Professional Documents
Culture Documents
32-Wu, M.H.Q (2010) - Tension Stiffening in Reinforced Concrete - Instantaneous and Time-Dependent
32-Wu, M.H.Q (2010) - Tension Stiffening in Reinforced Concrete - Instantaneous and Time-Dependent
time-dependent behaviour
Author:
Wu, Mark Han Qing
Publication Date:
2010
DOI:
https://doi.org/10.26190/unsworks/23066
License:
https://creativecommons.org/licenses/by-nc-nd/3.0/au/
Link to license to see what you are allowed to do with this resource.
February 2010
UNSW
THE UNIVERSITY OF NEW SOUTH WALES, SYDNEY, AUSTRALIA
I hereby declare that this submission is my own work and that, to the best of my knowledge
and belief, it contains no materials previously published or written by another person, nor
material which to a substantial extent has been accepted for the award of any other degree or
diploma at UNSW or any other institute of higher learning, except where due
acknowledgment is made in the thesis. Any contribution made to the research by others, with
I also declare that the intellectual content of this thesis is the product of my own work, except
to the extent that assistance from others in the project’s design and conception and linguistic
expression is acknowledged.
______
Mark Han Qing Wu
Dedicated to my Lord and Saviour:
Jesus Christ
The research reported in this thesis was undertaken in the School of Civil and Environmental
Engineering at University of New South Wales, Sydney. Sincere thanks is given to Professor
Ian Gilbert, who has supervised me through the entire research period with his gentle heart
and great suggestions. He has also cultivated my ability to think carefully about the academic
program. I also express my deep gratitude to Professor Stephen Foster, for his constructive
help in the computational modeling aspects of this study, without which this thesis would not
I would also like to thank the staff at the Heavy Structure Laboratory at the Randwick sub-
campus of UNSW, including the staff Ron Moncay, John Gilbert and Frank Scharfe, all of
whom have offered great and faithful assistance, so that the experimental program could be
carried out successfully. I am also grateful to the administrative staff at the School of Civil
and Environmental Engineering, for their consistent help over this 3.5 years academic period.
Finally I would like to thank my parents, who have always showed their steadfast love and
1. Wu, H. Q., and Gilbert, R. I. (2009a) "Modeling short-term tension stiffening in reinforced
31(10), 2380-2391.
reinforced concrete slabs." Proceedings of the 24th Biennial Conference of the Concrete
9(2), 151-158.
ABSTRACT
The research presented in this thesis is concerned with the time-dependent behaviour of
reinforced concrete and the mechanisms of tension stiffening. The research program
comprises both experimental investigation and numerical modeling. The aims are to provide a
comprehensive understanding about tension stiffening under short-term and long-term loading.
Of particular interests are the effects of concrete creep and shrinkage on the change of tension
The experimental program involved testing axially loaded reinforced concrete prisms and
simply-supported reinforced concrete beams and slabs under short-term and long-term
loadings. The experimental results have demonstrated that tension stiffening is greatly
influenced by the drying shrinkage of concrete, the extent of cracking and the deterioration
with time of the bond between the reinforcement and the concrete.
A two-dimensional continuum-based finite element model has been developed. The model
incorporates instantaneous and time-dependent constitutive laws for concrete, steel and
importantly the bond interface between them. A shrinkage-related bond model has been
proposed to accurately model the loss of tension stiffening under long-term loading.
The finite element model has been used to simulate the response of the test specimens and the
numerical results are in close agreement with the experimental results. A parametric study has
also been undertaken using the finite element model; the effects of different parameters
(concrete cover, reinforcing ratio, etc.) on time-dependent tension stiffening have been
TABLE OF CONTENTS
CHAPTER 1 INTRODUCTION 1
1.1 GENERAL ......................................................................................................................... 1
1.2 OBJECTIVES OF THE CURRENT STUDY.................................................................... 5
1.3 THESIS STRUCTURE...................................................................................................... 6
3.3.3 Time-dependent tension stiffening in the cracked reinforced concrete beam .................. 69
REFERENCES 312
VI
LIST OF FIGURES
Figure 4-23: Concrete force distribution of STN12 around crack NO. 3. ............................. 107
Figure 4-24: Steel stress versus maximum crack width (uniaxial tension short-term
tests). ..................................................................................................................... 108
Figure 4-25: Average strain versus load response of LTN12A. ............................................ 110
Figure 4-26: Average strain versus load response of LTN12C.............................................. 111
Figure 4-27: Initial and final concrete and steel stresses LTN12A (P = 40 kN).................... 111
Figure 4-28: Initial and final concrete and steel stresses LTN12C (P = 40 kN). ................... 112
Figure 4-29: Crack numbers and locations (LTN12A). ......................................................... 113
Figure 4-30: Crack numbers and locations (LTN12C). ......................................................... 113
Figure 4-31: Average strain versus load response of LTN12B.............................................. 115
Figure 4-32: Average strain versus load response of LTN12D. ............................................ 116
Figure 4-33: Initial and final concrete and steel stresses LTN12B (P = 20 kN). ................... 116
Figure 4-34: Initial and final concrete and steel stresses LTN12D (P = 20 kN).................... 117
Figure 4-35: Crack numbers and crack locations (LTN12B)................................................. 117
Figure 4-36: Crack numbers and crack locations (LTN12D). ............................................... 118
Figure 4-37: Tension stiffening strain versus time for long-term tested prisms. ................... 119
Figure 4-38: Maximum crack width versus concrete age of LTN12A and LTN12B. ........... 120
Figure 5-1: Dimension and detail of BSTN2-16 and BSTS2-16. .......................................... 123
Figure 5-2: Dimension and detail of BSTN3-16 and BSTS3-16. .......................................... 124
Figure 5-3: Dimension and detail of all slabs. ....................................................................... 124
Figure 5-4: Set up for short-term flexural tests. ..................................................................... 126
Figure 5-5: Set up for long-term flexural tests....................................................................... 127
Figure 5-6: High-resolution laser transducers underneath the short-term testing slabs......... 128
Figure 5-7: Dial gauges underneath the long-term testing slabs............................................ 129
Figure 5-8: Casting and curing of the slabs............................................................................ 130
Figure 5-9: Creep coefficients for the flexural tests............................................................... 133
Figure 5-10: shrinkage strain development for the flexural tests........................................... 134
Figure 5-11: Moment versus mid-span deflection (BSTN2-16) ............................................ 136
Figure 5-12: Moment versus mid-span deflection (BSTN3-16) ............................................ 136
Figure 5-13: Moment versus mid-span deflection (SSTN4-12)............................................. 137
Figure 5-14: Variation of forces in steel at different stages of loading (BSTN2-16). ........... 140
Figure 5-15: Variation of forces in steel at different stages of loading (BSTN3-16). ........... 141
IX
Figure 5-16: Variation of forces in steel at different stages of loading (SSTN 4-12)............ 142
Figure 5-17: Crack numbers and crack locations (BSTN2-16).............................................. 143
Figure 5-18: Crack numbers and crack locations (BSTN3-16).............................................. 144
Figure 5-19: Crack numbers and crack locations (SSTN-4-12)............................................. 144
Figure 5-20: Moment versus mid-span deflection for BSTS2-16.......................................... 147
Figure 5-21: Moment versus mid-span deflection for BSTS3-16.......................................... 148
Figure 5-22: Moment versus mid-span deflection for SSTS4-12. ......................................... 148
Figure 5-23: Variation of forces in steel at different stages of loading (BSTS2-16). ............ 150
Figure 5-24: Variation of forces in steel at different stages of loading (BSTS3-16). ............ 151
Figure 5-25: Variation of forces in steel at different stages of loading (SSTS4-12). ............ 152
Figure 5-26: Crack numbers and locations (BSTS2-16)........................................................ 154
Figure 5-27: Crack numbers and locations (BSTS3-16)........................................................ 154
Figure 5-28: Crack numbers and locations (SSTS4-12) ........................................................ 154
Figure 5-29: The effects of initial shrinkage on the tension stiffening deflections................ 157
Figure 5-30: The effects of reinforcing ratio on the tension stiffening deflections. .............. 157
Figure 5-31: The effects of initial shrinkage on the maximum crack width of flexural
members. ............................................................................................................... 159
Figure 5-32: The effects of reinforcing ratio on the maximum crack width of flexural
members. ............................................................................................................... 159
Figure 5-33: The change of mid-span deflection with time (SLTN4-12A) ........................... 161
Figure 5-34˖ the change of mid-span deflection with time (SLTN4-12B).......................... 161
Figure 5-35: The change of tensile bar forces with time (SLTN4-12A and SLTN4-
1B)......................................................................................................................... 163
Figure 5-36: Crack numbers and locations (SLTN4-12A)..................................................... 164
Figure 5-37: Crack numbers and locations (SLTN4-12B)..................................................... 164
Figure 6-1: Four-node plane isoparametric element. ............................................................. 169
Figure 6-2: Biaxial strength envelop of Foster and Marti (2003). ......................................... 173
Figure 6-3: stress-strain relationship of concrete ................................................................... 174
Figure 6-4: Discretization of continuous retardation spectrum of solidifying
constituent. ............................................................................................................ 179
Figure 6-5: Two node truss element....................................................................................... 183
Figure 6-6: Stress-strain relationship for the truss element.................................................... 185
X
Figure 7-20: Tensile strain contours for s1-a and s2-a........................................................... 233
Figure 7-21: Bond stress distributions for s1-a and s2-a........................................................ 233
Figure 7-22: Longitudinal (a) and cross-sectional (b) reinforcement of the test beams;
notation of cross-section (c) (Gribniak 2009). ...................................................... 234
Figure 7-23: Set-up of creep tests and variation of creep coefficient with time .................... 237
Figure 7-24: Moment versus mid-span deflection response of s-1 and s-1R......................... 239
Figure 7-25: Stress and strain distributions at the mid-span sections of Gribniak’s
first series beams just before loading (FEM). ....................................................... 241
Figure 7-26: Tension stiffening deflection versus mid-span moment of Gribniak’s
first series beams (experimental results). .............................................................. 242
Figure 8-1: Construction of beam specimens in the numerical experiment........................... 247
Figure 8-2: Finite element model of beam specimens in the numerical experiment. ............ 248
Figure 8-3: Construction of slab specimens in the numerical experiment. ............................ 249
Figure 8-4: Finite element model of slab specimens in the numerical experiment................ 249
Figure 8-5: Instantaneous moment versus mid-span deflection responses of series A
short-term specimens............................................................................................. 252
Figure 8-6: Tension stiffening deflection versus mid-span moment responses of
series A short-term specimens............................................................................... 253
Figure 8-7: Mid-span deflection versus time responses of series A long-term
specimens. ............................................................................................................. 254
Figure 8-8: Instantaneous moment verse mid-span deflection responses of series B
short-term specimens............................................................................................. 257
Figure 8-9: Tensile strain contour of series B short-term specimens at M = 30kN.m. .......... 257
Figure 8-10: Tension stiffening deflection verse mid-span moment responses of
series B short-term specimens............................................................................... 258
Figure 8-11: Mid-span deflection versus time responses of series B long-term
specimens. ............................................................................................................. 259
Figure 8-12: Instantaneous moment verse mid-span deflection responses of series C
short-term specimens............................................................................................. 261
Figure 8-13: Tension stiffening deflection verse mid-span moment responses of
series C short-term specimens............................................................................... 262
XII
LIST OF TABLES
Table 2-1: Parameters defined in CEB-FIP bond model (CEB-FIP 1997) .............................. 42
Table 4-1: Testing procedure of the uniaxial tension test specimens. ..................................... 80
Table 4-2: Reinforcing bar properties. ..................................................................................... 83
Table 4-3: Instantaneous concrete properties for uniaxial tension tests................................... 85
Table 4-4: Measured loads and strains for STN12................................................................... 89
Table 4-5: Measured loads and strains for STN16................................................................... 89
Table 4-6: Crack widths at different loading stage (STN12). .................................................. 94
Table 4-7: crack widths at different loading stage (STN16) .................................................... 95
Table 4-8: Measured loads and strains for STS12. .................................................................. 98
Table 4-9: Measured loads and strains for STS16. .................................................................. 98
Table 4-10: Crack widths at different loading stage (STS12)................................................ 103
Table 4-11: Crack widths at different loading stage (STS16)................................................ 103
Table 4-12: Crack width at different stages (LTN12A) ......................................................... 113
Table 4-13: Crack width at different stages (LTN12C) ......................................................... 113
Table 4-14: Crack width at different stages (LTN12B). ........................................................ 118
Table 4-15: Crack width at different stages (LTN12D) ......................................................... 118
Table 5-1: Testing procedure of all the specimens ................................................................ 131
Table 5-2: Instantaneous properties of concrete (flexural specimens)................................... 132
Table 5-3: Deflections of short-term non-shrunk flexural members...................................... 137
Table 5-4: Average tensile forces in the concrete and the steel over the constant
moment region (BSTN2-16, BSTN3-16 and SSTN4-12)..................................... 139
Table 5-5: Crack width at different stages (BSTN2-16) ........................................................ 145
Table 5-6: Crack width at different stages (BSTN3-16) ........................................................ 145
Table 5-7: Crack width at different stages (SSTN4-12) ........................................................ 145
Table 5-8: Deflections of short-term shrunk flexural members. ............................................ 149
Table 5-9: Average tensile forces in the concrete and the steel over the constant
moment region (BSTS2-16, BSTS3-16 and SSTS4-12)....................................... 153
Table 5-10: Crack width at different stages (BSTS2-16)....................................................... 155
XV
1.1 GENERAL
Deflection and crack control are both critical design requirements that must be considered in
the design of reinforced concrete structures at the serviceability limit states. They are often the
critical design considerations, especially for lightly reinforced flexural members, such as one-
way and two-way slabs. With the advent of higher strength materials, thinner slabs and longer
spans are possible, but the slenderness of modern slabs is limited by the deflection and
cracking behaviour under service loads. Structural design engineers need a sound
understanding of the interaction between the concrete and the reinforcement under service
loads and must have access to reliable methods for including the effects of concrete cracking
and the time-dependent deformation caused by creep and shrinkage in their calculations.
The prediction of deflection and crack widths of flexural members is complicated by the
difficulties in estimating the time-dependent stiffness after cracking. After the concrete first
cracks, the load-deformation response of the member becomes nonlinear. The deflection and
the crack widths increase significantly with increasing load and with increasing time after
loading. This decrement of structural stiffness and the corresponding increment of deflection
When a plain concrete element is subjected to tension, cracking occurs when the tensile stress
in the concrete reaches the tensile strength of concrete. After cracking, the tensile force
tensile force can be carried across the crack by friction arising from aggregate interlock.
After cracking in a reinforced concrete element, the tensile force is carried by the
reinforcement crossing the crack. Either side of the crack, the tensile force in the
Between the cracks in a reinforced concrete element, the tensile stress in the reinforcement
varies with part of the tensile force being transferred to the tensile concrete by bond, resulting
in the phenomenon known as tension stiffening. The bond-slip relationship between the steel
and the concrete greatly affects the transfer of tension to the concrete and the contribution of
the tensile concrete to the member’s stiffness. As a consequence, therefore, the bond-slip
relationship plays an important role in the determination of service load behaviour, including
the spacing of cracks, the crack widths, and the deflection, both in the short-term and in the
long-term.
Tension stiffening has been defined variously as the contribution of the intact concrete
between the primary cracks to the stiffness of the member or the ability of the intact concrete
between the primary cracks to carry part of the resultant tensile force. It is particularly
significant in relatively lightly reinforced members, where the actual stiffness may be several
times larger than the stiffness calculated on the basis of fully-cracked cross-sections, where
the tensile concrete is ignored and only the embedded tensile reinforcement is taken into
ȡ<0.003), tension stiffening may be responsible for more than 50% of the total stiffness of the
member ˄ Gilbert 2007 ˅ . It is generally agreed that tension stiffening decreases with
increasing external force. However, the decay of tension stiffening is not very well understood,
particularly the change in tension stiffening with time due to the combined effects of time-
dependent cracking, bond deterioration, tensile creep in the concrete and drying shrinkage. It
is generally agreed that tension stiffening reduces with time to about 50% of its initial value,
but this has not been satisfactorily confirmed and the magnitude and rate of change of tension
reinforced concrete elements have been reported in the literature. In the extensive
experimental program conducted by Beeby and Scott (2002), a number of axially loaded
reinforced concrete prisms were tested subjected to sustained service loads, and the concrete
and steel strains were measured with time. However, they did not measure the concrete drying
shrinkage both before and during the period of sustained loading and so a reliable
explanation for the decay of tension stiffening with time is not convincing. Drying shrinkage
crucial consideration in the study of the change in tension stiffening with time (Bischoff 2001).
stiffening in the analysis of reinforced concrete structures. Due to the lack of understanding of
the time-dependent behaviour of tension stiffening, most of these approaches try to model
tension stiffening based on experience. For example, Eurocode 2 (1992) proposes a method
based on using a distribution factor to obtain the effective stiffness and gives reduction factors
4
(0.5) to account for the decrease of bond with time on the decay of tension stiffening.
Therefore, it is suggested by the author that a more accurate numerical model must be
developed to predict time-dependent tension stiffening. The method must take appropriate
In the finite element context, tension stiffening is often modeled using a smeared crack
unloading branch after cracking (Barros et al. 2001, Ebead and Marzouk 2005, Gilbert and
Warner 1978, Guptam and Maestrini 1990, Lin and Scordelis 1975, Prakhya and Morley 1990,
Nayal and Rasheed 2006). Alternatively, tension stiffening has been included by adjusting the
constitutive relationship for the tensile reinforcement (Gilbert and Warner 1978). An
model the bond stress on the interface of steel and concrete as a function of slip between them.
Since tension stiffening is basically generated from the bond, this approach seems to be a
more realistic model of the state of stress and strain between the cracks in the tension zone.
This research program aims to offer a clear picture of the variation of tension stiffening with
time and the associated mechanisms, and hence give a better approach to model this
behaviour. In this thesis, several series of laboratory experiments investigating the time-
dependent change in tension stiffening are described and the results are presented. Reinforced
concrete prisms were tested in axial tension, including both short-term tests where the prisms
were gradually loaded to failure and long-term tests where the prisms were subjected to
prolonged periods of sustained service load. Short-term and long-term tests were also
undertaken on simply-supported reinforced concrete beams and slabs. The details of the
experimental work and results are presented to give a clear picture of the variation of tension
5
stiffening with time. The influences of creep, shrinkage, and cracking on the deterioration of
tension stiffening with time are carefully investigated and discussed. In addition, a two
dimensional non-linear finite element program, with a proposed bond model developed
concurrently with the experimental program, is presented. The proposed bond model is based
on the CEB-FIP bond model but incorporates other factors, such as steel and concrete stress,
concrete cracking and shrinkage. The program has been used to analyse the experimental
work and to conduct a parametric study, in which various parameters that may affect time-
including the mechanisms of tension stiffening and the change in tension stiffening with time
due to changes in either the sustained load or with increasing load. The thesis also outlines the
development of a numerical method that can accurately predict the time-dependent change in
term and long-term behaviour of tension stiffening in prisms subjected to uniaxial tension,
and in beams and slabs subjected to flexure, and to identify the mechanisms associated
constitutive laws, that can be used to predict the behaviour of cracked reinforced concrete
members and to satisfactorily model tension stiffening, both in the short-term and in the
6
long-term.
3 To demonstrate the applicability of the numerical model by comparing the finite element
stiffening on both macro and micro scales. A literature review is also presented of previous
work on the development of numerical models of the bond mechanism, and the creep and
Chapter 3 presents the author's understanding of tension stiffening behaviour and the factors
affecting the change of tension stiffening with time. The general responses of uniaxial tension
members and flexural members are discussed. The effects of creep and shrinkage are included
in the discussion.
The experiments undertaken at the Randwick Heavy Structures Laboratory of the University
of New South Wales are presented in Chapters 4 and 5. Details of the specimens, the test set
up, the construction and testing procedures and the measurements taken are described.
Shrinkage before or after cracking is an important parameter that was carefully measured
during these experiments. The results from the uniaxial tension tests and flexural tests are also
In Chapter 6, material constitutive laws for concrete and steel bars are introduced in a two
7
dimensional finite element program. The formulation of each elements and the numerical
solution scheme are also presented. In addition, a bond-interface model is proposed to model
the change in bond between the concrete and the steel bar.
In Chapter 7, the finite element model introduced in Chapter 6 is used to analyse the tested
specimens in the experimental program and the numerical results are compared with the
experimental results. The model is further verified by analysing test specimens presented
Chapter 8 consists of a series of numerical experiments conducted by using the finite element
program. A parametric investigation is undertaken to further study the factors affecting time-
dependent tension stiffening. The results are discussed and the influence of each parameter on
Chapter 9 summarises the entire thesis and presents the major findings of the study.
2.1 OVERVIEW
This chapter gives a review of the current state of knowledge relating to tension stiffening.
The chapter starts with a description of tension stiffening in both uniaxial tension members
and flexural members under instantaneous monotonic loading, (the effects of creep and
shrinkage on the time-dependent behaviour will be discussed in Chapter 3). Section 2.3
provides a review of the previous experimental work investigating and characterizing tension
stiffening, the majority of which is considering only the behaviour under short-term loading.
Section 2.4 describes the methods that are used to predict tension stiffening in practice,
including the procedures specified in building codes and those in the finite element context.
Section 2.5 and 2.6 summarise and discuss previous work on the bond mechanism and the
effects of creep and shrinkage, as these factors are deemed to have a significant effect on the
bond, creep and shrinkage, and the numerical modeling methods are also reviewed in Section
Tension stiffening is the ability of tensile concrete to contribute to the stiffness of a reinforced
concrete member. There are a number of different ways to represent this phenomenon in
practice. Previous studies have considered tension stiffening in a reinforced concrete member
at either a macro scale or a micro scale. In this thesis, the micro scale is referred to as the level
concerning the stress (strain, or slip) distribution between two primary cracks. The macro
scale is based on a spatially larger level that explicitly contains several primary cracks and
implicitly considers member stiffness and the global average structural response.
Consider an axially loaded singly-reinforced concrete tension member (Figure 2-1a). The
member is subjected to an incremental uniaxial tensile force P at both ends of the protruding
bar, as shown. The tensile force is gradually transferred from steel bar to surrounding concrete
by bond at the concrete-steel interface. Before cracking, the concrete tensile stress increases
linearly as the applied load is increased. At an applied load Pcr, first cracking occurs at the
weakest cross-section and this is usually assumed to occur when the concrete tensile stress
reaches the lower characteristic value of the direct tensile strength fct. The concrete stress at
this cracked section suddenly drops to zero and at this section the entire force is carried by the
tensile steel bar (Figure 2-1b). Within a distance S from the crack, the concrete stress is
gradually built up due to the bond at the concrete-steel interface, until at distances greater than
Pcr Pcr
s s
Vc
b) At first crack
P>Pcr P>Pcr
Vc
A relatively small increase in load will cause a second crack to develop at a cross-section at
some distance x t S from the first crack, thereby reducing the concrete stress in the vicinity of
that crack. Eventually, under increasing load, primary cracks form at somewhat regular
intervals along the member and the primary crack pattern is established. The concrete tensile
stress c at each crack is zero, rising to a maximum value (less than the tensile strength of the
11
concrete) mid-way between adjacent cracks, as shown in Figure 2-1c. For normal strength
After the crack pattern stabilises and no more primary cracks form, the average tensile stress
in the concrete decreases progressively under increasing load. This phenomenon can be
interface. The cover-controlled cracks are the internal cracks that radiate from the bar
deformations and are contained within the concrete cover (Goto 1971). The bond stress is
reduced by the formation of cover-controlled cracks in-between two primary cracks and hence
Figure 2-2 shows the micro responses, including concrete stress, bond stress and slip
distributions at the steel-concrete interface between two primary cracks. Bond stress is zero at
the cracked sections, because the concrete and the steel bar are not in contact. The slip at the
primary crack is at a maximum and decreases to zero mid-way between the two cracks, as
shown. Bond stress increases rapidly adjacent to each crack and then decreases to zero mid-
way between the cracks. After the crack pattern has stabilised, an increment of external load
causes a local decrease of the bond stress due to cover-controlled cracks and a widening of
existing primary cracks associated with increasing slip. Normally the deterioration of bond
represented as a smeared average decrease of bond stress between the two cracks.
12
Figure 2-2: Local bond stress and slip distribution between two primary cracks at crack
stablised stage.
Figure 2-3a shows the load versus average elongation response of a uniaxial tension member.
This is the macro-scale behaviour of tension stiffening under short-term load. The macro
response consists of three phases, namely linear elastic, primary crack formation and cover-
controlled cracking (stabilised crack) stages. Also plotted in the diagram is the bare bar
response, where the concrete is assumed to carry no tension and the entire tensile force is
carried by the steel bar at every section throughout the length of the member. The
13
instantaneous tension stiffening in the diagram can be denoted as the instantaneous tension
stiffening strain tsi, which is the difference between the bare bar strain and the actual member
strain at any particular load P (Figure 2-3a). The tension stiffening strain is a maximum tsi.max
at first cracking Pcr and then gets progressively smaller as the external load increases.
Fields and Bischoff (2004) introduced the concept of a tension stiffening factor or bond factor t
which can be defined as the tension stiffening strain at load P (tsi) divided by the maximum
tension stiffening strain tsi.max at P = Pcr. Figure 2-3b shows a typical relationship between the
tension stiffening factor and the average member strain, where t reaches its maximum value
(1.0) at first cracking and then diminishes with increasing strain eventually approaching zero as
The overview of the micro and macro instantaneous responses of tension stiffening in a
uniaxial tension member shows that they are compatible with each other. The global response
is attributed to the local mechanisms of bond and cracking. The decrease of bond stress due to
both primary cracking and cover-controlled cracking triggers the decrease of concrete stress
Figure 2-4a shows a typically cracked portion of a reinforced concrete beam subjected to an
applied moment Ms, greater than the cracking moment Mcr. The bottom fibres are in tension
while the top fibres are in compression. At the cracking moment, the tensile stress reaches the
flexural tensile strength of the concrete on the tension side of the beam, and primary cracks
occur. As moment increases, the cracks stabilise at a spacing about S. Under the moment Ms
even though the moment is constant along the beam, the second moment of area varies
depending on the crack locations and the curvature values along the beam. The second
moment of area of the cross-section is higher at the sections away from the cracks and lower
close to the cracks. The position of neutral axis is therefore not parallel to the geometric axis
of the member, as shown. However, it is usual to assume an average neutral axis location that
is at a constant depth below the compressive surface of the member (Figure 2-4a).
15
The distribution of concrete tensile stress, bond stress and slip in the tensile zone in-between
the cracks are usually considered to be similar to those of the uniaxial tension member, as
shown in Figure 2-1 and 2-2. When the direct tensile strength of concrete in the weakest
section of the tensile zone is reached at moment Mcr, and the first crack occurs, the tensile
stress of concrete at the cracked section reduces to zero. The concrete tensile stress gradually
increases away from the cracks as the bond stress develops. Additional primary cracks form
16
with increasing applied moment until the primary crack pattern has stabilised. Thereafter
cover-controlled cracks appear at the concrete-steel interface and these locally break down the
bond stress and hence reduce the tensile stress in the concrete.
An area of concrete Act symmetrically located around the tensile steel can be defined in the
cracked regions of a beam to carry the concrete tensile stress c between the cracks and to
model the tension stiffening effect in flexural members. The dimensions of this average
tensile concrete area have been proposed by Clark and Speirs (1978), Gilbert (1983), Castel et
al. (2006) and others. Gilbert indicated that Act is dependent on the magnitude of the
maximum applied moment Ms, the area of the tensile reinforcement Ast, the amount and the
types of tensile steel bars, the tensile strength of concrete fct, and the loading type (short-term
or long-term loading). Clark and Speirs suggested that Act is also affected by tensile bar
diameters db and concrete cover cb. Other parameters, such as the location of the neutral axis
and the location of the reinforcements in the tensile area of the beam, were taken into account
Globally, tension stiffening in reinforced concrete flexural members can also be interpreted in
terms of bending moment M versus deflection (or average curvature ) response, as shown
in Figure 2-4b. The cracking development phases shown in the graph are analogous to that of
the uniaxial tension member. As soon as the bending moment exceeds the cracking moment
Mcr, a first crack forms and the response starts to become nonlinear. Tension stiffening in this
case can be characterized as the instantaneous tension stiffening deflection (tsi) between the
actual response and the fully-cracked response. The fully-cracked response is obtained by
assuming that every cross-section is fully-cracked and the flexural rigidity is EcIcr, where Ec is
the elastic modulus of concrete and Icr is the second moment of area of a fully-cracked cross-
17
section. In this case, tension stiffening factor t can be represented as the ratio between the
tension stiffening deflection tsi (or the average curvature tsi) at any applied moment and its
maximum value tsi.max (or tsi.max) at first cracking (Bischoff 2005). Experimental
investigations suggest that the instantaneous tension stiffening in beams or slabs decreases
with increasing moment. This is illustrated in Figure 2-4b where the actual response
Many short-term tests have been carried out to investigate the instantaneous behaviour of
reinforced concrete members after cracking. Many of these tests were undertaken on axially
loaded tension members where tension stiffening can be most directly measured. Most of
these tests have not considered the influence of initial concrete shrinkage prior to loading,
which may greatly affect the measured test results (Bischoff 2001). Chapter 3 provides a
detailed explanation of the effects of creep and shrinkage on the short-term tension stiffening.
In additional to the effect of initial shrinkage on the experimental results, there are other
factors that may significantly influence short-term tension stiffening behaviour. Early
experimental work by Clark and Cranstown (1979) showed that the decrease of tension
stiffening under short-term loading is not only a function of strain but also bar spacing in
reinforced concrete slabs. They suggested the decay of tension stiffening was greater when the
bars were widely spaced in slabs. This is more significant when the bar spacing exceeds a
critical value of 1.5 times the slab depth. Kishek (1983) undertook an experimental study on
18
the effect of orientation of reinforcement on tension stiffening in slabs under uniaxial bending.
The angle between the reinforcing bars and the principal bending moment direction varied
from 0 to 60 degrees. The results show that the tension stiffening increases as the angle of
reinforcement with the principal bending moment increases. This is attributed to the fact that
with greater angle, the tensile stresses in the reinforcement in the principal bending direction
are smaller. Furthermore, the experimental study by Sooriyaarachchi et al. (2005) showed that
there was no significant influence on tension stiffening when the bar diameter was varied
Based on their experiments, Al-Fayadh (1997) and Abrishami and Mitchell (1996) concluded
that the use of high-strength concrete can increase tension stiffening but mainly at the crack
formation stage and the linear-elastic stage. However, they found that this influence decreases
gradually during the crack stabilisation stage. A similar conclusion was reported by Ouyang et
al. (1997) who found that the tension stiffening effect declines faster in high-strength concrete
members than in the normal-strength members as the average strain increases. They suggested
that this may be attributed to the more brittle cracking process in high-strength concrete.
Abrishami and Mitchell also indicated that crack spacing in high strength concrete specimens
Gilbert (2007) discussed the tension stiffening deflections that were obtained from short-term
tests on a number of slabs with different reinforcing ratios. He concluded that for members
with a relatively small reinforcement ratio, tension stiffening provides a large proportion of
In addition, other experimental works (Bischoff and Paixao 2004) suggested that the tension
stiffening contribution to the overall stiffness is greater in members reinforced with glass fibre
19
reinforced polymer (GFRP) than in members containing steel reinforcement. This is because
the GFRP reinforced concrete has higher cracking strain than the steel reinforced concrete,
thus the crack pattern in FRP reinforced concrete may not even be stabilised at service loads.
This thesis is only concerned with the tension stiffening in steel reinforced concrete elements,
Relatively few experimental programs have been undertaken to investigate the long-term and
series of long-term beam tests in which the specimens were subjected to sustained load for up
to 2 years, and the change of concrete and steel strains and the development of cracking were
recorded throughout the tests. Stevens concluded that the tensile force in the concrete is about
one third of that developed just before cracking, and it reduced with time at a reducing rate.
He suggested that the rate of reduction is akin to that of the relaxation of concrete stress in the
The bond break-down with time due to internal cracks or cover-controlled cracks was first
discussed by Illston and Stevens (1972). In their study, a technique of drilling holes from each
crack on the side faces near the level of reinforcement was adopted. A vacuum pump was used
to draw the resin through the cracks to measure the internal crack width. The crack width at
the surface of the beam was 2.5 times that at the bar. However, this ratio reduced to 20% after
2 years of sustained loading. The beams under sustained loading also exhibited about 30%
loss of bond after 1.25 year. They suggested that under sustained load there is a length of bond
breakdown near the surface crack, which can cause the opening of existing surface cracks (or
primary cracks), initially restrained by the steel bars. It is interesting to note that the
20
increasing load can result in the instantaneous breakdown of bond and hence the loss of
tension stiffening, while under constant sustained loading, breakdown of bond is also
observed experimentally. The mechanism for this long-term decay of bond or tension
stiffening can be attributed to the time-dependent properties of the material, particularly the
deflection caused by concrete shrinkage and the restraint to shrinkage provided by the
embedded reinforcement.
The most relevant and recent experimental work was undertaken by Beeby and Scott (2002).
sustained load levels. In the macro scale, they used Demec gauges to measure the average
elongation over several primary cracks; in the micro scale, strain gauges were used to measure
the local variation of strains along the tensile steel bars. They concluded that tension
stiffening decreases to about 50% of its initial value very rapidly (within 20 days) under
sustained loading. They attributed this phenomenon to several mechanisms including creep
and shrinkage of concrete, and development of internal cracks. Figure 2-5 depicts the test set
According to the literature, there are a variety of factors that may change tension stiffening in
reinforced concrete, including concrete strength, bar size, bar spacing, bar orientation, creep,
and shrinkage. However, most of the tests mentioned in the preceding sections did not
measure the creep and shrinkage characteristics of the concrete used in the experimental
programs. It has been conclusively demonstrated that initial shrinkage can greatly affect the
measured tension stiffening response in reinforced concrete tension members (Bischoff 2001).
Therefore, in the absence of measurements of shrinkage and creep, the reliability of many of
the previous tests is in doubt. The loss of tension stiffening is largely due to the breakdown of
bond and cracking of the concrete, both of which are affected by the development of
shrinkage in the concrete. Drying shrinkage can cause a significant increase in existing crack
widths and a decrease in the average crack spacing due to the formation of new crack (Nejadi
22
and Gilbert 2004a). Therefore, the effect of shrinkage on tension stiffening can not be ignored
Despite the previous work, the magnitude and the rate of change of tension stiffening under
sustained load are still not well understood. Illston and Stevens found that the loss of bond is
only about 30% after 1.25 years of sustained load while Beeby and Scott detected a 50% loss
of tension stiffening within the first 20 days. The tremendous difference in these observations
may be ascribed to the neglect of shrinkage in the experimental evaluation and also the
The behaviour of tension stiffening, especially its time-dependent responses and mechanisms
are not well understood, and there exists a range of opinions on the matter. The aim of the
current research is to give a clear picture of tension stiffening and its variation with time. As
there are a very limited number of tests incorporating the effects of shrinkage and creep on
tension stiffening reported in the literature, an experimental program with several uniaxial
tension and flexural members has been conducted in this study. The development of creep and
shrinkage has been monitored with time. The full details of the program are discussed in
Chapters 4 and 5.
Various methods are specified in building standards to consider tension stiffening in the
calculation of deflection or crack width. The Australian Standard (AS3600-2009) adopts the
well-known Branson’s equation (Branson 1968), which takes account of tension stiffening by
23
calculating the so-called effective stiffness that is an empirical estimate of the member
stiffness between the fully-cracked stiffness and the uncracked stiffness, depending on the
level of moment at the critical sections. The Eurocode 2 (2004) provides an approach that
state in the reinforcement under consideration. Both of these approaches are based on a
the member.
moment of area Ief of a beam or a slab after first cracking is obtained from Equation 2.1. The
short-term deflection or curvature can then be derived based on the elasticity theory, using the
3
§M ·
I ef I cr I I cr ¨¨ cr* ¸ d I ef . max
¸ 2.1
© Ms ¹
where Icr is the second moment of area of the cracked transformed section, which depends on
the modulus ratio and reinforcing ratio; I is the second moment of area of the uncracked
cross-section about the centroidal axis; Ms* is maximum bending moment at the section, based
on the short-term serviceability load or construction load; and M cr is the cracking moment.
M cr Z f cf‘ V cs P / Ag Pe t 0 2.2
where Z is the section modulus of the uncracked section, referring to the extreme fibre at
which cracking occurs; f’cf is the characteristic flexural tensile strength of concrete; cs is the
maximum shrinkage-induced tensile stress on the uncracked section at the extreme fibre at
24
which cracking occurs, and for a singly reinforced section may be taken as:
2.5 U w 0.8 U cw
V cs E s H cs* 2.3
1 50 U w
where cs* is the final design shrinkage strain; w is the web reinforcement ratio for the tensile
steel (= (Ast + Apt) / bwd); and cw is the reinforcement ratio for the compressive steel (= Asc /
bwd ); Asc and Ast are compressive and tensile reinforcement area respectively; and Apt is the
area of any prestressing steel in the tension zone. Equations 2.2 and 2.3 indicate that the
cracking moment Mcr is affected by the shrinkage strain of concrete and the reinforcement
kcs is specified as kcs >2 1.2 Asc / Ast @ t 0.8 , to account for the effects of creep and shrinkage,
where Asc / Ast is taken at — (a) mid-span, for a simply supported or continuous beam; or (b)
the support, for a cantilever beam. In the absence of more refined methods, the long-term
deflection of a beam or slab many be taken as kcs times the short-term deflection due to the
Equation 2.1 is also specified in the ACI code (ACI318-08) but in that standard no account is
taken of shrinkage (ie. cs = 0). Consequently ACI318-08 can overestimate stiffness and
The approach adopted by Eurocode 2 is to calculate the curvatures at typical sections and then
determine the deflection by integration. The curvature of a section after first cracking is
calculated by:
25
N 9N cr 1 9 N uncr 2.4
account of the bond property of the reinforcement (1.0 for ribbed bars and 0.5 for plain bars);
2 is the coefficient that take accounts for the duration of loading or of repeated loading (1.0
for single, short-term loads and 0.5 for repeated or sustained loads); cr is the curvature at the
section ignoring concrete in tension; uncr is the curvature on the uncracked transformed
It is generally agreed that both of these approaches give reasonable results for members with
reinforcing ratio in excess of about 0.5%. However, compared with experimental member
Branson’s equation. This is because when the reinforcement ratio is low, the maximum
moment Ms* in Branson’s equation is not significantly greater than the cracking moment Mcr,
and Ief remains close to the uncracked value I. Thus, the instantaneous deflection is
underestimated after first cracking (Bischoff 2005, Ghali 1993, Gilbert 2007). In contrast, the
Eurocode approach provides more accurate and reliable results particularly for lightly
In order to compensate for this limitation of Branson’s equation Gao et al. (1998), Ghali et al.
(2001), Toutanji and Saafi (2002) proposed modifications to Branson's equation for deflection
prediction by introducing empirical factors that effectively decrease the ratio between gross
moment of inertia and cracked moment of inertia. Bischoff (2005) proposed an alternative
expression for Ief which may be derived from the Eurocode 2 approach (Equation 2.4).
Substituting the notations in equation 2.1 into equation 2.4 and rearranging gives the
26
following formula:
I cr
I ef
§ I ·§ M · 2 2.5
1 E 1 E 2 ¨1 cr ¸¨ cr ¸
¨ I ¸¹© M ¹
©
on the decay of tension stiffening by the inclusion of cs in Equation 2.2. Compared with
AS3600-2009, Eurocode 2 approach predicts the final value of tension stiffening as half of the
short-term value by setting the coefficient 2 = 0.5, when the specimen is under sustained load.
Under sustained loads, creep and shrinkage complicate the response of a cracked member, and
these are the most difficult issues to account in design. The simplified design methods in the
various codes account for long-term loss of tension stiffening, but do not specifically describe
In a finite element context, methods to take account of tension stiffening can be generally
categorized into two groups: (1) modifying the constitutive relationship of the tensile concrete
(Barros et al. 2001, Ebead and Marzouk 2005, Gilbert and Warner 1978, Guptam and
Maestrini 1990, Lin and Scordelis 1975, Prakhya and Morley 1990, Nayal and Rasheed 2006);
(2) modifying the constitutive relationship of the tensile reinforcement (Gilbert and Warner
1978). As shown in Figure 2-6, the former approach models tension stiffening with an
adjusted softening curve for concrete in tension, and the degree of softening is greater in the
finite element layer that is closer to the reinforcement layer. Perfect bond is usually assumed
between concrete and steel elements, even though in reality bond stress is always
27
a) Gradually unloading response of the tensile b) Modified stress-strain diagram for tensile
concrete after cracking (Lin and Scordelis) steel after cracking (Gilbert and Warner)
Figure 2-6: Alternative stress-strain diagrams for concrete and steel in tension to represent
tension stiffening in the finite element context.
The latter approach introduces an additional stress carried by the tensile steel bars that is
dependent on the strain levels, and assumes that concrete after cracking carries no stress
anywhere. It can be seen that both approaches do not correctly represent the true mechanisms
of tension stiffening. Therefore, these approaches are not adopted herein for the numerical
A more realistic method for modeling tension stiffening is to use a suitable bond-slip
relationship at the concrete-steel interface. The action of bond produces tensile stress in the
concrete around the bar. Balazs (1993) used the ascending branch of the bond-slip relationship
from CEB-FIP Code to describe the bond zone and obtain the steel stress distribution between
the cracks. However, the formula that he gives to calculate crack width or spacing is not
computationally friendly. Instead of assuming a bond stress versus slip relationship, Somayaji
28
and Shah (1981) assumed a function to represent the bond stress distribution between the
cracks in order to obtain crack width and to simulate tension stiffening. A constant bond stress
distribution within the cracked zone of a reinforced concrete member is often assumed. This
simplifies the mathematics (Floegl and Mang 1982, Marti et al. 1998).
The tension chord model developed by Marti et al. assumes that the bond stress is constant
and therefore that the steel stress distribution along the tension element is linear as shown in
Figure 2-7a. The concrete stresses are zero at the crack, and away from the cracks tensile
force is transferred from the reinforcement to the surrounding concrete by bond. The method
was extended to model the cracked membrane element (Kaufmann 1998, Kaufmann and
Marti 1998) for finite element implementation. For ordinary ribbed bars a stepped, perfectly
rigid-plastic bond shear stress-slip relationship was proposed in tension chord model, as
illustrated in Figure 2-7b. Marti et al suggested b0 = 2.0fct and b1 = fct before and after bar
yielding, while Gilbert (2006) recommended that b0 = 2.5fct in the short-term, and b0 =
For the tension chord element between two consecutive cracks spaced at Srm shown in Figure
2-7a, the steel stress at the crack sr can be expressed as a function of the average strain m
W b 0 S rm
V sr EsH m , ( V sr d f y ) 2.6a
db
W b 0 S rm W b1 S rm § W b 0 E s · E s 2
S rm
f y EsH m ¨ ¸ W b 0W b1 2
db d b ¨© W b1 E sh ¸¹ E sh db
V sr fy 2 , 2.6b
W b0 E s
W b1 E sh
29
( V s. min d f y V sr )
§ f y · W b1 S rm
V sr f y E sh ¨¨ H m ¸ ( f y V s. min )
© E s ¸¹ db 2.6c
Where db is the bar diameter, Esh = (fu – fy)/(u – fy/Es) = hardening modulus of steel, and s.min
is the minimum steel stress between the cracks. Note that in serviceability condition, steel
stress normally remains below the yielding stress, hence only Equation 2.6a is applicable.
The maximum crack spacing occurs when c.max = fct, and is given as
f ct d b
S r . max 2.7
2W b 0 U
where = Ast / Act is the reinforcement ratio of the tension chord element and db is the
reinforcing bar diameter. The crack spacing in a fully developed crack pattern is limited by
S r . max
d S rm d S r . max 2.8
2
Before the yielding of the steel, the average instantaneous steel strain is obtained from the
1
Hm T 0.375 f ct Act 2.10
E s Ast
The tension stiffening strain is obtained by subtracting the average strain from the bare bar
strain at the cracks, and according to the tension chord model it is constant after all the
primary cracks have stabilised, because the bond stress stays constant.
30
0.375 f ct Act
H tsi H sr H m 2.11
E s Ast
The tension chord model simplifies the analysis by assuming the bond stress is constant,
hence giving rise to linear concrete and steel stress distributions. In reality the bond stress
distribution is nonlinear and significantly dependent on the level of steel stress and the extent
of slip. A more realistic model that can be used to correctly predict tension stiffening or
concrete tensile stress must incorporates a local bond stress-slip model that takes account of
these parameters. In this way, the interaction between the concrete and the steel bar can be
appropriately modeled and the tensile stress in the concrete will not remain constant after first
cracking. In the next section, a review and discussion is given of the bond stress-slip
There are generally three different scales in which the bond behaviour is characterized and
modeled in engineering practice, namely the rib scale, bar scale and member scale, (Lowe
1999) as illustrated in Figure 2-8. In the member scale, the whole member is considered as an
individual element in the analysis. The reinforcing bar is treated as being smeared over the
whole element and perfect bond between concrete and steel is assumed. The bond behaviour
can hardly be looked into in the member scale, as the combination of shear, flexure or torsion
on the member can affect the accurate investigation. In addition, this model can hardly be
applied in a finite element context as it can only simulate one particular structural element.
a) Member scale
c) Rib scale
The rib scale, at the other extreme, characterizes the bond response by discretizing the bond
zone around each rib. The bond effect in this scale is greatly affected by concrete behaviour at
33
the vicinity of the bar, such as fracture and crushing of concrete mortar and aggregates, load
transferred between concrete mortar and aggregate, etc. The bond behaviour is also greatly
influenced by the non-homogeneous response as the internal cracks on the top of the ribs
develop. The difficulty to investigate bond effects at this scale is that the material non-
demonstrated in the experiments. For the implementation of finite element, the rib-scale
model may require a sophisticated meshing solution and algorithm, and a need to introduce
material non-homogeneity in the model, which creates great complexity for the analysis.
Therefore, numerous pull-out tests examining the bond response have been conducted at the
scale of the reinforcing bar, and the mechanisms of bond at this scale have been widely
investigated. Most of the pull-out tests have been undertaken with embedment lengths
significantly larger than the rib spacing (usually over 6db). The bond response is assumed in
continuum within the bond zone, where the material properties of the concrete and the steel
bar are usually assumed to be homogenous. The size of the bond zone is normally smaller
than the crack spacing and therefore no primary cracks are allowed to form and disturb the
tests. External tensile force is imposed at the ends of the steel bar, and the parameters, such as
concrete strength, bar diameter, etc, on bond behaviour are considered. A typical example is
shown in Figure 2-9 with the usual idealization shown in Figure 2-9b.
The idealised continuum and the measured average bond stress-slip relationship greatly
simplify the problem and give good approximation of the global response of the whole
structure. Furthermore, the bond model identified in this scale can be directly applied to the
finite element model, together with the application of bond link elements. Accordingly, the
local bond stress-slip relationship that is proposed in this thesis is based on the steel bar scale.
34
Figure 2-9: Typical pull-out test configuration and its idealised continuum bond zone.
The characteristics of bond zone response are associated with the localised damage that
occurs through three distinct mechanisms acting in parallel during the loading process (Lutz
and Gergely 1967), namely chemical adhesion, friction, and mechanical interlock between the
concrete and steel bar. The first two mechanisms are primarily responsible for the bond
response of plain bars, where roughness of the bar surface is the primary bond strength
parameter. In this case, when the bond stress at typical section of the member exceeds the
For deformed steel bars, the mechanical interlock between concrete and steel has proven to be
35
the dominant mechanism. This phenomenon was demonstrated by Lutz et al. (1966). They
concluded that the load transferred between steel and concrete is significantly affected by
mechanical interaction. As the bar deformations bear onto the concrete, local fracture
(splitting) of concrete occurs. The presence of splitting crack can significantly leads to the
loss of tension stiffening, as demonstrated by Abrishami & Mitchell (1996). Lutz and Gergely
(1967) proposed that the slip can be induced by the wedging action at the front of the rib. A 30
to 40 degree effective rib face angle is formed by the crushing of concrete. Goto (1971)
investigated the wedging action by injecting ink into cracks around the reinforcing bar when
the specimens were loaded to the maximum force. Observation showed that besides primary
cracks, cover-controlled cracks (radial cracks) on the interface are initiated and inclined to the
bar axis at angle between 45 and 80 degrees (60 degrees on average). As the cover-controlled
cracks initiate, the bearing effect becomes the major mechanism causing damage to the bond
zone, where relatively little slip occurs. The bond stress induced by friction between concrete
and steel is a minimum in this phase. Splitting cracks along the steel bar axis form and
propagate from the steel interface to the surface of the concrete specimen. These splitting
cracks result from circumferential stresses within the concrete at the vicinity of the bond
interface. This is caused by the wedges bearing on the concrete at the front of each rib and the
radial stress perpendicular to the bar axis (Tepfers 1979). As soon as the circumferential stress
exceeds the tensile strength of concrete, longitudinal splitting crack forms. Therefore the bond
Figure 2-11 shows the major mechanisms that govern the bond stress-slip relationship at
different stages (fib 2000). As the bond stress stays below 20% to 80% of the tensile strength
of concrete, bond stress is primarily provided by the chemical adhesion and friction. The bond
stress-slip response is relatively stiff at this stage. For plain bar, as soon as the adhesion
breaks down, bond failure occurs. In the case of deformed bars, after the chemical adhesion is
broken down, mechanical interlock with the ribs bearing onto the concrete becomes the
governing mechanism. The steel bar starts to crush the concrete and internal radial cracks start
to form from the tip of the ribs onto the surrounding concrete, and the bond stress-slip
response begins to become non-linear. As soon as the splitting cracks reach the surface of
concrete where little confinement is provided, bond failure occurs suddenly. If sufficient
confinement is provided around the bond zone (usually by transverse reinforcement), splitting
failure is delayed. After bond stress reaches its maximum value, it drops off with increasing
slip. Mechanical interaction decreases at this stage and the bond stress is mainly provided by
the residual friction on the rough surface between the concrete and steel. Eventually, due to
37
the loss of shear resistance from surface friction, the steel bar can be completely pulled out
Many experimental programs have been undertaken to evaluate the effects of a wide range of
properties (bar sizes, anchorage length, bar spacing), and concrete properties (concrete
strength, concrete cover), etc. Few of these studies considered all parameters determining
bond effect, furthermore the testing methods and boundary conditions of the tests often varied
considerably.
It is generally agreed that bond strength increases as the concrete strength increases. It is
38
recognized that the bond strength is approximately proportional to the square root of concrete
also indicated that bond strength also depends on the bar diameter. The bond strength of a
19mm bar is 10% greater than that of 25mm bar, while the bond strength of 31mm bar is 10%
lower than that of 25mm bar. Morita and Fuji (1985) also reported that the measured slip in
their experiment was proportional to the bar diameter. The larger-size reinforcing bar is able
The steel stress state and concrete stress state have also been shown to significantly affect the
bond capacity. Shima et al. (1987) conducted an experimental study on steel reinforced
concrete cylinders loaded in tension, with an anchorage length of 50db. Bar strain was measure
using foil gauges glued to the bar along the length without removing the ribs from the
reinforcing bar. Slip was obtained from the integral of the measured strain along the length of
the bar. They suggested that the local strain in the steel was an indicator of the surrounding
concrete state, and the value of steel and concrete strain or stress varies along the bar
according to the embedded length of the pull-out test. The local bond-slip relationship along
the reinforcing bar can also differ according to the boundary conditions of the tests (Kankam
1997). Most of the other experimental work assumes a constant average bond stress along a
relatively short embedded length. The influence of steel stress on local bond response has not
The bond characteristics in the post-yielding range of the steel bar are also dependent on steel
stress or strain state. Under typical in-service conditions, the steel strain is less than yield
strain, and therefore the bond response after the yielding of steel is not discussed further here.
39
Bond capacity is greatly affected by confining pressure on the bond zone. Confining pressure
reduces radial stresses that lead to splitting in the bond zone (Tepfers 1979). If good
confinement is provided, the bond response can be very significantly enhanced and bar pull-
out can result without splitting. Eligehausen et al. (1983) concluded that the pull-out failure
response in a confined specimen is associated with higher bond strength and bond capacity
The confinement actions can be attributed to the surrounding concrete, the transverse
reinforcement and the lateral pressure. Giuriani et al. (1991) evaluated previous pull-out tests
with transverse reinforcing bars and proposed the minimum amount of transverse
reinforcement that is sufficient to prevent slitting failure of the pull-out test. Early
investigations on the influence of lateral pressure on bond strength (Untrauer and Henry 1965)
demonstrated that the increase of bond strength due to external pressure is more significant at
higher stress states, and the increase is proportional to the product of square root of the
normal pressure and the square root of the concrete strength. A similar conclusion has also
40
been found by other researchers (Navaratnarajah and Speare 1987; Robin and Standish 1984).
Apart from the bond stress-slip relationship, another significant bond response is the radial
stress-displacement relationship. The measurement of radial stress has not often been made in
previous experimental studies. Two of the most comprehensive tests regarding this issue were
conducted by Gambarova et al. (1989) and Malvar (1992). In the former study, a preformed
splitting crack with prescribed opening on the specimen was maintained under the monotonic
loading process, thus the radial deformation was maintained constant and the confinement
stress was monitored. Malvar applied constant confining stress on the external surface of
concrete under increasing load until splitting cracks occurred, during which the dilation or
radial deformation was recorded. Gambarova et al. suggested that with increasing crack
opening, the radial stress reached its maximum value when the slip slightly exceeded the
value of slip corresponding to the shear bond strength, but the peak value of the confinement
stress is not strongly dependent on the opening of the splitting crack. Malvar concluded that
the dilation of the bond zone is a reaction to the damage of concrete surrounding the bond
zone, and thereby can be represented by the concrete material state. In addition the radial
expansion of the pull-out specimens decreases with increasing confinement pressure. The
influence of radial stress (or confinement pressure) is more obvious when the tests reach the
post-cracking range.
The previous experimental studies have also indicated that bond response can be affected by
the deformation pattern of the ribbed bar, including the rib spacing, height, face angle and
pattern. It is found that the bars with increased relative rib area (the rib area perpendicular to
the bar axis normalized by the bar surface area between ribs) exhibit increased bond stiffness
(Hamad 1995). In addition, splitting failure can be prevented with the use of bars with larger
41
rib area. Darwin and Graham (1993) concluded that the bond stress-slip relationship was a
function of the relative rib area of the bars, and independent of the specific combination of rib
In this section, numerical models developed for the bond stress-slip relationship are reviewed
and discussed, such as CEB-FIP bond model, Bigaj’s bond model, Lowes’ model,
thermodynamic model and plasticity bond model. The sophistication of each model depends
on the way the bond zone mechanical interaction is simulated and the parameters that are
incorporated in the model. Each model is associated with different algorithm for finite
element implementation.
One of the most commonly used models is the non-linear relationship proposed by CEB-FIP
Mode Code (1997) and illustrated in Figure 2-13. The bond stress-slip relationship under
monotonic loading is decomposed into four phases. The bond stress in each phase is given by
identifying certain slip parameters s1, s2, s3 in the diagram (Figure 2-13). The first phase
(ascending part) is defined as a power function for the slip range s<s1, followed by a plateau
b = max for s1< s < s2. The curve then decreases linearly to the residual frictional bond
resistance f for a slip range s2< s < s3. For s > s3, the bond stress is assumed constant and
equal to f (Table 2-1). In order to distinguish the splitting bond failure and pullout bond
failure, CEB-FIP bond model simply specifies different values for these key stresses and slips
0.4
§ s1 · for 0 d s d s1
W b W max ¨¨ ¸
¸
© s2 ¹
W b W max for s1 d s d s 2
§ s s2 ·
W b W max W max W f ¨¨ ¸ for s 2 d s d s 3
¸
© s3 s2 ¹
Wb Wf for s 3 d s
Figure 2-13: Bond stress-slip relationship of CEB-FIP bond model (CEB-FIP 1997).
Unconfined Confined
Bond conditions Bond conditions
Good Others Good Others
S1 0.6mm 0.6mm 1.0mm
S2 0.6mm 0.6mm 3.0mm
S3 1.0mm 2.5mm Clear rib space
0.4 0.4
max 2 f cp f cp 2.5 f cp 1.25 f cp
f 0.15max 0.4max
Tepfer (1979) proposed a so-called thick-walled cylinder to characterize the bond zone. The
interaction between the concrete and the steel bar is represented by the shear stresses and
accompanied radial stresses. In his theory, three failure stages are involved in the bond stress-
slip relationship, namely uncracked elastic stage, partly cracked stage, and entirely cracked
stage (Figure 2-14), which characterises the propagation of radial stress from the interface.
Den Uijl and Bigaj (1996) extended Tepfers’ theory by incorporating a softening behaviour of
concrete caused by the internal radial cracks, where cracked concrete is defined on the basis
of the fictitious crack model (Hillerborg 1983) and the bilinear softening model by Roelfstra
43
and Wittmann (1986), rather than assuming it has no capacity after cracking as Tepfers
suggested. The number of radial cracks, which is related to the total radial and circumferential
assuming dry friction at the interface, so that the bond stress is proportional to radial stress
with the orientation of the interface. To distinguish the bond responses under splitting failure
and pull-out failure, Den Uijl and Bigaj suggested that the angle between radial displacement
and slip is a constant value. In the case of pull-out failure, they proposed that the cone angle
Figure 2-14: Three failure stages of bond stress-slip relationship by Tepfer (1979).
define the influence of confining pressure, concrete damage state, steel strain state, slip
history, bar size and bar spacing. The product of these multipliers contributes to the total peak
bond strength. For the convenience of finite element implementation, a non-local model is
used to include the adjacent concrete and steel material states into the bond elements. One of
the deficiencies of the model is that the multiplier to account for the damage of surrounding
44
concrete is defined based on their own concrete constitutive model. Thus it can hardly be used
with other concrete models, unless this multiplier is modified according to the corresponding
concrete model.
Ragueneau et al. (2006) describes the constitutive model of the bond interface elements based
on a thermodynamic framework. The model also couples continuum damage and plasticity
theory to consider the cracks parallel to the bar direction and inelastic strain due to sliding.
The mode- fracture failure and the frictional sliding are coupled, with a single damage
variable in the Helmholtz free energy description, to represent the energy transformed from
variation of the damage yield function, within the elasto-damage domain. For the frictional
sliding part, the constitutive relationship is represented by the plasticity yield function. The
The plasticity-based bond model (Cox and Herrmann 1998) is established based on the
assumed yield surface, hardening and softening rule and non-associated flow rule of the bond
stress-slip relationship. Similar to the thermodynamic bond model, the fundamental factor that
determines the evolution of the bond response in this model is attributed to a damage factor.
The damage factor here is a function of the plastic slip (slip which occurred during the initial
splitting phase of the loading) and the yield function and the evolution law. The yield criteria
of the model consist of an exponential criterion (for large damage value) and a power criterion
(for small damage value) as a linear combination through a weighting function, which is
dependent of the magnitude of the damage factor. Therefore the yield surface changes from a
non-associated flow rule (Cox 1994) is used to represent the bond stress versus radial dilation
45
response, which takes place near the peak bond stress. This radial dilation reduces as the
confinement pressure increases. A strong point of this model is that it does not require
establishing the geometric relationship between radial and shear stress. Instead, these two
bond stresses are automatically related through the variation of the plastic work in terms of
There were quite few experimental works that focused on the long-term bond-slip response.
Under low-level sustained loading, creep can cause additional slip on the concrete-steel
interface. Creep therefore softens the stiffness of the ascending part of bond stress-slip
relationship. This primary mechanical aspect of bond creep at the steel-concrete interface has
Rostasy and Kepp (1982) attempted to characterize the parameters of bond creep by their
long-term pull-out tests. Their testing series were classified with different variables, such as
concrete strength, sustained bond stress before the instantaneous bond strength, history of
bond stress, and relative area of ribs on the bar. They concluded that bond creep is
independent of concrete strength and bond stress states. In addition, the effect of aging on
bond creep can be ignored if the first loading age is greater than 35 days. This can greatly
simplify the numerical analysis. They presented the irrecoverable creep slip in terms of a
linear creep strain theory, where the bond creep displacement is proportional to the elastic
s sc t s siM b t 2.12
Where ssc(t) is the creep slip, ssi is the instantaneous slip induced by a certain level of short-
46
term bond stress, and ijb(t) is the bond creep coefficient, which performs non-linearly with
respect to time. If the creep coefficient in Equation 2.12 is non-aging, it can be depicted as a
Koch and Balazs (1993) suggested that the time-dependent slip is higher, if the applied pull-
out force is higher, which indicates that the long-term slip is associated with internal or
In a sufficiently long specimen, the breakdown of bond with time is primarily the result of
restraint to concrete shrinkage by the reinforcing bar (or other boundary conditions) and the
effects of bond creep are secondary. If the restraint to concrete shrinkage is great enough,
cracks (either primary cracks or cover-controlled cracks) will start to propagate and cause a
local degrading of bond stress. This process is further complicated since it is also related to
the steel stress and concrete stress state, as addressed by Koch and Balazs. The experimental
work in the literature was mainly involved with small size specimens with embedment length
much less than the crack spacing. They do not reflect the influence of local damage of bond
due to shrinkage and cracking. A more comprehensive and realistic long-term bond stress-slip
relationship must consider the effect of shrinkage on the loss of bond at different steel stress
reviewed. Expressions for creep and shrinkage strains, together with the instantaneous
The concrete strain at any time at a point in a specimen at constant temperature can be
expressed as the sum of instantaneous İe(t), creep İcp(t) and shrinkage İsh(t) components.
Instantaneous and creep strains are induced by the stress in the concrete whereas shrinkage
strain is stress-independent. Hence the total strain in a specimen is the superposition of the
three components
H (t ) H e (t ) H cp (t ) H sh (t ) 2.13
If the concrete element is under constant stress and it remains less than about 0.4 of the
compressive strength (which usually is the case under service condition), creep strain is
proportional to the stress and the creep isochrones are approximately linear (Figure 2-15). At
higher stress level, creep strain increases at a faster rate than the applied stress and becomes
non-linear with respect to the stress. Therefore, at lower stress level, one can calculate the
creep strain approximately proportional to the stress level and Equation 2.13 can be expressed
as
H (t ) VJ t , t ' H sh (t ) 2.14
in which J(t,t’) is the compliance function representing the mechanical strain (instantaneous
plus creep) at time t caused by a unit constant stress that has been acting since time t’. J(t,t’)
can also be expressed as the sum of elastic compliance function 1/E(t’) and the creep
1 M (t , t ' )
J (t , t ' ) 1 / E t ' C t , t ' 2.15
E (t ' )
where E(t’) is the elastic modulus characterizing the instantaneous deformation at age t’; and
ij(t,t’) is called the creep coefficient which is the ratio between the creep strain and the
/fcp
1.0
Failure
limit
t=0
t=
0.5
Linear
creep range
Concrete strain
Research on creep (Bažant and Wittmann 1983) showed that creep in the concrete is mainly
due to the breakage of bonds and contacts between colloidal sheets in the hydrated cement gel.
There are generally two types of creep, namely basic creep and drying creep. The former is
the creep strain occurring in a sealed moist environment, due to the breakage of bonds at
highly stressed sites within the colloidal microstructure of the calcium silicate hydrate gels in
the hardened cement paste. Pickett (1942) first observed drying creep, known as the Pickett
effect. The Pickett effect describes the fact that with drying, the deformation of a concrete
specimen under sustained load can exceed the sum of the drying shrinkage deformation of a
load-free specimen and of the deformation of a specimen that does not dry, i.e. is sealed. The
49
basic mechanism for drying creep is that drying can cause creep strain per unit stress to
become larger and thus produce additional deformation. Bažant and Xi (1994) suggested that
a major portion of the Pickett effect is attributed to the micro bond breakage within the
microstructure of calcium silicate hydrates by the drying effect, which can further increase
creep strain rate with time (Bažant et al. 1997).The other factors that may influence drying
creep were found and given by Bažant and Wu (1974), Bažant and Chern (1985), Wittmann
Creep is affected by many factors, including the maximum aggregate sizes, the water-cement
ratio, the concrete strength, the type of cement, the property of aggregates, temperature,
humidity, the degree of hydration, pore water content, specimen size and stress state, etc. The
detailed explanation of the effects of these parameters on creep can be found in the following
literature (Bazant et al. 1982; Bazant et al. 1976; Bažant and Chern 1985; Bažant and
Wittmann 1983; Domone 1974; Gamble and Parrott 1978; Gamble 1982; Gilbert 1988;
Equation 2.15 explicitly indicates that an important property of concrete creep is aging, which
means that creep strain is greater when the age of the first loading is reduced. The creep rate
under sustained load is greater during the first few weeks for concrete loaded at an early age.
The aging of creep is caused by the chemical process of hydration in the concrete, which can
produce tri-calcium silicate hydrate gel and gradually fill the pores of hardened cement paste
hence stiffening the microstructure within the concrete and greatly reducing the creep strain.
Figure 2-16 shows the typical curves for the development of J(t,t’) with different loading ages
t’. J(t,t’) increases at a faster rate in the concrete loaded at earlier ages.
50
J(t,t')
t'=3days
30days
300days
3000days
30000days
1/E0
1 hour 30 years
log(t-t')
For concrete in tension, previous studies (Brooks and Neville 1977; Illston 1965) illustrate
that the compliance is generally observed to be as large as for uniaxial compression. However,
the mechanism and magnitude of tensile creep are still not well quantified. In this thesis, the
Analogous to the creep strain, shrinkage strain can also be categorized into two different types:
drying shrinkage and autogenous shrinkage. Drying shrinkage commences as soon as the
concrete specimen begins drying. It results from increasing capillary tension of pore water
and solid surface tension of pore walls, caused by diffusion of pore water out of the specimen.
When the concrete specimen is immersed in water, swelling occurs, which is normally an
order of magnitude less than drying shrinkage and thus can not compensate for the drying
shrinkage. Autogenous shrinkage is the change of specimen volume caused by the chemical
compared to drying shrinkage. It is only about 5% of the total drying shrinkage. The
51
autogenous shrinkage development stops when the concrete dries to a relative humidity 80%
sh
Upper bound
t0 (start of drying) t
Figure 2-17: Development of drying shrinkage strain.
Thus shrinkage strain is affected by the water-cement ratio, the relative humidity, the ratio of
exposed surface area to volume, the aggregate, etc. It develops very fast in the first several
generally agreed that shrinkage strain approaches a finite upper bound as time approaches
The modeling of creep and its aging effect in a concrete structure is often a complex
procedure. Due to the time-varying stress history in almost every reinforced concrete
specimen, one can not directly derive creep strain at a typical time by equation 2.14. An
alternative way is to integrate the creep strain caused by the stress history d(t’). Each
infinitesimal stress at t’ produces the mechanical strain at time t according to its compliance
function d(t’) = J(t,t’)d(t’). The total strain at time t caused by a variable stress history can
52
then be obtained by summing these infinitesimal strains together with the stress-independent
t
Ht ³ J t , t ' dV t '
0
H sh t 2.16
Equation 2.16 (Stieltjes integral) relies on the principal of superposition. Volterra (1913) first
introduced this method for aging material analysis, and McHenry (1943) first applied it to
constitutive equation that relates the stress and strain histories. Differentiating equation 2.16
V t t
wJ t , t '
H t ³ dV t ' 2.17
Et 0
wt
It has been demonstrated that the principle of superposition can give accurate results as long
One major disadvantage of this approach is that a large amount of creep data is required.
Due to the complication of creep aging effect, simplified methods, such as rate of creep
method (RCM) (Dischinger 1937; Whitney 1932) have been developed. The RCM assumes
parallel creep curves for concrete with different loading ages (Figure 2-18). That is the rate of
53
change of creep is assumed to be independent of the age at first loading. Therefore, only one
creep versus time curve is needed to obtain the creep strain for different stress histories. The
creep strain at time t caused by a constant stress applied at time t’ can be calculated by
subtracting the creep strain induced from t0 (age at first loading) to t’ from the total creep
strain at t
V
H cp (t , t ' ) >M (t , t 0 ) M (t ' , t 0 )@ 2.18
Ec
1 M (t , t 0 ) M (t ' , t 0 )
J t, t' 2.19
Et Ec
1 V t
H t V t M t , t 0 2.20
Et Ec
cp
F
(t1,t0)
B E
Actual
Response
(t1,t') D
Assumed
Response
A C
t0 t' t Time
Integrating equation 2.20 with time or using the method of superposition, one can obtain the
total mechanical strain at time t. The greatest advantage of this method is that it requires no
previous stress history if the elastic modulus is given as a constant. However, larger error may
occur using this method, particularly for a decreasing stress history (Bazant and Najjar 1973).
One of the simplest methods to predict creep in concrete for manual calculation is to adopt
some type of effective modulus. The convenience of this method is that it does not require a
step-by-step numerical integration. McMillan (1916) first introduced the Effective Modulus
Method (EMM), in which the creep for time t is obtained by transforming equation 2.15 to
1 E (t ' )
E e (t , t ' ) 2.21
J t, t ' 1 M (t , t ' )
However, this simplification obviously ignores the fact that equation 2.15 is only applicable
when the stress is held constant, which is normally not the case for the long-term response of
structures. EMM fails to take the aging effect into account and this method is only useful for a
single step stress history. If the concrete stress varies with time, this method yields inaccurate
solutions.
(AEMM) was introduced and formulated by Trost (1967) and Bazant (1972). This method
takes the aging effect on creep into account by simply reducing the creep coefficient ij(t,t0) if
the stress is gradually applied. An aging coefficient (t,t0) (with the range from 0.6 to 0.9) is
used and the magnitude of the reduced creep coefficient is given as (t,t0)ij(t,t0). Then the
creep strain at time t due to a gradually applied stress (t) at the time interval t – t0 can be
55
expressed as:
'V t
H cp (t ) F t , t 0 I t ,t 0 2.22
E t0
Therefore, for a continuous stress history, as shown in Figure 2.21, by assuming the change of
stress due the time interval t - t0 as (t) = (t0) – (t), the total strain at time t can be
calculated as the sum of strains produced by (t0), the strains produced by (t) and the
shrinkage strain.
V t0
H (t ) >1 M t , t 0 @ 'V t >1 F t , t 0 M t , t 0 @ H sh t
E t0 Et
V t0 'V t 2.23
H sh t
Ee t , t 0 Ee t , t 0
E t0
in which E e t , t ' is called the age-adjusted effective modulus.
1 F t, t0 M t, t0
(t 0)
(t0)
(t)
t0 t time
The aging coefficient F (t , t ' ) is dependent of the age at first loading, the duration of load, the
size and shape of the specimens and the rate of change of stress and so on (Gilbert 1988). For
56
a relaxation problem, it is approximately 0.8 when the time reaches infinity. For manual
calculation, AEMM is the most satisfactory alternative considering both efficiency and
accuracy.
An improved basic creep formulation based on the principal of superposition is called the
theory of solidification (Bažant and Prasannan 1989a). The method describes all the
viscoelastic behaviour of concrete, including aging, by a series of Kevin chains. The aging
aspect of creep is due to growth of the volume fraction v(t) of the load-bearing portion of
solidified matter (i.e. hydrated cement), the properties of which are assumed to be age-
g(v,t)
(a)
(c)
(b)
Figure 2-20: Model for the role of solidification in creep: (a) Kevin Chain description for
viscoelastic component; (b) microscopic creep compliance functions of the solidified matter;
(c) schematic representation of the solidification creep model.
57
Figure 2-20 shows the model for the role of solidification in creep of this theory. It is assumed
that the volume v of the hydrated cement grows by deposition of layers of solidified matter, as
shown in Figure 2-20c. ıg(v,t) is the stress at time t in the layer which solidifies when the total
volume of the solidified matter is v. At the moment it solidifies, the layer (dv) must be stress-
free, i.e. ıg[v(t),t] = 0. Therefore, the non-aging viscoelastic stress-strain relationship for the
t
Hv t Hv W ³ ) t t ' V g >v W , dt '@ 2.24
W
in which ıg[v(),t’] = 0; v is the viscoelastic strain due to solidified matter and (t - t’) is the
microscopic creep compliance function of the solidified matter, representing the strain at age t
function and only with one variable, namely the duration of time t - t’.
Further formulation of Equation 2.24 shows that ıg can be eliminated, and a macroscopic
J t 1 t
Hv t ³ ) t t ' dV t ' 2.25
vt vt 0
t t'
in which ) w) t t ' / wt , (t) is regarded as the viscoelastic microstrain.
According to the analysis of test results, it is found that another component f, called flow,
must also be included in the concrete creep, as shown in Figure 2-20c. It is viscous rather than
1 t
H f (t ) ³ < t t ' dV t ' 2.26
ht 0
58
where 0 is called the effective viscosity of the solidified mater, as a constant; and h(t) is the
volume fraction of the solidified mater associated with the viscous strains.
V t V t
H f t 2.28
K0h t Kt
According to the previous description, the total strain of concrete may be expressed as:
V
H cp H cp H sh , H cp Hv H f 2.29
E0
in which /E0 is the ultimate elastic strain resulting from the deformation of the mineral
aggregate pieces in concrete and the microscopic elastic particles in the hardened cement
paste. Bažant and Prasannan suggested that the effective modulus E(t), which in ordinary
in a very short-time period, so that the age-dependent part of the ordinary elastic strain is
treated to be an apparent elastic deformation. Bazant and Bawja (Bazant and Baweja 1995a)
proposed that E0 = 1.6Ec28 where Ec28 is the elastic modulus of concrete at 28 days.
According to Equations 2.25, 2.28 and 2.29, the creep compliance function can be depicted as
59
1
t
§) (t t ' ) 1 ·
J (t , t ' ) ³ ¨¨ ¸dt
K (t ) ¸¹
2.30
E 0 0 © v(t )
The empirical functions in Equations 2.25 and 2.28 were introduced by Bažant and Prasannan
C t, t ' >
q 2 Q t , t ' q3 ln 1 t t '
n
@ q 4
§t·
ln¨ ¸
© t' ¹
2.32
m n 1
§ O0 · n W t '
t
Q t, t ' ³t ' ¨© W ¸¹ O0 W t ' n dW 2.33
in which q2, q3, q4, n, m, Ȝ0 are empirical constants. For all types of concrete, experimental
data indicated:
1 / r t '
ª § Q f t' ·
r t'
º
Q t, t' Q f «1 ¨¨ ¸¸ » 2.35
«¬ © Z t , t ' ¹ »¼
where:
8
0.12
r t' 1.7 t ' 2.36
Z t, t' t'
m
>
ln 1 t t '
n
@ 2.37
The computational implementation of solidification creep theory will require transforming the
60
formulation into a rate-type law based on a rheological model with non-aging properties. The
means to obtain this formulation will be presented in Chapter 6 as this method will be used
3.1 OVERVIEW
In this chapter, an analytical study is undertaken to describe the time-dependent behaviour of
reinforced concrete and the change of tension stiffening with time. The effects of creep and
shrinkage are taken into consideration, together with the deterioration of bond and the effects
of time-dependent cracking. The response of a uniaxial tension member will be analysed first
since tension stiffening in this case is more directly measurable. Thereafter the analysis of a
concentrically reinforced concrete member subjected to axial tension (Figure 2-1), both before
and after cracking. As described earlier (section 2.2.1) the instantaneous response (curve OAB
in Figure 3-1) is linear up until first cracking at P = Pcr and non-linear after cracking. Before
cracking, the instantaneous tension stiffening strain tsi (which is the difference between the
strain in the specimen and strain in the bare bar) increases with load, but after cracking, tsi
62
decreases as P increases. The load at which cracking occurs depends on the tensile strength of
the concrete at the time of loading and also on the amount of drying shrinkage that has
occurred prior to loading. Early shrinkage is restrained by the embedded reinforcing bar and a
significant tensile stress may exist in the concrete before the external load P is applied to the
member.
Figure 3-2 illustrates the effects of creep and shrinkage on an uncracked cross section of a
uniaxial tension member. If the load P is held constant with time, creep strain develops in the
concrete and the average axial strain changes. In fact, because the stress in the tensile concrete
is relatively small, creep strain is relatively small and the change of stress and strain in the
concrete is relatively small. In a member that does not shrink, tensile creep causes a softening
63
of the concrete in tension and a relatively small change in the load-deformation response with
time (as shown by the dashed curve labeled ‘after creep only’ in Figure 3-1). Tensile creep of
the concrete sheds some of the tensile force carried by the concrete into the bonded
reinforcement, thereby reducing tension stiffening with time and the ‘after creep’ response is
Due to Shrinkage:
Shrinkage of the concrete while the member is under load causes a significant change in the
load-deformation response (which is shown in Figure 3-1 as curve O’A’D). For an uncracked
(Figure 3-2b). While reinforcement is provided, it restrains the free shrinkage of concrete. The
force would be applied to the concrete (denoted as T in Figure 3-2b). Due to this restraint to
shrinkage, the shortening of the section reduces to sx. Therefore, before cracking, shrinkage
causes the member to shorten and, at low tensile loads, the load-deformation curve moves to
the left in Figure 3-1, as shown by the dashed curve O’A’. Restraint to shrinkage also causes a
gradual build-up of tensile stress in the concrete and this reduces the cracking load from Pcr to
Pcr.sh.
After first cracking (P > Pcr), cracking and the deterioration of bond caused by cracking
resulting in slip at the concrete-steel interface are the physical mechanisms that are primarily
responsible for the reduction in tension stiffening due to restraint to shrinkage. Without
cracking, incremental increases in stress caused by the restraint to shrinkage would cause an
increase in tension stiffening, but the formation of new cracks and the resulting bond slip
results in a subsequent overall (net) drop in tension stiffening. Therefore, at load P (> Pcr) in
64
Figure 3-1, the instantaneous tension stiffening strain tsi is represented by the horizontal
distance BE and this reduces to CE due to creep and further reduces to DE after both creep
and shrinkage. It is often assumed that tension stiffening reduces with time under sustained
loads to about 50% of its instantaneous value, but this is yet to be conclusively demonstrated.
a) Due to creep
b) Due to shrinkage
Figure 3-2: Time-dependent deformation and stress on an uncracked section of the uniaxial
tension member.
load was discussed in Section 2.2.2. The time-dependent response is described here.
The fundamental difference between a uniaxial tension member and a flexural member is the
strain gradient and the load carrying mechanism. In the uniaxial tension member, the stress
65
and strain distribution on the cracked section would not change with time, for concrete carries
no stress throughout. This is not the case for the cracked section of a flexural member, where
part of the concrete section is in compression. In a cracked reinforced concrete beam, the
and on a cracked section are described separately. The analyses described in the following
sections are based on the assumptions that plane sections before loading remain plane after
loading. Therefore both the short-term and long-term strain distributions on a section are
Consider the uncracked singly reinforced rectangular section shown in Figure 3-3a subjected
to a relatively small sustained moment Ms , so that the maximum tensile stress in the concrete
is always less than the tensile strength of the concrete (i.e. the section is uncracked). In this
case, the time-dependent change in strain on the cross-section is caused by creep. In the top
fibres of the concrete cross-section, compressive creep causes a gradual increase in the
compressive strain with time. In the bottom fibres, tensile creep causes the increase in the
tensile strains. The increase in tensile strain is restrained by the tensile reinforcing bars. Strain
compatibility requires that the tensile steel strain increases as the tensile concrete strain at the
steel level increases causing an increase in the tensile steel stress and an increase in the force
the concrete tensile stress and a lowering of the neutral axis (see Figure 3-3a).
66
a) Due to creep
b) Due to shrinkage
Due to Shrinkage:
The effects of concrete shrinkage on an uncracked and unloaded singly reinforced concrete
cross-section are shown in Figure 3-3b. Shrinkage of the concrete compresses the
reinforcement, which in turn imparts an equal and opposite tensile force on the concrete at the
steel level. The magnitude of the restraining force depends on the quality of bond and the
amount of slip. For uncracked members, the bond may be reasonably assumed to be good
with very little slip. Due to the eccentricity of the tensile restraining force T, a curvature is
induced on the cross-section and a gradual warping of the beam occurs, as shown in Figure 3-
3b. The magnitude of T is also dependent on the area of tensile reinforcement and the
geometry of the cross-section. It also depends on whether or not the concrete has cracked, i.e.
67
on the magnitude of the applied moment. Therefore, although shrinkage is independent of the
tensile force is great enough, cracking may be induced with time in a previously uncracked
member.
On a fully-cracked section, the tensile concrete loses its ability to carry tensile stress and all
the tensile force is carried by the reinforcing bars. Thus, under sustained bending moment,
only the concrete in the compressive zone undergoes significant creep. As creep strain
develops in the compressive zone, the top fibre strain increases gradually, the neutral axis
moves downwards and the compression zone becomes deeper. Aging of the concrete causes
the creep strain distribution at the section to become non-linear and, as a result of strain
compatibility, the instantaneous (or stress related elastic) strain also become nonlinear during
the period of sustained loading, as does the compressive stress distribution. The stress at top
fibre will reduces marginally with time and this affects the final creep strain in the top fibre.
As a result, under constant sustained moment, the tensile steel strain has to increase by a small
amount to compensate for the reduced lever arm between the resultant compressive and
a) Due to creep
b) Due to shrinkage
Due to Shrinkage:
After cracking, when subjected to the eccentric shrinkage induced restraining force imposed
by the reinforcement, the relatively small area of intact concrete above the crack must resist
the shrinkage induced restraining force imposed by the reinforcement T. Equilibrium and
small (as indicated in Figure 3-4b). The shrinkage induced curvature on the cracked section is
clearly significantly greater than that on an uncracked section, as can be seen by comparing
A cracked reinforced concrete beam contains both cracked and uncracked sections and the
response of the uncracked and cracked sections. The deformation of the beam increases with
time, with the rate of change being greatly dependent of the magnitude of shrinkage and creep
during the time period under consideration and the extent and severity of cracking.
If a beam begins to shrink prior to loading, as is commonly the case, shrinkage warping
occurs and the member deflects by an amount Gsh0,, when the applied moment is still zero (i.e.
M = 0), shown as point O’ in Figure 3-5. Therefore, the experimentally measured response of
70
a member that has already begun shrinking must start from a positive initial deflection, as is
shown in Figure 3-5. Because of the initial tensile stress induced by early shrinkage in the
tensile concrete, the moment required to cause first cracking Mcr.sh will be less than Mcr (as
indicated in Figure 3-5) and the moment-deflection response is given by the curve O’A’B’C’.
The initial deflection of the fully-cracked member due to early shrinkage (Gsh0.cr) is
significantly larger than that of the uncracked member (Gsh0), so early shrinkage causes the
line representing the fully-cracked cross-sectional response to move significantly to the right
(shown as line O’’D’ in Figure 3-5). At any value of applied moment greater than the cracking
moment (such as Ms in Figure 3-4), the instantaneous tension stiffening deflection Gtsi is the
distance BE if no shrinkage has occurred prior to loading and is the distance B’E’ Gtsi(t) if
shrinkage has occurred prior to first loading. It is likely that the magnitude of the tension
stiffening deflection is dependent on the amount of shrinkage that occurs prior to first loading,
as of course is the magnitude of the cracking moment, but this is yet to be conclusively
demonstrated.
For a simply-supported beam subjected to constant sustained loads, the bending moment Ms at
instantaneous mid-span moment versus deflection response of the member is shown as curve
OABC in Figure 3-6 (identical to curve OABC in Figure 3-5). The instantaneous fully-cracked
response (ignoring the tensile concrete) is also shown as line OD in Figure 3-6. If the member
does not shrink with time (i.e. Hsh remains at zero), creep causes an increase in deflection with
time and the member response shifts to curve OA’B’C’ in Figure 3-5a. The increase in
deflection with time is due to creep and may be expressed as 'G(t) = Gi M /D’, where Gi is the
71
instantaneous deflection, M is the creep coefficient and D’ is a factor that depends on the
amount of cracking and the reinforcement quantity and location. For reinforcement ratios
typical of beams, D’ is in the range 1.0 – 1.4 prior to cracking and in the range 4 - 6 when
cracking is extensive. With regard to the fully-cracked response (calculated assuming fully-
cracked cross-sections with no contribution from the tensile concrete), creep causes a
softening of the response shown as line OD’ in Figure 3-6a, with the slope of the line
When shrinkage after first loading is included, the deflection increases even further with time
due to shrinkage warping and the time-dependent response of the beam is shown as curve
O’A’B’C’ in Figure 3-6b. At zero loads (M = 0), the deflection increases due to shrinkage
warping of the uncracked member and the point O moves horizontally to O’. Due to the
restraint to shrinkage provided by the bonded reinforcement, tensile stress is induced with
time and this has the effect of lowering the cracking moment from Mcr to Mcr sh, as shown in
Figure 3-6b. For any flexural members subjected to a sustained moment in the range Mcr.sh <
Ms d Mcr, cracking will occur with time and the increase in deflection will be exacerbated by
the loss of stiffness caused by time-dependent cracking. In practice, many slabs are loaded in
this range. The fully-cracked response after creep and shrinkage is shown as line O’’D’ in
Figure 3-6b. The shrinkage induced deflection of the fully cracked member (ignoring the
tensile concrete) at zero loading (M = 0) is greater than that of the uncracked member and the
fully-cracked response is shifted horizontally from point O to point O’’ as shown. The slope
of fully-cracked response in Figure 3-6b is softened by creep and the slope of the line O’’D’
in Figure 3-6b is the same as the slope of line OD’ in Figure 3-6a.
With time, deflection increases from B to B’ while the fully-cracked response is also moved
72
from E to E’. The time-dependent tension stiffening deflection for the long-term tests
becomes from BE (tsi) to B’E’ (ts(t)). The difference between tsi and ts(t) in the figure
An alternative approach to analyse the singly reinforced section in sustained bending for
cracked and uncracked section is given by Gilbert (1988). This method will be used to analyse
the experimental results, to obtain the fully-cracked response of a flexural member. The
details of this method and a case study are presented in the Appendix B.
3.4 SUMMARY
This chapter provides qualitative description of the time-dependent responses of uniaxial
tension members and flexural reinforced concrete members. It is noted that the loss of tension
stiffening with time is associated with concrete cracking, either primary cracks or the
formation of additional cover-controlled cracks that cause local bond damage between
concrete and steel. Shrinkage is deemed to be a significant factor since the restraint to
shrinkage from the reinforcing bar can produce additional tensile stress in the concrete which
may indirectly induce cracking. Tensile creep plays a minor role in the decay of tension
stiffening with time as it gradually releases the tensile stress in the concrete.
The discussion concludes that the way to predict time-dependent tension stiffening in
reinforced concrete structures consideration must be given to the effects of creep and
shrinkage, as well as the effects of the cracking of concrete. An alternative approach to predict
the time-dependent responses of the cracked uniaxial tension member and flexural member
are proposed by the author. The approach is similar to that outlined in Eurocode 2 (1992), but
modified to include the effects of time-dependent slip at the concrete-steel interface caused by
73
4.1 OVERVIEW
The experimental program involved the testing of reinforced concrete prisms in axial tension
and flexural members (beams and slabs) in bending. The tension prism tests were undertaken
before the flexural tests, to first measure the effects of tension stiffening in reinforced
concrete in direct tension, before considering the imposition of the strain gradient associated
with bending. This chapter describes the testing process and presents the results of the
uniaxial tension tests. Chapter 5 discusses the flexural specimens and testing program.
To quantify tension stiffening in reinforced concrete members under increasing load and
To investigate the change in tension stiffening with time in a cracked member under
sustained service loads and, if significant, to measure the effect of creep and shrinkage on
To investigate the effects of the time-dependent change in tension stiffening on the crack
75
The primary test variables in the experiments were the load history, the reinforcement quantity,
and the creep and shrinkage characteristics of the concrete, which were measured throughout
the tests, from the commencement of drying until the end of the test. In addition, the
externally applied forces and reactions and the time-dependent displacements of the specimen
were recorded electronically throughout the test period, including the elongation for the
uniaxial tension members, the vertical deflection for the flexural members, the local steel
strains between the primary cracks, the crack width and the crack spacing.
dimensions, with details given in Figure 4-1. Each concrete prism was of square cross-section
(100mm by 100mm) and 1100 mm long and contains a single ribbed reinforcing bar running
longitudinally through the centroid of the cross-section. The tensile force P was applied to the
ends of the reinforcing bar protruding from each end of the concrete prism. Part of the tensile
force is transferred into the concrete from the reinforcement bar through bond at the steel-
concrete interface. Over the middle 600 mm length of each specimen, 25 strain gauges were
attached to the reinforcing bar at 25mm centres in order to monitor the local steel strains.
Typical formwork boxes for the uniaxial tension specimens, prior to the placement of concrete,
are shown in Figure 4-2, where strain gauges were attached to the reinforcing bar without
removing the ribs so as to minimise the effect of the strain gauge on bond.
Four of the specimens (STN12, STN16, STS12 and STS16) were tested under short-term
monotonically increasing load, with loading continuing into the post-yield range up to an
76
average steel strain of 0.3%. These four tests are referred to as the short-term tests. The other
four specimens (LTN12A, LTN12B, LTN12C and LTN12D) were subjected to a constant
sustained service load for a period of about 50 days and are referred to as the long-term tests.
The first two letters in the designation of each specimen indicate the test duration; “ST” for
short-term and “LT” for long-term. The third letter indicates whether or not the specimen
commenced drying and began to shrink prior to the application of load; “S” if yes and “N” if
no. The next two digits indicate the reinforcing bar diameter, either 12 mm or 16 mm. For the
four long-term tests, the final letter A, B, C and D distinguishes the different load cases.
To investigate the influence of reinforcing ratio on tension stiffening, two bar sizes were used,
namely 12mm and 16mm diameter bars. Therefore, the reinforcement ratio for each specimen
and STS16) is shown in Figure 4-3. The tests were undertaken in an Instron universal testing
machine. The specimens were positioned vertically within the testing machine as shown. The
member elongation was measured by attaching a pair of LVDTs (linear variable displacement
transducer) to a mounting frame that was clamped firmly to the specimen at both the front and
back surfaces. The gauge length of the LVDTs over which the elongation of the specimen was
measured was 900mm. The tests were undertaken in displacement control, with the rate of
tensile force was recorded throughout via the pressure transducer connected to the machine
grips. Loads, elongation and steel strains (strain gauges) were all recorded electronically using
For the long-term tests, the rig shown in Figure 4-4 was designed and constructed. The
specimen was positioned vertically by fixing the reinforcing bar protruding from the
specimen through holes in the cross-heads both top and bottom. A constant sustained tensile
force was imposed on each specimen through an adjustable anchor-support at the top of the
rig and monitored using a load cell connected to the reinforcing bar at the base of the rig.
Five Demec targets were attached on opposite sides of each tension member at 250 mm
centres, so that the averaged elongation of the specimen at specific loads could be obtained.
The strain gauges attached to the reinforcing bar were connected to the HBM amplifier and
the readings were taken manually at regular time intervals throughout the period of sustained
loading. Crack widths were also measured throughout the test using a microscope with a
79
cured under wet burlap. The burlap was kept wet for 28 days to facilitate strength gain of the
concrete and to delay the commencement of drying shrinkage. The Demec gauge targets were
glued onto the concrete surface after removing the burlap. For the short-term tests, specimens
STN12 and STN16 were tested immediately after wet curing, so that relatively little shrinkage
had occurred at the time of testing. Specimens STS12 and STS16 were uncovered and
For long-term test specimens, LTN12A and LTN12B were cast from one batch of concrete
(the same as the short-term specimens), while LTN12C and LTN12D were cast from a
80
different batch. All specimens were cured under wet burlap until the commencement of the
test. After curing, the specimens were put into the testing rig and the load was imposed by
manually adjusting the anchorage supports. LTN12A and LTN12C was loaded up to 40kN (all
cracks stablised), while LTN12B and LTN12D were subjected to 20kN (which was close to
the cracking load), and then the loads were maintained constant for 50 days. Table 4-1 shows
The basic concrete properties were measured on companion specimens at the onset of each
test. The compressive strength and the elastic modulus of concrete were measured on standard
100 mm diameter concrete cylinders. The indirect concrete tensile strength was measured on
standard cylinders using the Brazil tests and the flexural tensile strength was measured on
The creep and shrinkage characteristics of the concrete were also measured on companion
specimens throughout the test. Unrestrained concrete prisms, each 600mm long with
100×100mm cross section, were used to measure the development of shrinkage strain with
time (Figure 4-5b). Brass plugs at 250 mm centres were embedded in opposite surfaces of the
prism during casting, and a Demec gauge measure was used to record the average shrinkage
strain between the two plugs in the prism. The deformation between the brass plugs was
a) Creep specimens
b) Shrinkage specimens
The standard creep rigs were set up as shown in Figure 4-5a. Two standard 150mm diameter
cylinders were loaded in the rig with a maximum 6MPa compressive stress applied by the
hydraulic jack at the bottom of the creep rig. The stress was then kept constant. Two other
83
standard cylinders were left unloaded and drying besides the rigs. The creep strain at any time
can therefore be calculated by subtracting the shrinkage strain of the unloaded cylinders and
the instantaneous strain of the loaded cylinders from the total strain measured on the loaded
cylinders at that time. Creep tests were initiated at the commencement of drying and at the
The stress-strain curves of the deformed reinforcing bars were obtained in the Instron
universal testing machine. The tensile force and the position of the upper rod were recorded.
Therefore stress can be calculated by dividing the tensile force by the bar area, and the
average strain can also be derived by dividing the measured elongation by the gauge length.
There are basically two different sizes of deformed 500 Grade reinforcing bars used in the
experimental program, namely N12 and N16 bars. Three N12 and three N16 bars that were
from the same provider were tested in the Instron machine. The measured yield stress and
elastic modulus were then averaged and the results are given in Table 4-2.
The concrete used in this study was supplied by a pre-mixed concrete producer (BORAL
x Slump = 80mm
The compressive strength fcp, the indirect tensile strength fct, the flexural tensile strength fcf
and the elastic modulus of concrete Ec are presented in Table 4-3. For the short-term
specimens, these properties were measured at the same days of the tests; while for the long-
term specimens, they were measured at both the beginning and the end of the sustained
loading time. There were two identical companion specimens were tested to measure these
instantaneous properties at the time of each test. In general, the concrete strengths increase
with time by less than 16% during the sustained loading period.
Figure 4-6 shows the development of the creep coefficient c with time for the plain concrete
elements measured in the companion tests. The creep coefficient is the ratio of the creep
strain to instantaneous strain soon after first loading at different concrete ages. There were
two identical creep tests conducted for each batch of concrete as shown. The results are then
Specimen f cp f ct f cf Ec
STN12
21.56 2.04 3.05 22400
STN16
STS12
24.73 2.15 3.08 21600
STS16
(a) Short-term uniaxial tension specimens
f cp f ct f cf Ec
Specimen
Start End Start End Start End Start End
LTN12A
21.56 24.88 2.04 2.44 3.05 3.54 22400 23600
LTN12B
LTN12C
27.73 29.02 2.85 3.02 4.00 4.40 20500 22300
LTN12D
(b) Long-term uniaxial tension specimens
c 1.6 1.6
c
1.2 1.2
0.8 0.8
creep test 1
creep test 1 creep test 2
0.4 creep test 2 0.4 Averaged
Averaged
0.0 0.0
20 40 60 80 20 40 60 80
Concrete age (days) Concrete age (days)
(a) Batch 1 (STN12, STN16, STS12, STS16, (b) Batch 2 (LTN12C and LTN12D)
LTN12A and LTN12B)
Figure 4-6: Creep coefficients for the uniaxial tension tests.
86
For the short-term tests, the averaged creep coefficients at the onset of tests are 0.17 (STN12
and STN16) and 1.13 (STS12 and STS16) respectively. For the long-term tests, the averaged
creep coefficients by the end of the sustained loading periods are 1.38 (LTN12A and LTN12B)
Figure 4-7 shows the development of shrinkage strains after drying commenced. There were
also two companion shrinkage tests conducted for each batch of concrete. The measured
shrinkage strains are then averaged and plotted in the figures. For the short-term tests, the
averaged shrinkage strain at the onset of tests are -28 x 10-6 (STN12 and STN16) and -249 x
10-6 (STS12 and STS16) respectively. For the long-term tests, the averaged shrinkage strain
by the end of the sustained loading periods are -310 x 10-6 (LTN12A and LTN12B) and -356 x
400 400
Shrinkage strain ( ×10- 6
300 300
200 200
shrinkage test 1
100 100 shrinkage test 2
Averaged
0 0
0 20 40 60 80 0 20 40 60 80
Concrete age (days) Concrete age (day
Specimen STN12 and STN16 were subjected to monotonically increasing deformation soon
after the end of moist curing. The elastic modulus of concrete at the time of testing at age 32
days was Ec = 22400 MPa (and with Es = 200000 MPa, n = 8.93) and the tensile strength of
concrete was f ct = 2.04 MPa. Prior to loading, the measured drying shrinkage strain in the
unreinforced companion member was small at Hsh0 = -28 x 10-6 and the corresponding creep
coefficient was c = 0.17. The calculated strains in the specimen (accounting for the restraint
provided by the reinforcement – see Appendix A Equation A.1) were -25 x 10-6 for STN12 and
The measured axial load versus strain relationships for STN12 and STN16 are given in Figure
4-8 and 4-9 respectively, together with the bare bar response. Initially, prior to first cracking
in portion O’B of the curves in the figures, the specimens are at their stiffest and the load-
strain curves are steep. At first cracking (Point B when P = Pcr = 21.1kN and 23.0kN), there is
an abrupt change of stiffness and the stiffness continues to degrade under increasing
deformation as further cracks occur (portion BC). As the load increases, the tension stiffening
strain gradually reduces. In total, 5 primary cracks occurred in each specimen as loading
progressed, as indicated by the numbered peaks in the curves in Figure 4-8 and 4-9.
Tables 4-4 and 4-5 provide values of average axial strain Hs.avg and tension stiffening strain Htsi
88
at selected values of total applied load P, together with the average force carried by the
concrete and the steel, Pconc and Psteel, respectively, within the prisms. Psteel is derived by
multiplying the measured average axial strain with the elastic modulus and the cross-section
area of the steel bar. Subtracting Psteel from the applied load P gives Pconc.
It appears that with increasing load P, Htsi or Pconc drops off rapidly, whereas Psteel keeps
increasing. The decrease in Htsi or Pconc does not only occur at the crack formation stages, but
also after the cracks have stabilised. For example, Pconc of STN12 decreases from 4.4kN to
1.8kN as the load increases from 45kN to 55kN. Note that all the cracks had formed when P
reached 40kN. For STN16, Pconc decreases from 7.1kN to 5.0kN as the load increases from
65kN to 85kN after all the cracks had formed. The mechanism to induce this drop-off in
70
60
Axial Load (kN)
50 Experimental C
Hts 5
40 Bare bar
30 4 Crack numbers
Pcr B1 2 3
20 = 21.1 kN
10 -6
Average Axial Strain (x10 )
0 O’
-500 0 500 1000 1500 2000 2500 3000
-25
120
100 Experimental
C
Axial load (kN)
Bare bar
80
60
crack number
40 B
P cr = tsi
20 -6
23.0kN Average Axial Strain (×10 )
O'
0
-500 0 500 1000 1500 2000 2500 3000
The variation of steel strains over the 600mm gauge length of the reinforcing bar were
measured as the test progressed using strain gauges and the variation in force in the
reinforcing steel was determined from the measured strains. The variation in concrete force
was obtained by subtracting the force in the steel at each strain gauge location from the total
90
Figure 4-10 and 4-11 show the variation in the tensile forces carried by steel and concrete at
different loading stages for the short-term specimen STN12 and STN16. As can be seen, the
formation of a primary crack causes a large local decrease of force in the concrete and an
increase of force in the steel bar. All the tensile force is carried by the steel at the cracked
section. At a certain distance S from the crack, the tensile force carried by the concrete is
unaffected by the crack as shown in Figure 4-10b and Figure 4-11b. Of course, at cracking,
the sudden loss of stiffness causes a drop-off in the applied load and the force carried by the
concrete reduces accordingly as can be seen by comparison of Figure 4-10a and b and Figure
4.11a and b.
For example, just before first cracking of STN12 at P = 21.1kN (Figure 4-10a), the tensile
force in the concrete is almost uniform and equal to about 19kN, while the force in the steel is
about 2.1kN. Soon after first cracking (Figure 4-10b), the applied load drops to 17.8kN. At the
cracked section, steel carries the entire load and the force in the concrete drops to zero. At
distances greater than S | 140mm from the first crack, the concrete force is again uniform and
equal to about 16kN. As the load increases to 22.4kN (Figure 4-10c), the maximum tensile
force in the concrete exceeds 20kN, which allows the concrete tensile stress to exceed the
tensile strength of the concrete at the second weakest cross-section, thus a second primary
25 25
S=140mm
20 20
Force (kN)
Force (kN)
15 15
10
Steel 10
Steel
Concrete Concrete
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(a) Just before first cracking, P = 21.1 kN. (b) Just after first cracking, P = 17.8 kN.
25 25
Steel
20 20 Concrete
Force (kN)
Force (kN)
15 15
Steel
10 10
Concrete
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(c) Just before 2nd crack, P = 22.4 kN. (d) Just after 2nd crack, P = 17.4 kN.
30 30
Steel Steel
25 25
Concrete Concrete
Force (kN)
20
Force (kN)
20
15 15
10 10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(e) Just before 3rd crack, P = 21.8 kN. (f) Just after 3rd crack, P = 19.6 kN.
60
50
50
40
40
Force (kN)
Force (kN)
30
30
Steel Steel
20 Concrete
Concrete 20
10
10
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(g) At P = 40.0 kN. (h) At P = 50.0 kN.
Figure 4-10: Variation of forces in steel and concrete at different loading stages (STN12).
92
25
25
Steel S=150mm
20 Concrete
20
Force (kN)
Force (kN)
15 15
Steel
10 10
Concrete
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(a) Just before first cracking, P = 23.0 kN. (b) Just after first cracking, P = 20.8 kN.
30 25
Steel Steel
25 Concrete 20 Concrete
Force (kN)
20
Force (kN)
15
15
10
10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(c) Just before 2nd crack, P = 24.3 kN. (d) Just after 2nd crack, P = 22.7 kN.
30
30 Steel
Steel
25 Concrete
25 Concrete
Force (kN)
20
Force (kN)
20
15 15
10 10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm Distance along prism, mm
(e) Just before 3rd crack, P = 24.9 kN. (f) Just after 3rd crack, P = 22.8 kN.
60 100
Steel 90
50 Concrete 80
Force (kN)
70
Force (kN)
40
60
30
50
20 40 Steel
30 Concrete
10
20
0 10
200 300 400 500 600 700 800 900 0
Distance along prism, mm 200 300 400 500 600 700 800 900
Distance along prism, mm
(g) At P = 40.0 kN.
(h) At P = 75.0 kN.
Figure 4-11: Variation of forces in steel and concrete at different loading stages (STN16).
93
Thereafter, the development of new primary cracks follows a similar pattern (Figure 4-10e
and f) until all the cracks have formed at P = 40kN (Figure 4-10g). It is found that there is no
dramatic change of local steel forces or concrete forces when 40kN < P < 50kN (Figure 4-
10h), just a gradual reduction in average concrete force with increasing load.
Similar to STN12, STN16 also undergoes the same pattern for the development of local forces
The crack locations and crack numbers for STN12 and STN16 are shown in Figure 4-12 and
4-13. The crack numbers indicate the order of the formation of each crack, corresponding to
5 2 3 1 4
5 3 4 1 2
The maximum crack spacings for STN12 and STN16 are 320mm and 240mm respectively,
while the average crack spacings are 195mm and 165mm. STN12 (with the smaller
reinforcing ratio) exhibits slightly greater crack spacings than STN16. Using Equation 2.7 of
the Tension Chord model, the maximum crack spacings for STN12 and STN16 are calculated
94
as 263mm and 192mm. It appears that the measured maximum crack spacing is greater than
the one predicted by the Tension Chord model. It is often assumed that the average spacing
can be taken as the average of maximum crack spacing Sr.max and minimum crack spacing
Sr.max/2 (Equation 2.8). Therefore the calculated average crack spacings according to the
Tension Chord model are 197.3mm and 144.2mm, which are close to the measured average
values.
The crack widths at various stages of loading are given in Tables 4-6 and 4-7, together with
the steel stress (Vs) at the crack at each loading stage. Generally crack width increases as the
steel stress increases. The data shows that after cracks have stabilised, the maximum crack
width is about 1.5 – 2.0 of the average crack width. For example, at, Vs 440MPa, the
average crack widths for STN12 and STN16 are 0.200mm and 0.245mm, while the maximum
crack widths are 0.375mm and 0.375mm respectively (1.875 and 1.53 times of the average
values). In addition, the maximum crack width seems not to be affected by the bar diameter.
At similar steel stresses after cracks have stabilised, the measured maximum crack widths are
Maximum
Width of Width of Width of Width of Width of Average
Vs
Load crack
1st crack 2nd crack 3rd crack 4th crack 5th crack crack width
Stage (MPa) width
(mm) (mm) (mm) (mm) (mm) (mm)
(mm)
After 1st 104 0.013 - - - - 0.013 0.013
After 2nd 113 0.025 0.05 - - - 0.0375 0.05
After 3rd 114 0.05 0.075 0.05 - - 0.0583 0.075
After 4th 196 0.10 0.25 0.125 0.05 - 0.1313 0.25
After 5th 299 0.125 0.325 0.10 0.225 0.025 0.160 0.325
75 kN 373 0.15 0.325 0.225 0.15 0.125 0.215 0.325
90 kN 447 0.20 0.375 0.25 0.25 0.15 0.245 0.375
105 kN 522 0.25 0.50 0.25 0.275 0.225 0.3 0.50
Specimen STS12 and STS16 were identical to specimen STN12 and STN16 respectively,
except that they were tested 3½ weeks later at age 57 days when the drying shrinkage had
increased to Hsh = -249 x 10-6 in each specimen At the time of testing at age 57 days, Ec =
21600 MPa (i.e. n = 9.26) and fct = 2.15 MPa. The creep coefficient associated with the initial
period of shrinkage was Mc = 1.13. The calculated average strain in the specimen prior to the
H sh -6
H sh 0 209 u 10 6 (STS12) and -185×10 (STS16)
1 n U *
The average load versus strain measured using LVDTs throughout the tests is plotted in
96
Figures 4-14 and 4-15, together with the bare bar responses for the specimens. Obviously the
bare bar response for the shrunk members is identical to that of the non-shrunk member, as
It is noted that the average axial strain in the specimens before loading (P = 0 kN) is negative
due to early shrinkage. The early shrinkage in the experiment also resulted in a significant
reduction in the load at first cracking Pcr (13.0 kN in STS12 compared to 21.1 kN in STN12
and 11.6kN in STS16 compared to 23.0kN in STN16). A total of 5 and 6 primary cracks
occurred in the specimen STS12 and STS16 respectively, as shown in the figures. The
responses of STS12 and STS16 after first cracking were similar to the non-shrunk specimens,
with the tension stiffening strain gradually diminishing and the responses of the prisms
approaching the bare bar responses as the applied load were increased.
70
60
Axial Load (kN)
C
50 Experimental Hts
40 Bare bar
Pcr.sh = 13.0 kN
30 5
2 3 4
20 B 1
10 -6
O’ Average Axial Strain (x10 )
0
-500 0 500 1000 1500 2000 2500 3000
-209
120
100
C
Axial Load (kN)
80
60
Crack numbers
40
B 6 STS16
20 345 Pcr.sh = 11.6 kN Bare Bar
12
0
O’
-500 0 500 1000 1500 2000 2500 3000 3500
-185
-6
Average Axial Strain (x10 )
Tables 4-8 and 4-9 provide values of average axial strain Hs.avg and tension stiffening strain Htsi
at selected values of total applied load P, together with the average force carried by the
concrete and the steel, Pconc and Psteel, respectively, within the shrunk prisms. Similar to the
non-shrunk members, the values of Htsi and Pconc decrease as the applied loads increase at both
crack formation stage and at the stabilised crack stage. However, compared with the non-
shrunk members, the values of Hs.avg and Psteel of the shrunk members have negative
magnitudes before cracking due to initial shrinkage, while tension stiffening is greater in
terms of the values of Htsi and Pconc before cracking. After first cracking, this difference seems
to be less significant, as the values of Htsi and Pconc drop off quickly as the applied loads
increase.
98
Average strain, Hs.avg (x 10-6) -209 -171 -154 347 934 1452 2243
Tension stiffening strain Hts (x 10-6) 209 613 729 538 393 318 191
Avge force in concrete, Pconc (kN) +4.7 13.9 16.5 12.2 8.9 7.2 4.3
Avge force in steel bar, Psteel (kN) -4.7 -3.9 -3.5 7.8 21.1 32.8 50.7
Average strain, Hs.avg (x 10-6) -185 -149 -140 179 469 879 1461 2049
Tension stiffening strain Hts (x 10-6) 185 398 429 319 277 240 156 65
Avge force in concrete, Pconc (kN) 7.4 16.0 17.2 12.8 11.1 9.7 6.3 2.6
Avge force in steel bar, Psteel (kN) -7.4 -6.0 -5.6 7.2 18.9 35.3 58.7 82.4
The variations of steel and concrete forces in the middle 600 mm of the prism at various
20
20
S=120mm
15
15
Force (kN)
Force (kN)
10 Steel
10
Concrete Steel
5 Concrete
5
0
200 300 400 500 600 700 800 900 0
200 300 400 500 600 700 800 900
-5
Distance along prism, mm -5
Distance along prism, mm
(a) Just before first cracking, P = 13.0 kN. (b) Just after first cracking, P = 11.1 kN.
20
20
15
15
Force (kN)
Force (kN)
Steel Steel
10 10
Concrete Concrete
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
-5 -5
Distance along prism, mm Distance along prism, mm
(c) Just before 2 crack, P = 15.0 kN. nd
(d) Just after 2nd crack, P = 13.3 kN.
25
25
Steel
Steel
20 Concrete 20
Concrete
Force (kN)
Force (kN)
15 15
10 10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
-5 -5
Distance along prism, mm Distance along prism, mm
th
(e) Just before 4 crack, P = 18.1 kN. (f) Just after 4th crack, P = 15.9 kN.
60 60
Steel 50
50
Concrete
40
Force (kN)
40
Force (kN)
30
30 Steel
20 Concrete
20
10
10
0
0 200 300 400 500 600 700 800 900
200 300 400 500 600 700 800 900
-10
-10
Distance along prism, mm Distance along prism, mm
(g) At P = 40.0 kN. (h) At P = 50.0 kN.
Figure 4-16: Variation of forces in steel and concrete at different loading stages (STS12).
100
20 20
S=105mm
15 15
Force (kN)
Force (kN)
10 Steel 10 Steel
Concrete Concrete
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
-5
-5
-10
Distance along prism, mm Distance along prism, mm
-10
(a) Just before first cracking, P = 11.6 kN. (b) Just after first cracking, P = 11.0 kN.
25 25
Steel
20 20
Concrete
15 15
Steel
Force (kN)
Force (kN)
10 Concrete 10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
-5 -5
Distance along prism, mm Distance along prism, mm
-10 -10
(c) Just before 2nd crack, P = 13.8 kN. (d) Just after 2nd crack, P = 12.7 kN.
25 Steel 25
Concrete Steel
20 20
Concrete
15 15
Force (kN)
Force (kN)
10 10
5 5
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
-5 -5
Distance along prism, mm Distance along prism, mm
-10 -10
(e) Just after 3rd crack, P = 13.0 kN. (f) Just after 4th crack, P = 15.6 kN.
80 100
80
60
Force (kN)
Force (kN)
60
40
40 Steel
Steel Concrete
20 Concrete
20
0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along prism, mm
Distance along prism, mm
(g) At P = 50.0 kN.
(h) At P = 90.0 kN.
Figure 4-17: Variation of forces in steel and concrete at different loading stages (STS16).
101
Consider specimen STS12. Just before 1st cracking at P = 13.0kN (Figure 4-16a), the force in
the concrete is about 16.0kN, and the force in the steel is about -3.0kN. Soon after first
cracking at P = 11.0kN, the distance S away from the crack (within which concrete force is
built up through bond) is about 120mm, which is less than that of STN12 (140mm). It is
interesting to note that at the region where bond stress is not affected by the crack, the force in
the concrete exceeds the total applied load (due to the restraint to early shrinkage) and the
steel force still remains negative. As the load P increases, the local steel forces between the
cracks increase rapidly and then become positive at the crack stabilised stage (P > 40kN). A
comparison between STN12 and STS12 in terms of local concrete and steel forces at P =
40kN indicates that there is little difference between the two specimens at crack stabilised
stage. For example, the maximum forces in the concrete in-between the cracks for STN12 and
STS12 at P = 40kN are both around 19kN (Figure 4-10g and Figure 4-16g).
The crack locations and crack numbers for STS12 and STS16 are shown in Figure 4-18 and 4-
19. The maximum crack spacings for STS12 and STS16 are 245mm and 145mm, while the
average crack spacings are 177.5mm and 125mm respectively. Compared to the non-shrunk
members, the shrunk members exhibited smaller crack spacings. It has been already discussed
that the crack spacing Srm is dependent on the distance S, as indicated in the diagrams of local
steel forces and concrete forces (Figures 4-10, 4-11, 4-16, and 4-17). The values of S are
smaller in the shrunk member than in the non-shrunk member, as well as the values of Srm.
One can say that the influence of primary cracking on the local deterioration of bond around
the crack is more significant in the non-shrunk members than in the ones with significant
shrinkage. In the shrunk member, at first cracking, the tensile stress can be quickly transferred
102
to the surrounding concrete with a shorter distance S from the first crack, which indicates that
the second crack will occur in a closer distance to the first crack.
5 1 2 4 3
1 3 6 4 2 5
The crack widths at various stages of loading are given in Table 4-10 and 4-11, together with
the steel stress (Vs) at the crack at each loading stage. At the crack stabilised stage, the
measured maximum crack width is about 1.3 – 1.5 of the average crack width. For example,
at, Vs 440MPa, the average crack widths for STS12 and STS16 are 0.385mm and 0.288mm,
while the maximum crack widths are 0.525mm and 0.425mm respectively (1.36 and 1.48
times of the average values). With initial shrinkage, the member with less reinforcement ratio
(STS12) exhibited greater maximum crack width. At a steel stress of Vs 440MPa the
maximum crack width for STS12 is 0.525mm and for STS16 at Vs 440MPa, the maximum
Maximum
VsWidth of Width of Width of Width of Width of Average
Load crack
1st crack 2nd crack 3rd crack 4th crack 5th crack crack width
Stage (MPa) width
(mm) (mm) (mm) (mm) (mm) (mm)
(mm)
After 1st 98 0.025 - - - - 0.025 0.025
After 2nd 118 0.05 0.075 - - - 0.063 0.075
After 3rd 132 0.125 0.20 0.125 - - 0.15 0.20
After 4th 141 0.15 0.25 0.15 0.15 - 0.175 0.25
After 5th 265 0.175 0.35 0.225 0.175 0.05 0.195 0.35
40kN 354 0.125 0.40 0.50 0.375 0.30 0.34 0.50
50kN 442 0.25 0.425 0.525 0.425 0.30 0.385 0.525
The curves in the following pages show the average concrete tensile stress versus the average
axial strain for each short-term test prism. Figure 4-20 illustrates the influence of initial
shrinkage while Figure 4-21 shows the influence of reinforcing ratio. The average concrete
stress herein is calculated by dividing the average concrete tensile force Pconc with the cross
section area.
104
2.5
STN16 (Experimental)
STN12 (Experimental) 2
2
STS16 (Experimental)
STS12 (Experimental)
1.5
1.5
1
1
0.5
0.5
0
0
-500 0 500 1000 1500 2000 2500
-500 0 500 1000 1500 2000 2500
Average strain ×10-6
Average strain ×10-6
(a) STN12 and STS12 (U = As/Ac = 0.011). (b) STN16 and STS16 (U = As/Ac = 0.021).
Figure 4-20: The effects of shrinkage on the average concrete tensile stress.
The dots at the left hand end of each curve in Figures 4-20 signify the situation at the
beginning of loading, where an initial tensile stress existed in the concrete due to shrinkage
(Hsh) prior to the application of the external load P. The tensile strength of concrete is a highly
variable material property and, in these four specimens, the concrete tensile stress at first
cracking was 2.00 MPa and 2.11 MPa for STN12 and STN16, respectively, and 1.67 MPa and
1.76 MPa for STS12 and STS16, respectively. The concrete in both specimens with
significant initial shrinkage (STS12 and STS16) had a significantly lower tensile strength than
the concrete in the other two specimens (despite being almost 28 days older at the time of
testing). It is generally known that tensile strength of concrete increases with time. As STS12
and STS16 were tested in 31/2 weeks after STN12 and STN16, the tensile strength of STS12
and STS16 must also be greater. The fact that the calculated tensile stresses at first cracking
for STS12 and STS16 are less than STN12 and STN16 may be attributed to an
underestimation of restraint to initial shrinkage measured during the drying period prior to
loading. Therefore, the dots on the left hand end for the curves of the shrunk member may
After initial cracking, the average concrete stress in all specimens gradually reduced as the
deformation increased. There is minor difference of average concrete tensile stress between
the shrunk and non-shrunk specimens at in-service load levels (Figure 4-20). Although it
might appear that the concrete tensile stress (in effect the tension stiffening) decreases at a
slightly slower rate if significant shrinkage occurs prior to first cracking, but further testing is
2.5 2.5
Average concrete tensile stress (MPa
1 1
0.5 0.5
0 0
0 500 1000 1500 2000 2500 -500 0 500 1000 1500 2000 2500
-6
Average strain ×10 Average strain ×10-6
The average concrete tensile stress versus axial strain curves for STN12 and STN16 (Figure
4-21a) are remarkably similar, as are the curves for STS12 and STS16 (Figure 4-21b). While
the average tensile stress in the concrete after cracking appears to be not significantly affected
by the bar diameter (and reinforcement ratio U = As/Ac), the contribution of the cracked tensile
concrete to the stiffness of the member is a larger proportion of the total stiffness when the
This is illustrated in Figure 4-22 where the tension stiffening strain is plotted against the
average axial strain caused by load. After cracking, the tension stiffening strain in STN12 is
significantly greater than in STN16 with the ratio between them under service conditions (500
x 10-6 < Hs.avg < 2000 x 10-6) in the range 1.7 to 1.8. It is noted that the inverse of the ratio of the
106
reinforcement ratios for each specimen is 1.79. It appears that the tension stiffening strain is
almost inversely proportional to the reinforcement ratio, with tension stiffening becoming
900
strai (×10 )
-6
-6 800
×10 STN12 (Experimental)
700
stiffeningstrain
STN16 (Experimental)
600
tensionstiffening
500
400
300
Tension
200
100
0
0 500 1000 1500 2000 2500
-6
Average strain ×10
Figure 4-22: Tension stiffening strain versus axial strain (STN12 and STN16).
In order to investigate the local deterioration of bond at the crack stablised stage, the tensile
forces in the concrete around the 3rd crack at 40kN and 50kN in STN12 (Figure 4-10g and h)
are plotted in Figure 4-23. The bond stress is proportional to the gradient of the tensile stress
distribution. The gradient of the tensile stress in the concrete decreases as the applied load
increases, which means that the bond stress decreases. Since no other primary crack has
occurred, this decrease of bond stress must be attributed to local damage of bond due to
Force (kN)
30
At 40kN
25 At 50kN
20
3rd crack section
15
10
5
0
400 500 600 700
Distance along the prism (mm)
The maximum crack widths measured in the experiments and given in Tables 4-6, 4-7, 4-10
and 4-11 are plotted against steel stress at each crack in Figure 4-24 and a line of best fit for
each specimen is shown. It appears that shrinkage strain in the concrete prior to cracking has a
significant effect on maximum crack widths (as shown in Figure 4-24a and 24b). For a
particular steel stress, the crack widths are greater when the specimen has been exposed to
significant shrinkage prior to loading. The effect of changing the reinforcement ratio on crack
widths is shown in Figures 4-24c and 24d. Figure 4-24c compares the steel stress versus
maximum crack width curves for the two specimens without significant shrinkage. From
these tests it appears that reinforcement ratio has little effect. However, for the two specimens
with significant initial shrinkage, the specimen with the smaller reinforcement ratio has
500 500
400 400
300 300
200 200
STN12
100 STS12 100 STN16
STS16
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
400 400
300 STS16
300
STN12
200 200
STS12
STN16 100
100
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
For the long-term test, the average strain was measured at pre-selected times throughout the
period of loading using Demec gauges attached on the opposite sides of the concrete surface.
On each side, 5 Demec points were fixed to the concrete surface at 250mm centres and 4
readings were taken at each time instant. The readings were averaged to obtain the average
axial strain of the specimen (i.e. the sum of the elastic, crack opening, creep and shrinkage
Specimen LTN12A was moist cured until two days before the test started (at age 33 days).
109
The measured elastic modulus of concrete at the time of first loading was Ec = 22400 MPa (n
= 8.93) and the tensile strength of concrete was ft = 2.04 MPa. The creep coefficient for this
initial period of drying (up to age 33 days) was 0.3 (i.e. E e = 18060 MPa, n * = 11.1). The
initial concrete tensile stress caused by restraint to shrinkage prior to loading is Vsh0= 0.11
MPa. The initial drying shrinkage strain was Hsh0 = -52 x 10-6 and, accounting for restraint
provided by the reinforcing bar, the strain in the concrete was -46 x 10-6. Specimen LTN12C
was tested right after moist curing (at age 28 days), therefore the shrinkage prior to loading is
The shrinkage strains at the end of each sustained load period were -310 x 10-6 for LTN12A
and -356 x 10-6 for LTN12C respectively. The creep coefficients associated with the sustained
Both specimens were initially loaded at 5kN increments up to a load P = 40kN, at which load
5 cracks had developed, and this load level was then held constant for the duration of the tests
Figures 4-25 and 4-26 plot the axial load versus measured average strain responses of
LTN12A and LTN12C. It can be seen that in both tests, the loads were maintained constant at
40kN, after all primary cracks had formed. During the long-term loading period, the graph
indicates the increment of axial average strain with time. Since the bare bar strain did not vary
with time under constant axial force, tension stiffening strain must decrease with time.
For specimen LTN12A, the average axial strain immediately after application of the 40 kN
load was Hs.avg(0) = 1409 x 10-6 and the tension stiffening strain was Hts(0) = 361 x 10-6. The
110
average stress in the steel is therefore Vs(0)= Es Hs.avg(0) = 282 MPa and the average force
carried by the steel bar is Ps(0) = As Vs(0) = 31.84 kN. This means the average tensile force in
the concrete is Pc(0) = P - Ps(0) = 8.16 kN and the average concrete stress is Vc(0) = Pc(0)/Ac
= 0.825 MPa. After 50 days under load, the average axial strain had increased to Hs.avg (50) =
1556 x 10-6 and the tension stiffening strain had reduced to Hts(50) = 214 x 10-6. The
corresponding steel and concrete forces and stresses are Vs(50)= Es Hs.avg(50) = 311 MPa;
Ps(50) = As Vs(50) = 35.17 kN; Pc(50) = P - Ps(50) = 4.83 kN and the average concrete stress
is Vc(50) = Pc(50)/Ac = 0.49 MPa. It appears that for specimen LTN12A, initially with a fully
developed crack pattern, the average tensile stress in the concrete and the tension stiffening
strain reduced to 59.4% of their initial value in the first 50 days under sustained load.
As for specimen LTN12C, similar behaviour is found in the experiment. Within the first 50
days, tension stiffening strain Hts dropped from 291 x 10-6 to 207 x 10-6. Meanwhile, the
average tensile stress in the concrete Vc decreased from 0.67MPa to 0.47MPa. Both values
dropped to about 70% of its initial value during the sustained loading period.
60
LTN12A load sustained for 50 days
50
Axial Load,P (kN)
Bare bar
40
30
20
10
Average strain (x 10-6)
0
-250 250 750 1250 1750
-46
60
40
30
20
10
Average strain (x 10-6)
0
0 500 1000 1500 2000 2500
The variation of steel strains over a 600mm gauge length of the reinforcing bar was measured
using strain gauges as the time under load increased. Figures 4-27 and 4-28 show the variation
in the tensile stresses carried by concrete and steel immediately after the application of the full
sustained load and after 50 days under the sustained load for specimens LTN12A and
LTN12C, respectively.
3.0 400
Initial concrete stress (time = 0) 350
2.5
Final concrete stress (t = 50 days)
Stress (MPa)
300
Stress (MPa)
2.0
250
1.5 200
1.0 150
Initial steel stress (time = 0)
100
0.5 Final steel stress (t = 50 days)
50
0.0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along specimen (mm) Distance along specimen (mm)
(a) Concrete stress. (b) Steel stress.
Figure 4-27: Initial and final concrete and steel stresses LTN12A (P = 40 kN).
112
3.0
400
2.5 350
Initial concrete stress (time = 0)
Final concrete stress (t = 50 days) 300
2.0
Stress (MPa)
Stress (MPa)
250
1.5 200
The results show that within the 50 days under sustained loads, significant decrease of
concrete stress and increase of steel stress are observed. For example, the maximum
instantaneous stress in the concrete of LTN12A had dropped from 1.65MPa to 0.7MPa at the
end of the test. Since the external applied loads remained constant during the tests and no
other primary cracks were observed, the loss of tensile stress in the concrete must be
The crack patterns of LTN12A and LTN12C both immediately after loading and after 50 days
under sustained load are shown in Figure 4-29 and 4-30. The maximum crack spacings for
LTN12A and LTN12C are 295mm and 290mm, while the average crack spacings are 194mm
3 1 4 2 5
5 1 2 4 3
The crack widths at different stages for the specimens are tabulated in Table 4-12 and 4-13. It
appears that the maximum crack width of LTN12A and LTN12C increased by 80% and 16%
Specimen LTN12B was cast, cured and tested together with LTN12A, and LTN12D was cast
cured and tested with LTN12C. Therefore, the creep and shrinkage characteristics of the
concrete in LTN12B and LTN12D were the same as those of LTN12A and LTN12C
respectively, both before and after the loading commenced. Both specimens were loaded up to
20kN, and this load level was then held constant for 50 days. Initially, a single crack occurred
in specimen LTN12B under the instantaneous load, and no cracking occurred in LTN12D.
During the period of sustained loads, three additional cracks developed in LTN12B and one
crack occurred in LTN12D primarily due to the tensile forces provided by the reinforcing bar
Figure 4-31 and 4-32 show the load versus measured average axial strain responses of
LTN12B and LTN12D. During the long-term loading period, the graph also indicates the
increment of axial average strain with time and the decrease of tension stiffening strain.
For specimen LTN12B, the average axial strain immediately after application of the 20 kN
load was Hs.avge(0) = 89 x 10-6 and the tension stiffening strain was Hts(0) = 796 x 10-6. The
average stress in the steel is therefore Vs(0)= Es Hs.avge(0) = 17.8 MPa and the average force
carried by the steel bar is Ts(0) = As Vs(0) = 2.01 kN. This means the average tensile force in
the cracked concrete is Tc(0) = P - Ts(0) = 17.99 kN and the average concrete stress is Vc(0) =
Tc(0)/Ac = 1.82 MPa. At this time, only one crack had developed. After 50 days under load,
the average axial strain had increased to Hs.avge(50) = 239 x 10-6 and the tension stiffening
strain had reduced to Hts(50) = 646 x 10-6. The corresponding average steel and concrete forces
115
and stresses are Vs(50)= Es Hs.avge(50) = 47.8 MPa; Ts(50) = As Vs(50) = 5.40 kN; Tc(50) = P -
Ts(50) = 14.59 kN and the average concrete stress is Vc(50) = Tc(50)/Ac = 1.48 MPa. Several
additional cracks occurred with time due to restraint to shrinkage. For this relatively lightly
loaded member, the average tensile stress in the concrete and the tension stiffening strain
reduced to 81.2% of its initial value in the first 50 days under sustained load.
For specimen LTN12D, tension stiffening strain Hts dropped from 647 x 10-6 to 487 x 10-6
during the testing period. Meanwhile, the average tensile stress in the concrete Vc decreased
30
Load sustained for 50 days LTN12B
25
Bare bar
Axial load,P (kN)
20
15
10
5
Average strain (x 10-6)
0
-200 0 200 400 600 800
Load (kN)
30
Load sustained for 50 days
25 LTN12D
Bare bar
20
15
10
5
Average strain (x 10-6)
0
0 200 400 600 800
Figures 4-33 and 4-34 show the variation in the tensile stresses carried by concrete and steel
immediately after the application of the full sustained load and after 50 days under the
sustained load for specimens LTN12B and LTN12D, respectively. The formation of new
primary cracks with time lead to significant decreases in tension stiffening in LTN12B and
3.0 300
Initial concrete stress (time = 0) Initial steel stress (time = 0)
2.5 250
Final concrete stress (t = 50 days) Final steel stress (t = 50 days)
Stress (MPa)
Stress (MPa)
2.0 200
1.5 150
1.0 100
0.5 50
0.0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along specimen (mm) Distance along specimen (mm)
(a) Concrete stress. (b) Steel stress.
Figure 4-33: Initial and final concrete and steel stresses LTN12B (P = 20 kN).
117
3.0 300
Initialconcrete stress (time = 0) Initial steel stress (time = 0)
2.5 250
Final concrete stress (t = 50 days) Final steel stress (t = 50 days)
Stress (MPa)
200
Stress (MPa)
2.0
1.5 150
1.0 100
0.5 50
0.0 0
200 300 400 500 600 700 800 900 200 300 400 500 600 700 800 900
Distance along specimen (mm) Distance along specimen (mm)
The crack pattern and crack width of both long-term tests are shown in Figure 4-35 and 4-36,
both immediately after loading and after 50 days under sustained load. The measured crack
widths at different stages of loading are summarised in Table 4-14 and 4-15.
265 835
1 2 3 4
825 275
Figure 4-37 shows the tension stiffening strain versus concrete age for the long-term tests.
The amount of tension stiffening strains ts at the beginning and the end of the tests are also
labeled in the graphs. The rate of reduction of ts does not appear to be influenced by the
loading levels. For example, with the identical -356 × 10-6 shrinkage occurring, tension
stiffening strain in LTN12C (P = 40kN) decreased by 29% (ts decreased from 291 × 10-6 to
207 × 10-6), while in LTN12D (P = 20kN), this value equals to 25% (ts decreased from 647 ×
119
10-6 to 487 × 10-6). This is a result as the decrease in tension stiffening with time does not
appear to depend on the number of primary cracks, lest rather on the development of
microcracks at the steel-concrete interface. Both primary cracks or microcracks cracks are
The maximum crack widths measured in the experiments of LTN12A and LTN12B are plotted
against concrete age in Figure 4-38. It appears that at the stabilised crack stage (LTN12A),
crack width could be largely increased by both slip between the cracks caused by cover-
controlled cracking and the shrinkage of the concrete between the existing cracks. On the
contrary, the crack width of LTN12B widened with time at slower rate.
Tension stiffening strain (×10 -6
900
Tension stiffening strain (×10 -6)
900
646 800 LTN12C
800 647
700 LTN12D
700 487
796 600
600
500
500 361 LTN12A
400 291
400 LTN12B
300
300
200
200
214 100
100 207
0 0
33 43 53 63 73 83 28 38 48 58 68 78 88
Figure 4-37: Tension stiffening strain versus time for long-term tested prisms.
120
0.8
Maximum crack width(mm) LTN12A
0.7
0.6 LTN12B
0.5
0.4
0.3
0.2
0.1 Concrete age (days)
0
0 20 40 60 80 100
Figure 4-38: Maximum crack width versus concrete age of LTN12A and LTN12B.
4.7 SUMMARY
The experimental program and results of the time-dependent uniaxial tension tests have been
presented in this chapter. The experimental program has considered the significance of
shrinkage and the measurement of both creep and shrinkage was undertaken throughout the
tests. It has been demonstrated by the tests that shrinkage, both before and after loading, has a
For the short-term tests, shrinkage before first loading is restrained by the embedded
reinforcement and a tensile restraining force is imposed on the concrete before any external
load is applied. This restraining force reduces the external load required to cause cracking and
induces an initial negative strain in the member. It is also observed that shrinkage has a minor
effect on tension stiffening at the stablised crack stages. Although the shrunk tension
specimens exhibited a reduced rate of change of tension stiffening decreasing rate and slightly
larger crack width after cracking, compared with the non-shrunk members, the amount of
121
tension stiffening strain does not vary too much. The short-term test results also indicate that
the tension stiffening strain is almost inversely proportional to the reinforcement ratio.
Tension stiffening becomes more significant as the reinforcement ratio in the tension prism
decreases.
For the long-term tests, the results show that tension stiffening dropped off to around 70% of
its initial values in the tested prisms after 50 days under sustained loading. It was found that
the decreasing rate of tension stiffening is not significantly influenced by the loading level.
However the major mechanisms that cause the loss of tension stiffening at different loading
stages could be different. For the specimen loaded at the crack formation stage, additional
primary cracks can significantly bring down the stress in the concrete. For the specimen
loaded at the stabilised crack stage, the loss of tension stiffening is attributed to local bond
breakdown due to microcracking. In both situations, the restraint due to shrinkage is the factor
that causes formation of new primary cracks and formation of cover-controlled cracks that
further break down bond between the steel and concrete. Therefore, consideration of drying
5.1 OVERVIEW
In this chapter, the testing program for the beams and slabs is described and the results are
presented. The construction details of the test specimens and their curing and testing
progresses are first presented. Similar to the previous chapter, the material properties
measured in the experiment, including the instantaneous properties, and the creep and
shrinkage characteristics of the concrete, are next presented. The test results from the
specimens, including the moment versus deflection responses, the variation of forces in the
steel and the measured crack spacings and crack widths are then presented. Finally the results
are discussed. The effects of shrinkage and creep, the effects of primary cracks and cover-
controlled cracks on the change of tension stiffening and the time-dependent change of
span of 2.4m and tested in four point bending. Figure 5-1, 5-2 and 5-3 show the dimensions
and details of the beam and slab specimens. Each beam had a cross section that was 200mm
wide by 400mm deep. All beam specimens were reinforced with a single layer of tensile
123
reinforcement. BSTN3-16 and BSTS3-16 were reinforced with three N16 deformed tensile
bars in the bottom of the section, while BSTN2-16 and BSTS2-16 had two N16 bottom bars.
The clear cover of the tensile bars is 50mm. All the beams were subjected to short-term
All the four slabs (SSTN4-12, SSTS4-12, SLTN4-12A, and SLTN4-12B) were identically
designed with a cross-section that was 800mm wide by 140mm deep. They were singly
reinforced containing four N12 bars at 200mm spacing in the bottom of the specimens as
shown in Figure 5-3. The clear cover of the bottom reinforcing bars is 20mm. SSTN4-12 and
The first letter in the designation of each specimen indicates the specimen type, “B” for beam
and “S” for slab. The followed two letters represent the testing duration; “ST” for short-term
and “LT” for long-term. The fourth letter indicates whether or not the specimen had
125
developed significant drying shrinkage prior to the application of load; “S” if yes and “N” if
no. The next two digits are referred to as the number of tensile reinforcing bars and the bar
diameter. For instance, “2-16” means two 16mm bars reinforced on the tension side of the
specimen. For the two long-term slabs test specimens, the final letter A and B represent the
Local strains were also measured using strain gauges attached to the longitudinal tensile
reinforcing bars that are closest to the side surface of the specimens, thereby capturing the
influence of cracking on the local strain distribution in the tensile reinforcement. The strain
slabs. Figure 5-4 illustrates the set up for the short-term tests. Each specimen was simply
supported by a roller at one end and a pin at the other end. The loads were applied at the third
span points and monitored by a load cell placed on the bottom of the actuator. All loading
plates and support plates were 100 mm wide and extended across the full width of the
specimen.
A 0.3mm/min prescribed displacement was imposed by the Instron machine from the top and
transferred through a spreader beam to the two loading points on the top of the specimen. The
vertical displacements of the specimen at mid-span and at the two third-span points were
monitored with high-resolution laser transducers placed underneath the specimen (Figure 5-6).
All the measurements, including local steel strains, vertical displacement and the applied
loads, were recorded electronically by the HBM amplifier and recorded through the data
126
logger.
a) Beam set up
b) Slab set up
Figure 5-4: Set up for short-term flexural tests.
127
The set up for the two long-term slab tests are shown in Figure 5-5. The specimen was placed
on top of two firm steel supports (one pin and one roller). At the two third-span points of the
test specimen two cross-beams were located. Pre-cast concrete blocks were carefully
128
positioned on the top of the cross-beams to produce two constant sustained point loads at the
third span points. Dial gauges were placed underneath the slabs to monitor the change in
deflection at mid span and at the third points, as shown in Figure 5-7. The strain gauges were
also connected to the HBM amplifier and the readings of steel strains are recorded throughout
the long-term testing period. Crack widths were also measured with the microscope on the
Figure 5-6: High-resolution laser transducers underneath the short-term testing slabs.
129
cured in a water tank to keep them wet for 14 days, so that little if any drying shrinkage
occurred during the curing period. Then on the 14th day all the beams were immediately
moved out of the tank to allow drying to start. BSTN2-16 and BSTN3-16 were placed on the
existing set up (Figure 5-4a) and tested within 24 hours. In the contrast, BSTS2-12 and
BSTS3-12 however were also simply supported, but remained unloaded for a period of 46
days while drying shrinkage developed. After this initial drying, the specimens were tested in
the loading frame. During the drying period, the change of mid-span deflection of these two
beams was monitored using dial gauges, in order to correctly characterize the influence of
The four slabs were cast together from one batch of concrete, and cured in the water tank in
order to minimize initial shrinkage. Then on the 14th day, SSTN4-12 and SSTS4-12 were
moved out of the water tank. SSTN4-12 was immediately tested (Figure 5-4b), but SSTS4-12
was allowed to dry and unloaded for 45 days. The mid-span deflection of SSTS4-12
throughout the drying period was monitored with a dial gauge. On day 15, SLTN4-12A and
SLTN4-12B were taken out of the tank and the long-term tests commenced immediately
(Figure 5-5). Each of the long-term test specimens as loaded with concrete blocks on the top,
and the loads were maintained unchanged for 70 days. Specimens SLTN4-12A was subjected
131
to a total superimposed sustained load of 40kN, while specimen SLTN4-12B of 30kN. Table
5-1 shows the curing and testing procedure for all the flexural members in the experimental
program.
Because the creep strain starts to develop when the concrete is first loaded, whether by
external applied load or by internal restraint to shrinkage, the specimens were kept moist until
the time of loading so that creep due to load and due to restrained shrinkage started at the
same time. In this way the creep coefficient measured on the companion specimen would be
applicable to both the load and the shrinkage induced stress. In practice, creep and shrinkage
strains are likely to develop before any service loads are applied, and this may significantly
creep and shrinkage. The methods to measure creep strain and shrinkage strain were identical
to those described in Chapter 4 for the uniaxial tension specimens (Figure 4-5). The test
cylinders and prisms are given in Table 5-2. The concrete strength of the beams increased by
approximately 50% during the drying period. For the slabs, the concrete strength increased by
about 15% throughout the sustained load test. The elastic modulus of the concrete also
increased by 28% for the beams and by 16% for the slabs during the drying period. The
corresponding increase in the tensile strength, both indirect tensile strength and flexural
tensile strength, was about 10% for both beams and slabs.
Specimen f cp f ct f cf Ec
BSTN2-16
26.0 2.2 3.5 25000
BSTN3-16
BSTS2-16
40.0 2.4 3.9 32000
BSTS3-16
SSTN4-12 28.6 1.8 3.0 25000
SSTS4-12 32.6 2.0 3.3 29000
(a) Short-term flexural specimens
f cp f ct f cf Ec
Specimen
Start End Start End Start End Start End
SLTN4-12A
28.6 33.7 1.8 2.2 3.0 3.3 25000 29500
SLTN4-12B
For the concrete in the beams and slabs, creep was measured on two identical specimens. The
concrete was first loaded at age 14 days as soon as the sample started drying. The calculated
creep coefficients of the plain concrete from the creep tests for both beams and slabs are
plotted in Figure 5-9. Also plotted in the diagrams are the averaged values from the two creep
tests. According to the diagrams, the averaged c at the ends of the drying days for the short-
term beams and slabs tests were 0.92 and 1.37 respectively. By the end of long-term tests of
c 1.2
c 1.6
1.0
1.2
0.8
Creep test 1
0.6 0.8 Creep test 2
Creep test 1
0.4 Averaged
Creep test 2
0.4
0.2 Averaged
0.0 0.0
14 34 54 74 0 20 40 60 80
Concrete age (days) Concrete age (days)
Figure 5-10 shows the measured drying shrinkage strains and their averaged values for the
beams and slabs. The averaged shrinkage strain for specimen BSTS2-16 and BSTS3-16
during the drying period (day 60) was -220 × 10-6. In the time interval between the two short-
term slab tests (from day 15 to day 59), the averaged drying shrinkage strain was -320 × 10-6
and at the end of the long-term tests (day 85), the averaged drying shrinkage strain was -400 ×
10-6.
134
300 500
Shrinkage strain (×10-6
200
100 Shrinkage test 1
Shrinkage test 2 100
Averaged
0
0
0 20 40 60 80 100
0 20 40 60 80 Concrete age (days)
Concrete age (days)
(b) Slabs
(a) Beams
Figure 5-10: shrinkage strain development for the flexural tests.
BSTN2-16, BSTN3-16 and SSTN4-12 were tested soon after the end of moist curing before
significant shrinkage could occur. They were cured in a water tank for 14 days and then tested.
Minimum drying shrinkage and shrinkage-induced initial deflection sh0 occurred before the
tests.
Figure 5-11, 5-12 and 5-13 display the global moment versus mid-span deflection responses
from the tests, together with the calculated fully-cracked response. The fully-cracked response
is obtained by assuming that all the sections along the member are fully-cracked and the
tensile concrete carries no tension. The mid-span deflection of the fully-cracked beam is
obtained by integrating the curvature (M/EcIcr) over the span of each member (2.4m). Thus,
135
the instantaneous tension stiffening deflection tsi at a particular mid-span moment is the
difference between the fully-cracked deflection and actual member deflection, as seen in the
diagrams.
As can be seen, the member behaviour is approximately linear-elastic before first cracking
(curve OB) and behave in a nonlinear manner after cracking (curve BC). The maximum
tension stiffening deflection tsi.max occurs at the onset of first cracking (i.e. at M = Mcr). Then,
as the load is increased, tsi gradually diminishes and the moment-deflection curve approaches
Table 5-3 summarises the experimentally measured mid-span deflections exp, the calculated
fully-cracked deflections fc and the corresponding tension stiffening deflections tsi after first
cracking for each specimen. The magnitude of tension stiffening deflections decreases as
applied moment increases, and this indicates the loss of tension stiffening. As the applied
moments increased from Mcr to 2.0Mcr, tsi decreases by 60%, 56% and 75% for BSTN2-16,
BSTN3-16 and SSTN4-12 respectively. As the moment approaches the magnitudes where the
Moment (kN.m)
60
C
50
40
M cr = 22.1kN.m tsi
30
B
20 BSTN2-16
tsi.max
Fully-cracked
10
O Mid-span deflection (mm)
0
0 1 2 3 4 5 6
70
C
60
50
40 M cr = 26.6kN.m
tsi
B
30
BSTN3-16
20 tsi.max
Fully-cracked
10
O Mid-span deflection (mm)
0
0 1 2 3 4 5
25
Moment (kN.m)
C
20
15
M cr =9.1kN.m
tsi tsi
B SSTN4-12
10
Bare bar
Fully-cracked
tsi.max
tsi.max
5
Mid-span deflection (mm)
0 O
0 3 6 9 12 15 18
Figure 5-14, 5-15, and 5-16 show the variations of steel forces in the middle 800 mm of each
specimen (i.e. the constant moment region) at various stages of loading. The steel force was
obtained directly from the measured local steel strains. The variation of steel force depends on
the locations of the cracks, with the tensile force in the steel at a peak at each crack. Away
from the cracks, tensile stress in the concrete is built up through bond, and the tensile stress in
the steel decreases. Since the strain gauges on the reinforcement in the flexural tests were
located at 40mm centres, there were only 3 – 6 measurements of local steel strains between
adjacent cracks. Therefore, it is difficult to accurately access the actual distance S away from
the primary cracks, where bond is affected. Future research can try to improve this by using
more strain gauges at closer centres without removing the ribs or reducing the local bond
mechanism.
Unlike the uniaxial tension members, the tensile stress in the concrete at typical sections
along the flexural members can not be directly obtained from the measured steel strains,
because the area of concrete under tension, as well as the tensile stress gradient, is also
uncertain. Therefore, another approach to represent the quantity of local tension stiffening is
to calculate the average tensile force carried by the concrete between the cracks.
Table 5-4 tabulates the average tensile bar force Ts.avge in BSTN2-16, BSTN3-16 and SSTN4-
12 at certain loading stages after first cracking. Ts.avge was determined by averaging the
measured tensile force in the steel over the constant moment region, and the average tensile
force carried by the concrete Tc.avge was determined by subtracting the average force in the
steel from the maximum force in the steel at each primary crack (where concrete is assumed
to carry no tensile force). The results show that after cracking, Tc.avge decreases with
139
increasing moment, which is indicated by the reduction of the proportion Tc.avge/Ts.max. For
example, as the applied moment increases from 16.0kN.m to 20.0kN.m in SSTN4-12, the
value of Tc.avge/Ts.max decreases from 0.15 to 0.12. Notice that all the cracks had formed after
M = 16.0kN.m, thus the average tensile stress in the concrete drops off at the stabilised crack
Table 5-4: Average tensile forces in the concrete and the steel over the constant moment
region (BSTN2-16, BSTN3-16 and SSTN4-12).
20 30
25
15
Force (kN)
Force (kN)
20
10 15
10
5
5
0 0
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
30 40
Force (kN)
Force (kN)
30
20
20
10
10
0 0
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(c) Just after 3rd crack, M =.26kN.m (d) After 6th crack, M = 28.8 kN.m.
70 90
60 80
Force (kN)
Force (kN)
50 70
40 60
30 50
20 40
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
20 30
25
15
Force (kN)
Force (kN)
20
10 15
10
5
5
0 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(a) Just before first cracking, M = 24.6 kN.m. (b) Just after first cracking, M = 26.6 kN.
30 50
25 40
Force (kN)
Force (kN)
20
30
15
20
10
5 10
0 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(c) After 3rd and 4th cracks, M = 28.2 kN.m. (d) At M = 40 kN.m.
60 70
50 60
Force (kN)
Force (kN)
40 50
30 40
20 30
10 20
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
20 30
25
15
Force (kN)
Force (kN)
20
10 15
10
5
5
0 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(a) Just before first cracking, M = 8.5 kN.m. (b) Just after first cracking, M = 9.1 kN.m.
30 40
25
30
Force (kN)
Force (kN)
20
15 20
10
10
5
0 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
50 60
40 50
Force (kN)
Force (kN)
30 40
20 30
10 20
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
Figure 5-16: Variation of forces in steel at different stages of loading (SSTN 4-12).
143
The crack patterns of BSTN2-16, BSTN3-16 and SSTN4-12 at the end of the tests are shown
in Figure 5-17, 5-18 and 5-19. Within the constant moment regions, the numbers of cracks in
each specimen are 5, 6 and 7. The crack spacings shown in the figures are measured at the
levels of the tensile reinforcing bars. The maximum crack spacings of BSTN2-16, BSTN3-16
and SSTN4-12 in the constant moment region are 243mm, 153mm and 157mm respectively.
The average crack spacings are 154mm, 121mm and 122mm respectively. Similar to the
observation in uniaxial tension tests, the crack spacing in the members with the smaller
reinforcement ratio is greater than for the members with larger , i.e. more cracks had
formed in the latter. This indicates that in the flexural tests, the length S away from the
primary cracks where significant bond stresses develop is also dependent on the
reinforcement ratio.
The cracked widths at the different loading levels are tabulated in Table 5-5, 5-6 and 5-7.
Only the crack widths within the constant moment regions are shown, together with the
calculated steel stress at the cracked sections (Vs). Generally the crack width increases as the
steel stress Vs increases. Compared with the uniaxial tension members, the difference between
the maximum crack width and the average crack width is not that significant in the flexural
members. The ratio of maximum crack width over average crack width after all the cracks had
formed is about 1.25. In addition, the study of the crack width on the beams shows that at the
same steel stress level, maximum crack width seems independent of the reinforcement ratio.
145
Average Maximum
Vs Width of Width of Width of Width of Width of Width of Width of
Load crack crack
1st crack 2nd crack 3rd crack 4th crack 5th crack 6th crack 12th crack
Stage (MPa) width width
(mm) (mm) (mm) (mm) (mm) (mm) (mm)
(mm) (mm)
12kN.m 253 0.05 0.075 0.1 0.1 0.075 0.075 - 0.08 0.075
16kN.m 338 0.175 0.175 0.15 0.15 0.125 0.1 0.075 0.136 0.175
20kN.m 422 0.225 0.25 0.2 0.225 0.225 0.15 0.125 0.20 0.25
5.6.1.2 Members with significant initial shrinkage (BSTS2-16, BSTS3-16 and SSTS4-12)
Specimens BSTS2-16, BSTS3-16 and SSTS4-12 were allowed to dry for a period of time (45-
46 days) before they were tested. At the time of testing days, significant shrinkage strains (-
220 × 10-6 for beams and -320 × 10-6 for slabs) had developed prior to loading. The
corresponding creep coefficients were 0.92 and 1.37 respectively. The initial deflection Gsh.0
caused by shrinkage warping was also measured by the dial gauges located underneath the
146
mid span of the specimens and the results were Gsh.0 = 0.17mm, 0.14mm and 0.65mm for
BSTS2-16, BSTS3-16 and SSTS4-12, respectively. Therefore, at zero moment (M = 0), the
The moment versus mid-span deflection measured throughout the tests are plotted in Figure
5-20, 5-21 and 5-22. Also plotted are the corresponding fully-cracked responses. It has been
mentioned in Section 3.3.3 that the fully-cracked response for short-term flexural members
with initial shrinkage must be shifted to the right by an amount of Gsho.cr, in order to consider
the effect of initial shrinkage on the deflection of a fully-cracked member. The procedure to
calculate Gsho.cr can be found in Appendix B. The values of Gsho.cr are also shown in the figures.
It is also clear in these diagrams that the structural behaviour moves from linear to nonlinear
at the onset of cracking. At the linear-elastic stage (OB), the specimens are relatively stiff and
little deflection has occurred compared with the nonlinear stage (BC). Early shrinkage before
loading can greatly affect the member response. Compared with the specimens without early
shrinkage, the cracking moment Mcr.sh is significantly reduced. For example, the cracking
moment of BSTS2-16 (17.0kN.m) is much smaller than BSTN2-16 (22.1kN.m). The tension
stiffening deflection tsi also gradually diminish after cracking and the experimental responses
Table 5-8 provide values of measured mid-span deflections exp, calculated fully-cracked
deflections fc and instantaneous tension stiffening deflections tsi at selected values of total
applied moment M. Similar to the members without shrinkage, the values of tsi decrease with
increasing applied moments after first cracking. As the applied moment increases from Mcr to
147
2 Mcr, tsi drops by 55%, 36% and 45% for BSTS2-16, BSTS3-16 and SSTS4-12 respectively.
Moment (kN.m)
60
BSTS2-16
50
Fully-cracked C
40
30
M cr.sh =17.0kN.m tsi
B
20
tsi.max
10
O
sh0.cr = 0.5mm Mid-span deflection (mm)
0
0 1 2 3 4 5 6
sh.cr = 0.17mm
70
Moment (kN.m)
C
60
50
40
tsi
M cr.sh =21.4kN.m
30
B BSTS3-16
20 tsi.max
Fully-cracked
10
O sh0.cr = 0.5mm Mid-span deflection (mm)
0
0 1 2 3 4 5 6
0.14
sh.cr = 0.17mm
25
Moment (kN.m)
SSTS4-12
20 Bare bar
Fully-cracked C
15
M cr.sh = 6.0kN.m tsi
10
B
5 tsi.max
Mid-span deflection (mm)
O sh0.cr = 2mm
0
0 3 6 9 12 15 18
sh0 = 0.65mm
The variations of steel forces in the middle 800 mm of the beams at various stages of loading
are shown in Figure 5-23, 5-24 and 5-25. Before loading, since the reinforcing bars were
subjected to compressive force due to the shrinkage in the concrete, the steel strains were
negative. Therefore, the shrinkage-induced compressive strains must be subtracted from the
measured steel strains to obtain the actual strains at a typical moment. The procedure to
calculate the time-dependent initial compressive steel strains of the uncracked beam or slab
10 30
5 20
Force (kN)
Force (kN)
0 10
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
-5 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
-10 -10
Distance from the mid span (m) Distance from the mid span (m)
(a) Just before first cracking, M = 16.7kN.m. (b) Just after first cracking, M = 17.0 kN.m.
30 50
40
20 Force (kN)
Force (kN)
30
10 20
10
0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0
-10 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(c) Just after 2nd crack, M = 18.9 kN.m. (d) At M = 28.8 kN.m.
70 90
60 80
Force (kN)
Force (kN)
50 70
40 60
30 50
20 40
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
20 30
20
Force (kN)
10
Force (kN)
10
0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
-10 -10
Distance from the mid span (m) Distance from the mid span (m)
(a) Just before first cracking, M = 19.4kN.m. (b) Just after 3rd cracking, M = 21.4 kN.m.
30 50
40
20
Force (kN)
Force (kN)
30
10 20
0 10
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0
-10 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
(c) Just after 4thcrack, M = 28.4 kN.m. (d) At M = 40 kN.m.
70 70
60 60
Force (kN)
Force (kN)
50 50
40 40
30 30
20 20
10 10
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
20 20
15 15
Force (kN)
Force (kN)
10 10
5 5
0 0
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
-5 -5
Distance from the mid span (m) Distance from the mid span (m)
(a) Just before first cracking, M = 5.9kN.m. (b) Just after first cracking, M = 6.0 kN.m.
40
25
30
Force (kN)
Force (kN)
15 20
5 10
0
-5-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m)
Distance from the mid span (m)
(c) Just after 3rd crack, M = 8.0 kN.m. (d) At M = 12 kN.m.
50 60
40 50
Force (kN)
Force (kN)
30 40
20 30
10 20
-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4
Distance from the mid span (m) Distance from the mid span (m)
Table 5-9 tabulates the calculated average tensile forces in the concrete and the steel over the
constant region for BSTS2-16, BSTS3-16 an SSTS4-12. The comparison with Table 5-4
shows that the average tensile force in the concrete Tc.avg is larger in the members that had
16.64kN, while it is only 7.9kN in BSTN2-16 at the same moment level. The increase of the
average tensile force in the concrete in the member with early shrinkage is attributed to tensile
Table 5-9: Average tensile forces in the concrete and the steel over the constant moment
region (BSTS2-16, BSTS3-16 and SSTS4-12).
The crack locations and crack numbers for BSTS2-16, BSTS3-16 and SSTS4-12 are shown in
Figure 5-26, 5-27 and 5-28, and the crack widths at various stages of loading are given in
Table 5-10, 5-11 and 5-12, together with the steel stress (Vs) at the crack at each loading stage.
The members with early shrinkage exhibited similar numbers of cracks to the members
without shrinkage, except for BSTS2-16 which had two more cracks than BSTN2-16. The
maximum crack spacings in the constant moment regions for BSTS2-16, BSTS3-16 and
SSTS4-12 are 122mm, 162mm and 168mm respectively, and the average crack spacings are
110mm, 133mm and 139mm respectively. The difference of crack spacings between the
members with early shrinkage and without shrinkage can be hardly told.
154
Average Maximum
Vs Width of Width of Width of Width of Width of Width of
Load crack crack
1st crack 2nd crack 3rd crack 5th crack 7th crack 8th crack
Stage (MPa) width width
(mm) (mm) (mm) (mm) (mm) (mm)
(mm) (mm)
12kN.m 253 0.125 0.075 0.125 0.125 0.075 0.05 0.096 0.125
16kN.m 338 0.175 0.1 0.175 0.15 0.125 0.075 0.133 0.175
20kN.m 442 0.275 0.175 0.275 0.25 0.25 0.15 0.229 0.275
The tension stiffening deflections given in Table 5-3 and 5-9 are plotted against the calculated
maximum steel stresses at the cracks in Figure 5-29 and 5-30. Figure 5-29 demonstrates the
influence of initial shrinkage on the short-term tension stiffening deflections. The dots at the
left hand end of each curve in the figure signify the situation at the onset of first cracking,
where the tension stiffening deflections are at maximum. It appears that with the same steel
156
stresses at cracks, the tension stiffening deflections are not significantly affected by the
amount of early shrinkage. For the beam specimens, tension stiffening deflections appear to
be slightly greater in the beams without shrinkage (Figure 5-29 (a) and (b)). This is typically
obvious in the specimens BSTS2-16 and BSTN2-6. At the same stress level after first
cracking, the calculated tension stiffening deflection in BSTN2-16 is about 0.2mm greater
than that in BSTS2-16 at the same steel stress level. For BSTN3-16 and BSTS3-16, tension
stiffening deflection in the former specimen is only about 0.1mm greater than that in the latter.
In the slabs, the difference of tension stiffening deflection varies at different steel stress levels.
As steel stress is less than 220MPa, SSTN4-12 exhibits greater tension stiffening.
Nevertheless, the tension stiffening deflection is higher in SSTS4-12 has the steel stress
exceeds 220MPa.
The initial shrinkage increases the tensile stress in the concrete before cracking, and thus
increases tension stiffening. After cracking, some tensile stress in the concrete between the
cracks caused by shrinkage still exists. After first crack develops, the tensile stress in the
concrete away from the crack in the member with shrinkage will be greater than for a similar
member without shrinkage, if good bond conditions exist. However, if this tensile stress is
greater than the tensile strength of concrete at the weakest section, a second crack occurs and
tension stiffening is gradually broken down. On the contrary, if the tensile stress produced by
restraint to shrinkage is not great enough or bond condition is not good, additional cracks will
2 2
Tension stiffening
Tension stiffening
deflection (mm)
deflection (mm)
1.6 1.6
1.2 1.2
0.8 0.8
BSTN2-16 BSTN3-16
0.4 0.4
BSTS2-16 BSTS3-16
0 0
0 100 200 300 400 0 100 200 300 400
Tensile steel stress at crack (MPa) Tensile steel stress at crack (MPa)
6
Tension stiffening
deflection (mm)
SSTN4-12
4
SSTS4-12
0
0 100 200 300 400 500
Tensile steel stress at crack (MPa)
2
Tension stiffening
2
Tension stiffening
deflection (mm)
deflection (mm)
1.6
1.6
1.2
1.2
0.8
0.8
BSTS2-16
BSTN2-16 0.4
0.4 BSTS3-16
BSTN3-16
0
0
0 100 200 300 400
0 50 100 150 200 250 300 350
Tensile steel stress at crack (MPa)
Tensile steel stress at crack (MPa)
Figure 5-30 illustrates the effects of reinforcing ratio on short-term tension stiffening
deflections after cracking. It can be seen that the tension stiffening deflection is greater in the
specimens with the smaller reinforcing ratio. The tension stiffening deflections of BSTN2-16
and BSTS2-16 (with reinforcement ratio U = As/Ac= 0.005) is approximately 20% higher than
that of BSTN3-16 and BSTS3-16 (U = 0.0075) at the same steel stress level. This is similar to
The maximum crack widths measured in the experiments and given in Tables 5-5 to 5-7, and
Tables 5-10 to 5-12 are plotted against steel stress in Figure 5-31 and a best fit curve for each
specimen is also plotted. Unlike the uniaxial tension members, the effect of initial shrinkage
on maximum crack width is not very significant. At a particular steel stress at the crack
section, the crack widths of the shrunk and non-shrunk specimens are of little difference, as
seen in Figure 5-31. The effect of reinforcement ratio on crack widths is also not significant
as shown in Figures 5-32. For the members without initial shrinkage, BSTN3-16 has slightly
greater maximum crack width than BSTN2-16 at all steel stress levels. For the members with
significant initial shrinkage, BSTS2-16 has greater crack width than BSTS3-16 when the steel
stress is less than 200MPa. However, at higher steel stress levels, BSTS2-16 exhibits smaller
500
Steel stress at crack (MPa)
500
400
300
200
100 SSTN4-12
SSTS4-12
0
0 0.1 0.2 0.3
Maximum crack width (mm)
(c) SSTN4-12 and SSTS4-12
Figure 5-31: The effects of initial shrinkage on the maximum crack width of flexural
members.
500
Steel stress at crack (MPa)
500
400 400
300 300
200 200
100 BSTN2-16 100 BSTS2-16
BSTN3-16 BSTS3-16
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Maximum crack width (mm) Maximum crack width (mm)
SLTN4-12A and SLTN4-12B were tested after being cured in the water tank for 15 days. The
shrinkage strains in the specimens before the tests were very small and were ignored in the
analysis. Pre-cast concrete blocks were placed on the top of the specimen to provide sustained
dead loads. The dead loads were transferred as point loads to the specimens via two spandrel
beams at the third span points of the slabs to create constant moment regions (see Figure 5-5
(b)). SLTN4-12A and SLTN4-12B were subjected to two equal point loads of 20.0kN and
15kN, respectively, at the third span points, and the sustained moments in the constant
moment region of each slab were 17.9kNm and 13.9kNm, respectively. The shrinkage strain
and creep coefficient at the end of the long-term loading were -400×10-6 and 1.47.
Figure 5-33 and 5-34 show the time-dependent change of mid-span deflections of SLTN4-12A
and SLTN4-12B under sustained service loads. Both slabs did not begin to shrink until the
day of first loading. If the tensile concrete is entirely ignored (fully-cracked), the
instantaneous curvature on each cross-section can be determined using modular ratio theory
and the time-dependent curvature on each cross-section due to creep and shrinkage can be
calculated using the age-adjusted effective modulus method (Appendix B). By integrating the
curvature diagram at any time instant, the mid-span deflection of the fully cracked member
(here called the fully-cracked deflection fc(t) ) can be estimated. The tension stiffening
deflection ts(t) at any time after first loading is obtained by subtracting the measured
deflection of the member at mid-span from the fully-cracked deflection at particular moments.
20
Deflection (mm
15
10
5 SLTN4-12A
first 15days
0
0 10 20 30 40 50 60 70
Time (days)
15
Deflection (mm
10
5
SLTN4-12B
first 15days
0
0 10 20 30 40 50 60 70
Time (days)
Table 5-13: The measured and calculated mid-span deflection for SLTN4-12A and SLTN4-
12B.
SLTN4-12B 13.9 7.89 12.3 4.33 12.69 15.8 3.11 14.6 17.5 2.9
At first loading, the estimated tension stiffening deflections for SLTN4-12A and SLTN4-12B
were 46% and 55% of the measured total deflections. After 15 days under sustained loading,
the tension stiffening deflection ts(t) decreased by 23% for SLTN4-12A and 28% for SLTN4-
12B. From time 15 days to the end of the test, the tension stiffening deflection decreased by
only a small amount, and at t = 70 days, ts(t) had decrease to 75% and 67% of its
The measured forces in a steel bar derived from the strain gauges for both slabs are plotted in
Figure 5-35 and the average steel and concrete tensile forces are tabulated in Table 5-14. The
steel stress away from the crack appears to increase relatively quickly after first loading (in
the first 15 days under load) and the concrete tensile force suffers a corresponding reduction,
but in the period t = 15 days to t = 70 days the changes are relatively small. In addition, the
decreasing rate of the average tensile force in the concrete seems independent of the loading
levels.
163
50 50
t = 0 days
Tensile bar force (kN)
t = 15 days
40
30 30
t = 0 days
t = 15 days
20 20
t = 70 days
10 10
1 8 3 10 5 2 9
3 8 2 1 6 9
Crack number
Crack number
0 0
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
Distance from mid-span (m) Distance from mid-span (m)
Table 5-14: Comparison between fully cracked section and measured steel forces
Figure 5-36 and 5-37 show the crack patterns for SLTN4-12A and SLTN4-12B at the
beginning and the end of the tests. For both SLTN4-12A and SLTN4-12B, 3 new primary
cracks were observed to develop during the first 15 days under load, and the formation of
these new cracks significantly contributed to the drop off in tension stiffening. For the last 55
days under load, no new primary cracks were observed in either slab, and the very small
reduction in tension stiffening in this period may be caused by a gradual deterioration of bond
164
a) At first loading
b) After 70 days
Table 5-15 and 5-16 gives the measured crack widths during the tests. The maximum crack
width had increased by about 70% - 80% during the sustained loading period.
5.7 SUMMARY
Four beams and four identical slabs were tested to investigate the time-dependent tension
stiffening in short-term and long-term bending. The experimental program was specified in
this Chapter. The tests have demonstrated the significance of drying shrinkage on the
deflection of beams and slabs and have identified the gradual loss of tension stiffening in
Similar to the uniaxial tension tests in Chapter 4, it can be concluded that for both short-term
and long-term tests, the loss of tension stiffening shall be directly attributed to the formation
of new cracks and bond deterioration, which can be caused by applied bending moment or
restraint to shrinkage. Short-term test results show that the shrinkage prior to loading can
166
reduce the cracking moment and cause initial deflection. However, early shrinkage appears to
have relatively little effect on tension stiffening and crack width after first cracking.
The long-term tests indicates that tension stiffening can be greatly reduced due to the
formation of new cracks in the first 15 days and then bond deterioration further breaks down
tension stiffening thereafter but at a much slower rate. Since the formation of new cracks is
attributed at least in part to the restraint to shrinkage by the tensile reinforcing bars, the decay
of tension stiffening with time is very much associated with the amount of shrinkage.
CHAPTER 6 FINITE ELEMENT
FORMULATION AND MATERIAL
CONSTITUTIVE LAW
6.1 OVERVIEW
Numerical methods are widely adopted to model concrete structures. Nonlinear finite element
methods have been used for many years to analyse reinforced concrete structures and have
proven to provide good accuracy. Various methods have been developed to model the
interaction between concrete and steel reinforcement and the associated tension stiffening (see
Chapter 2). The difficulties of modeling reinforced concrete structures can be attributed to the
modeling can generally predict structural behaviour and model the complexities in a manner
and at a level that is not available in other approaches. The complicating factors when using
finite element method to model reinforced concrete structure can be categorized as below:
x The nonlinear stress-strain behaviour of concrete under multiple stress state (including
x The development of failure criteria for concrete under multiple stress states.
168
x The interaction between concrete and steel as they work together, namely the bond
In this study, the finite element program RECAP (Foster 1992, Foster and Gilbert 1990,
Foster and Marti 2003) has been adopted and further developed by the author. RECAP
includes a number of 2-D and 3-D concrete models, together with steel and bond elements.
The major contribution by the author is the development of a 2-D bond interface element that
is demonstrated to provide a good model of bond throughout the serviceability loading range.
The bond model is time-dependent and has been calibrated to accurately model the decay of
tension stiffening with time. The corresponding concrete, steel and bond element formulations
The four-node isoparametric element is often used to model concrete. A typical isoparametric
In the element, two coordinate systems are established, namely the global coordinate system
xOy and the local coordinate system O. The values of and on the edge of the elements
are [ r1 andK r1 . The global coordinate of any point within the element can be related
to the coordinate of each node by the shape functions. The relationship are given by
4
x ¦N x
i 1
i i 6.1a
4
y ¦N y
i 1
i i 6.1b
1
N1 1 [ 1 K 6.2a
4
1
N2 1 [ 1 K 6.2b
4
170
1
N3 1 [ 1K 6.2c
4
1
N4 1 [ 1K 6.2d
4
A feature of isoparametric elements is that the shape functions that are used to derive element
coordinates are also the ones that are used to obtain element displacements. The relationship
between interior element displacements and displacements at the nodes can be described as
4
u ¦N u
i 1
i i 6.3a
4
v ¦N v
i 1
i i 6.3b
Given the displacements, the strains of any point in the element can be derived as the first
wu ½
° °
H x ½ ° wx °
° ° ° wv °
^H ` ®H y ¾ ® ¾
°J ° ° wy °
¯ xy ¿ ° wu wv °
° wy wx °
¯ ¿
u1 ½
° °
wN 1 wN 2 wN 3 wN 4 ½° v1 ° 6.4
° 0 0 0 0 °°u 2 °
° wx wx wx wx °° °
° wN 1 wN 2 wN 3 wN 4 °°v 2 °
® 0 0 0 0 ¾® ¾
° wy wy wy wy °°u 3 °
° wN 1 wN 1 wN 2 wN 1 wN 3 wN 3 wN 4 wN 4 °° v °
° wy wx °¿°u °
3
¯ wx wy wx wy wx wy
° 4°
°v °
¯ 4¿
>B @>G @
where [B] is called the strain-displacement matrix, in which the shape functions are
differentiated with respect to global coordinates x and y. Since the shape functions are
represented by local coordinates and , transformation needs to be made between the two
ª wx wy º ª 4 wN i 4
wN i º
« w[ w[ » «¦ w[ xi ¦ w[ yi »
>J @ « » « i 41 i 1
» 6.5
« wx wy » « wN i x wN i »
4
«¬¦ ¦ yi
«¬ wK wK »¼ i 1 wK
i
i 1 wK
»¼
The derivatives of the shape functions about the x and y coordinates can be written as
ª wN i º wN i ½
« wx » ° °
« wN » >J @1 °® wwN[ °¾ 6.6
« i» ° i°
«¬ wy i »¼ °¯ wK °¿
and inputted into Equation 6.4 to obtain the strain-displacement matrix [B].
172
In order to construct the global element stiffness matrix ([Dc]), the material constitutive law
matrix obtained from the material coordinate system ([Dc12]) needs to be transformed by using
ª c2 s2 cs º
« 2 »
>TH @ « s c 2
cs » 6.7
« 2cs 2cs c 2 s 2 »
¬ ¼
in which c cos T and s sin T . is the angle between the global coordinate system and the
material principal strain direction. The material constitutive matrix [Dc12] in the principal
Finally applying the principal of virtual work gives the element stiffness matrix as:
The biaxial concrete strength envelope proposed by Foster and Marti (Foster and Marti 2003),
and shown in Figure 6-2, is used. Other biaxial stress-strain relationships such as that
proposed by Kupfer et al (1969) could also have been used. In Figure 6-2, 1c and 2c are the
principal stresses, and fcp is the uniaxial compressive strength of the in-situ concrete.
173
For concrete under tension, the bilinear softening curve proposed by Petersson (1981) is used
(Figure 6-3a). Cracking of concrete is introduced in the element as soon as the principal
tensile stress exceeds the tensile strength of the concrete. In order to deal with the mesh
element, the “crack band theory” approach (Bazant and Oh 1983) is used to depict the
softening branch of the tensile stress-strain relationship in the fracture zone immediately
adjacent to the crack. According to Petersson (1981), the three parameters required to define
1 2 18 E c G f
D1 ; D2 D 3 D1 ; D 3 6.10
3 9 5 f ct2 lt
where Ec is the elastic modulus of concrete; lt is the element size; fct is the tensile strength of
concrete and Gf is the fracture energy of concrete under softening which in this thesis is given
0.7
§f ·
Gf D f ¨¨ cp ¸¸ , ( f cp :MPa and G f :N/mm) 6.11
© 10 ¹
and f is a coefficient related to the aggregate size and may be taken as:
In this study, the concrete compressive strength in one principal direction is given by:
f c* E f cp 6.13
175
where is a scaling factor for the compressive curve in one principal direction and it depends
on the stress in the other principal direction, as shown in Figure 6-3b. The concrete strain at
H cp* EH cp 6.14
In the compression-compression state, is greater than 1.0 and depends on the ratio of major
1.0
E2 for 0 d D ' d 0.48
1D ' 6.15a
2.4
1.15 1 5.5 1
E2 for 0.48 D ' d 1.0 6.15b
1 D ' / 5.5
where 2 is the confining scaling factor for minor principal direction. The confining scaling
E1 D ' E 2 6.16
In the tension-compression state, the major principal stress is tensile and it reduces the
compressive strength in the minor principal direction (Berlarbi and Hsu 1991; Miyakawa and
Kawakami 1987; Vecchio and Collins 1982). In this study, the equation for the scaling factor
1
E d 1.0
H 6.17
0.8 0.34 1
H cp
nK
Vc f cp 6.18
n 1 K nk
in which:
Ec Hc
n ;K 6.19
E c E cp H cp
and cp is the strain corresponding to the peak stress and Ecp is the secant modulus at the peak
f cp
of the curve E cp . The parameter k in Equation 6.18 is the decay factor for post peak
H cp
k 1 .0 for H c d H cp 6.20a
f cp
k 0.67 t 1.0 for H c ! H cp 6.20b
62
For finite element implementation, Foster and Marti (Foster and Marti 2003) used the
equivalent uniaxial strains concept from Darwin and Pecknold (1977) to subtract the
Poisson’s effect in the material constitutive matrix and allow the usage of uniaxial stress-
H 1P ½ 1 ª 1 Q 12 º H 1 ½
® ¾ «Q »® ¾ 6.21
¯H 2 P ¿ 1 Q 12Q 21 ¬ 21 1 ¼ ¯H 2 ¿
177
where 1 and 2 are the strains in principal directions, 1 and 2 are the equivalent uniaxial
strains in principal directions, 12 and 21 are the Poisson’s ratios. Therefore the stress-strain
V c1 ½ ª E c1 0 º H 1P ½
® ¾ « 0 ® ¾
E c 2 »¼ ¯H 2 P ¿
6.22
¯V c 2 ¿ ¬
where Ec1 and Ec2 are the secant moduli in the principal directions. The orthogonal elasticity
ª E c1 Q 12 E c1 0 º
>D@c12 1 «Q E Ec 2 0 »
« 21 c1 » 6.23
1 Q 12Q 21
«¬ 0 0 1 Q 12Q 21 Gc12 »¼
The shear modulus in equation 6.23 is given by Attard et al. (Attard et al. 1996):
1
1 Q 12Q 21 Gc12 >Ec1 1 Q 12 Ec 2 1 Q 21 @ 6.24
4
For cracked concrete elements, both Poisson’s ratios in Equation 6.23 are taken as zero, thus
ª E c1 0 0 º
>D@c12 « 0
« Ec 2 0 »» 6.25
«¬ 0 0 Gc12 »¼
A rate-type creep law based on the application of Kelvin chains has been incorporated into the
finite element program RECAP by Chong et al. (2008). The rate-type law is based on the
solidification theory that has been reviewed in Section 2.6.2.4. The great advantage of rate-
178
type law over the integral-type law is that it does not require huge computer storage for
tracing the previous stress or strain history in the structure, but only the current states of
variables, hence it greatly reduces the computational effort at each time step.
In this model, the creep of concrete is described by N numbers of Kelvin Chain units. Each
unit contains a linear spring modulus and a viscosity dashpot (Figure 2-16). The constitutive
N
E P J P K P J P V, J ¦
P
JP
1
6.26
Where, E P and J P are the elastic moduli and viscosities of unit . Integrating the above
equation can generate the constitutive relation between viscoelastic microstrain at time t and
KP 1
V ¦ AP (1 e
( t t ') / t P )
J ), WP and AP 6.27
EP EP
where is the so called retardation time of the -th Kevin chain unit. Hence the creep
¦ AP (1 e
( t t ') / W P )
C (t , t ' ) ) 6.28
In equation 6.28, must be determined in order to solve the ill-posed problem since different
retardation times can give rise to good fits of the experimental creep curves. Bažant and Xi
(1995) formulated the solidification theory with a continuous retardation spectrum. They also
specified the way to convert this continuous spectrum into a discrete spectrum for numerical
the retardation spectrum is shown in Figure 6-4. A in equation 6.28 is the retardation
spectrum for the discretization of the Kelvin chains and is presented in terms of a continuous
L(W ) «
> @
ª 2n 2 (3W ) 2 n 3 n 1 (3W ) n º (3W ) 3
» q2
¬« >1 (3W ) n@ 3
¼» 2
6.30
«
> @
ª n(n 2)(3W ) n 3 n 1 (3W ) n n 2 (3W ) 2 n 3 º (3W ) 3
» q2
«¬ >
1 (3W ) n @
2
»¼ 2
Notice in Figure 6-4, the negative infinite area A0 of the retardation spectrum (Bazant et al.
1997) can be obtained by subtracting equation 6.29 from the microscopic creep compliance
function.
180
Chong et al. (2008) adopted the exponential algorithm proposed by Bazant (Bazant 1982;
Bazant and Prasannan (1989b), and incorporate it into the finite element formulation in the
program RECAP.
crack and total strain formulation. Therefore, the total stress of every concrete element is only
related to the total strain at every load step and independent of any previous stress-strain
history. Therefore the creep strain that is calculated from the solidification theory should also
'J i 1
'H v t i 1 6.31
Q i 1 / 2
where the subscripts i and i+1/2 indicate the reference to time ti and the time in the middle of
N
'J i 1 ¦J
j 1
j i 1 J ji G'VA0 6.33
ª 1 v 0 º
G « v 1 0 »
« » 6.34
«¬ 0 0 2 1 v »¼
The modified form of the viscoelastic microstrain at time ti+1 for the jth Kelvin chain given by
GV i 1 1 Oj
Jj i 1
J j e 'y i
1 e 'yi G'V 6.35
i
Ej Ej
where
'y j
't 1 e
'y j ; Oj ; and 'V V i V i 1 6.36
Wj 'y j
The volume of the solidified matter at mid-time of a logarithmic time step, vi+1/2, is then given
by:
1
ª 1 q º
vi 1 / 2 « 3» 6.37
¬ t i 1 / 2 q 2 ¼
Then the change in the viscous, non-recoverable, component of creep is evaluated from
'H f t i 1 GV i 1 / 2
6.38
't K i 1 / 2
K i 1 / 2 q 41t i 1 / 2 , into equation 6.38 then the change in the viscous strain per finite element
GV i 1 / 2 q 4 't
'H f t i 1 6.39
t i 1 / 2
Consequently the total creep strain can be obtained by adding both the viscoelastic strain and
viscous strain change components into the total creep strain from the previous time step:
H v t i 1 H v t i 'H v t i 1 6.40a
H f t i 1 H f t i 'H f t i 1 6.40b
In this study, since shrinkage is considered to be independent of stress, the shrinkage strain at
Ash t
H sh t 6.41
Bsh t
Thus, the total time-dependent strain at time t due to creep and shrinkage in the concrete is
given as:
H0 t H v t H f t H sh t 6.42
For the sake of finite element implementation, 0(t) is treated as an inelastic strain and then
converted into pre nodal forces P0 applied to the finite element nodes. These equivalent nodal
forces are not computed explicitly in the computational algorithm. However, the pre-strains
0(t) are accounted by being deducted from the total strains before the stresses of the elements
are evaluated. Therefore the pre nodal forces are actually indirectly considered.
183
The two node truss element is adopted in this study for modeling the reinforcement used in
this study. Assume a reinforcing bar i-j with a cross section area of A, length of l, shown in
Figure 6-5. The angle between the truss and the global x axis is given as .
The longitudinal strain of the truss can be easily represented as long as the nodal deformations
ui ½
°v °
H
1
> c s c s@°® i °¾ 6.43
l °u j °
°¯v j °¿
where c and s are the cosine and sine values of ; u i , vi , u j , v j are the nodal deformation
(Figure 6-5).
According to Hooker’s law, the stress in the truss element can be calculated as:
ui ½
°v °
V EH
E
> c s c s @°® i °¾ 6.44
l °u j °
°¯v j °¿
Using virtual work theory the nodal force and nodal deformation of the truss element are
ª c2 cs c2 cs º
« »
AE « cs s 2
cs s 2 »
>K @ 6.45
l « c 2 cs c2 cs »
« »
¬« cs s2 cs s 2 ¼»
An elastic-plastic stress-strain relationship for the reinforcement is adopted for the finite
element analysis. The relationship is shown in Figure 6-6, in which fsy is the yielding stress
and sy is the corresponding yielding strain. Since this research is only focused on the in-
service load range for concrete structures, the steel bars generally have not yielded and the
The zero width interface element is designed to model the bond mechanism between the
concrete and the reinforcing bar in a reinforced concrete structure. An interface element was
firstly introduced by Goodman et al. (1968) to model the joint behaviour in systems of rock
blocks and joints. For reinforced concrete structures, the element layout and its connectivity
The relative displacement between node set 1 (consisting of nodes1 and 4) and node set 2
(nodes 2 and 3) represents the slip between the concrete and the steel and is given by
where the superscript “+” and “-” denote the upper and lower faces of the interface element,
respectively, and the subscripts t and n represent shear and normal movement, respectively.
The relative nodal displacements ui are linked to the continuous displacement field u by
2
'u ¦ N 'ui i >B@^u e ` 6.47
i 1
The matrix [B] in equation 6.47 relates the continuous field relative displacements u to the
ª N 1 0 N2 0 N2 0 N1 0º
>B@ « 0 N 1 »¼
6.48
¬ N1 0 N2 0 N2 0
The element stiffness matrix is computed by applying the principal of virtual work and is
given by
6.49
Ae
where Ae is the tangential contact surface area between the interface element and the adjacent
materials and [Db] is the constitutive model relating the bond stress to the relative
displacements between the reinforcing steel and concrete, for example, Ae is the surface area
of the reinforcing bar encased in the concrete and [Db] is given by:
ª E bt 0 º
>Db @ « 0 Ebn »¼
6.50
¬
where Ebt is the bond stress-slip modulus for the current stress state and Ebn is the normal
bond stress-slip modulus. To maintain compatibility between the reinforcing steel and the
concrete in the normal direction, a stiff value for Ebn is used in this study. In the tangential
direction the stiffness is obtained from the relevant constitutive law. Ebt is taken as the secant
For the four-node linear bond element, integration of equation 6.49 is undertaken explicitly
giving:
188
where c is the sum of the bar circumferences of all bars on a layer and L is the length of the
bond element. The bond element stiffness matrix [Kb] in the global coordinate system is
obtained by:
ª cos I sin I º
>Te @ « sin I cos I »¼
6.54
¬
The bond model must correctly predict the relationship between local bond stress and slip,
and must therefore consider the effects that may influence this relationship (section 2.5).
However most of the pull-out tests on which the available bond-slip relationships have been
calibrated have been undertaken on short anchorage lengths with various boundary conditions,
In the following, the CEB-FIP bond model (section 2.5.4) is modified by incorporating the
effects of local damage, primary cracking and concrete confinement pressure for modelling
short-term tension stiffening behaviour under in-service loads (Wu and Gilbert 2009a). At
service load levels, when the steel stress-strain relationship is linear-elastic (i.e. the steel stress
is less than the yield stress), slip normally does not exceed S1 in Figure 6-8. The basic bond-
slip relationship (ignoring cracking and steel stress state) is therefore based on the ascending
D
§s·
Wb W max ¨¨ ¸¸ 6.55
© s1 ¹
in which max is the maximum bond stress equal to 2.5 f cp if good concrete confinement is
A number of investigators have proposed to incorporate concrete cracking or damage into the
bond constitutive law (Bolander et al. 1992; Lowes et al. 2004; Soh et al. 2003), by reducing
190
the bond stiffness as a result of damage to the surrounding concrete elements. On the basis of
experimental results, Maekawa et al (2003) proposed a bond model that is related to the local
steel strain.
As shown in Figure 2-2, soon after first cracking, the drop in tension stiffening is substantially
attributed to the formation of primary cracks, where at each crack the bond stress drops to
zero and the slip is substantial. After the formation of the primary cracks (at the crack
stabilization stage), the loss of tension stiffening can be best modeled by reducing the bond
stress with increasing local steel stress (to account for cover controlled cracking). Two
governing parameters 1 and 2 are introduced here to take into account the local concrete
damage at primary cracks and the effect of steel stress on cover-controlled cracks, respectively.
O1 1 D c 6.56
In Equation 6.56, Dc is the non-local averaged damage parameter of the concrete elements in
the vicinity of the reinforcement and represents the severity of the cracks. According to the
basic concept of continuum damage theory, the magnitude of damage in part of the concrete
material can be represented by an internal variable Dc. In this study, damage occurs as soon as
concrete cracks in the major principal direction (İ1 > İtp). According to the concrete stress-
strain relationship in tension (Figure 6-3a), concrete is undamaged (Dc = 0) before the stress
reaches its peak value fct, and soon after cracking, damage starts to occur until eventually the
material has lost its integrity (when Dc = 1). Therefore, Dc can be given as
191
Dc 0 when H 1 d H tp ;
V 1c
Dc 1 when H tp H 1 d D 3H tp ;
Ec H 1 6.58
Dc 1 when D 3H tp H 1
In Equation 6.57, s is local steel stress in MPa. It can be seen that the bond stiffness
decreases linearly as 250MPa < s <500MPa, and it diminishes after s >500MPa. The basic
assumption made for this equation is that the bond stiffness is reduced linearly with increasing
steel stress at in-service stage. Equations 6.58 have been calibrated based on the short-term
test results presented in Chapter 4 and 5. Figure 6-9 gives a typical graph of the parameter 2
versus steel stress s within the in-service region for normal-ductility ribbed steel bars.
In Equation 6.55, max and s1 are treated as constants, which are related to the bond
confinement condition. The test data in Marvar (Malvar 1992) shows that the bond strength
envelope increases and its radial deformation decreases significantly with the application of
confinement pressure. This test data is adopted here to alter the peak bond strength max
according to the average confinement stress pc of concrete around the steel bar elements. A
pc
O3 1 6.59
f cp
192
2 1.2
1
O2 2.14 0.0044V s
0.8
0.6
0.4
0
0 100 200 300 400 500
Steel stress ı s
Consequently, the modified bond stress-slip law under short-term monotonic load can be
written as:
D
O3 §s·
Wb O1O2W max ¨¨ ¸¸ 6.60
© s1 ¹
In order to incorporate these parameters into the bond stress-slip relationship in the finite
element program, a non-local analysis is undertaken at the end of each load step. As shown in
Figure 6-10, at each integration point in the bond interface elements, the surrounding concrete
damage value and confinement stress within a radius R is averaged by a weighting function
(r), where r is the distance between the bond element integration point and the source
concrete element integration points within the radius R (as shown in Figure 6-10a). Therefore,
the weighted average damage parameter Dc and confinement stress pc can be written as:
193
Dc ³ Z' r D
V
c r , T dV 6.61a
pc ³ Z' r pc r, T dV 6.61b
V
Zr
Z' r
6.62
V
³ Z r dV
As indicated in Figure 6-10b, the weight function in this study is taken as:
ª § 2r · 2 º
Zr exp « ¨ ¸ » 6.63
«¬ © R ¹ »¼
§ § 2r · 2 ·
Exp¨ ¨ ¸ ¸
¨ ©R¹ ¸
© ¹
a): Non-local region of bond and concrete b): Weight function of non-local
element integration point analysis
The weighting function is largest at the nearest integration point and decreases as the distance
r increases. The weighted averaged parameters Dc and pc are then substituted into
194
Equations 6.56 and 6.59 to give the governing parameters 1 and 3. In this study, the concrete
In order to model the long-term behaviour and mechanism of tension stiffening, the
Chapter 2, few experimental programs have been undertaken to investigate the long-term or
As mentioned earlier, the time-dependent loss of tension stiffening at crack initiation stage in
a tension member is mainly attributed to the formation of new primary cracks caused by the
restraint provided by the steel bars to shrinkage. At this stage, the steel stress usually is less
than 250MPa and the local deterioration of bond is not severe. However, if the applied load
remains constantly after the primary crack pattern has stablised (usually when 250MPa < s
<400MPa), the bond break-down between the primary cracks with time becomes the
dominant factor causing the loss of tension stiffening. As a result, the time-dependent bond
stress is dependent on both internal and external actions. The breakdown of bond with time is
primary the result of restraint to concrete shrinkage by the reinforcing bar, but is also due to
creep.
In this study, the time-dependent bond model is considered to be associated with drying
shrinkage of the surrounding concrete and the steel stress state. In order to correctly model the
interaction of shrinkage and steel stress on the decrease of bond stress with time, a long-term
D
O3 §s·
Wb O1O2 O4W max ¨¨ ¸¸ 6.64
© s1 ¹
k1
§H t ·
O4 1 ¨¨ sh ¸¸ and O 4 t 0.5
© H cs ¹
6.65
§ V ·
k1 3.0¨1 s ¸
© 500 ¹
where H cs is the design shrinkage strain after 30 years that is specified in AS3600 (2009), and
V s is the local tensile steel stress in MPa. In this study, Equation 6.64 is only applied when
the structure is under long-term loading as it only accounts for the long-term loss of bond,
which means in short-term analysis the effect of 4 is not taken into account (4 = 1).
Figure 6-11 shows the shrinkage effects on bond given by Equation 6.65. The proposed
shrinkage-related bond model gives rise to a sequence of curves 4 that depict the bond stress
deterioration at different steel stress levels. The influence of concrete shrinkage on bond
breakdown is less significant at lower steel stresses, as shown in Figure 6-11. The lower limit
of 4 is 0.50, which indicates that ultimately reduction of bond stress-slip stiffness due to
restraint to drying shrinkage is assumed to be 50%. The accuracy of this approximation will
be verified in Chapter 7 by using the proposed bond model to model the experimental work.
196
4 1.2
0.8
200MPa 100MPa
0.6 300MPa
400MPa
0.4
0.2
0
0 100 200 300 400 500
Shrinkage strain (×10-6)
Figure 6-11: Bond breakdown with the amount of shrinkage by the proposed model.
In the finite element program RECAP, a variety of solution schemes are available. One of the
most prevailing method is called the Standard Newton Raphson Method (SNRM) is adopted.
In the SNRM, the tangential stiffness matrix [K0] is formed at the beginning of the first
iteration (Figure 6-12), and then the displacement can be approximated by:
The strain and stress in each element is then obtained from the initial displacement and the
constitutive law of the material. Next, the corresponding nodal forces {P1} are formed. The
tangential stiffness matrix [K1] according to the displacement [1] is formed and the
197
incremental displacement [2] under load {P1} = {P} - {P1} are calculated by:
Thus the approximated displacements at the second iteration are given by:
until the convergence criteria is satisfied (either [i+1] - [i] or {Pi}) is less than a preset
minimum limit. The solution procedure for a finite element model consists of a number of
loading steps, each of which contains several iterations as described above. For the analysis in
this study, convergence is set at 1 percent for displacement and force norms with a maximum
of 100 iterations for one load step. For the time analysis under constant load, the period of
loading is discretized into a predetermined number of time steps, and the analysis is
In this study, instead of using tangent stiffness matrix (Figure 6-12), a secant stiffness matrix
is adopted in Equations 6.66 – 6.69. In the analysis of a structure where softening occurs, the
secant stiffness approach offers better numerical stability than the tangent stiffness approach.
This is because a negative tangent stiffness may lead to numerical instability, while the
change of secant stiffness under softening is still gradual and a positive stiffness is guaranteed.
198
6.6 SUMMARY
A 2-dimensional continuum based finite element package RECAP has been modified and used
for the analysis of the experimental work in this thesis. The power and accuracy of the finite
element method has made it one of the most attractive approaches to analyse reinforced
This chapter presents the concrete, steel bar and bond interface elements and their constitutive
relationships that have been written into RECAP. The concrete model is a hyperelastic model
that is based on the total secant modulus at each load step. The cracking of concrete is
demonstrated by using the crack band theory that relates the fracture energy of concrete with
the element size in order to solve the mesh sensitivity issues. Creep is modelled with the rate-
type solidification creep theory and incorporated into the program by using an incremental
stress-strain relationship. Creep and shrinkage strains developing with time are treated as
199
The bond mechanism between concrete and reinforcing bar is modelled with the zero-width
interface element. A modified bond model is proposed by the author. The bond stress in this
study is not only related to slip, but also some local parameters (including concrete damage,
steel stress, confinement and shrinkage, etc). A non-local analysis is performed at the
beginning of each load step to obtain relevant local bond parameters from the adjacent
The Newton Raphson Method is used to solve the nonlinear finite element program. The
The finite element program has been used to analyse the test specimens described in Chapter
4 and 5, and the results are compared with the experimental measurement in the next chapter.
The program has also been verified by modeling some experimental specimens tested by
others. The proposed model is used to undertake a parametric investigation of the factors that
affect the time-dependent behaviour of cracked reinforced concrete and the magnitude of
tension stiffening.
CHAPTER 7 FINITE ELEMENT ANALYSIS
OF EXPERIMENTAL WORKS
7.1 OVERVIEW
The finite element program presented in the previous chapter has been used to analyse a
number of specimens tested experimentally and the results are presented in this Chapter. The
specimens tested by the author in Chapter 4 and 5 are modeled first. Some other specimens
tested in other studies and reported in the literature have also been analysed (Gribniak 2009,
Nejadi and Gilbert 2004b). The numerical results are then compared with the experimental
results in order to verify the finite element model. Good correlation between numerical and
experimental results shows that the finite element program with the proposed bond model is
concrete structures.
The finite element mesh designed for the analysis of the uniaxial tension prisms (discussed in
Chapter 4) is shown in Figure 7-1. Symmetry requires only one quarter of each prism to be
201
modeled. An element size of 5 mm by 5mm was adopted, so as to reliably model the local
stress redistribution between adjacent primary cracks. The prisms are modeled using 1332
nodes, 1100 concrete elements, 110 two node steel reinforcement truss elements and 110 zero-
width bond interface elements. The concrete elements are isoparametric quadrilateral plane
stress elements with numerical integration performed using 2 × 2 gauss quadrature. The steel
elements are connected to the concrete elements with overlapping nodes by the zero-width
bond interface elements. The finite element solution is dependent on the mesh size. Relatively
small elements are necessary to accurately model the discrete nature of cracking and bond-slip
at the steel-concrete interface. However, for the specimen sizes considered here, a further
reduction in the element size resulted in relatively little increase in accuracy. In the modeling,
difference being assumed. Therefore, first cracking could be initiated in the concrete element
with the lowest assigned tensile strength and the tensile stress in surrounding elements in the
The material properties used as input data for the finite element model were those measured
in the experimental program. Relevant equations can be found in section 6.2.2.1. Table 7-1
gives the major instantaneous material parameters for concrete. Notice that the value of 2 and
3 are dependent on both fracture energy and the finite element size. Other property
Figure 7-1: Finite element model configuration for uniaxial tension prisms.
Table 7-1: Parameters for the instantaneous constitutive law of concrete (uniaxial tension
specimens).
In order to obtain the appropriate input data for the solidification creep model, the creep
coefficient is transformed into the compliance function J(t,t’), indicating the strain at age t
caused by a unit constant stress applied at age t’. Figure 7-2 plots the compliance function
curves obtained for this study. The empirical material constants q2, q3 and q4 in Equations
203
6.37 and 6.39 were determined by fitting the compliance data into the curves (Table 7-2). The
Dirichlet series was discretized into seven Kelvin chain units for storing the viscoelastic strain
history. The elastic moduli E of the -th Kelvin chain can be obtained by substituting the
predefined (retardation time) into Equations 6.29 and 6.30. The negative retardation
spectrum area A0 can be calculated by subtracting the sum of 1/E from the microscopic
creep compliance function. In this study, the retardation time is taken as 0.0001, 0.001, 0.01,
0.1, 1, 10, and 100 days. The corresponding values of E are given in Table 7-3.
1.4E-04
J(t,t') (MPa-1)
1.2E-04
1.0E-04
8.0E-05
6.0E-05
Experiment result
4.0E-05
2.0E-05 fit in curve
0.0E+00
0 10 20 30 40 50 60
t-t' (days)
(a) For the first batch (STN12, STN16, STS12, STS16, LTN12A and LTN12B)
1.2E-04
J(t,t') (MPa-1)
1.0E-04
8.0E-05
6.0E-05
Table 7-2: Major creep solidification data for uniaxial tension tests
A0 q2 q3 q4
Specimens
(MPa-1) (μ/MPa) (μ/MPa) (μ/MPa)
*First batch: STN12, STN16, STS12, STS16, LTN12A and LTN12B; Second batch: LTN12C
and LTN12D.
Table 7-3: The retardation times and the correlated E for uniaxial tension tests
Eμ μ
Specimens
(MPa) (days)
0.15348 0.0001
0.10448 0.001
0.072 0.01
First batch 0.05026 0.1
0.03555 1.0
0.02548 10.0
0.01850 100.0
0.160 0.0001
0.109 0.001
0.075 0.01
Second batch 0.052 0.1
0.037 1.0
0.026 10.0
0.019 100.0
The Newton Raphson solution scheme is used in the analysis to capture the nonlinear
behaviour of the structure. Because this study is only concerned with the tension stiffening
effect under in-service conditions, the analysis was terminated when the steel stress reached
205
the yield stress (500MPa), where tension stiffening is assumed to have disappeared.
Figure 7-3 shows the numerical and experimental results for the axial force versus average
strain diagram for the short-term tension prisms. The responses obtained from the finite
element model using the proposed bond model and the CEB-FIP bond model (only for STN12
and STN16) are shown in the figure. The average strain is calculated from the finite element
model by dividing the axial deformation between the two monitored nodes A and B, as shown
In general, the proposed bond model gives excellent correlation with the experimental
response, especially at the cover-control crack stage. At first cracking (P = Pcr), there is an
abrupt change of stiffness and the stiffness continues to degrade under increasing deformation
as further cracks occur. As the load increases, the tension stiffening strain gradually reduces.
By contrast, the CEB-FIP bond model tends to overestimate tension stiffening under
increasing loads.
To further investigate the difference between proposed bond model and CEB-FIP bond model,
tension stiffening factor t of STN12 and STN16 are calculated. Figure 7-4 shows the tension
stiffening factor versus average strain curves for STN12 and STN16. The tension stiffening
factor predicted using the proposed bond model also shows good correlation with the test
results, with t decreasing as the average strain increases. The results predicted using the
CEB-FIB bond model tend to increase with increasing strain after all the cracks have formed
stiffness is more significant in the specimen with the smaller reinforcement ratio (STN12).
The tension stiffening factor obtained from the experiments declines to almost zero as the
bare bar steel stress approaches the yield stress of the steel.
Load (kN)
60
Load (kN)
60
50
50
40 40
30 30
20 Experimental result
Proposed bond model 20
Experimental result
CEB-FIP bond model
10 Bare bar response 10 Proposed bond model
Bare bar response
0 0
0 500 1000 1500 2000 2500 -500 0 500 1000 1500 2000 2500
Average strain (×10-6)
Average strain (×10-6)
120
Load (kN)
120
Experimental result
100 100 Proposed bond model
80 Bare bar response
80
60 60
40 Experimental result
Proposed bond model 40
20 CEB-FIP bond model
Bare bar response 20
0
0
0 500 1000 1500 2000 2500
-500 0 500 1000 1500 2000 2500
Average strain (×10-6)
Average strain (×10-6)
Figure 7-3: Comparison between experimental and finite element results for the short-term
uniaxial tension tests (Load versus average strain).
207
Experimental result
1.0 proposed bond model 1.0 Proposed bond model
CEB bond model CEB bond model
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
Average strain (×10-6) Average strain (×10-6)
Figure 7-4: Comparison between experimental and finite element results for the short-term
uniaxial tension tests (tension stiffening factor versus average strain).
The significant difference between the proposed and CEB-FIP bond models is caused by the
fact that the latter directly links bond stress to slip which keeps increasing as external load
increases. However, in reality, bond stress is also related to the local concrete damage due to
The analysis output from the finite element model was converted into GID postprocessor in
order to show the local results in a graphical format. Figure 7-5 and 7-6 show the output from
the GID postprocessor, including tensile strain contours, the variation of steel force, bond
stress and slip distributions along each member at different load levels. The principal tensile
strains at the crack locations are much greater than the strain corresponding to the onset of
cracking (about 0.0001). In between the cracks; all the concrete strains remain in the elastic
range (< 0.0001) and the concrete remains uncracked (intact). Also at the cracks the steel
force reaches its maximum, and mid-way between cracks, it is at a minimum. The
numerically-generated graphs show that, with increasing load, the steel force tends to be more
uniform along the bar (due to the degradation of bond) as the tension stiffening effect
208
decreases. The proposed bond model gives rise to reduced bond stress with increasing slip as
the load is increased within the service load range ( 250MPa < s < 400MPa). However, the
variation of steel force measured during the tests is not smooth (Section 4.6), because the
breakdown in bond is quite local at the location of cover-controlled cracks and this is reflected
by the relatively large differences in strains measured at adjacent strain gauges. Therefore, the
proposed bond model reduces the average bond stress with increasing steel stress without
capturing the very local effect of damage on bond caused by cover-controlled cracking.
209
(a) Tensile strain contour of STN12 (at 50kN) (b) Tensile strain contour of STS12 (at 50kN)
Force (kN)
60
Axial force (kN)
60
50 50
40 40
30 30
20 20
40kN
10 40kN
50kN 10
50kN
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Distance from the middle section (m) Distance from the middle section (m
(c) Steel force distribution of STN12 (d) Steel force distribution of STS12
4.0 50kN
Bond stress (MPa)
4.0
2.0 2.0
0.0 0.0
-2.0 0 0.1 0.2 0.3 0.4 0.5 0.6 -2.0 0 0.1 0.2 0.3 0.4 0.5 0.6
-4.0 -4.0
40kN
-6.0 50kN -6.0
Distance from the middle section(m)
Distance from the middle section (m)
(e) Bond stress distribution of STN12 (f) Bond stress distribution of STS12
Slip (mm)
0.3 0.3
40kN
0.2 50kN
0.2
Slip (mm)
0.1 0.1
0.0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6
0 0.1 0.2 0.3 0.4 0.5 0.6
-0.1
-0.2 40kN
50kN -0.2
-0.3 Distance from the middle section(m)
Distance from the middle section (m)
Figure 7-5: Finite element results for STN12 and STS12 (local response).
210
(a) Tensile strain contour of STN16 (90kN) (b) Tensile strain contour of STS16 (90kN)
Force (kN)
100
Force (kN)
100
80 80
60 60
76kN 76kN
90kN 90kN
40 40
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Distance from the middle section (m)
Distance from the middle section (m)
(c) Steel force distribution of STN16 (d) Steel force distribution of STS16
2.0
2.0
0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
-2.0
-2.0
-4.0
-4.0 -6.0
Distance from the middle section (m)
Distance from the middle section (m)
(e) Bond stress distribution of STN16 (f) Bond stress distribution of STS16
0.2
0.2 76kN 76kN
90kN 90kN
0.1
0.1
Slip (mm)
Slip (mm)
0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
-0.1 -0.1
-0.2 -0.2
Distance from the middle section (m) Distance from the middle section (m)
Figure 7-6: Finite element results for STN16 and STS16 (local response).
211
Figure 7-7 plots the tension stiffening strain versus time curves of the long-term uniaxial
tension tests generated by the finite element models (with both the CEB-FIP bond creep
model and proposed bond model). The experimental results of LTN12A and LTN12C (Figure
7-7 (a) and (b)) indicate that during the long-term loading period, tension stiffening strain
decreases gradually, which can also be accurately demonstrated by the results of the proposed
shrinkage-related bond model. The CEB bond creep model in fact predicts responses that
generally indicate a larger instantaneous tension stiffening strain (as discussed for the short-
term analysis) and an increase of tension stiffening strain with time. The restraint to concrete
shrinkage gradually builds up the tensile stress in the concrete and the shortening of the entire
member, however, it also induce cover-controlled cracks and a breakdown of bond and
consequently member stiffness. The CEB bond creep model only relates the deterioration of
bond stress with creep (time period after first loading) and fails to consider the effects of
response for the uniaxial tension members under long-term sustained tensile loads. The
proposed shrinkage-related bond model takes account of this effect and hence provides the
The finite element results of LTN12B and LTN12D are quite different from those of LTN12A
and LTN12C in terms of the development of tension stiffening strain with time. The former
were under relatively small sustained loads that initially produce one or no cracks and the
steel stresses were relatively small. The latter were fully-cracked under sustained loads and
steel stresses were relatively large. The loss of tension stiffening strains in LTN12B and
LTN12D is the result of new crack formed with time due to the restraint to shrinkage by the
reinforcing bar. The dots in Figure 7-7 (c) and (d) indicates the time at which a new crack
212
occurred in the models, and a sharp drop of tension stiffening strain is observed thereafter.
Both the proposed bond model and CEB bond model capture the new crack formation
properly. However, the experimental results show a steadier decrement of tension stiffening
strain compared with the finite element results. The formation of new cracks during the tests
appears to be a more gradual process, while in the finite element analysis an abrupt change
occurs as the new crack form (typical at one numerical step). Nevertheless, in the long-term,
after the development of time-dependent cracking the experimental and finite element results
The loss of tension stiffening in the long-term tension members can also be seen in the local
responses shown in Figures 7-8 and 7-9. For LTN12A or LTN12C, no additional cracks are
monitored by the model but clear decreases of local steel stresses and bond stresses are found
(Figure 7-8 (c), (d), (e), and (f)). For LTN12B or LTN12D, one new crack formed during the
sustained loading period (no crack had formed under instantaneous loads), which causes a
significant redistribution of local stresses and loss of tensile stress in the concrete around the
cracks. The finite element model is a valuable research tool to undertake a detailed
700 600
Proposed bond model Experiment result
600 CEB-FIP bond model 500 CEB bond creep model
Proposed bond model
500
400
400
300
300
200
200
100
100
Concrete age (days) Concrete age (days)
0 0
33 43 53 63 73 83 28 43 58 73 88
(a) Tensile strain contour of LTN12A after 60 days (b) Tensile strain contour of LTN12C after 60 days
Force (kN)
60
Force (kN)
60
50
50
40 40
30 30
20 Start of test 20
Start of test
10 After 60 days 10 After 60 days
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Distance from the middle section (m) Distance from the middle section (m)
(c) Steel force distribution of LTN12A (d) Steel force distribution of LTN12C
2.0 2.0
0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
-2.0 -2.0
-4.0 -4.0
(e) Bond stress distribution of LTN12A (f) Bond stress distribution of LTN12C
0.1
Slip (mm)
0.0
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6
0 0.1 0.2 0.3 0.4 0.5 0.6
-0.1 -0.1
-0.2
-0.2
Distance from the middle section (m) -0.3 Distance from the middle section (m)
Figure 7-8: Finite element results for LTN12A and LTN12C (local response).
215
(a) Tensile strain contour of LTN12B (after 60 days) (b) Tensile strain contour of LTN12D (after 60 days)
30
Force (kN)
Force (kN)
Start of test 30 Start of test
After 60 days After 60 days
20 20
10 10
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Distance from the middle section (m) Distance from the middle section (m)
(c) Steel force distribution of LTN12B (d) Steel force distribution of LTN12D
2.0 2.0
0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
-2.0 -2.0
-4.0 -4.0
-6.0 Distance from the middle section (m) -6.0 Distance from the middle section (m)
(e) Bond stress distribution of LTN12B (f) Bond stress distribution of LTN12D
0.2 0.2
Start of test Start of test
After 60days After 60 days
0.1 0.1
Slip (mm)
Slip (mm)
0.0 0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
-0.1 -0.1
Figure 7-9: Finite element results for LTN12B and LTN12D (local response).
216
The finite element models for both beams and slabs are depicted in Figure 7-10. Due to
symmetry, only half of the span of each member is modeled. Four-node isoperimetric
elements, two-node truss elements and interface elements are also used to model the concrete,
reinforcing bars and bond interface, respectfully. An element size 10mm by 10mm is adopted,
which leads to a large number of nodes and elements of one model. Larger element size can
be adopted to reduce the number of nodes and elements and improve the computational speed.
However, in this analysis, in order to represent a more discrete-like cracking pattern in the
numerical results, smaller element sizes are chosen. The steel plates at the supports are
modeled by using very stiff linear elastic elements. A vertical roller is placed under the steel
plate to model the boundary condition at the support, while horizontal rollers are placed on
the outer right edge of the model at mid-span. The concrete elements used in the models are
assigned with a randomized tensile strength equal to the tensile strength of concrete ± 2%.
The Newton-Raphson solution scheme is also used for the finite element analysis of the
beams and slabs. For the short-term analysis, small load steps are adopted to capture the
nonlinear performance at the onset of cracking. The analysis is terminated at the load when
the steel is about to yield. For the long-term analysis, the load is maintained constant after
instantaneous loading; and the time analysis is undertaken at each time instant until the end of
the test. The mid-span deflection of the specimens is the vertical displacement at Node A in
Figure 7-10.
217
The major parameters of concrete instantaneous constitutive law for the flexural tests are
given in Table 7-4. Figure 7-11 demonstrates the calculated compliance function J(t,t’) in
accordance with the measured creep coefficient and the data used to calculate the numerical
model. The calculated creep solidification parameters A0, q2, q3 and q4 are given in Table 7-5.
The corresponding elastic moduli E P and retardation time W P of the Kelvin chain units are
Table 7-4: Parameters for concrete instantaneous constitutive law (flexural tests)
BSTN2-16
26.0 2.2 25000 0.333 36.6 163
BSTN3-16
BSTS2-16
40.0 2.4 32000 0.333 41.1 185
BSTS3-16
SSTN4-12 28.6 1.8 25000 0.333 50.0 226.6
SSTS4-12 32.6 2.0 29000 0.333 68.0 320.6
SLTN4-12A
28.6 1.8 25000 0.333 50.0 226.6
SLTN4-12B
A0 q2 q3 q4
Specimens -1
(MPa ) (μ/MPa) (μ/MPa) (μ/MPa)
Table 7-6: The retardation times and the correlated E of the Kevin chain units for flexural
tests.
Eμ μ Eμ μ
Specimens Specimens
(MPa) (days) (MPa) (days)
0.21772 0.0001 0.34758 0.0001
0.15734 0.001 0.23601 0.001
0.10813 0.01 0.16219 0.01
Beams 0.07526 0.1 Slabs 0.11289 0.1
0.05308 1.0 0.07963 1.0
0.03794 10.0 0.05691 10.0
0.02748 100.0 0.04122 100.0
219
1.0E-04
6.0E-05
4.0E-05
Experiment result
2.0E-05
fit in curve
0.0E+00
0 10 20 30 40
t-t' (days)
(a) Beams
1.4E-04
J(t,t') (MPa-1)
1.2E-04
1.0E-04
8.0E-05
6.0E-05
Experiment result
4.0E-05
2.0E-05 fit in curve
0.0E+00
0 20 40 60 80 100 120
t-t' (days)
(a) Slabs
Figure 7-12 illustrates the moment versus mid-span deflection diagrams for the short-term
flexural members. The figure compares the results from the tests and the finite element
models using the proposed bond model, with excellent agreement observed for each specimen.
Since the fully-cracked responses are the same for both experimental and finite element
results, the effect of tension stiffening observed in the experimental results can be accurately
220
simulated by the finite element results. In addition, the finite element model is capable of
predicting the initial deflections before loading due to the gradual shrinkage warping of the
member, and also the decreased cracking moment caused by restraint to initial shrinkage
(Figure 7-12 (b), (d) and (f)). In general, the model accurately simulates the tension stiffening
Locally, as can be seen in Figure 7-13, 7-14 and 7-15, the models provide an insight into the
mechanisms of tension stiffening for the short-term flexural members. Similar to the
responses for the uniaxial tension members (Figure 7-5 and 7-6), the tensile strain contours
illustrate the cracking patterns of the flexural members. At cracked sections, the steel forces
are at maximum, and then drop to a minimum in-between the cracks, as shown in the steel
force distribution diagrams. Before loading, the restraint to shrinkage by the reinforcing bar
results in tensile force in the concrete and compressive force in the steel bar. Therefore, the
steel forces near the left support where the moment is low in the pre-shrunk members
(BSTS2-16, BSTS3-16 and SSTS4-12) are negative. This tensile force in the concrete due to
restraint to shrinkage, working together with the applied moment, may be large enough to
cause additional primary cracks. The distributions of bond stress and slip are also affected by
the formation of new cracks under increasing loads, as shown in the figures. Slip reaches its
peak and bond stress drops to zero at the cracks. As the moments keep increasing and no more
cracks are formed, slip continues to increase due to the widening of existing cracks. Whereas
the bond stress in-between the cracks seems not to vary too much, which can be indicated by
the fact that the gradient of local steel force does not change too much as well.
221
Moment (kN.m)
Moment (kN.m)
60 60
C Experimental result
50 C 50
Fully-cracked
FE result
40 40
30 30
20 Experimental result 20
FE result
10 10
Fully-cracked response
0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Mid-span deflection (mm) Mid-span deflection (mm)
70 Moment (kN.m) 70
60 60
50 50
40
40
30
Experimental result 30
20 Experimental result
FE result 20
FE result
10
Fully-cracked response Fully-cracked response
10
0 O
0 1 2 3 4 5 0
Mid-span deflection (mm) 0 1 2 3 4 5 6
Mid-span deflection (mm)
25 25
Moment (kN.m)
20 20
15 15
10 10
Experimental result
FE result Experimental result
5 5 FE result
Fully-cracked response
Fully-cracked response
0 0
0 3 6 9 12 15 18 0 3 6 9 12 15 18
Mid-span deflection (mm) Mid-span deflection (mm)
(a) Tensile strain contour of BSTN2-16 at 40kN.m (b) Tensile strain contour of BSTS2-16 at 40kN.m
Tensile bar force (kN)
80
crack crack crack crack crack
40
30
20
10
0
-10
0.1 0.3 0.5 0.7 0.9 1.1 1.3
Distance from left support to mid span (m) 0.1 0.3 0.5 0.7 0.9 1.1 1.3
Distance from left support to mid span (m)
(c) Steel force distribution of BSTN2-16 (d) Steel force distribution of BSTS2-16
6 6 at 28.8kN.m
at 28.8kN.m
at 40kN.m
at 40kN.m at 51.2kN.m
4 4
at51.2kN.m
Bond stress (Mpa)
2 2
0 0
0.1 0.3 0.5 0.7 0.9 1.1 1.3 0.1 0.3 0.5 0.7 0.9 1.1 1.3
-2 -2
-4 -4
-6 -6
Distance from left support to mid span (m) Distance from left support to mid span (m)
(e) Bond stress distribution of BSTN2-16 (f) Bond stress distribution of BSTS2-16
0.2 0.2 at 28.8kN.m
at 28.8kN.m
at 40kN.m at 40kN.m
at 51.2kN.m at 51.2kN.m
0.1 0.1
Slip (mm)
Slip (mm)
0.0 0.0
0.1 0.3 0.5 0.7 0.9 1.1 1.3 0.1 0.3 0.5 0.7 0.9 1.1 1.3
-0.1 -0.1
-0.2 -0.2
Distance from left support to mid span (m)
Distance from left support to mid span (m)
Figure 7-13: Finite element results for BSTN2-16 and BSTS2-16 (local response).
223
(a) Tensile strain contour of BSTN3-16 at 60.8kN.m (b) Tensile strain contour of BSTS3-16 at 60.8kN.m
70
at 60.8kN.m
50
50
40
30 30
20
10
10
0
-10
0.1 0.3 0.5 0.7 0.9 1.1 1.3
Distance from left support to mid span (m) 0.1 0.3 0.5 0.7 0.9 1.1 1.3
Distance from left support to mid span (m)
(c) Steel force distribution of BSTN3-16 (d) Steel force distribution of BSTS3-16
6 at 40kN.m 6
at 40kN.m
at 51.2kN.m at 51.2kN.m
at 60.8kN.m 4 at 60.8kN.m
4
Bond stress (Mpa)
Bond stress (Mpa)
2
2
0
0
0.1 0.3 0.5 0.7 0.9 1.1 1.3
0.1 0.3 0.5 0.7 0.9 1.1 1.3
-2
-2
-4
-4
-6 Distance from left support to mid span (m)
Distance from left support to mid span (m)
-6
(e) Bond stress distribution of BSTN3-16 (f) Bond stress distribution of BSTS3-16
0.2 0.2 at 40kN.m
at 40kN.m
at 51.2kN.m at 51.2kN.m
at 60kN.m
at 60.8kN.m
0.1 0.1
Slip (mm)
Slip (mm)
0.0 0.0
0.1 0.3 0.5 0.7 0.9 1.1 1.3 0.1 0.3 0.5 0.7 0.9 1.1 1.3
-0.1
-0.1
Figure 7-14: Finite element results for LTN12B and LTN12D (local response).
224
(a) Tensile strain contour of SSTN4-12 at 20kN.m (b) Tensile strain contour of SSTS4-12 at 18kN.m
60 60
at 12kN.m at 12kN.m
crack crack crack crack crack crack crack crack crack crack
at 16kN.m at 16kN.m
50
Tensile bar force (kN)
50 at 20kN.m at 20kN.m
10 0
0 -10
0.1 0.3 0.5 0.7 0.9 1.1 1.3
0.1 0.3 0.5 0.7 0.9 1.1 1.3
Distance from left support to mid span (m)
Distance from left support to mid span (m)
(c) Steel force distribution of SSTN4-12 (d) Steel force distribution of SSTS4-12
6 at 12kN.m 6 at 12kN.m
at 16kN.m at 16kN.m
at 20kN.m at 20kN.m
4 4
Bond stress (Mpa)
2 2
0 0
0.1 0.3 0.5 0.7 0.9 1.1 1.3 0.1 0.3 0.5 0.7 0.9 1.1 1.3
-2 -2
-4 -4
-6
Distance from left support to mid span (m) -6
Distance from left support to mid span (m)
(e) Bond stress distribution of SSTN4-12 (f) Bond stress distribution of SSTS4-12
0.2 at 12kN.m 0.2 at 12kN.m
at 16kN.m at 16kN.m
at 20kN.m at 20kN.m
0.1 0.1
Slip (mm)
Slip (mm)
0.0
0.0
0.1 0.3 0.5 0.7 0.9 1.1 1.3
0.1 0.3 0.5 0.7 0.9 1.1 1.3
-0.1
-0.1
-0.2
Distance from left support to mid span (m) -0.2
Distance from left support to mid span (m)
Figure 7-16 plots the mid-span deflection versus time results of SLTN4-12A and SLTN4-12B
from the experiment and finite element models. Generally the finite element model shows
good ability to capture the test responses that the mid-span deflection increases significantly
in the first 15 days under loads, which implicitly means that the change of tension stiffening
Figure 7-17 shows the development of the crack patterns of the slabs in terms of tensile strain
contour derived by the finite element models. It can be seen that in the first 15 days, one
additional crack is formed in the half length of both SLTN4-12A and SLTN4-12B (Figure 7-
17 (a) and (b)), which is deemed to be the major factor for the loss of stiffness. Thereafter, the
models give no more cracks till the end of the test and the loss of tension stiffening with time
20
Deflection (mm)
15
15
10
10
first 15days Experimental result
5 5 first 15days Experimental result
FE result
FE result
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time after first loading (days) Time after first loading (days
Figure 7-16: Comparison between experimental and finite element results for the long-term
flexural tests (mid-span deflection versus time).
226
(a) Tensile strain contour of SLTN4-12A after load (b) Tensile strain contour of SLTN4-12B
(c) Tensile strain contour of SLTN4-12A after 15 (d)Tensile strain contour of SLTN4-12B after 15
days days
Figure 7-17: Finite element results for SLTN4-12A and SLTN4-12B (local response).
So far, the finite element program, with the proposed bond model, introduced in Chapter 6, is
able to accurately predict time-dependent tension stiffening in both macro and micro scales.
The model appropriately takes account of the effect of shrinkage and creep in the concrete,
together with the basic mechanisms (cracking and bond deterioration) that cause the loss of
tension stiffening. In order to further verify that the finite element model can be applied in a
reinforced concrete structures, the experimental works conducted by Nejadi and Gilbert
long-term loads, in order to investigate the effects of shrinkage on crack width. Herein, the
author uses the experimental data to investigate the change of tension stiffening with time in
their tested slabs and to further verify the proposed shrinkage-related bond model.
227
Figure 7-18: Details of cross section for the slabs (Nejadi and Gilbert 2004).
Figure 7-18 shows the cross-sectional dimensions of the slabs constructed by Nejadi and
Gilbert. The slabs were simply supported over a span of 3.5m and each subjected to a constant
sustained distributed load q over the entire span. Table 7-7 tabulates the design details for
each slab. The specimens were all cast on the same day from the same batch of concrete and
moist cured for 14 days prior to the tests. The sustained loading commenced on the 14th day
and then the loads were held constant for 394 days. The constant loads were created by
placing pre-cast concrete blocks on the top of the slabs, together with the self-weight. The
shrinkage and creep characteristics of the concrete were also monitored during the tests, as
well as the instantaneous concrete and steel material properties. The measured material
properties were fully documented by Nejadi and Gilbert (2004). During the test, dial gauges
were used to monitor the mid-span deflections of the slabs, and the measured results are used
The finite element mesh used for the slabs of Nejadi and Gilbert were similar with that shown
in Figure 7-10, with only half length of each slab being modeled due to its symmetry. Each
slab model is composed of 2832 isoparametric concrete elements (10mm by 10mm size), 177
steel truss elements and 177 bond interface elements. The uniformly constant distributed loads
were applied as nodal loads at every node on the top of the slabs.
fcp = 24.8MPa; fct = 2MPa; Ec = 24950MPa; v = 0.2; Es = 200000MPa; fsy = 500MPa; and Gf =
75N/m.
Table 7-8: Creep coefficient and shrinkage strain (x10-6) (Nejadi and Gilbert 2004).
Hsh 0 14 109 179 403 591 731 772 784 816 825
229
According to Table 7-8, with first loading and the onset of drying both at 14 days, creep
coefficient c and shrinkage sh strain gradually increased with time. By the end of the tests,
the measured c and sh are 1.71 and 825 x10-6 respectively. For the finite element model,
parameters Ash and Bsh are taken as 950 and 45days for the shrinkage model in Equation
52.8MPa-1. Eight Kelvin chain units were used and the calculated retardation time and
elastic modulus E for each Kelvin chain are given in Table 7-9.
Table 7-9: Kelvin chain properties for solidification creep model for Nejadi and Gilbert’s
slabs.
E (MPa) (Days)
0.0848 0.0001;
0.0721 0.001
0.0621 0.01
0.0541 0.1
0.0478 1
0.0428 10
0.0388 100
0.0356 1000
Figure 7-19 shows the numerical and experimental mid-span versus time responses of s1-a
and s2-a in Nejadi and Gilbert’s test. The finite element model results are in good agreement
with the measured experimental results as shown. It appears that the mid-span deflections of
the slabs increased significantly within the first 50 days then tended to be steadier thereafter.
230
In the long-term, the mid-span deflections of the slabs increased by up to 150% of the
instantaneous values. Shrinkage warping was concerned as the major contribution to the time-
35
35
Deflection (mm)
30
30
25
25
20
20
15
15
10 Experimental result Experimental result
10
5 FE result FE result
5
0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Time after first loading (days) Time after first loading (days)
Figure 7-19: mid-span deflection versus time for s1-a and s2-a.
In order to analyse the contribution of tensile concrete to the stiffness of the slabs and how it
changes with time, the fully-cracked deflections at mid-span were calculated in 15, 50, and
394 days after first loading. The relaxation method presented in Appendix B was used to
calculate the responses of the fully-cracked sections and the fully-cracked deflections of the
slabs with time. Then the tension stiffening deflection at different loading stages can be
obtained by subtracting the actual mid-span deflections from the fully-cracked deflections.
Table 7-10 summarises the calculated tension stiffening deflections from both experimental
and numerical results at different stages of the loading history for the slabs considered here.
It is seen in Table 7-10 that for most of the slabs tension stiffening deflections dropped
dramatically in the first 15 days after instantaneous loading by 30% - 50%. Thereafter, the
decrease of tension stiffening deflections slowed down. This decrease of tension stiffening is
quite similar to the observation of the slabs tested by the author, although the rate of decrease
of tension stiffening was slightly different. The total tension stiffening deflections over
231
sustained loading period also varied among the slabs. For the slabs under consideration, the
loss of tension stiffening deflections by the end of the test varied from 46% - 94%. The major
mechanism to cause the loss of tension stiffening was shrinkage-induced cracks developing
with time.
Table 7-10: Calculated tension stiffening deflections for Nejadi and Gilbert’s slabs.
s1-b 5.28 9.98 10.11 6.99 7.50 7.49 6.98 5.34 5.48
s2-a 9.87 5.25 5.07 2.25 1.91 2.34 2.41 0.3 1.44
s2-b 6.81 7.34 7.03 4.95 4.87 5.26 3.90 3.2 1.97
s3-a 11.35 4.94 4.75 1.71 2.39 2.06 1.93 1.36 0.88
s3-b 8.34 6.45 6.46 4.00 3.31 4.58 2.70 2.58 2.00
For the slabs subjected to higher applied moments (s1-a, s2-a and s3-a), the decrement of
tension stiffening deflections with time seems more significant than the slabs subjected to
lower applied moments (s1-b, s2-b, and s3-b). For instance, the experimental tension
stiffening deflection of s3-a decreased by 65.4% after 15 days and by 72.5% at the end of the
test, while the corresponding changes for s3-b were 38.0% after 15 days and 60.0% at the end.
Notice this phenomenon was not quite remarkable in the tests undertaken by the author. A
comparison between the results for s1-a (Ast = 220mm2) and s2-b (Ast = 330mm2) shows that
under the same loading level, the reinforcing ratio has little effects on the rate of change of
232
tension stiffening deflections. Tension stiffening deflection of s1-a decreased by 38.7% after
15 days and 48.3% at the end, while that of s2-b decreased by 32.6% after 15 days and 56.4%
at the end.
The crack patterns of s1-a and s2-a at instantaneous loads, after 50 and 394 days are indicated
by the tensile strain contours in Figure 7-20. The graphs indicate that new primary cracks
were formed during the sustained loading period. For the first 50 days under loading, there
were 3 and 2 new cracks discovered in the numerical models of s1-a and s2-a, respectively,
and then after 394 days, 1 more crack had developed in each specimen. Apparently the figure
shows that the time-dependent new cracks were normally formed at lower-moment regions at
some distance away from the existing cracks. Restraint to shrinkage is regarded as the major
factor triggering the time-dependent cracking and causing the loss of tension stiffening.
considered as another factor contributing to the loss of tension stiffening in the slabs. Derived
from the finite element results, the bond stress distributions at different loading stages (after
50 days and 394 days) along the specimens for s1-a and s2-a are shown in Figure 7-21. The
bond stresses in-between the existing cracks in the slabs. Regardless of the influence of new
cracks on bond, the maximum bond stress in-between the existing cracks in s1-a dropped
from 3.48MPa to 3.01MPa from 50 days after first loading to 394 days after first loading,
while for s2-a, it dropped from 3.52MPa to 3.11MPa during the same time interval.
233
2 2
0 0
0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7
-2 -2
-4 -4
-6 -6
Distance from left support to mid span (m) Distance from left support to mid span
(four couples of twin specimens), each with a single span of 3m, were tested under the four-
point loading. The design dimensions of the beams are shown in Figure 7-22.
Figure 7-22: Longitudinal (a) and cross-sectional (b) reinforcement of the test beams;
notation of cross-section (c) (Gribniak 2009).
The experimental program comprised of two series of beams. In the first series (S-1 and S-2),
four N10 bars were used as the tensile reinforcement, whereas in the second series (S-3 and S-
4), two N14 bars were used. Within the same series, a twin specimen designated with extra
‘R’ was produced. ‘R’ means the twin specimen that had heavy top reinforcement of three
attempt to eliminate the shrinkage curvature on the specimens. However the longitudinal
restraint to shrinkage within the specimen will be significantly greater. Table 7-11 summaries
Free shrinkage measurements were performed on prisms of 100 × 100 × 400mm and 280 ×
300 × 350mm in size. The latter was in fact the fragments of the beams, so that the size effects
235
on shrinkage measurements can be eliminated in this situation. Steel gauge studs, with the
base 200mm, were glued or embedded in the concrete to measure the companion free
shrinkage strains.
The finite element configuration of the beams is similar to that shown in Figure 7-10, whereas
in this case not only tensile reinforcement but compressive and transverse reinforcement are
also provided. The tensile steel truss elements are connected to the concrete elements with the
bond interface elements, while the compressive and transverse steel truss elements are
embedded to the concrete elements by sharing the same nodes. The tensile strength and the
elastic modulus of the concrete were calculated from the measured compressive strength
using the formulas specified in AS3600 (2009) and given in Table 7-12.
The other instantaneous material properties for concrete and steel bars are given as:
= 0.2; Es = 212000MPa; fsy = 566MPa (N10) and 542MPa (N14); and Gf = 92.5N/m.
236
Table 7-12: Calculated tensile strength and elastic modulus of the concrete at the tests.
In the experiment, free shrinkage measurements were initiated 24 hours after casting or 3-4
days after casting. Table 7-13 gives the shrinkage deformations of the 280 × 300 × 350mm
prisms at the tests of each beam and the corresponding shrinkage parameters (Ash and Bsh)
used for the model. As for creep, creep strains were also measured in companion tests on 100
× 100 × 400 mm prisms at a stress level corresponding to 40% of the strength. The
measurement of creep strain was initiated on the 7th day after casting. Figure 7-23 illustrates
the measured creep coefficient versus time. The author used this creep coefficient data and
converted it to the compliance function to derive the solidification creep parameters for the
finite element model. The parameters are taken as: q2 = 75/MPa, q3 = 1.2/MPa, and q4 =
15/MPa. The negative infinite retardation area of the solidification spectrum is taken as A0 =
20.0MPa-1. Ten Kelvin chain units were used and the calculated retardation time and elastic
Figure 7-23: Set-up of creep tests and variation of creep coefficient with time
(Gribniak 2009).
Table 7-13: Shrinkage deformations of 280 × 300 × 350mm prisms (Gribniak 2009) and the
corresponding modeling parameters.
Table 7-14: Kelvin chain properties for solidification creep model for Gribniak’s beams.
E (MPa) (Days)
0.46344 0.0001;
0.31468 0.001
0.21626 0.01
0.15053 0.1
0.10617 1
0.07589 10
0.05495 100
The experimental and numerical mid-span deflection versus mid-span moment responses of
two of the beams are shown in Figure 7-24. Excellent agreement is achieved using the finite
element model. As it is known that with free shrinkage prior to loading, initial deflections
were induced. This initial deflection is very dependent on the amount of shrinkage and the
number of tensile and compressive steel bars. As the beam sections remained uncracked
before loading, shrinkage in the sections with larger compressive reinforcement can induce a
reversed curvature compared with those with little compressive reinforcement, thus the beams
deflections (negative) before loading, even though this initial deflections could be very small
and may be eliminated due to self weight of the specimens (Figure 7-24b). In the test,
shrinkage induced initial curvature or deflection was not measured. Therefore, the initial
negative deflections derived from the finite element models were added to the experimentally
On the fully-cracked sections, since the tensile concrete is ignored, the amount of
By using the method presented in Appendix B, the initial mid-span deflections of each beam
assumed fully-cracked are calculated and the fully-cracked responses are also plotted in
Figure 7-24.
50 Experimental response
50 Experimental response
FE response
Fully-cracked response FE response
40
Moment (kN.m)
40 Fully-cracked response
30 Moment (kN.m) 30
20 20
10 10
Mid-span deflection (mm) Mid-span deflection (mm)
0 0
0 2 4 6 8 10 12 -1 1 3 5 7 9 11
Figure 7-24: Moment versus mid-span deflection response of s-1 and s-1R.
Table 7-15 summarises the cracking moments Mcr obtained from both the experimental and
numerical results for all the eight beams. The finite element model has predicted Mcr with
good accuracy compared with experimental results. It can be clearly seen that the beams with
heavy top bars had increased cracking moment compared to the beams with small top bars.
This can be illustrated by looking at the local responses of the beams. Figure 7-25 shows the
finite element results of stress and strain distributions at the mid-span sections of the first
series of beams. The y axis is the distance from the bottom to the top of the sections, whereas
the x axis is the strain or stress values just before loading. Due to shrinkage and the restraint
to free shrinkage by the reinforcing bars, stress and strain develop at the section. The restraint
forces on the concrete are significantly affected by the amount of reinforcement as known.
240
For the beams heavily reinforced with compressive bars (S-1R and S-2R), the shrinkage
strains on the top levels of the sections are greatly restrained by the reinforcement. Thus larger
restraining tensile force is applied on the concrete on the top levels of S-1R and S-2R,
compared with that of S-1 and S-2. Therefore the stress (strain) distributions of S-1R and S-
2R are actually in reversed directions compared with S-1 and S-2. The shrinkage induced
tensile stress on the bottom sides of the beams is also reduced before loading commenced.
Therefore, s-1R and s-2R exhibited higher cracking moments (Table 7-15).
Table 7-15: Measured and modeled Cracking moments for Gribniak’s beams.
Mcr (kN.m)
Beam
EXP FEM
S-1 16.8 17.0
S-1R 19.8 20.0
S-2 15.9 17.0
S2-R 17.9 21.0
S-3 15.8 16.0
S3-R - 19.0
S-4 13.9 15.0
S-4R 16.0 17.0
241
300 300
Distance from the bottom (mm)
(a) s-1 and s-1R (strain) (b) s-2 and s-2R (strain)
300 300
Distance from the bottom (mm)
200 200
(c) s-1 and s-1R (stress) (d) s-2 and s-2R (stress)
Figure 7-25: Stress and strain distributions at the mid-span sections of Gribniak’s first series
beams just before loading (FEM).
Gribniak’s test, the tension stiffening deflection versus mid-span moment diagrams for the
first series beams are plotted in Figure 7-26. The figure shows that the tension stiffening
deflections in S-1R is slightly greater than that in S-1 after first cracking, whereas little
difference is found between the results of S-2 and S-2R. It seems that even though the
provision of compressive reinforcement can reduce the cracking moment, influence on the
short-term tension stiffening still is not very clear. In chapter 8, a parametric study on the
discussion is made.
Tension stiffening deflection (mm
4 4
s-1 s-2
3 s-1R 3 s-2R
2 2
1 1
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Moment (kN.m) Moment (kN.m)
7.6 SUMMARY
The finite element program incorporated with the proposed bond model has been used to
analyse the specimens tested by the author. Each specimen was decomposed into numbers of
elements, including concrete, steel truss and bond interface elements for the analysis. The
details of the configuration of each finite element model were presented. Shrinkage and creep
of the specimens measured at the tests were carefully examined and modeled, as well as the
The finite element modeling results are presented in two scales, namely the global scale and
the local scale. The global scale results investigated tension stiffening behaviour in terms of
the tension stiffening strain or tension stiffening deflection under both instantaneous loads and
sustained loads. But then the local scale results focused on the mechanisms of tension
stiffening through the concrete-steel interactions. Excellent correlation was achieved between
243
To further verify the accuracy and applicability of the finite element program, the long-term
and short-term tests conducted by Nejadi and Gilbert (2004) and Gribniak (2009) were
modeled by the author. Both of the tests had considered and measured shrinkage and creep.
Again good correlation was achieved. For the long-term slabs tested by Nejadi and Gilbert,
both the experiment and the model predicted a rapid decrease of tension stiffening in the first
15 days by 30%-50%, and then a more gradual decrease to 54% - 6% of the instantaneous
values. The major factor of reducing tension stiffening is attributed to the formation of new
crack due to shrinkage of the concrete and secondary bond deterioration. The testing and
modeling results for Gribniak’s beams show that the provision of compressive reinforcement
can greatly resist the effect of shrinkage on the uncracked beam section and reduce cracking
moments.
In general, the finite element model has been shown to accurately capture the tension
stiffening behavior under both short-term and long-term loading. The model considers the
effects of shrinkage and creep in the concrete, which can significantly reduce the effect of
tension stiffening.
CHAPTER 8 PARAMETRIC STUDY
8.1 OVERVIEW
In the forgoing chapters, it has been demonstrated that the magnitude of tension stiffening at
any time depends on the amount of shrinkage in the concrete before and after cracking. The
numerical model introduced in Chapter 6 has shown to accurately and rationally model the
mechanisms of tension stiffening and the correlation with experimental results is good.
However, the physical experiments conducted by the author are necessarily limited and a
comprehensive laboratory based study of various parameters affecting tension stiffening has
not been possible. For example, the time-dependent tension stiffening of the beams may be
number of numerical specimens are devised and modeled, and a parametric study is
undertaken using the 2-D finite element model described in Chapter 6. Both beam and slab
specimens are considered. The major parameters considered in the parametric study are:
In each series of numerical experiments, only a single parameter is varied and all the other
parameters are kept constant. Both short-term and long-term analyses are undertaken. For the
short-term analysis, shrinkage is allowed to develop prior to loading. For the long-term
analysis, shrinkage is only allowed to develop after first loading, when the loads are sustained
for a prescribed period. The variation of tension stiffening with time of each specimen is then
evaluated in terms of the change in the tension stiffening deflection. It should be remembered
that for a given specimen under consideration, the increase of tension stiffening deflection
indicates the reduction of beam or slab deflection. Therefore the examination of the effects of
these parameters on tension stiffening deflection reflects their effects on member behaviour
under service loads. Results of the parametric studies are discussed and conclusions are drawn
A benchmark specimen is used as a reference for both the beam and slab specimens. The
fcp = 25MPa; fct = 2MPa; Ec = 25000MPa; v = 0.2; Es = 200000MPa; fsy = 500MPa; and Gf =
246
75N/m.
The creep and shrinkage characteristics of the benchmark specimens are given in Tables 8-1.
Table 8-1: Creep coefficient and shrinkage strain (x10-6) for the benchmark specimens.
E (MPa) (Days)
0.348 0.0001;
0.236 0.001
0.162 0.01
0.113 0.1
0.080 1
0.057 10
0.041 100
Shrinkage is assumed to commence at the 14th day and accumulate up to 400×10-6 after 100
days. Accordingly parameters Ash and Bsh are taken as 960×10-6 and 140days for the shrinkage
model (Equation 6.41). The benchmark creep coefficient after 100 days is taken as 2.0. The
40.0/MPa, and A0 = 37.3MPa-1. Table 8-2 summarises the retardation time and elastic
Figure 8-1 shows the beam specimens analysed in the numerical experiments. All the beams
are simply-supported on the top of stiff plates. Each plate is of 100mm length and 250mm
width, located at 100mm from each end of the beam. Each beam is 250mm wide, with a span
247
L of 3m. The effective depth of the tensile reinforcement in the beam is dst = 280mm. The
bottom cover to the tensile reinforcement cb is treated as a variable in the experiment. Some
of the beams also contain compressive reinforcements Asc at the top of the members (with an
effective depth dsc = 30mm) and stirrups over the length of the beams. A uniformly distributed
load q is applied on the top of the beams so that the concrete at the bottom of the beam is in
tension. For the short-term tests, q is increased monotonically until the steel is about to yield.
For the long-term tests, q is sustained for a period of 100 days. The level of the sustained load
In the finite element model, only half the length of the beam is modeled, as shown in Figure
8-2. The uniformly distributed applied load is applied as nodal loads on the concrete elements
on the top of the beams. It is not necessary to adopt too fine a mesh layout. In these numerical
experiments, 25mm by 25mm isoparametric concrete elements are adopted to model the
concrete, together with reinforcing bar truss elements and bond interface elements. The
compressive reinforcing and stirrup truss elements are directly attached to the surrounding
concrete element nodal points. The benchmark beam model consists of 658 nodal points, 552
Figure 8-2: Finite element model of beam specimens in the numerical experiment.
The slab specimens are also simply-supported over a length of 3m, with uniformly distributed
loads imposed on the top of the slabs (Figure 8-3). Each slab has a cross section 400mm wide
and D depth. The effective depth of the tensile reinforcing bars dst is taken as 130mm. The
mesh layout for the slabs is shown in Figure 8-4. The quadrilateral concrete element size is
also taken as 25mm × 25mm. The benchmark slab specimen model consist of 376 nodal
points, 276 concrete elements, 46 tensile reinforcing truss elements and 46 bond interface
elements.
249
Figure 8-4: Finite element model of slab specimens in the numerical experiment.
The various parameters investigated in each series of numerical test are provided in Tables 8-3
and 8-4. The letter (‘A’ to ‘I’) represents the different testing series. Each series contains 2 – 4
specimens. The parameters investigated in each test series are highlighted with the dark
shading in the table. Normally it is specimen No.1 in each series that is defined as the
benchmark specimen. It should be noted that the maximum bond stress (max) alter according
to the various concrete strengths selected in the test series B. The way to modify the values of
Table 8-3: Various parameters for the beam specimens in the numerical experiment.
Table 8-4: Various parameters for the slab specimens in the numerical experiment.
that series can be clearly investigated. The mid-span deflections of the specimens are directly
obtained from the finite element program and the results are used to derive tension stiffening
deflections. The results of tension stiffening and its variation with time are reported in terms
investigated in Series A, which includes 3 beams and 3 slabs under short-term loading, and
the same numbers of beams and slabs under long-term loading. Figure 8-5 shows the moment
versus mid-span deflection responses of the short-term tests. Notice that since the effective
depths of the tensile steel bars for each beam series or slab series are identical, the calculated
fully-cracked responses are also identical. The numerical results show that deformation of the
members occurs before loading due to initial shrinkage strain (-400×10-6). Shrinkage prior to
loading also brings down the cracking moment from Mcr to Mcr.sh (from 18kN.m to 13.4kN.m
for beams and from 7.3kN.m to 6.2kN.m for slabs) for the specimens with shrinkage. The
results indicate that the variation of bottom concrete cover has little effect on Mcr.sh provided
the initial shrinkage for each specimen is the same. It also appears that the short-term mid-
span deflections of the beams or slabs with different concrete cover are quite similar.
Moment (kN.m)
50 30
Moment (kN.m)
40
20
30
C
Cbb == 20mm
20mm C b = 15mm
S-A-S-1
20 C b = 30mm
B-A-S-2 10 C b = 20mm
S-A-S-2
C b = 40mm
B-A-S-3 C b = 25mm
S-A-S-3
10
Figure 8-6 plots the tension stiffening deflection versus mid-span moment responses of the
series A short-term test specimens. Notice that at zero moment, tension stiffening deflection is
greater than zero due to initial shrinkage. The results show that with different bottom concrete
cover, tension stiffening deflections at the same moment level are slightly affected by the
253
bottom concrete cover. The magnitudes of tension stiffening deflections are slightly greater in
the specimens with larger bottom covers. For instance, for the beams specimens, at M =
26.9kN.m, the tension stiffening deflection for the beam specimen with cb = 20mm equals to
1.68mm, which is less than that in the beam specimen with cb = 40mm that equals to 1.94mm.
However, the difference between the beams with cb = 30mm and cb = 40mm is not that
significant as shown in Figure 8-6a. For the slabs specimens, at M = 16.3kN.m, the tension
stiffening deflection changes from 2.45mm to 2.60mm, as cb increases from 15mm to 25mm.
It could be perceived that with even larger bottom cover, the magnitude of tension stiffening
4 7
Tension stiffening deflection (mm
6
3 5
4
2
3
CCb
b = 2 CCb
b =
=15mm
1 =20mm
20mm 20mm
CB-A-S-2
b = 30mm
CS-A-S-2
b = 20mm
1 CS-A-S-3
CB-A-S-3
b = 40mm b = 25mm
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
Mid-span moment (kN.m) Mid-span moment (kN.m)
Figure 8-7 plots the development of long-term mid-span deflections for beams and slabs
under sustained loads in test series A. The beams and slabs are under sustained uniformed
distributed loads, which give rise to a constant mid-span moment M = 25.8kN.m and
10.7kN.m for beams and slabs respectively. Similar to the short-term specimens, the mid-span
deflections for the long-term test specimens do not vary significantly as the concrete cover
254
varies. With time, the development of mid-span deflection of each beam or slab is similar,
with the curves for the slabs are almost the same, as shown in Figure 8-7b. For the beams
specimens Beam 1 and Beam 2 (cb = 20mm and cb = 30mm) are quite similar, whereas beam
4 (cb = 40mm) exhibits greater long-term stiffness under the same applied moment. The finite
element models have given rise to less numbers of primary cracks in Beam 4 than Beam 1 and
Beam 2, which are regarded as the major reason for the improved stiffness in Beam 4.
10 20
Mid-span deflection (mm)
8 16
6 12
4 B-A-L-1
C b = 20mm 8 C b = 15mm
S-A-L-1
B-A-L-2
C b = 30mm C b = 20mm
S-A-L-2
B-A-L-3
C = 40mm S-A-L-3
C b = 25mm
2 b
4
Table 8-5 summarises the mid-span deflections ms(t), and the calculated tension stiffening
deflections ts(t) at different stages under loading (t = 0, 15 days and 100 days). It is found
that by increasing the bottom concrete cover in reinforced concrete beams, the time-dependent
tension stiffening deflections are increased. As shown in the table, for the beam specimens, at
t = 15 days, ts(t) increases from 1.27mm to 1.71mm as the bottom cover increases from
20mm to 40mm. The rate of decrease of tension stiffening with time is also decreased with the
increased bottom cover. For example, for the beam with cb = 20mm, ts(t)/ tsi = 0.64 on t = 15
days and 0.38 on t = 100 days, whereas for the beam with cb = 40mm, the corresponding
255
values are 0.72 and 0.56. However, this phenomenon is not apparent for the slab specimens.
Even though the magnitude of time-dependent tension stiffening deflection ts(t) slightly
increases with increased bottom cover, the rate of decrease of tension stiffening seems
Deflections (mm)
Cb Instantaneous t = 15 days t = 100 days
Specimen
(mm) ts(t)/ ts(t)/
msi tsi ms(t) ts(t) ms(t) ts(t)
tsi tsi
Beam 1 20 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 30 3.61 2.08 5.51 1.34 0.65 8.42 0.82 0.40
Beam 3 40 3.31 2.37 5.14 1.71 0.72 7.91 1.33 0.56
Slab 1 15 6.99 3.61 10.9 2.31 0.64 17.2 1.48 0.41
Slab 2 20 6.79 3.81 10.76 2.45 0.64 17.12 1.56 0.41
Slab 3 25 6.73 3.87 10.73 2.48 0.64 17.13 1.55 0.40
It is generally agreed that concrete cover is an important factor in the development of the
crack pattern and crack spacing. With less bottom cover, more cracks will form and hence
break down the local bond stress at the interface between concrete and steel bars. The
magnitude of tension stiffening will drop off with the breakdown of bond at the cracked
implicitly improved by larger concrete cover. For example, the analytical model have shown
that Beam 3 (cb = 40mm) exhibits less cracks than Beam 1 (cb = 20mm) and Beam 2 (cb =
30mm). Study conducted by Salem, et al (1999) drew the same conclusion in their study of
the effect of concrete cover on tension stiffening of reinforced concrete members. They
concluded that spitting cracks due to lack of concrete cover can greatly affect the bond effect
The test specimens in Series B are assigned with the different concrete tensile strengths fct. A
group of four beams and four slabs was tested under short-term loading, and another group
with the identical number of beams and slabs was subjected to long-term loads. In this case, it
is also necessary to adjust the corresponding peak bond stress max for the modeling input,
since the bond strength is dependent on the concrete strength (max equals to 2.5 times the
square root of fcp ). Thus, max increases with the increased fct. Table 8-6 shows the values of
max for each specimen in this series. In addition, according to the change of fct, the parameters
1 2 and 3 defining the descending branch of Petersson’s equation (Equation 6.10) are also
modified.
Table 8-6: max for the series B test specimens (All in MPa).
Figure 8-8 illustrates the moment versus mid-span deflection responses of the short-term
specimens of the series B tests. An increase of tensile strength obviously leads to increased
cracking moment Mcr and decreased deflections after cracking. Specimens with lower tensile
strength exhibit more cracks (smaller crack spacing), which results in reduced stiffness and
less tension stiffening. Figure 8-9 shows the tensile strain contour of the beams with fct =
2.5MPa and fct = 3.0MPa at M = 30kN.m, where the darkened areas demonstrate the crack
locations (tensile strain greater than the cracking strain). There are 7 primary cracks found in
257
50 30
Moment (kN.m)
40
20
30
f ct = 1.5MPa
S-B-S-1 2.0MPa
f ct = 1.5MPa
B-B-S-1 f ct = 2.0MPa
1.5MPa
20 S-B-S-2
f ct = 2.0MPa
B-B-S-2 10 fS-B-S-3
ct = 2.5MPa
f ct = 2.5MPa
B-B-S-3 fS-B-S-4
ct = 3.0MPa
10 f ct = 3.0MPa
B-B-S-4
Mid-span deflection (mm) Mid-span deflection (mm)
0 0
0 3 6 9 12 15 0 6 12 18 24 30 36
In reality, the elastic modulus Ec and the strength of concrete are normally interdependent
properties. However, this test series tries to focus on the influence of fct and Ec was held
constant for all specimens. Based on the formulas provided in Appendix B, the fully-cracked
response is independent of fct but dependent on Ec, therefore the fully-cracked deflections for
the beams or slabs are identical. The fully-cracked deflections of each specimen were
calculated, as well as the tension stiffening deflections. The results for each short-term test
specimen are plotted in Figure 8-10. The magnitude of tension stiffening deflection is
obviously greater in the specimen with largest tensile strength. For instance, at a typical post-
cracking moment M = 16.3kN.m, tension stiffening deflection ts = 2.45mm, 2.96mm and
258
3.89mm for the slabs with fct = 2.0MPa, 2.5MPa and = 3.0MPa, respectively.
5 8
fB-B-S-1
ct = 1.5MPa fS-B-S-1
ct = 1.5MPa
2.0MPa
4
fB-B-S-2
ct = 2.0MPa
6 fS-B-S-2
ct = 2.0MPa
1.5MPa
fB-B-S-3
ct = 2.5MPa fS-B-S-3
ct = 2.5MPa
3
fB-B-S-4
ct = 3.0MPa fS-B-S-4
ct = 3.0MPa
4
2
2
1
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
Moment (kN.m) Moment (kN.m)
Figure 8-11 shows the development of mid-span deflections with time for the long-term
specimens in series B. The specimens were subjected to the same amount of loading as those
in Series A long-term tests. The specimens with smaller tensile strength undergo larger
deformation with time. Table 8-7 tabulates the mid-span deflections and tension stiffening
deflections for the long-term specimens at instantaneous loading, 15 days and 100 days after
first loading. The mid-span deflections by the end of the test are about 2 - 3 times of the
instantaneous deflections. With larger tensile strength, the proportion of final deflection to
instantaneous deflection ms(t)/ msi is also larger. For instance, the values of ms(t) / msi (t =
100 days) for Beam 1, Beam 3 and Beam 4 (fct = 2.0MPa, 2.5MPa and 3.0MPa) are 2.30, 2.73
and 3.48, respectively. This is attributed to the fact that at the instantaneous moment (M =
25.8kN.m), the specimen with the larger tensile strength suffers less cracking and thus the
this reduces the stiffness of the member and increases the long-term deflections significantly.
259
10 20
Mid-span deflection (mm)
6 12
f ct = 1.5MPa fS-B-L-1
ct = 1.5MPa
B-B-L-1
4 f ct = 2.0MPa 8 fS-B-L-2
ct = 2.0MPa
B-B-L-2
fS-B-L-3
ct = 2.5MPa
f ct = 2.5MPa
B-B-L-3
2 4 fS-B-L-4
ct = 3.0MPa
f ct = 3.0MPa
B-B-L-4
Time (days) Time (days)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Deflections (mm)
Instantaneous t = 15 days t = 100 days
Specimens fct (MPa)
ts(t)/ ts(t)/
msi tsi ms(t) ts(t) ms(t) ts(t)
tsi tsi
Beam 1 2.0 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 1.5 4.60 1.08 6.33 0.52 0.48 8.92 0.32 0.30
Beam 3 2.5 2.88 2.80 4.96 1.89 0.68 7.88 1.36 0.49
Beam 4 3.0 1.91 3.77 3.71 3.14 0.83 6.65 2.59 0.69
Slab 1 2.0 6.99 3.61 10.9 2.31 0.64 17.2 1.48 0.41
Slab 2 1.5 7.90 2.70 11.72 1.49 0.55 17.91 0.77 0.29
Slab 3 2.5 6.34 4.26 10.38 2.83 0.66 16.65 2.03 0.48
Slab 4 3.0 5.52 5.08 9.52 3.70 0.73 15.76 2.92 0.58
With regard to the change of tension stiffening deflection with time, after t = 15 days or 100
days, both the magnitude of tension stiffening deflection ts(t) and the ratio of time-dependent
the tensile strength increases. For example at t = 100 days, the values of ts(t) for Beam 1,
Beam 3 and Beam 4 (fct = 2.0MPa, 2.5MPa and 3.0MPa) are 0.75mm, 1.36mm and 2.59mm,
respectively. Meanwhile, the values of and ts(t)/tsi for these specimens are 0.38, 0.49 and
0.69, respectively. Therefore, the specimen with larger tensile strength exhibits greater tensile
stiffening deflection and slower rate of change in tension stiffening with time. Note that in this
260
study, the tensile strength of concrete is treated as a constant while in reality it is also a time-
dependent parameter, which normally increases with time. It is reasonable to assume that the
Series C consists of specimens designed with different values for the elastic modulus of
concrete Ec. Four beams and four slabs are chosen for both the short-term and long-term
numerical experiments. The design details are given in Tables 8-3 and 8-4.
Figure 8-12 shows the moment versus mid-span deflection responses for the short-term
specimens in series C. Before loading, the initial deflections of each specimen caused by
shrinkage warping vary. The specimens with larger concrete elastic modulus exhibit slightly
smaller initial deflections. For the short-term beams and slabs under consideration, the initial
Beams: 0.66mm (Ec = 15000MPa), 0.55mm (Ec = 25000MPa), 0.49mm (Ec = 35000MPa),
Slabs: 1.50mm (Ec = 15000MPa), 1.24mm (Ec = 25000MPa), 1.11mm (Ec = 35000MPa) and
concrete. Namely the larger is the elastic modulus of concrete, the smaller is the initial
curvature or the initial deflection. After cracking, the deformation responses of the beam or
261
slab specimens with Ec are shown in Figure 8-12. The slope of the post-cracking response
increases as Ec increases, with the specimens with Ec = 15000MPa clearly showing the
50 30
Moment (kN.m)
40
20
30
E c = 15000MPa
BES1 ESES1
c = 15000MPa
20 ESES2
E c = 25000MPa
BES2 c = 25000MPa
10
E c = 35000MPa
BES3 ESES3
c = 35000MPa
10 E c = 45000MPa ESES4
c = 45000MPa
BES4
Mid-span deflection (mm) Mid-span deflection (mm)
0 0
0 3 6 9 12 15 0 6 12 18 24 30 36
Since the fully-cracked response is dependent on the magnitude of concrete elastic modulus,
in this series, the fully-cracked deflections vary among different specimens. Figure 8-13 plots
the tension stiffening deflection versus mid-span moment responses for the short-term
specimens considered herein. For the beam specimens, the influence of concrete elastic
modulus on the quantity of short-term tension stiffening deflection can hardly be told. Beam 3
and Beam 4 (Ec = 35000MPa and 45000MPa) show similar amount of tension stiffening,
while both Beam 1 and Beam 2 (Ec = 25000MPa and 15000MPa) exhibit slightly greater
tension stiffening deflection than the formers. The slabs also show a similar phenomenon.
Tension stiffening deflections in Slab 3 and Slab 4 (Ec = 35000MPa and 45000MPa) are quite
similar, while Slab 1 and Slab 2 (Ec = 25000MPa and 15000MPa) possess greater values of
tension stiffening deflections. Note that this difference only becomes significant as the
reinforcing steel are about to yield. Before the applied moment is greater than 2 – 3 times of
262
the cracking moment, the effect of concrete elastic modulus on short-term tension stiffening
4 8
EBES1
c = 15000MPa
EBES2
c = 25000MPa ESES1
c = 15000MPa
3 EBES3 6 ESES2
c = 25000MPa
c = 35000MPa
EBES4
c = 45000MPa ESES3
c = 35000MPa
4 ESES4
c = 45000MPa
2
1 2
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
Moment (kN.m) Moment (kN.m)
For the long-term specimens in this series, all the beams and slabs were initially loaded to
25.8kN.m and 10.7kN.m (moment at mid-span), respectively, and held constant with time.
Figure 8-14 shows the development of mid-span deflections during the loading history for the
long-term beams and slabs in this series. No significant difference of mid-span deformations
with time is observed for the specimens with the concrete elastic modulus ranging from
of Beam 2 and Slab 2 (Ec = 15000MPa) are slightly larger than the other specimens.
Table 8-8 gives the time-dependent tension stiffening deflections of the long-term test
specimens in this series. With increased concrete elastic modulus, the magnitude and rate of
change of the time-dependent tension stiffening deflection decreases. For instance, ts(t)/ tsi
for Slab 1 (Ec = 25000MPa), Slab 3 (Ec = 35000MPa) and Slab 4 (Ec = 45000MPa) at the age
t = 15 days are 0.64, 0.47 and 0.29, respectively. The time-dependent loss of bond stress and
263
tensile stress in the concrete at the level of reinforcing steel does not directly depends on the
elastic modulus of concrete Ec. However, the magnitude and the rate of change of tension
stiffening is related to Ec, because the fully-cracked response is highly affected by Ec. The
time-dependent fully-cracked response is weaker in the specimens with smaller Ec, therefore
20
Mid-span deflection (mm)
10
6 12
EBEL1 SEL1
E c = 15000MPa
c = 15000MPa
4 8 SEL2
E c = 25000MPa
EBEL2
c = 25000MPa
SEL3
E c = 35000MPa
EBEL3
c = 35000MPa
2 4 SEL4
E c = 45000MPa
EBEL4
c = 45000MPa
Deflections (mm)
Instantaneous t = 15 days t = 100 days
Specimens Ec (MPa)
ts(t)/ ts(t)/
msi tsi ms(t) ts(t) ms(t) ts(t)
tsi tsi
Beam 1 25000 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 15000 4.20 2.10 5.90 1.60 0.76 8.79 1.20 0.57
Beam 3 35000 3.60 1.75 5.53 0.82 0.47 8.42 0.58 0.34
Beam 4 45000 3.38 1.78 5.59 0.52 0.29 8.41 0.28 0.16
Slab 1 25000 6.99 3.61 10.9 2.31 0.64 17.2 1.48 0.41
Slab 2 15000 8.72 3.49 12.26 2.37 0.68 18.5 1.85 0.53
Slab 3 35000 6.51 3.39 10.66 1.64 0.48 16.93 1.27 0.37
Slab 4 45000 6.08 3.34 10.35 1.53 0.46 16.79 0.81 0.24
264
The effect of the quantity of tensile reinforcement Ast on tension stiffening is investigated by
the Series D tests. There are 3 beams and 3 slabs modeled in both the short-term tests and
The global responses of the short-term tests are plotted in Figure 8-15. With the same amount
of initial creep and shrinkage in the concrete, little difference of stiffness is found before first
cracking of the members. Only slight variation of initial deflections due to shrinkage warping
is found (Figure 8-15), which indicates that shrinkage warping is dependent on the amount of
reinforcement, because the restraining force at the level of the steel bars is proportional to the
reinforcement area. Therefore, shrinkage warping on the uncracked beams or slabs is more
significant for members with larger reinforcement area. Hence the greater the tensile
reinforcement area is, the larger the initial deflection prior to loading. For example, the initial
mid-span deflections for the beams with Ast = 400mm2, 309mm2 and 226mm2 are 0.55mm,
After cracking, the stiffness in the members with less tensile reinforcement is obviously
smaller than the stiffness of members with larger tensile reinforcement area. The flatten out of
the moment deflection responses for Beam 3 and Slab 2 as the moment increases is caused by
Moment (kN.m)
Moment (kN.m)
50 30 2
AS-C-S-1
st = 314mm
2
AS-C-S-2
st = 452mm
40 2
AS-C-S-3
st =603mm
20
30
2
B-C-S-1
A st = 400mm
20
B-C-S-2
A st = 309mm
2 10
2
10 B-C-S-3
A st = 226mm
Figure 8-16 shows the tension stiffening deflection versus mid-span moment response for the
short-term tests in series D. It appears that with decreased tensile reinforcement area Ast (or
reinforcing ratio ȡ), the magnitude of tension stiffening deflection is more significant both
before and after cracking. This is also true in terms of the contribution of tension stiffening to
the total stiffness of the member. For instance, at M = 26.9kN.m the values of fc/ms (ratio
between fully-cracked deflection and mid-span deflection) for the beams with ȡ = 0.005,
0.004 and 0.003 are 1.52, 1.67 and 1.73, respectively. Therefore, in a member with small
reinforcement ratio, such as slab, the contribution of tension stiffening to the total member
stiffness is more significant, and the importance of accurately modeling tension stiffening in
deflection calculation increases. Besides, the figure shows that at the time when yielding is
about to occur, tension stiffening deflection tends to drop off dramatically. The short-term slab
5 2 8
A st = 400mm
B-C-S-1
2 2
A st = 309mm
B-C-S-2 S-B-S-1
A st = 314mm
4 2 2
B-C-S-3
A st = 226mm 6 S-B-S-2
A st = 452mm
2
A st =603mm
S-B-S-3
3
4
2
2
1
0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30
Mid-span moment (kN.m) Moment (kN.m)
Figure 8-17 shows the development of the long-term deflections of the series D specimens
under constant sustained loads (with mid-span moment M = 25.8kN.m for beams and M =
10.7kN.m for slabs) for a period of 100 days. Similar to the short-term deflections, the long-
terms deflections are greater in the members with smaller tensile reinforcement areas.
The calculated tension stiffening deflections ts(t) for the long-term specimens in this series
are given in Table 8-9. It seems that the magnitudes of ts(t) are also greater in the specimens
with smaller reinforcing ratio, either after t = 15 days or t = 100 days. For example, after t =
15 days ts(t) equals 2.94mm, 2.31mm and 1.82mm for the slabs with ȡ = 0.005, 0.0075 and
0.01, respectively. However, the proportion of ts(t) to tsi is not as affected by the
reinforcement ratio, as seen in the table. As discussed in Chapter 7, the experimental results of
the long-term slabs tested by Nejadi and Gilbert (2004) also demonstrated that the rate of
change of tension stiffening is not significantly affected by the reinforcing ratio. The proposed
bond model only relates the long-term decay of bond stress to the quantity of shrinkage and
the level steel stress, but not the reinforcement area. However, the magnitude of local steel
stress is also dependent on the reinforcement area. At the same level of bending moment M,
267
the sections in the members with larger reinforcement areas have smaller steel stresses, which
can lead to a smaller rate of change of bond stress with time (Equation 6.65). In addition, with
larger reinforcement area, the shrinkage induced compressive force on the reinforcing bars is
also larger. Thus if the crack spacing in a specimen has already stabilised under instantaneous
loads, the effect of reinforcement area on the rate of change of tension stiffening with time is
12 25
Mid-span deflection (mm)
8
15
6
2 2
A st = 400mm
B-C-L-1 10 A st = 314mm
S-C-L-1
2 2
4 A st = 309mm
B-C-L-2 A st = 452mm
S-C-L-2
2 2
A st = 226mm
B-C-L-3 5 S-C-L-3
A st =603mm
2
Time (days) Time (days)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Deflections (mm)
Instantaneous t = 15 days t = 100 days
Specimen ȡ
ts(t)/ ts(t)/
ms tsi ms ts(t) ms ts(t)
tsi tsi
Beam 1 0.005 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 0.004 4.75 2.32 6.69 1.58 0.68 9.71 1.04 0.45
Beam 3 0.003 5.93 3.26 8.50 1.99 0.61 11.86 1.23 0.38
Slab 1 0.0075 6.99 3.61 10.9 2.31 0.64 17.2 1.48 0.41
Slab 2 0.005 9.63 4.49 13.97 2.94 0.66 20.61 1.98 0.44
Slab 3 0.01 5.52 3.01 9.23 1.82 0.60 15.33 1.08 0.36
268
The numerical models in Series E were established to look into the significance of tensile bar
diameter db on time-dependent tension stiffening. There are only 2 beams and 2 slabs that are
modeled for both the long-term and short-term tests in this series. According to Table 8-3 and
8-4, the beams or slabs are designed with the similar amount of tensile reinforcement area but
different diameters of bars. Therefore, the value of c for the bond interface elements in
6
Moment (kN.m)
50
S-D-S-1
412
214
5
40 410
610
S-D-S-2
4
30
3
20 214
B-D-S-1
410
2
B-D-S-2
10 1
Mid-span deflection (mm)
0 0
0 3 6 9 12 15 18 0 10 20 3
Mid-span moment (kN.m
The short-term global moment versus mid-span deflection responses are shown in Figure 8-15.
The results indicate that there is very slight difference for the members reinforced with bars of
different diameters, either before cracking or after cracking. Since the reinforcement area and
the depth for the reinforcing bars are similar for each pair of specimens, the fully-cracked
As shown in Figure 8-19, the quantity of short-term tension stiffening deflection also exhibits
little difference between the members with different bar diameters. It can be concluded by the
269
numerical test results that the bar diameter has a very minor effect on the quantity of short-
term tension stiffening in reinforced concrete beams or slabs. This observation is similar to
5 6
B-D-S-1
214 S-D-S-1
412
214
4 5
410
B-D-S-2 410
610
S-D-S-2
4
3
3
2
2
1 1
0 0
0 10 20 30 40 50 60 0 10 20 30
Mid-span moment (kN.m) Mid-span moment (kN.m)
Similar responses are also found in the results of long-term tests in this series. The time-
dependent tension stiffening deflections ts(t) seems not to be influenced by bar diameters,
both in terms of its magnitude and its proportion of the instantaneous tension stiffening
12 25
10 20
8
15
6
214
B-D-L-1
10 412
S-D-L-1
410
B-D-L-2
4 610
S-D-L-2
5
2
Time (days) Time (days)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Deflections (mm)
No. of Instantaneous t = 15 days t = 100 days
Specimen db
bars ts(t)/ ts(t)
ms tsi ms ts(t) ms ts(t)
tsi / tsi
Beam 1 2 14 4.63 2.43 6.72 1.55 0.64 9.73 1.02 0.42
Beam 2 4 10 4.49 2.57 6.68 1.59 0.62 9.66 1.09 0.43
Slab 1 4 12 6.99 3.61 10.9 2.31 0.64 17.2 1.48 0.41
Slab 2 6 10 6.62 3.67 10.6 2.23 0.61 16.88 1.46 0.40
reinforced concrete beams. On an uncracked section in bending, tensile creep increases the
tensile stress in the steel bars and reduces the tensile stress in the concrete, hence reducing
tension stiffening. In this test series, four beams and four slabs are considered for both the
short-term and long-term tests, while the same numbers of specimens for the long-term tests.
The only variable parameter is the creep coefficient ijc. In order to adjust the creep
solidification model to obtain the desired ijc, only the value of q4 needs to be varied. For
example, q4 for B-E-S-1 (ijc = 2.0) is taken as 40.0/MPa, and for B-E-S-2 (ijc = 0.5) as
Figure 8-21 shows the global moment versus deflection responses for the short-term
specimens in series F. In this case, creep only occurs together with shrinkage prior to loading.
For the short-term beams and slabs under consideration, the initial deflections before loading
are:
Beams: 0.43mm (ijc = 0.5), 0.47mm (ijc = 1.0), 0.51mm (ijc = 1.5), and 0.55mm (ijc = 2.5);
Slabs: 0.96mm (ijc = 0.5), 1.05mm (ijc = 1.0), 1.15mm (ijc = 1.5) and 1.24mm (ijc = 2.5).
Moment (kN.m)
Moment (kN.m)
50 30
c = 0.5
S-E-S-2
c = 1.0
S-E-S-3
40
c = 1.5
S-E-S-4
20 c = 2.0
30 Series1
c = 0.5
B-E-S-2
c = 1.0
B-E-S-3
20 c = 1.5
B-E-S-4
10
c = 2.0
Series2
10
Mid-span deflection (mm) Mid-span deflection (mm)
0 0
0 3 6 9 12 15 18 0 6 12 18 24 30 36
It seems that with same amount of shrinkage, increased initial deflections can be induced by
increasing the quantity of creep in the concrete, even though its effect is not very significant.
This is because the instantaneous stress and strain at zero moment is quite small and only
gradually accumulates with increasing shrinkage. The redistribution of stress and strain on the
uncracked sections due to creep is much less significant compared with that caused by
shrinkage. Therefore, the short-term responses of the members are quite similar both before
Figure 8-22 plots tension stiffening deflection versus mid-span moment responses for the
short-term members in series F. In this case, on the fully-cracked sections, only the concrete
in the compression zone carries small stresses and tensile creep is not applicable. Therefore
the initial deflections of the fully-cracked response for the members with various creep
coefficients are similar. The values of instantaneous tension stiffening deflections for the
c = 0.5
S-B-S-1 5 c = 0.5
S-B-S-1
3 c = 1.0
S-B-S-2 c = 1.0
S-B-S-2
4
S-B-S-3
c = 1.5 c = 1.5
S-B-S-3
S-B-S-4
c = 2.0
3
S-B-S-4
c = 2.0
2
2
1
1
0 0
0 10 20 30 40 50 60 0 10 20 30
Moment (kN.m) Moment (kN.m)
Figure 8-23 shows the mid-span deflection versus time responses of series F long-term
specimens. The beams and slabs were loaded up to 25.8kN.m and 10.7kN.m then held
constant for 100 days. Clearly the members with increased creep during the sustained loading
also exhibit increased long-term deflections. However, the amount of calculated time-
dependent tension stiffening deflections ts(t) as given in Table 8-11 seem to be little affected
by concrete creep. The increase of actual deflection due to creep is largely due to the increase
of curvatures on the cracked sections. Despite the fact that the relative increase of curvature
on a cracked section due to creep is less than that on an uncracked section, the magnitude of
curvature on the cracked section is much larger and dominates. On the uncracked sections in-
273
between the cracks, creep increases the depth of compressive zone and reduce the area of
concrete in tension, meanwhile the total tensile force increases due to the shortening of lever
arm. This may leads to an increase of tensile stress in the concrete and in the tensile steel bars.
However bond deterioration and tensile creep in the concrete break down the concrete tensile
stress with time which cancel out the effect of compressive creep. Therefore, the quantity of
creep has only a very minor impact on the long-term behaviour of tension stiffening.
25 c = 0.5
S-E-L-2
12 c = 0.5
B-E-L-2
Mid-span deflection (mm)
c = 1.0
B-E-L-3 Mid-span deflection (mm) c = 1.0
S-E-L-3
c = 1.5
S-E-L-4
10 c = 1.5
B-E-L-4 20
c = 2.0
Series2
c = 2.0
Series2
8
15
6
10
4
5
2
Time (days)
Time (days)
0
0
0 20 40 60 80 100
0 20 40 60 80 100
Deflections (mm)
Instantaneous t = 15 days t = 100 days
Specimen ijc
ts(t)/ ts(t)
msi tsi ms ts(t) ms ts(t)
tsi / tsi
Beam 1 2.0 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 0.5 3.77 1.92 4.91 1.34 0.70 6.90 0.87 0.45
Beam 3 1.0 3.69 2.00 5.21 1.24 0.62 7.52 0.76 0.38
Beam 4 1.5 3.80 1.89 5.46 1.18 0.63 8.06 0.70 0.37
Slab 1 2.0 6.99 3.61 10.90 2.31 0.64 17.2 1.48 0.41
Slab 2 0.5 7.02 3.58 9.47 2.40 0.67 13.9 1.33 0.37
Slab 3 1.0 7.03 3.57 10.02 2.32 0.65 15.1 1.33 0.37
Slab 4 1.5 6.95 3.66 10.44 2.33 0.64 16.1 1.40 0.38
274
As previously mentioned, shrinkage has a significant effect on both the short-term and long-
term behaviour of tension stiffening. Series G of the numerical experiments examines this
effect. Four identical beams and four identical slabs are chosen for both the short-term and
long-term numerical tests, with the only difference as the level of shrinkage. The design
details are given in Table 8-3 and 8-4. The numerical way to modify the values of shrinkage
strain in the finite element models consists in replacing the values of Ash in Equation 6.41. For
example, the final shrinkage strains for Beam 1 and Beam 2 are -400 and -200 , the
As for the short-term analysis, since shrinkage only occurs before loading and bond
deterioration due to shrinkage is not significant, the long-term decay parameter of bond stress
Ȝ4 in Equation 6.64 is not applicable (taken as 1.0) in the finite element models. The global
moment versus deflection responses are plotted in Figure 8-24. The curves in the figure are
generally parallel with each other. At the same moment level, the member with greater initial
in each specimen. An increase of shrinkage prior to loading results in a larger initial deflection
and a smaller cracking moment. The initial deflections of the short-term beams and slabs in
Beams: 0.30mm (sh = -200 ), 0.42mm (sh = -300 ), 0.55mm (sh = -400 ), and 0.67mm
Slabs: 0.62mm (sh = -200 ), 0.93mm (sh = -300 ), 1.24mm (sh = -400 ), and 1.55mm
Moment (kN.m)
50
Moment (kN.m)
30
40
20
30
B-F-S-2
sh = 200
S-F-S-2
sh = 200
B-F-S-3
sh = 300
20 S-F-S-3
sh = 300
sh = 400
Series2 10 Series1
sh = 400
Series4
sh = 500
sh = 500
Series4
10
Mid-span deflection (mm) Mid-span deflection (mm)
0 0
0 3 6 9 12 15 18 0 6 12 18 24 30 36
The fully-cracked responses after loading are also parallel with each other as the other
parameters (such as elastic modulus) remain the same during the short-term loading period.
Figure 8-25 plots the tension stiffening deflections at different loading levels for the series G
short-term specimens. The results demonstrate that before cracking, the magnitude of tension
stiffening of the member is increased with increasing initial shrinkage. However, as the
moment increases and cracks start to form, the significance of initial shrinkage on tension
stiffening deflection becomes less obvious. At the descending branches of the curves in Figure
8-25, it appears that tension stiffening deflection is in fact greater in the members with less
initial shrinkage.
276
4 6
shc=
= 200
0.5
Series1 5 shc=
= 200
0.5
Series1
3 shc=
= 300
1.0
Series2 shc=
= 300
1.0
Series2
4 sh
shc=
= 400
1.5
Series3 c=
= 400
1.5
Series3
Series4
shc=
= 500
2.0 Series4
shc =
= 500
2.0
2 3
2
1
1
0 0
0 10 20 30 40 50 60 0 10 20 30
Moment (kN.m)
Moment (kN.m)
The analytical results indicate that for the specimens with significant initial shrinkage, such as
Beam 4 and Slab 4 (İsh = -500 ), tension stiffening and its contribution to the total stiffness
of the specimens are less than the others at higher bending moments. A careful examination of
the tensile strain contours (Figure 8-26a) shows that these specimens actually have more
primary cracks formed, which leads to a further decrease of tension stiffening. This is because
the restraint to initial shrinkage by the steel bars has induced significant tensile stress on the
concrete at the steel bar level before loading. As illustrated by Figure 8-26b, the maximum
tensile stress at the mid span section of Slab 4 (due to shrinkage and self-weight) prior to
loading is 1.75MPa, which is quite close to the cracking stress (2.0MPa). It is possible that
with more shrinkage, cracking can even occur before loading. Comparatively the maximum
120
Slab 2 Slab 4
40
(a) Tensile strain contours at 20.8kN.m (b) Concrete stress at mid-span before loading
Figure 8-26: Mid-span deflection versus time responses of series G long-term specimens.
It has been mentioned that shrinkage is the major factor for the long-term increase of
deflection and loss of tension stiffening. Figure 8-27 shows the development of mid-span
deflections of series F long-term specimens. The specimens were subjected to the same
constant loads as in series E tests. The mid-span deflections are very dependent on the total
amount of shrinkage developed during the sustained loading period. The larger the quantity of
shrinkage is, the greater the final deflection. The decay of long-term tension stiffening of the
specimens is given in Table 8-12. The decrease in tension stiffening deflections ts(t) depends
on the quantity of shrinkage. For instance, ts(t) of Beam 1 (İsh = -400 ), eventually drops to
0.38 of its instantaneous value tsi, while for Beam 4 (İsh = -500 ) ts(t) / tsi drops to 0.26.
When the developed shrinkage strain is less than -400 , the effect of shrinkage on the decay
of tension stiffening is not so great. This is because with sh = -500 , more primary cracks
form by the end of the tests, which leads to significant decrease of tension stiffening with time.
Since the proposed bond model relates the bond stress to shrinkage, bond stress decreases
more rapidly in the specimens with greater amount of shrinkage. Thus tension stiffening may
278
B-F-L-2
sh = 200 25 sh = 200
S-F-L-2
12
Mid-span deflection (mm)
Deflections (mm)
Instantaneous t = 15 days t = 100 days
Specimen sh ()
ts(t)/ ts(t)/
msi tsi ms ts(t) ms ts(t)
tsi tsi
Beam 1 -400 3.69 1.99 5.58 1.27 0.64 8.49 0.75 0.38
Beam 2 -200 3.65 2.03 5.48 1.17 0.58 7.63 0.83 0.41
Beam 3 -300 3.72 1.97 5.54 1.20 0.61 8.06 0.79 0.40
Beam 4 -500 3.81 1.87 5.82 1.10 0.59 9.14 0.49 0.26
Slab 1 -400 6.99 3.61 10.90 2.31 0.64 17.20 1.48 0.41
Slab 2 -200 6.86 3.74 10.46 2.35 0.63 15.06 1.88 0.50
Slab 3 -300 7.12 3.48 10.85 2.25 0.65 16.16 1.68 0.48
Slab 4 -500 7.13 3.47 11.37 2.04 0.59 18.65 0.90 0.26
The analysis on the experimental data provided by Gribniak (2009) in Chapter 7 shows that
the provision of compressive reinforcement can reduce the tensile stress due to shrinkage in
the tensile concrete and hence reduce the reduction in the cracking moment caused by early
specimens under short-term loading. The beams in series H were designed with different
quantities of compressive reinforcement Asc. The details of the design parameters are given in
Table 8-3. Four of the specimens were under short-term loading after being dried for 100 days
(to allow shrinkage strain to accumulate to -400×10-6 and the other four were under sustained
50
Initial
40 Specimen deflections
(mm)
30 Beam 1 0.24
2
Beam 1 (A sc = 226mm )
B-G-S-1
2
20 Beam 2 (A sc = 402mm )
B-G-S-2 Beam 2 0.05
2
B-G-S-3
Beam 3 (A sc = 509mm )
10 B-G-S-4
Beam
2
4 (A sc = 628mm ) Beam 3 -0.05
Figure 8-28: Instantaneous moment verse mid-span deflection responses of series H short-
term specimens.
Figure 8-28 demonstrates the moment versus mid-span deflection responses for the short-term
tested specimens in series H. Test specimens with heavy compressive reinforcement (for
example Beam 4, Asc = 628mm2) exhibit less initial deflection (or even upwards deflections)
and larger cracking moments. Furthermore, the post-cracking deflections of these specimens
are also smaller compared with the ones reinforced with light compressive steels. The method
to predict short-term deflection of beams in the latest revision of Australia Standard AS3600-
2009 has taken into account the influence of compressive reinforcement by reducing the
Figure 8-29 demonstrates the tension stiffening deflection versus mid-span moment responses
for the short-term specimens in this series. Before cracking, the magnitude of tension
stiffening deflection varies for the beams with different compressive reinforcement area.
However, there is no regular pattern that can be observed to see how tension stiffening
difference of tension stiffening deflection among the specimens tends to be less significant, as
4 Beam 1 (A sc = 226mm )
2
B-G-S-1
sh = -400
2
Beam 2 (A=sc-200
B-G-S-2 = 402mm )
sh
2
3 Beam 3 sh
B-G-S-3 (A=- = 509mm )
sc 300
2
B-G-S-4
Beam 4 sh
(A=- = 628mm )
sc 500
0
0 10 20 30 40 50 60
Mid-span moment (kN.m)
Figure 8-29: Tension stiffening deflection verse mid-span moment responses of series H
short-term specimens.
Figure 8-30 shows that the long-term deflections of the beams and how they are affected by
the provision of compressive reinforcement. The beams and slabs were under the same
constant loads as those in the previous test series. It can be seen that the long-term
attributed to the fact that compressive reinforcement can restrain creep in the compressive
281
zone and reduces the effect of shrinkage warping, and hence increases the time-dependent
stiffening of the beams. Table 8-13 demonstrates that with increased quantities of compressive
steel, the loss of tension stiffening deflection with time is reduced. This is especially the case
for Beam 3 and Beam 4. Since the top compressive reinforcement can help reduce the tensile
stress developed on the bottom of the beam due to shrinkage, the shrinkage-induced cracking
is also reduced. Therefore, the loss of tension stiffening in the long-term is less significant in
these specimens.
2
Beam 1 (A sc = 226mm )
B-G-L-1
12
Mid-span deflection (mm)
2
Beam 2 (A sc = 402mm )
B-G-L-2
2
10 B-G-L-3
Beam 3 (A sc = 509mm )
2
B-G-L-4
Beam 4 (A sc = 628mm )
8
2
Time (days)
0
0 20 40 60 80 100
Figure 8-30: Mid-span deflection versus time responses of series H long-term specimens.
Deflections (mm)
Asc Instantaneous t = 15 days t = 100 days
Specimens
(mm2) ts(t)/ ts(t)/
msi tsi ms ts(t) ms ts(t)
tsi tsi
Beam 1 226 3.56 2.01 5.17 1.29 0.64 7.41 0.71 0.35
Beam 2 402 3.58 1.92 4.94 1.32 0.69 6.75 0.83 0.43
Beam 3 509 3.47 1.99 4.71 1.46 0.73 6.28 1.05 0.53
Beam 4 628 3.57 1.85 4.71 1.35 0.73 6.18 0.90 0.49
282
tension stiffening is examined in the Series I tests. The idealization of the transverse
reinforcement in the finite element models is the one-dimensional truss element that shares
the same nodes with the connected concrete elements. The developments of mid-span
deflections in both the short-term and long-term models are shown in Figure 8-31 and 8-32. It
seems that the influence of transverse reinforcement on the deflections on the beams is very
minimal. Since the calculated fully-cracked response of the beams is only dependent of the
area of tensile and compressive reinforcement, the difference of tension stiffening deflections
among the beams is small. It may be concluded that the inclusion of transverse reinforcement
does not induce any unpredictable effects on the behaviour of time-dependent tension
stiffening.
Moment (kN.m)
50
40
30
B-H-S-1
Beam 1 (No stirrup)
20 B-H-S-2
Beam 2 (2 8@100)
B-H-S-3
Beam 3 (2 10@100)
Beam 4 (2 12@100)
B-H-S-4
10
Figure 8-31: Instantaneous moment verse mid-span deflection responses of series I short-
term specimens.
283
12
2
Time (days)
0
0 20 40 60 80 100
Figure 8-32: Mid-span deflection versus time responses of series I long-term specimens.
8.4 SUMMARY
The effects of a variety of parameters on time-dependent tension stiffening have been
investigated by means of numerical experiments conducted with the finite element program.
Night series of tests have been undertaken, with one specific parameter being valued and
investigated in each series. The finite element program is able to predict the tension stiffening
effect and is ideal to generate the large amount of numerical results presented herein. Tension
stiffening has been conveniently represented by the tension stiffening deflection, as the
difference between the calculated mid-span deflection and the calculated fully-cracked
deflection.
It has been concluded that certain parameters, such as concrete tensile strength, concrete
elastic modulus, tensile reinforcement area and compressive reinforcement area, can
substantially affect the magnitude of tension stiffening. The magnitude of tension stiffening is
larger in members with less tensile reinforcement area or reinforcing ratio, but its rate of
change under long-term loading seems little affected by this parameter. Tension stiffening is
284
also sensitive to the concrete tensile strength. The greater the concrete strength is, the greater
the contribution of concrete to the global stiffness of the members. Moreover, the long-term
rate of change of tension stiffening decreases as the tensile strength of concrete increases. It
has been concluded that a increase in concrete elastic modulus increases the stiffness of a
reinforced concrete specimen and the rate of change of tension stiffening for the specimens
with larger concrete elastic modulus under long-term loads is greater. The provision of
compressive reinforcement can reduce the deflection of the structure by reducing the effect of
shrinkage warping on the structure. It also reduces the maximum shrinkage-induced tensile
stress that can cause further cracking and helps resist the decay of tension stiffening with time.
Tension stiffening can also be improved by providing larger bottom concrete cover. The study
demonstrates that the rate of change of tension stiffening deflection in a reinforced concrete
beam decreases as the bottom cover increases, although this effect was not observed in the
case of slabs.
The influence of tensile bar diameter and transverse reinforcement on time-dependent tension
stiffening is not significant. The numerical results indicate that very minimal change of
tension stiffening can be observed with either a change of tensile bar diameter or the amount
of transverse reinforcement.
Creep and shrinkage are the main parameters that vary with time. Shrinkage is the most
significant parameter that affects time-dependent tension stiffening. Shrinkage warping of the
flexural members results in initial deflection prior to loading and a decrease in the cracking
moment. The specimens with larger initial shrinkage exhibited slightly less tension stiffening
deflection under short-term loading. In the long-term, new primary cracks due to the restraint
to shrinkage are regarded as the major factor for the loss of tension stiffening. The larger the
285
shrinkage strain, the larger the loss of tension stiffening. Creep on the other hand has
relatively minor impact on the short-term and long-term behaviours of tension stiffening,
compared to shrinkage. According to the numerical results, the short-term and long-term
tension stiffening deflections of the specimens with different amount of creep are not
significantly different.
CHAPTER 9 CONCLUSIONS AND
RECOMMENDATIONS
9.1 SUMMARY
The present study aims to contribute to a theoretical understanding of tension stiffening in
reinforced concrete structures and its variation with time. It has undertaken experimental and
numerical studies to achieve this objective. Tension stiffening under both long-term and short-
term loading has been investigated. Shrinkage and creep, of the concrete have been taken into
the account in the analysis. These time-dependent effects have been largely ignored in the
previous studies of tension stiffening. Numerical works were undertaken using the finite
element program RECAP (Foster 1992, Foster and Gilbert 1990, Foster and Marti 2003). A
bond model has been proposed so as to simulate the effect of bond deterioration and its effect
on tension stiffening. The finite element program was used to model laboratory specimens
tested by the author and others. To first demonstrate the versatility of the finite element
program and demonstrate other possible parameters that may affect time-dependent tension
stiffening, a number of numerical experiments were conducted and the results were analysed
and discussed.
287
9.2 CONCLUSIONS
The mechanisms and factors affecting tension stiffening have been investigated in this study
and the results have been carefully presented. It is been demonstrated that initial shrinkage
before loading has significant effects on the structural behaviour. For uniaxial tension tests,
shrinkage causes initial shortening of the member, while for flexural tests, shrinkage induces
initial deflection. Ignoring this deformation prior to loading can lead to misinterpretation of
the experimental results. The experimental program conducted in this research carefully
examines the effects of initial shrinkage prior to loading, and the results are significant.
Before cracking, the restraint to shrinkage by the reinforcing bars produces significant tensile
force in the concrete. Therefore, members with initial shrinkage had much lower cracking
loads than the members without shrinkage. It has also been demonstrated by the experiments
that after cracking initial shrinkage have relatively minor influence on tension stiffening under
increasing loads. Nevertheless, for the flexural members considered in the numerical
experiments, it seems that initial shrinkage can actually reduce the tension stiffening
deflection after cracking. This is caused by the restraining force to shrinkage that is applied to
the concrete which can trigger additional cracks under the increasing moment. For the short-
term uniaxial tension members, larger crack width was observed on the members with initial
shrinkage. However, this is not the case for the short-term flexural members, where crack
width was not significantly influenced by initial shrinkage. It is also observed that shrinkage
prior to loading may induce more cracks and reduce crack spacing, which has been found in
Compared with shrinkage, creep has a relatively minor effect on short-term behaviour of
tension stiffening. This is observed in the numerical experiment. The short-term specimens
288
with different initial creep strains exhibited almost identical behaviour before and after
cracking. Creep before loading can slightly increase the initial deflection triggered by
shrinkage warping. However, after cracking the significance of creep on the behaviour of a
In the long-term tests where the load was maintained constant after cracking of the member,
the experimental results show that the contribution of the tensile concrete decreases with time.
The analysis of the long-term slab tests indicates that tension stiffening dropped off
dramatically in the first 15 days under loading, and then it decreased at a slower rate thereafter.
The test results also indicate that the long-term decreasing rate of tension stiffening is not
remarkably affected by the loading levels. A similar decreasing rate pattern was found from
the analysis of earlier test results (Nejadi and Gilbert 2004). The restraint to shrinkage by the
reinforcing bars can induce the time-dependent formation of primary cracks and cover-
controlled cracks, and hence is regarded as the major mechanism for the loss of tension
stiffening with time. On the other hand, in the uniaxial tension tests, tensile creep can shed the
tensile stress carried by the intact concrete to the steel bars and cause a gradual softening of
the load-deformation response. In flexural members, compressive creep causes the lowering
of neutral axis and slightly increases of total tensile force carried by the tensile steel bars.
A bond model has been proposed and incorporated into the finite element program. The bond
model was calibrated using the results obtained from the experimental program. In the finite
element models, concrete, steel and bond mechanism are represented by three different types
solidification creep theory (which had previously been developed in RECAP) in terms of a
series of Kevin chain units. Shrinkage was introduced by a simple age-related function. Both
289
the creep and shrinkage strains were treated as inelastic strains in the finite element analysis.
bond stress-slip relationship, in which bond stress is not only dependent on slip but also
concrete damage and steel stress states. For the long-term analysis, bond stress in the
proposed bond model also decreases with respect to the amount of shrinkage, as shrinkage is
considered as the major mechanism to cause long-term bond damage. The finite element
program was then adopted to simulate the experimental works and gave excellent results in
terms of both global and local responses compared with the experimental responses.
Finally, the parametric study has led to the following extra conclusions:
amount of tension reinforcement area or the reinforcing ratio. The numerical study has
members both under short-term and long-term loadings. With time, the decreasing rate
of tension stiffening does not vary too much among the members with different tensile
reinforcement area.
ii. The provision of compressive reinforcement can give rise to a stiffer response for the
beams by reducing the curvature resulting from restraint to shrinkage by the tensile
reinforcing bars and reduces the tensile stresses in the cracked concrete and the
development of bond degradation. Therefore it enhances both the short-term and long-
iii. The tensile strength of concrete has a direct impact on time-dependent tension
stiffening. The increased tensile strength of concrete can also improve the capacity of
290
concrete to carry tension and resist shrinkage-induced cracking. Thus both short-term
and long-term tension stiffening are increased in the concrete with higher tensile
strength. For the long-term specimens, the rate of change of tension stiffening seems
iv. The long-term rate of change of tension stiffening is dependent on the elastic modulus
of concrete. Namely the larger the concrete elastic modulus, the greater the rate is.
v. The bottom cover of concrete can increase the amount of tension stiffening in the
reinforced concrete beams. It also reduces the rate of change of tension stiffening
vi. Certain other parameters, such as transverse reinforcement, tensile bar diameter, etc.
The long-term decay of tension stiffening is a significant factor that should be considered in
the structural design, since stiffness of the structure is often the governing design criteria.
Shrinkage is the major factor leading to the loss of tension stiffening with time. In addition,
early shrinkage before loading can also contribute to the long-term increase of total deflection.
Thus the design approach that is used to predict long-term stiffness of reinforced concrete
beams and slabs must incorporate shrinkage. In Appendix A, an effective stiffness based
method is proposed by the author to predict the short-term and long-term stiffness of a
uniaxial tension prism. The method is based on the formula given by Eurocode 2 (1992) but
Chapter 4 were chosen to be analysed by this method and good correlation has been observed.
The method may then be extended for calculating the time-dependent stiffness of beams or
slabs.
members with different amount of initial shrinkage are required to confirm the
ii. Experimental work could also be undertaken to investigate the time-dependent tension
undertaken to extend the 2-D finite element model to cover 3-D simulations for the
confinement and radial stress) on the bond mechanism can be more accurately
iii. Further numerical studies to investigate the influence of different reinforcement types
(or example FRP bar, plain bar, etc.) on tension stiffening could be conducted. The
bond stress-slip relationship will need to be calibrated for different bar types.
292
iv. Various other parameters (loading history, bar surface characteristics, boundary
tension prisms given in Appendix A considering both shrinkage and creep can be
further simplified and extended to the analysis of beams and slabs in practice,
culminating in the development of reliable and rational approaches for the analysis
sustained loads has been proposed by Gilbert and Wu (2009). Recommendations are made for
including tension stiffening in the design of reinforced concrete elements at the serviceability
limit states. The method is presented here and might be further extended to be applied for the
Before cracking
Consider the uniaxial loaded concrete prism shown in Figure 2-1. When the applied load P is
less than the time-dependent cracking load, i.e. P < Pcr.sh, the member remains uncracked
throughout. Using the age-adjusted effective modulus method, the average member strain at
time t, s.avg(t) and the steel and concrete stresses in the uncracked section (s(t) and c(t),
P § nU · § n* n · H sh
H s ,avg (t ) ¨¨ ¸¸¨¨1 ¸¸ + A.1
© 1 nU ¹© n (1 n U ) ¹ 1 n U
* *
E s As
where sh is the shrinkage strain that occurs before time t; is the reinforcement ratio (As /
Ac); n is the modular ratio (Es / Ec), n* is the effective modular ratio (Es / Ee); n * is the age-
adjusted effective modular ratio ( E s / E e );Es and Ec are the elastic moduli of the steel and
coefficient for concrete for the time interval 't (between first loading and time t); and F is an
aging coefficient used to account for the gradual reduction in the creep coefficient with age
when the stress is gradually applied over the time interval 't and is often approximated by
The shrinkage induced tensile stress in the concrete on the uncracked cross-section ( V sh 0 ) is
obtained by setting P = 0kN in Equations A.1 and A.2 and the reduced cracking load after
shrinkage (Pcr.sh) is reached when c(t) in Equation A.2 equals the tensile strength of the
H sh E s U
V sh 0 and Pcr . sh Ac (1 nU )( f c V sh 0 ) A.3
1 n *U
After cracking
Cracking and cracking induced deterioration of bond caused by slip at the concrete-steel
interface are the physical mechanisms that are primarily responsible for the reduction in
tension stiffening under either increasing load or due to restraint to shrinkage and temperature
would cause an increase in tension stiffening, but the formation of new cracks and the
resulting bond slip results in a subsequent overall (net) drop in tension stiffening.
The average strain caused by the applied load P in a cracked tension member can be estimated
using an effective rigidity (EcAe), which is between the rigidities of the fully cracked section
ignoring the concrete (EcnAs) and of the uncracked section (EcAt). The area At is the
transformed area of the uncracked cross-section (Ac+nAs). The approach is similar to that
295
outlined in Eurocode 2 (1992), but modified to include the effects of time-dependent slip at
the concrete-steel interface caused by shrinkage and by time-dependent cracking. The average
strain caused by load in an axially loaded tension member after cracking at time t may be
approximated by
where s.cr is the steel strain at a crack, i.e. ignoring the concrete in tension; s.uncr is the steel
strain on the uncracked transformed section; and is a coefficient accounting for load level,
the degree of cracking and the amount of shrinkage. For deformed bars under short-term
loading, Eurocode 2 (1992) suggest that = 1 – (Pcr / P)2. However, to provide better
agreement with the instantaneous and time-dependent test data obtained in this study and
D
] 1 Pcr .sh / P A.5
where Pcr..sh is the load required to cause cracking, suitably modified to account for the tension
induced by restraint to shrinkage (and given by Equation A.3) and the index depends on the
level of shrinkage that has occurred prior to time t (sh) and may be taken as
If the magnitude of shrinkage strain is between 100 and 300 microstrain, then may be
In terms of an effective cross-sectional area, Ae, the average strain caused by load at time t is
P P P
(H s.avg ) i ] (1 ] ) A.7
E c Ae E c nAs E c At
nAs
Ae D
§ nA ·§ P · A.8
1 ¨¨1 s ¸¸¨ cr . sh ¸
© At ¹© P ¹
If the member has been subjected to drying shrinkage, (sh), the initial compressive strain on
the uncracked section ( H sh / 1 n * U must be added to the average strain caused by load to
P H sh
H s.avg A.9
E c Ae (1 n * U )
At any load P > Pcr.sh, the instantaneous tension stiffening strain tsi is the difference between
the strain in the bare bar ( P / AsEs) and the average member strain, s.avg (see Figure 3-1):
P P H sh
H tsi ( ) A.10
E s As E c Ae (1 n * U )
At each crack the reinforcement carries the entire load P, but the average force carried by the
Ns E s H s.avg A.11
and the average force in the concrete and the average concrete stress are therefore
Nc P Ns and Vc N c / Ac A.12
When the concrete on a cracked member shrinks by an amount sh over a time period t, the
restraint provided by the reinforcement introduces additional tensile stress in the concrete
between the primary cracks. However, this shrinkage induced tension causes additional time-
dependent cracking, and time-dependent slip at the concrete-steel interface and, consequently,
a net reduction in tension stiffening with time. To model the increase in cracking with time,
Eurocode 2 (1992) suggests that, for long term loading, the tension stiffening strain reduces to
50% of its initial value (i.e. 0.5tsi), irrespective of the level of loading. In reality, early
shrinkage causes a reduction in the cracking load and this affects the value of ] obtained
from Equation A.5 and consequently the value of Ae given by Equation A.8.
Working Examples
In the tests conducted by the author, two of the specimens (STN12 and STS12) were tested
under short-term monotonically increasing load (the short-term tests), while the third
specimen (LTN12A) was loaded to a typical service load level and then subjected to a
constant sustained loads for a period of about 50 days (the long-term test).
Short-term tests
Specimen STN12 was subjected to a monotonically increasing load soon after the end of
moist curing. The elastic modulus of concrete at the time of testing at age 32 days was Ec =
298
22400 MPa (and with Es = 200000 MPa, n = 8.93) and the tensile strength of concrete was ft
= 2.04 MPa. Prior to loading, the measured drying shrinkage strain was -28 x 10-6. The creep
coefficient associated with this initial period of shrinkage was M = 0.3 and therefore E e =
18060 MPa ( n * = 11.1). From Equation A.3, the initial concrete tensile stress caused by
restraint to shrinkage prior to loading and the reduced cracking load are Vsh0= 0.06 MPa and
Specimen STS12 was identical to specimen STN12 except that it was tested 3½ weeks later at
age 57 days when the drying shrinkage had increased to -249 x 10-6. For STS12, at the time of
testing at age 57 days, Ec = 21600 MPa (i.e. n = 9.26) and ft = 2.15 MPa. The creep coefficient
associated with the initial period of shrinkage was M = 1.13 and therefore E e = 11350 MPa
and n * = 17.6. From Equation A.3, the initial concrete tensile stress caused by restraint to
shrinkage prior to loading and the reduced cracking load are Vsh0= 0.47 MPa and Pcr.sh = 18.3
The average strain versus applied load curves measured using LVDTs throughout the short-
term tests are plotted in Figure A-1, together with the theoretical curves obtained using
Equation A.9. The agreement between the experimental and theoretical curves is excellent.
299
70
Pcr = 21.5 kN (experimental)
60 = 21.6 kN (theoretical – Eq. A.3)
Axial Load (kN)
Experimental
50 Hts
Theoretical
40 Bare bar P (kN) Hs.avg (exp) Hs.avg (theory)
30 33.9 1097x10-6 1031x10-6
40.0 1531x10-6 1401x10-6
Pcr 50.0 2083x10-6 1941x10-6
20
10 -6
Average Axial Strain (x10 ) (Hs.avg)
0
-500 0 500 1000 1500 2000 2500 3000
-25
(a) STN12
70
Pcr = 13.0 kN (experimental)
60 = 18.3 kN (theoretical – Eq. A.3)
Axial Load (kN)
50 Experimental
Theoretical
40 P (kN) Hs.avg (exp) Hs.avg (theory)
Bare bar 33.9 1098x10-6 1143x10-6
-6
30 40.0 1455x10 1466x10-6
-6
50.0 1994x10 1951x10-6
20
10
Average Axial Strain (x10 )
-6 (Hs.avg)
0
-500 0 500 1000 1500 2000 2500 3000
-209
(b) STS12
Figure A-9-1: Axial load versus average axial strain for STN12 and STS12.
Table A-1: Experimental and theoretical average concrete stresses after cracking
Applied Load, P Average concrete stress (MPa) Applied Load, P Average concrete stress (MPa)
on STN12 (kN) Experimental Theoretical on STS12 (kN) Experimental Theoretical
40 0.84 0.84 40 0.84 0.69
50 0.65 0.62 50 0.81 0.60
From the steel strain gauge readings, the average steel stress was obtained, and knowing the
applied load at each instant, the average concrete stress was determined for each specimen.
The measured and theoretical concrete stresses are compared in Table A-1 for several values
of applied load after the primary crack pattern had developed. Agreement is considered to be
good.
300
Long-term tests
Specimen LTN12A was moist cured until two days before the test started (at age 33 days).
The specimen was initially loaded in 5 kN increments up to a load P = 40 kN. The load was
then sustained for a period of 50 days (i.e. from age 33 days to age 83 days). The measured
elastic modulus of concrete at the time of first loading was Ec = 22400 MPa (n = 8.93) and the
tensile strength of concrete was ft = 2.04 MPa. The measured shrinkage strain in the concrete
just before first loading was -52 x 10-6 and the creep coefficient for this initial period of
drying (up to age 33 days) was 0.3 (i.e. E e = 18060 MPa, n * = 11.1). From Equation A.3, the
initial concrete tensile stress caused by restraint to shrinkage prior to loading and the reduced
cracking load are Vsh0= 0.11 MPa and Pcr.sh = 21.1 kN.
From Equation A.6, D= 2.5 and the theoretical load verses average strain curve at first loading
is shown in Figure A-2, together with the experimental curve obtained at first loading.
Agreement at the time of first loading is again excellent. Immediately after the application of
P = 40.0 kN, the measured average axial strain was 1409 x 10-6 while the calculated value was
1400 x 10-6.
The shrinkage strain at the end of the sustained load period was -310 x 10-6. The creep
coefficient associated with the sustained load period was 1.38. Therefore for the period
between age 33 days and age 83 days, E e = 10650 MPa ( n * = 18.8). From Equation A.3, the
concrete tensile stress caused by restraint to shrinkage after 50 days and the reduced cracking
load are Vsh0= 0.58 MPa and Pcr.sh = 15.9 kN. From equation A.6, D= 4.0 and at a sustained
load of P = 40 kN, the calculated average strain at age 83 days (obtained from Equation A.9)
is 1475 x 10-6 and this agrees reasonably well with the measured member strain at this time of
301
1556 x 10-6.
60
Age (days) = 33 83
Axial Load, P (kN)
50
'Hexp = 147 x 10-6
40
30
Experimental
20
Theoretical
10 -6
Bare bar
Average Strain (x10 )
0
-250 0 250 500 750 1000 1250 1500 1750 2000
Figure A-2: Axial load versus average axial strain forLTN12A.
APPENDIX B TIME-DEPENDENT ANALYSIS OF A
CRACKED OR UNCRACKED BEAM SECTION
An alternative approach for analysing a time-dependent reinforced concrete beam section is
given by Gilbert (1988). The method is also called a relaxation procedure (Bresler and Selna
1964). The basic theory of this method is to assume that the total strain throughout the section
remains constant during any time interval, meanwhile creep and shrinkage strain components
vary. Consequently the instantaneous concrete strains shall be changed, as well as concrete
stresses.
Consider a singly-reinforced concrete beam section in Figure B-1. The beam section has a
cross section of b (width) × D (depth) and is reinforced with Ast bottom tensile bars. The
effective depth of the tensile reinforcement is dst. Assume that the section is imposed with a
relatively small moment M so that it remains uncracked throughout. By using the method of
transformed section, the area of the steel reinforcement is transformed into an equivalent
concrete area, AstEs /Ec = nAst, where n is the modular ratio between steel and concrete, Es /Ec.
Figure B-2 shows the transformed section and the strain distribution of the uncracked section
By assuming that the plane section remains plane, at moment M, the instantaneous strain
distribution at any point with a distance y from the top fibre can be given as
Hi H oi yN i B.1
where oi is the instantaneous strain at the top fibre and i is the instantaneous curvature of the
Vi EcH i E c H oi yN i B.2
By integrating the first moment of the stress block about the top fibre, the bending moment is
calculated as
Mi ³
V i ydA E c H oi B E c N i I B.2
where B is the first moment of the transformed area about the top fibre, and I is the second
304
Therefore the values of oi and i can be obtained through the following expressions
BM i
H oi B.3
E c AI B 2
AM i
Ni B.4
E c AI B 2
where A is the area of the transformed section. The relaxation approach used here is to allow
the stress along the section to change with time due to the change of instantaneous strain.
Thus the equilibrium of the section is not satisfied. To regain the equilibrium along the section,
a change of axial force N and bending moment M must be applied. The time-dependent
variation of top fibre strain and curvature produced by M and N (called restraining forces)
Be M I e N
o B.5
E e Ae I e Be2
Ae M B e N
B.6
E e Ae I e Be2
where A e is the area of the age-adjusted transformed section; B e and I e are the first and
second moments of the area of the age-adjusted transformed section about the top fibre; E e is
The restraining actions on the section due to the change of creep and shrinkage strain can be
obtained as
N E e >M Ac oi Bc i sh Ac @ B.7
305
M E e >M Bc oi I c i sh Bc @ B.8
where Ac, Bc and Ic are the properties of the concrete section about the top fibre. The loss of
stress in the concrete at any point with a distance y from the top fibre due to relaxation is
relax >
E e M oi y i sh @ B.9
The stress that is needed to restore the equilibrium with the applied actions is:
restore >
E e 'H o y'N @ B.10
The actual change of concrete stress with the time interval is the sum of relax and restore
The analysis of fully-cracked section is also based on the same assumption that cross section
remains plane and the total strain of the cross section remains unchanged with time. Another
assumption is made in this case that the depth of concrete compression zone dn remains
constant with time and is calculated from the short-term analysis. This assumption indicates
that the short-term and time-dependent change of stress or strain can be calculated separately
and added together by the means of superposition. Little error was found under this
assumption.
Under the sustained moment M (greater than cracking moment Mcr), cracking occurs
306
underneath the neutral axis of the section and concrete losses its ability to carry any tensile
stress. Equation B.7 and B.8 can be adopted to calculate the restraining actions to restore
equilibrium of the section when the relaxation method is used. However, the section
properties, such as A, B, and I, (and Ac, Bc, and Ic) in the equations are calculated according to
the concrete compression zone dn rather than the entire depth D. The change of strain
distribution with time can be calculated by equation B.5 and B.6, where A e , B e and I e are the
Accordingly, the change of concrete stress with any time interval throughout the depth of the
section can be determined by Equations B.9, B.10, and B.11. The change of steel stress can
WORKING EXAMPLES
Some of the tested beams and slabs presented in Chapter 5 are chosen to be analysed by the
BSTS2-16
For the short-term specimens (BSTS2-16), the fully-cracked response needs to be changed
according to the shrinkage prior to loading, which herein are calculated using the analytical
method. The calculated fully-cracked responses were then used to analyse instantaneous
Ec = 25000MPa (day 14); Es = 204000MPa; c = 0.92; = 0.8; sh = -220 × 10-6; (before
For a fully-cracked beam section, the depth of the compression concrete zone dn must be
derived before the properties of the cracked section can be determined. In this study, dn is
calculated by considering the equilibrium of the concrete and steel stresses through the
dn 90.0mm
and the properties of the uncracked part of the cracked concrete cross section (compressive
The age-adjusted effective modulus during the time interval of drying is (Equation 2.23)
25000
Ee 14401MPa
1 0.8 u 0.92
In terms of the age-adjusted transformed section, the properties are calculated as:
Equation B.7 and B.8 give the restraining actions on the cracked cross-section due to creep
N >
14401u 0.92 18003 u 0 810230 u 0 220 u 10 6 u 18003 @ 57.0kN
M >
14401u 0.92 810230 u 0 4.86 u 10 7 u 0 220 u 10 6 u 810230 @ 2.57 kN .m
Therefore the variation of time-dependent strain at the top fibre and curvature are obtained
308
Integrating the increment of curvature across the entire span (2.4m) of the beam gives the
change of mid-span deflection during the drying period of the beam, assuming all the sections
are fully-cracked.
After being dried for a period of time, BSTS2-16 was subjected to monotonic applied moment.
At a specific applied moment M (M > Mcr.sh), the fully-cracked response needs to be shifted
with an amount of sh0.cr (Figure 3-5), which indicates that the fully-cracked deflection is the
sum of sh0.cr and the one calculated without considering initial drying shrinkage (short-term
fully-cracked deflection). Thereafter tension stiffening deflection for the test beam can be
calculated by subtracting the measured mid-span deflection from the calculated fully-cracked
SLTN4-12A
This analytical method can also be adopted to derive the long-term fully-cracked responses of
a flexural member. The slab SLTN4-12A tested by the author and presented in Chapter 5 is
chosen as a working example in this case. The basic properties for specimen SLTN4-12A are
Ec = 25000MPa (Day 15); Es = 200000MPa; c = 1.47; = 0.8; sh = -400 × 10-6 (70 days
309
after first loading), Ast = 452mm2; b = 800mm, D = 140mm, dst = 114mm, and Mi = 17.9kN.m
Under instantaneous moment, the depth of compressive zone for the cracked cross-section of
SLTN4-12A is calculated as
dn 28mm
The properties of the fully-cracked transformed section about the top fibre are calculated as
The properties of the concrete compressive zone about the top fibre are:
At first loading, the instantaneous top fibre strain and curvature in the constant moment
723565 u 17.9 u 10 6
H oi 612 u 10 6
25000 25935 u 5.28 u 10 7 723565 2
25935 u 17.9 u 10 6
Ni 2.2 u 10 5
25000 25935 u 5.28 u 10 7 723565 2
25000
Ee 11489MPa
1 0.8 u 1.47
Therefore by using equation B.7 and B.8 and assuming applied moment remains constant
with time, the restraining actions in the constant moment region are calculated as
310
N >
11489 u 1.47 22319 u 612 u 10 6 311341 u 2.2 u 10 5 400 u 10 6 u 22319 @ 218kN
M >
11489 u 1.47 311341 u 612 u 10 6 5.79 u 10 6 u 2.2 u 10 5 400 u 10 6 u 311341 @ 2.5kN .m
In terms of the age-adjusted transformed section, the properties are calculated as:
The time-dependent increment of top fibre strain and curvature can be calculated from
equation B.5 and B.6 (notice that in the calculation below, the quantity
2
E e Ae I e B e 2.07 u 1016 N.mm4.)
which indicates that during the sustained loading period, shrinkage warping induced time-
dependent curvature of the fully-cracked section along the specimen. Since the values of 0
and are dependent on the magnitude of applied moment Mi, as the restraining forces are
related to the instantaneous curvature and strain, the time-dependent curvature along the
length of the specimen varies. Therefore, in this study, a number of sections at regular
intervals along the specimens were chosen, and time-dependent curvatures of these sections
are then used to calculate the fully-cracked deflections at mid-span. As for SLTN4-12A, after
70 days under sustained load can be calculated by integrating the curvature diagram and is
given as
Then the tension stiffening deflection after 70 days can be calculated by subtracting the
311
measured deflection at mid-span from the calculated fully-cracked deflection. Similarly, given
the concrete properties at different loading stages (such as creep and shrinkage strains), the
stages. Then the long-term tension stiffening deflections at different loading stages can be
calculated accordingly.
312
REFERENCES
CEB-FIP (2004). "Eurocode 2: Design of concrete structures Part 1-1: General rules and rules
Standards Australia (AS), (2001). "Australian Standards for concrete structures" AS 3600-
CEB-FIP, (1993). CEB-FIP Model code (1990 design code), Comitte Euro-International du
Attard, M. M., Minh, N. G., and Foster, S. J. (1996). "Finite element analysis of out-of-plane
Balazs, G. L. (1993). "Cracking analysis based on slip and bond stresses." ACI Materials
Barros, M. H. F. M., Martins, R. A. F., and Ferreira, C. C. (2001). "Tension stiffening model
Bazant, Z., and Prasannan, S. (1989b). "Solidification theory for concrete creep II.
Bazant, Z. P. (1982). "Mathematical models for creep and shrinkage of concrete." Symposium
Modelling of Creep and Shrinkage of Concrete, Z. P. Bazant, ed., John Wiley & Sons, Inc.,
Bazant, Z. P., Abrams, M. S., Chern, J. C., and Gillen, M. P. (1982). "Normal and refractory
of Portland Cement and refractory concretes." Report NP-2437, Electric Power Research
Bazant, Z. P., Asghari, A., and Schmidt, J. (1976). "Experimental study of creep hardened
cement past at variable water content." Materials and Structures (RILEM, Paris), 9, 279-290.
314
Bazant, Z. P., and Baweja, S. (1995a). "Justification and refinements of Model B3 for
Bažant, Z. P., and Chern, J. C. (1985). "Concrete creep at variable humidity: constitutive law
Bažant, Z. P., Hauggaard, A. B., Baweja, S., and Ulm, F.-J. (1997). "Microprestress-
solidification theory for concrete creep: I: aging and drying effects." Journal of Engineering
Bazant, Z. P., and Najjar, L. J. (1973). "Comparison of approximate linear methods for
Bazant, Z. P., and Oh, B. H. (1983). "Crack band theory for fracture of concrete." Materiaux
Bazant, Z. P., and Prasannan, S. (1989a). "Solidification theory for concrete creep I.
Bažant, Z. P., and Wittmann, F. H. (1983). "Creep and Shrinkage in Concrete Structure." John
Bažant, Z. P., and Wu, S. T. (1974). "Creep and shrinkage law for concrete at variable
Bazant, Z. P., and Xi, Y. (1995). "Continuous retardation spectrum for solidification theory of
Bažant, Z. P., and Xi, Y. (1994). "Drying creep of concrete: Constitutive model and new
27(165), 3-14.
Beeby, A. W., and Scott, R. H. (2002). "Tension stiffening of concrete (Behaviour of tension
Belarbi, A., and Hsu, T. T. C. (1991). "Constitutive laws of reinforced concretein biaxial
with steel and fiber reinforced polymer bars." Journal of Structural Engineering, 131(5), 752-
762.
Bischoff, P. H., and Paixao, R. (2004). "Tension stiffening and cracking of concrete reinforced
with glass fiber reinforced polymer (GFRP) bars." Canadian Journal of Civil Engineering,
31(4), 579-588.
Bolander, J., Jr., Satake, M., and Hikosaka, H. (1992). "Bond degradation near developing
concrete structures." Symposium on creep of concrete, ACI Special Publication SP-9, No.5,
115-128.
Brooks, J. J., and Neville, A. M. (1977). "A comparison of creep, elasticity and strength of
Castel, A., Vidal, T., and Francois, R. (2006) "Effective tension active cross-section of
reinforced concrete beams after cracking." Materials and Structures, 39, 115-126.
structures using the cracked membrane model and solidification theory." Computers and
Clark, L. A., and Cranstown, W. B. (1979). The influence of bar spacing on tension stiffening
London.
Clark, L. A., and Speirs, D. M. (1978). Tension stiffening in reinforced concrete beams and
Collins, M. P., and Porasz, A. (1989). " Shear strength for high strength concrete." Bull. NO.
193, Design Aspects of High Strength Concrete, Comite Euro-International du Beton: 75-83.
Committee, A. C. (2005). "Building code requirements for structural concrete " ACI318-05,
Cox, J. V. (1994). "Development of a plasticity bond model for reinforced concrete - Theory
Cox, J. V., and Herrmann, L. R. (1998). "Development of a plasticity bond model for steel
Darwin, D., and Graham, E. K. (1993), "Effect of Deformation Height and Spacing on Bond
Darwin, D., and Pecknold, D. A. (1977). "Nonlinear biaxial stress-strain law for concrete."
American Society of Civil Engineers, Journal of the Engineering Mechanics Division, 103(2),
229-241.
Den Uijl, J. A., and Bigaj, A. J. (1996). "A bond model for ribbed bars based on concrete
und das Kriechen des Betons bei Bogenbruecken." Bauingenieur, 18(39/40), 595-621.
Domone, P. L. (1974). "Uniaxial tensile creep and failure of concrete." Magazine of Concrete
Edwards, A. D., and Yannopoulos, P. J. (1979). "Local bond-stress to slip relationships for hot
rolled deformed bars and mild steel plain bars." Journal of the American Concrete Institute,
76(3), 405-420.
318
Eligehausen, R., Popov, E. P., and Berbero, V. V. (1983). "Local bond stress-slip relationships
of deformed bars under generalized excitations: experimental results and analytical model."
Berkeley, California.
structures Part 1-1: General rules for buildings." DD ENV 1992-1-1m European Prestandard,
Brussels, Belgium.
Fields, K., and Bischoff, P. H. (2004). "Tension stiffening and cracking of high-strength
Floegl, H., and Mang, H. A. (1982). "Tension stiffening concept based on bond slip."
American Society of Civil Engineers, Journal of the Structural Division, 108(ST12), 2681-
2701.
Foster, S. J., and Marti, P. (2003). "Cracked membrane model: Finite element
Franke, L. (1976). "Einfluss der Belastung-Sdauer auf das Verbundverhalten von Stahl in
Gambarova, P. G., Paolo Rosati, G., and Zasso, B. (1989a). "Steel-to-concrete bond after
Gamble, B. R., and Parrott, L. J. (1978). "Creep of concrete in compression during drying and
319
Gao, D., Benmokrane, B., and Masmoudi, R. (1998). "A calculating method of flexural
properties of FRP-reinforced concrete beam: Part 1: crack width and deflection.Tech. Rep."
Ghali, A., Hall, T., and Bobey, W. (2001). "Minimum thickness of concrete members
reinforced with fibre reinforced polymer bars." Can. J. Civ. Eng., 28(4), 583-592.
Gilbert, R. I. (2006). "Design for flexural crack control - recent amendments to the Australian
Standard AS3600." 1st International Structural Specialty Conference, CSCE, Calgary Alberta,
Candada.
Gilbert, R. I., and Warner, R. F. (1978). "Tension stiffening in reinforced concrete slabs."
Gilbert, R. I., and Wu, H. Q. (2009). "Time-dependent stiffness of cracked reinforced concrete
elements under sustained actions." Australian Journal of Structural Engineering, 9(2), 151-
158.
Giuriani, E., Plizzari, G., and Schumm, C. (1991). "Role of stirrups and residual tensile
Goodman, R. E., Taylor, R. L., and Brekke, T. L. (1968). "Model for mechanics of jointed
rock." American Society of Civil Engineers Proceedings, Journal of the Soil Mechanics and
Goto, Y. (1971). "Cracks Formed in Concrete Around Deformed Tension Bars." ACI Journal
Guptam, A. K., and Maestrini, S. R. (1990). "Tension-stiffness model for reinforced concrete
Hillerborg, A. (1983), Analysis of one single crack. Fracture Mechanics of Concrete. Ed. F. H.
Illston, J. M. (1965). "The creep of concrete under uniaxial tension." Magazine of Concrete
Illston, J. M., and Stevens, R. F. (1972). "Long-term cracking in reinforced concrete beams."
Proceedings of the Institution of Civil Engineers (London). Part 1 - Design & Construction,
Kankam, C. K. (1997). "Relationship of bond stress, steel stress, and slip in reinforced
plane shear and normal forces," Swiss Federal Institute of Technology, Zurich.
Kaufmann, W. and Marti, P., "Structural Concrete: Cracked Membrane Model," ASCE
Kishek, M. A. (1983). "Tension stiffening and crack widths in reinforced concrete beam and
slab elements."
Koch, R., and Balazs, G. L. (1993). "Slip increase under cyclic and long term loads." Otto-
Graf-Journal, 4, 160-191.
Kupfer, H., Hilsdorf, H. K., and Rusch, H. (1969). "Behaviour of concrete under biaxial
Lin, C.-S., and Scordelis, A. C. (1975). "Nonlinear analysis of RC shells of general form."
322
101(3), 523-538.
Lorman, W. R. (1940). "The theory of concrete creep." Preceedings of ASTM, 40, 1082-1102.
Lowes, L. N., Moehle, J. P., and Govindjee, S. (2004). "Concrete-steel bond model for use in
finite element modeling of reinforced concrete structures." ACI Structural Journal, 101(4),
501-511.
Lutz, L. A., and Gergely, P. (1967). "Mechanics of bond and slip of deformed bars in
Lutz, L. A., Gergely, P., and Winter, G. (1966). "Mechanics of bond and slip of deformed
Maekawa, K., Pimanmas, A., and Okamura, H. (2003). Nonlinear Mechanics of Reinforced
Marti, P., Alvarez, M., Kaufmann, W., and and Sigrist, V. (1998). "Tension chord model for
McHenry, D. (1943). "New aspect of creep in concrete and its application to design."
323
Miyakawa, T., and Kawakami, T. (1987). "Nonlinear behaviour of. cracked reinforced
concrete plate element under uniaxial compression." JSCE Proceedings, 378, 249-258.
deformed reinforcing bars in concrete under lateral pressure." Magazine of Concrete Research,
39(140), 161-168.
Nayal, R. and Rasheed, H. A. (2006). "Tension stiffening model for concrete beams reinforced
with steel and FRP bars." ASCE Journal of Materials in Civil Engineering, 18(6), 831-841.
Nejadi, S., and Gilbert, I. (2004a). "Shrinkage cracking and crack control in restrained
Nejadi, S., and Gilbert, I. (2004b). "An experimental study of flexural cracking in reinforced
concrete members under sustained loads." UNICIV Report. University of New South Wales.
No. R-435.
Neville, A. M., Dilger, W. H., and Brooks, J. J. (1983). Creep of plain and structural concrete,
Ouyang, C., Wollrab, E., Kulkarni, S. M., and Shah, S. P. (1997). "Prediction of cracking
70-78.
324
Petersson, P. E. (1981). "Crack Growth and Development of Fracture Zone in Plain Concrete
interface model for cohesive brittle materials: Application to bond slip in RC structures."
Robins, P. J., and Standish, I. G. (1984). "Influence of lateral pressure upon anchorage bond."
Roelfstra, P. E., Wittmann, F. H. (1986), "Numerical method to link strain softening with
Salem, H., Hauke, B., Maekawa, K. (1999). "Fracture of concrete cover. Its effect on tension
Shima, H., Chou, L.-L., and Okamura, H. (1987). "Micro and macro models for bond in
39(2), 133-194.
325
Soh, C. K., Liu, Y., Dong, Y. X., and Lu, X. Z. (2003). "Damage model based reinforced-
Somayaji, S., and Shah, S. P. (1981). "Bond stress versus slip relationship and cracking
response of tension members." Journal of the American Concrete Institute, 78(3), 217-225.
Sooriyaarachchi, H., Pilakoutas, K., Byars, E. (2005). "Tension stiffening behaviour of GFRP-
Einschlub der Mitwirkung des Betons auf Zug in Abhangigkeit von Last und Zeit." Diss. TU
Braunschweig.
Tepfers, R. (1979). "Cracking of concrete cover along anchored deformed reinforcing bars."
Thorenfeldt, E., Tomasewicz, A., and al, e. (1987). "Mechanical properties of high strength
Toutanji, H. A., and Saafi, M. (2000). "Flexural behaviour of concrete beams reinforced with
glass fiber-reinforced polymer (GFRP) bars." ACI Structural Journal, 97(5), 712-719.
probleme bei Beton Und Spannbeton", Beton- und Stahlbetonbau, 62(10), 230-238.
326
Untrauer, R. E., and Henry, R. L. (1965). "Influence of normal pressure on bond strength."
Vecchio, F. J., and Collins, M. P. (1982). "The response of reinforced concrete to in-plane
Vecchio, F. J., and Collins, M. P. (1986). "Modified compression-field theory for reinforced
concrete elements subjected to shear." Journal of the American Concrete Institute, 83(2), 219-
231.
Whitney, C. S. (1932). "Plain and reinforced concrete arches." American Concrete Institute --
Wu, H. Q., and Gilbert, R. I. (2009a) "Modeling short-term tension stiffening in reinforced
31(10), 2380-2391.
reinforced concrete slabs." Proceedings of the 24th Biennial Conference of the Concrete