for design should be elaborated. There is a concern that
if new materials are all the time introduced, this will lead to a delay in practical
implementation of these materials because of difficulties to develop and adapt
code clauses to a very dynamic situation. However it is not possible and also not
especially clever to hinder people to invent and develop materials and
techniques.
The great variety of materials and production methods result in combinations,

which make it only possible to treat the bond problem on a general base in a
State-of-the-Art review. For practical application each special type of barírod
must be tested individually and the actual bond stress-slip relation established.
Then the relation enables to estimate the bond behavior in concrete members.
Information completing tests on representative FRP reinforced concrete members
should also be performed.
LOCAL BOND-SLIP LAW
The bond in its different stages up to the ultimate failure load (ascending
branch) and further under deformation controlled conditions until the load taking
capacity drops to zero (descending branch), can be studied with pull-out tests
with short bond length with central and also eccentric placement of barhod. The
short bond length represents an increment of the anchorage length. The local
bond slip law determined in pull-out tests can be subdivided in different stages
and has different failure possibilities, FIGURE 1, [8].
Obb29Yï 0543115 05T
Stage I: For small values of bond stress the bond is assured by chemical
adhesion and no slip occurs.
Stage II: For larger bond stress values the chemical adhesion breaks down, the
deformations of the rod surface now transfer bond forces by mechanical friction
provided by lugs and surface deformations and induce bearing stresses in the
concrete. Transverse micro cracks originate at the tips of lugs allowing barhod
to slip, but the wedging action of the lugs remains limited. FRP reinforcement
is softer than steel and will not cause so easy and so many micro cracks in
concrete at lug tips. The bond performance of FRP reinforcement in this stage
will in some case probably be better then that of steel rebars.
Stage III: At this stage the bond induced bearing forces subtend an angle, a,
with the bar axis, FIGURE 2, [9]. These bond forces or stresses can be resolved
into radial and tangential components. The radial forces are counterbalanced by
a concrete ring under tension. When tensile strength of the concrete is reached
the concrete ring will crack where it is thinnest. In case of pull-out specimen
with eccentric placement of bar the cover along the bar will crack. The slip of
the barlrod will considerably increase when this crack appears, [9], [lo], [ l i ]
and [12].
The load at appearance of the cover crack along the bar can be estimated
assuming a concrete ring around the bar and calculating the failure load in partly
cracked elastic stage (lower bound solution) and in plastic stage (upper bound
solution), FIGURE 3, [8]. There exist different more precise approaches in
between the boundaries for estimation of the load at cover cracking.
Stage IV: Four types of failure can be distinguished at this stage:
1) Concrete cover splitting failure at or above the load which causes the cover
crack along the bar: The cracked concrete cover along the bar will split away
by the pressure exerted by the bond forces from the anchored reinforcing FRP
barhod. An ultimate splitting failure pattern will be formed. The failure load
can be estimated by balancing the radial pressure from the bar against the tensile
strength of concrete effective on the areas of the minimum possible ultimate
failure pattern. When the cover is split away the load will drop to zero. The
descending branch of the stress-slip curve shows a sudden loss of load.
2) Shear failure in concrete along the lugs of the bar: If the splitting resistance
of the surrounding concrete is high enough, then bond failure will occur as shear
failure along the perimeter of the bar lugs. This bond strength is the maximum
possible for rebars, and can be studied on pull-out specimens of short bond
length. 7-wire strand of steel might show even higher bond shear resistance for
large slip values. This is due to compaction of crushed concrete by the winded
wires so the shear failure zone is moved outward and thereby the resistance
increases. This is not the case for FRP strands because of soft surface and
considerable Poisson contraction in comparison to that of steel strand. The
descending branch shows a slow loss of load combined with increasing slip.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
0662949 054333b T4b
498 Tepfers
3) Shear failure due to shearing off the deformations of the FRP rod: if the
splitting resistance of the surrounding concrete is high enough, then bond failure
will occur as shear failure through the rod deformations and will be determined
by the shear resistance of the resin in eventual combination with fiber
reinforcement in the deformations. The concrete has to have higher shear
resistance than the shear resistance of the lugs. For the descending branch the
coefficient of sliding friction and its change in the shear failed surface will
determine the shear resistance and the slope. In cases where the shear failure is
a combined concrete - resin bar surface deformation failure the sliding frictional
shear resistance can be rather stable. Normally the frictional resistance falls with
increasing slip.
4) Shear failure in the surface part of the bar: Bars with bonded sand to the
surface layer, which gives very good bond to concrete might be pealed of by
failure in the resin layer underneath the sand layer. It must be observed that the
bond stress, due to smaller perimeter, is higher in this layer than in the concrete
- sand layer interface. The descending branch may show a fast drop in load.
In situations with special cover geometries 1) and with thick concrete covers and
good confinement 2), 3) and 4), the Stage III wtih cover crack along the bar
may not appear before the Stage IV.
Analyzing the development length of prestressing force a barírod push in

situation should be studied.
Bars which have transverse diameter reduction due to large Poisson contraction
and with low transverse compressive strength can by the pulling force be
squeezed against the surrounding over strong concrete and pulled out.
In FIGURE 4, [ 131, the bond stress-slip relations in pull-out tests are compared
for Arapree 20x2 111111 strip, 7-wire steel strand and CFCC strand, Ks600 steel
rebar (standard Swedish) and FRP C-bar. It can be concluded that Arapree and
CFCC have very good bond for lower loads. However the Poisson contraction
causes their relations to fall below those for steel rebar and strand. C-bar has
the same bond-slip relationship as Ks600 rebar. However the related rib area of
C-bar is only 0.08 compared to that of Ks600 rebar, O . 13 determined according
to DIN 488.The good bond performance of C-bar is probably due to softer
surface than steel and thereby reduced micro cracking. The concrete
compressive strength was about 45 MPa.
BONDALONGANCHORAGELENGTH
The pull-out test represents an increment of the anchorage length and shows
what happens when the tensile load in the bar is increased up to the ultimate.
The bond stress-slip relation is determined in this test in all stages for the
ascending and the descending branch. Depending on the stress in the anchored
bar, the concrete strength and the member geometry, adequate bond stress-slip
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
stages will exist along the anchorage length. These and their distribution along
the anchorage length will govern the different failure modes and the anchorage
capacity, FIGURE 5, 181.
Where the concrete cover is cracked along the bar the slip increases and results
in plastification of the bond stress, length 1, in FIGURE 5 . In certain cases this
cracked length can be extended along the whole anchorage length. Along the
cover cracked part of anchorage length the ultimate crack pattern resistance
governs the bond capacity.
For the determination of stress distribution along the anchorage length part 12+3
in FIGURE 5 , without cover cracks a theory based on displacement along the
bonded surface of the reinforcing bar can be used. The theory is also
appropriate for determination of maximum bond stress and anchorage capacity
of prepregs glued on concrete surface.
The bond and the bar tensile stress distributions along the anchorage length
without cover cracks can be calculated using a theory based on the modulus of
displacement, K (MN/m3). The approach was first mentioned by Bleich [14].
Bleich used the theory to determine the individual rivet loads in long riveted
joints. In the calculation the rivets are replaced by a continuous medium. The
shear deformations in the medium are characterized by the modulus of
displacement. The tangent or secant modulus for the relationship between bond
stress and slip in a pull-out test with short bond length and Confinement
constitutes the modulus of displacement. The modulus of displacement is a
linear function of the compressive strength of the concrete and depends on the
roughness of the bar surface. The determination of stress distribution is rather
insensitive to changes in the modulus of displacement. However the stress
calculations may give differing stress peaks close to bar ends. The calculated
high stress gradients appear only along very short bar length at the bar ends.
The intense stress gradients will be reduced by the plastification of concrete.
FRP rodslbars have lower modulus of elasticity than steel and therefore lower
modulus of displacement for the same concrete resulting in reduced bond stress
variation compared to steel rebars. The theory was applied to reinforced
concrete by, Granholm [15], Losberg [16], FIGURE 6 and Tepfers [8].
ULTIMATE FAILURE CRACK PATTERN AND BOND RESISTANCE
In cases where concrete gives enough confinement the tensioned anchored bar
will be at failure pulled out and the shear resistance in the bar to concrete
interface will determine the failure load. If the shear resistance is the lowest
between the ribs and the bar core, as it might be for some types of FRP
rodshars in strong concrete, this shear resistance will govern. After the
maximum resistance is reached a lower resistance based on the sliding friction
in the shear failure interface is possible.
In cases where the confinement is limited by the concrete cover cracking

resistance against radial pressure from the anchored bar the cover will crack
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
500 Tepfers
-
along the bar. The cover cracked concrete parts around the bar will now hold
0662949 0543118 8bï
and anchor the bar. The final splitting failure of concrete cover will follow a
minimum ultimate failure crack pattern. The tensile strength of the concrete
applied on the crack pattern will determine the possible resistance for radial
bond forces. In some special cases an ultimate failure crack pattern may develop
directly without giving a prewarning by showing cover crack along the bar.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Different modes of failure exist.
Mode A) The resistance of ultimate minimum crack pattern is higher than that
which is obtained at cracking of concrete cover along the bar. The cover will
be cracked along the whole anchorage length and the increased slip will cause
plastification of bond forces and give a prewarning of failure by showing
extensive cracking along the anchorage length.
Mode B) The resistance of the ultimate minimum crack pattern including cracks
along the bar is less the that which is obtained at cracking of concrete cover
along the bar. These failures will allow plastification of bond stress only at the
loaded end of the anchorage length where the cover is cracked. The failure load
will be determined by optimizing and adding the plastificated ultimate bond
resistance from the cover cracked part with the unevenly distributed bond
resistance from the not cover cracked part of the anchorage length. This type
of failure will give a prewarning by showing extensive cracking at the most
loaded end of the anchorage length. The failure will start by splitting off the
concrete cover at bar ends following a minimum resistance ultimate failure crack
pattern. Then this failure will spread along the anchorage length.
Mode C) The resistance of the ultimate minimum crack pattern without cover
cracks along the bar is less than that which is obtained at cracking of concrete
cover along the bar. These failures will have zipper character and start as soon
as the ultimate minimum failure crack pattern resistance is reached The failures
will be brittle and will not give a prewarning of failure by showing extensive
cracking. This type of failure can be expected for FRP prepregs glued on
concrete surface.
The failure modes are shown in FIGURE 7, [9], for an overlap splice.
INFLUENCING FACTORS ON BOND OF FRP REINFORCEMENT
Form of the transverse section of the rodlbar - round, flat rectangular or

specially shaped -has importance for bond and anchorage of the rodlbar. Flat
bars might exert increased splitting forces from the flat surfaces on concrete in
anchorage zones in comparison to circular bar sections. When splitting occurs
the slip increases and ultimate bond resistance decreases. Shapes with flat
surfaces can be provided with shear connectors of FRP strings for improved
anchorage in the concrete [ 171.
The size and type of lugs or surface deformations constituting the roughness of
0662749 0543339 7T5 W
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
FRP rod/bar are important for bond. Small dense surface deformations give a
very intense bond transfer at low loads. However, for higher loads, when the
rodlbar becomes thinner due to Poisson effect or a splitting crack develops in
concrete cover along the reinforcing rod/bar, these small deformations may lose
their grip in concrete very suddenly with bond failure as result.
It has importance for bond resistance if the barhod lugs or surface

deformations are made up of only resin, of resin mixed with special
strengthening fiber reinforcement in the lugs or of resin containing the tensile
load transferring longitudinal continuous fibers in the surface deformations. This
latter is achieved by braiding the fibers or by winding the rod by a separate
fiber filament. When the longitudinal fibers are brought out of the direction of
the bar axis by braiding or by press-deforming the longitudinal fibers (by
winding a fiber bundle around) the axial modulus of elasticity of the rod
decreases. The type of bond failure will be governed by rod/bar surface
deformation type.
If the shear resistance of FRP rod lugs determine the ultimate load an increase
of the thickness of the lugs should raise the shear resistance. The distances
between the lugs are filled up with concrete which takes the lug pressure. If
these distances becomes too small the failure will change to become a concrete
shear (local compression) failure. It should be observed that the bond shear
stresses at the rot of the lugs are higher than at the top of the lugs were the
concrete bond shear failure takes place. This is because of different perimeters
at lug rot and top. There is an optimum situation for maximum bond, which
should be identified.
The Poisson contraction of the rod/bar, when tensioned, has influence on the
bond to concrete. A pull-out or push in situation (at development of prestressing
force) has importance on the bond. Push in situations takes advantage of the
Hoyer effect. The Hoyer effect may be more pronounced for FRP rods/bars but
the efficiency less because of softer resin than steel.
It is known that transverse pressure improves the anchorage capacity of steel

rebars. There are indications that the same yields for FRP barslrods. Probably
transverse pressure contributes less to the anchorage capacity for FRP barhod
because it is not as hard as steel. It has been observed that transverse expansion
of FRP barhod due to temperature increase is much more than that of concrete.
However this expansion does not seem to cause splitting and cracking of
concrete because of softness of the bar/rod. This inàicate that the transverse
pressure at anchorage is less efficient for FRP barhod than for steel rebar.
Because the resin of the rod/bar takes the transverse compression, long term and
temperature effects should be considered, when the favorable bond effect of this
pressure is used.
The modulus of elasticity of the bar has influence on the ultimate bond load.
It has been observed that certain types of FRP bar with lugs give higher bond
resistance than steel rebars does. The cause for this is probably the fact that bars
œ 0662949 0543320 417 œ
502 Tepfers
with modulus of elasticity alike that of concrete at anchorage by lugs do not

create as much stress concentrations in concrete as steel rebar does, because
steel is much harder than concrete. This means that anchorage of steel
reinforcement by lugs give local stress concentrations in concrete from which
failure develops.
Bond splitting effect on surrounding concrete induced by the bond forces from
anchored rod/bar. The FRP bardrods may have less tendency to split the
surrounding concrete also due to surface softness and reduced micro cracking
of concrete.
The position of rodlbar during casting in concrete, top or bottom bar position,
and horizontal, inclined or vertical bar position, has influence on the bond
resistance between concrete and rod/bar . This influence should be determined
for each type of FRP barhod.
It has been observed that wedging effect increases the anchorage capacity for
seven wire steel strands above that of deformed rebar with increasing slip. This
wedging effect does not appear for FRP strands due to considerably lower
surface hardness than steel has and also a considerable Poisson effect.
Required concrete covers and distances between roddbars for necessary

anchorage. Needed cover thicknesses for environmental and fire protection of
the FRP rodsíbars should be determined. Indications exist that covers with
thickness > 44 is satisfactory for anchorage of FRP prestressed strands.
Differences in thermal elongation between the FRP unit and concrete,

especially in the transverse direction to the axis of the rod/bar, might influence
the bond. Too high transverse thermal elongation of the rod/bar might give rise
to splitting cracks in concrete surrounding the rod/bar. However there are
indications that the FRP rods are enough soft not to cause splitting off the
concrete cover, when trying to expand. The thermal expansion of bar diameter
may also be counteracted by Poisson effect at tensioning the rod. If surface
deformations or lugs are of normal size the difference in thermal elongation
between concrete and FRP rod should not cause that the bar looses its grip at
decrease of temperature.
Environmental influence on bond between FRP rod/bar and concrete should

be taken in consideration. Adsorption of water leading to strength and modulus
deterioration as well as expansion of rod might influence the bond. The
environment, normal in concrete with pH-values 12-13, might degrade the rod
surface or fibers especially at elevated temperature. Sometimes less resistant
fibers are protected by resistant thick resin layer. Here it is important to know
if the bond transfer is through a thicker resin layer or there can be counted for
a more direct support from stiffer fibers against concrete through a thin layer
of resin. It is important than to know if resin or fibers or both can be
environmentally degraded, how fast and in which extent. FRP units are
composites made from two constituent materials - fibers and resin. The qualities
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
W 0bb2949 0543323 353 œN
of the composite is what will be used, but to understand them better under
influence of environment, the qualities of the two constituent materials should
also be studied. An appropriate mixing law should be used to state the material
qualities of the composite. However the real composite súucture with defects
should be considered.
Anchorage and splice lengths for non-prestressed rods/bars have to be

determined. FRP bars are made today to be used also as non-prestressed
reinforcement. These bars have limited lengths and sometimes have to be
spliced. The necessary splice lengths must be determined taking into account the
tendency of the bars to split the concrete. Influence of the concrete strength on
the anchorage and splice lengths and concrete cover dimensions should be
determined.
Necessary development lengths and anchorage lengths for bonded prestressing

strands taking into consideration the pull-out and push in stress-strain
relationships and strengths. Long term redistribution of bond stresses and the
influence of elevated temperature are important to know. The consequences on
the concrete anchorage zone concerning bursting, spalling and splitting should
be known.
There is also a psychological effect. By recognizing FRP reinforcement units

alike steel reinforcement, engineers might feel confidence in some FRP units
and in others not. However this fact does not automatically imply that these
units are bad. FRP reinforcement is not steel. It is a different material than steel
and has to be treated as such for a successful technical use.
CONCLUSIONS
The work of CEB, TG 2/5 "Bond Models" with the State-of-the-Art-Report is

going on. The information inflow closure will be after the Third International
Symposium on Non-Metallic (FW)Reinforcement for Concrete Structures" in
Sapporo, 14-16 October 1997. The presentation of the report is expected to be
at the FIP Congress at a CEB session in Amsterdam May 1998. On the existing
knowledge a basic thinking will be elaborated which should be behind design
rules for bond of FRP reinforcement in concrete. For time being the available
information is not sorted but the following observations can be mentioned.
Non-Prestressed Reinforcement
Most FRP reinforcing bardrods have modulus of elasticity only 30-70%of that
of steel. If the stress in these tensile reinforcement FRP bardrods would be used
to the same level as in steel reinforcement, the deflections of the concrete
members would be unacceptable. This concludes that for the FRP bars/rods, in
spite of having ultimate tensile strength much in excess of that of steel bars,
oniy 30-70% of the steel stress in serviceability limit stage ( S U ) can be used
due to deformation requirements. Because of the low stress level the anchorage
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
ûbb2947 0542753 4 b l
504 Tepfers
of these bars will not be a problem. However there are cases where the
deformations are not dimensioning and ultimate limit state yields. In such cases
the bars might be used to higher tensile stresses. Then the necessary anchorage
should be provided.
Prestressed Reinforcement
FRP roddstrands are favorable to use as prestressing reinforcement due to low

modulus of elasticity. This fact limits the loss of prestress due to elastic strain,
creep and shrinkage of concrete. The FRP rodsístrands should have limited
creep of its own. For these bonded rodslstrands the development length should
be determined taking into account a " push-in" situation and the anchorage length
taking into account a pull-out situation. The development length often is about
400 mm long for a strand corresponding to 7 wire steel strand. The force
transfer should not be too intense because this might cause problems to cope
with splitting, bursting and spalling forces in the anchorage zone of the concrete
member.
REFERENCES
[i] CEB-FIP Model Code 1990. Bulletin d'information No 213/214.Thomas

Telford, London 1993. p.437.
[2] Comité Euro-Intemational du Béton (CEB). Bulletin d'information No

151. "Bond action and Bond Behavior of Reinforcement". State-of-the-
Art Report. CEB, Case Postale 88, CH-1015 Lausanne, April 1982. p.
153.
[3] Comité Euro-International du Béton (CEB). Bulletin d'information No

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
181. "Anchorage Zones of Prestressed Concrete Members". State-of-the-

Art Report. CEB, Case Postale 88, CH-1015 Lausanne, April 1987. p.
153.
[4] Comité Euro-Intemational du Béton (CEB), Riga Technical University

(RTU), Intemational Conference "Bond in Concrete - from research to
practice" Proceedings Volume 1, 2 and 3. Riga, Latvia October 15-17,
1992.
[5] Machida Atshiko (Editor), Second Research Committee on CFRM, Japan

Society of Civil Engineers. State-of the Art Report on Continuous Fiber
Reinforcing Materials. Concrete Engineering Series 3. Tokyo, October
1993. pp. 15-26.
[6] International Research on Advanced Composites in Construction (IRACC

1996). Bologna Workshop, June 1996. Final Report. CO-FORCE
International Consultants (IRACC). < http://www.iper.net/co-
Obb2949 0542754 3Tô
force/iracc.hûn> . Bologna, August 1996. p. 35.
[7] International Research on Advanced Composites in Construction(IRACC

1996).Current & Pending Research Project & Application of FRP in the
Construction Industry. CO-FORCE International Consultants (IRACC).
> . Bologna, November 1996.
<http://www.iper.net/co-force/iracc.htm
[8] CEB, TG 2/5 "Bond Models", Party 1, Convener Gambarova P.: Bond
Models - Chapter 1. Milano, January 10, 1997. p. 40.
[9] Tepfers, R.: A Theory of Bond Applied to Overlapped Tensile

Reinforcement Splices for Deformed Bars. Pub1 73:2. Division of
Concrete Structures, Chalmers University of Technology, Göteborg,
May 1973, p. 328.
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
[lo] Tepfers, R. : Lapped tensile reinforcement splices. American Society of

Civil Engineers, Journal of the Structural Division, January 1982, New
York. pp. 283-301.
[Il] Tepfers, R.: Cracking of concrete cover along anchored deformed

reinforcing bars. Magazine of Concrete Research Vol. 31, No 106,
March 1979, pp. 3-12.
[12] Tepfers, R.: Bond stress along lapped reinforcing bars. Magazine of
Concrete Research, Vol. 32, No 112. September 1980. pp. 135-142.
[131 Tepfers R.,Karlsson M.: Pull-out and tensile reinforcement splice tests
using FRP C-BARS. Chalmers Uiversity of Technology, Division of
Building Technology, Work No 13, Publication 97:2. Göteborg, June
1997.
[14] Bleich Fr. : Theorie und Berechnung der Eisernen Brücken. Verlag Julius
Springer. Berlin 1924. pp 306-314.
[ 151 Granholm Hj.: Cracking in reinforced concrete. Chalmers University of

Technology, Division of Building Technology. Göteborg 1958. p. 28.
[161 Losberg A.: Transmission of force and stress distribution at anchorage

and curtailment of reinforcement. Chalmers University of Technology,
Division of Building Technology, Work No 608. Göteborg 1963. p. 51.
[171 Kettil P.: Composite beams of fibre reinforced plastic profile and
concrete. Chalmers University of Technology, Division of Building
Technology, Work No 8. Graduation thesis: E-95:1. Göteborg,
September 1995. p. 80.
H Obb2949 0542755 234
506 Tepfers
Fig. 1-Local bond-slip law for reinforcement
Fig. 2-Schematic representation of how the radial components of the bond

forces are balanced by tensile stress rings in concrete in anchorage zone
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 3-Plots of bar-concrete pressure and bond strength versus concrete cover,
according to dofferent models, and fitting of test results at cover cracking, bar
diameter 0 = 16 mm.
W Obb2949 054275b I170
25
S
fa
a
5.
n
s”
.
I
U
O
m ARAPREE
O 0.1 0.2 0.3 0.4
Slip (mm]
Fig. 4-Comparison of pull-out test results with different types of barslrods
Fig. 5-Bond behavior in anchorage; complete split in I,, partial splitting in ,I

frictional bond or chemical adhesion in ,I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
508 Tepfers
= 0662949 0542757 007 9
- X
Tensile stress it1 rrbar .
Fig. 6-Distribution of tensile stress in anchored rebar and bond stress along
anchorage length
O i s t r i b u l i o n o f bond s l r e s s e s
Mode o f TU
failure A cover crack
Mode o f
f a i l u r e Ei
I &cover crack
I-LJ 1
I
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Fig. 7-Overlapped tensile reinforcement. Distribution of bond stresses of Failure

Modes A, B, C; fbu=smallest ultimate failure crack pattern bond stress; fbc=bond
stress which initiates cover crack
Obb2949 0542758 T43
SI (Metric) Tables 509
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
*
CONVERSION FACTORS- -INCH-POUND TO SI (METRIC)
To couveri from to multiply by
inch millimeter (nun) 25.4Et

foot meter (m) 0.3048E
yard meter (m) 0.9144E
mile (statute) kilometer (km) 1.609
Area
square inch square millimeter (nun’) 645.1

square foot square meter (II?) 0.0929
square yard square meter <m2> 0.8361
Volume (capacity)
ounce milliliters (Id) 29.57

galion cubic meter (III’)$ 0.003785
cubic inch cubic millimeter (nun3) 16390
cubic fmt cubic meier (m3) 0.02832
cubic yard cubic meter (m’)S 0.71546
Force
kilogram-force newton (N) 9.807

kipforce kilo newton (kN) 4.448
pod-force newton (N) 4.448
Pressure or stress (force per area)

kilogramforce/square meter pascal (Pa) 9.807
kip-forceisquare inch (ksi) megapascal (MPa) 6.895
newtodsquare meter (Nlm’) pascal (Pa) 1.WE
pound-forcdsquare fmt pascal (Pa) 47.88
pound-forcekquare inch (psi) kilopasai ( P a ) 6.895
Bending moment or torque
iiich-poünd-fûrm newton-meter (Nem) O . 1130

foot-poun&force newton-meter (Nam) 1.356
meter-kilogramforce newionmeter (Nem) 9.807
- ~~
O662949 0542759 98T

510 SI (Metric)Tables
To couveri from to multiply by
ouncemass (avoirdupois) gram (8) 28.34

pound-mass (avoirdupois) kilogram (kg) 0.4536
ton (metric) megagram (mg) I.WE
ton (short, 2000 Ibm) kilogram (kg) m.2
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
M a s per volume
pound-masslcubic foot kilogramkubic meter (kglm’) 16.02

pound-masslcubic yard kilogradcubic meter (kglm’) 0.5933
pound-mass/gallon kilogradcubic meter (kg/m3) 119.8
Temperatures
d e g r m Fahrenheit (F) degrees Celsius (C) = (tF - 32)ll.S

degrees Celsius (C) degrees Fahrenheit (F) tE = 1.8tC +
32
This selected list g i v a practical conversion factors of units found in wncreie technology. T k r e f e m e sources for information on SI
uni0 and more exact wnversion factors are ASTM E 380 ami E 621. Symbols of metric units are given in parenIlmis.
t E Indicates that the factor given is exact.
Oiu liter (cubic decimeter) equals O.ûû1 m3 or 1ûûû cm’.
P T k quatiom conven one t~nperarurereading to a m k r and inclwle the necessary scale correcliom To conven a differem in
tmperaNre from Fahrtnkitdegms to Celsius degrees, divide by 1.8 only, ¡.e.. a cbange from 70 to 88 F represents a change of 18 F
or 1811.8 = 10 C deg.
H 0662949 0542760 bTL m
INDEX Index 511
Aarup, B., 459 Design, 1

Analysis, 431 Detailing, 1
Anchorage, 1, 171, 215,347,367,405 Development, 1,261
Applications, 459 Development length, 473
Azizinamini, A., 243 DeVries, R.A. 367
Dilation; 81
Balázs, G.L., 125 DiMarco, R., 105
Bashandy, T., 367
Bigaj, A.J., 195 Eligehausen, R., 45
Bond, 1,45, 215,405,459,493 Engström, B., 21 5
Bond (concrete to reinforcement), 23, 125, Epoxy resins, 391,405,431
243,261,391,431,473
Bonding, 195,319,347,367 Fatigue (materials), 125, 391
Bond models, 1,81, 145 Fatigue tests, 125
Bond slip, 319 Fiber concrete, 459
Bond splitting, 145, 195 Finite element method, 1
Bond strength, 23, 171, 299 Fire resistance, 459
Bond stress, 105, 171, 195 Fracture, 431
Bridge anchorages, 319 Friction, 431
Bridge specifications, 261 French, C.W.. 261
Bridge structure, 319
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Building codes, 243, 261 Gambarova, P.G., 23
Giuriani, E., 171
Cairns, J., 405 Grundhoffer, T., 261
Caps (supports), 319
Cleary, D.B., 391 Hamad, B.S., 473
Coatings, 391 Hasan, H.O., 391
Concrete, 45,493 Hayashi, T., 431
Concrete durability, 195 High performance, 459,473
Concrete width, 391 High-strength, 459
Confined concrete, 195 High-strengthconcrete, 215, 243,473,493
Confinement, 215 Huang, 2.. 215
Cover, 215
Cox, J.V., 81 Ichinose, T., 431
Cracking (fracturing), 171,299,431 Ingham, J. M., 319
Cracks, 45 Interface. 81
Crack width and spacing, 105, 391 Itani, M.S., 473
Creazza, G., 105
Cyclic creep, 125 Jensen, B.C., 459
Cyclic loads, 125, 319, 391 Jirsa, J.O., 367,405
Joints, 459
Darwin, D., 299
Deflections, 391,405 Lap connections, 243,261,391,405
Deformation, 125, 299 Leon, R., 261
Deformed reinforcement, 243, 261,347, 391 Lin, W., 431
denllijl, J.A., 145, 195
-
D 0662749 0542761 538

512 Index
Magnusson, J., 215 Strength, 23,431

Masonry, 347 Stress, 105
Mayer, U., 45 Structural engineering, 243, 261, 391
McCabe, S.L., 1,405
Mechanics, 81 Tensile strength, 105,299
Mendis, P.A., 261 Tension stiffening, 45,405
Modulus, 105,493 Tepfers, A., 493
Test, 23
Ozbolt. J., 45 Tholen, M.L., 299
Transfer length, 145
Pantazopoulou, S.J., 1 Transverse reinforcement, 367
Phenomenological, 8 1
Plasticity, 1, 81 Validation, 81
Plizzari, G.A., 171 Variability, 261
Poisson's ratio, 493 Vintzileou, E., 3473
Priestley, M.J.N., 319
Psilla, N., 347 Walraven, J.C., 195
Pull-out, 81, 215
Yielding, 45
Ramirez, J.A., 391 Yield strength, 299
Reinforced concrete, 81,459,473
Reinforced concrete structures, 23, 367 Zuo, J., 299
Reinforcing bars, 125
Reinforcement, 1,45
Reinforcing steel, 243, 261, 299,347,493
Relative rib area, 261
Reliability, 261
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---
Ribbed bar, 215

Rib pattern, 45
Rosati, G.P., 23
Rotation capacity, 45
Russo, S., 105
Schumm, C.E., 23
, Seible, F., 319
Seismic loading, 319
Silica fume, 459,473
Siviero E., 105
Slippage, 105, 125,215, 319, 367
Splicing, 243, 261, 391
Splitting (cracks). 23, 81,215
Sritharan, S.,319
Steel fiber, 459
Stirrups, 171, 431
Strand (prestressing), 145
strand (spacing), 145
--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---

Aci Bond & Development of Reinforcement

Uploaded by

Copyright:

Available Formats

Aci Bond & Development of Reinforcement

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

Aci Bond & Development of Reinforcement

Uploaded by

Copyright:

Available Formats

What is the main topic discussed in the document?

What is the main topic discussed in the document?

What is the purpose of organizing the symposium discussed?

What is the purpose of organizing the symposium discussed?

--```,``,,`,````,``,,,`,````,,-`-`,,`,,`,`,,`---

Obb2949 05425bb 258

Bond and Development

DISCUSSION of individual papers in this symposium may be submitted in

AMERICAN CONCRETE INSTITUTE

Printed in the United States of America

Editorial production: Jane Carroll

Library of Congress catalog card number: 98-86927

The sessions held in Seattle, Washington, were organized into a symposium

Roberto Leon, Chairman, AC1 Committee 408

A Tribute to Dr. Peter Gergely

Peter Gergely was a professor of structural engineering at Corneil University

Peter made major contributions to a wide variety of structural engineering

In recognition of his outstanding contributions, Dr. Gergely received many

James O. Jirsa, Chairman

EVALUATION OF BOND PERFORMANCE IN REINFORCED

BOND AND SPLïïïING: A VEXING QUESTION

CONTRIBUTION OF CONCRETE BETWEEN CRACKS AT INELASTIC

STEEL STRAINS AND CONCLUSIONS FOR THE OPTIMIZATION

A DILATATIONAL-INTERFACE MODEL FOR BOND

TENSION STIFFENING AND CRACKING BEHAVIOR

BOND UNDER REPEATED LOADING

BOND MODELLING OF PRESTRESSING STRAND

CONFINEMENT ROLE IN ANCHORAGE CAPACITY

BOND OF RIBBED BARS MODELED THROUGH CONCRETE

PULL-OUT BOND BEHAVIOR OF RIBBED BARS IN NORMAL

PREVENTING BRITTLE FAILURE OF TENSION SPLICES

BOND OF EPOXY-COATED REINFORCEMENT

OVERVIEW OF RESEARCH TO IMPROVE THE DEVELOPMENT

BOND SLIP OF BRIDGE COLUMN REINFORCEMENT ANCHORED

PULLOUT OF HORIZONTAL REINFORCEMENT EBEDDED

EFFECTS OF TRANSVERSE REINFORCEMENT AND BONDED LENGTH

PERFORMANCE OF CONCRETE BRIDGE DECKS AND SLABS

DESIGN RECOMMENDATIONS FOR EPOXY-COATED

FUNDAMENTAL ANALYSIS OF RC LAP SPLICES

BOND PROPERTIES OF HIGH-STRENGTH

BOND STUDIES OF REINFORCING BARS

BOND OF FRP REINFORCEMENT IN CONCRETE:

Evaluation of Bond Performance in Reinforced

by S. L. McCabe and S. J. Pantazopoulou

Keywords: Anchorage; bond; bond models; design; detailing; development; finite

sponse of a given anchorage as reference for design purposes. Complete bond-slip

this parameter on the anchorage length is continually changing. An added layer of

For standard configuration, this equation reduces to

the reinforcing steels considered in the two cases.

fbd = qlq2q3fl'd (12)

In the above, factors aiadjust the length based on:

For straight bars = 0.7 2

i = 4: Accounts for confinement provided by traverse steel:

ANALYTICAL BOND MODELS

A traditional approach taken in modeling the behavior is to use experimental

In the above, E, is the elastic modulus of concrete, H is the heaviside step

and a non-associated flow rule:

[qPIT = <h>[sgn(z), 91 (25)

individual contributions in detail (in a narrow layer of interface elements which is