Sie sind auf Seite 1von 269

BOND STRESS AND SLIP MODELING

IN NONLINEAR FINITE ELEMENT ANALYSIS


OF REINFORCED CONCRETE STRUCTURES

Youai Gan

A Thesis submitted in conformity with the requirements


for the Degree of Master of Applied Science
Graduate Department of Civil Engineering
University o f Toronto

O Copyright by Youai Gan 2000


1+1 National Library
of Canada
Bibliothque nationale
du Canada
Acquisitions and Acquisitions et
Bibliographie Services services bibliographiques
395 Wellington Street 395, rue Wellington
W w a O N KIAON4 OttawaON KlAOFJQ
Canada Canada

The author has granted a non- L'auteur a accord une licence non
exclusive licence aiIowing the exclusive permettant la
National Library of Canada to Bibliothque nationale du Canada de
reproduce, loan, dismbute or sel1 reproduire, prter, distribuer ou
copies of this thesis in microform, vendre des copies de cette thse sous
paper or electronic formats. la forme de microfiche/fiim, de
reproduction sur papier ou sur format
lectronique.

The author retains ownership of the L'auteur conserve la proprit du


copyright in this thesis. Neither the droit d'auteur qui protge cette thse.
thesis nor substantial extracts fiom it Ni la thse ni des extraits substantiels
may be printed or otherwise de celle-ci ne doivent tre imprims
reproduced without the author's ou autrement reproduits sans son
permission. autorisation.
ABSTRACT

The stress transfer behavior between reinforcing steel and surrounding


concrete through bond and slip plays an important role in the response of
reinforced concrete structures, especially in their hysteretic response. This
thesis studies the bond-slip relationship under both rnonotonic loading and
cyclic loading. Two types of bond elements, contact elements and linkage
elements, are developed and bond stiffness matrices are derived. Four
bond-slip models are presented based on the findings of previous
experimental studies. The degradation of bond resistance is included in the
bond-slip models. The two bond elements and four bond-slip models are
added into the nonlinear finite element program TRIX99. The finite
element models are then used to study the behavior of reinforced concrete
walls subjected to cyclic lateral load. The accuracy of the models are
assessed by cornparison of the finite element numerical response with
experimental data fiom four reinforced concrete shear walls tested under
cyclic loading.
The NLFEA prograrn with the nonlinear material models for concrete,
reinforcing bar and bond-slip is capable of reproducing the important
features of the measured Iiysteretic response of reinforced concrete walls
with a variety of cyclic loading histones and configurations of walls. The
prograrn is able to successfully predict the load-deflection values of the
reinforced concrete structures under cyclic loading. The proposed finite
element method provides an efficient method for evaluating the rnonotonic
and cyclic response of the reinforced concrete structures.
ACKNOWLEDGMENT

I would like to thank my supervisor, Dr. Frank J. Vecchio, Professor


of Department o f Civil Engineering at the University of Toronto. He
provided the subject for my thesis, the knowledge, the wisdom, the
guidance, the patience and support throughout the course of my research
which allowed me to complete al1 my tasks.
Table of Contents

CHAPTER 1 Introduction
Background
Scope
Plan of Study

CHAPTER 2 Review of previous studies


Introduction
Mechanics of Bond Slip
Factors Mecting Bond Behavior
2.3.1 Monotonic Loading
2.3-2 Cyclic Loading
2.3.3 Loads
2.3 -4 Failure Modes
2.4 The Experimental Study of Bond Stress
and Slip
2.4.1 Bond Stress and Slip Under Monotonic
Loading
2.4.2 Bond Stress and Slip Under Cyclic
Loading
2 -5 The Study of Bond Stress and Slip by
Theoretical Anal ysis Methods
2.5.1 Constitutive Equation of Bond Stress
and Slip
2.5.2 The Solution of Constitutive Equation of
Bond Stress and Slip
Table of Contents

2.6 Code Development


2.7 The Analysis of Bond Stress and Slip
by Finite Element Method

CHAPTER 3 The Modeling of Bond Stress and Slip


Analysis by Finite le ment Metbod
3.1 Introduction
3-2 Contact Element
3.3 Linkage Element
3.4 Temperature Changes, Prestrains and Support
Displacements
3.5 Transformation Matrix
3-6 Material Properties
3 -6 1 Concrete Properties
3.6-2 Reinforcement Properties
3.7 Bond Stress and Slip Models of Finite
Element Analysis
3-7.1 Models of Bond-Slip Element
3.7.2 Models of Local Bond Behavior
3.7.2.1 Mehlhorn Model (Model 1)
3.7.2.2 Eligehausen Model (Model 2)
3.7.2.3 Moharned Harji Model (Model 3)
3.7.2.4 Proposed Model (Model 4)
3.7.2.5 Model of Bond Stress-Slip Relationship
with Partial Confining Pressure
Table of Contents

3.7.2.6 Models 2,3, and 4 under Cyclic


Loading
3 -7.2.7 Effect of Hooks in Reinforcement
3.7.2.8 Normal Bond Stiffiiess

CHAPTER 4 Evaluation of Models by Cornparison


Between Finite Element Analysis and
Experimea t 124
4.1 Verification of the Program 124
4 -2 Expenmental Details 125
4.2.1 Dimensions and Reinforcement Amangement
of the Specimens 125
4.2.2 Materiai Properties of the Specimen 126
4.2.3 Loading History of the Specimens 127
4.3 Study of Element Mesh 128
4.4 Comparison of the Analytical and
Experimental Results 129
4 -4.1 Shear Wall- 1 130
4.4.2 Shear Wall-2 134
4.4.3 Shear Wall-3 138
4.4.4 Shear Wall4 141

CHAPTER 5 Discussion and conclusions


5.1 Discussion
5.2 Conclusions
Table of Contents

References
List of Figures
Fig. 2-1 Specimen details in pull-out test before the 1970's
Fig. 2-2 Specimen details in pull-out test d e r the 1970's
Fig. 2-3 (a) Splitting mode of thin cover.
Fig. 2-3 (b) Splitting mode of thick cover.
Fig. 2-4 Load-slip curves.
Fig. 2-5 Details of the beam tested by Lutz.
Fig. 2-6 Bond stress distribution in constant moment region near a crack.
Fig. 2-7 Longitudinal variation in stresses along steel-concrete interface.
Fig. 2-8 Separation between concrete ahd bar near a pnmary crack.
Fig. 2-9 Ratio of crack widths at the steel and at the concrete surface.
Fig. 2-10 Tension and bond stress distribution for eccentric pull-out tests.
Fig. 2-1 1 Bar tension and bond stress distributions in beam at a crack.
Fig. 2-12 Effect of concrete strength on eccentric pull-out tests.
Fig. 2-13(a) Details of specimens and loads.
Fig. 2-13(b) Relationship between bond and steel stress for ail three
specimens.
Fig. 2-14 Types of specimens tested by Morita.
Fig. 2-15 Bond stress distribution in pull-out specimen type 1.
Fig. 2-16 Bond stress distribution in pull-out specimen type 2.
Fig. 2-17 Bond stress distribution in pull-out specimen type 3.
Fig. 2-1 8 Local bond stress versus local slip relationship in pull-out test.
Fig. 2-19 Test specimen used by other researchers for studying bond
behavior.
Fig. 2-20 Test specimen used by Jiang for studying bond behavior.
Fig. 2-21 Local bond stress and local slip relationship.
Fig. 2-22 Average characteristic bond stress versus lug inclination.
Fig. 2-23 The bars profiles a to d.
Fig. 2-24 Average bond stress-slip characteristic for bars of senes a to d.
Fig. 2-25 Effect of concrete strength and cover thickness on slip.
Fig. 2-26 Bond slip versus specirnen size.
Fig. 2-27 Bond slip of square specimen versus steel stress.
Fig. 2-28 Bond stress-slip relationship.
Fig. 2-29 Bond stress-slip curve for beam and pull-out specimen.
Fig. 2-30 Bond stress distribution in the specimen.
Fig. 2-3 1 Bond stress and slip relationship proposed by Hayashi et al.
Fig. 2-32 Coefficient of local maximum bond stress deterioration.
Fig. 2-33 Bond stress distribution along 25 mm plain round bar.
Fig. 2-34 Bond stress distribution for 25 mm cold-worked nbbed bar.
List of Figures

Fig. 2-35 Bond stress distribution for 25 mm hot-rolled ribbed bar.


Fig. 2-36 Bond stress-steel stress relationship for plain round bar.
Fig. 2-37 Bond stress-steel stress-slip relationship for mild steel bar.
Fig. 2-38 Bond stress-steel stress-slip relationship for cold-worked ribbed
bar.
Fig. 2-39 Bond stress-steel stress-slip relationship for hot-rolled ribbed bar.
Fig. 2-40 Bond stress-slip relationship for cold-worked nbbed bar.
Fig. 2-4 1 Bond stress-slip relationship for hot-rolled ribbed bar.
Fig. 2-42 Bond stress-slip response for pull-out failure.
Fig. 2-43 Bond stress-slip response for splitting failure.
Fig. 2-44 Analytical bond model for pull-out failure.
Fig. 2-45 Analytical bond model for splitting failure.
Fig. 2-46 Experimentally derived local bond stress-slip curves.
Fig. 2-47 Idealized local bond stress-slip curve for rnonotonic loading.
Fig. 2-48 Effect of lateral pressure on local bond stress-slip relationship.
Fig. 2-49 Effect of rate of slip on the local bond stress-slip relationship.
Fig. 2-5 1 Specimen tested by Tassios.
Fig. 2-52 Bond stress-slip relationship for monotonic loading.
Fig. 2-53 Steel stress distributions for monotonic loading.
Fig. 2-54 Local bond stress versus local slip curves in rnonotonic loading.
Fig. 2-55 Bond stress distributions for various loading levels of cyclic
loading.
Fig. 2-56 Bond stress distributions for various cyclic loading levels.
Fig. 2-57 Slip distributions for various cyclic loading levels.
Fig. 2-58 Local bond stress-local slip relationship under cyclic loading.
Fig. 2-59 Relationship of bond stress versus slip under cyclic loading.
Fig. 2-60 Local bond stress-slip relationship for repeated loading.
Fig. 2-6 1 Local bond stress-slip experimental curves for cyclic loading of
confined specimens by Eligehausen et al.
Fig. 2-62 Local bond stress-slip curves for deformed bars with hooks by
experiments.
Fig. 2-63 Free body of a reinforced concrete member.
Fig. 2-64 Slip distribution along the bar.
Fig. 2-65 Bond stress distribution along the bar.
Fig. 2-66 Cornparison of bond stress dismbution between Yang and
Samayaji.
List of Figures

Fig. 3-1 Relative displacements A.


Fig. 3-2 Linkage element.
Fig. 3-3 Relation of local coordinate and global coordinate system.
Fig. 3-4 Compressive softening parameter f3 of concrete.
Fig. 3-5 Constitutive response of cracked concrete in compression.
Fig. 3-6 Constitutive response of cracked concrete in tension.
Fig. 3-7 Strength enhancement due to biaxial compression.
Fig. 3-8 Stress-strain relation for biaxially compressed concrete.
Fig. 3-9 Poisson's ratio of concrete in compression.
Fig. 3-1 0 Hysteresis model for concrete.
Fig. 3- 1 1 Stress-strain relationship of reinforcement.
Fig. 3- 12 Hysteresis model for reinforcement.
Fig. 3-13 Relationship of local bond stress and slip.
Fig. 3-14 Influence of the direction of casting of concrete corresponding to
the pull-out direction.
Fig. 3-15 Influence of the relative rib area of the reinforcement bar
normalized to a,=0.065.
Fig. 3-16 Relationship of bond strength and -
confining pressure on the
contact surface of a steel bar.
Fig. 3- 17 Relationship of bond strength and gap between steel bar and
concrete.
Fig. 3- 18 Angle between the steel bar and the crack of the concrete.
Fig. 3-19 Influence of angle between steel bar and the crack and distance
between the linkage element to a nearest crack on f342.
Fig. 3-20 The bond stress and slip relationship of Eligehausen model.
Fig. 3-2 1 The bond stress and slip relationship of Model 3 (Harjli).
Fig. 3-22 The bond stress and slip relationship of proposed model (Model 4).
Fig. 3-23 The bond stress and slip relationship with partially confiniag
pressure.
Fig. 3-24 The bond stress and slip relationship under cyclic loading
proposed by Eligehausen et al.
Fig. 3-25 Analytical local bond stress and slip model for deformed bars with
hooks.

Fig. 4-1 Finite element mesh of a specimen used b veriQ the TRIX.
Fig. 4-2 Distribution of tensile force in bar under load =20 kN
(linkage element)
List of Figures

Fig. 4-3 Distribution of tensile force in bar under load =20 k N


(contact element)
Fig. 4-4 Distribution of tensile force in bar under load =70 kN
(linkage element)
Fig. 4-5 Distribution of tensile force in bar under load =70 kN
(contact element)
Fig. 4-6 Cross-sections of walls ( K u z m ~ o v i c 1994).
,
Fig. 4-7 Front and side view of specimen (Chio, 1995).
Fig. 4-8 Front and side view of reinforcement-Wall-3 (Chio, 1995).
Fig. 4-9 Side view of reinforcement-Wall-3 (Chio, 1995).
Fig. 4-1 0 Front and side view of reinforcement-Wall4 (Chio, 1995).
Fig. 4-1 1 Side view of reinforcement-Wall4 (Chio, 1995).
Fig. 4-12 Stress-strain curves of concrete used in Wall-3 and Wall-4.
Fig. 4- 13 Stress-strain curves for reinforcement steel.
Fig. 4- 14 Loading history of the specimen Wall-1 .
Fig. 4-15 Loading history of the specimen Wall-2.
Fig. 4- 16 Loading history of the specimen Wall-3.
Fig. 4-17 Loading history of the specimen Wall-4.
Fig. 4- 18 Coarse mesh for Wall- 1.
Fig. 4- 19 Fine mesh for Wall- 1.
Fig. 4-20 Finer mesh for Wall-1.
Fig. 4-2 1 Mesh for Wall-l when considering bond-slip between interfaces of
al1 steel bars and concrete of web.
Fig. 4-22 Final FE mesh for Wall-1. .
Fig. 4-23 FE mesh for Wall-2 and Wall-3.
Fig. 4-24 FE mesh for W a l l 4
Fig. 4-25 Computed load-displacement response of Wall-1 using coarse
smeared elements.
Fig. 4-26 Computed load-displacement response of Wall-1 using fine and
finer smeared elements.
Fig. 4-27 Load-displacement of Wall-1 assuming perfect bond.
Fig. 4-28 Load-displacement of Wall-l for bond-slip Model 1
(linkage element).
Fig. 4-29 Load-displacement of Wall-1 for bond-slip Model 2
(linkage element).
Fig. 4-30 Load-displacement of Wall-1 for bond-slip Model 3
(linkage element).
List of Figures

Fig. 4-3 1 Load-displacement of Wall-1 for bond-slip Model 4


(linkage element).
Fig. 4-32 Load-displacement of Wall-1 for bond-slip Model 1
(contact element).
Fig. 4-33 Load-displacement of Wall-1 for bond-slip Model 2
(contact element).
Fig. 4-34 Load-displacement of Wall-1 for bond-slip Madel 3
(contact element).
Fig. 4-35 Load-displacement of Wall-1 for bond-slip Model 4
(contact element).
Fig. 4-36 Load-displacement envelope of Wall-1 computed assuming
perfect bond (PB), computed using bona-slip Model 2 with
contact elements (CB2), and as measured during test.
Fig. 4-37 Load-displacement of Wall- l fiom test data (Kumanovic, 1994).
Fig. 4-38 Load-displacement of Wall- 1 assuming perfect bond.
Fig. 4-97 Load-displacement of Wall-2 for bond-slip Model 1
(linkage element).
Fig. 4-40 Load-displacement of Wall-2 for bond-slip Mode12
(linkage element).
Fig. 4-41 Load-displacement of Wall-2 for bond-slip Model 3
(1 inkage e lement).
Fig. 4-42 Load-displacement of Wall-2 for bond-slip Model 4
(linkage element).
Fig. 4-43 Load-displacement of Wall-2 for bond-slip Model 1
(contact element).
Fig. 4-44 Load-displacement of Wall-2 for bond-slip Model 2
(contact element).
Fig. 4-45 Load-displacement of Wall-2 for bond-slip Model 3
(contact element).
Fig. 4-46 Load-displacement of Wall-2 for bond-slip Model 4
(contact element).
Fig. 4-47 Load-displacement envelope of Wall-2 computed assuming
perfect bond (PB), computed using bond-slip Model 2 with
contact elements (CB2), and as measured during test.
Fig. 4-48 Load-displacement of Wall-2 fiom test data (Kmanovic, 1994).
Fig. 4-49 Load-displacement of Wall3 assuming perfect bond.
Fig. 4-50 Load-displacement of Wall3 for bond-slip Model 1
(linkage element).
List of Figures

Fig. 4-5 1 Load-displacement of Wall-3 .for bond-slip Model 2


(linkage element).
Fig. 4-52 Load-displacement of Wall-3 for bond-slip Model 3
(linkage element).
Fig. 4-53 Load-displacement of Wall-3 for bond-slip Mode14
(linkage element).
Fig. 4-54 Load-displacement of Wall-3 for bond-slip Model 1
(contact element).
Fig. 4-55 Load-displacement of Wall-3 for bond-slip Mode12
(contact element).
Fig. 4-56 Load-displacement of Wall-3 for bond-slip Model 3
(contact element).
Fig. 4-57 Load-displacement of Wall-3 for bond-slip Model 4
(contact element).
Fig. 4-58 Load-displacement envelope of Wall-3 computed assuming
perfect bond (PB), computed using bond-slip Model 2 with
contact elements (CB2), and as measured during test.
Fig. 4-59 Load-displacement of Wall-3 fiom test data (Kuunanovic, 1994).
Fig. 4-60 Load-displacement of Wall4 assuming perfect bond.
Fig. 4-6 1 Load-displacement of Wall-4 for bond-slip Model 1
(linkage element).
Fig. 4-62 Load-displacement of Wall-4 for bond-slip Model 2
(1inkage element).
Fig. 4-63 Load-displacement of Wall4 for bond-slip Model 3
(linkage element).
Fig. 4-64 Load-displacement of Wall-4 for bond-slip Model 4
(linkage element).
Fig. 4-65 Load-displacement of Wall4 for bond-slip Model 1
(contact element).
Fig. 4-66 L~ad-displacementof Wall4 for bond-slip Model 2
(contact element).
Fig. 4-67 Load-displacement of Wall4 for bond-slip Model 3
(contact element).
Fig. 4-68 Load-displacement of Wall4 for bond-slip Model 4
(contact element).
Fig. 4-69 Load-displacement envelope of Wall4 computed assuming
perfect bond (PB), computed using bond-slip Model 2 with
contact elements (CB2), and as measured during test.
List of Figures

Fig. 4-70 Load-displacement of Wall4 fiom test data (Kuzmanovic, 1994).


Fig. 4-71 Load-displacement of Wall-1 computed assuming perfect
bond (PB), and computed using contact elements with
bond-slip Model 2 (CB2) under rnonotonic loading.
Fig. 4-72 Load-displacement of Wall-2 computed assuming perfect
bond (PB), and computed using contact elements with
bond-slip Model 2 (CB2) under rnonotonic loading.
Fig. 4-73 Load-displacement of Wall-3 computed assuming perfect
bond (PB), and computed using contact elements with
bond-slip Model 2 (CB2) under rnonotonic loading.
Fig. 4-74 Load-displacement of Wall4 computzd assuming perfect
bond (PB), and computed using contact elements with
bond-slip Model 2 (CB2) under rnonotonic loading.
Fig. 4-75 Bond slip envelope of Wall-1 (Mode! 2, linkage element).
Fig. 4-76 Bond slip envelope of Wall4 (Model 3, linkage element).
Fig. 4-77 Bond slip envelope of Wall-1 (Model 4, linkage elementj.
Fig. 4-78 Bond slip envelope of Wall-1 (Model 2, contact element).
Fig. 4-79 Bond slip envelope of Wall-1 (Model3, contact element).
Fig. 4-80 Bond slip envelope of Wall-1 (Model 4, contact element).
Fig. 4-81 Bond slip envelope of Wall-2 (Model 2, linkage element).
Fig. 4-82 Bond slip envelope of Wall-2 (Model 2, contact element).
Fig. 4-83 Bond slip envelope of iVall-3 (Model 2, linkage element).
Fig. 4-84 Bond slip envelope of Wall-3-(Mode1 2, contact element).
Fig. 4-85 Bond slip envelope of Wall4 (Model 2, linkage element).
Fig. 4-86 Bond slip envelope of Wall4 (Model 2, contact element).
Fig. 4-87 Bond slip at the south bottom of Wall-1 (Model 2, contact
element).
Fig. 4-88 Bond slip at the north bottom of Wall-1 (Model 2, contact
element).
Fig. 4-89 Bond stress at the south bottom of Wall-1 (Model 2, contact
element).
Fig. 4-90 Bond stress at the north bottom of Wall-1 (Model 2, contact
element).
Fig. 4-91 Steel stress at the south bottom of Wall-1 (Model 2, contact
element).
Fig. 4-92 Steel stress at the north bottom of Wall-1 (Model 2, contact
element).
List of Figures

Fig. 4-93 Bond slip at the south bottom of Wall-2 (Model 2, contact
element).
Fig. 4-94 Bond slip at the north bottom of Wall-2 (Model 2, contact
element).
Fig. 4-95 Bond stress at the south bottom of Wall-2 (Model 2, contact
element).
Fig. 4-96 Bond stress at the north bottom of Wall-2 (Model 2, contact
element).
Fig. 4-97 Steel stress at the south bonorn of Wall-2 (Model 2, contact
element).
Fig. 4-98 Steel stress at the north bottom of Wall-2 (Model 2, contact
element).
Fig. 4-99 Bond slip at the south bottom of Wall-3 (Model 2, contact
element).
Fig. 4-100 Bond slip at the north bottom of Wall-3 (Model 2, contact
element).
Fig. 4-101 Bond stress at the south bottom of Wall-3 (Model 2, contact
element).
Fig. 4-102 Bond stress at the north bottom of Wall-3 (Model 2, contact
element).
Fig. 4-1 03 Steel stress at the south bottom of Wall-3 (Model 2, contact
element).
Fig. 4-104 Steel stress at the north bottom of Wall-3 (Model 2, contact
element).
Fig. 4-105 Bond slip at the south bottom of Wall4 (Model 2, contact
element).
Fig. 4-106 Bond slip at the north bottom of Wall-4 (Model 2, contact
element).
Fig. 4-107 Bond stress at the south bottom of Wall-4 (Model 2, contact
element).
Fig. 4- 108 Bond stress at the north bottom of Wall-4 (Model 2, contact
element).
Fig. 4-109 Steel stress at the south bottom of Wall4 (Model 2, contact
element).
Fig. 4-1 10 Steel stress at the north bottom of Wall4 (Model2, contact
element).
Fig. 4-1 11 Deflection of Wall-1 after failure (assurning perfect bond).
Fig. 4-1 12 Deflection of Wall-1 after failure (linkage element, Mode! 2).
List of Figures

Fig. 4-1 13 Deflection of Wa Fter fai Lure (linkage elernent, Model 3).
Fig. 4-1 14 Deflection of Wa Fter fai Lure (linkage element, Model4).
Fig. 4- 115 Deflection of Wa Fter fai Lure (contact element, Model 2).
Fig. 4-1 16 Deflection of Wa Fter fai lure (contact element, Model3).
Fig. 4- 1 17 Deflection of Wa Eter fai lure (contact element, Model 4).
Fig. 4-1 18 Deflection of Wa Rer fai Lure (assuming perfect bond).
Fig. 4-1 19 Deflection of Wa Rer fai Lure (linkage element, Model 2).
Fig. 4-120 Deflection of Wa Eter fai lure (linkage element, Mode1 3).
Fig. 4-1 2 1 Deflection of Wa !ter fai lure (linkage element, Model 4).
Fig. 4-122 Deflection of Wa Rer fai Lure (contact element, Model 2).
Fig. 4-1 23 Deflection of Wa Rer fai lure (contact element, Model 3).
Fig. 4-1 24 Deflection of Wa Fter fai lure (contact element, Model 4).
Fig. 4-1 25 Deflection of Wa fier fai lure (assum ing perfect bond).
Fig. 4-126 Deflection of Wa Fter fai .ure (linkage element, Model 2).
Fig. 4-127 Deflection of Wa 3er fai ure (linkage element, Model 3).
Fig. 4-128 Deflection of Wa 3er fai ure (linkage element, Model 4).
Fig. 4-129 Deflection of Wa 3er fai,.ure (contact element, Model 2).
Fig. 4-130 Deflection of Wa 3er fai s r e (contact element, Model 3).
Fig. 4- 13 1 Deflection of Wa !ter fai .ure (contact element, Model 4).
Fig. 4- 132 Deflection of Wa 3er fai ure (assuming perfect bond).
Fig. 4- 133 Deflection of Wa 3er fai ure (linkage element, Model 2).
Fig. 4-134 Deflection of Wa 3er fai .ure (linkage element, Model 3).
Fig. 4-135 Deflection of Wa 3er fai ure (linkage element, Model 4).
Fig. 4-136 Deflection of Wa 3er fai ure (contact element, Model 2).
Fig. 4-137 Deflection of Wa 3er fai ure (contact element, Model 3).
Fig. 4-1 38 Deflection of Wa 3er fai ure (contact element, Model 4).
List of Tables
Table 2-1 Confining pressure in experiments of Figure 2-48 (a)
Table 2-2 Confining pressure in experiments of Figure 2-49 (a)

Table 3-1 Ultimate relative displacements Au (mm).

Table 4-1 Matenal data.


Table 4-2 Specimen parameters and analytical maximum slips (mm).
Table 4-3 Uniaxial compressive strength of the Wall-1 and Wall-2.
Table 4-4 Uniaxial compressive strength of the Wall-3 and W a l l 4
Table 4-5 Characteristics of reinforcement steel.
Table 4-6 Comparison of ultimate loads acting at top of Wall-1 fiom test
data and FE analysis 0.
Table 4-7 Maximum and minimum slips of Wall-1 by FE analysis (mm).
Table 4-8 Comparison of ultimate loads for Wall-2 fiom test data and
FE analysis (kN).
Table 4-9 Ultimate loads for Wall-3 fiom test data and FE analysis 0.
Table 4-10 Comparison of ultimate loads acting for Wall4 from test data
and FE analysis (kN).
Table 4-1 1 Comparison of ultimate loads for the four test walls calculated
assuming perfect bond and bond-slip Mode1 2 with contact
bond elements under rnonotonic loading 0.

Table 5-1 Comparison of ultimate loads of walls fiom test data and FE
analysis 0.
BOND STRESS AND SLIP MODELING IN
NONLINEAR FINITE ELEMENT ANALYSIS OF
REINFORCED CONCRETE STRUCTURES

CHAPTER 1
INTRODUCTION

1.1 Background
Bond stress is the shear stress acting parallel to an embedded bar on the
surface between the reinforcing bar and the concrete. Bond slip is the
relative displacement between the bar and the concrete. Reinforced
concrete depends on the combined action of the concrete and its
embedded reinforcement for satisfactory operation as a construction
material. This action is produced by the interaction between both of its
components, plain concrete and reinforcing bars. The transfer of forces
across the interface between these two materials is completed by bond
action behveen them, so bond plays a very important role in most aspects
of reinforced concrete behavior. To better understand bond behavior, there
have been a number of studies specifically aimed at examining behavior
of bond stress-slip by way of both expenment and theory. Several laws to
describe the behavior have been developed on experimental ground.
Constitutive laws between the local bond stress and the local bond slip on
the interface have been formulated for theoretical analysis purposes.
The finite element method has been widely used for analyzing the
response of reinforced and prestressed concrete structures. niere are two
distinct ways of representing cracks in a finite element procedure. The
discrete crack mode1 represents cracks as interslement discontinuities.
There is restriction on the crack propagation direction depending on the
mesh layout. The smeared crack approach represents cracks as a change in
the material property of the element over which the cracks are assumed to
be smeared. In the simplified analysis of reinforced and prestressed
concrete stmctures, complete compatibility between concrete and
reinforcement or prestressed tendons is usually assurned. It means that
perfect bond is presurned. For models with smeared steel, the perfect bond
relationship is the easiest to adopt since it simply involves overlaying the
constitutive matrix of the steel elements with concrete elements. For the
models with discrete reinforcement elements, perfect bond also represents
a very easy solution, since the displacements of the nodal points are the
sarne for both concrete elements and reinforcement elements. Actually,
this assumption is only valid in regions where only low transfer stresses
between the two components exist. In the regions where high transfer
stresses occur between the interfaces of the two components, especially
for the regions near cracks, there are different strains in concrete and
reinforcement. As result of this, relative displacements, which are called
bond slips, occur between concrete and reinforcement. Because bond
stresses are related to bond slips between the two components, the
assumption of petfect bond in a cracked region would require infinitely
high reinforcement strains to explain the crack widths. It would also cause
significant error in the predicted load-deflection response, stresses and
strains of the reinforced structures when the bar slip is large.
Most of the work done in the past to understand bond behavior
focussed on experimental and analytical study of concrete prism
specimens with an embedded steel bar. In most cases, the model was
used to explain the behavior of the specimens fiom which the model was
derived. This does not constitute real verification of the model.
Theoretical methods proposed by some researchers are only applicable
for very simple members such as concrete prism with an embedded steel
bar, and can not be used for reinforced concrete structures. However,
little work has been done towards verimg the bond stress-slip models
obtained fiom experimental investigations for reinforced concrete
structures and studying the effects of bond slip on behavior of reinforced
concrete structures. Such a study is vital in verifjring bond stress-slip
models and investigating behavior of reinforced concrete structures.

1.2 Scope
The scope of this thesis is to study the bond stress-slip models under
both rnonotonic loading and cyclic loading; to verie these models by
comparing analytical response with the experimental results of reinforced
concrete shear walls; to study the effects of bond slip on reinforced
concrete shear walls; and to provide a kind of method of nonlinear finite
element analysis for reinforced concrete structures when bond slip is
considered.

1.3 Plan of Study


Published experimental and theoretical research under both
rnonotonic and cyclic loading are studied in the first part of this thesis
(Chapter 2), to investigate factors which affect bond behavior and to
provide bond-slip models. Chapter 3 studies the modeling of bond stress
and slip analysis by finite element method. First, two types of bond-slip
elernents, contact element and linkage element, are studied and
formulations are presented for use in 'finite element analysis; then four
types of bond-slip models are described and implemented into the
nonlinear finite element program TRIX99. Chapter 4 evaluates the bond-
slip models and elements by cornparison between finite element analysis
results and experimental data of a set of four shear walls tested at the
University of Toronto. Chapter 5 discusses the bond-slip elements and
models presented in chapter 4, and gives the conclusions of this study.
CHAPTER 2
REVIEW OF PREVIOUS STUDIES

2.1 Introduction
Many researchers have studied bond by performing experiments on
strain-gaged reinforcing bars embedded in a variety of concrete
specimens. In these tests, the axial strain in the bars was measured
directly, but bond stresses could only be detennined indirectly from the
slope of the steel strain curve, while the bond slips were determined fkom
displacements obtained by numerical integration of the difference of
strains between steel bar and concrete in the interface.
Study of bond behavior between steel bar and surrounding concrete
began in the 1960s. Most of the studies perfonned in 1960s and 1970s
were experimental investigations under monotonic loading. Study of bond
behavior under cyclic loading has been done since 1980s. Because there
exist many variables that affect the bond stress and bond slip relationship,
and because measurement of bond stress and bond slip is difficult and
sensitive to experimental errors, it is very difficult to get a generally
applicable law from experimental investigations. Some researchers
developed the theoretical method to study bond behavior by analytical
models based on some assumptions since 1980s.
The bond stress between the reinforcing bar and its surrounding
concrete is very complicated. It is dependent on the slip between the steel
bar and the concrete, and the stress in the reinforcing bar as well as many
other factors. Two alternative basic hypotheses have been used in the Fast;
in one bond stress is considered to be a linear function of slip ( Ngo and
Scordelis, 1967), while in the other it is considered to be a nonlinear
relationship between bond stress and slip.
Early experimental studies of bond were concerned with determining
bond failure strengths and the influence of surface deformations on them
by push-in test. Later some researchers found that bond failure occurs at a
higher stress for a push-in test than for the normal pull-out test. The
explanation for this is simple: firstly, the compressive axial stress
developed in the bar during a pushin test causes an increase in bar
diameter because of Poisson's effect, which, in tum, causes an increase in
the radial pressure between bar and concrete. Since friction is an
important element in bond, this increase in pressure leads to an increased
bond strength; secondly, cracking is an important reason to cause
degradation of bond strength. Because there is no cracking in a push-in
test, this leads to the increase of bond strength.
In the ordinary pull-out test, the test specimen is typically a cylinder or
prism with a bar embedded in it. Before the 19709s9the bar was pulled
fi-om one side while the concrete was held by the reaction pressure on the
same end as shown in Fig. 2-1. Since the bar is in tension and concrete in
compression, differential strains force a relative slip at very low steel
stresses. This is different to what occurs in an actual structure. The most
senous weakness of this pull-out test is that the concrete in compression
eliminates transverse tension cracking. After the 19709s, the method of
pull-out test was improved. A specimen is pulled firom both sides of the
bar (Fig. 2- 2).
Fig. 2-1 Specimen details in pull-out test before the 1970's.

Fig. 2-2 Specimen in pull-out test after the 1970's.

Most researchers believe that there are different relationships for local
bond stresses versus local bond slips at different points of the interface of
the steel bar and the concrete. Since the properties of the interface will not
be different, this change can only corne fiom stress effects. Some
researchers (Morita, 1985; Narnmur and Naaman, 1989; Edward and
Yannopoulos, 1979) thought that the relationship is a material property
and, therefore, independent of location. They thought that there exists a
unique bond slip relationship which depends only o n matenal properties
and steel geometry.

2.2 Mechanics of Bond Slip


The strength of the bond between a rebar and the surrounding concrete
is generally made up of three components:
1. Chernical adhesion;
2. Friction;
3. Mechanical interlock between reinforcement and concrete. This
includes the bearing of lugs on concrete and shear strength of concrete
section between lugs.
Bond of plain bars depends primarly on the first two mechanisms,
although there is some mechanical interlocking due to the roughness of
the bar surface. However, even a low .bar stress causes slip suficient to
break the adhesion immediately adjacent to a crack in the concrete. Over
the slipping length o d y the friction drag remains, and the highest adhesive ,

bond stress can act only close to this slipping portion. Bond resistance is
thus an ultimate bond stress.over a short length where adhesion is about to
fai1.
Deformed bars change this behavior. Adhesion and fiction still assist,
but the primary resistance has been changed to mechanical interlocking
for superior bond properties. With deformed bars, a pull-out specimen
nearly always fails by splitting; the concrete splits into two or three
segments rather than failing by crushing against the lugs or by shearing on
the cylindrical surface which the lugs-tend to strip out. Splitting results
from the wedging action of the lugs against the concrete. It should be
noted that a split on one face of a concrete member does not represent .
complete failure. When stimps are present, significant bond resistance
remains.
Clear cover over a reinforcing bar will be significant in connection with
splitting resistance. Thin cover can be easily split like Fig. 2-3 (a); a thick
cover can greatly delay splitting if bars are not closely spaced laterally. If
a number of bars are closely spaced with thick cover in a beam, a splitting
failure will occur as shown in Fig. 2-3@).

Fig 2 3 (a) Splitting mode of thin cover

Fig. 2-3 (b) Splitting mode of thick cover.

Lutz and Gergely (1967) investigated the mechanics of bond and slip.
For deformed bars, initially, chemical adhesion combined with
mechanical interaction prevents slip. Afier adhesion is destroyed, as slip
occurs, the rib of a bar restrains this movement by bearing against the
concrete between the ribs. Friction, which would occur after the slip of
plain bars, does not occur here because of the presence of the ribs.
Slip of a deformed bar can occur in two ways: (1) the ribs c m push the
concrete away from the bar (wedging action), and (2) the ribs can cmsh
the concrete.
Tests indicate that the movement of the ribs is about the same for al1
ribs with a face angle greater than about 40 degrees. In bars with ribs
having face angles near 90 degrees, the ribs can not push concrete
outward (no wedging action). The friction between the rib face and the
concrete is suficient to prevent relative movement at the interface when
face angles are larger than 40 degrees. The slip is due almost entirely to
tlie crushing of the porous concrete paste in front of the ribs if the rib face
angles are larger than about 40 degrees. Bars with ribs having face angles
less than about 30 degrees exhibit a markedly different load slip
relationship. Here the friction between the rib face and the concrete is not
sufficient to prevent relative movement. Thus, slip is due mainly to the
relative movement between the concrete and the steel bar dong the face of
the rib, and secondarily to some crushing of the concrete.
For the usual case of good fictional properties and a nb face angle
greater than 40 degrees, slip occurs by progressive crushing of the porous
concrete paste structure in front of the rib.
The slip resistance on reloading is considerably higher than the slip
resistance initially, as shown in Figure 2-4.
Lutz et al also experime&ally investigated the cracking effects on the
bond stress distribution in concrete beams. The details of the beam used in
tlie test are shown in Fig. 2-5. Before cracking, the bar adheres chemically
to the concrete. This adhesion acts until slip or movement of the steel
relative to the adjacent concrete occurs. Tests showed that adhesion would
be lost in tension rather than in shear. Before slip and cracking occur,
tension can exist on the interfae between the bar and concrete as a result
of the large Poisson's ratio of the steel and concrete (Poisson ratios of 0.3
and 0.15 for the steel and concrete, respectively). Due to the loading, no
bond stresses exist between the concrete and reinforcing bar in the middle
third of the bearn. In the outer thirds of the beam, because there is no slip,
the steel and concrete elongate the sarne amount over any given length,
but due to the difference in the elastic moduli of the two materials, a
larger change in stress occurs in the steel than in the concrete, leading to
bond stresses.

O 2 4 6 8 10
distance dong bar liom loaded end(in.j

Fig. 2-4 load-slip curves (unit of loads: lb).

Fig. 2-5 Details of the beam tested by Lutz.


M e r a flexural crack forms in the constant moment region of the
beam, the stress condition in the vicinity of the crack is very complex.
The bond stress distribution near a crack is sketched in Figure 2 6 . As
shown in Fig. 2-7, the circumferential tensile stresses in the concrete
around the bar are very small prior to flexural cracking. However, near
transverse cracks bonding forces cause large circumferential tensile
stresses. Also, radial tensile stresses, acting normal to the concrete steel
interface, destroy contact near the crack and allow separation and slip
between the bar and the concrete.

Fig. 2-6 Bond stress distribution


in constant moment region near a crack.

--Radial stress
.= 80 - --Circumferential stress
-ss --Bondstress
--Longitudinal stress

crack
O 1 2
1 xld

Fig. 2-7 Longitudinal variation


in stresses dong steel-concrete interface.
In the case of plain bars this separation would mean complete loss of
bond in the region next to the crack. The concrete stress at the bar surface
and parallel to the bar would disappear, and the crack width at the bar
would be essentially the same as the width at the surface of the concrete.
However, when the reinforcing bar has transverse nbs, separation does
not produce complete unloading of the concrete adjacent to the bar
inasmuch as the bar nbs prevent much of the opening of the crack at the
bar (Fig. 2-8). Some unloading does occur, allowing the crack to open at
the surface of the bar. This opening is caused partly by the unloading of
.
the concrete between the crack and the nearest bar rib when the crack
forms, producing a relative contraction of the concrete, and partly by the
bearing deformation under the ribs. Another factor conbibuting to the
opening is the inclination of the bar ribs, since the separation of the bar
and the concrete is accompanied by a movement along the inclined bar rib
face. The bond stress near a transverse crack is transferred only by bearing
of the concrete against the face of the ribs. With an increase in the load
above the cracking load, additional extemal cracks form. At each of the
surface cracks the steel stress will reach a local peak; between cracks the
steel stress is lower as part of the tension force is carried by the concrete.
The transfer of forces produces bond stresses.
The interna1 transverse cracks tend to occur before longitudinal cracks .

if the steel stress is high relative to the bond stress (in the flexural region),
while splitting would occur fust when the bond stress is large relative to
the steel stress (in the anchorage region). Fig. 2-9 shows the ratio of the
crack width adjacent to the bar to the one at the concrete surfaces. A large
variation in the crack widths occurs fiom the bar to the concrete surface
with increasing steel stress. .

P~~.IMIYor
surfa crack

Fig. 2-8 Separation between concrete and bar near a primary crack.

0.6 -
O
3 0.5 - -
ws width of crack at bar surhce
f WC - wvidth of crack at concrete surface

0.3
1t,

O 20 40 60 80 100
*el a r e s at free ends ( ksi )

Fig. 2-9 Ratio of crack widths at the steel and at the concrete surface.

Ingraffea et al (1984) studied the fi-acture mechanics of bond in


reinforced concrete. They thought there are several possible contributions
to bond behavior. These are: elastic deformation; crushing at points where
concrete bears on steel ribs; secondary radial cracking; and longitudinal
*

splitting cracking. They researched the radial secondary cracking. This


cracking was found not to follow the principles of linear elastic fracture
mechanics, and was therefore modeled using a nonlinear discrete-crack
finite element approach. They proposed a new method for finite element
modeling of nonlinear fracture and introduced the "tension-softening
element" for the purpose of modeling bond stress-slip in practical
engineering problems. The characteristics of this element were derived by
studying bond behavior allowed by secondary cracking in the tension pull
specimen.
They concluded that secondary cracking is an important mechanism
allowing bond slip to occur. Secondary and primary cracking are
interrelated and secondary cracks are generally confined to a region close
to the location where a reinforcing bar crosses a primary crack.

2.3 Factors Affecting Bond Behavior

2.3.1 Monotonic Loading


Under static monotonically increasing loads, the main factors that
affect bond behavior are the concrete strength, yield strength of steel bar,
bar size, cover and bar spacing, geometry of bar, surface condition of bar,
transverse reinforcement, and concrete casting position.

2.3 -2 Cyclic Loading


Al1 parameters that are of importance under rnonotonic loading are
also of importance under cyclic loading. In addition, the bond stress range,
bond slip range, and type of loading are important factors affecting bond
behavior under cyclic loading.

2.3.3 Loads
Loads c m be subdivided into monotonic and cyclic loads. Monotonic
loading implies that slip is always increasing. Cyclic loading implies slip
reverses in direction many times during the load history. Cyclic loading i s
divided into two general categories. The first category is low-cycle
loading or low-cycle, high-stress loading. That is loading containing few
cycles but having a large range of bond stress. The second category is
high-cycle loading or fatigue loading.
The bond behavior under cyclic loading can further be divided
according to the type of stress applied. The first is unidirectional loading,
which implies that the bar stress does not change sense ( tension to
compression) during a loading cycle. The second is stress reversal, where
the bar is subjected alternatively to tension and compression.

2.3.4 Failure Modes


Under rnonotonic loading, two types of bond failures are typical. The
first is direct pull-out of the bar, which occurs when large confinement is
provided to the bar. The second type of failure is a splitting of the
concrete cover when the cover or confinement is insufficient to obtain a
pull-out failure.

2.4 The Experimental Study of Bond Stress and Slip

2.4.1 Bond Stress and Slip Under Monotonic Loading


Peny and Thompson (1966) studied the bond stress distribution in
beams and eccentric pull-out specimens. Bond stress distribution curves
are shown in Fig. 2-10 for eccentric pull-out specimens for different loads.
The point of maximum bond stress moved away from the loaded end as
the force in the bar increased.
For beams, even when shear is zero and moment is constant, large local
bond stresses exist adjacent to each flexural crack. At the crack, most of
the tension is carried by the bars and the steel stress is maximum. Between
cracks, the concrete carries tension and the steel stress drops off. Thus
bond must take stress out of the steel after a crack and put it back in the
steel just before the next crack is reached. Bond stress distribution curves
were plotted in Fig. 2-1 1 for bearns near a crack. It was noted that the
point of maximum bond stress was aiways about 1.5 inches fiom the crack
regardless of the load.
700, 1

O 2 4 6 8 10
distance along bar from loaded end (in.)

Fig. 2-10 Tension and bond stress distribution for eccentric


pull-out tests (f,' =2500 psi, unit of loads: Kips).
500
400
.- 300
& zoo
2 100
L
C
UJ O
w
5-100
-200
-300
400 - -

O 1 2 3 4 5 6 7 8 9
distance along bar from crack-in.

Fig. 2-1 1 Bar tension and bond stress distributions


in beam at a crack (f,'=5000 psi, unit of loads: Kips).
The eEect of concrete strength on the distribution of bond stress in the
eccentric pull-out specimens is s h o w in Fig. 2-12. The point of
maximum bond stress was found to shift toward the unloaded end for the
lower concrete strengths. As the maximum bond stress was reached at any
point along the bar, the bar slipped and caused the bond stress on the side
of the loaded end to reduce gradually to the vzlue of fnctional drag
between the bar and the concrete.

O 1 2 3 4 5 6 7 8 9
distance fiom foaded end - in.

Fig. 2-1 2 Effect of concrete strength on eccentric pull-out tests.


(unit of concrete strength: psi)

The authors cornpared relationships between bond stress and steel


stress for the three types of specimens shown in Fig. 2-13. Three bond
curves were plotted from steel siress distribution curves having
approximately the same steel stress at point zero. The maximum bond
'

stress in each case was not significantly different, but they occurred at
different locations. It was seen that the bar cut-off point occurred in a zone
of rapidly changing moment in one bearn, whereas the crack in the other
beam was in a zone of constant moment. The bar cut-off specimen (Beani
a) is different from others because horizontal shearing stresses exist at the
level of the bar, while there are no shearing stresses in others. These
differences probably caused the non-similarity of the bond stress
distribution for the two beam tests. When bars are cut off in a tension
zone, a flexural crack f o m s prematurely at the discontinuity. The sudden
change in bar stress produces large bond stress locally and causes
splitting.

Fig. 2-13 (a) Details of specimens and loads.

Cut off point, crack

-
Or free end

-
- pullout
beam a
beam b

-200 4
-300 J
distance along bar-in.

Fig. 2-13 (b) Relationship between bond


and steel stress for al1 three specimens.

Monta and Fujii (1985) developed a bond stress-slip mode1 used iii
finite element analysis. They conducted extensive tende tests using
deformed bars of 51 mm, 25 mm, 19 mm, 7 mm and 3 mm diameters to
investigate the efTect of bar size upon slip behavior. The following
equation was obtained fiom regression analysis of the test data :
z = 888*(Nd)
''sO @g/cm2) (2- 1
where z is the local bond stress, A is the local bond slip, and d is the
diameter of the steel bar. Using equation (2-l), stress distributions were
calculated numerically. The results indicate that if the distance fiom the
loaded end is normalized with respect to bar diameter, the stress
distributions are almost identical for bar diameters of 3 mm to 51 mm.
This means that the extent of the intemal cracking zone is probably
proportional to the bar diameter.
The other basic aspect of bond behavior is the occwence of radial
forces caused by wedging action of the ribs. Due to this action, interna1
longitudinal cracks occur in the vicinity of the bar. With the increase of
the slip, splitting of the surrounding concrete develops along the bar axis.
The local bond behavior is strongly affected by development of these
longitudinal cracks. The peak stress of local bond is ofien dependent on
the resisting capacity against splitting of entire concrete cover. The
provision of transverse reinforcement is effective only to sustain the bond
up to a large slip.
Morita et al (1985) investigated the effect of location on the
development of bond stress and bond slip (specimen is shown in Fig. 2-
14). Figures 2-15 to 2-17 show the bond stresses along the bar axis at
several loading stages for specimens type 1, type 2 and type 3. The
reduction of bond stress transfer in the neighborhood of primary cracks at
the pull-out end (the region of about 5d) is clearly demonstrated, while at
the push-in end, bond stress transfer near the primary crack is much more
effective throughout the load application. Fig. 2-18 shows the bond stress
versus bond slip relationship at some locations. The closer the location to
the loaded end, the more deteriorated the bond resistance in pull-out tests.

Fig. 2-14 Types of specimen tested by Morita.


(type 1-pull only; type 2-push only, type 3-push and pull)

-c- P=99ton

-
+el23 ton
P=149 ton

Free
end

O i/ 7 t

O 20 40 60 80
didance from pulled end ( cm )

Fig. 2-15 Bond stress distribution in pull-out specimen type 1 .

-+- P=27 ton

-
-P=51
P=99 ton
ton

Pushed
end
O 20 40 60 80
didance frorn free end (cm)

Fig. 2-16 Bond stress distribution in push-in specimen type 2.


-t-f i 147 ton
+P=l23 ton
+-ton
-x--75 ton
*Ml ton

h h end

distance from pulled end ( cm )

Fig. 2-17 Bond stress distribution in combined


pull-out and push-in specimen type 3.

O 0.2 0.4 0.6 0.8 1


local slip ( mm )

Fig. 2- 18 Local bond stress versus local slip relationships in pull-out test.

Nilson (1972) studied the spacing and width of cracks, and the
distribution of concrete stresses in partially cracked members. Usually,
bond stress-slip relationships were studied using pull-out specimens of the
type s h o w in Figure 2-19. This type of test is intended to simulate
conditions in the tension zone of a concrete beam between primary
flexural cracks and beiow the neutral axis. With this type of specimen, it .

is possible to measure the variation in steel strain, and hence to measure


the bond stress distribution, and to measure slips at the two ends.
However, it is difficult to measure local slip, strain distribution in the
surrounding concrete, and the extent of the secoridary cracks. Actually, in
experimental investigations, the exact measurement of concrete strains at
the interface is almost impossible to achieve. Therefore the strains are
measured at a certain distance away fkom the interface. In many studies of
bond slip behavior done using prismatic specimens, concrete strains dong
steel bar are measured at some distance away from the interface to give
the interface slip. The different methods used in measuring the local slip
lead to widely varyiny test results for the bond stress-slip relationship.
Jiang et al (1984) developed a new type of test specimen (Fig.2-20).
They measured directly the interface slip using a microscopic technique.
Since the steel-concrete interface was exposed, the local slip and concrete
strain were measured directly. They found the cracks generally initiated
fiom the location of the first lug to the two ends and extended at an
inclination of about 60 degrees fiom the axis of the bar towards the side of
the specimen with increasing loads. The inclination of the secondary
cracks other than the first mes was generally larger. They assumed that
resistance to relative slip, until the formation of the first secondary crack,
will be largest at the lugs nearest the primary cracks. The peak bond stress
70 (near the fust lug) was empirically related to the steel stress ad at the
end as:
TO =0.034 0, ( 1 - 0 . 0 1 ~ )ksi (2-2)
Jiang et al found that the maximum slip is not at the end of the bar but at the
location where the secondary cracks were ubserved. The relationship
between the bond stress and slip at the steel-concrete interface is not unique
but varies fiom location to location as shown in Fig. 2-2 1.
P* P/2 P/2
Fig. 2- 1 9 Test specimen Fig. 2-20 Test specimen used
used by other researchers by.Jiang for shidying bond behavior.
for studying bond behavior.
1.2 ,

Fig. 2-21 Local bond stress and local slip relationship.

For a one-dimensional model, a parabolic bond stress distribution was


assumed
r = ro [ 1- ( I - ~ X L ) ~ ] (2-3)
Soretz and ~ ~ i z e n b e i(1979)
n studied the influence of the rib
dimension of reinforcing bars to bond behavior. They considered the
influence of the following three parameters related to the formation of
lugs: the height and spacing of the lugs , the inclination of the lugs with
respect to the bar axis, and the cross-section of the lugs. They arriveci at
the following conclusions: Fustly, the bond resistance shows no
significant dependence on the pattern of the ribbed bars with an identical
related nb area. ~econdly,a reduction of the rib height seems to be
advantageous, since the danger of longitudinal cracks in the reinforced
concrete stnicture due to the splitting effect is considerably reduced. An
increase of lug height will cause a decrease of the bendability of the bars,
and the tendency for brittle &acture of the steel bar. To maintain the
necessary bond characteristic, they suggested keeping the lug spacing at
0.3d and the minimum height of lugs at 0.03d (diameter of the reinforcing
bar). Thirdly, when the inclinations of the lugs with respect to the bar axis
are from 45 degrees to 90 degrees, the bond characteristic improves
slightly but that the influence of the rib area is three-fold as important as
that of the lug inclination with increasing inclination of the lugs (Fig. 2-
22). Fourthly, the inclination of the nb flanks towards the surface of the
core from 45 degees to 90 degrees has no significant influence on either
the bond stresses and bond failure due to splitting or on the maximum slip
value (Figs. 2-23, 2-24). Fifthly, changing the rib cross section from a
rectangle to a 45 degrees trapezoid has no significant influence and
changing to a very flat triangle has only a slight influence on the bond
characteristics. The splitting effect is not influenced at all.
4 0 5 0 6 7 0 8 0 9 0
lug inclination (deg)

Fig. 2-22 Average characteristic bond stress versus h g inclination.

Fig. 2-23 The bar profiles a to d.

profile a
Y)
II) profile b
C
er profile c
II)

profile d

"
Ot O 0.1 0:2 0.3 0.4 0.5 0.6 0.7
slip (mm)

Fig. 2-24 Average bond stress-slip characteristic for bars of series a to d.

The crack width specified in various codes is the crack width at the
surface of concrete, and it differs significantly fiom the crack width at the
steel-concrete interface. Experirnental evidence shows that the ratio of the
crack width at the concrete face to that at the steel-concrete interface
varies with the steel stress (shown in Fig. 2-7).
Mirza and Houde (1979) showed that crack formation is inherently
subjected to a greater experimental scatter than other properties of
concrete. The crack spacing can Vary between 1 to 2 times the minimum
crack spacing. If one considers the influence of the crack spacing, it is
normal to expect the crack spacing to be 50 per cent larger or smaller than
the average measured value. The crack spacing was principally govemed
by the concrete cover thickness and was approximately equal to 3c (cover
thickness).

(1) f i 3 0 0 0 psi
( 2 ) fc=6000 psi

O IO 20 30 40 50
end sips (0.0001 in.)

Fig. 2-25 Effect of concrete strength and cover thickness on slip.

-t-fs=15 ksi
+f s--30 ksi
+f s=45 ksi

O 1 2 3 4 5 6 7 8
square specimen size (in)

Fig. 2-26 Bond slip versus specmen size.


O 10 20 30 40 50
steel stress (ksi) .

Fig. 2-27 Bond slip of square specimen versus steel stress.

Mina and Houde examined the influence of bar size, concrete


strength and the thickness of the concrete cover. They showed that the slip
is seen to increase linearly with an increase in the steel stress, and to
increase with an increase in concrete strength and an increase in the
specimen dimensions up to a certain size (Figures 2-25 to 2-29). [In
Figures 2-25 and 2-27, S2*2 means that the dimension of a specimen is 2
in x 2 in]. The slip can be explained only by the interna1 cracking of the
first layer of concrete surrounding the bar and by the bending andlor
cracking of the small concrete teeth near bar lugs. For small specimens,
where the crack spacing is small, the slips are small. For large specimens,
the cracking increases with the specimen cross section up to a point where
no transverse cracks are observed. This means that over the length
provided, the force transferred to the concrete does not exceed the tensile
capacity of the concrete section.
The tests done by Mirza and Houde showed that the concrete strength
has an insignificant effect on the observed slip values. The influence of
the steel stress and the section geometry can be expressed by an equation
of the fonn:
SI ip= k 1f S k Z ( ~ ~ s ) H (2-4)
the value of the slip is in 10" inches, and the steel stress f, is in ksi.

O 1 2 3
bond slip (0.001 in)

Fig. 2-28 Bond stress-slip relationship


(x is the distance fiom loaded end, and bond stress ratio is rlf,'ln).

beam
pull

O 2 4 6 8 10 12
bond slip (0.0001 in)

Fig. 2-29 Bond stress-slip curve for beam and pull-out specimen.

The bar slip increased almost linearly with the steel stress and with the
value of the ratio (AC~AS)'" between 45 and 60.Beyond this value, the
larger concrete area imposes a greater. restraint on the steel bar, thereby
causing a decrease in the slip at the steel concrete interface.
Slip has been considered to result from a gradua1 deterioration of the .

concrete in front of the lugs as a result of high bearing and shearing


stresses. Mirza and Houde thought the slip at the steel concrete interface
can be explained only by the bending of the comb-like structures of the
first concrete layers surrounding the bar.
They concluded that the magnitude of force transferred from the steel
to the concrete is dependent on the embedment length, the concrete cross
section and the concrete strength. The bond stress at the steel- concrete
interface is seen to reach the maximum value at a slip of some value.
Before the peak value is reached, the -1ationship between the localbond
stress(r) and the local slip (A) can be expressed by the following
polynomial :
9 2
A +1.39*1012A 3 -0.33* 1015A 4
~=1.95*10~~-2.35*10 (2-5)
Differentiation of both sides of equation (2-5) with respect to A, yields:
12 2 15 3
A -1.32*10
d~dA=1..95*10~-4.70*10~~+4.17*10 A (2-6)
They found fiom tests that the bond stress level was not to be related to
the distance &om the end face before the maximum bond stress was
reached. The maximum bond stress was attained at al1 locations when the
slip value was between 0.001 inch and 0.0012 inch. The bond stress-slip
relationship is thus applicable directly at any point along the bar.
The bond stress slip behavior past the peak point was dependent on the
distance from the end face. For points at some distance fiom the end face
(3 to 4 inches or more) the bond stress was almost constant for al1 slip .

values. For smaller distances, bond stresses decreased progressively with


increasing slip. This loss of bond transfer capacity near by the end faces
can be attributed to the splitting cracks observed.
Hayashi et al (1985) investigated bond behavior near a crack. They
concluded that near cracks, both bond stifiess and maximum bond stress
deteriorate. The local maximum bond stress deteriorates near the crack in
proportion to the distance from the crack. When the distance from the
crack exceeds 4 bar diameten, no bond deterioration is observed. The
thickness of concrete cover does not effects the bond behavior as iong as
the minimum thickness is 2.5 bar diameters.
Figure 2-30 shows the bond stress distribution in the test specimen. The
maximum bond stress is about 40 kgicm2 (about 4 MPa) for al1 specimens.
Figure 2-31 shows the relationship between local bond stress and slip. It
can be seen that the relationship between local bond stress and local slip
varied with distance fiom the loaded end.
80

O S 10 15 20 25
distance from fme end (cm)

Fig. 2-30 Bond stress distribution in the specimen.


Hayashi et al proposed a bond detenoration model for finite element
method analysis of pull-out bond tests and of experiments on reinforced
concrete beams subjected to bending moment and shear force. They
proposed three rnaterial models for bond as s h o w in the following figures.

slip (mm)

a b c
Fig. 2-3 1 Bond stress and slip relationship proposed by Hayashi et al.
(stress unit: kgkm2)

Figure 2-3 l a is derived fiom the relationship between bond stress and
slip at some distance from a crack. Therefore, the bond deterioration near
the crack is not considered. The second model (2-3 1b) considers the bond
deterioration near cracks. Where the bond stiffness changes, bond
stiffnesses are multiplied by a,which is determined by the response at
the distance fiom the crack. For the third model (2-31c), the maximum
bond stress of the basic bond stress slip relationship is limited to 40
kg/cm2 (4 MPa). The relationship between a and the distance from the
crack proposed here is illustrated in Fig. 2-32. When L is larger than 4 bar
diameter, a is equal to 1.
Kankam (1997) investigated the relationship of bond stress, steel
stress and slip in reinforced concrete for mild steel, cold-worked steel and
hot-rolled steel bars.
a

Fig. 2-32 Coefficient of local maximum bond stress deterioration.


The steel strain distribution curve for the plain round bar showed
very little change in form throughout the range of loading. This implied
that the px-imary mechanism of force transfer fiom the embedded plain
round bar to the surrounding concrete, resulting in the modification of the
steel stress between the two loaded ends, remain unchanged. The
maximum point of each strain distribution curve occurred at the loaded
end of the bar for a plain round bar, indicating a general decrease in slope
towards the central anchor point. For b&h types of ribbed high yield bar, a
typical curve was largely parabolic and characterized by double
curvatures and showed large differences in the magnitude of strain
between the ends and central anchor point. These large sbain differences
in the ribbed bars showed a consistent increase with an increasing appiied
load.
Curves of the slip distribution of the embedded plain round bar and the
liigh-yield ribbed bars were approximately parabolic in the initial stages of
low applied load but later, as the load increased, the tendency was towards
a linear distribution. The magnitude of slip of the nbbed bars was less
than that of the plain round bar. The difference in the magnitude of slip
shown by the high-yield nbbed bars and the plain round bar could be
attributed to the different surface patterns of the bars and the relative
resistance developed against the movement of each reinforcing bar
relative to its surrounding concrete.

Fig. 2-33 Bond stress distribution along 25 mm plain round bar.

Fig. 2-33 shows bond stress distribution curves corresponding to a


number of load increments. It was noted that although the change in the
curves of the bond stress distribution with respect to load increments
within the service range did not confirm to any unique pattern in the plain
bars, the maximum point of each cume always occurred at, or very close
to, the loaded end and the minimum point at the central anchor point. The
magnitude and distribution of the bond stress changed only slightly as the
load increased.

Fig. 2-34 Bond stress distribution for 25 mm cold-worked ribbed bar.

Fig. 2-35 Bond stress distribution for 25 mm hot-rolled nbbed bar.

For the ribbed bars (Figures 2-34 and 2-35), as the tensile load
increased, the form of the distribution curve generally changed slightly.
Nevertheless, it was impossible to describe the curves by any simple
consistent fonn. In general, there was; with increasing load, a consistent
increase in the bond stress at almost al1 points.
2.5 -,
Slifl.04 mm Slip=.OS min

Fig. 2-36 Bond stress-steel stress relationship for plain round bar.

From Fig. 2-36, it c m seen that the bond stresses decreased linearly
with the increase of steel stress when the bond slip was constant.
Fig. 2-37 is a typical curve of the fundamental relationship of bond
stress versus bond slip for different steel stresses for the 25 mm plain
round bar. In the early stages of relatively low slip, the curve showed a
linear increase of the bond stress with the slip, but later the relationship
changed to nonlinear following large increases in the slip. As the
reinforcing bar continued to slip, causing its surface aspenties to corne in
contact with different points of the surrounding concreta, the bond stress
at any given value of the slip was less for greater values of the steel stress.
This decrease in the bond stress would be expected due to the radial
contraction of the bar and cocsequent reduction of the confining pressure
of the surrounding concrete. The following empirical equation for bond
stress-slip relationship was proposed by Kankam:
T=(~Q-kfs)~o-8 (2-7)
b,kl are constants that depend on certain parameters such as concrete
strength, bar size, surface texture of bar, and type of loading; f, is steel
stress.
The authors proposed following values for and kl :
l~=4l.7
k* =0.2

O 2 4 6
slip (0.01 ml$

Fig. 2-37. Bond stress-steel stress-slip relationship


for mild steel bar (f, is steel stress, unit: MPa).

For cold-worked and hot-rolled ribbed bars, it was found more


appropriate to examine the relationship between bond stress and slip,
since the steel stress was found to increase from the anchored midpoint to
the pulling ends at different positions along the bar. The bond stress for a
'

given arnount of slip increased in magnitude with the distance from the
loaded ends of the specimens. At each position, the bond stress increased
with local slip. Near the center of the embedded bar, the relationship wa
first linear when the local slip was small, but became nonlinear as the slip
increased. At al1 other positions, the local relationship was generally
nonlinear.

Fig. 2-38 Bond stress-steel stress-slip relationship for


cold-worked tibbed bar (f, is steel stress, unit: MPa).

O
I
O 1 2 3 4 5 6 7
slip (0.01mm)

Fig. 2-39 Bond stress-steel stress-slip relationship for


hot-rolled ribbed bar (f, is steel stress, unit: MPa).

The initial dope of the bond stress versus slip curve was diflerent for
each of the positions and, in fact, increased consistently fkom the loaded
ends to the center of the embedded bar. The authors proposed the
following equation of local bond stress and bond slip for nbbed bars:
Cold-worked ribbed bar: T = ( ~ ~ - O . S X ) A ~ - ~ (2-9)
Hot-rolled ribbed bar: T=(~S-O.~X)A~-~ (2- 1O)

O 2 4 6 8
dip (0.Olmm)

Fig. 2-40 Bond stress-slip relationship for cold-worked ribbed bar.

O 1 2 3 4 5 6 7
siip (0.01mm)

Fig. 2-4 1 Bond stress-slip relationship for hot-rolled ribbed bar.

Nikon (1968) studied the spacing, width of cracks, and the distribution
of concrete stress in partially cracked members. In spite of the
considerable scatter of experimental data, Nikon recognized a definite
trend and fitted the following equation by using the least square's
method:
6 2*+1.986*10
~ 3 . 6 0 6 ~ 1 0 ~ ~ - 5 . 3 59A 10 12 3
A (2-1 1)
where r is the nominal bond stress in psi and A is the local bond slip in
inches. Differentiation of both sides of equation (2-9) with respect to A,
yields:
12 2
dr/d~=3.606*10~-10.712*10~~+5.985*10
A (2-1 2)
which represents the stifiess of the concrete layers transfemng the forces
to the steel bar.
Later, Nikon (1972) reported results obtained using embedded gages to
measure the concrete strain distribution in tension specimens. The
concrete displacement curve was derived directly by intergrating the
concrete strains measiued at points at a distance of 0.5 inch from the bar.
The following bond stress slip relationship was proposed:
t=3 100(1.43~+1
SO)Afc (2- 13)
where the bond stress r<(1.43x+1.50)f, ;
and x is the distance fkom the loaded end in inches, and P, is the concrete
strength in psi .
The typical pull-out tests that many researchers have used are
nonuniform in the local bond stress distribution along the reinforcement.
In these tests, the actual bond stress varies significantly along the
embedment length, and results were reported as the average bond stress
versus slip measured at one end of the specimen. Abrishami and Mitchel .

(1996) proposed a new testing technique that simulates a uniform bond


stress distribution. A suitable combination of pull-out and push-in forces
can simulate a uniform bond stress distribution along the reinforcing bar
embedded in the concrete. This test method bas enabled a more accwate
detennination of the bond stress versus slip response. It simulates a
uniform bond stress by adjusting the top and bottom forces in the
embedded bar such that the strain distribution along the reinforcing bar,
measured by strain gauges, is essentially linear. This linear variation of .

strains results in approximately unifonn bond stress over the embedment


length. Figures 2-42 and 2-43 show the bond stress versus slip response
obtained fiom tests simulating a uniform bond stress distribution for
specimens failing by pull-out and splitting, respectively. They concluded
that the relationship of bond stress versus bond slip response is linear for a
brittle failure due to shearing of the concrete along lugs. On the other
hand, the relationship is nonlinear for a bond splitting failure that exhibits
a more ductile response. Afier splitting cracks form, the predicted bond
stress distribution is almost uniform and does not seem to be affected as
much by the type of specimen loading.
They proposed the analytical relationships (Fig. 2-44 and 2-45) of the .

bond stress versus slip response which capture the key behavioral features
of these two types of bond failure mechanisms, and derived analytical
solutions of both pull-out failures and splitting failures in pull-out tests,
push-in tests, and a combination of pull-out and push-in tests. The
predicted bond stress distribution is dependent on the length of
embedment, bar size, concrete properties, and the size of the specimen. As
the length of the specimen decreases, the bond stress becomes more
unifonn. Because of this, early attempts by the other researchers to
determine bond strength under nearly uniform bond stresses involved
short embedment lengths. However, these short embedment lengths gave
rise to umealistically high bond strength results. On the other hand, the .
use of a longer embedment length in a simple pull-out test gives a large
variation between the maximum and minimum bond stress. Hence, taking
the average bond strength for these specimens is not representative of the
actual bond strength.
9 1

O Y
O 1 2 3 4
Average slip (mm)

Fig. 2-42 Bond stress-slip response for pull-out failure.

O Y ,
O 1 2 3 4
Average dip (mm)

Fig. 2-43 Bond stress-slip response for splitting failure.

Hota et a1 (1997) investigated the interface properties between a straight


steel fiber and two matrix materials. They concluded that the average
interface bond strength and the average interface bond stiffbess for a
matrix material containing polymer (10 percent by weight of cernent)
were approximately twice the values measured using a mortar matrix
without polymer. The energy required for debonding and pull-out was
also approximately doubled by adding 10 percent polymer by weight of
the cernent.

awrage slip

Fig. 2-44 Analytical bond model for pull-out failure.

awrage slip

Fig. 2-45 Analytical bond model for splitting failure.

Ayyub et al (1994) studied bond srength of welded wire fabric


(WWF) by pull-out tests and considered the following factors which may
affect the bond strength: penetration and strength of welds between
transverse and longitudinal wires; size of transverse wires; number of
transverse wires, and weld strength; and use of bundled longitudinal
wires. They concluded that the pull-out capacity of the WWF depends .

mainly on the bond strength between the WWF longitudinal wires, the
anchorage effects of the transverse wires, and the shear strength of welds
between the transverse and longitudinal wires. M e r the first slippage of
the longitudinal wires, the load is resisted only by the transverse wires and
the welds between the transverse and longitudinal wires.
The contribution of the adjacent longitudinal wires to the pull-out
resistance was very small, and the majority of the pull-out specimens failed
by bond between the wires and concrete.
Some researchers (Esfahani and Orangun, 1998; Thompson, 1968)
investigated the bond stress between concrete and reinforcing bars in
splices in beams. The test parameters included concrete compressive .

strength, cover to reinforcement, spacing between bars, splice length, and


the type of steel bars. Based on the tests of 22 rectangular simply
supported beams, Esfahani and Rangan proposed the following equations
to calculate the cracking bond strength of short length specimens:
For concrete with compressive strength less than 50 MPa,
7, = 4.9[(C/d +OS)/(C/d +3 -6)fa (2- 14)
For concrete with compressive strength equal to or greater than 50 MPa,
7, = 8.6[(C/d +OS)/(C/d +5.5) f, (2- 15)
In equations (2-14) and (2- 15), C is the minimum cover of concrete.
Equation (2-15) was obtained based on results of specimens made of
concrete with compressive strength of 50 and 75 MPa, and bars with rib
face angle between 40 to 47 degrees. For bars with rib face angle between
23 to 27 degrees, tests showed that the right side of equation (2-15)
should be multiplied approximately by 0.85.
The bond strength of splices was calculated by following expression:
r, = r, [(l+l/~)/(l.85+0.024M-~)]
(0.88+0.12C,J& ) (2-16)
wliere &,, is the minimum of side cover C, , bottom cover C,, and half
spacing between bar center (Cs +d)/2 (Cs is the clear spacing of
reinforcement), Gdis the median value of &, C, and (Cs+d)/2.
M was given by following expression :
M=cosh [0.0022L* (rf, /dlo" ] (2-17)
where r is a constant which depends on the type of reinforcement.
Eligehausen, Popov and Bertero (1983) investigated the bond stress-
slip relationship by extensive experimental study. They extended the slip
range up to very large values. Various bond stress-slip diagrarns fiom
experimental results are shown in figure 2-46. A typical diagram can be
idealized as a sequence of linear segments, as shown in Figure 2-47.
Up to a certain value of stress (ri in Figure 2-47), bond is due to
chernical adhesion of the cement paste on the surface of the steel bar and
practically no slip take place; typical values of rl range from 0.5 to 1.O
MPa. For r>ri adhesion breaks down and bond is provided by fiction and
wedging action between the cernent paste and the rnicroscopic anomalies
(pitting) of the bar surface and also, in the case of deformed bars, by
mechanical interlock of the deformations and the surrounding concrete.
Due to these interlock forces, at a stress level -2 (which is a function of
the tensile strength of concrete as shown in Figure 2-47) bond cracks
form. At approximately the same time, separation of concrete f?om the
reinforcing bar takes place in the region of primary (flexural) cracks. This
separation causes an increase in the circumference of the concrete surface
previously in contact with the bar and, as a result, circumferential tensile
stresses develop. These stresses, in combination with the radial .
component of the force carried by the ribs or indentations, lead to splitting
cracks. At the stress FT~ (Figure 2-47), these cracks propagate up to the
extemal face of the member and, if there is not enough confinement ,
bond is destroyed and a splitting failure occurs (Figure 2-46 and 2-47). On
the other hand, if the presence of adequate confining reinforcement
inhibits the propagation of the splitting cracks, the bond stress can reach
substantially higher values,t( in Figure 2-47).
7 , 1

O 1 2 3 4 5 6
slip (mm)

a. Local bond stress-slip curves without confinement.

b. Local bond stress-slip curves with confinement.


Fig. 2-46 Experimentally derived local bond stress-slip curves.
slip ( mm ) 4
Ar A2 A3 4,
Fig. 2-47 Idealized local bond stress-slip curve for monotonie loading.

The slopes of the consecutive branches of the bond stress-slip diagram


gradually decrease; in other words, the value of the relative slip .

corresponding to a given increment of bond stress increases. Along the


branch defined by the stresses 73 and r, , a gradua1 deterioration of the
concrete lugs (keys) between adjacent ribs occurs until, at a value of
,,TF , tliese lugs fail in shear .
The descending branch of the bond stress-slip diagram (A>&, )
corresponds to a complete deterioration of concrete between adjacent ribs,
and for A>& , the moderate amount of residual bond stress (r4 ) is due
exclusively to fiction at the cylindncal surface defined by the tips of the
ribs. The stress 4 can remain practically constant for high values of slip,
as shown in the experimental curves of Figure 2-46. The value of the slip
& almost coincides with the spacing of the ribs, since when a nb is
displaced to the position occupied by the adjacent one when loading
started. The only remaining rnechanism of the bond transfer is fiction at
the cyl indrical failure surface.
The bond stress and slip diagram for rnonotonic loading remains
approximately the same for loading in tension, as well as in compression,
provided that an adequate degree of Confinement exists. In the case of
unconfined zones, such as the areas of a beam-column joint which lie
outside the column reinforcement, the bond stress-slip curve is different
for tension and compression. In addition, when the stress in the bar
exceeds its yield strength, the lateral contraction (for tensile loading) or
expansion (for compressive loading) of steel bars will cause a decrease or
increase, respectively, in bond strength. It has been found that this lateral
deformation cannot affect the bond strength more than 2030%, even for
very large steel strains.
Figure 2-48 shows the influence of transverse (confining) pressure,
resulting either h m compressive radial stresses (such as those acting
beam-column joints where beam bars*are anchored or pass through) or
from confinement. As can be seen in the Figure 2-48, both the maximum
bond stress and residual stress due to friction (r4) increase with the
confining pressure, but in a nonlinear fashion. These data indicate the
favorable effect of confinement with regard to bond conditions. Indeed,
the presence of confining reinforcement inhibits a premature, brittle type
of bond failure due to splitting, and in addition it increases bond strength.
O 2 4 6 8 10 12
slip (mm)

5 10
confining pressure (MPa)

Fig. 2-48 Effect of lateral pressure on local bond stress-slip relationship.

Table 2-1. Confining pressure in experiments of Figure 2-48 (a )

curve confining pressure (MPa)


1 O
There are numerical values for the various parameters of the model,
based on the tests by Eligehausen, Popov and Bertero (1983 and 1986) on
concrete specimens (fc=30 MPa) with 25 mm diarneter deformed bars,
having a rib spacing of 10.5 mm and a relative rib area of 0.66. The scatter
in the experimentally derived values of the various parameters was also
pointed out by these investigators, and as a result they decided to suggest
empirical coefficients for correcting the values given in Figure 2-47
wlienever the shape of the deformed bar, the concrete strength and the
spacing of bars are different from those used in their tests. Values of,,t
in the literature Vary from about 10 to 21 MPa, while values of A, ,
which show considerable scatter, Vary from 0.25 to 2.5 mm.
With regard to the influence of confining pressure, they suggested the
relationship shown in Figure 2-48 @), while for the effect of rate of
application of slip, shown in Figure 2-49 (a) for three different rates, the
approximation 2-49 @) was suggested. It is noted that an increase in the
slip rates of 100 times results in increases of r, and r3 about 15%.

O 2 4 6 8 10 12
dip (mm)
-2 -1 O 1 2
log of rate of slip rs

(b)
Fig. 2-49 Effect of rate of slip on the local bond stress-slip relationship.

Table 2-2. Confming pressure in experiments of Figure 2-49 (a )

0.034 0.02
2.4.2 Bond Stress and Slip Under Cyclic Loading
Bond behavior under cyclic loading is affected by the following factors:
1. Concrete compression strength.
2. Cover thickness and bar spacing.
3. Bar size (bar diameter).
4. Anchorage length.
5. Geometry of bar deformations (ribs).
6. Steel yield strength.
7. Amount and position of transverse steel.
8. Casting position and vibration.
9. Strain (or stress) range.
1 0. Type and rate of loading (strain rate).
11.Temperature.
12.Surface condition-coating.
It has been pointed out that the influence of many of foregoing
parameters on bond resistance is only qualitatively understand. Parameters
(1)-(3), (6)-(9, and (12) appear to be the ones mostly affecting bond
under monotonic loading, while (9), -(IO) and the value of maximum
imposed bond stress, in addition to previous parameters, are very
important under cyclic loading conditions.
Tassios and Koroneos (1984) investigated local bond stress-slip
relationship by an optical experimental method (the Moire method) in
rnonotonic and cyclic loading. They proved that the point of maximum
local bond stress moved towards to loaded end with an increase of load
(Fig. 2-52).
Fig. 2-5 1 Specimen tested by Tassios.

Fig. 2-52 Bond stress-slip relationship for rnonotonic loading.


(unit of fs: MPa)

Fig. 2-53 Steel stress distributions for rnonotonic loading.


(unit of fs: MPa)
O 10 20 30 40 50
local siip ( mm11000 )

Fig. 2-54 Local bond stress versus local slip curves


in rnonotonic loading.

Fig. 2-55 Bond stress distributions for various loading


levels of cyclic loading. (unit of fs: MPa)
Fig. 2-56 Bond stress distributions for various cyclic loading levels.
(unit of fs: MPa)

Fig. 2-57 Slip distribution for various cyclic loading levels.


(unit of fs: MPa)

From Fig. 2-54, it can be seen that a c o m m o n bond stress-slip curve


may be traced with a reasonable scatter of points away fiom the loaded
end (x:L<0.75). On the contrary, for points near the loaded end
(x:L)0.75), a much lower and distinctive local bond stress-slip
relationship is found which gradually tends to zero bond stress levels for
finite slip values. f, in Figures 2-52 to 2-57 is the stress in the steel bar at
the loaded point.
Under cyclic loading, bond deteriorated with increasing number of
loading cycles. Figures 2-55 to 2-57 show the distribution of bond stress
and bond slip along the steel-concrete interface for several levels of
cyclic loading. Residual steel stresses, bond stresses and irreversible slips
are observed at zero external loading. It was noted that the considerable
difference in bond stress distribution for equal external loading depends on
loading history.
The constitutive law of bond stress-slip under cyclic loading was given
experimentally by means of the Moire method in Figure 2-58.

-80 -40 -2Q O 20 40 W


slip ( 0.001 mm )

Fig. 2-58 Local bond stress-lacal slip relationship under cyclic loading .
slip

Fig. 2-59 Relationship of bond stress versus slip under cyclic loading.

Tassios et al (1979 and 1981) studied bond behavior under cyclic


loading. They developed the constitutive curve for local bond stress versus
local slip under cyclic loading shown in Fig. 2-59. Parts of OAM and ON
are the constitutive response under monotonic loading. Unloading from H
follows the bond stress path HCDE, where HC is parallel to OA; CD is at
the negative residual bond strength level -Tri and DE is along the
monotonic compression curve. Unloading fiom E follows the stress path
EFGH, where EF is parallel to OA; FG is at the positive residual bond
strength level 7,; and GH is parallel to OA and passes through the
previously most tensioned point, i.e., point B. Further loading fiom H to 1
moves along the monotonic curve, and unloading fiom 1 follows the stress
path IJKL, where KL is parallel to OA and passes through the previously
most compressed point, i.e., point E. The residual positive and negative
bond strengths were not kept constant during load cycling, but were
considered to deteriorate with increased number of loading cycles. The
authors did not give the numerical value of the residual bond strengths at
either the initial stage or during the cyclic loading.
Ismail et al (1972) studied bond deterioration under low cyclic loads
by pnsm tests and beam tests. They arrived at the following conclusions:
Firstly, the most important factor affecting stress transfer was the peak
stress reached in the preceding cycles. If the peak stress was increased, the
stress transfer at lower stresses was reduced in subsequent cycles. The
most significant reduction was at stress levels well below the peak stress.
Secondly, a small number of repetitions of load cycles with constant peak
stress produced a gradua1 deterioration in stress transfer, but the reduction
was minimal in cornparison with the reduction associated with increases
in peak stress. Thirdly, the bond stresses were higher in those specimens
in which compressive stresses were applied.
Siva Hota et al (1997) studied the bond stress-slip response between
reinforcing bars and fiber reinforced concrete (FRC) under monotonic and
cyclic loading by pull-out test. They tested the four matrix types: SIFCON
(a fiber reinforced cernent composite with a relatively large fiber volume
fraction between 5 percent to 20 percent), fiber reinforced concrete (FRC,
2 percent fibers), plain concrete (PC), and confined concrete (CC). Tests
showed that al1 SIFCON specimens failed by friction pull-out,
accompanied by large slip, while the concrete surrounding the
reinforcement remained together. On the other hand, the failure of the FRC
specimens started with some fiictional pull-out but eventually ended up
with a splitting type of failure. Al1 of the plain concrete specimens failed
by splitting, while al1 the CC specimens failed in a cone-shaped manner.
The addition of steel fibers to a cernent matrix was shown to improve the
overall bond properties of a reinforcing bar. High fiber content in the
matrix can lead to higher bond strength, higher ductility, reduced post-peak
degradation of the pull-out load versus displacement curve, increased
energy absorption, and ultimately a significantly increased safety level
against collapse and failure. The increase in compressive strength of the
matrix increased the bond strength. Since the tensile strength of concrete is
relatively low, cracking occurs early in concrete. At the onset of cracking,
the role of the steel fibers is to prevent m e r crack opening and to resist
additional tensile forces which the concrete matrix itself can not sustain.
This controls the failure in the concrete matrix itself, thereby presewing
the bond strength between the reinforcing bar and the surrounding concrete
matrix. The higher the volume fiaction of fibers, the higher the bond
strength and the higher the area under the bond stress versus slip response.
The fibers may not much delay the formation of the first crack, but they
keep crack width at small values and prevent the sudden opening of cracks.
The plain concrete specimen with no fibers had very low slip values at
maximum load, while the SIFCON specimen had values up to five times
larger. The presence of the fibers is significant in the increase pull-out
energy and ductility.
The bond stress-slip relationship depends on a number of parameters
such as the concrete compressive strength, concrete cover, anchorage
length, and the rib area at rnonotonic load and first loading cycle. Under
cyclic loading, at low bond stresses, inclined cracks propagate fiom the tip
of the ribs. The transfer of forces across the interface between concrete and
steel bar occurs and is caused by bearing and adhesion. As the loading is
increased, inclined cracks begin to form.
For unidirectional loading, the use of fibers played a major role in
slowing down degradation of the specimen's load canying capacity by
delaying the onset of cracks and slowing down the opening of existing
cracks.
For fully reversed cyclic loading, the degradation of bond strength and
bond stifhess occurs faster than for the unidirectional cyclic loading. The
degradation depends on the maximum slip value at which reversa1 of
loading occurs in either direction. The larger the slip at which reversa1
occurs, the more the damage that results in the reversed loading.
For the same number of cycles, the hlly reversed cyclic loading af5ected
the bond strength and stiffness more severely than the unidirectional cyclic
loading.
Eligehausen, Popov and Bertero (1983) studied the bond stress-slip
relationship under cyclic loading by extensive tests. They showed that the
envelope of bond stress-slip hysteresis loops for repeated loading lies very
close to the curve resulting fiom monotonic loading as shown in Figure 2-
60. It is worth pointing out that even after 20 cycles of consecutive loading
and unloading, the envelope of the .loops remains quite close to the
rnonotonic loading curve, which means that the mechanism of bond
deterioration remains the sarne as that described in the previous section.
O 1 2 3 4 5 6
slip (mm)

Fig 2-60 Local bond stress-slip relationship


for repeated loading.

slip

Fig. 2-6 1 Local bond stress-slip expec-imental curves


for cyclic loading of confined specimens by Eligehausen et al.

Figure 2-61 shows bond stress-slip hysteresis loops derived fiom cyclic
loading tests of specimens with one 25 mm deformed bar and confinement
reinforcement. It is first pointed out that, as in the case of repeated loading,
the residual slip during unloading @ranch EF in Figure 2-61 ) is quite
large, which may be attributed to the fact that the elastic part of slip
consists of the concrete deformation only, which is just a small portion of
the total slip. Whenever the sign of the bond stress reverses, the slope of
the curve remains significant up to a level of stress (ri); this increased
stifhess is due to friction between the bar and the surrounding concrete.
When the frictional resistance is overcome, the bar begins to slip in the
opposite direction (with respect to that of initial loading OAE) until the
ribs of the bar again corne into contact with the surrounding concrete (point
1in Figure 2-61). It is understood that the foregoing applies when the level
of Loading is such that concrete lugs between the adjacent ribs (see also
Figure 2-61) have been ground, thus creating gaps between the side face of
the ribs and the surrounding concrete.
Reloading in the opposite direction (branch IAIE1in figure 2-61) is now
taking place at a significantly increased slope and the path followed is
similar to that of monotonie loading. However, if the maximum previously
attained (absolutz) value of bond stress is higher than 70-80% of r, ,the
new envelope (OAIBiCl in Figure 2-61) has reduced ordinates with respect
to original one (OABC). This reduction in available bond resistance is
inore pronounced as the values of slips between which cyclic takes place
increase, and also as the number of cycles increases.
Whenever at a certain point the sign of loading changes (in a
deformation-controlled test, when the sign of the applied slip changes), the
unloading and friction branches (IKLPIiIN) are similar to the previous ones
(EFGHI). Further loading is now taking place dong a new envelope
(OA'B'C') whose ordinates are reduced with respect to the initial one
(OABC). If the level of loading is hi& enough for shear cracks to fonn in
concrete lugs between adjacent ribs, only a portion of these lugs can
contribute to the resistance of the system. Hence, the envelope OA'BfCf
has lower ordinates than the previous one (OA'iBfIC'i). Moreover, if
unloading takes place at a point along the descending branch of the bond
stress-slip curve which corresponds to a pronounced deterioration of
concrete lugs due to shear, the fictional resistance (rfu) will be higher than
previous value (rr ), since at this stage the interface between the bar and the
surrounding concrete is rougher. This characteristic may be verified if the
corresponding branches of the two loops in Figure 2-61 are cornpared.
It is seen fiom the foregoing discussion of the bond degradation
mechanism under cyclic loading that most of the damage occurs during the
first loading cycle. During subsequent cycles, a gradua1 smoothing of crack
interface occurs, causing a reduction of mechanical interlock and fiction
forces.
Eligehausen et ai (1983) studied experimentally the behavior under
cyclic loading of specimens with 25 mm diarneter deformed bars, having a
clear length of 5d and hooks at their ends. The typical local bond stress-slip
curves for a hooked bar specimen are shown in Fig. 2-62. It is clearly seen
that the available bond resistance under rnonotonic loading remains almost
constant, even for very large values of slip, in contrast to what happens in
bars without hooks. Furthemore, during successive reversed loading cycles
a significant drop of bond resistance is observed for values of the slip lower
than the previously attained peak. M e r this value is reached, bond
resistance is soon recovered, and the corresponding curve at large values of
slip lies quite close to the rnonotonic loading curve, even afier 10 loading
cycles.
Figure 2-61 shows bond stress-slip hysteresis loops derived fiom cyclic -

loading tests of specimens with one 25 mm deformed bar and confinement

-4 O 4 8 12 16 20
Slip ( mm )

Fig. 2-62 Local bond stress-slip curves


for deformed bars with hooks by experiments.
Theoretical Analysis Method
2.5 The Study of Bond Stress and S1ip.b~

2.5.1 Constitutive Equation of Bond Stress and Slip


The increment of the local slip dA within an infinitesimal bar length dx at
the location x can be defined as the difference between the bar strain E,

and the concrete strain E ~ ~ :

ciA/dx=~,-~~~ (2- 18)


The differentiation of the equation (2-18) with respect to x gives:
d2A/d2x=de, /dx-d&,/dx (2- 19)

Fig. 2-63 Free body of a reinforced concrete member.

As shown in Figure 2-63, by cutting the segment at x and taking a fiee


body, it gives :
P=Pc+Ps=AcEc~a+ASES~SX
= &E,(~,,+nps,) (2-20)
The differentiation of the equation (2-20) with respect to x gives:
&Ec (d&,/dx+npds,/dx)=O (2-2 1a)
ds,,/dx=-npds,/dx (2-2 1b)
substituting equation (2-2 1) into equation (2- 19) gives:
d2A/d2x=(l+np)d&, /du (2-22)
r,~d=(xd2/4)a, (2-23a)
r,=(d/4) a,=(Esd/4) ds- /dx (2-23b)
substituting equation (2-23) into equation (2-22) gives:
d2A/d2x=[4(i+np)/(&d)] rx (2-24)
Equation (2-24) is a differential equation of local bond slip and local
bond stress. From this equation, it can be seen that some of the factors that
influence the relationship between bond stress and slip are elastic modulus
of concrete and steel bar, percentage of reinforcement and diameter of steel
bar.

2.5.2 The Solution of the Constitutive Equation of Bond Stress and Slip
Because the measurement of local bond stress and local slip along a
stressed reinforcing bar and the surrounding concrete is difficult and very
sensitive to expenmental errors, there has been a wide variation in deduced
bond stress versus bond slip relationships derived from the results of
various experiments. Some researchers developed analytical models based
on various assumptions.
Shah et al (1 98 1) assumed bond stress distribution as:
D 2 ~ d 2 x~= e +Be-'
" +C (2-25)
where x is the distance fiom the loaded end of the specimen to any section.
From this assumption, it is obvious that the relationship of local bond
stress versus local slip is a function of the distance of the section fiom the
cracked face.
For the pull-out specimen in Figure 2-63, Shah et al solved the
differential equation (2-25) by boundary conditions:
A= ~ e +Be-'
* +cx2/2 +Dx +E (2-26)
the five constants A-E are related to cross sectional dimensions A, and 4,
total perimeter of steel bars, constitutive relationships of steel bar and
concrete, the transfer length L,and the length of segment or crack spacing.
The above assumption rnakes the location of the peak bond stress close
to the center of the transfer length. Test results show that peak bond stress
occurs at the section closer to crack face than to the other end of transfer
length. The relationship between bond stress and local slip is not unique
but varies from location to location. Based on these test results, Yang and
Chen (1988) assumed the following relationship for bond stress versus
slip:
T(X) = KA(x) +(ex2 +D) +Ecos(x/~*x&) (2-27)
where K is the bond constant which is the bond stress per unit slip; C, D,
and E are constants to be determined by boundary conditions; L, is the
transfer length, and x is the distance from the loaded end of the specimen
or the cracking spacing to any section.
Yang et al employed the boundary conditions to determined the
theoretical solution of equation (2-24) as following:
~ (c
A(y)=ZA{ch(cy) - ( ~ ~ )ch / 2) +p2[ch(c )-11 [l -cos(yd2)] -1 )
(2-28)
t(y)=ZAK {ch(cy) -ch(c )+[ch(c ) -1] cos(yd2)) (2-29)
where y=x/Lb (2-30a)
a 4 ( l+np)l(dE,) (2-30b)
c=(a~)'.'~, (2-3Oc)
p=(2/x)c (2-30d)
ZA={sh(c) -ch(c) +p[ch(c) -11) QU (2-30e)
eo is the steel strain at loaded end of the specimen or cracking face of the
actual structure.
Fig. 2-64 Slip distribution along the bar.

Fig. 2-65 Bond stress distribution along the bar.

Fig. 2-64 and 2-65 show the curves o f A and T of the Yang and Chen
theoretical solution, respectively, with a L t =0.3 mm and c=2,3,4. It can
be seen that the peak values always occur near y=0.7 and move slightly
toward y=l with very weak dependence to an increasing c. Figure 2-66
shows the cornparison of solution of Yang and Chen with the solution of
the Somayaji and Shah method.
Fig. 2-66 Cornparison of bond stress distribution between Yang and Samayaji.

The authors compared the analytical results with those measured in


experiment.
Hamayoun et al (1996) developed an analytical method to predict the
ultimate tensile strength of fiber reinforced concrete when the failure is
govemed by the strength of the fiber concrete interface, and to account for
the interfacial slip between the concrete and the fiber. The model of
analysis is based on the concept of fracture mechanics with the presence of
a traction-free interfacial crack between the fiber and concrete. The model
is developed based on the assumption that both the fiber and the concrete
behave elastically.
For elastic fibers with L/r exceeding 100, the interface crack and the
fiber lengths have no major effect on the pull-out strain energy rate (L is
fiber length, and r is fiber radius); the interface crack becomes significant
if the fiber is ngid (when L/r is equal to or less than 25); and increasing the
fiber spacing increases the strain energy release rate.
2.6 Code Development

The AC1 Building Code (1983) bond strength of reinforcing bars of


No.1 l (36 mm) or smaller sizes in tension is given as:
~ , ~ ~ = ( f ~ &f;'40) [. 5 / ( ~ .Abfy)
~4 PSI . (2-3 1a)
where 0.04 Abf?=0.004df?=12 in (305 mm). (2-3 1b)
The corresponding equation proposed by AC1 cornmittee 408 is
500&)(60,000/fy)
~~~~=(fsd/4)[(Kf~"-~)/(5 psi (2-32a)
in (305 mm)
where (5500Ab )/(~f,'~~)*(60,000/f,)>=12 (2-32b)
K=(C+Kw)<=3d (2-32~)
KpAtrfY/(1500S)<=d (2-32d)
S(stimip spacing)<=Ld/2 (2-32e)
Mina (1987) investigated the variability of bond strength of bottom-
tension reinforcement in concrete beams. He used the Monte Carlo
technique to generate the variability of the ultimate strength of bond
between concrete and reinforcement zind to estimate statistical properties
of the bond strength by analyzing the simulated sample by considering
.
following parameters : ratio C/d , where C is the smaller of the bottom or
side cover to the center of flexural tension reinforcement, and d is the
diameter of longitudinal bar; ratio of transverse reinforcement A&f/sd ;
concrete strength; and grade of flexural steel. The author proposed the
following equation for calculating bond strength :
rh=(fyid/4){ [ 1.2+3C/d+(Ayfyt)/(500sd)]/[d(fyI/(.625fspr)O)]} psi
(2-33a)
where (AUf,J/(500sd)~=3, (2-33b)
C/d<=2.5 (2-33d)
Cs/(Cbd)<=3 (2-33e)
fYland fF are the yield stresses of the longitudinal and transversal
reinforcement, respectively, and f, is the splitting strength of concrete.
Due to the nonuniformity in the bond stress distribution, no theoretical
method currently exists for evaluating bond strength of a reinforcing bar
embedded in concrete. Hence an "average bond stress" or "average bond
strength" is used in codes of practice. Several studies have show that
there is a significant variation of the actual bond stress distribution, with
the maximum bond stress in some cases being much greater than the
average bond stress. In addition, it has been shown that the bond stress
distribution varies greatly as slip develops.
Mirza estimated the ultimate bond strength for varied material strengths
and steel properties by a Monte Car10 analysis, and compared the
theoretical strengths with the results fiom AC1 408 method (equation 2-32)
and AC1 Building Code method (equation 2-3 1). It was found that AC1
408 method produced bond strengths having less variation with the
theoretical value than did the AC1 building Code method.

2.7 The Analysis of Bond Stress and Slip by Finite Element Method
Ngo and Scordelis (1967) were the fust to use the finite element
method to analyze reinforced concrete structure accounting for bond-slip
effects using a linear relationship between bond stress and bond slip. Later
in 1968, Nilson studied the nonlinear analysis of reinforced concrete by
finite element method. Nilson considered nonlinear bond stress-slip
relationships, nonlinear material properties, and the influence of
progressive cracking. A third degree polynomial was employed as the
relationship of bond stress versus slip. That is:
6 2
s=3606*103~-5356*10A +1986*109A3 (2-34)
Shava et al (1989) developed a elastic-plastic cracking constitutive
model for the analysis of reinforced concrete by considering the
nonlinearities due to tensile cracking, aggregate interlock, plasticity in
compression, yielding of steel, bond stress and slip, and tension stiffening.
Mehlhom et al (1985) developed one- and two-dimensional
isopararnetric contact elements and used these elements to analyze the
reinforced structures.
Shiro and Morita (1987) thought that the bond model used in linkage
elements requires quite cumbersome calculations to determine the global
behavior and did not include the mechanism of stress transfer by wedging
action. They assumed that the concrete deformation due to interna1
secondary cracking occurs at the interface and it was involved as a part of
the local slip of the bond model. They proposed a combined link element
which consists of the slip link element ( linkage element) and wedge link
element to account for wedging action.
Keuser et al (1987) investigated the properties of the elements and the
quality of the results which are affected by the displacement function of
the elements, the density of the element mesh, and bond stress-slip
relat ionship.
The quality of the results depends strongly on the design of the element
mesh and the stiffnesses of individual elements. For the evaluation of an
element stiffness matrix, two basic assumptions are made; one conceming
the displacement function and other regarding material behavior.
Differences between an exact solution and results of an finite element
analysis are caused by the fact that finite elements actually model only part
of the solution with the same or a lower order than the displacement
function used for the finite elements. Higher order effects are only
approximated.
The influence of the displacement function is investigated by an energy
consideration.
For a loading structure, the intemal energy in the contact area of steel
and concrete is:

In finite element analysis, the interna1 energy is:

The energy values for elements with different displacement functions


and for various numbers of degrees of fieedom were calculated using
equation (2-36) . These results were compared to theoretical values which
were obtained by equation (2-35).
The authors studied the results of five prescribed linear or cuwed
functions and concluded that:
1. The bond linkage element with its constant displacement function is
not very well suited to mode1 non-constant slip curves. The
approximation of antisymmetric portions is especially poor. The
results can be improved by increasing the number of degrees of
fieedom, but the convergence is still rather slow because the bond
stress and slip curves are approximated by stepwise functions.
2. The contact element with a linear displacement function models
linear slip curves exactly and even gives good approximations of the
nonlinear slip curves with only a few elements.
3. Results for the contact element can be improved by use of quadratic
or higher order displacement functions, but the difference is much
smaller than between the use of a constant displacement function
and a linear one.
Allwood and Bajman (1996) also modeled nonlinear bond stress-slip
behavior for the finite element method. They ignored the adhesion effect,
and assumed that the contribution of friction and interlock is proportional
to radial pressure. The radial pressure between bar and concrete cornes
fiom the total effect of:
1. Pressure on the bar generated by the shrinkage of concrete during
setting.
2. Changes in diameter of the bar due to Poisson's ratio as axial stresses '

are developed in a bar.


3. Lateral confming stress in the surrounding concrete.
They developed a new approach for bond stress and slip analysis within
the finite element method. The b a i s of the method is to analyze the
concrete and steel separately and then to bring the two solutions together
by a rapidly converging iterative process that adjusts the bond stresses
linking the two components together.
Pochanart and Harrnon (1989) studied the bond stress-slip model
including fatigue under cyclic loading. By the load-controlled test and slip-
controlled test, they proposed an analytical model of bond stress-slip for
rnonotonic load and cyclic load, including a reduced bond stress-slip.
envelope .
CHAPTER 3
THE MODELING OF BOND STRESS-SLIP
BY FINITE ELEMENT METWOD

3.1 htrodunt'ion
Accurate prediction of the nonlinear response of reinforced concrete to
loads using the finite element method is dependent on the knowledge of
several complex phenomena such as the behavior of concrete and steel,
relationship of bond stress versus slip at the steel-concrete interface,
aggregate interlock at the cracks and time-dependent phenomena such as
creep and shrllikage.
There are two basic approaches on which the bond stress-slip modcl are
applied: (1) rnicroscopic analysis of the stress state in the neighborhood of
the reinforcement (Gajer and Dux in 1990, Ingraffea et al in 1984), and (2)
macroscopic analysis of the global behavior of a member or a structural
assemblage (Ngo and Scordels, 1967; Nilson, 1968; Keuser and Mehlhorn,
1987).
In rnicroscopic analysis, the stress transfer mechanism by bond should
be treated as the local contact in front of bar ribs, not as the one
dimensional bond-slip at the interface. The interface slip is due to cmshing
of concrete within the limited zone in fiont of ribs, and can be derived fiom
material properties and the configuration of the bar. It is essential to mode1
appropriately the radial action of bond, as well as shear transfer parallel to
the axis of the bar. Stress States in the concrete, especially in the
neighborhood of the reinforcement, should be evaluated accurately, and
initiation and propagation of interna1 cracks, both longitudinal splitting and
cone-shaped cracks, should be followed analytically.
Thus, the local bond stress versus the local slip relationship cm be
denved as the result of the analysis. Therefore, the bond stress-slip
relationship depends on the local stress state and in general varies from
location to location.
In macroscopic analysis, the bond stress versus the slip relationship may
be directly modeled by determining the constitutive law of a one- or two-
dimensional bond element using experimental data or the results of
microscopie analysis. The stress state of the concrete near the bar is not
more realistic, but it is a fictitious one to sirnplifi the behavior.
Reinforcement can be introduced in a finite element mesh in three ways.
In the first method, steel is represented as discrete bar or beam elements
connecting the nodes of the finite element mesh. In the second method (the
embedded steel element), steel is placed anywhere in the finite element
mesh by embedding the steel bar element within the concrete element and
enforcing displacement compatibility through interpolation and
transformation. In the third method (the smeared steel element), a
composite steel concrete material matrix is employed, and this requires a
minimum number of elements and nodes. Third method can be employed
advantageously in structures where steel is distributed throughout. The
drawback of the latter two representations is that bond stress and slip can
not be modeled. The discrete bar elernents can mode1 bond stress and slip
through the provision of linkage elements or special bond-slip elements.
The stress strain curve usually employed for steel is either elastic-perfectly
plastic or elastic-hardening .
The nonlinearity due to bond stress-slip is usually modeled either by
using linkage elements or by special contact elernents. The parameters for
these elements are obtained fiom experimental bond stress-slip
relationships.
Tension stiffening of concrete can be modeled in two ways. In the fust
method, it is assumed that the tensile stress in concrete reduces gradua-lly
to zero &er tensile cracking. In the second method, a modified stress
strain curve for concrete is used. The dowel action of steel can be modeled
by using a normal stifhess for bond-slip elements. Aggregate interlock is
usually modeled by assuming a positive shear modulus after cracking.
In the finite element method, the elements are connected to each other
at the nodal points. The adjacent elements have identical displacements at
the common nodes; referred to as the displacement compatibility. If the
bond slip between concrete and reinforcement is taken into account, the
condition of displacement compatibility will not be satisfied. Special
interface elements will have to be used in conjunction with the discrete
concrete elements and reinforcement elements, while constitutive laws will
be required to mode1 bond stress and slip between these two components.
In finite element analysis, the linkage element, developed by Ngo and
Scordelis in 1967, has been used most commonly for modeling bond-slip
behavior. This element connects one node of a concrete element with one
node of an adjacent bar element. The linkage element has no physical
dimensions, so two connected nodes have identical coordinates before slip
occurs in the interface between steel bar and concrete. For plane stress
problems, a linkage element consists of two springs, one parallel and the
another one normal to longitudinal axis of the reinforcing bar.
Compared with the linkage element, the contact element (also called
bond-zone element), which was developed by de Groot et al in 1981, is
completely different. The most important differences are that contact
element has the dimension along the steel-concrete interface (it does not
have physical dimension in other two directions) and it provides a
continuous contact surface between - steel bar and concrete. The contact
surface between the steel bar and the concrete in the immediate vicinity of
the steel bar is modeled by a bond stress-slip law which considers the
special properties of the bond zone.

3.2 Contact Element


The bond between steel and concrete is modeled by a double node contact
element. The order and type of the contact elements should be compatible
with the order and type of the steel bar and concrete elements. For example,
if the concrete element is a two dimensional isoparametric element with nine
nodes and the steel bar element is a one-dimensional isoparametric element
with three nodes, a one dimensional parametic contact element with three
double nodes is required to connect it to the concrete and steel elements.
in the contact interface, the two elements connected by a contact
element have independent element nodes, and double nodes of the contact
element also have independent element nodes. One group of nodes of the
contact element is connected to the steel bar element, and another group of
nodes of the contact element is comected to the concrete element. In the
unloaded state, these double nodes have identicai coordinates. In the loaded
state, there are displacements between the double nodes so the double
nodes have the different final coordinates.
In this thesis, the contact element considered is the two double-noded
element with a linear displacement function for the plane problems. Two
nodes of a concrete element are connected to the two nodes of a steel tniss
bar element by the two double nodes of a contact element.
It is assumed that the relative displacernent of any point in a contact
elernent after loading in local coordinate system is :

matrix form

The relationship between nodal relative displacements A and nodal


displacements { 6 } is :

deformed
element
undefked i
element .

Fig.3-1 Relative displacements A.


The joint coordinates in the local coordinate system to element k are:
Double nodes local coordinate r
i and j .O

mandn lk

The relative displacements of two double nodes i ,j and m, n are:


7 -

Substituting Eq. 3-4 into Eq. 3-2, the following relationship between the
displacements of any point and the relative displacements of two double
nodes in a contact element is got:
{f) =N(a)=PJl [Al -' {A} (3-5)
Substituting Eq. 3-2 into Eq. 3-5, the relationship between the
displacements of any point and the displacements of two double nodes in a
contact element is obtained:
{f) =N[AI -' ICI( 8 1 (3-6)
In the local coordinate system, the relationship between contact stress
and relative displacements in a contact element is

(3-7)
where [G] is the constitutive law matrix between contact stresses and
relative displacements in the contact interface in the local coordinate system.
($1 is the initial stress vector in a contact interface in the local coordinate
system.
I f one considen the special case in which the contact behavior in the two
coordinate directions is independent of each other, the constitutive law
matrix will be
-
O

Gtt-

Thus the strain energy per unit volume is :


U=% (oT{O )=% ( f )
([G]{f )+{ao))
The potential energy expression of element k is :
nplc=u-(p)rnfcT{8)l~=xI
( { ~ } ~ [ ~ ] { f } l c +{f
V O ~ } k T {go) k)d~tc-(p)rnk~
(5) ~t

(3-9)
Substituting Eq. (3-6) into Eq. (3-9), one obtains:
n P= j W~ r r 6 1 * [cT([AI- ' ) T ~ T [ ~ -l~ [CI
~ ~( 6[) ~ ~

+@) k T r ~ ~ T ( [ ~ ~ k -1ld) dvk


T ~4p)&1
T{~G IC (3- 1O)

The sum of the potential energy for z structure with n contact elements is :

+ 2 (6)
1=1
1
i T [ ~ ] Tvol
T O
([A] i 'OTIN] {a ) i dvi 2
r=l
(O) i T{~}m
i

+ 2 {d)iT[clT Id
I=I
T O
([A] i-')T[~]{a ) i dvi- 2
i=l
(d) i T ( p ) i
wliere {p), i and {p) i are the nodal load vectors for the contact element i in
the local and global coordinate system. (P) is the nodal loading vector of
the whole structure in global coordinate system. {dli is the nodal
displacement vector of the element in the global coordinate system.

(d) i =

and {D) is the nodal displacement vector of the whole structure in the
global coordinate system.
-
Dl
D2
.....
{Dl= DI
Di+1
......
Dm1
Dn
-
NOW let [BIi = N [A], [Cl
Then @3]iT=[clT([~]
i )
-1 T
mT
The element stifhess matrix of the element K in the local coordin

The initial load vector is


{r) *=I, {~o~Icd~k
~ ] k =

The load vector is :


W k = w mk + { ~ } m k

Then the element stifiess equation of element k in the local coordinate


system is:
[SI mi^ { 8)L= (RIL: (3- 19)
Equation (3-1 1) becomes
we get the structural s t i f i e s s equation as following from (3-1 8):

[SI i and [SI are the element stifiess matrix o f the element i and the
stiffness matrix o f the whole structure in the global coordinate system.
From above equation, it c m be shown that:
3.3 Linkage Element
A linkage element is required at each node to connect between steel bar
elements and concrete elements. The element has no physical dimensions.
It connects two nodes with identical coordinates and can be conceptually
thought of as consisting of two linear springs. One is parallel to the steel
bar axis and another is normal to bar axis. Both nodes occupy the sarne
coordinate in space before loading, but they undergo relative displacement
by the deformation of the linkage springs, resulting in different coordinates
after loading. The component of linkage force in the direction of the bar
axis gives the bond force, and the normal component of linkage force gives
the radial splitting force.

concrete

concrete
I
reinforcing bar
Fig. 3-2 Linkage element.
The relative displacements at nodes i j are :

The radial and tangential stresses are given by:

or fd=[Kl{ A1 (3-33)
Thus the strain energy per unit volume is :
u=%{ G } ~{A )=i/2{s)T[c]T[~~
[CI {) (3 -34)
The potential energy expression of element k is :
npk=u-(p}mkT (6 }k=x{}? [c]~[K][cI (6)t -{p)mkT ( 6 )k (3-3 5 )

The sum of the potential energy for a structure with n linkage elements is:
where {dli and {D) are the nodal displacement vectors of element i in the
local coordinate system, and of the overall structure in the global system,
respectively. {ph) and (pli are the nodal loading vectors of linkage
element i in the local coordinate system and global coordinate system,
respectively.
Letting XiddDi =O for i=1,2 .... n.
One obtains the stifbess matrix of element k in the local coordinate
system as following:

and the stifbess matrix of element k in the global system is:


[S]k=[TIk [clT[K][CI[ ~ ] k = [ ~ ] t [ ~ ] r n [ ~ ] k (3-38)
and the stifiess equation of element k is :

The stiffiless matrix of the whole structure is:

The stiffhess equation of the whole stnicture is:


[SI {DJ=fFf (3-41)
Substituting the matrices [KI, [Cl and [Tl (see section 3.5
Transformation Mamx) into above equations, we get:

[SI, =
O -Kt O
3.4 Temperature Changes, Prestrains and Support Displacements
The effects of temperature changes, prestrains and support
displacements are considered as following:
Firstly, one calculates the strains (initial strains) of the concrete element
and the steel element at the interface due to temperature changes,
prestrains and support displacements. Then one calculates the nodal initial
dis?lacements due to the initial strains. Assuming the initial nodal
displacements are (Do), the stifiess equation of the whole structure can
3.5 Transformation Matrix
As shown in Figure 3-2, the relationship between the displacements
{6Ic the nodal forces (RIk in local coordinate system, and between the
displacements (dlr and the nodal forces {FIi, in global coordinate system ,
are :
{6)k=[T]k {d)k (3-47a)
{R)k =[T]k {F)k (3-47b)
so Tm1,
{~I~=[TI~ (3-48a)
and {F)l~=[T]i, {R)k (348b)
where [Tlk is the transformation matrix of elemtnt k.

I
Fig. 3-3 Relation of local coordinate
and global coordinate system.
3.6 Material Properties
The Modified Compression Field Theory (MCFT) was proposed by
Vecchio and Collins (1986) for predicting stresses, strains and deformation
response of reinforced concrete subjected to in-plane shear and normal
stresses. The theory presented new constitutive laws for modeling the
material response of concrete and steel reinforcement, based on the results
of 30 reinforced concrete panels tested in pure shear or in combinations of
shear and normal loads. The theory has yielded excellent agreement with
many experimental results tested by many researchers. Vecchio (1989)
developed a nonlinear finite element program (TRIX) for plane stress
analysis of reinforced concrete membranes according to the formulations
of the MCFT. In the program, cracked reinforced concrete was treated as
an orthotropic material using a smeared, rotating crack modeling approach.
The solution procedure used was based on a secant-stifThess formulation,
giving good numerical stability. Vecchio and other researchers improved
the mode1 of material properties after that and produced the 1999 version
TRIX99. In tliis thesis, the bond element was implemented into the
TRIX99 nonlinear finite element program for considering the slip between
steet reinforcement and concrete.

3.6.1 Concrete Properties


For cracked concrete subjected to a tension-compression stress state,
the constitutive law adopted was according to the MCFT. The strength
reduction factor B for concrete in compression, illustrated in Fig. 3-4, is
p= ll(0.85- 0 . 2 7 ~ ~<=1
1 ~ .O~ ) (3-50)
where E, and ~2 are the strains in the principal tensile and principal
compressive directions, respectively. Thus, the stress-strain relationship of
concrete in compression, illustrated in Fig. 3-5, is
fc2= fp )- (&dgpl2 1
[2 (&hp for O> EZ >sp (3-51)
f c 2 = f p [ ~ - { ( & 2 - ~ p ) / ( 2 ~ ~ - &for
p ) }E2~] > E ~ > ~ E ; (3-52)
where E, ' is peak compressive strain of uncracked concrete.
B

Fig. 3-4 Compressive softening parameter P of concrete.


G2
ognestad
.- - -...-..- - ..
.- .-- . .*.
- 0 . .
O
8- o.. Parabola

Fig 3-5 Constitutive response of cracked concrete in compression.

The peak compressive stress of the cracked concrete, f,, is:


fp= p f c f (3-53)
occurring at the peak strain
gp=PE
: (3-54)
The stress-strain relationship for concrete in tension, s h o w in Fig. 3-6,
is given by
fcl =Em C I for 0-1 (3-55)
fcl= f , '/[li- ( ~ o o E ~] ) ~ - for
~ EI (3-56)
For concrete in a biaxial compression state, strength enhancement was
modeled using a relationship approximating the Kupfer et a1 (1969) model.
The strength enhancement factor for concrete in the 3-direction, due to the
stress fci acting in the 1- direction, is
Kc2 = 1 +O .92(fcl/fi)- 0.76(f~~/f')~ (3-57)
Cut-off due to local
stress conditions at
crack

El

Fig 3-6 Constitutive response of cracked concrete in tension.

The peak stress, f and strain at peak stress, EN, are


f p2 = Ks2 f: (3-58)
Ep2 = I(c2 f: (3-59)
The strength enhancement factor for concrete in the 1-direction, due to the
stress f,- acting in the 3- direction, is
- 0.76(fc2/fc')2
Kcl = 1 +O .92(fC2/fcr) (3-60)
The peak stress, fPi,and strain at peak stress, are
fpi= Ki fc' (3-6 1 )
pl = LIfc' (3-62)
Fig. 3-7 Strength enhancement due to biaxial compression.

Eo Ep 2%

Fig 3-8 Stress-strain relation for biaxially compressed concrete.

The constitutive law for biaxially compressed concrete is based on the


mode1 of Kent and Park (1971), which was modified later by Scott et al.
j1982), shown in Fig. 3-8. The formulation of constitutive law for biaxial
concrete is
f=z= fp2 [2 (E&Z )- ( ~ 2 / & ~l22 ] for O> 2 > EN (3 -63)
fc2=fp2[1 + &(tz2 - zp2)] <=O. 2f p2 for EZ < (3-64)
where &=O S / [ (3+0.29fc')/(145f,' -1000) ] + fc1/450+ (365)
The stress-strain relation for concrete confined by rectangular hoops is :
fc= fp [2 ( d s p)- ( d & pl2 1 for O> E, > sp (3-66)
f,= fp [1 + &(gC - gp )] <=O.2f p for cc < E, (3-67)
where
Kc = 1+QSffi/f,' (3-68)
The peak stress f, and strain at peak stress %are
f p =&fC' (3-69)
sp=I(c f l (3-70)
&=0.5/[ (3+0.29f,' )/(145f,' - 1OOO)+ O . 7 5 * ~ , * ( h ' / ~ t , ) ~ - ~ - (3-7 1)
where fyhis the yield strength of hoop reinforcement (MPa), Qs is the ratio
of volume of hoop reinforcement to volume of concrete core measured to
outside of the hoops, h' is the width of concrete core measured to the
outside of the peripheral hoop (mm), and sh is the center-to-center spacing
of hoop sets (mm) .
Concrete in compression exhibits a lateral expansion charactenzed by a
progressively increasing Poisson ratio. The mode1 used for the Poisson
effect, which was proposed by Vecchio [see Fig. 3-91, is
~ 1=
2 Vo for O> ~2 >6 2 (3-72)
~ 1 2 = v g [ 1 + 1 . 5 ( 2 ~ ~ / ~ - 1for
)~] ~ 2 ~ 0 . 5 ~ ~ (3-73)
v21 = vo for O> >0 . 5 ~ ~ (3-74)
v2l = vo [1+ 1.5(2~,
/ cp-112] for <0 . 5 ~ ~ (3-7 5)
where both viz and vzl do not exceed 0.5; vo is the initial value of the
Poisson ratio; Vij is the expansion in the i-direction due to the concrete
stress in the j-direction, fcj.For concrete in tension, pnor to cracking, the
Poisson ratio was considered constant at vo. M e r cracking, the Poisson
ratio was equated to zero for expansion normal to the tensile direction
only. For example, if 1-direction is the tensile direction, then v2,=0, v+O.

Fig. 3-9 Poisson's ratio of concrete in compression.

( a ) Compression response.
( b ) Tension response.
Fig. 3-10 Hysteresis model for concrete.

For the stress-strain relationship under cyclic loading, a plastic offset


formulation was used [Vecchio 19991.The hysteresis model for concrete is
shown in Fig. 3-10.
On a reloading cycle where the concrete plastic strain in effect is E$,

the concrete compressive stress is calculated as:


fc(&)=0 for E, >ccP or E= >O (3-76a)
fc(&c)= ( 8 c - ~ c P ) f m / ( & m - & ~ for EcP> Ec >Ecm (3-76b)
f, (& )=fb& ) for E, CE, (3-76c)
where E, is the maximum compression strain attained during previous
loading, f, is the stress corresponding to E
, , and fk(cC) is the stress
calculated from the base curve for a strain cc. If E, < E ~then
, E, and fc,
are updated to E
', and fm' respectively.
At each load stage, the instantaneous plastic strain gCP'is calculated as
follows:
scP'= E , - E ~ [0.87(cj~J- 0.29 ( d c J 2 ] Q 1 . 5 ~ ~ (3-77a)
ccP' = cc- 0.00 1305(&,/0.002) cc< 1 . 5 ~ ~ (3-77b)
where cpis the strain corresponding to the peak stress in the base curve. If
the instantaneous plastic strain ccP'exceeds the plastic offset G ~ then
, the
latter is updated accordingly.
On a reloading cycle, when the active plastic strain is E?, the concrete
tensile stress is calculated by
( ~<~zC-< tzEI.
fc(ec)=( E ~ - E ~ ~ ) ~for~ tzCP E ~ ~ (3-78a)
fc (CC
)=fbt(sc ) for ec >cm (3-78b)
where cm is the maximum tensile strain attained during previous loading,
f, is is the stress corresponding to cm ,and fb<(ec
) is the stress calculated
fiom the base curve for a strain .
On the unloading cycle i where the concrete plastic strain in effect is E?',
the concrete compressive stress is calculated as:
N=Ec (ccp'- E ~ * ) / [ ~ ~ ~ cm3)]
+E~(E~- (3-79a)
For WN-30:
fco=EC(~, - - Bcm*lN-'
fcl=fms+Ec(~cemY)+fd
For N<=l and N>=20:
fc l = E c ( ~ ~ES')
For any value of N:
fc2=fci-i+Ec(~c- &ci-1)
fci=the most negative value ( fcl,f ; )
On the unloading cycle i where the concrete plastic strain in effect is G~',
the concrete tensile stress is calculated as:
N=Ec (smY- -
- E ~ * ' ) fa*]
Ec(~tmw (3-8 1a)
For 1<N<20:
-
f c ~ = E & ~E~)~/N(&~~
~ - 4p' )N-1 (3-8 1b)
- &=)+fd
7

fci=fb*-EC(ct, (3-8 1C)


For N<=l and N>=20:
-
fcl=Ec(~cacp') (3-8 1d)
For any value of N:
fd=fci- 1+E,(E, - E ~,)- (3-8 1e)
fci=the maximum value ( cl,G) (3-82)

3.6-2 Reinforcement Properties


The stress-strain relationship of steel bars is the same for tension and
compression. The rnonotonic stress-strain curve for reinforcement consists
of three regions: the linear region, the yieid plateau, and the strain
hardening region. A trilinear curve was used for modeling stress-strain
relationship of reinforcement as shown in Fig. 3-1 1.
fs=EreS for %ccy (3-83)
fs=fy for (3-84)
fs=f, +ESh(&.- E& )<fu for es>&h (3-8 5)
where f, is the yield stress of reinforcement; Esis the elastic modulus of
reinforceinent; Es is the strain-hardening modulus; and E* is the strain at
the commencement of strain hardening.
Fig 3 - 11 Stress-strain relationship of reinforcement.

The reloading and hysteretic response of the reinforcement is modeled


after Seckin (1971) with some minor simplifications. The first loading
circle is the rnonotonic response. The unloading relationship is
f s ( ~ i )=f s i-I+w~i
- ci-,) (3-86)
where E, is the unloading modulus.
E, = Es for - (gm O go)

E,=Es[l.OS-O.OS(&m-~)/&y] for E y < ( ~ m ~ ) < 4 ~ y

E, = 0.85Es foi ( E ~ - E ~ ) > ~ E ~

The stresses upon reloading are


&(&i )=Er( ci- EO ) +( Em -ET ) ( ci- EO ) N / ~ (&mg EO ) N-l]
where
N=( E, -5) ( Em- QI )/[fk EX Ei- &O )]
Fig. 3-12. Hysteresis mode1 for reinforcement.

In a positive cycle, cm is the maximum positive strain attained during


previous cycles, f, is the stress corresponding to E, as determined from the
hysteresis curve, and & is the tangent stiffness at E, . The parameter is
the plastic offset strain corresponding to the zero stress point for the
present cycle; it is redefined whenever the stress passes through zero.
In a negative cycle, the sarne formulations apply except that E, is the
maximum negative strain previously attained. The stress f, and stifhess
E, are evaluated accordingly.

3.7 Bond Stress-Slip Models for Finite Element Analysis

3 -7.1 Models of Bond-Slip Element


Perfect bond was assumed in lower strain regions. Bond slip was
considered in high strain regions. For the plane stress problem, the
concrete element used was either a rectangular or a triangular constant
strain element, and the steel element was a tmss bar element. For the
reinforced concrete shear wdls which were analyzed in this paper, in the
top beam and top part of the web, the slips were very small so the smeared
matrix constant strain rectangular element with four nodes was adopted.
High strains and large slips occurred in the bottom part of the web, so
discrete steel truss elements and concrete elements as well as bond-slip
elernents in the interface between steel and concrete were used. In the part
of the bottom beam which was comected with the web, bond slip was
perhaps significant in the stages of large displacements, so bond-slip
elements were employed in this part of the bottom beam.
An incremental load method was adopted for the nonlinear analysis
procedure. An initially uncracked element is loaded incrementally until the
principal tensile stress exceeds the tensile strength of the concrete at one or
more locations. At each loading step, concrete cracking and its propagation
are checked, then material properties and stifiess coefficients of each
element are calculated according to the current stress-strain states. If the
average value of the principal tensile stress in a element exceeds the tensile
strength, then a crack is established in this element, and the concrete
element stifhess of the element in the crack direction is modified
according to the stress-strain curve of the concrete in tension.

3 -7-2 Models of Local Bond Behavior

3.7.2.1 Mehlhorn Model (Model 1)


This mode1 was proposed by Mehlhom et al. The relationship of local
bond stress and local slip is s h o w in Figure 3-13 . The cwves are defined
as :

D
Fig. 3- 1 3 Relationship of local bond stress and slip.
(AB and CD are unloading and reloading curve)

At a particular stage of monotonic loading, the bond stress is calculated as


s =p fct[ 5 NA^) - 4.5 1 d A / A , ~+1.4 ( A / A ~ )]~ for 1 A/ <Ai
(3-90a)
r =1.9 P fct for AU> A>=Al (3-90b)
r -1.9 P fct for -Au >A<=Al (3-90c)
r =O for 1 >=A~ (3-90d)
P= Pi f32 P3 P 4 (3-90e)
In the case of cyclic loading, the first loading cycle is the rnonotonic
response. During an unloading and reloading cycle, the bond stress is
calculated as :
t =O for O<A< A~ or Ap <A<O (3-9 1a)
r = ( A - A ~T&(A,+-A~
) for Ape cA<Amc (3-9 1 b)
r =(A-Ap) qJ(A,,,--Ap) for Am-<A < Ap (3-9 1c)
r =f(A) for A >= Ae or A <= Am- (3-9 1d)
where fci is the concrete tensile strength; f(A) is the bond stress-slip curve
of the rnonotonic loading given in Eq. 3-78; & and A,,. are the maximum
and minimum slips attained during al1 previous loading stages d o n g the
concrete-steel interface, respectively; and r ,and r ,
. are the bond
stresses corresponding to bond slips A+ and &. , respectively . APCand
A~ are the plastic slips at the bond slips A= A,+ and A= &- respectively.
They are calculated as
/Em
A~ =Am+-im+ (3-92a)
-
Ap =Am- rm-/EbO (3092b)
where EbOis the inclination slope of the unloading or reloading curve,
equal to the inclination slope of the rnonotonic loading curve at slip A=O. It
cm be calculated as
EbO=dddA 1 A=o=SP fct/AI (3-93)
where Al is the limit value for the slip. When the slip is greater than Al
,the bond stress remains constant until total bond failure occurs at slip=Au.
AI depends on the lateral pressure to the axis of the reinforcing bar. It can
be expressed in the following formula:
Al = 0.06mm+0.004q (Wmm2) (3-94)
where q is the confining pressure on the reinforcing bar in the contact
surface.
The factor Pi considers the size distribution of the aggregates (pl
ranges between 0.8 and 1.25) and the position of the concrete during
casting. The influence of the concrete casting is shown in Fig. 3-14
according Martin and Doerr (198 1) tests. When the directions of concrete
casting and of bar slip are identical, q=o0. When the directions of concrete
casting and of bar slip oppose each other, cp=180? When the directions of
concrete casting and of bar slip are perpendicular each other, q1=90~.

Fig. 3- 14 Influence of the direction of casting


of concrete corresponding to the pull-out direction.

The factor B2 considers the geometry of the bar surface, expressed by


the relative rib area, which is defined as the quotient of the rib area and the
bar surface between two ribs. The relative n b area is calculated as
a,=k*Fr*sin<p/(scdsi)+ih/st (3-95)
where k is the number of lug series ,
Fr is the area of the longitudinal section of one lug,
<p is the h g inclination towards the bar axis,
d is the nominal diameter of the steel bar,
si is the distance between lugs,
i is the number of longitudinal ribs,
h is the height of longitudinal rib, and
st is the pitch of twist of twisted bars.
Because the local bond stress-slip relationship is based on Doerr's tests
for ribbed reinforcement of 16 mm diameter, the influence of the geometry
of the bar surface is related to the 16 mm diameter bar. For this bar, the
relative rib area is ar=0.065. Marin (1981) studied the bond behavior of
reinforced concrete with varying geometry of the surface of the tested bar.
For relative displacements between 0.01 mm to 0.50 mm, Marin proposed
a bond stress-slip relationship normalized with respect to the compressive
concrete strength. The mean values related to the relative rib area &+.O65
can be represented by the cuve given in Fig. 3-15 .

Fig 3-15 Influence of the relative nb area of


the reinforcement bar normalized to ar=0.065
Related to the relative rib area, the ultimate relative displacement Au is
given in Table 3-1 for various concrete compressive strengths.
Table 3-1. Ultimate relative displacements Au(mm)
The factor B3 considers the transverse pressure or gaping based on the
expenmental research of Doerr (1978) and Eligehausen et al (1 983). Doerr
investigated this influence under rotationally symmetric pressure p leading
to the pressure q in the contact surface. With increasing confining pressure,
the bond strength increases. Eligehausen et a1 (1983) studied the influence
of confined pressure in only one direction. The results are summarized in
Fig. 3-16.

Fig 3-1 6 Relationship of bond strength and confining pressure


on the contact surface of a steel bar (q,=q for rotational
symmemc pressure, q,= uniaxial pressure).

There are no experimental data available to consider the influence of


lateral tension on the bond strength. It is obvious that lateral tension
applied to the reinforcing bar will decrease the bond strength. Mehlhorn et
al (1985) recommended that the factor P3 depends on the gap size At as
shown in Fig. 3-17. The h is the height of the lug.
The lateral pressure or tension on a reinforcing bar, produced by
concrete pressure or tension stress around the bar, is variable during the
loading procedure. Hence, it is not convenient to use the above curves. The
following formula can be obtained by the least squares method.
The increase of bond strength due to uniaxial confined pressure on the
contact surface of reinforcing bars is given by:

The decrease of bond strength due to gap At between concrete and


reinforcing bars is:

Fig. 3- 17 Relationship of bond strength


and gap between steel bar and concrete.

The factor P4 considers local damage of the concrete in the bond zone
caused by the development of secondary cracks. The influence of the
strain cs of the steel bar and the ratio of the yield stress f, to ultimate
strength f, of the steel bar are considered to express the damage of the
bond near the crack. The directions of the steel bar and the crack are also
considered by the test result. Multiplying these two influencing factors
leads to Eq. 3-84 to consider the effects of bond damage.
p 4 1 =1- (f&ni) (&o/Esy)

P42 = sin+(l-sine) d(4d) for x<4d


p 4 2 =1 for x>4d
P 4 =P41 p42 (3-99)
where 0 is the angle between the steel bar and the nearest crack of the bar
as shown in figure 3-18 ,and x is the distance between the bond element to
the nearest crack.

7 -
cracks
steel bar

Fig. 3-18 Angle between the steel bar and the crack of the concrete.
P42

Fig. 3-1 9 Influence of angle between steel b;u and the crack
and distance between the linkage element to a nearest crack on Bq2 .

For uncracked concrete, Pd2=l. For cracked concrete, in a theoretical


analysis, it 1s impossible to determine the positions of cracks accurately, so
Eq. 3-86 can not be used to calculate Pd2. In the analyses presented in this
thesis, it is assumed that P 4 2 =l.
The secant method is used for considering nonlinear characteristics of
bond stress-slip relationship. The stifiess of a bond element is calculated
as following:
a) On the initial loading increment
For linkage elements:
Km=Ak*Ew
For contact elements:
KpEm
b) On the other loading increments
For linkage elements:
K,=Ak dddA (3-101a)
For contact elements:
Km=dddA (3-101b)
Where Ai, is the contributary area of bond strength of element k. This
area for each linkage element, at a concrete node, is computed based on the
tributary surface area of the steel bars associated with each element
connected at that node. .

3X2.2 Eligehausen Model (Model 2)


This model is very different fiom Mehlhorn model. It allows for much
larger slips than does Mehlhorn model, and it utilizes larger bond
strengths. Eligehausen, Popov and Bertero (1 983), basing their
formulations on the findings of an extensive experimental program
descnbed previously, suggested a non-linear curve for the bond stress-slip
relationship. Other researchers complemented the model. Soroushian and
Chio (1989) obtained a relationship between the peak bond stress and the
bar size. Pochanart and Harmon (1989) examined the effect of bar surface
geometry and developed a relationship between the surface geometry of
the bar and the peak fictional bond stress. The Eligehausen model can be
defined by the following expressions:
Under monotonically increasing loading, The local bond stress and slip
relationship with suficient confined pressure to the reinforcing bars
(causing pull-out failure) is shown in Figure 3-20.
The bond stress and slip relationship can be defined by the following
expressions:
7=71 for A<=Al (3-102)
where rl is the maximum bond stress value, and Al is the bond slip
corresponding to the bond stress r=zlm.

Fig. 3-20 The bond stress and slip relationship of Eiigehausen Model.

This branch is followed by a plateau branch( r=tt=tl ) corresponding to a


slip field Ai<=A<=A2 .
T=TZ for Al<=A<=A2 (3-103)
The third branch is linearly decreasing up to the value r=ti , which is
achieved at A=A3 .
F r 2 -(A- A2)K A39 A 2 ) ( 1 2 - ri) for A2<=A<=A3
The fourth branch is a horizontal branch where the bond stress remains
constant and equal to its minimum value T = T ~( the ultimate fiictional bond
resistance).
T-q- for b A 3 (3-105)
where
ri =tz =(20-d/4)(f,'/30)~-' ( m a ) (3-1 06)
~~=(5.5-0.07S/H)(f,'/27.6)~-~(MP) (3- 107)
Ai= (f,'/3 o)'.~ ( mm ) (3-108)
A2==3.0 mm (3- 109)
A3= S (3-1 10)
a=0.4 (3-1 11)
where S and H are the clear spacing and height of lugs on the bar,
respectively.
For splitting failure, the local bond stress and slip c w e s are given by
the dashed line in Figure. 3-20. The curves are expressed by following
expressions:
T 1, =.rzs =O.748 (f,'~/d)*.~
<=r1 (MPa) (3-1 12)
234 (f;'~/d)O.~
Tfs =O. <=Q W a ) (3-1 13)
-
- (fJ30) 0.5e (Va)ln(tlsk1) ( m m )
AIs=Ale (Va)in(~ld~1) (3-1 14)
A2s =A2 (3-1 15)
A3s =A3 (3-1 16)
where c is the smaller of the concrete-cover thickness, and one-half the bar
spacing. 71, is the maximum bond stress value in splitting failure, and AISis
the corresponding bond slip to bond stress r=ti,. AZs is the maximum slip
of the plateau branch ( tfstis~the
~ = t ~ ).~ = ~ ultimate fiictional bond
resistance in splitting failure, A3, is the corresponding bond slip to bond
stress F T ~ , .

3.7.2.3 Mohamed Harjli Model (Model 3)


Mohamed Harjli et al (1995) proposed the same shape of bond stress and
slip curve for pull-out failure but with a different value of a and with
slightly different values of .ri and TC. The expressions are as follows :
71 =r2=2.575 (f,
r
)0.5 (MPa (3-1 17)
T~ =0.3ki (3-1 18)
Al= 0.75A- (3-1 19)
A2= 1.75A- (3-120)
A3= S (3-121)
a=O .3 (3- 122)
A, is the slip at which the peak bond resistance Tm&, is reached and
A, =0.189S+0.18 (3-123)
or more simply,
Ama.=0.2s
For splitting failure, the local bond stress and slip curves are given by
the dashed line in Figure 3-21. The curve is expressed by following
expressions:
T=T I @A I for A<=Als (3-124)
T=O for A>Als (3-125)
ri,=(0.249+0.29 1dd) (f, r )0.5 <=q (MPa) (3-1 26)
Als=A1e(Va)i n ( t l d ~ 1 )-
- 0.75A,, e (Va)i n ( ~ 1 d ~ i ) (3-127)
T~~=T~~=O (3-1 28)
where ri, is the maximum bond stress value in splitting failure, and AIS
is the bond slip corresponding to bond stress -1,.

Fig. 3-21 The bond stress and slip relationship of Model 3 (Harjli).

3X2.4 Proposed Model (Model 4)


This model is proposed by the author of this thesis according to scme
test results fiom published papers. This model is modified version of the
Eligehausen Model (Model 2). The local bond stress and slip relationship
is same as with Model 2 for pull-out failure, and some modifications are
made for splitting failure.
For splitting failure, the local bond stress and slip curves are a v e n in
the dashed line in Figure 3-22. They are defined by the following
expressions:
~ -1 d ut A l SIa for Ac=Ais (3- 129)
-@-
F ~ I S AIS)/( A ~ s AIS
- ) (% - T ~ )s for Al,<=A<=A2. (3- 130)
I-Tfs for 1A>A2, (3-13 1)
Fig. 3-22 The bond stress and slip relationship
of proposed model (Model 4).

sk ,ns,and A2 can be calculated usinp following formulations :


71s =T~*. 1 52lS (3-1 34)
A2,=2.0 mm (3-1 35)

3.7.2.5 Model of Bond Stress and Slip Relationship With Partial Confining
Pressure

According to the CEB (1993) code, largely based on the Eligehausen,


Popov and Bertero (1983) model described previously, the parameters for
confined concrete are applicable whenever the transverse pressure p>=7.5
MPa or closely spaced transverse reinforcement is present, satisfjing the
Where LA, is the cross area of stimps over a
conditions LAmv>=nAS.
length equal to the anchorage length; n is the number of the longitudinal
bars enclosed by the s h p s ; and A, is the area of one longitudinal bar. For
O< LAJIIA~ 4 . 0 or O c p < 7.5 MPa, a linear interpolation between the
values of confined and unconfined concrete are be used to derive the mode1
parameters.

Fig. 3-23 The bond stress and slip relationship


with partially confining pressure.

The local bond stress and slip relationship is shown in Figure 3-23. It is
defined as:
~lsp=~ls+(~~-~ls)P
~ 2 s p = ~ 2 s (~3-72s)P
f

rrsp=~+(~rrrJP
A1sp=A1s +(Ar Al n ) P
A2sp=AzS+(A3A2s)p

A3Sp=A3

where p is the confining pressure factor (O<=P<=l .O) .


3.7.2.6 Models 2 ,3 and 4 Under Cyclic Loading
On the basis of their cyclic loading tests, described in the previous
section, Eligehausen et al proposed the analytical mode1 shown in Figure 3-
24 together with corresponding experimental curves.
The f ~ sloading
t cycle is the rnonotonic response. The secant method is
used in the nonlinear finite element analysis. The stifiess of a bond
element is calculated as :
On the initial loading increment:
For the linkage elements:
G,=AkGo
For the contact elements:
G,=Go
On other loading increments:
For the linkage elements:
G,=Akdh
For the contact elements:
Gn=dA
The initial bond stiffness is attained at A/A1=l/l00. Then
Go=dr/dA=tia*100'" (3- 144)
Unloading is considered to take place at a constant slope of bond
stifkess. The same slope of bond stifhess is retained during reloading
towards the envelope.
nie unloading curve at point A is curve AD and the subsequent
horizontal branch DE. Denoting rPm as the maximum bond stress value
which is reached at the end of the previous loading phase, the horizontal
branch subsequent DE is afTected by the ordinate run=0.25zp,.

Fig. 3-24 The bond stress and slip relationship under cyclic
loading proposed by Eligehausen et al .

The reduction of bond resistance with increasing slip and number of


loading cycles is calculated using damage index D, which is equal to zero
for no damage and to one when bond breaks down completely ( ~ 0 )The
.
index D is a hnction of the ratio Efi, where E is the hysteretic energy
dissipation at the stage of unloading, and & is the energy corresponding to
the area under the rnonotonic bond stress and slip curve up to the value A3.
In calculating the value of E, only 50% of the energy due to friction is taken
into account, and the remaining 50% is assumed to be spent in overcoming
the friction resistance without causing any bond degradation.
During cycle i, the stifhess Ei can be calculated using the following
equations:
Ei=O (3-145)
Ei=Ei.i+AEi (3- 146)
AEi =O for (3-147)
AEi = rimA1(Aim /Al ) ('*) /(1+a)+0.5*0.25timAim
for &.8maxcAun<=A1 (3- 148)
AEi = ~ ~ ~ A ~ / ( l 0.5(rli
+ a ) + +rh)(A -Al) +0.Sf 0.2SimAU.
for AI< A, <=A2 (3-149)
AEi =riiA /(l +a)+0.5(rl i+tti)(A3Al)+O . 5 ( ~ ~ + 7), ( A h -&)
+ O S *0.25zkA, for A2 <Ah <=A3 (3- 150)
iAl/(l+a)+O. 5 ( 1i+~2i)(A3d1)+0
~ ~(T~~+T~)(A~-A~)
+tli(Aun-A3) +0.5 * 0 . 2 5 ~ ~ A , for A,>A3 (3-1 5 1)
For pull-out failure, & is given by:
b = r 1Ai/(l +a)+r (A3A1)+0S ( tl+~f)(A3-A2) (3-152)
For splitting failure, & is given by:
&=TI A d ( 1 +a)+0.5(~ls+~2s)(A2s-Als)+0.5(~
1&tfs)(A~s-A2s)

(3-1 53)
Di=l-e a (3- 154)
a=-l .z(E~/Eo)'.' (3455)
where Aim is the maximum absolute value of the bond slip reached during
the first half-cycle loading of cycle i if the present loading stage is in the
second half-cycle loading of cycle i ,and is the maximum absolute value of
the bond slip in the second half-cycle loading of cycle i-l (previous loading
cycle ) if the present loading stage is in the first half-cycle loading of cycle i
. t h is the absolute value of the bond stress corresponding to bond-slip
Ah.
For pull-out failure, t l i , rzi and ru are bond stresses corresponding to
bond slips Ai ,A2 and A3 in the cycle i , respectively. They are calculated
as:
TI^ =72i = ~ (1-Di
l ) (3-1 56)
Tfi =tf(1-Di ) (3-157)
For splitting failure without confining pressure, 7 i i , T l i and ra are bond
stresses corresponding to bond slips Ais , AZr and A3, in the cycle i ,
respectively. They are calculated as:
~ l =Tls
i (l-Di ) (3.158)
~ 2 =TZ=
i (1-Di ) (3- 159)
tfi=rfs (1-Di ) (3-160)
For splitting failure with partial confining pressure, Tli , rzi and TG are
bond stresses corresponding to bond slips AI, ,A2, and A3, in the cycle i ,
respectively. They are calculated as: *

rii (1-Di ) (3-161)


~ 2 =Tzse
i (1-Di ) (3.162)
tfi = ~ f s(1-Di ) (3- 163)
Dunng the unloading in cycle i ,bond stress is calculated as :
Ti =O .2Srim for O < = ) A ~ I < ~ A ~ ~(3-164)
I
ri = ( A i - A i ? r i d ( A i , - ~ ~ P > for 1 ~~ ~ = ~ 1
~ A ~ ~ ~
(3-165)
r i =f(Ai) for IhiI >=IhimI (3-166)
where Aim is the maximum (absolute) bond slip in the loading (or
reloading) stage of cycle i , and r, is the bond stress corresponding to
bond slip A,. A: is the plastic bond slip at bond slip Ah . It is defined
as:
A r = Aim TU^ / Go (3-167)
On reloading in cycle i ,the bond stress is calculated as :
r , =f(Ai)>=0.25~U, for ri>=O (3-168a)
r i =f(Ai)<=-O.25rh for T ~ < O (3- 168b)
where r, is the bond stress corresponding to the maximum absolute value
of bond slip at the previous loading (or reloading) stage. In equations (3-
152) and (3-154), f(Ai) 1s the bond stress calculated from the base curve
considering the bond degradation of cyclic loading of cycles i.

3.7 -2-7 Effect of Hooks in Reinforcement


From the experimental results of Eligehausen, Bertero and Propov (1 983),
the local bond stress and slip curves for bars with hooks under cyclic loading
are shown in Figure 3-25. They are given by the following expressions:
For monotonic loading:
r = 2 2 . 0 ( ~ / ~ ~ ) a" ) for IAI<=A, (3-169)
~ 2 2 . 0 (h4Pa) for IA~;A* (3- 170)
On a unloading and reloading cycle, bond stress is calculated as :
t =O for O c A c A"+ or Ap <Ac0 (3- 17 1)
r = ( A - ~ p )7mJ(Am+-A9 for A~ <A<Am+ (3- 172)
( A - )( A - A ) for A, <A < A~ (3- 173)
r =f(A) for A >= Am+ or A >= Am+ (3- 1 74)
where f(A) is the bond stress aiid slip curve of rnonotonic loading given
in (3-155) and (3-156). Am+ and A,.are the maximum and minimum slips
attained during al1 previous loading stages along the concrete-steel
interface respectively; r+, and r .
, are the bond stresses corresponding to
Am+ and Am. respectively. A~ and Ap are the plastic slips at the bond slip
A= Am+ and A= Am. respectively. They are calculated as
A~ =Am+ /Go
Ap =Am- 0 ~ /Go
~ -

Vi
2 O '
C
Y)
w
E
-5 1
O
p -10

-1 5

-20

-25
-10 -8 -6 -4 -2 O 2 4 6 8 10
slip (mm)

Fig. 3-25 Analytical local-bondstress and slip mode1


for deforrned bars with hooks.

3.7.2.8 The Normal Bond Stiflhess


Because there is a lack of experimental research regarding the local
bond stifiess between a steel bar and concrete in the direction normal to
the interface, the assumption made by M. Keuser et al (1983) is used in
this paper. The normal bond stifniess is assumed to equal 100 times the
tangent bond stifhess.
Gp100 Gr OC Ku=lOOKm (3-1 77)
CHAPTER 4
EVALUATION OF MODELS BY COMPARISON BETWEEN
FINITE ELEMENT.ANALYSIS AND EXPERIMENT

4.1 Verification of the Program


The expenmental results fkom bond specimens, given in the paper by
Keuser et al (1987), are used to ver@ the program. The specimen and
finite element mesh are shown in Figure 4-1. Linkage elements and contact
elements were altematively used at the interface between the concrete and
the steel bar. The specimen is a concrete cylinder of 75 mm diameter, with
a 16 mm deformed bar. Material data are given in Table 4-1. It should be
noted that this is a 3-D problem actually, so it is certain to cause some
degree of error due to simplification to the plane stress problem. The
parameters and maximum bond slips (for load= 70 kN) of the four bond
slip models are given in Table 4-2.
The distributions of steel force for low stress and high stress conditions
are plotted in Figures 4-2 to 4-5. From the Figures, it can be seen that
analytical results of two kinds of elements (linkage and contact elements) are
aimost identical. The analytical results of al1 models agree very well with
experimental results for low steel stress (load=20 kN), but the results fkom
Models 2 and 4 agree better than those fiom Models 1 and 3. For the high
steel stress conditions (load=70 kN),the analytical results of Model 1 differ
from the test results significantly; the farther away fkom the loaded end, the
larger the difference. Model 1 gives much larger slips than do the other
models, and it under-estimates the bond strength. The analytical results of
Models 2 and 4 match very well with the test results. The results of the finite
element calculation generally show satisfactory agreement with experimental
data for Models 2,3, and 4.
Table 4- 1 Material data.
Concrete 1 Steel Bar 1
E'=37.2 MPa fo,420MPa
Gt=2.7 MPa Eh=2050 MPa

Table 4-2 Specimen parameters and analytical maximum slips (mm ).


Element type Mode1 1 Mode12 1 Mode1 3 -
Mode1 4 -
1
Au=0.530 A1=0.178 A 104 Ai= O. 178
link A,-4.290 A,, =O. 124 A-4.097 &=O. 124
contact A-- 4 . 2 8 9 A,, 4.120 =0.093 &, 4 . 1 2 0

4 -2 Experiment Details
4.2.1 Dimensions and Reinforcement Arrangement of the Specimens
A set of four specimens of low-rise reinforced concrete walls was
designed and tested by Eun Hee Choi and Sasha Kurmanovic under
guidance of Professor J.F.Bonacci at the University of Toronto. The main
difference between the four walls was in the design of boundary elements.
Wall- 1 had uniformly distributed web reinforcament without boundary
elements. Concentrated vertical reinforcement was added near the vertical
edges of Wall-2 and Wall-3, but only Wall-3 had transverse (horizontal)
hoops enclosing the concentrated vertical reinforcement. Wall- 1, -2, and -3
had a rectangular cross section for the wall web. Wall4 had a small flange
on each side; the area of concentrated reinforcement was the same as in the
Wall-3 (showri in Figure 4.6).
A top and a bottom beam were designed for each wall. The vertical
reinforcement in the walls was anchored to the top and bottom beams to
achieve the full yielding capacity. The bottom beam was 3500 mm long,
500 mm deep, and 500 mm thick, and the top beam was 2325 mm long,
300 mm deep, and 400 mm thick.
The dimensions and reinforcing bars -ofthe four specimens are s h o w in
Figures 4.7 to 4.1 1 .

4.2.2 Material Properties of Specimens


In order to produce a strong base beam and relatively weak wall, two
different types of pre-mixed concrete were used : high strength concrete
for the base beams and normal strength concrete for the wall and the top
beam. Cylinder samples for each type of concrete were taken, cured,
ground and tested. Uniaxial compression tests were performed, and the
results are summarized in Table 4-3, Table 4-4 and Figure 4-12.

Table 4-3 Uniaxial compressive strength of the Wall-1 and Wall-2


7 days 28days Test day Test day
Moist cured Moist cured Moist cwed Air cured
Base beam 50 MPa O--e-
75 MPa -----
Top beam and wall 23 MPa 32 MPa 33 MPa 36 MPa
Table 4-4 Uniaxial compressive strength of the Wall-3 and Wall4
7 days 14days 1 7 days after Test day
Base beam -II--
48 MPa 66 MPa
Top beam and 26 MPa ---- 39 MPa

Four different types of bars were used to meet the guidelines of the AC1
and CSA Standard. A list of the bars and their characteristics are presented
in Table 4-5 and Figure 4-13.

Table 4-5 Characteristics of reinforcement steel.


Types of bars Area Yield stress Ultimate stress
(mm2) (MPa) (MPa)
D4 25.7 593 643

4.2.3 Loading History of Test Specimens


Most specimens tested under lateral loading are subjected to a point
load at one end of the top beam, and supported at the opposite corner of the
base beam. Therefore, the test specimens c m be interpreted as being
loaded diagonally. The method of loading adopted in this experiment is
different fiom the method commonly used in other studies. A steel plate
with stiffeners and an adapter-plate for an actuator were attached to the top
beam in this series of tests. The lateral force was applied at the adapter-
plate that was located in the middle of the steel plate and uniformly
distributed through the steel plate.
The loading history is outlined as below:
1. A 140kN lateral force was applied for one complete reverse cycle.
2. For the second cycle, load was applied until top beam displaced Imm.
3. Cycle 2 was repeated.
4. The wall was loaded until fust yielding of reinforcement was detected in
the load-deflection response. Then it was unloaded and reloaded in the
reverse direction for the same deflection. It was unloaded to finish this
cycle.
5 . Cycle 4 was repeated.
6. The sixth cycle proceeded in the sarne rnanner as cycle 4, but the top
deflection imposed was about two times larger than the deflection of
cycles 4 and 5.
7. Cycle 6 was repeated.
8. The load was applied in a similar manner as cycle 6, but the top
deflection imposed was about two times larger as the deflection of cycle
6.
9. Cycle 8 was repeated.
10. Load was applied until the wall failed.
The loading history of the four specimens is plotted in Figure 4-14 to
Figure 4-17.

4.3 Study of the Element Mesh


To study the effect of element meshes on the FE results, three kinds of
smeared element meshes (shown in Figures 4-18 , 4-19 and 4-20 ) were
used in the analysis for Wall-1. The top load-deflection curve of Wall-1 is
given in Fi-gure 4-25 for a coarse mesh (Figure 4-18) and in Figure 4-26
for a fine mesh (Figure 4-19). The loads calculated using the coarse mesh
were 10 -1 5 percent larger thm with the fine mesh; hence, the coarse mesh
is too course to attain the sufficiently accurate result. The results with the
fine mesh show only about a 1 percent -difference fiom the those obtained
with a finer mesh (Figure 4-20); hence, it is not necessary to use the finer
mesh to achieve so little improvement in accuracy at the expense of
computing time two to three times greater.
The element mesh of Wall- 1, shown in Figure 4-2 1, was used to study
the bond element response in the finite element model. From the analytical
results, the following can be concluded. Compared to the slip between the
vertical bars and surrounding concrete in the lower half of the web, the
slips between the horizontal bars and surrounding concrete and the slips
between the vertical bars and surrounding concrete in the upper half of the
web are negligible. Hence, in this thesis, the bond elements are employed
only between the vertical bars and surroundiog coiicrete in the lower half
of the web (Figs. 4-22 to 4-24).

4.4 Cornparison of the Analytical and Experimental Results


In order to verify the nonlinear bond stress and slip models described
above, the experimental results f b m the four shear walls tested at
University of Toronto are compared to analytical results. The walls are
fixed against translation in both the horizontal and vertical directions at the
base of the bottom beam. The measured deflection history is imposed
uniformly at the top of the walls.
The success of the analytical models is evaluated by comparing the
measured and computed load-deflection-responseof the top of the wall.
4.4.1 Shear Wall-1
The elernent mesh used is shown in Figure 4-22, and the ultimate loads
acting at the top of the wall are given in Table 4-6 for the FE analysis and
test data. From the ultimate loads of Table 4-6, it can be seen that there is
good correlation between the results calculated using TFUX and
experimental results, and the differences varying fiom 0.2% to 3-7% . The
contact elements give marginally better results than do the linkage
elements, but the m e r e n c e between results attained from the two types of
elements is small.

Table 4-6. Cornparison of ultimate loads acting at top of Wall-1


fiom test data and FE analysis ( kN )
1 FE analysis 1 Test data
Element Type Model 1 Model 2 Model 3 Mode14 N o slip*
linkage 409 408 385 - 403 425 400
contact 403 404 399 403

* assumes perfect bond between the concrete and the steel bars.

The computed load-deflection responses of Wall-1 (at the top of the


wall) are plotted in Figures 4-27 to 4-35 for the five bond slip models, and
two types of elements. The load-deflection envelope computed assuming
perfect bond, and computed using bond-slip Model 2 with contact
elements, and test response are plotted in Figure 4-36. The expenmentally
determined response is given in Figure 4-37. Although Model 1 gives a
ultimate load that is very close to test result, the curves of the load-
deflection are significantly different each other. The convergence of Model
1 is very poor for large slips/deflections (top deflections larger than 4 mm
for the linkage elements and 6 mm for the contact elements). This mode1
did not converge for deflections greater than about 6 mm for the linkage
elements and 10 mm for the contact elements, so the ultimate loads and
other results obtained fiom Model 1 are not reliable. For other three
Models, the results fiom the two types of bond elements do not have
apparent differences. There are not significant differences between the
perfect bond Model and Models 2,3 and 4 when top deflections of the wall
are smailer than 10 mm, but Models 2 and 4 give the better results when
top deflections of the wall are larger than 10 mm. Al1 bond-slip Models
give larger loads than those measured from test for top deflections of the
wall from about 10 mm to 16 mm, but the wall is failing in the analytical
models for a top deflection over 20 mm.
The calculated hysteretic response exhibits the same trend as the
measured data and successfully represents the amplitude of the measured
data except for the last cycle. The only notable discrepancies between
calculated and measured behavior are the degree of pinching evident in the
load deflection hysteresis and failure cycle.
The bond slip envelopes of the Wall-1 are plotted in Figures 4-75 to 4-
80 for Models 2 to 4. The bond slip at a node is defined by the
displacement of the steel tmss element node less that of the concrete
element node which is connected to uss element node by the bond
element. The maximum slips and minimum slips are given in Table 4-7.
During the first three cycles (when the peak deflections at the top of the
wall are not more than 1 mm), the bond slips are very small, and
significant bond stresses exist in the interfaces between the steel bars of
the bottom web and the surrounding concrete. In these loading stages, the
shear wall works as if there is perfect bond between the steel bars and the
surrounding concrete. The large active slips occur at the top of the bottom
beam, and large negative slips occur in the tensile side of a zone near 40%
height of the web wall. When the outside concrete of the bottom web is
uncracked or when cracks are very small, the maximum slips occur at the
interfaces between the outside tensile steel bars and the surrounding
concrete. As the cracks of the outside concrete at the web bottom become
large, the maximum slips move toward the interfaces between inside
tensile steel bars and the surrounding concrete. In the tensile zone near
40% height of the web wall, where the concrete cracks are large, the
displacement of a concrete node is larger than that of the corresponding
steel bar node causing large negative slip. In this area, the peak values of
maximum and minimum slips calculated using contact elements are larger
than those attained using linkage elements. nie maximum and minimum
slips calculated using contact elements are much smaller than those
atained using linkage elements in the cycle 1 to 7. This means that the
maximum and minimum slips caiculated using contact elements are
smaller than those attained using linkage elements when the top deflections
are small, and the maximum and minimum slips calculated using contact
elernents are larger than those attained using linkage elements when the top
deflections are large. The difference bekeen the maximum and minimum
slips obtained fkom Models 2 and 4 is not much when linkage elements are
used, but the difference is large when contact elements are used. The slip
envelopes of Mode1 2 are the same for linkage and contact elements.
Table 4-7 Maximum and minimum slips of Wall-1
by FE analysis (mm)
1 Bond Model ( Model 2 1 Model 3 1 Model 4 1
max min max min max min
Linkage elemenq 0.53 -0.45 0.84 -2.42 0.53 -0.59
Contact element 0.53 -0.45 2.94 -1.90 0.64 -1.36

Figures 4-87 to 4-92 give the bond slip, bond stress and steel stress at
the outside bottom of the web of Wall-1 for Model 2 using contact
elements. Very small slips are seen in the first three cycles. From cycles 4
to 7 (the peak top deflections of the wall are 2 mm to 4 mm), there are
apparent bond slips. The bond slips are large fkom cycles 7 to 9 (the peak
top deflection of the wall is 10 mm). The peak values of the bond stresses
do not change much fiom cycles 4 to 9. Actually, the peak values of the
bond slips in these cycles lie between Al and A2 , and the bond strength in
these peak values is constant (the horizontal branch FTI of bond stress-slip
curve) for Model 2. In the results coming from Models 3 and 4 , the peak
values of bond stresses decrease gradually from cycle 7 due to peak bond
slips that are larger than Al .
The steel stresses in the three outside vertical bars at the bottom reach
the yield stress during the 6th cycle (the peak top deflection is 4 mm).
Dunng the 8th cycle (the peak top deflection was 10 mm), the steel stresses
in the outside four vertical bars exceeded the yield stress and the stresses in
other vertical steel bars ranged between 40% and 92% of yield stress at the
bottom of the web except for three vertical bars in the middle of the web.
The steel stresses in the outside vertical bars at the bottom web dropped

133
significantly before failure due to bond damage, and the steel stresses in
the inside vertical bars near the horizontal middle of the web were very
large (the largest steel stress is about 1:2 f,). When the wall was about to
fail, the crack width was very large (the largest crack was more than 10
mm). There were large slips (over A2 ) in a zone of near 40% height of the
web at the tensile side, bond stresses were very small, and the steel stresses
dropped suddenly. The wall-then failed.
The maximum concrete compressive stress, at the compressive side of
the wall, was fd= f l when the failure was about to occur. The deflection of
Wall-1 after failure are shown in Figure 4-1 11 to 4-1 17 for assuming
perfect bond and bond-slip Models 2 to 4 and two bond elements. Wall-1
failed due to two factors: large steel bar slips (at the zone near the base of
the web when assuming perfect bond; at the zone near 40% height of the
web wall for Models 2 , 4 ; and at the zone near one-third height of the web
wall for Model 3 in tensile side), and concrete crushing at the compression
face. The computed failure mode was a combination of bond slip and
concrete crushing failure.

4.3.2 Shear Wall-2


The element mesh used to mode1 Wall-2 is shown in Figure 4-23. The
ultimate loads acting at the top of the wall are given in Table 4-8 for the FE
analysis and the test data. The analytical predictions fiom the finite
element method fit very well with the experimental results except for
Model 1; the differences range fiom 0.3% to 5.5% . The ultimate loads
predicted by Model 4 have only a 0.3% difference for contact elements,
and a 0.8% difference for linkage elements when with the compared test
data. The load-deflection response cornputed using contact elements is
closer to experimental results than thaf obtained using linkage elements,
but the difference in results attained using the two kinds of elements
small.
Table 4-8. Cornparison of ultimate loads for Wall-2
fkom test data and FE analysis ( kN )
FE analysis Test data
Element Type Model 1 Model 2 Mode13 Model 4 No slip*
1in kage 265 645 605 624 747 640
contact 600 642 623 633
linkagekest 0.4 14 1.008 0.945 0.974
contadtest 0.938 1.003 0.973 0.988 1.167
- - . -

* Assumes perfect bond between the concrete and the steel bars.
The computed load-deflection of Wall-2 (at the top of the wall) is
plotted in Figures 4-38 to 4-46 for the five bond models and two types of
elements. Load-deflection envelope computed assuming perfect bond, and
computed using bond-slip Model 2 with contact elements is compared to
the test data in Figure 4-47. The experimentally observed response is given
in Figure 4-48. The convergence of the Model 1 is very poor for large
slips/deflections (deflections larger than 4 mm for linkage elements and 6
mm for contact elements), hence, the ultimate loads and other results
obtained f'kom Model 1 are not reliable. For the other three models, the
results fiom the two types of bond elements show no obvious difference.
The loads calculated assuming perfect bond (i.e., no bond slip) are larger
by about 17% compared to those attained fiom the test. The wall fails
earlier in the analytical models than in the test. The load deflection curves
derived fiom Model 2 agree best with the experimental curves.
The analytical hysteretic loops follow the same trend as the observed
experimental data and successfully represented the amplitude of the
measured data except for the last cycle. The only notable discrepancies
between calculated and observed behavior are the degree of pinching
evident in the load-deflection hysteresis and failure cycle.
The bond slip envelopes for Wall-2 are plotted in Figure 4-8 1 and 4-82
for Model 2. It is evident in the figures that, in the first three cycles, the
slips are very small and the shear wall works as if there is perfect bond
between steel bars and surrounding concrete. The maximum slips are 0.82
mm and 0.87 mm, and minimum slips are -2.88 mm and -2.93 mm for
linkage and contact elements, respectively. The large active slips occur at
the top of the boaom bearn and large negative slips occur in the web zone
near one-third height of the wall up the bottom, due to large concrete
cracks (the largest crack is about 6 mm). When the outside concrete of the
bottom web is uncracked or when cracks are small, the maximum slips
occur at the interfaces between outside tensile steel bars and surrounding
concrete at the bottom of the web. As the cracks of the outside concrete of
the bottom web become large, the maximum slips move toward the
interfaces between the inside tensile steel bars and the surrounding
concrete at bottom of the web. The slip envelopes of Model 2 are the same
for the linkage and contact elements.
Figures 4-93 to 4-98 give the bond slip, bond stress and steel stress at
the outside bottom of the web of Wall-2 for Model 2 using contact
elements. In the fourth and fifth cycles (the peak top deflection of the wall
is 3.5 mm), there are obvious bond slips. The bond slips are large in cycles
6 and 8 (the peak top deflection of the wall is about 7 mm). The peak
values of the bond stresses do not change much fiom cycles 4 to 8, except
during cycle 7 when the peak values of the bond slips in these cycles lie
between Ai and A2 , and the bond strength in these peak values is constant
(the horizontal branch of bond stress and slip c w e ) for Model 2. In
cycle 7 , the maximum (or minimum) slip is less than that of the previous
cycle, so bond stresses are very small. In cycle 8, the bond stresses in most
elements of the web are smaller than in cycle 6 for the large bond slips. In
the results obtained fkom Models 3 and 4 , the peak values of the bond
stresses decrease greatly in cycle 8 due to peak bond slips larger than Ai .
For Model 2, the steel stresses of the three outside vertical bars at the
bottom of the web reach the yield stress during cycles 6 and 8. For Models
3 and 4, the steel stresses of the outside vertical bars at the bottom drop
due to large slips (the peak slips are larger than AI). When the wall is on
the verge of failing, the steel stresses of the inside vertical bars are very
large (the Iargest steel stress was about 1.1 fy), and there are large slips in
the section at one-third height of the web (over A2 ). Bond stresses are very
small, the steel stresses drop suddenly, and then the wall fails .
In al1 loading stages, the compressive stresses in the concrete are less
than the compressive strength of the concrete. The deflections of the Wall-
2 after failure are shown in Figures 4-1 18 to 4-124 for the various bond
models and bond elements. The computed failure of Wall-2 is due tu large
slips and concrete shear damage at the base of the web when assuming
perfect bond, and at a section at about one-third-height of the web for al1
other bond-slip models.
4.3.3 Shear Wall-3
The element mesh used to mode1 Wall-3 is s h o w in Figure 4-23. The
ultimate loads acting at top of the wall are given in Table 4-9 for the
various analyses and test data. From Table 4-9, it can be seen that the
ultimate loads calculated using al1 bond-slip models, except Model 1, are
reasonably similar to the experimental results, with the differences ranging
fiom 0.0% to 5.5% . Model 2 gives ultirnate loads which agree to within
0.0% for the contact elements and 1.9% for the linkage elements. In
general, the contact elements give better ultimate loads than do the linkage
elements. The ultimate load calculated assuming perfect bond is about 23%
larger than that obtained in the experiment.

Table 4-9. UItimate loads for Wall-3


fiom test data and FE analysis ( kN )
FE analysis Test data
Element Type Model 1 Model 2 Model 3 Model 4 No slip*
linkage 550 64 1 605 627 787 640
contact 562 652 611 632
linkagehest 0.859 1.000 0.945 0.980
contacthest 0.878 1.019 0.955 0.988 1.230
- - -- - - -

* assumes perfect bond between the concrete and the steel bars.
The computed load-deflection responses of Wall-3 (at the top of the
wall) are plotted in Figures 4-49 to 4-57 for the five bond-slip models and
the two types of element. The load-deflection envelopes computed
assuming perfect bond, and computed using bond-slip Model 2 with
contact elements, are plotted against the test result in Figure 4-58. Plotted
in Figure 4-59 is the experimental load-deflection response. The
convergence of Model 1 is very poor for large slipddeflections (deflections
larger than 3 mm), so ultimate loads and other results coming fiom Model
1 are not reliable. For the other three models, the results fiom the two types
of bond elements have some apparent differences. The predictions usine
contact elements give better results; the analytical results agree very well
with test data except for the failure cycle. The loads calculated assuming
perfect bond are about 20% larger than those attained fiom the test. The
wall fails earlier in the analytical models than in the test. The load
deflection curves for Wall-3 derived fiom Model 2 best agree with the
experimental curves.
The calculated hysteretic response exhibits the same trend as the
measured data and successfully represents the amplitude of the measured
data except for the last cycle. The only notable discrepancy between the
calculated and measured behavior is the degree of pinching evident in the
load deflection hysteresis.
The bond slip envelopes for Wall-3 are plotted in Figures 4-83 and 4-84
for Model 2 . It is seen in these figures that slips are very small and the
shear wall works as if there is perfect bond between the steel bars and the
surrounding concrete in the first three cycles. The maximum slips are about
0.24 mm and 1.08 mm, and minimum slips are -3.16 mm and -2.73 mm
for linkage elements and contact elements, respectively. The largest active
slips occur at the top of the bottom bearn and the largest negative slips
occur at the one-third height of the web wall due to large cracks (the widest
crack was 5.6 mm) at the tension side of the web.
Figures 4-99 to 4-104 give the bond slip, bond stress and steel stress at
the outside bottom of the web of Wall-3 for Model 2 with contact
elements. During cycles 4 and 5 (the peak top deflection of the wall is 3
mm), there is some apparent bond slip. The bond slips are large during
cycles 6 to 8. The peak values of the bond stresses do not change f?om
cycles 6 to 8, while the peak values of the bond slips in these cycles lie
between Ai and A2 , and the bond strength in these peak values is constant
(the horizontal branch F r , of bond stress-slip c u v e ) for Model 2. In the
8th cycle, the bond stresses in most elements in the web are very small due
to the large bond slips. In the results obtained fkom Models 3 and 4 , the
peak values of bond stresses decrease gradually fiom cycle 6 due to peak
bond slips being larger than AI .
The steel stresses in the outside vertical bars at the bottom web reach
the yield stress in the sixth cycle, and exceed it during the seventh cycle.
The steel stresses of the outside vertical bars at the bottom decrease greatly
when the wall is at the point of failure in the 8th cycle. For Models 3 and 4,
the steel stresses of the outside vertical ban at the bottom cirop earlier and
to a greater extent. When the wall is failing, large slips occur in the zone
near the one-third height of the web. Bond stresses are very small, and the
steel stresses &op suddenly.
At al1 loading stages, the compressive stress in the concrete is less than
the concrete compressive strength. The cracks and defiections of Wall-3
are shown in Figures 4-126 to 4-130 assurning perfect bond and bond-slip
Models 2 to 4, and two bond elements. The computed failure of Wall-3 is
due to large bond slips and concrete shear damage at the base of the web
when assuming perfect bond, and at about one-third height of the web for
al1 other bond-slip models.
4.3.4 Shear Wall4
The element mesh used to mode1 Wall4 is shown in Figure 4-24. The
ultimate loads acting at the top of the wall are given in Table 4-10 for
analytical and test results. From the ultimate loads of Table 4-10, it can be
seen that the correlation between the ultimate loads calculated using TRIX
and experimental results is good; the differences range fiom 1.7% to 3.2%
for al1 models except Model 1. Models 2 and 4 give ultimate loads that
agree exceptionally well with the test data; differences of 1.8% and 1.7%
for the linkage elements, and 2.6% and 2.4% for the contact elements,
respectively, for Models 2 and 4. The ultimate load calculated assuming
perfect bond is about 26% larger than the experimental result.

Table 4-10. Comparison of ultimate loads for Wall-4


fiom test data and FE analysis ( kN )
1 FE analysis ITest data 1
Elernrnt ~ ~Model 1
~ 1 e Model 1 1
( 2 Mode13 Model 4 No slip* 1 I
linkage
contact

* assumes perfct bond between th concrete and the steel bars.


The load-deflection responses of Wall4 are plotted in Figures 4-60 to
4-68 for the five bond Models and the two types of elements. The load-
deflection envelopes computed assuming perfect bond, computed using
bond-slip Model 2 with contact elements, and from the test are plotted in
Figure 4-69. The expenmental result is given in Figure 4-70. The
convergence of Model 1 is very poor for large slips/deflections (deflections
larger than 3 mm), so ultimate loads and other results obtained from Model
1 are not reliable. For the other three bond-slip models, the results corn the
two types of bond elements are not much diEerent, but the analytical
response obtained using contact element fits better with the expenmental
results. The analytical results using Models 2 to 4 agree very well with test
data except during the failure cycle. The loads calculated assuming perfect
bond are about 20025%larger than those attained from the test. The wall
fails earlier in the analytical models than in the test. The load deflection
cuves derived fiom ~ o d e i s2 and 4 agree best with the expenmental
Cumes.
The calculated hysteretic response exhibits the same trend as the
measured data and successfully represents the amplitude of the measured
data except for the last cycle. The only notable discrepancy between
calculated and measured behavior is the degree of pinching evident in the
load deflection hysteresis.
The bond slip envelopes of Wall4 ire plotted in Figure 4-85 and 4-86
for Model 2. It is seen in these figures that the maximum slips are about
0.5 mm and 1-84mm, and the minimum slips are -0.74 mm and -0.7 1 mm
for linkage elements and contact elements, respectively. The large active
slips occur at the top of the bottom bearn, and large negative slips occur
near the middle-height of the web due to large cracks (the widest crack is
5.7 mm) in the tensile side of the web.
Figures 4- 105 to 4-1 10 give the bond slip, bond stress and steel stress at
the outside bottom of the web of Wall-4 calculated using Model 2 with
contact elements. During the fust cycle, the slips are very small and the
shear wall works as if there is perfect bond between the steel bars and the

132
surrounding concrete. During cycles 2 and 3 (the peak top deflections of
the wall are 1 mm), there are apparent bond slips. During cycles 4 and 5
(the peak top deflections of the wall are between 4 mm and 5 mm), there
are large bond slips. The peak values of the bond stresses do not change
from cycle 4 until the wall is about to fail, while the peak values of the
bond slips in these cycles lie between Ai and A2 . n e bond strength in
these peak values is constant (the horizontal brmch r=rl of bond stress-slip
curve) for Model 2. When the wall is about to fail in the 8th cycle, the
bond stresses in most elements of the .web are very small for the large
bond slips. In the results obtained fkom Models 3 and 4, the peak values of
bond stresses decrease gradually from cycle 6 due to peak bond slips that
are larger than Ai .
The stresses in the outside vertical bars at the bottom tension side
reach the yield stress during the first half-cycle of cycle 6, and exceed it in
the second half-cycle of cycle 6 and first half-cycle of cycle 7. The steel
stresses at the bottom of outside vertical bars decrease greatly in the second
half-cycle of cycle 7. For Models 3 and 4, the steel stresses at the bottom
of outside vertical bars drop more rapidly. When the wall is near failing,
large slips occur at the mid-height of the web for Models 2 and 4 using
linkage elements, and at the bottom of the web for d l models using contact
elements and Model 3 (over A2 ) using linkage element. The bond stresses
are very small, and the steel stresses drop suddenly. The wall fails due to
large slips and shear damage.
The maximum compressive stress in the concrete is less than the
compressive strength during al1 loading stages. The computed deflections
of Wall-4 after failure are shown in Figures 4-132 to 4-138 for the various
bond models and bond elernents. The computed failure of Wall4 is due to
large slips and concrete shear damage- at the mid-height of the web for
Models 2 and 4 using linkage elements, and at the bottom of the web for
the other bond-slip Models and bond elements. If assuming perfect bond,
Wall4 fails due to concrete cnishing at the compression side, at the base of
the web.
The computed load-deflection responses of the four test specimens are
plotted in Figure 4-71 to 4-74 for Model 2 with contact bond elements and
assuming perfect bond under monotonic loading. The ultimate loads acting
at the top of the walls are given in Table 4-1 1. When displacements are
small, the differences between the results obtained from the two models
(assuming perfect bond, and Model 2 with contact elements) are small
because of the small bond slips. When displacements are larger, the
differences in the results are greater because of larger bond slips. When the
walls are near failing, the differences in the results fiom the two models
become smaller. The ultimate loads under rnonotonic loading are larger
than those under cyclic loading for both models, but the differences are
small. The ratios of the ultimate loads calculated assuming perfect bond to
those calculated using bond-slip Model 2, under rnonotonic loading, are
slightly smaller than those calculated under cyclic loading.
Table 4-1 1. Compatison o f ultimate loads for the four test walls
calculated assuming perfect bond and-bond-slip Model 2 with
contact bond elements under monotonie loading( kN )
Wall- 1 Wall-2 Wall-3 Wall4
Perfect bond 430 76 1 802 867
Mode1 2 + contact 408 647 659 703
PerfectModel2 1 .O54 1.176 1.217 1.233

Fig. 4-1 Finite element mesh of a specimen


used to verify the TRIX.
-test data
+mode1 1
+, mode12
mode13
-mode1 4

Fig. 4-2 Distribution of tensile force in bar under


Load =20 kN (linkage element)

-est data
-m-- mode11
,-mode1 2
m o d e 1 3
m o d e 1 4

Fig. 4-3 Distribution of tensile force in bar under


Load =20 kN (contact element )
t e s t data
+model 1
+mode1 2
+mode13
-x- mode14

Fig. 4-4 Distribution of tensile force in bar under


Load =70 kN (linkage element )

-test data
+mode1 1
+mode1 2
m o d e 1 3
-x- mode14

Fig. 4-5 Distribution of tensile force in bar under


Load =70 k N (contact element )
Wall- 1
--
Fig. 4-6 Cross-sections of walls (Kuzmanovic, 1994)
Fig. 4-9 Side view of reinforcement-Wall- (Chio, 1995)
Fig. 4-1 1 Side view of reinforcement-Wall-4 (Chio, 1 995)
W a l l
-Wall
W a l l
*base
-bas8
-base

Fig 4- 12 Stress-strain curves of concrete used in Wall-3 and Wall-4.

Fig. 4- 13 Stress-strain curves for reinforcement steel.


O 1 2 3 4 5 6 7 8 9 10
cycle

Fig. 4- 14 Loading history of the specimen Wall-1

O 1 2 3 4 5 6 7 8 9 1O
Cycle

Fig. 4-1 5 Loading historv of the s~ecimenWall-2


Fig. 4-16 Loading history of the specimen Wall-3

Cycle

Fig. 4-1 7 Loading history of the specimen Wall4


Fig. 4-1 8 Coarse rnesh for Wall-1

Number of nodes : 103


Number o f elements : 78
Dimensions o f web element: 22Smm x225mm
Fig. 4-19 Fine mesh for Wall-1

Number of nodes : 260


Number of elements : 222
Dimensions of web element : 112.5mmx 11ZSmm
Fig. 4-20 Finer mesh for Wall-l

Number of nodes : 497


Number of elements : 446
Dimensions of web element: 75mm x75mm
Fig. 4-21 Mesh for Wall-1 when considenng bond-slip
between interfaces of al1 steel bars and concrete of web.

Number of nodes : 466


Number of elements (linkage element model): 703
Number of elements (contact element model): 688
Dimensions of concrete web element : 1 LMmm x 1 l2.Smm
Length of truss element at the bottom steel hooks : 50 mm (about 5d)
* Thick Iine represents steel tmss elements and bond slip elements.
Fig. 4-22 Final FE mesh for Wall-1

1. For perfect bond model


Number of nodes : 260
Number of elements : 285
2. For other bond-slip models
Nurnber of nodes : 364
- mo$eQ:39?
Numberofekment~link;rgeeleme~t - - - - - - -

Number of elements (contact element model): 388


Dimensions of concrete web element: 1 12.5mmx1 12.5mm
Length of truss element at the bottom steel hooks : 50 m m (about 5d)
* Thick line represents steel truss elements and bond slip elements
Fig. 4-23 FE mesh for Wall-2 and Wall-3

For perfect bond model


Number of nodes : 3 16
Number of elements : 379
For other bond-slip models
Number of nodes : 463
Number of elements (linkage element model): 540
Number of elements (contact element model): 525
Length of m i s s element at the bottom steel hooks : 50 mm (about 5d)
* Thick line represents steel tniss elements and bond slip elements
Fig. 4-24 FE mesh for Wall-4

For perfect bond model


Number of nodes : 288
Number of elements : 339
For other bond slip models
Number of nodes : 417
Number of elements (linkage element model): 480
Number of elements (contact element model): 467
Length of truss element at the bottom steel hooks : 50 mm (about Sd)
* Thick line reDresents steel tmss elements and bond s l i ~elements
-26 -24 -22-20 -18 -16 -14 -12-10 4 -6 4 -2 O 2 4 6 8 10 12
Displacement (mm)

fin
msh

Fig. 4-26 Computed load-displacement response of Wall-1


using fine and finer smeared elements
'splacememt (mm)

Fig. 4-27 Load-displacement of Wall- 1 assuming perfect bond

-7 -6 -5 4 -3 -2 -1 O 1 2 3 4 5
R'spiacernent (mm)

Fig. 4-28 Load-displacement of Wall-1 for bond-slip Mode1 1


(linkage elements)
-26 -24 -Z-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 O 2 4 6 8 10
Dispiacement (mm)

Fig. 4-29 Load-displacement of Wall-1 for bond-slip Model 2


(linkage elements)

-26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 4 -2 O 2 4 6 8 10 12


Displacement (mm)

Fig. 4-30 Load-displacement of Wall-1 for bond-slip Model 3


(linkage elements)
-26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 O 2 4 6 8 10
Displacement (mm)

Fig. 4-3 1 Load-displacement of Wall-1 for bond-slip Model 4


(linkage elements)
1

-12 -10 -8 -6 4 -2 O 2 4 6 8 10 12
Displacement (mm)

Fig. 4-32 L.oad-displacementof Wall-1 for bond-slip Model 1


(contact elements)
R'splacement (mm)

Fig. 4-33 Load-displacement of Wall-l for bond-slip Mode1 2


(contact elements)

-26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 O 2 4 6 8 10 12


Displacement (mm)

Fig. 4-34 Load-displacement of Wall-1 for bond-slip M ~ d e l 3


(contact elements)
-26 -24 -22 -20 -18 -16 -14 -12 -10 -8 6 4 -2 O 2 4 6 8 IO 12
Displacement (mm)

Fig. 4-35 Load-displacement of Wall-1 for bond-slip Model 4


(contact elements)

-28 -24 -20 -16 -12 -8 -4 O


Displacement ( mm )
Fig. 4-36 Load-displacement envelope of Wall-1 computed assuming
perfect bond (PB), computed using bond-slip Model 2 with contact
elements (CB2), and as measured during test.
ispiacement (mm)

Fig. 4-38 Load-displacement of Wall-2 assuming perfect bond

-8 -6 4 -2 O 2 4 6 8
Oisplacement (mm)

Fig. 4-39 Load-displacement of WalI-2 for bond-slip Model 1


(linkage elements)
Displacement (mm)

Fig. 4-40 Load-displacement of Wall-2 for bond-slip Model 2


(iinkage elements)

- -6 4 -2 O 2 4 6 8
Dispiacement (mm)

Fig. 4-41 Load-displacement of Wall-2 for bond-slip Model 3


(linkage elements)
-2 O 2
spiacement (mm)

Fig. 4-42 Load-displacement of Wall-2 for bond-slip Model 4


(linkage elements)

-2 O 2
R'splacement (mm)

Fig. 4-43 Load-displacement of Wall-2 for bond-slip Model 1


(contact elements)
-2 O 2
asplacement (mm)

Fig. 4-44 Load-displacement of Wall-2 for bond-slip Model 2


(contact elements)

-2 O 2
Displament (mm)

Fig. 4-45 Load-displacement of Wall-2 for bond-slip Model 3


(contact elements)
-8 -6 -4 -2 O 2 4 6 8
R'spiacement (mm)

Fig. 4-46 Load-displacement o f Wall-2 for bond-slip Model 4


(contact elements)

-PB
+CB2
+Test

-16 -12 -8 4 O 4 8 12 t6 20
Displacement ( mm )

Fig. 4-47 Load-displacement envelope of Wall-2 computed


assuming perfect bond (PB), computed using bond-slip Model 2
with contact elements (CB2), and as rneasured during test.
-16 -14 -12 -10 -8 -6 -4 -2 O
Displacement (mm)

Fig 4-49 Load-displacement of Wall-3 assuming perfect bond

-6 -4 -2 O
Displacement (mm)

Fig 4-50 Load-displacement of Wall-3 for bond-slip Mode1 1


(linkage elements)
-12 -10 -6 -4 -2 O 2 4 6 8
Dispacement (mm)

Fig. 4-5 1 Load-displacement of Wall-3 for bond-slip Model 2

-1 0 -8 -6 4 -2 O 2 4 6 8
Displacement (mm)

Fig. 4-52 Load-displacement of Wall-3 for bond-slip Model 3


(linkage elements)
-12 -10 -8 4 4 -2 O 2 4 6

Displacement (mm)

Fig. 4-53 Load-displacement of Wall-3 for bond-slip Mode14


(linkage elements)

-12 -1O -8 -6 -4 -2 O 2 4 6 8
Displacement (mm)

Fig 4-54 Load-displacement of Wall-3 for bond-slip Mode1 1


(contact elements)
Displacement (mm)

Fig. 4-55 Load-displacernent of ~ a l l - 3for bond-slip Model 2


(contact elements)

-12 -10 -8 -6 -4 -2 O 2 4 6 8
DispiacemeM (mm)

Fig. 4-56 Load-displacement of Wall-3 for bond-slip Model 3


(contact elements)
Fig. 4-57 Load-displacement of Wall-3 for bond-slip Model 4
(contact elements)

-PB
+C82
+Test

Displacement ( m m )

Fig 4-58 Load-displacement envelope of Wall-3 computed


assuming perfect bond (PB), computed using bond-slip Model 2
with contact elements (CB2), and as determined fkom the test.
-10 -8 -6 4 -2 O 2 4 6 8 10 12 14 16
Dispiacement (mm)

Fig 4-60 Load-displacement of Wall-4 assuming perfect bond

-10 -8 4 -2 O 2 4 6 8 10 12 14 16
Displacement (mm)

Fig. 4-6 1 Load-displacement of Wall4 for bond-slip Mode1 1


(linkage elements)
-10 -8 -6 4 -2 O 2 4 6 0 10 12 14
Displacement (mm)

Fig. 4-62 Load-displacement o f Wall4 for bond-slip Model 2


pnkage elements)

-10 -8 -6 4 -2 O 2 4 6 8 10 12 14 16
'splacement (mm)

Fig. 4-63 Load-displacement of Wall4 for bond-slip Model 3


(linkage elements)
-10 -8 -6 4 -2 O 2 4 6 8 10 12 14
Displacement (mm)

Fig. 4-64 Load-displacement of Wall4 for bond-slip Model 4


(linkage elements)

-10 4 4 4 -2 O 2 4 6 8 10 12 14 16
Dispiacemecit (mm)

Fig. 4-65 Load-displacement of Wall-4 for bond-slip Model 1


(contact elements)
-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 l2 l4 l6
R'splacement (mm)

Fig. 4-66 Load-displacement of Wall4 for bond-slip Mode1 2


(contact elements)

Displacernent (mm)

Fig. 4-67 Load-displacement of Wall-4 for bond-slip Mode1 3


(contact elements)
-10 -8 4 4 -2 O 2 4 6 8 10 12 14 16
Oisplacement (mm)

Fig. 4-68 Load-displacement of Wall4 for bond-slip Model 4


(contact elements)

Fig. 4-69 Load-deflection envelope for Wall4 computed


assuming perfect bond(PB) ,cornputed using bond-slip Model 2
witb contact elements (CB2), and as measured dunng the test.
O 5 10 - 15
Displacement ( mm )

Fig. 4-7 1 Load-displacement of Wall- 1 calculated assuming


perfect bond (PB) and calculated using contact elements
with bond Model 2 (CB2) under rnonotonic loading.

Fig. 4-72 Load-displacement of Wall-2 calculated assuming


perfect bond (PB) and calculated using contact elernents
with bond Model 2 (CB2) under rnonotonic loading.
Fig. 4-73 Load-displacement of Wall4 calculated assurning
perfect bond (PB) and calculated using contact elements
with bond Model 2 (CB2) under rnonotonic loading.

O 5 10 15
Displacement ( mm )

Fig. 4-74 Load-displacement of Wall4 calculated assuming


perfect bond (PB) and calculated using contact elements
with bond Model 2 (CBZ)under rnonotonic loading.
4 5
Cycle

Fig. 4-75 Bond slip envelope of Wall-1 ( Model 2, linkage elements)

L
1

(
u
O 1 2 3 4 5 6 7 8 9 10

Cycle

Fig. 4-76 Bond slip envelope of Wall 1 ( Model 3, linkage elements)


4 5 6
Cycle

Fig. 4-77 Bond slip envelope of Wall-1 ( Model 4, linkage elements)

Fig. 4-7 8 Bond slip envelope of Wall-1 ( Model 2, contact elements)


4 5 6
Cycle
Fig. 4-79 Bond slip envelope of Wall-1 ( Model 3, contact elements)

4 5 6
Cycle
Fig. 4-80 Bond slip envelope of Wall-1 ( Model 4, contact elements)
O 1 2 3 4 5 6 7 8 9
Cycle

Fig. 4-81 Bond slip envelope o f Wall-2 ( Model 2, linkage elements)

4 S
Cycle

Fig. 4-82 Bond slip envelope of Wall-2 ( Model 2, contact elements)


Cycle

Fig. 4-83 Bond slip envelope of Wall-3 ( Model 2, linkage elements)

3 4
Cycle
Fig. 4-84 Bond slip envelope of Wall-3 ( Model 2, contact elements)
3 4
Cycle
Fig. 4-85 Bond slip envelope of Wall4 ( Model 2, linkage elements)

2
1
1.6

1.2 t
I
A

E
E
u
0.8 1
.-
0 0.4
V)

-0.4

-0.8 -
O 1 2 3 4 5 6 7 8
Cyc le
Fig. 4-86 Bond slip envelope of Wall-4 ( Model 2, contact elements)
4 5 6
Cycle

Fig. 4-87 Bond slip at the south bottom of Wall-1


(Model 2, contact elements)

Fig. 4-88 Bond slip at the north bottom of Wall-1


(Model 2, contact elements)
Cycle

Fig. 4-89 Bond stress at the south bottom of Wall-1


(Model 2, contact elements)

4 5 6
Cycle
Fig. 4-90 Bond stress at the north bottom of the Wall-1
(Model 2, contact elements).
Fig. 4-91 Steel stress at the south bottom of Wall-1
(Model 2, contact elements)

O 1 2 3 4 5 6 7 8 9 10
Cycle
Fig. 4-92 Steel stress at the north bottom of Wall-1
(Model 2, contact elements)
O 1 4 5
Cycle

Fig. 4-93 Bond slip at the south bottom of Wall-2


(Model 2, contact elements)

4 5
Cycle

Fig. 4-94 Bond slip at the north bottom


(Model 2, contact elements)
- -

4 5
Cycle
Fig. 4-95 Bond stress at the south bottom of Wall-2
(Model 2, contact elements)

' 1
O 1 2 3 4 5 6 7 8 9
Cple
Fig. 4-96 Bond stress at the north bottom of Wall-2
(Model 2, contact elements)
-500 ' I
O 1 2 3 4 5 6 7 8 9
Cycle
Fig. 4-97 Steel stress at the south bottom of Wall-2
(Model 2, contact elements)

4 5
Cycle
Fig. 4-98 Steel stress at the north bottom of Wall-2
(Model 2, contact elernents)
4

Cycle
Fig. 4-99 Bond slip at the south bottom o f Wall-3
(Model2, contact elements)

Cycle

Fig. 4-100 Bond slip at the north bottom of Wall-3


(Mode1 2, contact elements)
4
Cycle
Fig. 4-101 Bond stress at the south bottom of Wall-3
(Model 2, contact elements)

-5 J

O 1 2 3 4 5 6 7 8
Cycle
Fig. 4-102 Bond stress at the north bottom of Wall-3
(Model 2, contact elements)
O 1 2 3 4 5 6 7 8
Cycle
Fig. 4-1 03 Steel stress at the south bottom of Wall-3
(Model 2, contact elements)

Fig. 4-104 Steel stress at the north bottom of Wall-3


(Model 2, contact elements)
Cycle
Fig. 4-105 Bond slip at the south bottom of Wall4
(Model 2, contact elements)

O 1 2 3 4 5 6 7
Cycle

Fig. 4-106 Bond slip at the north bottom of Wall4


(Model 2, contact elements)
Cycle
Fig. 4-107 Bond stress at the south bottom of Wall-4
(Mode1 2, contact elements)

Fig. 4-108 Bond stress at the north bottom of Wall4


(Model2, contact elements)
4
Cycle
Fig. 4-109 Steel stress at the south bottom o f Wall-4
(Mode1 2, contact elements)

4
Cycle
Fig. 4-1 10 Steel stress at the north bottom of W a l l 4
(Mode1 2, contact elements)
Fig. 4- 1 1 1 Deflection of Wall-1 after failure (assuming perfect bond)
Fig. 4- 1 12 Deflections of Wall-1 after failure (linkage elements, Mode1 2)
Fig. 4-1 13 Deflections of Wall-1 after failure (linkage elements, Mode1 3)
Fig. 4- 1 14 Defiections of Wall- l afier failure (linkage elements, Mode1 4)
Fig. 4-1 15 Deflections of Wall-1 afier failure (contact elements, Mode1 2)
Fig. 4- 1 16 Deflections of Wall- 1 after failure (contact elements, Mode1 3)
Fig. 4- 1 17 Deflections of Wall- 1 afier failure (contact elements, Mode1 4)
Fig. 4-1 18 Deflection of Wall-2 after failure (assuming perfect bond)
Fig. 4-1 19 Deflections of Wall-2 afier failure (linkage elements, Mode1 2)
Fig. 4-120 Deflections of Wall-2 afier failure (linkage elements, Mode1 3)
Fig. 4-1 2 1 Deflections of Wall-2 after failure (linkage elements, Mode1 4)
Fig. 4-1 22 Deflections of Wall-2 after failure (contact elements, Mode1 2)
Fig. 4-123 Deflections of Wall-2 after failure (contact elements, Mode1 3)
Fig. 4- 124 Deflections o f Wall-2 after failure (contact elements, Model4)
Fig. 4- 125 Deflection of Wall-3 after failure (assuming perfect bond)
Fig. 4-1 26 Deflections of Wall-3 after failure (linkage elements, Model2)
Fig. 4-127 Deflections of Wall-3 afier failure (linkage elements, Mode1 3)
Fig. 4-1 28 Deflections of Wall-3 afier failure (linkage elements, Mode1 4)
Fig. 4-1 29 Deflections of Wall-3 afier failure (contact elements, Model2)
Fig. 4-130 Deflections of Wall-3 after failure (contact elements, Mode1 3)
Fig. 4- 13 1 Deflections of Wall-3 afier failure (contact elements, Mode1 4)
Fig. 4- 1 3 2 Deflection of Wall-4 after failure (assuming perfect bond)
Fig. 4-1 33 Deflection of Wall-4 after failure (linkage element, Mode1 2)
Fig. 4-1 34 Deflections of Wall-4 afier failure (linkage elements, Mode1 3)
Fig. 4-1 35 Deflection o f Wall-4 after failure (linkage element, Mode1 4)
Fig. 1-136 Deflections of Wall-4 after failure (contact elements, Mode1 2)
Fig. 4-137 Deflections of Wall-4 afier failure (contact elements, Mode1 3)
Fig. 4-1 38 Deflections of Wall-4 after failure (contact elements, Mode1 4 )
CNAPTER 5
DISCUSSION AND CONCLUSIONS

5.1 Discussion
For the four examined walls, during the first two cycles (peak
deflections not more than 1 mm), bar slips were negligible and the shear
walls behaved as if there was perfect bond between the steel bars and the
surrounding concrete.
Large positive slips occwed above the bottom beam and large negative
slips occurred at the section from one-third to one-half the height of the
web due to large cracks in the tension zone of the web. When the outside
concrete of the bottom web was uncracked or when cracks were srnall, the
maximum slips occurred at the interfaces between outside tension steel
bars and the surrounding concrete near the bottom regions of the web. As
the cracks on the outside concrete of the bottom web became larger, the
maximum slips moved toward the interfaces between inside tension steel
bars and the surrounding concrete at the base of the web.
The differences between the perfect bond mode1 and the experimental
data increased with the increase in stifniess of the wdl. The larger the
stiffness of the wall, the more significant the difference.
The damage index D of the bond resistance due to cyclic loading was
small. It was very close to zero for many of the elements, and about 8% to
20% only for a small number of the elements. It should be emphasized that
the damage index D does not compute accurately in the models used in this
paper. Due to the lack of experimental data on the reduction of bond
resistance under unidirectional cyclic loading, it is assumed that the
damage of bond resistance occurs under reversed cyclic loading only.
Actually, the damage of bond resistance also occurs under unidirectional
cyclic loading. Although the loading history of al1 four walls is reversed
cyclic loading, the slips are unidirectionai cyclic for a certain number of
elements; for example, the elements in the horizontal middle of the web.
This factor is very important for predicting the load-deflection response
accurately when top deflections are large, because the bond resistance in
the outside web of the tension side is very small in this situation. To
analyze the reinforced concrete structures more accurately for large slips
under cyclic loading, it is recommended that experimental research be
undertaken in order to obtain test data under unidirectional cyclic loading.
It is noted that Wall-2 and Wall-3 achieved the sarne ultimate load
during testing, and Wall4 with a barbell-shaped cross section had a
ultimate load slightly higher than that of a rectangular section with the
same amount and detailing of web and boundary reinforcement. Choi
concluded fiom the above data that the horizontal hoops did not affect the
shear strength of the wall, and the flange of the cross section only aflected
the shear strength slightly. It is this author's contention that his conclusions
were incorrect. Because the loading histories were different, the responses
of Wall-2, Wall-3 and Wall4 are not comparable. Actually, the effect of
the reinforcement hoops is considered in the theoretical models by
enhancement of the bond resistance and concrete strength in the area of the
hoops, and the ultimate loads predicted by finite analysis were very close
to those observed in the experiments. This indicates that the hoops
influenced the response of the walls. As for the function of the web
flanges, they affected the response greatly when uncracked, and the effect
become smaller when cracked. However, they still had the efFect of
undertaking compression at the compressive side of the wall.
For al1 four shear walls modeled by bond-slip Models 2 to 4, the
analytically computed hysteretic responses showed the same trends as
measured experimentally, and the computed ultimate lateral loads
correlated very closely to those experimentally recorded, as shown in
Table 5-1. The stifniess of the walls is diminished by increasing the
number of cycles of reversed cyclic bading. Residual deflections upon
unloading are significant and cumulative. Excellent agreement of the
Models 2, 3 and 4 with test results is observed for al1 experimental walls.
There are only two apparent differences between analytical responses with
experimental results. Firstly; the experimental load-deflection curves show
more pinching in the hysteretic loops than those of finite element analysis.
Secondly, the al1 four walls failed one cycle earlier in theoretical analysis
than in experiment. These may be related to two factors. First, the
hysteretic models used for the compression and tension response of
concrete are preliminary, and currently under further development.
Secondly, the hysteretic models of the bond stress-slip relationship may be
inaccurate. It is noted that the bond stress-slip relationship shows wide
scatter in experimental investigations, even under rnonotonic loading. Of
course, the hysteretic mode1 of the bond stress-slip relationship shows
even greater scatter under cyclic loading. It is very dificult to predict bond
stresses and slips of a reinforced concrete structure accurately at present.
The darnage of bond resistance is underestimated due to ignoring the
reduction of bond resistance under unidirectional cyclic slips. It results in a
computed response that is stiffer than seen in the actual structure, therefore
the ultimate loads occur earlier and the structures fail earlier than obsewed
in the experiment.
Table 5-1 Cornparison of ultimate loads of walls
fiom test data and FE analysis ( kN )
1 FE Analysis 1 Test data 1
Wall ink ka& Element Contact Element
Model 2 Model 4- Model 2 Mode14
Wall-1 load 408 403 404 403 400
FE/Test 1.020 1.008 1.O10 1.008
Wall-2 load 643 624 64 1 633 640
FE/Test 1.O08 0.974 1.003 0.989
Wall-3 load 64 1 627 652 632 640
FE/Test 1.O02 0.980 1.019 0.988
Wall4 load 692 69 1 698 696 680
FE/Test 1.018 1.017 1.O26 1.O24
Total FE/Test 1.012 0.995 1.015 1 1.002

Model 2 gives the closest fit to the experimental data. Model 4 also
gives excellent agreement with the data observed in experiment. The
analyses using Model 1 are convergent only when the slips are very small;
this model is not recommended for theoretical modeling of the bond stress-
slip relationship.
The analytical responses computed using contact elements match
slightly better with the experimental data than those cdculated using
linkage elements. The linkage element is not well behaved because its
constant displacement function can not model the non-constant slip field.
The contact element, with a linear displacement function, c m model the
linear slip field exactly. Also, it gives a better approximation of the
nonlinear slip field than does the lnkage element.

5.2. Conclusions
This thesis studied the bond stress-slip relationship under both
rnonotonic loading and cyclic loading. It presented formulations for four
types of bond-slip models and two types of bond-slip elements for use in
finite element analysis. These formulations were implemented into the
finite element program TRIX99. The thesis compared results kom
different analytical models and bond elements for a series of four
reinforced concrete shear walls. The bond-slip Model 2 (by Eligehausen et
al) and Model 4 (proposed by the author of this thesis), in conjunction
with contact elements, is recommended for use in analysis of reinforced
concrete structures.
The thesis documents a reliable analytical tool. The results of the
validation studies demonstrate that the analytical models are capable of
reproducing most of the important aspects of the measured cyclic
responses of reinforced concrete walls with a variety of cyclic loading
histones and wall configurations. The degradation of bond resistance is
included in the finite element program.
The analytical results obtained using the modified program T m 9 9
show excellent correlation with the experimental measurements of the
reinforced concrete walls with different cross sections and reinforcement
arrangements. The prograrn is able to successfully compute the load-
deflection values of the walls under cyclic loading. The proposed finite
element formulation provides an efficient method for evaluating the
rnonotonic and cyclic response of the reinforced concrete structures.
Although this study is restricted to 2-D prablems, the bond-slip element
can easily be adapted to 3-D problems and added to 3-D finite element
programs.
Due to the lack of experimental data on t!e reduction of bond resistance
under unidirectional cyclic loading, the darnage of bond resistance is
considered only under reversed cyclic loading in this thesis. To analyze the
reinforced concrete structures more accurately for large slips under cyclic
loading, more experimental research is needed in order to obtain the
unidirectional experimental data.
REFERENCES

AC1 Committee 408,1966. "Bond Stress- The State of The Art", AC1
Journal, Vol. 63 ,No. 11 ,November, pp. 1161-1188.
AC1 Committee 408, 1991. "Abstract of: State-of-the-art-report: Bond under
Cyclic Loads", AC1 Materials Journal, Vol. 88, No. 6, Nov.-Dec., pp. 669-
673.
Abrishami, H. H., and Mitchell, D., 1992. "Simulation of Uniform Bond
Stress," AC1 Material Joumal, Vol. 89, No. 2, pp. 161-168.
Abrishami, H. H. and Mitchell, D., 1996. "Analysis of Bond Stress
Distributions in Pull-out Specimens", Journal of Structural Engineering,
Vol. 122, No. 3, March , pp. 255-26 1.
Allwood, R. J. and Bajarwan, A. A., 1996. "Modeling Nonlinear Bond-Slip
Behavior for Finite Element Analysis of Reinf'orced Concrete Structures",
AC1 Structural Journal, VOL 93, No. 5, September-October, pp. 538-544.
7
Alsiwat, J. M. and Saatcioglu, M., 1992. "Reinfocement Anchorage Slip
under Monotonic Loading", Journal of ~hucturalEngineering, ASCE, Vol.
118, No. 9, Sept., pp. 242 1-2438.
Ayoub, A. and Filippou, F. F., 1999. "Mixed Formulation of Bond-Slip
Problems Under Cyclic Loads", Journal of Structnial Engineering, Vol.
125, No. 6, June, pp. 661-671.
Ayyub, B. M., Mutairi, N. A. and Chang, P., 1994. "Bond Strength of
Welded Wire Fabric in Concrete Bridge Decks", Journal of Structural
Engineering, Vol. 120, No. 8, August, pp. 2520-2531.
Azizinzmini, A., Stark, M., Roller, J. J., and Ghosh, S.K., 1993. "Bond
Performance of Reinforcing Bars Embedded in High-Strength Concrete",
AC1 Structural Journal, Vol. 90, No. 5, Sep.-Oct., pp. 554-56 1.

243
Chajes, M. J., Finch, W. W., Januszka, T. F., and Thomson, T. A. , 1996.
"Bond and Force Transfer of Composite Material Plates Bonded to
Concrete", AC1 Structural Journal, Vol. 93, No. 2, March-Apnl, pp. 208-
218.
Channakeshava, C. and Sundara, R. I., 1988. "Elasto-Plastic Cracking
Analysis of Reinforced Concrete ", Journal of Structural Engineering, Vol.
114, No. 11, Nov., pp. 2421.2438.
Charles K. K., 1997. "Relationship of Bond Stress, Steel Stress, and Slip in
Reinforced Concrete", Joumal of Structural Engineering, Vol. 123, No. 1,
Jan., pp. 79-85.
Cook, R. A., Doerr, G. T. and Klingner, R. E., 1993. "Bond Stress Mode1
for Design of Adhesive Anchors", AC1 Structural Journal, Vol. 90, No. 5,
September-October, pp. 5 14-524.
Edwards, A. D., and Yannopoulos, P. J., 1979. "Local Bond-Stress to Slip
Relationships for Hot Rolled Deformed Bars and Mild Steel Plain Bars,"
AC1 Journal, Vol. 76, No. 3, March, 1979, pp. 405-420.
Eligeliausen, R., Popov, E., and Bertero, V., 1983. "Local Bond Stress-Slip
Relationship of Deformed Bars under Generalized Excitations", Report No.
UCBEERC-83/23, Earthquake Engineering Center, University of
California, Berkeley.
Esfahani, M. R., and Rangan, B. v., 1998. "Bond Between Normal
Strength and High-Strength Concrete (HSC) and Reinforcing Bars in
Splices in Beams", AC1 Structural Journal, Vol. 95, No. 3, May-June, pp.
272-280.
Esfahani, M. R., and Rangan, B. V., 1998. "Local Bond Strength of
Reinforcing Bars in Normal Strength and High-Strength Concrete", AC1
Structural Joumal, Vol. 95, No. 2, March-April, pp. 96-106.
Fillippou, F. C., Popov, E. P. F. and Bertero, V., 1983. "Modeling of
Reinforced Concrete Joints under Cydic Excitations," Vol. 109, No. 1 1,
Nov., pp. 2666-2684.
Fillippou, F. C., 1986. "A Simple Model for Reinforcing Bar Anchorage
under Cyclic Excitations", Journal Structural Engineering, ASCE, Vol. 112,
No. 7, July, pp. 1639-1659 .
Fillippou, F. C . , Popov, E. P. F. and Bertero, V., 1983. "Effects of Bond
Deterioration on Hysteretic Behavior of Reinforced Concrete Joints",
UCBEERC-83/19, Earthquake Engineering Research Center, University of
Cd ifomia, Berkeley.
Filippou, F. C., Popov, E. P. F., and Bertero, V., 1986. "Analytical Studies
of Hysteretic Behavior of R C Joints7', Journal of Structural Engineering,
ASCE, Vol. 112, No. 7, July, pp. 16054622.
Gajer, G., and Dux, P. F., 1990. "Crack Band Based Model for FEM
Analysis of Concrete Structures", Journal of Structural Engineering, Vol.
116, No. 6, June, 1696-1714.
Gambarova, P. G., and Rosati, G. P., and Zasso, B. , 1989. "Steel-to-
Concrete Bond After Concrete Spiitting: Test Results", Materials and
Structures, Vol. 22, pp. 35-47.
Giuriani, E., Plinari, G. and Schumm, C., 1991. "Role of Stimps and
Residual Tensile Strength of Cracked Concrete on Bond", Journal of
Structural Engineering, ASCE,Vol. 117, No. 1, Jan., pp. 1-18.
Harnoush, S. A. and Salami, M. R., 1990. "Interfacial Strain Energy
Release Rate of Fiber Reinforced Concrete Based on Bond Stress-Slip
Relationship", AC1 Structural Journal, Vol. 87, No. 6, November-
December, pp. 678-686.
Harajli, M. H. and Mukaddam, M. A., 1988. "Slip of Steel Bars in Concrete
Joints under Cyclic Loading", Journal of Structural Engineering, ASCE,
Vol. 114, No. 9, Sept., pp. 20 17-2035.
Hayashi, S and Kokusho, S., 1985. ""BondBehavior in the Neighborhood of
the Crack", Finite Element Analysis of Reinforced Concrete Structures,
ASCE, pp. 364-373.
Hota, S., and Naaman, A. E., 1997. "Bond Stress-Slip Response of
Reinforcing Bars Embedded in FRC Matrices under Monotonie and Cyclic
Loading", AC1 Structural Journal, Vol. 94, No. 5, September-October , pp.
525-537.
Ingraffea, A. R., Gerstle, W. H., Gergely, P. and Saouma, V., 1984.
"Fracture Mechanics of Bond in Reinforced Concrete", Journal of the
Structural Division, ASCE ,Vol. 110, No. 4, Apnl ,pp. 87 1-890.
Inoue, N., Koshika, N., and Suzuki, N., 1985. "Analysis of Shear Wall
Based on Collins Panel Test", Finite Element Analysis of Reinforced
Concrete Structures, ASCE, pp. 288-300.
Ismail, M. A. F., and Jirsa, J. O., 1972. "Bond Deterioration of Reinforced
Concrete Subject to Low Cycle Loads", AC1 Joumal, Vol. 69, No. 6, June,
pp. 334-343.
Jiang , D. H., Shah, S. P. and Anderian, A. T., 1984. " Study of the
Transfer of T e n d e Forces by Bond", AC1 Journal, Vol. 81, No. 3, May-
June, pp. 251-259.
Kent, D. C . and Park, R., 1971. "Flexural Members with Confined
Concrete", Journal of the Structural Division, ASCE, Vol. 97, No. 7, July,
pp. 1969-1990.
Keuser, M., Mehlhom, G. and Cornelius, V., 1983. "Bond Between
Prestressed Steel and Concrete - Computer Analysis Using ADINA",
Computer & Structures ,Vol. 17, No. 5, pp. 669-676.
Keuser, M. and Mehlhom, G., 1987. "Finite Element Models for Bond
Problems", Journal of the Structural Division, ASCE, Vol. 113, No. 10,
October, pp. 2 160-2173.
Lutz, L. A. and Gergely, P., 1967. "Mechanics of Bond and Slip of
Defonned Bars in Concrete", AC1 Joumal , Vol. 64 , No. 11, November
pp. 711-721.
Mehlhom, G. and Keuser, M. 1985. "Isoparametnc Contact Elements for
Analysis of Reinforced Concrete Structures", Finite Element Analysis of
Reinforced Concrete Structures, ASCE, pp. 329-347.
Mina, S. M., and Houde, J., 1979. "Study of Bond-Slip Relationships in
Reinforced Concrete", AC1 Joumal, Vol. 76, No. 1, January, pp. 19- 46.
Morita, S., and Fujii, S.," Bond-slip Models In Finite Element Analysis",
Finite Element Analysis of Reinforced Concrete Structures, ASCE, 1985,
pp. 348-363.
Mirza, S. A., 1987. "Bond Strength Statistics of Flexural Reinforcement on
Concrete Beams", AC1 Structural Journal, Vol. 84, No. 5, September-
October, pp 383-391.
Monti, G., Fillippou, F. C. and Spacone, E., 1997. "A Finite Element for
Anchored Bars under Cyclic Load Reversais", Journal Structural
Engineering, ASCE, Vol. 123, No. 5, May, pp. 614-623.
Morsi, M. E., Kianoush, M .R. and Tso, W. K., 1998. ''Nonlinear Analysis
of Cyclically Loaded Reinforced Concrete Structures", AC1 Structural
Journal, Vol. 95, No. 6, Nov.-Dec., pp. 725-739.
Naaman, A. E., Namur, G. G., Alwan, G. M. and Najm, H. S., 1991. "Fiber
Pullout and Bond Slip I : Analytical Study", Joumal of Structural
Engineering, ASCE, Vol. 117, No. 9, Sept., pp. 276992790.
Naaman, A. E., Namur, G. G., Alwan, G. M., and Najm, H. S., 1991. "Fiber
Pullout and Bond Slip II : Experimental Validation", Journal of Structural
Engineering, ASCE, Vol. 117, No. 9, Sept., pp. 279 1-2800.
Narnmur, G. J. and Naaman, A. E., 1989. ""Bond Stress Mode1 for Fiber
Reinforced Concrete Based on Bond Stress-Slip Relationship", AC1
Materials Journal , Vol. 86, No. 1, January-Febmary, pp. 45-57.
Neuenhofer, A., and Fillippou, F. C., 1997. "Evaluation of Nonlinear Frame
Finite Element Models", Journal Structural Engineering, ASCE, vol. 123,
No. 7, July, pp. 958-966.
Ngo, D. and Scordelis, A. C., 1967. "Finite Element Analysis of Reinforced
Concrete Beams", AC1 Joumal, Vol. 64, No. 3, March, pp. 152- 163.
Nilson, A. H., 1968. 'Nonlinear Analysis of Reinforced Concrete by the
Finite Element Method", AC1 Journal Vol. 65, No. 9, September, pp. 757-
766.
Nilson, A. H., 1972. " Intemal Measurement of Bond Slip ," AC1 Journal,
Vol. 69, No. 7, July, pp. 439-441.
Niwa, J., Chou, L. L., Shima, H., and Okamura, H., 1985, 'Wonlinear
Spring Element for Bond-Slip Relationship of a Deformed Bar", Finite
Element Analysis of Reinforced Concrete Structures, ASCE, pp. 374-383
Perry, E. S. and Thompson, J. N., 1968. "Bond Stress Distribution on
Reinforcing Steel in Beams and Pull-out Specimens", AC1 Journal, Vol.
65, No. 9, September, pp. 865-874.
Pochanart, S. and Hannon, T., 1989. "Bond-Slip Model for Generalized
Excitations Including Fatigue", AC1 Materials Journal, Vol. 86, No. 5,
September-October, pp. 465-474.
Rhem, G., and Eligehausen, R., 1979. "Bond of Ribbed Bars under High
Cycle Repeated Loads", AC1 Journal, Vol. 76, No. 2, Feb., pp. 297-309.
Russo, G., Zingone, G. ar.d Romano, F., 1990. "Analytical Solution for
Bond-Slip of Reinforced Bars in R. C. Joints", Journal of Structura1
Engineering, ASCE, Vol. 116, No. 2, Feb., pp. 336-355.
Russo, G., Zingone, G. and Romano, F., 1990. "Analytical Solution for
Bond-Slip of Reinforcing Bars in R. C. Joints", Journal of Structural
Engineering, ASCE, Vol. 116, No. 2, Feb., pp. 336-355.
Schnobrich, W. C., 1985. "The role of Finite Element Analysis of
Reinforced Concrete Structures", Finite Element Analysis of Reinforced
Concrete Structures, ASCE, pp. 1- 12.
Scordelis, A. C., 1991. "Past, Present and Future Developments", Finite
Element Analysis of Reinforced Concrete Structures II, ASCE, pp. 656-
666.
Scott, R.H., and Gill, P. A. T., 1987. "Short-Term Distribution of Strain and
Bond Stress dong Tension Reinforcement", Journal of Stmctural
Engineering, Vol. 65B(2), June, pp. 39-43.
Scott, B. D., Park, R. and Priestley, M. J. N., 1982. "Stress-Strain Behavior
of Concrete Confined by Overlapping Hooks at Low and High Strain Rates",
AC1 Journal, Vol. 79, No. 1, Jan.-Feb., pp. 13-27.
Shirai, N., 1985, "Applicability of Smeared-Type Reinforced Concrete
Model Based upon Anisotropic Constitutive Law", Finite Element Analysis
of Reinforced Concrete Structures, ASCE, pp. 384-400
Sittipunt, C. and Wood, S. L., 1995. "Muence of Web Reinforcement on
the Cyclic Response of Structural Walls", AC1 Structural Journal, Vol. 92,
No. 6, Nov.-Dec., pp. 745-756.
Somayaji, S. and Shah, S. P., 1981. "Bond Stress Versus Slip Relationship
and Cracking Response of Tension Members", AC1 Journal, Vol. 78, No.
3, May-June, pp. 217-225.
Soretz, S. and Holzenbein, H., 1979. " Influence of Rib Dimensions of
Reinforcing Bars on Bond and Bendability", AC1 Journal, Vol. 76, No. 1,
January, pp. 111-126.
Soric, Z. and Tallin, L. G., 1989. "Bond Stress /Deformation in Pull-out
Masonry Specirnens", Journal of Structural Engineering, ASCE, Vol. 115,
No. 10, Oct., pp. 2588-2602.
Soroushian, P., Obasaki, K. and Marikunte, S. , 1991. "Analytical Modeling
of Bonded Bars under Cyclic Loads", Journal of Structural Engineering,
ASCE, Vol. 117, No. 1, Jan., pp. 48-60.
Soroushian ,P., and Choi, K. B., 1989. "Local Bond of Deformed Bars with
Different Diameters in Confined Concrete", AC1 Structural Journal, Vol. 86,
NO.2, pp. 217-222.
Tassios, T. P. and Yannopoulos, P. J., 1981. "Analytical Studies on
Rzinforced Concrete Members Under Cyclic Loading Based on Bond
Stress-Slip Relationships", AC1 Joumal, Vol. 78, No. 3, May-June, pp.
206-2 16.
Tassios, T. P. and Koroneos, E. G., 1984. "Local Bond-Slip Relationships
by Means of the Moire Method, AC1 Joumal, Vol. 81, No. 1, January-
February, pp. 27-34.
Tsubaki, T., 1985. "Shear Transfer Across Interface in Boundary Element
Analysis of Concrete Structures", Finite Element Analysis of Reinforced
Concrete Structures, ASCE, pp. 265-276.
Vecchio, F. J., 1999. ccTowards Cyclic Load Modeling of Reinforced
Concrete", AC1 Stnictnial Journal, Vol. 96, No. 2, March-Apd, pp. 193-
202.
Vecchio, F. J., 1992. "Finite Element Modeling of Concrete Expansion and
Confinement", Journal of Structural Engineering, ASCE, Vol. 118, No. 9,
Sept., pp. 2390-2398.
Yang, S. and Chen, J., 1988. "Bond Slip and Crack Width Calculations of
Tension Members", AC1 Structural Journal, Vol. 85, No. 4, July-August ,
pp. 4 14-422
Yankelevsky, D. Z., 1985. "Bond Action Between Concrete and Deformed
Bar- A New Model", AC1 Journal, Vol. 82, No. 2, March - April, pp. 154-
161.
Yankelevsky, D. Z., 1985. "New Finite Element for Bond-Slip Analysis",
Vol. 111, No. 7, July, pp. 1535-1 542.

Das könnte Ihnen auch gefallen