02whole PDF

Time Dependent Flexural Analysis of
Reinforced Concrete Members
Prepared by
Noor Md. Sadiqul Hasan

B.Sc (Civil Engineering) and M.Sc (Structural Engineering)
A dissertation submitted for the degree of

Doctor of Philosophy (Structural Engineering)
School of Civil, Environment and Mining Engineering
The University of Adelaide
Australia
-August 2016-
ABSTRACT
Concrete is one of the most widely used building materials in the construction industry in
the world. Time dependent behaviour of concrete is the major concern for the structural
engineers due to its significant effect in the long term serviceability and durability.
Reinforced concrete (RC) members are prone to the effect of time dependent deformations
that are known as shrinkage and creep, can produce substantial deformations and
deflections to the structure.
The mechanics of quantifying the serviceability deflection of RC beams is complex due
to flexural cracking and the associated partial interaction (PI) behaviour of slip between
the reinforcement and adjacent concrete. Add the additional complexity of time dependent
concrete shrinkage to this partial-interaction (PI) behaviour and the problem becomes very
complex.
Current design and analysis techniques to quantify serviceability deflection of reinforced
concrete (RC) members are generally built on two major principles which are full
interaction (FI) through the use of moment curvature approaches; and a uniform
longitudinal shrinkage strain sh within the member to simplify the analysis technique.
Both of the premises are gross approximations and with regard to the first premise, RC
beams are subject to flexural cracking and the associated partial interaction (PI) behaviour
of slip between the reinforcement and adjacent concrete. Furthermore with regard to the
second premise, numerous tests have shown that sh varies along both the depth and width
of the beam and which is far from uniform. Hence there are two major sources of error in
the quantification of serviceability deflections of RC beams for design and which are due
to the PI mechanisms that occur in practice; and that due to the time dependent material
properties of creep and shrinkage.
This thesis deals with the development of PI numerical mechanics models with non-linear
shrinkage strain variations achieved from a moisture diffusion model developed in this
study and that is required to simulate the PI behaviour of RC beams in order to
considerably reduce the source of error occurred due to the application of numerical
mechanics model. Hence this new mechanics model will allow: the development of better
design mechanics rules for serviceability deflection; and also assist in the better
quantification of non-linear shrinkage and creep by removing or considerably reducing
I
the existing mechanics source of error. Importantly, this research provides mechanics
solutions for all the facets that control the serviceability time dependent behaviour of RC
beams and it is envisaged that these numerical mechanics solutions can provide
researchers with the tools to develop simple design procedures as they simulate the major
mechanisms influencing cracking and tension stiffening in reinforced concrete beams.
Current shrinkage test methodology is having some limitations that are all surfaces are
exposed to the environment and they are small scaled which leads to a uniformity of
shrinkage strain and which are not present in real size RC beams. Therefore in this thesis,
a new form of experimental setup for shrinkage have been proposed to better quantify the
shrinkage variations along both the width and depth of RC members with varying the sizes
and surface boundary conditions.
II
STATEMENT OF ORIGINALITY
I certify that this work contains no material which has been accepted for the award of any
other degree or diploma in my name, in any university or other tertiary institution and, to
the best of my knowledge and belief, contains no material previously published or written
by another person, except where due reference has been made in the text. In addition, I
certify that no part of this work will, in the future, be used in a submission in my name,
for any other degree or diploma in any university or other tertiary institution without the
prior approval of the University of Adelaide and where applicable, any partner institution
responsible for the joint-award of this degree.
I give consequent to this copy of my thesis when deposited in the University Library,
being made available for loan and photocopying, subject to the provisions of the Copyright
Act 1968.
The author acknowledges that copyright of published works contained within this thesis
resides with the copyright holder(s) of those works.
I also give permission for the digital version of my thesis to be made available on the web,
via the University’s digital research repository, the Library Search and also through web
search engines, unless permission has been granted by the University to restrict access for
a period of time.
Noor Md. Sadiqul Hasan Date
III
ACKNOWLEDGEMENT
At first I would like to express my deepest gratitude and sincere appreciation to my

principal supervisor Emeritus Professor Deric John Oehlers for his excellent supervision
during my PhD study. I also would like to express my sincere appreciation to my co-
supervisor Dr. Phillip Visintin for his wonderful supervision throughout the PhD research
were an enormous help to me in the completion of this work. Their invaluable advice,
continuing support, encouragement, patience and weekly regular meetings were generous
help throughout the PhD work reached me to the final level.
I would also like to thank Dr. Terry Bennett for his good collaboration in solving my
issues, providing guidance and his valuable advices throughout my PhD research.
The financial support from the Adelaide Scholarship International (ASI) through the
University of Adelaide (UoA) are highly appreciated.
Finally, I am very much grateful to my parents, family members and relatives for their
love, continuing support, motivation, understanding and encouragement during my PhD
start to the final day. A special thanks goes to my beloved Wife for her continuous support
when it was needed, distraction when it was required and motivation when I lost mine.
Last but not the least, I would like to dedicate my PhD thesis to my beloved parents.
IV
List of Publication
Based on the research work one journal paper has been submitted for publication in
Proceedings of the Institution of Civil Engineers – Structures and Buildings.
Hasan, N. M. S., Bennett, T., Visintin, P., Oehlers, D. J. (2016). “Mechanics of simulating
the serviceability deflection of RC beams allowing for partial interaction and non-linear
shrinkage” (Submitted to Proceedings of the Institution of Civil Engineers – Structures
and Buildings).
V
Table of Contents
ABSTRACT…………………………………………………………………………........І
STATEMENT OFORIGINALITY………………………………………………..........ІІІ
ACKNOWLEDGEMENT……………………………………………………………...ІV
LIST OF PUBLICATIONS……………………………………………………………...V
LIST OF FIGURES……………………………………………………………………..ІX
LIST OF TABLES……………………………………………………………………XІV
Chapter 1 Introduction................................................................................ 1
1.1 Introduction......................................................................................................... 1
1.2 Scope of the Research ......................................................................................... 6
1.3 Aims and Objectives of the Research ................................................................. 6
1.4 Structure of the Thesis ........................................................................................ 7
Chapter 2 Literature Review ...................................................................... 9
Introduction.................................................................................................................... 9
2.1 Shrinkage and its types ....................................................................................... 9
2.2 Effect of Shrinkage and Creep on Structures ................................................... 12
2.3 Prediction of Shrinkage Strains ....................................................................... 14
2.4 Using Moisture Diffusion to quantify Shrinkage ............................................. 17
2.4.1 Moisture and Humidity Diffusion in Concrete.......................................... 18

2.4.2 Shrinkage strain in Concrete ..................................................................... 22
2.4.3 Correlations in between with Moisture loss, Humidity and Shrinkage strain
of Concrete ................................................................................................ 25
2.5 Member Behaviour ........................................................................................... 37
2.5.1 Models to predict long term Deflections ................................................... 37

2.6 Summary ........................................................................................................... 41
VI
Chapter 3 Journal Paper on Non-linear Shrinkage ............................... 43
Chapter 4 Simulating Shrinkage Strain using Moisture Diffusion ....... 84
4.1 Introduction....................................................................................................... 84
4.2 Moisture diffusion equation .............................................................................. 84
4.3 Moisture diffusion coefficient .......................................................................... 86
4.4 Finite difference method on moisture diffusion analysis ................................. 87
4.5 Relationship between pore relative humidity and free shrinkage strain of
concrete ............................................................................................................... 88
4.6 Quantification of moisture diffusion coefficient .............................................. 90
4.7 Four way flow in rectangular beam .................................................................. 92
4.8 Three way flow in a beam ................................................................................ 94
4.9 Two way flow in a beam .................................................................................. 95
4.10 One way flow in a beam ................................................................................... 96
4.11 First simulation with Asad, Baluch et al. (1997) .............................................. 98
4.12 Second simulation with Jafarifar (2012)......................................................... 100
4.13 Third simulation with Kim and Lee (1999) .................................................... 105
Chapter 5 Proposed Experimental Work .............................................. 114
5.1 Introduction ................................................................................................. 114

5.2 Purpose of tests ........................................................................................... 114
5.3 Sizes of specimen ........................................................................................ 115
5.4 Testing for material properties .................................................................... 116
5.5 Standard shrinkage test ............................................................................... 116
5.6 Instrumentation in details ............................................................................ 116
5.7 Concluding remarks .................................................................................... 130
VII
Chapter 6 Long term Beam Deflection using Segmental Approach ... 131
6.1 Introduction..................................................................................................... 131
6.2 Partial interaction segmental analysis ............................................................. 132
6.3 Prior to cracking segmental analysis .............................................................. 133
6.4 Accommodation of cracking in the segmental approach ................................ 135
6.5 Partial interaction tension stiffening model .................................................... 137
6.6 Partial-interaction segmental model ............................................................... 143
6.7 Constant longitudinal shrinkage along depth and width................................. 144
6.8 Variation in longitudinal shrinkage strain along depth................................... 147
6.9 Variation in longitudinal shrinkage strain along depth and width .................. 149
6.10 Parametric study ............................................................................................. 151
6.11 Application to test specimens ......................................................................... 154
Chapter 7 Conclusions and Recommendations..................................... 158
References ................................................................................................................. 161
VIII
List of Figures
Figure 1.1: Concrete strain component under sustained compressive stress (Gilbert and
Ranzi 2011) ....................................................................................................................... 1
Figure 1.2: Thesis layout of this research .......................................................................... 7
Figure 2.1: Shrinkage stages and types of concrete ........................................................ 10
Figure 2.2: Shrinkage strain components in normal strength concrete (Sakata et al., 2004
cited in Gribniak et al., 2008) .......................................................................................... 11
Figure 2.3: Experimental and predicted shrinkage strain in a plain rectangular concrete
prism (80 × 150 × 500 mm) using different codes and shrinkage prediction models ..... 16
Figure 2.4: Experimental and predicted shrinkage strain in a standard concrete prism (50
× 50 × 300 mm) using different codes and shrinkage prediction models ....................... 17
Figure 2.5: Numerical moisture profiles compared with the experimental test results: (a)
Plain CC mix; (b) Plain RCC mix; (c) SFR-CC mix (2.5%); (d) SFR-RCC mix (2.5%)
(Jafarifar 2012) ................................................................................................................ 19
Figure 2.6: Shrinkage strain variation along the thickness of concrete specimen for two
different mixes and comparison with the analytical results (Kim and Lee 1998) ........... 23
Figure 2.7: Calculated relative humidity compared to modified experimental results due
to moisture diffusion only a) curing period = 3 days b) curing period = 28 days (Kim and
Lee 1999) ......................................................................................................................... 26
Figure 2.8: Relationship between relative humidity and moisture diffusion on different
moist curing period (Kim and Lee 1999) ........................................................................ 27
IX
Figure 2.9: Loss of moisture in concrete due to drying (Kim and Lee 1999) ................. 28
Figure 2.10: Comparisons between experimental and numerical results (ambient

temperature 200 C): a) Exposed at 3 days (w/c = 0.28) b) Exposed at 28 days (w/c = 0.28)
c) Exposed at 3 days (w/c = 0.40) d) Exposed at 28 days (w/c = 0.40) e) Exposed at 3 days
(w/c = 0.68) (Kang et al 2012) ........................................................................................ 29
Figure 2.11: Relationship between ultimate shrinkage and relative humidity for cement
paste and mortar specimens (Bissonnette, Pierre et al. 1999) ......................................... 30
Figure 2.12: Relation between shrinkage and weight loss for various types and sizes of
paste, mortar and concrete specimens (Bissonnette, Pierre et al. 1999) ......................... 30
Figure 2.13: Relationship between free shrinkage strain and interior relative humidity
(RH) of a) C30 concrete and b) C80 concrete (Zhang, J, Dongwei and Wei 2010) ....... 32
Figure 2.14: Calculated free shrinkage strains and measured relative humidity at different
depths of the slab from exposed surface against drying period (Zhang, J, Dongwei and
Wei 2010) ........................................................................................................................ 33
Figure 2.15: Relationship between free shrinkage strain and moisture loss (ACI209R-92
1997; Asad, M 1995) ....................................................................................................... 34
Figure 2.16: Experimental results of different concrete specimens drying shrinkage as a

function of weight loss (Granger, Torrenti and Acker 1997) .......................................... 36
Figure 4.1: Flow chart diagram to perform moisture diffusion modelling process ........ 89
Figure 4.2: a) Gilbert and Nejadi (2004) Beam B1a b) Sectional elevation of beam B1a at
A-A c) Four way flow in beam B1a d) Three way flow in beam B1a e) Two way flow in
beam B1a f) One way flow in beam B1a ........................................................................ 90
X
Figure 4.3: Moisture diffusivity vs moisture content or pore relative humidity, h for the
beams tested by Gilbert and Nejadi (2004) ..................................................................... 92
Figure 4.4: Shrinkage strain profile in a four way flow Gilbert and Nejadi (2004) Beam
B1a after 100 days, 250 days and 400 days of drying ..................................................... 94
Figure 4.5: Shrinkage strain profile in a three way flow Gilbert and Nejadi (2004) Beam
Figure 4.6: Shrinkage strain profile in a two way flow Gilbert and Nejadi (2004) Beam
B1a after 100 days, 250 days and 400 days of drying..................................................... 96
Figure 4.7: Shrinkage strain profile in a one way flow Gilbert and Nejadi (2004) Beam
Figure 4.8: Relationship between moisture content or pore relative humidity, h and
moisture diffusivity D ..................................................................................................... 98
Figure 4.9: Numerical simulation of experimental and predicted values of moisture loss at
1 cm from the drying surface using finite difference method ....................................... 100
Figure 4.10: Moisture diffusivity versus pore relative humidity or moisture content for
various types of concrete mixes (Jafarifar 2012) .......................................................... 101
Figure 4.11: Numerical moisture profiles compared with the experimental results
simulated using finite difference method: (a) Plain CC mix; (b) Plain RCC mix; (c) SFR-
CC mix; (d) SFR-RCC mix ........................................................................................... 104
Figure 4.12: Relationship between moisture diffusivity and moisture content for three
different types of concrete a) H (w/c = 0.28) b) M (w/c = 0.40) c) L (w/c = 0.68) with
moist cured for 3 days and d) H (w/c = 0.28) e) M (w/c = 0.40) f) L (w/c = 0.68) with
moist cured for 28 days ................................................................................................. 109
XI
Figure 4.13: Numerical simulation of experimental results for three different types of
concrete after moist cured for 3 days ............................................................................ 111
concrete after moist cured for 28 days .......................................................................... 113
Figure 5.1: Arrangements of demec gauge points along the depth at front (shown only)
and rear surfaces and at both ends through the depth as well as width of the prism for
measurements of shrinkage strains and total deformations of the prisms in one direction
moisture diffusion process............................................................................................. 119
and rear surfaces of the prisms and at both ends through the depth as well as width of the
prisms for measurements of shrinkage strains and total deformations of the prisms in two
direction moisture diffusion processes .......................................................................... 121
prisms for measurements of shrinkage strains along its depth and total deformations of the
prisms in three direction moisture diffusion processes ................................................. 123
Figure 5.4: Arrangements of demec gauge points along the width at top (shown only) and
bottom surfaces of the prisms and at both ends through the depth as well as width of the
prisms for measurements of shrinkage strains along its width and total deformations of the
prisms in three direction moisture diffusion processes ................................................. 125
prisms for measurements of shrinkage strains along its depth and total deformations of the
prisms in four direction moisture diffusion processes................................................... 127
XII
Figure 5.6: Arrangements of demec gauge points along the width at top (shown only) and
bottom surfaces of the prisms and at both ends through the depth as well as width of the
prisms for measurements of shrinkage strains along its width and total deformations of the
prisms in four direction moisture diffusion processes................................................... 129
Figure 6.1: A standard multi-crack segmental analysis ................................................ 133
Figure 6.2: Separating elements of RC beam (Concrete element) ................................ 135
Figure 6.3: Separating elements of RC beam (Reinforcement element) ....................... 135
Figure 6.4: Flexural properties (M/θ, M/χ and M/EI) ................................................... 136
Figure 6.5: Cracked segmental analysis ........................................................................ 137
Figure 6.6: Tension stiffening prism ............................................................................. 138
Figure 6.7: Local deformation of nth segment in prism ................................................. 140
Figure 6.8: Tension stiffening analysis ......................................................................... 141
Figure 6.9: Linear shrinkage strain over width and depth (Concrete element) ............. 146
Figure 6.10: Linear shrinkage strain over width and depth (Reinforcement element).. 146
Figure 6.11: Non-linear shrinkage strain over depth (Concrete element) ..................... 148
Figure 6.12: Non-linear shrinkage strain over depth (Reinforcement element)............ 149
Figure 6.13: Non-linear shrinkage strain over width and depth (Concrete element 2b1)
....................................................................................................................................... 150
XIII
....................................................................................................................................... 151
Figure 6.15: Non-linear shrinkage strain over width and depth (Reinforcement element)
....................................................................................................................................... 151
Figure 6.16: Influence of slice number along the width of beam on member deflection
....................................................................................................................................... 154
Figure 6.17: Predicted deflection of six beams tested by Gilbert and Nejadi (2004) ... 156
Figure 6.18: Influence of exposed surfaces on member deflection ............................... 157
List of Tables
Table 5.1: Specimen size details with V/S ratios for one up to four direction diffusion
processes ........................................................................................................................ 115
Table 5.2: Test details for material properties ............................................................... 116
XIV
Chapter 1 Introduction
1.1 Introduction
The versatility of concrete in terms of both application and material properties make it one
of the most widely used building materials in the world. When concrete is subjected to a
load, its deformation response, shown in Figure 1.1 is both immediate and time dependent.
The time dependent behaviour of concrete due to shrinkage and creep can produce
substantial time dependent deflections and increases crack widths. Significantly, the
increase in member deflection and crack width may lead to serviceability failures and
adversely affect the durability of RC structures.
Figure 1.1: Concrete strain component under sustained compressive stress (Gilbert and
Ranzi 2011)
Concrete shrinkage can be divided into four distinct phases: (i) Plastic shrinkage; (ii)
chemical shrinkage; (iii) thermal shrinkage; and (iv) drying shrinkage, which is the major
concern of this thesis. The major cause of each form of shrinkage is as follows:
1
 Plastic shrinkage occurs in wet concrete and is the result of capillary tension in the
water contained within the porous structure of concrete.
 Chemical shrinkage, which is also known as autogenous shrinkage occurs due to
chemical reactions during the cement hydration process. Chemical shrinkage is
independent on the size of specimen and the surrounding environment and it
occurs increasingly during the days and weeks after casting.
 Thermal shrinkage occurs due to the heat of hydration and thus gradually
dissipates in the first few hours or days after setting. Thermal shrinkage is
particularly important in the placement of mass concrete such as dams in which
significant heat is generated during hydration.
 Drying shrinkage occurs due to the loss of pore water throughout the lifespan of a
concrete structure, this form of shrinkage is caused by moisture movement
throughout the concrete and results in a reduction in the volume in concrete.
Drying shrinkage depends upon all the factors which affect the drying of concrete,
these include: the size and shape of members, the relative humidity of the ambient
environment and the mix characteristics of concrete. For example, high strength
concrete undergoes less drying shrinkage than normal strength concrete as these
concretes have a smaller quantity of free water after the hydration process is
complete. In contrast, chemical and thermal shrinkage can be significantly higher
in high strength concretes where increased hydration occurs. (Gilbert 1988, Gilbert
and Ranzi 2011).
Figure 1.1 shown above is a breakdown of the total concrete strain into its individual
components over time. In Figure 1.1 it can be seen that shrinkage strains begin to develop
2
during the drying process which begins at a time t = τd and that the magnitude of the
shrinkage strain is independent of the applied stress σc0 which is applied at time τ0.
Following loading at time t0 an instantaneous or elastic strain is developed due to the
application of stress σc0 and significantly this strain increases over time due to a variation
in the stiffness of the concrete. This variation is due to concrete creep the magnitude of
which depends on the applied load σc0. Concrete creep strains increases rapidly in the
period initially after application of load then the rate of increment slowed down
dramatically with time. After 2 -3 months of initial application of load produces about 50
percent of final creep strain whereas it reaches to about 90 percent in between 2-3 years
(Gilbert and Ranzi 2011).
However, for the serviceability analysis of an RC member it is only necessary to consider
drying shrinkage as the magnitude of autogenous, thermal and plastic shrinkage strains
are negligible in comparison (Holt and Leivo 2004, Gribniak, Kaklauskas et al. 2008).
Concrete time dependency is incredibly important in that it can lead to serviceability and
durability failures. Two major problems exist in the modelling of shrinkage: (i) how to
accurately defining the shrinkage strain (ii) how to incorporate shrinkage into the analysis
of a member. Issues surrounding (i) are the huge scatter of test results and which may be
due to the measurement methods and the prescription of a single shrinkage constant strain
to a block of concrete regardless of size. Issues surrounding (ii) arise because methods of
analysis commonly assume full interaction which does not occur in practice and hence fill
the gap between assumptions of analysis and experimental observations with empiricisms.
This means that the approaches do not simulate what is seen in practice and cannot directly
incorporate residual strains such as shrinkage. Moreover, they will have incredible
difficulty incorporating real material behaviours such as non-linear shrinkage.
3
In this thesis, the above mentioned issues are addressed by firstly simulating the
mechanics of drying shrinkage through a diffusion process and secondly simulating the
mechanics of concrete cracking using partial interaction (PI) mechanics allowing for the
real non-linear variation in shrinkage. The benefit of this approach is the only empiricisms
are the definition of material properties and the allowance for the real distribution of
shrinkage which is shown to be shape and size dependent.
Moisture diffusion in concrete greatly depends on pore relative humidity, member sizes
and concrete material properties. A moisture diffusion model to quantify the shrinkage
strain in concrete specimens been proposed by (Kim and Lee 1998, Kim and Lee 1999,
Mu and Forth 2009, Kang, Kim et al. 2011) who consider moisture diffusion as a nonlinear
diffusion problem. By using the nonlinear moisture diffusion theory it also requires a
moisture diffusion coefficient expressed as a function of moisture content can be
determined through the CEB FIP model code 1990 (CEB-FIP 1990) using concrete
material properties and also can be obtained from experiments (Akita, Fujiwara et al.
1997, Zhang, Hou et al. 2011).
Having addressed the issues surrounding the definition and quantification of non-linear
shrinkage variations in concrete member, this research aims to develop a moisture
diffusion model to predict the non-linear shrinkage variations along the concrete member.
Moisture diffusion from concrete emerges the shrinkage strain which shows non-linear
variations in practical experiments (Terrill, Richardson et al. 1986, Mu and Forth 2009,
Gilbert, Bradford et al. 2012) and it decreases with increasing of volume/exposed surface
area ratios (V/S) (Hansen and Mattock 1966, Almudaiheem and Hansen 1987). Current
design techniques suggest a uniform shrinkage profile throughout the member achieved
from an experimental test with smaller sizes of prism specimens compared to the real sizes
4
of RC members. Hence an experimental set up having different V/S ratios have been
proposed in Chapter 5 to show the influence of member sizes in shrinkage strain variation
in concrete member.
Having now modelled the mechanisms of drying shrinkage of concrete, this study also
aims to develop a member model for simulating the member deflection. Prior to flexural
cracking, the deflection of an RC beam can be derived from strain based approaches, such
as the use of flexural rigidities or moment-curvature, and it is fairly straightforward to
incorporate constant values of shrinkage (Visintin, Oehlers et al. 2013). This can be used
to predict the initial flexural crack and however once the initial flexural crack occurs then
the behaviour is governed by tension-stiffening as this partial-interaction behaviour
controls the crack spacing and crack widths that have a major effect on the deflection
(Oehlers, Mohamed Ali et al. 2011, Visintin, Oehlers et al. 2012, Visintin, Oehlers et al.
2013). A constant shrinkage strain, that is a longitudinal shrinkage strain that is constant
along both the depth and width of the member, can be added to these partial interaction
analyses (Visintin, Oehlers et al. 2013) to determine the effect of shrinkage. However,
numerical analyses and tests have shown that the shrinkage strain does vary along the
depth and width of the member (Terrill, Richardson et al. 1986, Mu and Forth 2009,
Gilbert, Bradford et al. 2012). The main aim of this study is to develop a numerical
procedure that allows for non-linear shrinkage strains as happens in practice and also to
determine whether a constant shrinkage strain is an adequate design simplification for the
non-constant shrinkage strains that can occur in practice.
5
1.2 Scope of the Research
The goal of the present study is to improve the accuracy of prediction in long term
deflection of RC members with incorporating the effect of creep and non-linear shrinkage
variations along both the depth and width of the member. In order to achieve this goal, a
numerical moisture diffusion model has developed to quantify the non-linear shrinkage
variations along the RC member for any sizes of member having any concrete material
properties with a variety of surface boundary conditions with any environmental
conditions. Hence a segmental approach has also been developed allowing for partial
interaction and non-linear shrinkage to quantify the long term deflection of RC members.
Hence a new form of complete experimental setup has proposed to quantify more
accurately the shrinkage variations through the depth and width of the member with
varying from one up to four different exposed surface conditions and also taking into
account the size effect according to the volume over exposed surface (V/S) area ratios.
1.3 Aims and Objectives of the Research
The main objectives of this research are outlined below:
 Develop a numerical diffusion model using the finite difference method to
quantify the non-linear variations in shrinkage strains throughout a RC member
for any environmental conditions with a variety of surface boundary conditions
and for any sizes of member and with any concrete material properties.
 Develop a mechanics based model for quantifying the long term behaviour of a
flexural member subjected to a sustained load. This approach is to be based on the
segmental approach previously developed at the University of Adelaide and in this
6
thesis the model is to be extended to allow for the non-linear shrinkage variations
along both the width and depth of the member.
 Proposing a new form of experiment to better quantify the shrinkage variations
within a member with varying from one up to four different exposed surfaces and
also considering the size variations with respect to the volume over exposed
surface area ratios (V/S) of the member.
1.4 Structure of the Thesis
This thesis comprises of seven chapters as also outlined below:
Time Dependent Flexural Analysis of Reinforced Concrete (RC) Members
Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6
Simulating Long term

Journal
Shrinkage Proposed Beam
Literature Paper on
Introduction Strain using Experimental Deflection
Review Non-linear
Moisture Work using
Shrinkage
Diffusion Segmental
Approach
Chapter 7 – Conclusions and Recommendations for Future Research
Figure 1.2: Thesis layout of this research
7
Chapter 1 provides the brief research background, significance of this research, aims
and objectives of this study.
Chapter 2 gives the detailed literature review on shrinkage and creep and its influence
on structures. Code and research based shrinkage strain prediction techniques of
varying complexity are critically reviewed.
Chapter 3 contains the journal paper on ‘Mechanics of Simulating the Serviceability
Deflection of RC Beams allowing for Partial-interaction and Non-linear Shrinkage’
where a novel technique has been developed to quantify the long term deflection of
RC beams allowing for the inclusion of non-linear shrinkage variations in concrete.
Chapter 4 describes the development and application of a moisture diffusion
modelling to simulate drying shrinkage of concrete subjected to any environmental
conditions. This chapter also includes a validation of the diffusion model with existing
experimental test results.
Chapter 5 provides recommended changes to shrinkage strain experimental tests for
concrete specimens which will allow for the determination of non-linear shrinkage
strain distributions.
Chapter 6 deals with the development of a new segmental approach to quantify the
long-term deflection of RC members associated with the effect of creep and non-linear
shrinkage strain variations of concrete members along both the depth and width of the
member. It also illustrates the parametric study on variation of long-term deflection of
RC beams with the effect of different exposed surfaces and also the validation of
numerical segmental model with available experimental test results of six RC beams.
Chapter 7 summarises the major findings of this study and recommends the scope and
opportunities for further research.
8
Chapter 2 Literature Review
Introduction
This chapter reviews literature surrounding the mechanisms for concrete shrinkage and
creep as well as existing prediction methods applied in practice. Prediction of shrinkage
strain in concrete using ten different codes and models including the most up to date
prediction techniques to predict the shrinkage strain from experimental test results are also
covered. Special attention is paid on the loss of moisture in concrete i.e. moisture diffusion
as this is the main mechanism driving long term drying shrinkage in structures. Finally,
this chapter reviews the most noteworthy literature on numerical and analytical modelling
to quantify long term deflections of reinforced concrete (RC) members such as beams or
slabs due to the effect of creep and shrinkage.
2.1 Shrinkage and its types
Shrinkage is the volumetric changes of concrete due to loss of water from the concrete
into the atmospheric environment or by internal chemical reactions. Concrete shrinkage
is a time-dependent strain and occurs in an unloaded and unrestrained specimen. There
are various types of shrinkage and that occurred in two stages one is in early ages (< 24
hours) and another one is in the long term (> 24 hours).
9
SHRINKAGE
Early Age (<24 hours) Long Term (>24 hours)
Drying Autogenous Drying Autogenous
Thermal Thermal Carbonation
Figure 2.1: Shrinkage stages and types of concrete
According to Gribniak, Kaklauskas et al. (2008), Gilbert and Ranzi (2011) and Lam
(2002) shrinkage of concrete can be considered to consist of four major forms: (i)plastic
shrinkage, (ii)chemical or autogenous shrinkage, (iii)carbonation shrinkage and (iv)
drying shrinkage. Plastic shrinkage occurs in the fresh concrete due to loss of moisture
from the freshly poured concrete to its surroundings whereas autogenous, carbonation and
drying shrinkage all occur in hardened concrete after setting. Autogenous or chemical
shrinkage occurs at the early stages due to chemical reactions within the cement paste and
binder includes the hydrations of cement without any moisture movement outside to the
environment. Carbonation shrinkage occurs due to the chemical reaction with carbon
dioxide in the air with various products of cement hydration. Thermal shrinkage is another
type of shrinkage which occurs in first few hours or days after setting time as the heat of
hydration dissipates gradually and the term endogenous shrinkage is used to refer to the
part of shrinkage of hardened concrete which is not related with drying that means the
sum of autogenous and thermal shrinkage. According to Holt and Leivo (2004), shrinkage
10
types and stages are as illustrated in the above Figure 2.1 where drying shrinkage in early
ages has a negligible effect compare to the long term.
Drying shrinkage in concrete increases with time at a gradually decreasing rate and takes
place in the long run even along the life time duration of a structure. The magnitude and
rate of development of drying shrinkage depend on all the factors that affect the drying of
concrete including the atmospheric relative humidity, the mix proportions such as water
content, water to cement ratio, types of cement, aggregate types, fine to coarse aggregate
ratio and also the shape and sizes of the member.
In terms of analysis of concrete structures two types of shrinkage are need to be
considered: drying shrinkage and autogenous shrinkage (Holt and Leivo 2004, Gribniak,
Kaklauskas et al. 2008). The ratio of drying and autogenous shrinkage is illustrated in the
Figure 2.2 below (Sakata and Shimomura 2004) and it can be found that drying shrinkage
is the governing one should be considered and autogenous shrinkage is negligible compare
to the total long term shrinkage after drying and which is only 10% to 20% of the long
term shrinkage strain (Silliman and Newtson 2006).
Figure 2.2: Shrinkage strain components in normal strength concrete (Sakata and
Shimomura 2004 cited in Gribniak et al., 2008)
11
2.2 Effect of Shrinkage and Creep on Structures
Reinforced concrete members such as beams or slabs are embedded with reinforcements
and also supported with joints which may provide the restraining effect to the shrinkage
and can produce shrinkage induced curvature in an unsymmetrically reinforced concrete
member. Hence, this shrinkage induced curvature can often leads to the significant
deflection of the member and which is load independent. Therefore, shrinkage is one of
the major concerning factors and that affects the time dependent deflection in RC flexural
members (Jayasinghe 2011, Jafarifar 2012, Gilbert and Ranzi 2011).
The effect of shrinkage on structural members depends on the drying conditions at the
exposed surfaces of RC members where the shrinkage profile should considered to be
non-linear during the long term analysis of concrete structures (Terrill, Richardson et al.
1986, Mu and Forth 2009, Gilbert, Bradford et al. 2012) and will be discussed briefly at
later sections of this chapter.
Creep of concrete can be defined as the time dependent strain in hardened concrete
subjected to sustained stress and this is both time dependent and stress dependent. Creep
is obtained normally by subtracting instantaneous strain after application of load and
shrinkage strain in an unloaded specimen from the measured total strain with the change
of time in a loaded specimen. Creep can be classified into two types, one is basic creep
and another one is drying creep. Basic creep is the time dependent deformation which
occurs in a loaded specimen without any moisture movement from the specimen to the
surrounding environment. Hence the drying creep is the additional creep which occurs in
a drying specimen due to the moisture movement between the specimen and the
environment (Gilbert 1988, ACI209R-92 1992, He 2013).
12
The magnitude and the rate of development of creep are influenced by many factors such
as concrete mix properties, aggregate type and sizes, water to cement ratio in the concrete,
environmental humidity and temperature, member sizes and the loading conditions.
Creep decreases with increase in concrete strength and better quality of concrete provides
lesser amount of creep. Creep also decreased with increment of aggregate content,
maximum aggregate sizes and using of stiffer aggregate in the concrete mixes. Reduced
water cement ratio uses in concrete mix composition may reduce the creep of concrete.
The magnitude of creep depends on the age of concrete and it decreases as the age at first
loading increases. Creep decreases as the relative humidity increases and shows vice
versa. Temperature rises also increases the creep as the deformability of cement paste is
increased by a temperature rise and drying being accelerated.
Creep is also dependent on the sizes of the members and it increases in thinner specimen
such as in a thin slab specimen as surface area to volume ratio increases. Creep is mainly
dependent on the stress level and when the sustained stress is less than half of the
compressive strength of concrete the creep strain is proportional to the stress level and is
known as linear creep (Gilbert 1988, Gilbert and Ranzi 2011).
There are several methods to analyse the creep effects on structural members are available
which includes the effective modulus method (EMM), the age adjusted effective modulus
method (AEMM), the rate of creep method (RCM) and the rate of flow method (RFM)
and among these first two of them used widely for the analysis of structures with regards
to the effect of creep (Jayasinghe 2011). However, in this thesis we will discuss about the
EMM which has been used to analyse the creep effect in RC members to quantify the long
term deflections.
13
The effective modulus method (EMM) suggested by Faber (1928) is the simplest and
oldest method applied in this research and the elastic modulus of concrete in a time t after
first loading at time t0 at any point can be defined as Ec (t, t0) and using effective modulus
method can be written as
𝐸 (𝑡, 𝑡 )
𝐸𝐶 (𝑡, 𝑡0 ) = 1 +𝑐𝜑(𝑡,0𝑡 Equation 2.1
0)
Where t0 is the time at initial loading and φ is the creep coefficient of concrete at time t
after initial loading at time t0. Hence the change in elastic modulus in concrete due to creep
would be applied to both in compression and tension zone of concrete at the segmental
analysis in chapter 5 for time dependent deformations of the RC members (Gilbert and
Ranzi 2011).
2.3 Prediction of Shrinkage Strains
Several models have been developed to predict shrinkage strains in concrete based on the
results of experimental studies by the prominent researchers. These models have been
adopted by codes of practices in different countries can be named as code based model
and rest of them are remained as research based model which also can be used to predict
the shrinkage strain in concrete specimens. These model including ACI 209R-92 model,
ACI model modified by Huo, AS 3600-2009, AASHTO LRFD 2012 model, Bazant-
Baweja B3 model, CEB MC90, CEB MC90-99, FIB 2010 model, GL2000 model and
Sakata 1993 model are been discussed and compared with experimental shrinkage test
results from Bazant NU-ITI database (Bazant and Li 2008) and shrinkage test results of
concrete prisms by Al-Saleh and Al-Zaid (2006).
14
Concrete deformations due to shrinkage is quite difficult to predict as the actual behaviour
of concrete with the effect of shrinkage are very complex and involves several physical
mechanisms; also influenced by so many parameters associated with it. Numerous
research has been done to quantify and predict the deformation of concrete due to
shrinkage effect and also various analytical and empirical models have been developed to
predict the shrinkage strain in concrete specimens (Al-Saleh 2014). Ten of the most
utilised shrinkage prediction models including the current code of practices are described
and also considered for the comparison with experimental test results of two rectangular
concrete prisms, one is plain concrete prism and the other one is standard prism having
sizes of 80 mm × 150 mm × 500 mm and 50 mm × 50 mm × 300 mm respectively (Al-
Saleh and Al-Zaid 2006).
Figure 2.3 below represents the measured and calculated shrinkage strains in a rectangular
plain concrete prism of sizes 80 mm × 150 mm × 500 mm and which are cured in a
humidity and temperature control room having humidity of 50% and temperature of 280C.
It can be seen from the figure that the theoretical shrinkage strains calculated by using
ACI 209, ACI 209 model modified by Huo and AASHTO LRFD 2012 model provide
well prediction of shrinkage strain. Whereas CEB FIP MC90, CEB MC90-99, FIB 2010,
B3 model shown similar trend by achieving good correlation with measured values at the
beginning of drying but underestimated at later ages. GL2000 and Sakata 1993 model
exhibit closer estimation to the test results initially but as drying period increases the
shrinkage strain also increases with increasing rate and finally overestimates the
experimental shrinkage strains. AS 3600-2009 is found to have the worst approximation
from early age up to the end of drying.
15
Figure 2.4 depicts the comparison between measured and calculated shrinkage strains in
standard concrete prism having sizes of 50 mm × 50 mm × 300 mm stored in a similar
environmental conditions with the rectangular concrete prism. Similar trends for
shrinkage strains in rectangular plain concrete prism can be seen in this standard specimen
except for the GL2000 model shows higher estimation than the Sakata 1993 model
compared to rectangular plain prisms and however, both of them provide the
overestimation of shrinkage strain. Hence finally, AS 3600-2009 gives the poorest
approximation of shrinkage strain among all the models and code approaches.
900
800
700
Shrinkage strain in micorstrain, mm/mm
600
500
400
Experimental results
ACI 209 model
300 CEB FIP 90

B3 model
Sakata model adopted in JSCE 1996
200 GL 2000 model
AS 3600-2009
CEB MC 90-99
100 FIB 2010 model
ACI modified by Huo (2001)
AASHTO (2012)
0
0 25 50 75 100 125 150
Time elapsed in days
Figure 2.3: Experimental and predicted shrinkage strain in a plain rectangular concrete
prism (80 mm × 150 mm × 500 mm) using different codes and shrinkage prediction
models
16
1000
900
800
Shrinkage strain in micorstrain, mm/mm

700
600
500
Experimental data
400 ACI 209 model
CEB MC90
B3 model
300
Sakata model adopted in JSCE 1996
GL 2000 model
200 AS 3600-2009
CEB MC 90-99
FIB 2010
100
ACI 209 modified by Huo (2001)
AASHTO 2012
0
0 25 50 75 100 125 150
Time elapsed in days
Figure 2.4: Experimental and predicted shrinkage strain in a standard concrete prism (50
mm × 50 mm × 300 mm) using different codes and shrinkage prediction models
2.4 Using Moisture Diffusion to quantify Shrinkage
Shrinkage occurs in the concrete mainly due to moisture diffusion from inner surface of
the concrete to the outside environment due to the effect of ambient relative humidity.
Many researchers have investigated the moisture diffusion of concrete experimentally and
numerically will be described in this chapter. Hence due to the moisture diffusion,
concrete exhibits the shrinkage strain which can be extracted by applying a mathematical
relationship in between them.
17
2.4.1 Moisture and Humidity Diffusion in Concrete
Moisture diffusion is affected by moist curing period of concrete and it shows higher
diffusion rate for poorly cured concrete and lower diffusion rate for properly cured
concrete. Since self-desiccation has significant role in the early ages of concrete so this
need to be considered especially for high strength concrete. Furthermore, moisture
diffusion progresses more rapidly at high temperature compare to the low temperature
environment conditions. Both finite difference and finite element method using moisture
diffusion equation has shown good agreement with experimental results for one direction
and two directional moisture diffusion processes (Kim and Lee 1999, Kang, Kim et al.
2011).
An experimental and numerical simulation of one direction moisture movement (exposed
top surface only and rest of the surfaces are sealed) modelled for 180 mm ×150 mm ×150
mm plain conventional concrete (CC), plain rolling compacted concrete (RCC) and steel-
fibre-reinforced (SFR) concrete. This investigation was performed in a chamber of
constant relative humidity and temperature of 40±3% and 25±3ºC respectively and
gravimetric method was used for measuring the moisture content inside the concrete
specimen. ABAQUS 3D 8-noded solid element called DC3D8 was used for finite element
(FE) simulation with the experimental test results and moisture diffusion coefficient and
surface factors were back-calculated by using FE analysis. It can be seen that higher
moisture diffusivity achieved when moisture content is above 90% and surface factor does
not have significant effect except near the drying surface of the concrete specimen.
Moisture diffusivity and surface factor varies from 0.001 to 30 mm2/day and 3 to 10
mm/day for plain CC and 5-10 mm/day for plain RCC. Figure 2.5 below depicts the
comparison of experimental and numerical non-uniform moisture profiles for four
18
different types of concrete. It can be found that moisture content increases through the
depth of the concrete block away from the drying top surface and decreases over the drying
period in days. Hence it can also be seen that numerical model results shown good
agreement with experimental test results (Jafarifar 2012).
Figure 2.5: Numerical moisture profiles compared with the experimental test results: (a)
Plain CC mix; (b) Plain RCC mix; (c) SFR-CC mix (2.5%); (d) SFR-RCC mix (2.5%)
(Jafarifar 2012)
19
Zhang, Qi et al. (2009) also performed experimental investigation with 200 mm × 200
mm × 800 mm concrete prism for measuring the relative humidity at early ages in various
depths starting from 2.5 cm to 18 cm exposed from drying top surface only and rest of the
five surfaces are sealed. A numerical model considering both the cement hydration and
moisture diffusion was also developed. It can be found that good agreement achieves with
numerical modelling and finite difference method and which well predicts with the
experimental results. Furthermore, the model results revealed that self-desiccation is more
significant in high strength concrete compare to the normal strength concrete in the
reduction of relative humidity. It can also be seen that humidity reduction is mainly due
to the moisture diffusion for normal strength concrete but for high strength concrete, it is
affected by both cement hydration and moisture diffusion together. Moisture diffusion
coefficient is a critical parameter to identify and is required for measuring the changes of
internal relative humidity due to water movement in concrete. Therefore, Zhang, Hou et
al. (2011) determined the moisture diffusion coefficient at early ages for low and high
strength concrete. It can be found that this coefficient decreases with increasing of drying
periods and provide higher values for low strength concrete compare to the high strength
one. Moisture diffusion coefficient varies from 8×10-8 to 7×10-10 m2/s for low strength
concrete and 2×10-8 to 4×10-10 m2/s for high strength concrete from 3 to 28 days drying
period after concrete casting.
However non-linear shrinkage exhibits with the effect of non-uniform moisture loss and
it increases linearly with the increment of moisture losses in concrete. The exposed
surfaces of concrete provide higher shrinkage than the inner part as because of higher
moisture lost from the surface layer compare to the central region of concrete.
Volume/surface ratio has a significant effect in shrinkage of concrete which reflects larger
20
the specimen size slower the moisture loss and hence lower the shrinkage strain and vice
versa. Non-linear diffusion theory demonstrates non-linear diffusion of concrete and
moisture diffusivity greatly depends on the pore relative humidity (RH) of concrete.
Experimental investigation revealed to express the diffusion coefficient as a function of
moisture content and the diffusion coefficient value of 0.23cm2/day and 0.30cm2/day was
suggested by different authors (Pickett 1946, Bazant and Najjar 1972, Sakata 1982, Mu
and Forth 2009). Moisture diffusion coefficient is also been determined by Akita, Fujiwara
et al. (1997) with respect to the relative mass decreases during certain drying period for
one-face and six-face drying of prismatic specimens and they revealed that the diffusion
coefficient was a function of moisture content.
Hence Bažant and Najjar (1971) have proposed non-linear diffusion theory over the linear
theory which can predict drying more realistically. They also provide mathematical
formulation of moisture/pore relative humidity dependent moisture diffusivity/ diffusion
coefficient which is then adopted by CEB-FIP model code applied in this study and also
used by other researchers (Kim and Lee 1998, Kim and Lee 1999, Kang, Kim et al. 2011)
to incorporate with their moisture diffusion model. This moisture diffusion coefficient
decreases sharply at the early stages of drying in between 90 to 60% of pore humidity
while it approaches approximately constant values below 60% of pore humidity inside the
concrete specimens. However, there is a scarce in literature to predict the variation of
moisture or relative humidity inside the concrete specimens with effect to the exposed
surfaces and ambient humidity and therefore more extensive research is required to
achieve the non-linear moisture profile in concrete with varying the specimen sizes and
surface conditions. Hence this moisture profile can be used to quantify the shrinkage
variations in concrete members and will be discussed in later sections.
21
2.4.2 Shrinkage strain in Concrete
Diffusion of moistures in concrete which varies with space and time yielded shrinkage
strain and it provides non-uniform shrinkage profile along the depth of the concrete from
exposed surfaces of concrete structures. Drying shrinkage strain was measured in a
concrete specimen having size of 30 cm × 30 cm × 15 cm. Five sides of the specimen were
covered with paraffin wax to avoid the moisture losses. Embedded strain gauges were
used to measure the strain along the exposed surface of the specimen at 2 cm, 5 cm, 8 cm
and 12 cm. Concrete specimen was kept at a constant temperature and humidity of 20±1ºC
and 68±2% RH after being moist cured for 7 days. Two types of concrete, one is with
admixture and another one is without admixture having the compressive strength of 44
and 28 MPa respectively was used for performing the test.
From Figure 2.6 (a) and (b), it can be seen that drying shrinkage strain varies significantly
along the depth from exposed surface and the shrinkage strain decreases from the exposed
surface to the inner part of the specimen. It can also be seen that the shrinkage strain
exhibits slow increment inside the concrete while it increases quickly near the exposed
surface of the specimen.
22
Figure 2.6: Shrinkage strain variation along the thickness of concrete specimen for two
different mixes and comparison with the analytical results (Kim and Lee 1998)
The analytical calculation also been performed with consideration of the creep of concrete.
Hence the suggested analysis method able to quantify the differential drying shrinkage
strain which is then compared with the test results. Thus the analysis method reflects well
agreement with the test results. Furthermore, it has been recommended by the authors to
consider the differential drying shrinkage strain for the analysis of thick structures and the
method of analysis with the diffusion equation and embedded strain gauges is also suitable
for measuring the differential drying shrinkage strain (Kim and Lee 1998).
Gilbert, Bradford et al. (2012) has also performed experimental investigation of shrinkage
strain for composite concrete slab specimens for a period up to 322 days. A total of 10
specimens divided into two categories, type A (with restrained using steel deck) and type
B (without restrained). Three different types of deck profiles which are re-entrant profile
RF55 and two trapezoidal or wave-form profiles KF40 and KF70 were used. Vibrating
wire strain gauges placed through the thickness of the slab were used to measure the
concrete strain and a DEMEC gauge was used to measure the strains by placing at the
upper and lower surface of the specimen. According to the test results, it can be seen that
23
the steel deck has provided significant restraint effect compare to the unrestrained one. An
analytical calculation using age adjusted effective modulus method considering non-linear
shrinkage profile for restrained specimen was performed to measure the shrinkage-
induced long term deflection and curvatures. The predicted concrete stress at bottom of
each slab is found almost 60% of the flexural tensile strength of concrete and that
quantified the significant reduction of cracking moment. Hence it can be concluded that
analytical method has shown excellent agreement with the experimental results.
According to Ong, Chandra et al. (2010) early age shrinkage strains varies along the depth
of mortar specimens with using image analysis technique. Two types of specimens having
the size of 75 mm × 75 mm × 285 mm and 100 mm × 100 mm × 285 mm were casted
with three different water/cement (w/c) ratios of 0.25, 0.30 and 0.35. Shrinkage strains
was measured from the top (trowelled/ exposed) surface along the depth (3 mm, 20 mm,
40 mm and 60 mm) of the mortar prism specimen started 30 minutes after adding water
and end up to 24 hours with using the innovative image analysis technique for both sealed
and unsealed specimen. The temperature and relative humidity during the whole test
period was kept constant at 30 ± 0.5ºC (86 ± 0.9ºF) and 65 ± 2%. From the test results, it
can be found that shrinkage strains along the depth of the specimen shows minor
differences for the sealed specimens but it has shown significant variation for the unsealed
specimens. In unsealed specimens, shrinkage strains decreased from the exposed top
surface to the depth of the mortar specimens. Moreover, it can be also seen that higher
shrinkage strains occurred to the lower w/c ratio specimens compared to the higher one.
That means high level of early-age shrinkage strains achieved for the lower w/c ratios
mortar specimens and vice-versa. However, a limited number of research has been found
on non-linear shrinkage behaviour in concrete and hence a further investigation is required
24
to quantify the non-linear shrinkage profile in concrete specimens with varying the sizes
and surface boundary conditions.
2.4.3 Correlations in between with Moisture loss, Humidity and Shrinkage strain
of Concrete
Shrinkage strain is directly related to the moisture/ humidity loss of concrete to the
atmospheric environment and many researchers investigated the relationship in between
them to quantify shrinkage strain from moisture /humidity loss of concrete will be
discussed in this section.
Figure 2.7 (a) and (b) below represents the relative humidity at 3 cm, 7 cm and 12 cm
distance of 10 cm × 10 cm × 20 cm prism specimens measured after 3 and 28 days moist
curing. These Figures shown the relative humidity occurred only for moisture diffusion
that been achieved by subtract the humidity due to self-desiccations. It can be found that
the effect of water/cement ratio on moisture diffusion is substantial and high water/cement
ratio concrete shows higher diffusion rate compare to the low water/cement ratio concrete.
Numerical finite element analysis been performed using non-liner moisture diffusion
theory and maximum moisture diffusion coefficient (which provides the best fit with
experimental results) which shows well agreement with experimental results.
25
(a)
(b)
Figure 2.7: Calculated relative humidity compared to modified experimental results due
to moisture diffusion only a) curing period = 3 days b) curing period = 28 days (Kim and
Lee 1999)
Hence the linear relationship between humidity and moisture diffusion in concrete and it
exhibits significant effect on moist curing periods and which is illustrated in Figure 2.8.
Internal relative humidity occurred only due to moisture diffusion varied slowly at each
location for 28 days moist curing compare to the 3 days moist cured concrete. Concrete
microstructure is affected by curing periods which effects the moisture diffusion as well.
26
Figure 2.8: Relationship between relative humidity and moisture diffusion on different
moist curing period (Kim and Lee 1999)
Moisture loss has direct relationship with moisture diffusion and indirect relationship with
self-desiccation. Hence according to Figure 2.9, high water/cement ratio concrete provides
higher amount of water loss in terms weight per unit exposed area compare to the low
water/cement ratio one. Moist-curing time has also significant effect on moisture loss of
concrete and it can be found that 3 days moist-cured concrete gives higher moisture loss
as because of high amount of evaporable free water content in capillary pore with
comparison of 28 days moist cured one (Kim and Lee 1999)
27
Figure 2.9: Loss of moisture in concrete due to drying (Kim and Lee 1999)
Finite difference method using moisture diffusion coefficient considering the concrete
porosity and temperature effect applied to compare with experimental results performed
by Kim & Lee, 1999 as illustrated in Figure 2.10. Gradient boundary conditions were used
using environmental humidity, surface humidity and surface factor of concrete. It can be
seen that numerical investigation shown good agreement with experimental results as
shown in Figure 2.10 (Kang, Kim et al. 2011).
28
Figure 2.10: Comparisons between experimental and numerical results (ambient
temperature 200 C): a) Exposed at 3 days (w/c = 0.28) b) Exposed at 28 days (w/c =
0.28) c) Exposed at 3 days (w/c = 0.40) d) Exposed at 28 days (w/c = 0.40) e) Exposed
at 3 days (w/c = 0.68) f) Exposed at 28 days (w/c = 0.68) (Kang et al 2012)
Hence the relationship between shrinkage and relative humidity for both cement paste and
mortar specimens represents in Figure 2.11. It can be seen that shrinkage has linear
relationship with relative humidity in between 48 to 100% for cement pastes. For mortar
specimens, the slope between 48 to 75% is lower and then it exhibits increment in between
the relative humidity range of 75 to 100% and the effect of water/cement ratio is
negligible. So it can be said that drying shrinkage is inversely proportional to the
environment relative humidity. Hence it can be observed that drying shrinkage is
approximately proportional to the water loss of specimens as shown in Figure 2.12
(Bissonnette, Pierre et al. 1999).
29
Figure 2.11: Relationship between ultimate shrinkage and relative humidity for cement
paste and mortar specimens (Bissonnette, Pierre et al. 1999)
Figure 2.12: Relation between shrinkage and weight loss for various types and sizes of
paste, mortar and concrete specimens (Bissonnette, Pierre et al. 1999)
30
(a)
(b)
31
Figure 2.13: Relationship between free shrinkage strain and interior relative humidity
(RH) of a) C30 concrete and b) C80 concrete (Zhang, J, Dongwei and Wei 2010)
Relationship between free shrinkage strain (εw) and interior relative humidity of concrete
has shown in above Figure 2.13 for two different types of concrete having 28 days
compressive strength of 34 MPa (C30) and 88 MPa (C80). Free shrinkage results can be
split into two parts whereas free shrinkage induced at stage I (RH = 100%) and free
shrinkage induced at stage II (RH < 100%). Linear relationship between shrinkage strain
and internal relative humidity can be achieved at stage II for both C30 and C80 concrete
and the relationship can be expressed as
εw = ε0 + k (100-RH) for RH < 100% Equation 2.2
Where εw is the free shrinkage strain, ε0 is the shrinkage at stage I (RH = 100%), k is the
shrinkage strain occurred due to one-unit humidity reduction. For this case, ε0 = 197 &
367 µm/m and k = 15.8 & 6.4 µm/m per percentage for C30 and C80 concrete
respectively.
32
Figure 2.14: Calculated free shrinkage strains and measured relative humidity at
different depths of the slab from exposed surface against drying period (Zhang, J,
Dongwei and Wei 2010)
Figure 2.14 displays above the analytical free shrinkage strain calculated by using
Equation 2.2 and internal relative humidity measured at various depths of 2.5, 5, 10 and
18 cm of 200×200×800 mm slab specimens sealed at five sides except top surface against
moisture movement for two types of concrete C30 and C80. Significant variation of
humidity reduction and shrinkage strain exhibits through the depth of slab. It should be
noted that shrinkage strain has been achieved only for stage II (RH < 100%) by using
Equation 2.2 but moisture induced shrinkage also occurred at stage I when RH = 100%.
Therefore, author recommend for further investigation in complete exploration of non-
linear relationship between internal relative humidity and shrinkage strain of concrete
(Zhang, Dongwei et al. 2010).
A series of experimental tests for different types of concrete (low, medium and high
strength) in sealed and dry conditions of concrete prism been performed for validation of
a theoretical model that including pore correction factor, developed both for autogenous
and drying shrinkage strain predictions and it provides well agreement with the
experimental results (Jun, Dongwei et al. 2010, Zhang, Hou et al. 2012, Zhang, Hou et al.
2012).
Experimental investigated moisture loss at a depth of 1 cm from drying surface is plotted
against free shrinkage strain in Figure 2.15. Hence it can be seen that the relationship
between free shrinkage strains and moisture loss in non-linear. It can be expressed by this
Equation based on a regression model:
(εsh) M = [M2.5/ (195 + M2.5)] × (εsh)ult Equation 2.3
33
This Equation is very similar to the form of ACI committee 209 Equation for the variation
of shrinkage with time (ACI209R-92 1997). Here, εsh is the free shrinkage strain, M is the
loss of moisture in percentage starting from 0% and (εsh)ult is the ultimate free shrinkage
strain which is considered 1600 micro strain for this case. The free shrinkage strain can
be able to predict with using Equation 2.3 for different values of moisture loss in concrete
(Asad 1995).
Figure 2.15: Relationship between free shrinkage strain and moisture loss (ACI209R-92
1997; Asad, M 1995)
Suwito, Ababneh et al. (2006) has performed analytical and numerical investigation on
moisture diffusivity and its effect on drying shrinkage of concrete. Finite element method
been used for the analysis of concrete slab specimen having a depth of 20 cm and 100 cm
long. The slab is considered to be fully saturated at the initial stage (RH = 100%) and
34
environmental humidity at the exposed top surface is 50% (rest of the surfaces are sealed).
It can be seen from the finite element mesh consisting of 3000 quadrilateral elements that
RH decreases from sealed surface to the exposed surface along the depth of the slab over
the period of 365 days and its converging to the boundary conditions at RH = 50%.
Shrinkage strain is increased through the depth from sealed surface to exposed face over
the total period of a year. Two types of curves have shown one is damaged and another
one is undamaged. Damage curve has given higher effect than the undamaged one. As
because drying shrinkage occurs due to moisture loss and it generates damage. In
consequence, damage increased the concrete diffusivity which accelerates the moisture
dissipation. Bazant, Sener et al. (1987) has also performed experimental and analytical
studies cracking effect on drying permeability and concrete diffusivity. They found that
cracked specimen dried more rapidly than the uncracked one and simultaneously cracked
specimen showed more weight loss compared to the uncracked one. Granger, Torrenti et
al. (1997) has developed a model which is in good agreement with shrinkage as a function
of moisture loss. Previous experimental result reveals that concrete loses its weight with
the effect of drying shrinkage and it shows non-linear relationship between shrinkage
strain and weight loss of concrete.
35
Figure 2.16: Experimental results of different concrete specimens drying shrinkage as a
function of weight loss (Granger, Torrenti and Acker 1997)
Above Figure 2.16 represents the drying shrinkage as a function of weight loss for six
different types of concrete having 28 days compressive strength of 30 MPa to 65 MPa.
Concrete specimens of having 16 cm diameter and 100 cm height are kept at controlled
ambient conditions of 50±5% RH and 20º±1º C. End of the specimens were sealed with
self-adhesive aluminium layers for preventing any loss of moisture. Only radial drying
was allowed and the shrinkage measurement was taken at the middle of 50 cm of the
cylinders. Drying shrinkage was achieved by subtracting the autogenous shrinkage
measured from completely sealed specimen by two self-adhesive aluminium layers to
prevent drying and kept under the same environment conditions. Shrinkage test was began
after 28 days when the cylinders were covered with a polyane film and itself covered with
two self-adhesive aluminium layers to prevent from moisture before 28 days. Among all
the six types of concrete, if we closely concentrate on Penly concrete having 28 days
compressive strength of 24.3 MPa; it can be seen that 1% weight loss occurred at the
36
drying shrinkage strain value of 100 micro strains. Concrete is gradually losing its weight
as the shrinkage strain increases and the graph shows linear behaviour up to the value of
2% weight loss then non-linear relationship exhibits to the maximum drying shrinkage of
400 micro strains at an amount of 2.7% weight loss of concrete.
However, the free shrinkage strain of concrete is a function of pore relative humidity and
can be described using Equation 2.4 from Bažant and Yunping (1994) has been used in
this current study.
𝜀𝑠ℎ = 𝑘𝑠ℎ 𝑓𝑠 (ℎ) = 𝑘𝑠ℎ (1 − ℎ); 𝑘𝑠ℎ = 𝜀𝑠0 𝑔𝑠 (𝑡) Equation 2.4
Where, 𝑘𝑠ℎ is the shrinkage coefficient; 𝜀𝑠0 is the ultimate shrinkage strain and 𝑔𝑠 (𝑡) is
𝐸(𝑡0)
the ratio of elastic modulus with time, i.e., 𝑔𝑠 (𝑡) = and E(t) can be estimated by
𝐸(𝑡)
𝑡
𝐸(𝑡) = 𝐸(28)√4+0.85𝑡 with 𝐸(28) = 33𝑤 1.5 √𝑓𝑐′
2.5 Member Behaviour
The shrinkage induced deformation affects the serviceability of structures by producing
initial cracks even before application of loading and increases the crack widths after
loading as drying commences that accelerates shrinkage and finally enhances shrinkage
induced excessive deformations in a concrete member at a later period and which may
cause the serviceability failure of a RC member.
2.5.1 Models to predict long term Deflections
There are several models to quantify long term deflections of RC members such as beams
or slabs. The prediction of long term deflections in a concrete member depends on so
many parameters including nonlinear concrete material properties, history of loading,
flexural cracking, time dependent effects due to creep and shrinkage, loading ages in
37
concrete and reinforcement ratios. Some codes such as AS 3600-2009 (Standards
Australia 2009), CEB FIP model code 90 (CEB-FIP 1990) and ACI 318-08 (American
Concrete Institute 2008) provide simplified method to quantify instantaneous and long
term deflection of RC members.
The instantaneous deflection using these methods are computed with using effective
second moment of area, Ieff or mean curvature of the member, Km and that are provided by
empirical equations. The long term deflection calculation due to creep and shrinkage can
be determined from the material properties of concrete i.e. design shrinkage strain and
creep coefficient of concrete and the principles of mechanics. The other simplified method
to quantify the time dependent long term deflection is to multiply the instantaneous
deflection by a deflection multiplying factor λ which is also given by an empirical
equation.
These simplified methods do not consider the ageing of concrete, loading history, bond
slip behaviour between steel and concrete, flexural cracking variation with time, tension
stiffening behaviour of concrete prism and steel in the cracked region. Hence according
to Gilbert (1999), the present simplified method may not be able to calculate the
instantaneous and time dependent long term deflection perfectly especially for lightly
reinforced concrete beams and slabs. Hence, there is a need to provide a better deign
procedures to predict the short and long term deflections of RC members with
incorporation of ageing of concrete, loading history, bond slip characteristics between
steel and concrete, crack propagation with time, steel in between cracked concrete need
to be considered (Gilbert and Ranzi 2011, Jayasinghe 2011).
Numerical models are also been developed to compute the long term deflection of
reinforced concrete beams or slabs where non-linear material properties of concrete,
38
flexural cracking, tension stiffening and bond slip behaviour between reinforcement and
concrete are been considered in the analysis. In such analysis RC member is divided into
elements and each elements containing flexural cracks been initiated. In this method the
RC member is divided into block element and each block is having two flexural cracks
where a bond slip relationship between steel and concrete can be presented and can be
said that there is no perfect bond between steel and concrete been applied (Fantilli, Ferretti
et al. 1998, Gravina and Warner 2003, Gravina and Smith 2008).
Gravina (2002) proposed a block model where a linear variation of strain is considered at
crack faces to compute the stress and tensile steel and tensile concrete strains are been
used to quantify the deflections of the member. This model is applicable for one layer of
reinforcement to predict the crack formation of RC beams and the similar trend can be
observed from the block model developed by Fantilli, Ferretti et al. (1998).
Oehlers, Liu et al. (2005) proposed a model to predict the deflection of RC beams due to
instantaneous loading considering the formation of flexural cracks, interaction between
steel and concrete, allowing slip of the bars and can be applicable for multiple layers of
steel but unable to predict long term deflection considering the effect of creep and
shrinkage.
A finite element method proposed by Pulmano and Shin (1987) to predict the
instantaneous and long term deflections of simply supported, continuous RC and
prestressed concrete beams using empirical equations to compute the effective bending
stiffness and the creep and shrinkage curvatures.
Bazant and Oh (1984) proposed a model to quantify the short term deformations of
members up to ultimate load and long term creep deformations of cracked beams by
making an assumption of concrete having nonzero tensile capacity and characterized by a
39
uniaxial stress strain diagram that has progressive micro cracking behaviour due to strain
softening. Prasada Rao, Jayaraman et al. (1994) proposed a model to calculate the time
dependent stress and strain distribution in concrete due to the effect of creep and shrinkage
and the relaxation of prestressing steel in a cracked prestressed concrete member without
considering the tensile strength of concrete.
Jayasinghe (2011) developed an analytical model to predict the long term deflections of
post-tensioned beams and one way spanning slabs under sustained loading conditions
including its rotations, crack widths and crack spacing considering the effect of creep and
shrinkage, bond slip behaviour between steel and concrete and prestress losses over time.
Gribniak (2009) proposed an innovative numerical model to derive a free of shrinkage
tension stiffening relationships using test data of shrunk bending RC members and
applying moment curvature relationships to investigate the short-term deformation of RC
flexural members due to the effect of shrinkage.
All of these mentioned above numerical and analytical models and simplified methods to
quantify the instantaneous and long term deflection of RC members are mainly based on
moment curvature (M/χ) approach where there is a linear strain profile and the full
interaction (FI) between reinforcement and concrete that is reinforcement does not slip
relative to the concrete is been considered. Therefore in these assumptions, it can be found
that the analysis techniques are unable to describe the crack widths and crack spacing or
widening properly as because these techniques are relied on the effective flexural rigidity
(EIeff).
Hence, a numerical model has been developed using partial interaction (PI) moment
rotation (M/θ) approach to predict the instantaneous deflection of RC flexural members
and that taken into consideration the bond slip behaviour in between with concrete and
40
steel, tension stiffening of the concrete prism and considering the flexural cracking
behaviour with crack widths and crack spacing of the member accurately (Oehlers,
Mohamed Ali et al. 2011, Visintin, Oehlers et al. 2012, Visintin, Oehlers et al. 2013). This
numerical partial interaction (PI) moment rotation (M/θ) analysis technique has been
extended by Visintin, Oehlers et al. (2013) to quantify the long term deflection of RC
members incorporating the effect of creep and shrinkage. But in this case, linear shrinkage
strain profile of the member has been considered which is quite straight forward to do the
analysis and in the practice shrinkage strain profile should be strongly non-linear as
mentioned earlier and also varies along the depth and width of the member. Therefore,
there is a need to develop a new model to quantify the long term deflection of RC members
with incorporation of creep and non-linear shrinkage variations in concrete member.
2.6 Summary
Based on the critical reviewing of all the literatures stated above the following conclusion
can be drawn:
The most common cause of cracking in concrete structures occur due to shrinkage. Direct
tension cracks due to restrained shrinkage can penetrate through the member and these are
always difficult to control and can create serviceability and durability problems to the
structures. In addition, shrinkage is detrimental to the structures which may result in
significant deflections to the member and the restrained shrinkage can cause tension in the
concrete and that result the cracks and if not controlled properly can lead to durability and
serviceability failures to the members. Current design techniques assume uniform
shrinkage profile but in practice it is strongly non-linear and also influence of member
size has significant effect in shrinkage strains. A limited number of research been
conducted to quantify the non-linear shrinkage behaviour of concrete. Therefore, a

41
numerical moisture diffusion model been developed to predict the non-linear shrinkage
behaviour of concrete along both its depth and width of the member. Hence an
experimental program has also been proposed to quantify the shrinkage behaviour of
concrete member varying the exposed surfaces with considering the member size effect
as well.
Finally, it can be clearly stated that a lack of research and knowledge is existing in this
shrinkage effect in RC structures and we can briefly mention that although a lot of research
has been performed on this time dependent shrinkage behaviour of concrete but still these
behaviours are not been clarified properly. However, there is a need to do more research
to explore completely the shrinkage effect of concrete which will be able to quantify the
shrinkage induced deformation of RC members effectively. There is no research been
conducted on using non-linear shrinkage profile along the depth and width of the member
to examine the influence of non-linear shrinkage on member behaviour and there are also
no design guidelines available for long term member deflection allowing for non-linear
shrinkage. Hence, a numerical partial interaction (PI) segmental moment rotation (M/θ)
approach is been developed to quantify the long term deflection of RC members and which
also considers the non-linear shrinkage behaviour along the depth and width of the
member.
42
Chapter 3 Journal Paper on Non-linear Shrinkage
This chapter deals with the journal article prepared from this study titled “Mechanics of
simulating the serviceability deflection of RC beams allowing for partial interaction and
non-linear shrinkage”.
In this chapter, it is shown that the mechanics of simulating the long term serviceability
deflection of RC beams is very complex as it requires four specific mechanics models:
(1) A diffusion model to quantify the variation of shrinkage strain with: time;
environmental conditions; beam shape; beam surface conditions; and with the width and
depth of the beam.
(2) A full interaction flexural segmental analysis that can cope with the variations in
shrinkage strain from the diffusion model to quantify: the behaviour of the beam in
uncracked regions; and also to predict the onset of flexural cracking that is the start of
partial interaction behaviour.
(3) A partial-interaction tension stiffening analysis that can cope with imposed shrinkage
strains from the diffusion model to quantify: primary crack spacings and widths; the onset
of secondary cracks and their spacings and widths; and crack opening stiffnesses.
(4) A partial interaction flexural segmental analysis that can cope with any imposed
variation in shrinkage strain to quantify the behaviour of the cracked sections of the beam.
This analysis gives: the moments at which primary and secondary cracks occur; variations
in crack spacings and widths; and mechanically correct moment-curvatures for
quantifying the deflections.
43
This research provides mechanics solutions for all the facets that control the serviceability
time dependent behaviour of RC beams. It is envisaged that these numerical mechanics
solutions can provide researchers with the tools to develop simple design procedures.
44
Publication:
Mechanics of simulating the serviceability deflection of RC beams allowing for
partial interaction and non-linear shrinkage
1
Noor Md. Sadiqul Hasan, 2Terry Bennett, 3Phillip Visintin
and 4Deric John Oehlers
1
Noor Md. Sadiqul Hasan
PhD Candidate
School of Civil, Environmental and Mining Engineering
South Australia 5005
AUSTRALIA
2
Dr. Terry Bennett
Senior Lecturer,
AUSTRALIA
45
Corresponding Author:
3
Dr.Phillip Visintin
Senior Lecturer,
AUSTRALIA
email: phillip.visintin@adelaide.edu.au
Tel. +61 8 8313 3710
Fax. +61 8 8303 4359
4
Emeritus Professor Deric J. Oehlers
AUSTRALIA
Submitted to Proceedings of the Institution of Civil Engineers – Structures and Buildings
46
Statement of Authorship

Submitted to Proceedings of the Institution of Civil Engineers – Structures and
Buildings
Noor Md. Sadiqul Hasan (Candidate)
Performed all analysis of the paper and wrote the manuscript.
I hereby certify that the statement of contribution is accurate and give permission for
the inclusion of the paper as thesis chapter
Signed.......................................................................................................Date...............
Terry Bennett
Helped in analysis of the paper and manuscript evaluation.
Signed.......................................................................................................Date...............
Phillip Visintin
Supervised the work development and edit the manuscript.
Signed.......................................................................................................Date...............
Deric John Oehlers
Helped in evaluation and in editing the manuscript.
Signed.......................................................................................................Date...............
47
Noor Md. Sadiqul Hasan, Terry Bennett, Phillip Visintin and Deric John Oehlers
Abstract
Existing analysis and design techniques to quantify serviceability member deflection of
reinforced concrete (RC) beams are generally built on two major premises: (1) full
interaction (FI) through the use of moment curvature approaches; and (2) a uniform
longitudinal shrinkage strain sh within the member to simplify the analysis. Both premises
are gross approximations. With regard to the first premise, RC beams are subject to
flexural cracking and the associated partial interaction (PI) behaviour of slip between the
reinforcement and adjacent concrete. Furthermore with regard to the second premise,
numerous tests have shown that sh varies along both the width and depth of the beam that
is, it is far from uniform. Hence the quantification of the serviceability deflections of RC
beams for design is subject to two major sources of error: that due to the PI mechanisms
that occur in practice; and that due to the time dependent material properties of shrinkage
and creep. This paper deals with the development of PI numerical mechanics models with
non-linear shrinkage strain variations required to simulate the PI behaviour of RC beams
in order to considerably reduce the source of error due to the mechanics model. This new
mechanics model will allow: the development of better design mechanics rules for
serviceability deflection; and also assist in the better quantification of creep and shrinkage
by removing or considerably reducing the existing mechanics source of error.
48
Keywords: reinforced concrete beams; deflection; concrete shrinkage; concrete creep;
partial interaction mechanics; segmental approach.
List of notation
Ac cross sectional area of concrete in prism
Ar cross section area of reinforcement
Arc total Ar in compression
Art total Ar in tension
B bond force at reinforcement/interface
Bn B in nth prism segment
b width of beam
C moisture diffusion coefficient
C1 maximum value of C
Cn force in concrete on left hand side of nth segment
D displacement of reinforcement relative to position at unstressed state
Db-long long term deflection of beam at mid-span
Db-short short term deflection of beam at mid-span
DFI D from full interaction analysis
DPI D from partial interaction analysis
d full depth of beam
drt distance of tension reinforcement to tension face of beam; half depth of
reinforced concrete prism
dNA depth of neutral axis from compression face
dNA-c dNA for concrete section
49
dNA-cn dNA for concrete slice n
dNA-r dNA for all the longitudinal reinforcement
d/dx slip strain
Ec concrete modulus allowing for creep
Er reinforcement modulus
E(t) concrete elastic modulus at a given time
E(t0) initial elastic modulus of concrete
EI flexural rigidity
F force profile
Fcc resultant force in concrete in compression
Fccn Fcc in nth slice
Fct resultant force in concrete in tension
Fctn Fct in nth slice
Frc resultant force in compression reinforcement
Frt resultant force in tension reinforcement
FI full interaction
f surface factor
fc compressive strength of concrete
fck Characteristic compressive strength
fct tensile strength of concrete
gs(t) E(t0)/Et
H pore humidity
Hc H when C(H) = 0.5C1
Hen environmental humidity
50
Hs surface humidity
ksh shrinkage coefficient
Lb length of beam
Ldef half length of symmetrically loaded segment; half length of concrete prism prior
to straining
Lper length of reinforcement perimeter
Ls length of prism segment
LT total length of prism after straining
m empirical constant for the diffusion coefficient
M moment
Mcr moment to cause cracking
Mcr-in Mcr to cause initial crack in uncracked beam
Mcr-pr moment to cause primary cracks
Mcr-sec moment to cause secondary cracks
n number of slices per half width
P resultant longitudinal force in reinforcement
Pcr resultant force in reinforcement at a flexural crack
Pcr-pr Pcr at the formation of primary cracks
Pcr-sec Pcr to cause secondary cracks
Pcr/cr crack opening stiffness
Pn P on left side of nth element
PI partial interaction
RC reinforced concrete
Scr crack spacing
51
Scr-pr primary crack spacing
T temperature
t time
w width of crack; 2cr
α0 empirical constant for the diffusion coefficient
 curvature
 slip of reinforcement
cr  relative to crack face; w/2
n slip at nth segment
 deformation profile
b interface bond slip
c-n concrete deformation within nth slice
rc contraction of compression reinforcement
rt extension of tension reinforcement
n change in slip in nth segment
 strain profile; strain
c concrete strain distribution
c-sh strain in concrete due to shrinkage
cn c in nth slice; mean concrete strain in nth prism segment
ct fracture strain of concrete; fct/Ec
r reinforcement strain distribution
rc strain in compression reinforcement
𝜀𝑠0 magnitude of the ultimate shrinkage strain
52
rt strain in tension reinforcement
rn mean reinforcement strain in nth prism segment
r-sh strain in reinforcement due to shrinkage
𝜀𝑠0 ultimate concrete shrinkage strain
sh shrinkage strain
shn shrinkage strain distribution in nth slice of concrete or nth prism
 rotation
 stress profile
c concrete stress distribution
cc stress in concrete at level of compression reinforcement
ct stress in concrete at level of tension reinforcement
r reinforcement stress distribution
rc stress in compression reinforcement
rt stress in tension reinforcement
b interface bond stress
bn b in nth segment
1. Introduction
Prior to flexural cracking, the deflection of an RC beam can be derived from strain based
approaches (Faber 1927; Whitney 1932; Bresler and Selna, 1964; Ghali et al., 1967;
Bazant 1972, Branson 1977; Gilbert 1988; Gilbert and Ranzi, 2011) , such as the use of
FI flexural rigidities (EI) or FI moment-curvatures (M/), and it is fairly straightforward
53
to incorporate constant values of shrinkage strain sh (Visintin et al., 2013a). This can be
used to predict the initial flexural crack and the deflection up to the formation of the initial
flexural crack. The mechanics is correct, that is there is no error associated with the FI
M/ mechanics, and the only major source of error is in estimating the shrinkage strain sh
and in particular the assumption of a uniform distribution.
Once the initial flexural crack occurs, then the behaviour is governed by tension-stiffening
as this PI behaviour controls the crack spacing and crack widths that have a major effect
on the deflection (Visintin et al., 2013a; Visintin et al., 2013b; Knight et al., 2013; Knight
et al., 2015). A constant shrinkage strain, that is a longitudinal shrinkage strain sh that is
constant along both the depth d and width b of the member, can be added to these PI
analyses (Visintin et al., 2013b; Knight et al., 2013; Knight et al., 2015) to determine the
effect of shrinkage. However numerical analyses and tests have shown that the shrinkage
strain does vary along the depth and width of the member (Terrill et al., 1986; Mu and
Forth, 2009; Gilbert et al., 2012).
The main purpose of this paper is to develop a mechanics based numerical procedure for
quantifying the serviceability deflection that directly simulates all regions of the RC beam
and that allows for non-linear shrinkage strains and creep as happens in practice. It is
shown that the following four distinct mechanics models are required to simulate the long
term serviceability deflection of RC beams:
Model 1. A Standard Diffusion Model (Kim and Lee, 1998; Kim and Lee, 1999;
Kang et al., 2011) to quantify the variations in the shrinkage strains sh within
54
beams for any environmental condition and for a variety of surface boundary
conditions. It is shown how the diffusion coefficients (Bazant and Najjar, 1971;
CEB-FIP 1990) can be derived from the CEB FIP model code 1990 using concrete
material properties.
Model 2. A FI Segmental Model (Visintin et al., 2013b; Knight et al., 2015) that can
incorporate the variations in sh derived from Model 1 above. It may be worth
noting that a standard FI M/ analysis gives exactly the same results as this FI
segmental analysis. However the FI segmental model is described as it can easily
incorporate PI as explained later. This FI segmental model quantifies the
occurrence of the first or initial flexural crack Mcr and, consequently, the
serviceability deflection prior to Mcr. After the initial flexural crack has formed,
reinforcement slip occurs at and adjacent to the crack so that the following PI
tension-stiffening model is required to predict the occurrence and spacing (Scr) of
subsequent flexural cracks (Visintin et al., 2013a; Visintin et al., 2013b; Knight
et al., 2013; Knight et al., 2015).
Model 3. A PI Tension Stiffening Model (Visintin et al., 2013a; Visintin et al.,
2013b; Knight et al., 2013; Knight et al., 2015) that can incorporate the non-linear
shrinkage strains sh from Model 1. This model predicts the occurrence of
subsequent cracks such as primary and secondary cracks and their crack spacings
Scr. Furthermore, it quantifies the crack opening stiffness Pcr/cr, that is the force
in the reinforcement at a crack Pcr as a proportion of the half crack width cr. To
complete the analyses, the PI properties Scr and Pcr/cr are required in the
following PI segmental model.
55
Model 4. A PI Segmental Model (Visintin et al., 2013a; Knight et al., 2015) can now
be used to predict the deflection of RC beams with the shrinkage strains sh from
Model 1 for any environmental conditions and a variety of RC surface conditions.
These analyses are then used to simulate 6 RC test specimens by Gilbert and
Nejadi (2004).
2. Diffusion model
2.1 Diffusion analysis
Drying of concrete can be represented in terms of pore humidity H at constant temperature
T and can be written as follows
𝜕𝐻
= 𝑑𝑖𝑣(𝐶 𝑔𝑟𝑎𝑑 𝐻) (1)
𝜕𝑡
where the dependence of the moisture diffusion coefficient C on the pore humidity H in
the concrete specimen can be determined empirically such as by the following
1−𝛼
0
𝐶(𝐻) = 𝐶1 (𝛼0 + 1+[(1−𝐻)/(1−𝐻 )]𝑚
) (2)
𝑐
where C1 is the maximum value of the diffusion coefficient, Hc is the pore relative
humidity when 𝐶(𝐻) = 0.5 𝐶1 . For typical dense concretes having low water-cement
ratios, the model parameters have the following values 𝐻𝑐 = 0.8, 𝛼0 = 0.05, 𝑚 = 15
(CEB-FIP 1990). The maximum diffusion coefficient is calculated as
56
𝐶1,0
𝐶1 = (3)
𝑓𝑐𝑘 /𝑓𝑐𝑘0
where 𝐶1,0 = 0.864 cm2/day, 𝑓𝑐𝑘0 = 10 MPa and the characteristic compressive strength
𝑓𝑐𝑘 may be estimated as 8 MPa less than the mean.
At high saturations, the process of moisture migration will involve bulk liquid transport
in addition to diffusion and these processes can be modelled using a two phase, liquid and
water vapour, formulation (Lewis et al., 1998). However for efficiency, a surface factor 𝑓
representing the relationship between moisture transfer from the concrete surface to the
atmosphere, can be applied to reduce the system to a single field equation. Therefore, the
boundary conditions for the one dimension case can be expressed as follows (Kang et al.,
2011)
𝜕𝐻
𝐶(𝐻) 𝜕𝑥 = 𝑓 ∙ (𝐻𝑒𝑛 − 𝐻𝑠 ) (4)
where Hs is the surface humidity and Hen is the environmental humidity. A value of 0.55
cm/day is assumed for the surface factor 𝑓 for normal strength concrete (Sakata 1982).
The free shrinkage strain of concrete is a function of pore relative humidity (Bazant and
Yunping, 1994)
𝜀𝑠ℎ = 𝑘𝑠ℎ (1 − 𝐻) (5)
where 𝑘𝑠ℎ is the shrinkage coefficient which is time dependent as follows
57
𝑘𝑠ℎ = 𝜀𝑠0 𝑔𝑠 (𝑡) (6)
with 𝜀𝑠0 is the ultimate shrinkage strain and 𝑔𝑠 (𝑡) is the ratio of the initial elastic modulus
E(t0) to the elastic modulus at a given time E(t) where t is the time in days and E(t) can be
estimated by
𝑡
𝐸(𝑡) = 𝐸(28)√4+0.85𝑡 (7)
where E(28) is equal to 33w1.5fc, w is the unit weight of concrete in lb.ft-3 and E(t) and
E(28) are in psi. In using Eq.7 units were converted to SI.
2.2 Flow in rectangular beams
The diffusion analysis has been applied to the test beams of Gilbert and Nejadi (2004)
which were subjected to a sustained loading for 400 days. The beams had the following
properties: length Lb = 3500 mm, effective depths d = 325 to 340 mm, width b = 250mm,
reinforced with either 2 No. 16 mm or 3 No. 16 mm diameter bars and a compressive
cylinder strength fc of 24.8 MPa. From these beam properties and using CEB-FIP Model
Code (1990) was derived the relationship between moisture diffusivity (moisture diffusion
coefficient) with moisture content (pore relative humidity of concrete) in Figure 1.
58
60
50
40
C (mm2/day)
30
20
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
H
Figure 1. Variation in moisture diffusion coefficient
The beams were divided into n slices per half width as in Figure 2 such that the behaviour
within one half was the mirror image of the other. The length of the beam Lb was
considered to be much larger than either the depth d or width b so that there was only
lateral moisture flow that is no moisture flow along the length. The environmental
humidity was taken as 40% and the ultimate shrinkage strain as 1800.
59
mid-width
b/2 b/2
A B
(n) (2) (1) (1) (2) (n)
C D
Figure 2. Beam cross-section for diffusion analysis
The results from a diffusion analysis when moisture could flow through all four
boundaries in Figure 2 are shown in Figure 3 at 400 days. In this analysis the cross-section
in Figure 2 has 4 slices per half width. The slice nearest to the mid-width is labelled 1 in
Figure 3 and that adjacent to the boundary A-C in Figure 2 labelled 4 in Figure 3. As
would be expected, the shrinkage strain distributions are: symmetrical about mid-depth;
they are greatest adjacent to the sides A-C and B-D in Figure 2 shown as Slice 4 in Figure
3 in comparison to those at mid-width that is Slice 1; and they are a maximum at the upper
and lower boundaries A-B and C-D in Figure 2.
60
Figure 3. Variation in shrinkage strain along beam width and depth
Importantly: the shrinkage strains in Figure 3 have the greatest magnitude and greatest
variations in the tension stiffening region labelled 2drt; and above the tension stiffening
region they are not even close to uniform as assumed in current design.
Sealing the side A-B in Figure 2 gives the variation in shrinkage strain in Figure 4 for
three intervals of time and where there are two slices per half width. Once again the critical
tension stiffening region within 2drt, critical as it controls crack spacing and crack widths,
has the greatest shrinkage strain and variation in shrinkage strain. However in the
remaining region the shrinkage strain can be considered to be uniform.
61
Figure 4. Variation in shrinkage strain with time
Finally sealing A-C, A-B and B-D in Figure 2 so that flow is only one way gives the
shrinkage strain after 400 days in Figure 5 labelled ‘One way’. These are compared with
the previous two cases to further emphasis that a uniform shrinkage strain is a very crude
assumption and that the critical tension stiffening region has the greatest variation and
magnitude.
62
Figure 5. Variation in shrinkage strain with sealed surfaces
It has been shown that the diffusion model can be used to quantify the longitudinal
shrinkage strains sh in an RC beam. It is now a question of incorporating these shrinkage
strains in a flexural analysis of a beam to determine their effect on deflection. Let us start
with the behaviour prior to the onset of flexural cracking. As flexural cracking has not
occurred there is no slip between the reinforcement and the adjacent concrete so that the
beam has a full interaction behaviour as follows.
3. Full-interaction segmental model
Consider a small length 2Ldef of a beam of length Lb as in Figure 6(b). The segment length
Ldef can be of any length and has the cross-section in Figure 6(a) where: b is the width of
63
the beam; d is the full depth; Arc is the total area of reinforcement in compression; and Art
is the total area of reinforcement in tension at a distance drt from the tension face. To
determine the flexural properties of the beam, such as its M/ relation or its flexural
rigidity EI, the segment is subjected to a constant moment M as shown by applying the
symmetric Euler-Bernoulli deformations A-A as shown in Figure 6(b). The deformations
cause rotations  of the segmental faces which have a neutral axis depth dNA.
b Ldef Ldef
A A
dN
Arc
d M M
Art Euler-Bernoulli
θ deformation θ
drt
A A
(a) (b)
Figure 6. Uncracked segment
As the segment in Figure 6(b) is symmetrically loaded, consider the left hand side as in
Figure 7(a). To illustrate the segment approach, it is first subjected to uniform concrete
shrinkage strain sh such that the shrinkage strain does not vary along the width b nor along
the depth d (Visintin et al., 2013a; Knight et al., 2015). If the concrete were free to shrink,
that is it is unrestrained by the reinforcement, then the concrete face would move from B-
B to C-C that is a distance shLdef as shown, whereas, the face of the reinforcement would
remain at B-B. Hence the Euler-Bernoulli deformation A-A would cause the compression
64
reinforcement to contract rc and the tension reinforcement to extend rt. Dividing these
deformations by Ldef gives the strain profile r in Figure 7(b) in which the neutral axis
depth is dNA-r as shown in Figure 7(a). Furthermore the Euler-Bernoulli deformation
causes the concrete at the compressive face to contract A-C and expand C-A on the tension
face such that the neutral axis depth is dNA-c; dividing these deformations by Ldef gives the
concrete strain profile c in Figure 7(b).
δ Ɛ σ F
Ldef
B C A dNA-c
Fcc
Frc =
dNA-r Ɛrc
δrc σcc σrc Arc (σrc – σcc)
Ɛr
M Ɛc
ƐshLdef σc σr
Ɛsh χ = θ/ Ldef Fct
θ
Frt =
δrt σct Art (σrt – σct)
A B C Ɛrt σrt
(a) (b) (c) (d)
Figure 7. Uncracked segmental analysis with uniform shrinkage strains
The Euler-Bernoulli deformation in Figure 7(a) gives the strain distributions in Figure
7(b) that are offset by sh as shown. This is exactly what would be obtained from a standard
FI M/ analysis but is presented in this form as it can later cope with non-linear strain
profiles and with flexural cracking. Having quantified the strain distributions in Figure
7(b), in which the slope is the curvature  which is /Ldef, the concrete stress profile c
and reinforcement stress profile r in Figure 7(c) can be derived from their material
moduli. Furthermore from the stress profiles, can be derived the resultant forces F in
65
Figure 7(d). To simplify the analysis, the concrete cross section in Figure 6(a) has an area
bd that is the concrete ‘voids’, at the position of the reinforcement, have been allowed for
by reducing the reinforcement force. Hence in Figure 7(d): Fcc is the resultant force in the
concrete in compression; Frc is the resultant force in the compression reinforcement that
is the cross sectional area of the compression reinforcement Arc multiplied by the
reinforcement stress rc less the stress in the concrete at the level of the reinforcement cc;
Fct is the resultant force in the concrete in tension; and Frt is the resultant force in the
reinforcement in tension.
For a fixed rotation  in Figure 7(a), it is a question of varying the neutral axis depth dNA-
r until there is longitudinal equilibrium in Figure 7(d) after which moments can be taken
to derive M for that particular . Dividing  by Ldef gives the equivalent curvature  and
the slope of M/ gives the equivalent flexural rigidity EI.
The above analysis is adapted in Figure 8 to incorporate non-linear shrinkage strain
distributions such as those in Figure 3. In this case, the cross-section in Figure 8(a) is
divided into four slices. It will be assumed that the shrinkage strains to the left of the mid-
width are a reflection of those to the right, that is there is symmetrical drying. Hence in
this example there are only two variations in shrinkage to consider which are labelled sh1
and sh2. Multiplying the shrinkage strains, such as those in Figure 3 but for two slices per
half width, by Ldef gives the concrete deformations c1 for Slice 1 and c2 for Slice 2 in
Figure 8(b). An Euler-Bernoulli deformation A-A at neutral axis depth dNA-r is applied
where it can now be seen that there are now two concrete neutral axis depths dNA-c1 and
66
dNA-c2 that is a neutral axis depth for each slice. The analysis follows the same procedure
as outlined in Figure 7.
δ Ɛ F
mid-width Ldef
B A
2Fcc2
Frc
slice (2) (1) (1) (2) dNA-r dNA-c2
2Fcc1
δc2 dNA-c1 Ɛc1
Ɛsh2 Ɛsh1 Ɛsh1 Ɛsh2
δc1 2Fct2
Ɛc2 2Fct1
θ
Ɛr Frt
A B
(a) (b) (c) (d)
Figure 8. Uncracked segmental analysis with non-linear shrinkage strains
The above analyses give the moment at which flexural cracking starts or initiates Mcr-in
and of course the flexural properties of uncracked members up to Mcr-in. As illustrated in
Figure 8, shrinkage causes a reduced moment at cracking as would be expected. In Figure
8(b), flexural cracking without shrinkage will occur when the strain (A-B)/Ldef equals the
fracture strain capacity of the concrete ct which can be taken as the concrete tensile
strength fct divided by its modulus. However shrinkage increases the strain to (A-C2)/Ldef
so that cracking will occur at a lower moment.
The above analyses can also be used until the tip of the flexural crack extends above the
tension reinforcement after which the tension reinforcement slips relative to the concrete
67
such that there is partial interaction. Once this occurs, behaviour is governed by the tension
stiffening mechanism as explained below.
4. Partial interaction tension stiffening model
The segment required to allow for partial interaction tension stiffening (Visintin et al.,
2013a; Knight et al., 2015) is illustrated in Figure 9(b). In this case, the length of the
segment 2Ldef is equal to the flexural crack spacing Scr. Furthermore, it is common practice
to allow for tension stiffening through the use of axially loaded RC prisms (Knight et al.,
2013) as shown where the depth of the prism 2drt is twice the distance of the tension
reinforcement from the tension face. The mean shrinkage strain within each prism shn in
Figure 9(a) is determined from the results of a diffusion analysis as in Figures 3-5 that is
the mean shrinkage within the depth 2drt.
b Ldef = Scr/2 Ldef = Scr/2

B A A B
(2) (1) (1) (2) crack crack

face face
d -2drt d - 2drt
tension stiffening prism
Pcr θ θ Pcr
drt Ɛ A B B A
2drt sh3 Ɛsh4 Ɛsh3
Δcr Δcr
(a) (b)
tension stiffening
prism
Figure 9. Cracked segmental analysis
68
Consider the tension stiffening prism in the segment in Figure 9(b), half the length of
which Ldef is shown by itself in Figure 10(b). To understand the mechanics, the prism is
divided into x segments of a very short length Ls that is Ls << Ldef. The cross section is
shown Figure 10(a) where the perimeter length of the interface between the reinforcement
and adjacent concrete is Lper. The bond-slip properties (b-b), that is the relationship
between the interface bond shear stress b and interface bond slip b, control the behaviour
across this interface of area LperLs. For example, if the bond stress between the
reinforcement and adjacent concrete is b such that the slip is b, then the bond force B
over the segment length is bLperLs and the bond stiffness is b/b.
Ldef mid-length
(1) (2) (n) (x)
A Ls
Lper
(b) Prism
Ec, Ac
A
Er, Ar
Ɛr-shLdef Ɛc-shLdef
(a)
A B C
DFI (c) FI
A B C
ƐshLdef
LT
(1) (2) (n) (x)
DPI
Pcr (d) PI
Δcr
crack face LT
Figure 10. Tension stiffening prism
First consider the FI case where the bond stiffness b/b is infinite which is shown in Figure
10(c) and in which the concrete shrinkage strain is sh. If there were no bond, that is no
69
restraint from the reinforcement, then the concrete face which was originally at A-A in
Figure 10(b) would shrink by shLdef, that is slip relative to the reinforcement shLdef, to C-
C in Figure 10(c) as shown. However, an infinite bond stiffness prevents interface slip
such that the concrete and reinforcement face B-B lies between the unstressed face of the
reinforcement A-A and the unstressed face of the concrete C-C. Applying equilibrium and
compatibility (Visintin et al., 2013a), it can be shown that the residual compressive strain
in the reinforcement due to shrinkage is given by
𝜀𝑠ℎ
𝜀𝑟−𝑠ℎ = 𝐸 𝐴 (8)
1+ 𝑟 𝑟
𝐸𝑐 𝐴𝑐
where: ErAr is the axial rigidity of the reinforcement; EcAc the axial rigidity of the concrete
prism; the shrinkage strain is taken as positive; and tensile strains are also taken as
positive. From compatibility in Figure 10(c), it can be shown (Visintin et al., 2013a) that
the residual tensile strain in the concrete due to shrinkage is
𝜀𝑐−𝑠ℎ = 𝜀𝑠ℎ − 𝜀𝑟−𝑠ℎ (9)
Furthermore, the displacement of the reinforcement face relative to its original position
whilst there is FI is given by
𝐷𝐹𝐼 = 𝐿𝑇 − 𝐿𝑑𝑒𝑓 = 𝜀𝑟−𝑠ℎ 𝐿𝑑𝑒𝑓 (10)
in which DFI is negative when in contraction.
70
Now consider the effect of PI that is slip between the reinforcement and concrete as
illustrated in Figure 10(d). The nth segment in Figure 10(b) which is also shown in Figure
10(d) is displaying now in Figure 11. At the position of the nth segment, the reinforcement
force and concrete force to the left of the segment is Pn and Cn respectively, the slip
relative to the concrete is n and the bond force due to this slip n is Bn. The parameters
Pn, Cn and n depend on the behaviour of the whole prism and will be determined later.
As the chosen segment length Ls is very small compared with Ldef, mean strains will be
dealt with. The force in the reinforcement to the left Pn reduces due to the bond to Pn-Bn.
Hence the mean strain in the reinforcement rn is due to the average of the forces Pn and
Pn-Bn. Similarly the mean strain in the concrete cn is due to the mean of the forces Cn and
Cn+Bn.
Ls base-line
D C B A
𝑥=𝑛−1 𝑥=𝑛
𝐶𝑛 = ෍ 𝐵𝑥 ෍ 𝐵𝑥 = 𝐶𝑛 + 𝐵𝑛
𝑥=1
𝐵𝑛
E 𝑥=1
𝑃𝑛 𝜀𝑟𝑛 𝑃𝑛 − 𝐵𝑛
(∆𝑛 ) E
𝜀𝑐𝑛
A
D C B
ƐshLs
ƐcnLs
ƐrnLs
𝛿∆𝑛 = (𝜀𝑠ℎ − 𝜀𝑐𝑛 + 𝜀𝑟𝑛 )𝐿𝑠
Figure 11. Local deformation of nth segment in prism
The concrete shrinkage strain causes the concrete face to displace shLs from D-D to B-B
as shown in Figure 11. The stress in the concrete cn increases this by cnLs as shown to C-
71
C. Similarly the strain in the reinforcement rn increases it by rnLs to E-E as shown. Hence
the increase in the slip in the nth segment is given by the following in which the term in
brackets is often referred to as the slip strain d/dx.
𝛿Δ𝑛 = (𝜀𝑠ℎ − 𝜀𝑐𝑛 + 𝜀𝑟𝑛 )𝐿𝑠 (11)
The application of the analysis depicted in Figure 11 to the first two segments in Figure
10(d) are shown in Figure 12. The initial crack is to the left of Segment 1, hence the force
in the concrete on the left hand side C is zero. Let the force in the reinforcement at the
crack be Pcr and the slip cr such that the width of the crack w is 2cr. From cr and the
bond-slip properties b-b can be derived the bond force B1 as shown in the equations
below. Hence the forces on the right of the prism and consequently the material strains c1
and r1, also in the equations below. Finally the slip-strain d1/dx, and increase in slip of
Segment 1 that is 1. The forces on the left hand side of Segment 2 are those already
derived on the right hand side of Segment 1. The analysis then follows the same steps as
for Segment 1 which are listed below Segment 2. And so on with further segments.
72
(a) Segment 1 (b) Segment 2
crack face
𝐵1 𝐵1 𝐵1 + 𝐵2
C=0
𝐵1 𝐵2
𝑃2 =
𝑃𝑐𝑟 = 𝑃1 𝜀𝑟1 𝜀𝑟2
𝑃𝑐𝑟 − 𝐵1 𝑃𝑐𝑟 − 𝐵1 𝑃𝑐𝑟 − 𝐵1 − 𝐵2
𝜀𝑐1 𝜀𝑐2
∆𝑐𝑟 = ∆1 ∆2
𝜏𝑏1 = 𝑓𝑛 (∆1 ) ∆2 = ∆1 + 𝛿∆1

𝐵1 = 𝜏𝑏1 𝐿𝑝𝑒𝑟 𝐿𝑠 = ∆𝑐𝑟 + (𝜀𝑠ℎ − 𝜀𝑐1 + 𝜀𝑟1 )𝐿𝑠
0 + 𝐵1 𝜏𝑏2 = 𝑓𝑛(∆2 )
𝜀𝑐1 =
2𝐴𝑐 𝐸𝑐
2𝑃𝑐𝑟 − 𝐵1 𝐵2 = 𝜏𝑏2 𝐿𝑝𝑒𝑟 𝐿𝑠
𝜀𝑟1 =
2𝐴𝑟 𝐸𝑟 2𝐵1 + 𝐵2
𝜀𝑐2 =
2𝐴𝑐 𝐸𝑐
𝑑∆1 /𝑑𝑥 = 𝜀𝑠ℎ − 𝜀𝑐1 + 𝜀𝑟1
2𝑃𝑐𝑟 − 2𝐵1 − 𝐵2
𝜀𝑟2 =
𝛿∆1 = (𝑑∆1 /𝑑𝑥) 𝐿𝑠 2𝐴𝑟 𝐸𝑟
𝛿∆2 = (𝜀𝑠ℎ − 𝜀𝑐2 + 𝜀𝑟2 ) 𝐿𝑠
Figure 12. Tension stiffening analysis
An iterative shooting method is used to find a solution. The first case to consider is when
only the initial crack has occurred from the analysis of the uncracked segment in Figure
6. Hence there is only one crack on the left hand side of the prism in Figure 10(d). For a
given force P1 in Figure 12(a) an estimated slip 1 is chosen. At each segment the slip n
and the slip-strain d1/dx are derived. It is a question of finding the xth element where FI
occurs which in this case is that the slip is zero and the slip-strain is r-sh-c-sh, which can
be derived from Eqs. 8 and 9, is also zero. The estimate 1 is varied until this occurs. Once
this boundary condition is achieved the position of the xth segment is the first primary
crack spacing Scr-pr. When the concrete strain in this xth segment cx reaches the concrete
strain capacity ct the reinforcement force to cause primary cracks Pcr-pr.
73
Having formed primary cracks at known spacings Scr-pr, this is used in a tension stiffening
analysis in which Ldef in Figure 10 is equal to Scr-pr/2 to determine the tension stiffening
behaviour between primary cracks. The analysis depicted in Figure 10 still applies except
that the boundary condition is no longer that for FI but instead that at mid-length, that is
Scr-pr/2 from the crack faces, the slip is zero. This analysis gives the relationship between
Pcr and cr in Figure 10(d) that is the crack opening stiffness Pcr/cr required for the
segmental analysis. Furthermore it gives the reinforcement force to cause secondary
cracks Pcr-sec that is when c is equal to ct should they occur and if so an analysis with a
prism length of Scr-pr/4 gives the behaviour in regions where there are secondary cracks.
It can be seen in Figure 11 and Eq. 11 that the shrinkage strain sh causes a more rapid
build up of slip and therefore bond force, so that shrinkage will reduce the crack spacings
and also the reinforcement force to cause cracking.
5. Partial-interaction segmental model
After flexural cracking from the segmental analysis in Figure 6, the properties of the beam
within the cracked regions are determined from a segmental analysis of a segment between
two adjacent cracks as shown in Figure 9.
Starting with primary cracks, the length of the segment in Figure 9 is Scr-pr which is
obtained from the above tension stiffening analysis and which allows for the effects of
shrinkage. Euler-Bernoulli deformations A-A are applied. Within the tension stiffening
region of depth 2drt, the force in the tensile reinforcement Pcr is now determined from rt
that is from the crack opening stiffness Prt-rt also previously obtained from the tension
74
stiffening analysis and which also allows for shrinkage within the tension stiffening
region. For the region of depth d-2drt above the tension stiffening analysis, the segmental
analysis follows that depicted in Figure 8 except that flexural cracking of the concrete can
occur at ct that is the progression of a flexural crack tip can continue. When the force in
the reinforcement is Pcr-pr that was obtained from the above tension stiffening analysis,
this gives the moment to cause primary cracks Mcr-pr. Hence the moment to cause primary
cracks is the lesser of Mcr-pr and Mcr-in (Visintin et al., 2013b) as both can occur in a FI
region. When the force in the reinforcement in Figure 9 reaches Pcr-sec, this gives the
moment to cause secondary cracks Mcr-sec should they occur. Hence in regions of the beam
where Mcr-sec is exceeded both primary and secondary cracks occur where the crack
spacing is now Scr-pr/2.
The analysis in Figure 9 can be used to derive the M/ at serviceability loads in members
with non-linear shrinkage strains not only along the depth of the beam d but also along
the width of the beam b. Dividing the rotations by Ldef gives the equivalent M/; this is
not an approximation in mechanics terms but in mechanics terms an exact solution that
directly copes with variations in shrinkage. The equivalent M/ can then be used in the
analysis of a loaded beam to determine its deflection and can cope with any shape of beam,
any cross-section properties, any environmental conditions and any surface conditions.
75
6. Application of shrinkage analyses
The above shrinkage analyses were applied to beams tested by Gilbert and Nejadi (2004).
The segmental analyses depicted in Figure 8 were first used to quantify the variation in
moment-curvature with time as follows
Figure 13. Moment-curvature with time
These were then used to quantify the deflection with time and compared with test results
in the following figure. It can be seen in Figure 14 that there is a reasonable correlation.
To further show the significance of the incorporation of a non-linear shrinkage profile, the
analysis for beam B1b in Figure 14 has been repeated in Figure 15(b) but with various
different shrinkage profiles as shown in Figure 15(a). That is the non-linear shrinkage
profile is compared to an assumed constant shrinkage strain of magnitude equal to that of
the minimum, maximum and mean obtained from the diffusion analysis. Finally these
76
values are compared to that obtained from a standard shrinkage test conducted on a
concrete prism with cross sectional dimensions of 600 mm by 160 mm. The results of the
analysis show the best fit with experimental data is obtained by considering a non-linear
shrinkage profile and that taking an average result from the diffusion analysis may lead to
a poor correlation with test data. While in the scenario considered the constant shrinkage
strain obtained experimentally yield reasonable correlation with test results this may not
always hold in practice. For example, as the beams exposed surfaces are changed a
significant variation in the shrinkage profile in Figure 5 is observed and this leads to a
substantial variation in long term deflection as in Figure 16.
Figure 14. Comparison of deflections with test results
77
Figure 15. a) Various shrinkage strain profiles used in the analysis at 400 days and b)
Deflection of beam B1b over time
Figure 16. Variation in deflection with exposed surfaces

78
7. Conclusion
It has been shown that the mechanics of simulating the long term serviceability deflection
of RC beams is very complex as it requires four specific mechanics models:
(1) A diffusion model to quantify the variation of shrinkage strain with: time;
environmental conditions; beam shape; beam surface conditions; and with the width and
depth of the beam.
(2) A full interaction flexural segmental analysis that can cope with the variations in
shrinkage strain from the diffusion model to quantify: the behaviour of the beam in
uncracked regions; and also to predict the onset of flexural cracking that is the start of
partial interaction behaviour.
(3) A partial-interaction tension stiffening analysis that can cope with imposed shrinkage
strains from the diffusion model to quantify: primary crack spacings and widths; the onset
of secondary cracks and their spacings and widths; and crack opening stiffnesses.
(4) A partial interaction flexural segmental analysis that can cope with any imposed
variation in shrinkage strain to quantify the behaviour of the cracked sections of the beam.
This analysis gives: the moments at which primary and secondary cracks occur; variations
in crack spacings and widths; and mechanically correct moment-curvatures for
quantifying the deflections.
Importantly, this research provides mechanics solutions for all the facets that control the
serviceability time dependent behaviour of RC beams. It has been shown that the problem
is very complex. However it is envisaged that these numerical mechanics solutions
although not suitable for routine design can provide researchers with the tools to develop
79
simple design procedures as they simulate the major mechanisms influencing cracking
and tension stiffening in reinforced concrete beams.
References
Bresler, B and Selna, L (1964) Analysis of Time Dependent Behaviour of Reinforced
Concrete Structures. Symposium of Creep in Concrete ACI Special Publication SP-
9(5): 115-128.
Bazant Z and Najjar L (1971) Drying of concrete as a nonlinear diffusion problem. Cement
and Concrete Research 1(5): 461-473.
Bazant, ZP (1972) Prediction of Concrete Creep Effects using Age-Adjusted Effective
Modulus Method. ACI Journal 69: 212-217.
Branson, DE (1977) Deformation of Concrete Structures. McGraw Hill, New York.
Bazant ZP and Yunping X (1994) Drying creep of concrete: constitutive model and new
experiments separating its mechanisms. Materials and Structures 27(1): 3-14.
CEB-FIP. CEB-FIP model code (1990) Design code 1994. Thomas Telford, London.
80
Faber, O (1927) Plastic Yield, Shrinkage and Other Problems of Concrete and their Effects
on Design. Minutes of the Proceedings of the Institute of Civil Engineers. 225(1928).
Part I, London, pp. 27-73.
Ghali, A, Nevile, AM and Jha, PC (1967) Effect of Elastic and Creep Recoveries of
Concrete on Loss of Prestress. ACI Journal 64(12): 802-810.
Gilbert, RI (1988) Time Effects in Concrete Structures. Elsevier Science Publishers B.V..
Gilbert RI and Nejadi S (2004) An experimental study of flexural cracking in reinforced
concrete members under sustained loads. UNICIV Report No. R-435, School of Civil
and Environmental Engineering, University of New South Wales, Sydney, Australia.
Gilbert RI and Ranzi, G (2011). Time-dependent behaviour of concrete structures. Spoon
Press.
Gilbert RI, Bradford MA, Gholamhoseini A and Chang ZT (2012) Effects of shrinkage
on the long-term stresses and deformations of composite concrete slabs. Engineering
Structures 40: 9-19.
Kim JK and Lee CS (1998) Prediction of differential drying shrinkage in concrete. Cement
and Concrete Research 28(7): 985-994.
81
Kim JK and Lee CS (1999) Moisture diffusion of concrete considering self-desiccation at
early ages. Cement and Concrete Research 29: 1921-1927.
Kang ST, Kim JS, Lee Y, Park YD and Kim JK (2011) Moisture diffusivity of early age
concrete considering temperature and porosity. KSCE Journal of Civil Engineering
16(1): 179-188.
Knight, D, Visintin, P, Oehlers, D and Jumaat, M (2013) Incorporating residual strains in
the flexural rigidity of RC members with varying degrees of prestress and cracking.
Advances in Structural Engineering 16(10): 1701-1718.
Knight, D, Visintin, P and Oehlers, DJ (2015) Displacement‐based simulation of time‐
dependent behaviour of RC beams with prestressed FRP or steel tendons. Structural
Concrete 16(3): 406-417.
Lewis RW, Schrefler BA and Abd Rahman N (1998) A finite element analysis of
multiphase immiscible flow in deforming porous media for subsurface systems.
Communication in Numerical Methods in Engineering 14: 135-149.
Mu R and Forth J (2009) Modelling shrinkage of concrete from moisture lost. Magazine
of Concrete Research 61(7): 491-497.
Sakata K (1982) A Study on moisture diffusion in drying and drying shrinkage of
concrete. Cement and Concrete Research 13: 216-224.
82
Terrill J, Richardson M and Selby A (1986) Non-linear moisture profiles and shrinkage in
concrete members. Magazine of Concrete Research 38(137): 220-225.
Visintin P, Oehlers DJ and Haskett M (2013a) Partial-interaction time dependent
behaviour of reinforced concrete beams. Engineering Structures 49: 408-420.
Visintin, P, Oehlers, DJ, Muhamad, R and Wu, C (2013b) Partial-interaction short term
serviceability deflection of RC beams. Engineering Structures 56: 993-1006.
Whitney, CS (1932) Plain and Reinforced Concrete Arches. ACI Journal 28: 479-519.
83
Chapter 4 Simulating Shrinkage Strain using Moisture
Diffusion
4.1 Introduction
This chapter describes the usage of the diffusion process to predict the non-linear variation
in shrinkage of a concrete prism. Firstly, the basic mechanics of moisture diffusion are
presented in a mathematical form and it is shown how the finite difference method can be
used to solve and develop a numerical solution to describe the diffusion process. It is then
shown how the material properties and boundary conditions required to describe the
diffusion process that can be extracted using generic concrete material properties and
which is described in the CEB-FIP model code 90. Hence it is also shown how these
properties relate the moisture or pore relative humidity of concrete to the shrinkage strain.
Finally, the diffusion model developed is then used to predict the shrinkage strain
variations of RC beam tested by Gilbert and Nejadi (2004) with varying the surface
boundary conditions. A wider validation is then conducted to show the general accuracy
of the approach to a wide range of concrete material properties.
4.2 Moisture diffusion equation
Moisture diffusion can be used to simulate with the movement of moisture in concrete and
taking drying shrinkage to be proportional to the moisture content or pore relative
humidity, h of concrete specimens it follows that it can also be used to predict concrete
shrinkage strains. According to Bažant and Najjar (1971), the specific water content of
concrete, w (mass per unit volume) should satisfy the partial differential Equation 4.1
which is known as Fick’s law,
84
𝜕𝑤
= 𝑑𝑖𝑣(𝐷 𝑔𝑟𝑎𝑑 𝑤) Equation 4.1
𝜕𝑡
where t = time and D = diffusion coefficient as a function of w and this Equation is only
applicable when the degree of hydration is uniform throughout the body of the specimen
and temperature T is constant and changing of material properties due to hydration is
negligible especially in old concrete. Drying of concrete can also be represented in terms
of pore humidity h at constant temperature T and a fixed degree of hydration, dh = k dw
where k = function of h which is equivalent to co-tangent of the slope of the desorption
isotherm w = w(h) and hence the equation can be expressed as,
𝑑𝑤 1 𝑑ℎ
= Equation 4.2
𝑑𝑡 𝑘 𝑑𝑡
Hence from Equation 4.1 and 4.2, the mass balance equation can be expressed as follows,
𝜕𝐻
= 𝑘 𝑑𝑖𝑣(𝑐 𝑔𝑟𝑎𝑑 𝑤) Equation 4.3
𝜕𝑡
where c = D/k which represent the permeability of the concrete and it is equal to the mass
flux due to a unit gradient of h and D is the moisture diffusion coefficient. According to
Powers and Brownyard (1946), k may be taken as a constant in the ranges of 0.20 to 0.95
for dense cement pastes and concretes. Hence from Equations 4.1, 4.2 and 4.3, the non-
linear moisture diffusion equation can be obtained as follows:
𝜕ℎ
= 𝑑𝑖𝑣(𝐷 𝑔𝑟𝑎𝑑 ℎ) Equation 4.4
𝜕𝑡
Where D is the moisture diffusion coefficient and it is the same as that in Equation 4.1
except it is considered as a function of h rather than w. In Equations 4.1-4.4 the moisture
diffusion coefficient D is dependent on the pore humidity h in the concrete specimen kept
for drying and is been investigated to provide a fundamental mathematical formulation in
Equation 4.4 used for diffusion modelling by finite the difference method.
85
4.3 Moisture diffusion coefficient
The moisture diffusion coefficient or moisture diffusivity, D is also dependent on the
moisture content or pore relative humidity h of concrete. According to the CEB-FIP
(1990) model code, the moisture diffusion coefficient is expressed as a function of the
pore relative humidity 0 < h < 1 given in Equation 4.5,
1−𝛼
𝐷(ℎ) = 𝐷1 (𝛼 + 1+[(1−ℎ)/(1−ℎ 𝑛
) Equation 4.5
𝑐 )]
where 𝐷1 is the maximum of 𝐷(ℎ)for h = 1.0, α = D0/D1, D0 is the minimum of D(h) for
h = 0.0, hc is the pore relative humidity at D(h) = 0.5 D1 and n is an exponent, α = 0.05,
hc = 0.80 and n = 15 are approximately assumed. D1 may be estimated from Equation 4.6,
𝐷1,0
𝐷1 = Equation 4.6
𝑓𝑐𝑘 /𝑓𝑐𝑘0
where 𝐷1,0 = 0.864 cm2/day, 𝑓𝑐𝑘0 = 10 MPa and the characteristic compressive strength
𝑓𝑐𝑘 may be estimated by the mean compressive strength 𝑓𝑐𝑚 and
𝑓𝑐𝑘 = 𝑓𝑐𝑚 − 8 𝑀𝑃𝑎 Equation 4.7
By solving Equation 4.4, the boundary conditions which correlate the surface moisture
with the environmental humidity at the atmosphere and on the exposed surface S must be
defined and according to Kang, Kim et al. (2011) can be expressed as:
𝜕ℎ
𝐷 (𝜕𝑥 ) = 𝑓(ℎ𝑒𝑛 − ℎ𝑠 ) Equation 4.8
s
86
4.4 Finite difference method on moisture diffusion analysis
Moisture diffusion occurs in the concrete if concrete is exposed to an environmental
humidity that is different than the internal humidity of concrete. A numerical analysis on
the moisture diffusion of concrete specimen has been performed by formulating the
moisture diffusion equation as in basic form given in an Equation 4.4 and the boundary
condition expressed in Equation 4.8 by means of finite difference method. Considering
the one dimensional case in Equation 4.4, moisture diffusion can also be expressed using
the Laplace transform given in Equation 4.9 (Kang, Kim et al. 2011)
∂h ∂2 h
=D(h) Equation 4.9
∂t ∂x2
where h is the pore relative humidity of concrete, D is the moisture diffusion coefficient
(mm2/day), t is the drying period in days and x is the spaces (mm).
Using a Taylor’s series of expansion, the Equation can be rewritten as follows in one
dimension but can be readily exposed to multiple dimensions (varying the exposed surface
conditions),
ℎ𝑖𝑡+1 = ℎ𝑖𝑡 + 𝜆 (ℎ𝑖+1

𝑡
− 2ℎ𝑖𝑡 + ℎ𝑖−1
𝑡
) Equation 4.10
∆𝑡
where, 𝜆 = D(h) ∆𝑥 2 Equation 4.11
At high humidity’s, the process of moisture migration will involve bulk liquid transport
in addition to diffusion and these processes can be modelled using a two phase (liquid and
water vapour) formulation according to Lewis, Schrefler et al. (1998), however for
efficiency a surface factor expressed using Equation 4.12 below can be applied to reduce
the system to a single field Equation for the one dimension case according to (Kang, Kim
et al. 2011)
𝜕ℎ
D(h) 𝜕𝑥 = 𝑓 ∙ (ℎ𝑒𝑛 − ℎ𝑠 ) Equation 4.12
87
Where, hen is the environmental humidity, hs is the humidity at the surface of the concrete
and the surface factor f represents the relationship between moisture transfers from
concrete surface to the atmosphere at the exposed surfaces while drying and depends on
the water to cement ratio, air flow speed, moisture gradient and surface textures of the
concrete specimens (Sakata 1982). A value of 5.5 mm/day can be used for normal strength
concrete which is determined by Sakata (1982) from experimental results.
Discretising hen and hs in Eq. 4.12 yields Equation 4.13 and 4.14 for left and right hand
side boundary conditions respectively.
𝑓∆𝑥
(−ℎ𝑖𝑡+1 − ℎ𝑒𝑛 )
𝑡+1 D(h)𝑡+1
𝑖
ℎ𝑖−1 = 𝑓∆𝑥 Equation 4.13
(−1− )
D(h)𝑡+1
𝑖
𝑓∆𝑥
(ℎ𝑖𝑡+1 + ℎ𝑒𝑛 )
𝑡+1 D(h)𝑡+1
𝑖
ℎ𝑖+1 = 𝑓∆𝑥 Equation 4.14
(1+ )
D(h)𝑡+1
𝑖
Which along with Equation 4.10 allow for the development of a numerical finite
difference solution to the moisture diffusion problem.
4.5 Relationship between pore relative humidity and free shrinkage strain of
concrete
According to Bažant and Yunping (1994) the free shrinkage strain of concrete is a function
of pore relative humidity and can be described using
𝜀𝑠ℎ = 𝑘𝑠ℎ 𝑓𝑠 (ℎ) = 𝑘𝑠ℎ (1 − ℎ); 𝑘𝑠ℎ = 𝜀𝑠0 𝑔𝑠 (𝑡) Equation 4.15
Where, 𝑘𝑠ℎ is the shrinkage coefficient; 𝜀𝑠0 is the ultimate shrinkage strain and 𝑔𝑠 (𝑡) is
𝐸(𝑡0)
the ratio of elastic modulus with time, i.e., 𝑔𝑠 (𝑡) = and E(t) can be estimated by
𝐸(𝑡)
𝑡
𝐸(𝑡) = 𝐸(28)√4+0.85𝑡 with 𝐸(28) = 33𝑤 1.5 √𝑓𝑐′
88
A flow chart diagram summarising the process of moisture diffusion modelling is shown
below in Figure 4.1.
Start
Geometry of the section, diffusion coefficient,

surface factor and environmental humidity
Divide the beam section depth into small segment

each of 1 mm
Setting the moisture content 100% initially
Fixing the boundary conditions at the exposed

surfaces in terms of environmental humidity and
surface factor
Iteration begins with using moisture diffusion

equation
Running the iteration until achieves the moisture

content values at ultimate drying period in days
Calculate the moisture loss in terms of moisture

content after drying
Convert the moisture content or pore relative humidity

into shrinkage strain
Processes will be applied for different exposure

conditions in different directional moisture diffusion
cases
End
Figure 4.1: Flow chart diagram to perform moisture diffusion modelling process
89
4.6 Quantification of moisture diffusion coefficient
Moisture diffusion coefficient has been extracted from Gilbert’s tested beam material
properties with using Equation 4.5 from CEB-FIP (’90) model code where moisture
diffusivity expressed as a function of pore relative humidity, h of concrete.
Gilbert and Nejadi (2004) tested six beams having same material properties of 25 MPa
concrete under sustained loading for a period up to 400 days. All the beams were of length
of 3500 mm, with a width of 250 mm and varying in depth from 325 to 340 mm. Beam
B1a represented in Figure 4.2a and sectional elevation of this beam also shown in Figure
4.2b.
250 mm
A
300 mm
40 mm
A
3500 mm
b/2 b/2 (b)
(a) b/2 b/2
12 nn 21 12 n n 21
1 1
d d d d
(c) (d) (e) (f)
Figure 4.2: a) Gilbert and Nejadi (2004) Beam B1a b) Sectional elevation of beam B1a
at A-A c) Four way flow in beam B1a d) Three way flow in beam B1a e) Two way flow
in beam B1a f) One way flow in beam B1a
90
Now let the beam be divided into four slices along its width and allowed to diffuse
moisture through the four surfaces of the beam except its two edges. As the beam is very
long in length compare to its depth and width so considering there will be no moisture
movement along the edges of the beam. Hence Figure 4.2c present the four way flow in
the beam and Figure 4.2d shows the three way flow in the beam where top surface is
sealed against moisture movement and moisture is moving through the three exposed
surfaces of the beam. The two way flow in Beam represents in Figure 4.2e where moisture
is moving along the top and bottom surfaces of the beam and rest of the surfaces are sealed
against moisture movement and then Figure 4.2f depicts the one way flow in the beam
where all the surfaces are sealed against moisture movement except the bottom surface
and so the moisture is moving only through the bottom surface of the beam.
For analysis the characteristic compressive strength of concrete has been determined using
Equation 4.7 for the concrete having mean compressive strength of 24.8 MPa. The
maximum moisture diffusion coefficient D1 for the relative humidity of h = 1.0 has then
been calculated using Equation 4.6 which is then be used in Equation 4.5 to develop the
relationship between moisture diffusion coefficient, D with moisture content or pore
relative humidity, h of concrete shown in S-shaped curve in Figure 4.3.
91
60
50
Moisture diffusivity, D mm2/day
40
30
Gilbert tests
20
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Moisture content or Pore relative humidity, h
Figure 4.3: Moisture diffusivity vs moisture content or pore relative humidity, h for the
beams tested by Gilbert and Nejadi (2004)
4.7 Four way flow in rectangular beam
As mentioned in earlier section, the beam in Figure 4.2a is very long in length L = 3500
mm compare to the depth d or width b of the beam and if this is the case, there will be no
moisture flow along the length of the beam and therefore moisture will be flowing through
the four surfaces except the edges of the beam. Hence moisture diffusion modelling has
been performed by using finite difference method with using Equation 4.9 and the
boundary conditions were used according to Equation 4.12 where environmental humidity
is been considered as 40 percent. Relationship between moisture diffusivity with moisture
content been achieved using Equation 4.5 from CEB-FIP (’90) model code as illustrated
92
in Figure 4.3. Then the moisture content can be converted into shrinkage strain of concrete
with using Equation 4.15 from Bažant and Yunping (1994) where ultimate shrinkage
strain is 1800µε. According to Figure 4.2c, where concrete member can be sliced up to n
components and may be considered into four slices along its width in this study and due
to symmetry only half of the member would be used for analysis and hence shrinkage
strain profile for slices named b1 and b2 is been plotted in Figure 4.4 after been drying for
the period of 100, 250 and 400 days. Experimental investigation by Gilbert and Nejadi
(2004) also confirms that the moisture diffusion model can reasonably well predict the
shrinkage strain variations in a beam as displayed in Figure 4.4 where shrinkage strain
increases over time while drying period accelerates. It can be found that shrinkage strains
are symmetrical about mid-depth and they are of maximum at top and bottom sections of
the beam. Also, they are greatest adjacent to the sides labelled ‘slice b1’comparison to
those at mid-width that is slice b2.
93
Figure 4.4: Shrinkage strain profile in a four way flow Gilbert and Nejadi (2004) Beam
B1a after 100 days, 250 days and 400 days of drying
4.8 Three way flow in a beam
Let us now consider the beam shown in Figure 4.2a is sealed against moisture movement
at the top surface and rest of the surfaces are exposed to moisture movement as it can be
seen in Figure 4.2d. Therefore, moisture will be flowing through the sides and bottom
surfaces of the beam. Boundary conditions are been used near to the exposed surfaces
according to Equation 4.12 and Neumann boundary condition (Walker, Leonard et al.
2009) has also been applied to the sealed surface to allow diffusion to commence from the
sealed surface over time. Using the Finite difference method and considering the
relationship between moisture diffusivity and moisture content represented in Figure 4.3,
the moisture content will be converted into free shrinkage strain of concrete according to
Equation 4.15 that is displayed in Figure 4.5 and which shows the shrinkage strain profile
at slices b1 and b2 for the three way moisture diffusion of a beam at the end of drying
period for 100, 250 and 400 days. It can be seen that the shrinkage strain has the greatest
variations and is the greatest magnitude at the bottom section of the beam and in the
remaining region the shrinkage strain can be considered to be uniform in this case.
94
Figure 4.5: Shrinkage strain profile in a three way flow Gilbert and Nejadi (2004) Beam
4.9 Two way flow in a beam
Let the beam in Figure 4.2a now be exposed form top and bottom surfaces to allow for
moisture movement which is to cause shrinkage strain variations along the beam and rest
of surfaces are sealed against moisture movement can be seen in Figure 4.2e. This
moisture diffusion process can be treated as two directional flow or two way flow in a
beam where identical values achieved along the slices of its width and the problem is
become one dimensional (1D). The boundary condition expressed in Equation 4.12 can
be applied to the top and bottom exposed surfaces at the moisture diffusion modelling
purposes. Hence the finite difference method is applied with consideration of the
relationship between moisture diffusion coefficient with moisture content or pore relative
95
humidity, h shown in Figure 4.3. Then the remaining moisture content after been drying
of 100 days, 250 days and 400 days which causes of moisture losses in the member would
be converted to the shrinkage strain variations along the depth of the beam as displayed
in Figure 4.6 with using Equation 4.15. It can be found remarkable increment of non-
linear shrinkage strain variations along the depth of the beam as the drying period
increases.
Figure 4.6: Shrinkage strain profile in a two way flow Gilbert and Nejadi (2004) Beam
4.10 One way flow in a beam
Now let the beam in Figure 4.2a is been sealed all surfaces except bottom surface against
any moisture movement to allow for shrinkage and so the moisture will be moving from
top region of the member through bottom surface of the beam to cause shrinkage and that
96
can be treated as one directional or one way flow (1D) in a beam as shown in Figure 4.2f.
As discussed earlier, Equation 4.12 would be used to set up the boundary conditions at the
exposed bottom surfaces considering environmental humidity of 40 percent and surface
factor f value of 5.5 mm/day (Sakata 1982). The Neumann boundary conditions (Walker,
Leonard et al. 2009) would be applied at the sealed top surfaces to allow the moisture
diffusion will be taking place from top which causes the shrinkage strain would be starting
from top surfaces of the member which increases over time and representing a non-linear
parabolic profile after 100, 250 and 400 days of drying displaying in Figure 4.7.
Figure 4.7: Shrinkage strain profile in a one way flow Gilbert and Nejadi (2004) Beam
97
4.11 First simulation with Asad, Baluch et al. (1997)
According to Asad, Baluch et al. (1997), a concrete repair material mortar specimen
having dimensions of 10 cm × 2 cm × 2 cm exposed one faces and rest of the surfaces are
sealed. Moisture loss measured at 1cm from the drying surface and simulated with non-
linear theory using non-linear moisture diffusivity provided by the author as expressed in
Equation 4.16 and also displayed in Figure 4.8.
ℎ 𝑏
𝐷 (ℎ) = 𝐷0 + 𝑎 (1−ℎ) Equation 4.16
Where 𝐷0 is the moisture diffusivity at oven-dry condition when ℎ = 0 % and is equal to
0.1175 cm2/day, and the regression parameters a and b are 0.05 and 1.878 respectively.
100
Moisture diffusivity, D (cm2/day)
10
0.1
Regressed values
Experimental values
0.01
0 10 20 30 40 50 60 70 80 90 100
Moisture content or pore relative humidity, h (%)
Figure 4.8: Relationship between moisture content or pore relative humidity, h and
moisture diffusivity D
98
Experimental and predicted values of moisture loss at 1 cm from the drying surface can
be represented in Figure 4.9 which has been simulated using the finite difference method
considering the non-linear relationship between moisture diffusivity with moisture content
from Figure 4.8 where the value of surface factor, f = 0.30 cm/day is the same used by the
author. For linear theory, considering constant diffusivity D = 0.45 cm2/day (where author
used D = 0.30 cm2/day) and f = 0.40 cm/day (where author used f = 0.30 cm/day) to
achieve the best fit. It can be observed that the prediction of moisture loss is found to be
in better agreement using moisture dependent diffusivity with comparison of constant
diffusivity. At the initial stages of drying, moisture loss is high due to the high value of
moisture diffusivity and after 10% of moisture loss the predicted value is in good
agreement with experimental values and this is because the bulk water flux also
participates in moisture transport in the form of pore water suction and elevation at the
early stages for nearly saturated conditions though diffusion mechanism in unsaturated
condition is mainly due to the vapour flux diffusion.
99
40
35
30
Moisture loss: %
25
20
15
10
Non-linear theory
5
Linear theory
Experimental values
0
0 5 10 15 20 25 30 35 40 45 50
Drying time: days
Figure 4.9: Numerical simulation of experimental and predicted values of moisture loss
at 1 cm from the drying surface using finite difference method
4.12 Second simulation with Jafarifar (2012)
According to Jafarifar (2012), concrete specimens were cast in a dimension of 180 mm ×
150 mm × 150 mm to measure the moisture content at various depths after 90 days of
water curing and drying starts at 1, 3, 5, 7, 14, 28, 42, 56, 70 and up to 84 days. Specimens
were kept at a chamber of constant humidity and temperature control room of 40 ± 3%
and 25 ± 3°c. Moisture measurement for various types of concrete conducted for this
experiment using the modified gravimetric method and moisture diffusivities as a function
of moisture content determined from the experimental results using the inverse analysis
technique as shown in Figure 4.10. The surface factor, f also derived from experimental
test results using the technique of inverse analysis and found to be in the range of 3-10
100
mm/day for conventional concrete CC mixes, and 5-10 mm/day for rolling compacted
concrete RCC mixes.
Figure 4.10: Moisture diffusivity versus pore relative humidity or moisture content for
various types of concrete mixes (Jafarifar 2012)
Figure 4.11 represents the numerical simulation of experimental moisture profiles using
the finite difference method with consideration of nonlinear relationship between moisture
diffusivity and moisture content obtained from Figure 4.10 and the range of surface factor
values for different concrete mixes derived by the author. These curves are the best fit
plotting achieved from numerical simulations to the experimental results of moisture
content values using the non-linear relationship of moisture diffusivity with moisture
content and by changing the moisture diffusivity and range of surface factor values
determined by the author.
101
It can be seen from Figure 4.11 (a) for plain CC mix, predicted moisture content values
are in good agreement with experimental results and for rest of the mixes from Figure 4.11
(b) to (d) for plain RCC mix, SFR-CC mix and SFR-RCC mix, moisture content lower
than 75-80%, it shows good accuracy with the test results. This is because the bulk water
flux also responsible for moisture transfers in the form of pore water suction and elevation
whereas vapour flux is considered only in the moisture diffusion mechanism.
0.9
Moisture content
0.8
0.7
0.6
Experimental 10 mm
0.5 Experimental 35 mm
Numerical 10 mm
Numerical 35 mm
0.4
0 10 20 30 40 50 60 70 80
Drying time, days
(a) Plain CC mix
102
1
0.9
Moisture content 0.8
0.7
0.6
Experimental 11 mm
Numerical 11 mm
Numerical 36 mm
0.4
0 10 20 30 40 50 60 70 80
Drying time, days
(b) Plain RCC mix
0.9
Moisture content
0.8
0.7
0.6
Experimental 11 mm
Numerical 11 mm
Numerical 35 mm
0.4
0 10 20 30 40 50 60 70 80
Drying time, days
103
(c) SFR-CC mix
Moisture content 0.9
0.8
0.7
0.6
Experimental 12.5 mm
Numerical 12.5 mm
Numerical 36 mm
0.4
0 10 20 30 40 50 60 70 80
Drying time, days
(d) SFR-RCC mix
Figure 4.11: Numerical moisture profiles compared with the experimental results
simulated using finite difference method: (a) Plain CC mix; (b) Plain RCC mix; c) SFR-
CC mix; (d) SFR-RCC mix
104
4.13 Third simulation with Kim and Lee (1999)
An experimental investigation has performed with three different types of concrete for a
specimen having dimension of 10 cm × 10 cm × 20 cm drying one side at cross-section of
10 cm × 10 cm and total depth of 20 cm and rest of the surfaces are sealed with paraffin
wax against moisture movement. Relative humidity measured at three different depths of
concrete 3 cm, 7 cm and 12 cm using humidity probe inserted into the concrete prism
(Kim and Lee 1999).
Figure 4.12 represents the relationship between moisture diffusivity and moisture content
was extracted using Equation 4.5 from CEB-FIP (1990) and which is been used to
simulate the experimental results shown in Figure 4.13 and 4.14. There are three types of
concrete having different water-cement ratios with two different moist curing conditions
are 3 and 28 days respectively.
0.4
H (w/c = 0.28), moist cured for 3 days
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
Moisture content or pore relative humidity, h
(a) H (w/c = 0.28) drying commence after 3 days moist cured

105
0.45
M (w/c = 0.40), moist cured for 3 days
0.4

0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
(b) M (w/c = 0.40) drying commence after 3 days moist cured
0.7
L (w/c = 0.68), moist cured for 3 days
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
106
(c) L (w/c = 0.68) drying commence after 3 days moist cured
0.3
H (w/c = 0.28), moist cured for 28 days
0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
(d) H (w/c = 0.28) drying commence after 28 days moist cured
107
0.35
M (w/c = 0.40), moist cured for 28 days
0.3

0.25
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1
(e) M (w/c = 0.40) drying commence after 28 days moist cured
0.7
L (w/c = 0.68), moist cured for 28 days
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
108
(f) L (w/c = 0.68) drying commence after 28 days moist cured
Figure 4.12: Relationship between moisture diffusivity and moisture content for
three different types of concrete a) H (w/c = 0.28) b) M (w/c = 0.40) c) L (w/c =
0.68) with moist cured for 3 days and d) H (w/c = 0.28) e) M (w/c = 0.40) f) L
(w/c = 0.68) with moist cured for 28 days
Figure 4.13 and 4.14 shows the numerical simulation of relative humidity
measured at three different locations inside the concrete prism at 3 cm, 7 cm and
12 cm from the drying surface. These curves are best fitted to the experimental
results considering the maximum moisture diffusion coefficient expressed through
the Figure 4.12 for different concrete mixes with two different moist curing
conditions at 3 days and 28 days.
These best fit curves are achieved by changing the moisture diffusivity and surface
factors which shows in good accuracy of predicted relative humidity with
comparison of experimentally measured relative humidity inside the concrete
specimen while drying. It can be appeared that the moisture diffusion model
developed in this study can well predict the moisture or humidity diffusion in these
different types of concrete specimens in 3 days moist cured exposed to the ambient
air at early ages as well as in 28 days moist cured concrete.
109
1
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
0.6 Experimental 12 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120
Drying period, days
(a) H (w/c = 0.28), moist cured for 3 days
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120
Drying period, days
110
(b) M (w/c = 0.40), moist cured for 3 days
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120
Drying period, days
(c) L (w/c = 0.68), moist cured for 3 days
concrete after moist cured for 3 days a) H (w/c = 0.28) b) M (w/c = 0.40) and c) L (w/c
= 0.68)
111
1
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120 140 160 180 200 220 240
Drying period, days
(a) H (w/c = 0.28), moist cured for 28 days
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120 140 160 180 200 220 240
Drying period, days
112
(b) M (w/c = 0.40), moist cured for 28 days
0.95
0.9
0.85
Relative humidity
0.8
0.75
0.7
Experimental 3 cm
0.65
Experimental 7 cm
Numerical 3 cm
0.55 Numerical 7 cm
Numerical 12 cm
0.5
0 20 40 60 80 100 120
Drying period, days
(c) L (w/c = 0.68), moist cured for 28 days
concrete after moist cured for 28 days a) H (w/c = 0.28) b) M (w/c = 0.40) and c) L (w/c
= 0.68)
113
Chapter 5 Proposed Experimental Work
5.1 Introduction
This chapter proposes a new experimental test setup which can better quantify the
variation in shrinkage strain along the width and depth of a cross section and can allow
for variation in surface permeability. Current shrinkage test methodologies have some
limitations that is all surfaces are exposed and they are small scale which leads to a
uniformity of shrinkage strain and which is not present in full size beams. Importantly the
approach considers the influence of member size in terms of the volume over exposed
surface area ratio (V/S) of the member. Experimental tests are proposed to be performed
to measure the long term (> 1 year) shrinkage strain and to obtain all material properties
required to simulate concrete shrinkage via moisture diffusion modelling as proposed in
Chapter 4.
5.2 Purpose of tests
The proposed test regime will provide experimentally the non-linear variation in shrinkage
strain along depth of the prism for one and two direction diffusion processes. It also
provides the non-linear shrinkage strain variations through depth as well as width of the
prism for three and four direction diffusion processes. Importantly, in the proposed
experiments no internal strain gauges are employed as the presence of internal
instrumentation restrains shrinkage and influences the measured results. Hence in the
proposed setup measurement of shrinkage strain is based on the use of DEMEC gauges to
quantify the deformation of the prism at the concrete surface. These readings can then be
114
correlated with the results of a moisture diffusion analysis as outlined in Chapter 4 in order
to quantify the variation in shrinkage strain throughout the width and depth of the
specimen.
5.3 Sizes of specimen
To variation in moisture diffusion and hence shrinkage with specimen size, it is proposed
that for a given concrete under consideration, tests be performed on specimen of size
ranging from 75 mm ×75 mm × 1200 mm to 300 mm × 800 mm × 1200 mm as in Figure
5.1 to Figure 5.6. In these tests the ends of the prism are to be sealed such that the prisms
represent a beam segment extracted from the central region of a beam span. When
considering a three directional diffusion process represent a T beam as in Figure 5.3 to 5.4
only in addition to the beam ends all but the bottom surface will be exposed, while for a
two direction diffusion process representing a slab, the top and bottom surfaces will be
exposed whereas bottom surface will be exposed only for one direction. Details specimen
sizes are provided in Table 5.1. In Table 5.1, specimen having V/S ratios of 75 (bolded)
will be casted in paired form to compare the shrinkage strain variations among themselves.
Table 5.1: Specimen size details with V/S ratios for one up to four direction diffusion
processes
Specimen size One direction Two direction Three direction Four direction
(mm) (V/S) (V/S) (V/S) (V/S)
75×75×1200 75 37.50 25 18.75
150×150×1200 150 75 50 37.50
150×300×1200 300 150 75 50
200×600×1200 600 300 120 75
300×800×1200 800 400 171.43 109.09
115
5.4 Testing for material properties
In addition to the shrinkage tests outlined above, it is suggested that additional tests are
required to measure the compressive and tensile strength of concrete at various ages of 3,
7, 14, 28, 56, 90, 180 and 365 days curing periods. Details are given in the table 5.2 below:
Table 5.2: Test details for material properties
Test name Specimen sizes (mm) Number of specimens

Compressive strength test 100 × 200 24
Cylinder split test 100 × 200 24
Flexural strength test 100 × 100 × 350 24
5.5 Standard shrinkage test
Standard shrinkage test of specimens are crucial to be performed as they are done with a
smaller sizes of concrete prismatic specimens compare to the real sizes of RC members
and all the surfaces are exposed to the environment which are not the case as it happens
always in practice. Therefore, standard shrinkage test of 6 prisms with specimen sizes of
75 mm ×75 mm × 280 mm will also be carried out to simulate with proposed experimental
test results.
5.6 Instrumentation in details
Detailed instrumentation processes are shown in Figure 5.1 to Figure 5.6 below for four
different diffusion processes. Hence Figure 5.1 and 5.2 depict the arrangement of demec
gauge points through the depth of the prism to measure the shrinkage strain along its depth
and total deformation of the prismatic specimens for one and two direction moisture
116
diffusion processes. Figure 5.3 and 5.5 shown the demec gauge points arrangement for
measuring the shrinkage strain at three different points of the prism along its depth for
three and four direction moisture diffusion processes respectively. In Figure 5.4 and 5.6,
it can be seen that the arrangement of demec gauge points to measure the shrinkage strain
along the width at three different locations of the prism for three and four diffusion
moisture diffusion processes respectively.
117
6 @ 12.5
mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm =
150 mm A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm A
A Section A-A
150×300×1200 mm 150×300 mm
118
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm A
A Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A
Section A-A
300×800×1200 mm 300×800 mm
and rear surfaces and at both ends through the depth as well as width of the prism for
measurements of shrinkage strains and total deformations of the prisms in one direction
moisture diffusion process
119
6 @ 12.5
mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm =
150 mm A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm A
A Section A-A
150×300×1200 mm 150×300 mm
120
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm A
A
Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A
Section A-A
300×800×1200 mm 300×800 mm
prisms for measurements of shrinkage strains and total deformations of the prisms in two
direction moisture diffusion processes
121
6 @ 12.5
mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm =
150 mm A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm A
A Section A-A
150×300×1200 mm 150×300 mm
122
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm A
A Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A Section A-A
300×800×1200 mm 300×800 mm
prisms for measurements of shrinkage strains along its depth and total deformations of
the prisms in three direction moisture diffusion processes
123
6 @ 12.5 6 @ 12.5 mm
mm = 75 mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm = 6 @ 25 mm =
150 mm 150 mm
A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm
A
A Section A-A
150×300×1200 mm 150×300 mm
124
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm
A
A Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A Section A-A
300×800×1200 mm 300×800 mm
Figure 5.4: Arrangements of demec gauge points along the width at top (shown only)
and bottom surfaces of the prisms and at both ends through the depth as well as width of
the prisms for measurements of shrinkage strains along its width and total deformations
of the prisms in three direction moisture diffusion processes
125
6 @ 12.5
mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm =
150 mm A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm A
A Section A-A
150×300×1200 mm 150×300 mm
126
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm A
A Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A Section A-A
300×800×1200 mm 300×800 mm
prisms for measurements of shrinkage strains along its depth and total deformations of
the prisms in four direction moisture diffusion processes
127
6 @ 12.5 6 @ 12.5 mm
mm = 75 mm = 75 mm A
A Section A-A
75×75×1200 mm 75×75 mm
6 @ 25 mm = 6 @ 25 mm =
150 mm 150 mm
A
A Section A-A
150×150×1200 mm 150×150 mm
6 @ 25 mm = 6 @ 50 mm =
150 mm 300 mm
A
A Section A-A
150×300×1200 mm 150×300 mm
128
6 @ 33.33 6 @ 100 mm
mm = 200 mm = 600 mm
A
A Section A-A
200×600×1200 mm 200×600 mm
6 @ 50 mm = 6 @ 133.33
300 mm mm = 800 mm A
A Section A-A
300×800×1200 mm 300×800 mm
Figure 5.6: Arrangements of demec gauge points along the width at top (shown only)
and bottom surfaces of the prisms and at both ends through the depth as well as width of
the prisms for measurements of shrinkage strains along its width and total deformations
of the prisms in four direction moisture diffusion processes
129
5.7 Concluding remarks
Current test methodology of shrinkage has major limitations and which are outlined
below:
1) Specimen sizes are very small (75 mm ×75 mm × 280 mm according to AS
1012.8.4:2015) compare to the real sizes of RC members which provide
uniformity of shrinkage strains and that does not happen in practice and all these
issues are also reported in ACI 209R-92 (1992).
2) It also considers the specimens are exposed in all surfaces which are not always
present in practice.
The proposed test setup will provide the better estimations of shrinkage strain for various
sizes of concrete prism specimens according to its volume over exposed surface area ratios
(V/S). It also benefits the size variations starting from 75 mm ×75 mm × 1200 mm to 300
mm × 800 mm ×1200 mm and measurement of shrinkage strains in terms of one up to
four different exposed surface conditions to correlate with real sizes in appropriate
exposed condition of RC members. However non-linear shrinkage strains will be achieved
from these experimental test results that normally happens in practice and could be utilised
for long term analysis of RC members.
130
Chapter 6 Long term Beam Deflection using Segmental
Approach
6.1 Introduction
This chapter deals with the development of a sophisticated numerical segmental approach
to quantify the long term deflection of RC beams. Prior to flexural cracking, the deflection
of an RC beam can be derived from strain based approaches, such as the use of flexural
rigidities or moment-curvature, and it is fairly straightforward to incorporate constant
values of shrinkage (Visintin, Oehlers et al. 2013). This approach can be used to predict
the formation of an initial flexural crack, however once the initial flexural crack occurs
then the behaviour is governed by tension-stiffening as this partial-interaction behaviour
controls the crack spacing and crack widths that have a major effect on the deflection
(Oehlers, Mohamed Ali et al. 2011, Visintin, Oehlers et al. 2012, Visintin, Oehlers et al.
2013).
Previous research has allowed for the incorporation of a constant shrinkage strain into a
partial interaction analysis, that is an analysis where the longitudinal shrinkage strain that
is constant along both the depth and width of the member (Visintin, Oehlers et al. 2013).
While the assumption of a constant shrinkage strain is convenient for numerical analysis,
tests have shown that the shrinkage strain does vary along the depth and width of the
member (Terrill, Richardson et al. 1986, Mu and Forth 2009, Gilbert, Bradford et al.
2012). The main purpose of this study is therefore to develop a numerical procedure that
allows for non-linear shrinkage strains which normally happens in practice and also to
determine whether a constant shrinkage strain is an adequate design simplification for the
non-constant shrinkage strains that can occur in practice.
131
A partial-interaction segmental numerical model is first described. It is then shown how
this model can be adapted to allow for variations in the shrinkage strain along the depth
and width of the member. A standard diffusion analysis (Kim and Lee 1998, Kim and Lee
1999, Kang, Kim et al. 2011) described in Chapter 4 is then used to quantify the variations
in the shrinkage strains within beams for known diffusion coefficients. It is also shown in
Chapter 4 how these diffusion coefficients (Bažant and Najjar 1971, CEB-FIP 1990) can
be achieved from CEB FIP model code 1990 using concrete material properties. The
analyses are then used to simulate test specimens of six RC beams under sustained load
up to 400 days tested by Gilbert and Nejadi (2004) and the predicted results provided good
correlation with experimental test results.
6.2 Partial interaction segmental analysis
The partial interaction (PI) moment rotation (M/θ) segmental analysis (Oehlers, Mohamed
Ali et al. 2011, Visintin, Oehlers et al. 2012, Visintin, Oehlers et al. 2013) of a segment
of a beam is illustrated in Figure 6.1a on a segment of the crack spacing S cr = 2Ldef and
which is extracted from a beam having total length of L >> d (depth of beam) or b (width
of beam) such that Euler-Bernoulli beam theory applies.
For analysis a constant moment of Mseg is applied to the beam which causes the ends of
the segment in Figure 6.1a rotate θ from A-A to B-B at each crack faces. The segment is
symmetrical along C-C and symmetrically assigned moment of Mseg and thereby both
ends rotate θ such that by symmetry only half of the segment needs to be considered only
for analysis as in Figure 6.2a.
Let us now consider the segmental analysis to be performed prior to flexural cracking of
the beam under the application of constant sustained load/ moment of Mseg.
132
Scr
δtop C ε σ P
A B B A
Prc
θ θ Pcc
Mseg Mseg
Pct
Prt
B A C A B
Δ
f Ldef Ldef
(a) (b) (c) (d)
Figure 6.1: A standard multi-crack segmental analysis
6.3 Prior to cracking segmental analysis
The left half portion of the segment shown in Figure 6.2a and Figure 6.3a below depict
the separation of RC beams with concrete and reinforcement element respectively prior to
cracking for segmental analysis been performed. Left half segment length of Ldef in Figure
6.2a rotate from A-A to B-B with an amount of rotation θ which causes deformation of
δtop at top fibre of the concrete by inducing a constant moment of Mseg. This deformation
profile can be converted to a strain profile as shown in Figure 6.2b by dividing the δ top
with deformation length of Ldef and these are real strains which can be measured by
attaching strain gauges on the member body. By achieving the strain distribution in the
segment, the stress profile shown in Figure 6.2c can also be determined using any
conventional material stress-strain relationship as because these are real strains and
eventually can be determined internal forces as displayed in Figure 6.2d of the member.
After determining all the internal forces, the top deformation of the concrete fibre δtop can
be varied for a fixed rotation value of θ which adjusts the neutral axis depth until
equilibrium of internal forces is achieved. The same procedure as described above need
133
to be applied to Figure 6.3a to achieve the strain, stress and force profile as shown in
Figure 6.3b, Figure 6.3c and Figure 6.3d respectively bearing in mind to keep exactly the
same top deformation of the concrete fibre δtop for a fixed rotation of θ in both cases.
By achieving the equilibrium of internal forces that is the sum of forces in both Figure
6.2d and 6.3d sum to zero and it provides a single point in Figure 6.4a moment rotation
(M/θ) curve and the procedure is repeated by increasing rotations until concrete strain
reaches to its flexural tensile cracking strain and the crack tip reaches the tensile
reinforcement which leads to be plotted the moment rotation curve up to the uncracked
portion. This crack can be initiated anywhere along the beam and where the moment
exceeds the cracking moment of the beam element. After achieving the uncracked portion
of moment rotation (M/θ) curve as shown in Figure 6.4a, this iteration process is repeated
with increasing rotations until reaches the serviceability limit.
The moment rotation (M/θ) relationship in Figure 6.4a can be converted into the
traditional moment curvature (M/χ) relationship by dividing with deformation length Ldef
as shown in Figure 6.4b and both of them are identical for uncracked segments but for
cracked element moment rotation (M/θ) allows the partial interaction (PI) tension
stiffening mechanism for calculating the forces at the tensile reinforcement whereas
conventional moment curvature (M/χ) relationship uses full interaction (FI) in between
reinforcement and surrounding concrete. Hence, the moment rotation (M/θ) approach
shown in Figure 6.4a converted to moment curvature (M/χ)
134
as shown in Figure 6.4b and the secant stiffness’s of the moment curvature (M/χ) are EI
as displayed in Figure 6.4c. Hence integrating the curvatures in the beam from Figure 6.4b
gives the deflections.
δtop ε σ P
A B C
θ Pcc
f
Mseg
Pct
B A Ldef C
(a) (b) (c) (d)
Figure 6.2: Separating elements of RC beam (Concrete element)
δtop ε σ P
A B C
Prc
θ
f
Mseg
Prt
B A Ldef C
Δ
f (b) (d)
(a) (c)
Figure 6.3: Separating elements of RC beam (Reinforcement element)
6.4 Accommodation of cracking in the segmental approach
Let us now considering the crack tip is just above the tension reinforcement and when
Mseg exceeds the cracking moment (Mcr) and that follows the cross-section of left hand
segment of Figure 6.1a for single concrete element and which is strain based moment
curvature (M/χ) approach that is full interaction (FI) in between reinforcement and
135
concrete. If when the crack tip goes above, tension stiffening takes over and is shown in
Figure 6.5a for multi concrete element. At this stage, partial interaction theory must be
used to describe the behaviour of the tensile reinforcement and the crack formation as
because the load developed in the reinforcement bar is now dependent on the slip of the
bar, Δ at the crack face which in turn depends on the bond slip (τ/δ) properties between
the reinforcing bars and the concrete surrounding it (Muhamad, Ali et al. 2011, Visintin,
Oehlers et al. 2012, Visintin, Oehlers et al. 2013, Visintin, Oehlers et al. 2013). The partial
interaction load slip behaviour can be determined through the well-established numerical
technique (Haskett, Oehlers et al. 2008, Oehlers, Mohamed Ali et al. 2011) and will be
described in the next section of this chapter. This analysis procedure as described above
can be readily expanded for multi-concrete element by slicing the member into n number
of slices shown here up to four slices along its width and which can be seen in Figure 6.5a
represents the multi-concrete element after tension stiffening mechanism is named as
partial interaction (PI) moment rotation (M/θ) approach will be discussed briefly in the
next section of this chapter.
M M M
uncracked uncracked
cracked cracked
θ χ (θ/Ldef) EI
(a) (b) (c)
Figure 6.4: Flexural properties (M/θ, M/χ and M/EI)
136
6.5 Partial interaction tension stiffening model
The segment shown in Figure 6.5b is required to allow for partial interaction tension
stiffening (Visintin, Oehlers et al. 2013, Knight, Visintin et al. 2015) and in this case, the
length of the segment 2Ldef is equal to the flexural crack spacing Scr. Furthermore, it is
common practice to allow for tension stiffening through the use of axially loaded RC
prisms (Knight, Visintin et al. 2013) as shown where the depth of the prism 2drt is twice
the cover to the tension reinforcing bar from the tension face. The mean shrinkage strain
within each prism shn in Figure 6.5a is determined from the results of a diffusion analysis
discussed in Chapter 4 that is the mean shrinkage within the depth 2drt.
b Ldef = Scr/2 Ldef = Scr/2

B A A B
(1) (2) (2) (1) crack crack

face face
d -2drt d - 2drt
tension stiffening prism
Pcr θ θ Pcr
drt Ɛ A B B A
2drt shn Ɛshn Ɛshn
Δcr Δcr
(a) (b)
tension stiffening
prism
Figure 6.5: Cracked segmental analysis
Consider the tension stiffening prism in the segment in Figure 6.5b, half the length of
which Ldef is shown by itself in Figure 6.6b and the prism is divided into x segments of a
very short length Ls that is Ls << Ldef ,to understand the mechanics behind the tension
stiffening model. The cross section is shown in Figure 6.6a where the perimeter length of
the interface between the reinforcement and adjacent concrete is Lper and the bond-slip
137
properties (b-b), that is the relationship between the interface bond shear stress b and
interface bond slip b, control the behaviour across this interface of area LperLs. For an
example, if the bond stress between the reinforcement and adjacent concrete is b such
that the slip is b, then the bond force B over the segment length is bLperLs and the bond
stiffness is b/b.
Ldef mid-length
(1) (2) (n) (x)
A Ls
Lper
(b) Prism
Ec, Ac
A
Er, Ar
Ɛr-shLdef Ɛc-shLdef
(a)
A B C
DFI (c) FI
A B C
ƐshLdef
LT
(1) (2) (n) (x)
DPI
Pcr (d) PI
Δcr
crack face LT
Figure 6.6: Tension stiffening prism
First consider the FI case where the bond stiffness b/b is infinite which is shown in Figure
6.6c and in which the concrete shrinkage strain is sh. If there were no bond, that is no
restraint from the reinforcement, then the concrete face which was originally at A-A in
Figure 6.6b would shrink by shLdef, that is slip relative to the reinforcement shLdef, to C-
C in Figure 6.6c as shown. However, an infinite bond stiffness prevents interface slip such
that the concrete and reinforcement face B-B lies between the unstressed face of the
138
reinforcement A-A and the unstressed face of the concrete C-C. Applying equilibrium
and compatibility (Visintin, Oehlers et al. 2013), it can be shown that the residual
compressive strain in the reinforcement due to shrinkage is given by
𝜀𝑠ℎ
𝜀𝑟−𝑠ℎ = 𝐸 𝐴 Equation 6.1
1+ 𝑟 𝑟
𝐸𝑐 𝐴𝑐
where: ErAr is the axial rigidity of the reinforcement; EcAc the axial rigidity of the concrete
prism; the shrinkage strain is taken as positive; and tensile strains are also taken as
positive. From compatibility in Figure 6.6c, it can be shown (Visintin, Oehlers et al. 2013)
that the residual tensile strain in the concrete due to shrinkage is
𝜀𝑐−𝑠ℎ = 𝜀𝑠ℎ − 𝜀𝑟−𝑠ℎ Equation 6.2
Furthermore, the displacement of the reinforcement face relative to its original position
whilst there is FI as shown in Figure 6.6c is given by
𝐷𝐹𝐼 = 𝐿 𝑇 − 𝐿𝑑𝑒𝑓 = 𝜀𝑟−𝑠ℎ 𝐿𝑑𝑒𝑓 Equation 6.3
in which DFI is negative when in contraction.
Now consider the effect of PI that is slip between the reinforcement and concrete as
illustrated in Figure 6.6d. The nth segment in Figure 6.6b which is also shown in Figure
6.6d is displaying now in Figure 6.7. At the position of the nth segment, the reinforcement
force and concrete force to the left of the segment is Pn and Cn respectively, the slip
relative to the concrete is n and the bond force due to this slip n is Bn. The parameters
Pn, Cn and n depend on the behaviour of the whole prism and will be determined later.
As the chosen segment length Ls is very small compared with Ldef, the mean strains will
139
be dealt with. The force in the reinforcement to the left Pn reduces due to the bond to Pn-
Bn. Hence the mean strain in the reinforcement rn is due to the average of the forces Pn
and Pn-Bn. Similarly the mean strain in the concrete cn is due to the mean of the forces Cn
and Cn+Bn.
Ls base-line
D C B A
𝑥=𝑛−1 𝑥=𝑛
𝐶𝑛 = ෍ 𝐵𝑥 ෍ 𝐵𝑥 = 𝐶𝑛 + 𝐵𝑛
𝑥=1
𝐵𝑛
E 𝑥=1
𝑃𝑛 𝜀𝑟𝑛 𝑃𝑛 − 𝐵𝑛
(∆𝑛 ) E
𝜀𝑐𝑛
A
D C B
ƐshLs
ƐcnLs
ƐrnLs
𝛿∆𝑛 = (𝜀𝑠ℎ − 𝜀𝑐𝑛 + 𝜀𝑟𝑛 )𝐿𝑠
Figure 6.7: Local deformation of nth segment in prism
The concrete shrinkage strain causes the concrete face to displace shLs from D-D to B-B
as shown in Figure 6.7. The stress in the concrete cn increases this by cnLs as shown to
C-C. Similarly the strain in the reinforcement rn increases it by rnLs to E-E as shown.
Hence the increase in the slip in the nth segment is given by the following in which the
term in brackets is often referred to as the slip strain d/dx.
𝛿Δ𝑛 = (𝜀𝑠ℎ − 𝜀𝑐𝑛 + 𝜀𝑟𝑛 )𝐿𝑠 Equation 6.4
The application of the analysis depicted in Figure 6.7 to the first two segments in Figure
6.6d are shown in Figure 6.8. The initial crack is to the left of Segment 1, hence the force
140
in the concrete on the left hand side C is zero. Let the force in the reinforcement at the
crack be Pcr and the slip cr such that the width of the crack w is 2cr. From cr and the
bond-slip properties b-b can be derived the bond force B1 as shown in the equations
below. Hence the forces on the right of the prism and consequently the material strains c1
and r1, also in the equations below. Finally the slip-strain d1/dx, and increase in slip of
Segment 1 that is 1. The forces on the left hand side of Segment 2 are those already
derived on the right hand side of Segment 1. The analysis then follows the same steps as
for Segment 1 which are listed below Segment 2 and so on with further segments.
(a) Segment 1 (b) Segment 2

crack face
𝐵1 𝐵1 𝐵1 + 𝐵2
C=0
𝐵1 𝐵2
𝑃2 =
𝑃𝑐𝑟 = 𝑃1 𝜀𝑟1 𝜀𝑟2
𝑃𝑐𝑟 − 𝐵1 𝑃𝑐𝑟 − 𝐵1 𝑃𝑐𝑟 − 𝐵1 − 𝐵2
𝜀𝑐1 𝜀𝑐2
∆𝑐𝑟 = ∆1 ∆2
𝜏𝑏1 = 𝑓𝑛(∆1 ) ∆2 = ∆1 + 𝛿∆1

𝐵1 = 𝜏𝑏1 𝐿𝑝𝑒𝑟 𝐿𝑠 = ∆𝑐𝑟 + (𝜀𝑠ℎ − 𝜀𝑐1 + 𝜀𝑟1 ) 𝐿𝑠
0 + 𝐵1 𝜏𝑏2 = 𝑓𝑛(∆2 )
𝜀𝑐1 =
2𝐴𝑐 𝐸𝑐
2𝑃𝑐𝑟 − 𝐵1 𝐵2 = 𝜏𝑏2 𝐿𝑝𝑒𝑟 𝐿𝑠
𝜀𝑟1 =
2𝐴𝑟 𝐸𝑟 2𝐵1 + 𝐵2
𝜀𝑐2 =
2𝐴𝑐 𝐸𝑐
𝑑∆1 /𝑑𝑥 = 𝜀𝑠ℎ − 𝜀𝑐1 + 𝜀𝑟1
2𝑃𝑐𝑟 − 2𝐵1 − 𝐵2
𝜀𝑟2 =
𝛿∆1 = (𝑑∆1 /𝑑𝑥) 𝐿𝑠 2𝐴𝑟 𝐸𝑟
𝛿∆2 = (𝜀𝑠ℎ − 𝜀𝑐2 + 𝜀𝑟2 ) 𝐿𝑠
Figure 6.8: Tension stiffening analysis
An iterative shooting method is used to find a solution. The first case to consider is when
only the initial crack has occurred from the analysis of the uncracked segment in Figure
141
6.3. Hence there is only one crack on the left hand side of the prism in Figure 6.6d. For a
given force P1 in Figure 6.8a, an estimated slip 1 is chosen. At each segment, the slip n
and the slip-strain d1/dx are derived. It is a question of finding the xth element where FI
occurs which in this case is that the slip is zero and the slip-strain is r-sh-c-sh, which can
be derived from Equations 6.1 and 6.2, is also zero. The estimate 1 is varied until this
occurs. Once this boundary condition is achieved the position of the xth segment is the first
primary crack spacing Scr-pr. When the concrete strain in this xth segment cx reaches the
concrete strain capacity ct, the reinforcement force to cause primary cracks Pcr-pr.
Having formed primary cracks at known spacings Scr-pr, this is used in a tension stiffening
analysis in which Ldef in Figure 6.6 is equal to Scr-pr/2 to determine the tension stiffening
behaviour between primary cracks. The analysis depicted in Figure 6.6 still applies except
that the boundary condition is no longer that for FI but instead that at mid-length, that is
Scr-pr/2 from the crack faces, the slip is zero. This analysis gives the relationship between
Pcr and cr in Figure 6.6d, that is the crack opening stiffness Pcr/cr required for the
segmental analysis. Furthermore it gives the reinforcement force to cause secondary
cracks Pcr-sec that is when c is equal to ct should they occur and if so an analysis with a
prism length of Scr-pr/4 gives the behaviour in regions where there are secondary cracks.
It can be seen in Figure 6.7 and Equation 6.4, that the shrinkage strain sh causes a more
rapid build up of slip and therefore bond force, so that shrinkage will reduce the crack
spacings and also the reinforcement force to cause cracking.
142
6.6 Partial-interaction segmental model
After flexural cracking from the segmental analysis in Figure 6.3, the properties of the
beam within the cracked regions are determined from a segmental analysis of a segment
between two adjacent cracks as shown in Figure 6.5.
Starting with primary cracks, the length of the segment in Figure 6.5 is Scr-pr which is
obtained from the above tension stiffening analysis and which allows for the effects of
shrinkage. Euler-Bernoulli deformations A-A are applied. Within the tension stiffening
region of depth 2drt, the force in the tensile reinforcement Pcr is now determined from rt
that is from the crack opening stiffness Prt-rt also previously obtained from the tension
stiffening analysis and which also allows for shrinkage within the tension stiffening
region. For the region of depth d-2drt above the tension stiffening analysis, the segmental
analysis follows that depicted in Figure 6.1 except that flexural cracking of the concrete
can occur at ct that is the progression of a flexural crack tip can continue. When the force
in the reinforcement is Pcr-pr that was obtained from the above tension stiffening analysis,
this gives the moment to cause primary cracks Mcr-pr. Hence the moment to cause primary
cracks is the lesser of Mcr-pr and Mcr-in (Visintin, Oehlers et al. 2013) as both can occur in
a FI region. When the force in the reinforcement in Figure 6.5 reaches Pcr-sec, this gives
the moment to cause secondary cracks Mcr-sec should they occur. Hence in regions of the
beam where Mcr-sec is exceeded both primary and secondary cracks occur where the crack
spacing is now Scr-pr/2.
The analysis in Figure 6.5 can be used to derive the M/ at serviceability loads in members
with non-linear shrinkage strains not only along the depth of the beam d but also along
the width of the beam b. Dividing the rotations by Ldef gives the equivalent M/; this is
143
not an approximation in mechanics terms but in mechanics terms an exact solution that
directly copes with variations in shrinkage. The equivalent M/ can then be used in the
analysis of a loaded beam to determine its deflection and can cope with any shape of beam,
any cross-section properties, any environmental conditions and any surface conditions.
6.7 Constant longitudinal shrinkage along depth and width
The M/θ approach can be applied to the cracked section in Figure 6.9a after having defined
the tension stiffening behaviour using partial interaction theory (PI) with incorporation of
shrinkage strain as stated above. Ldef in this Figure 6.9a will be equal to half of the crack
spacing (Scr-p/2) in regions where primary cracks take place and in regions where
secondary cracks occur will be equal to Scr-p/4. Figures 6.9 and 6.10 below represent the
concrete and reinforcement element separately from left half portion of Figure 6.1a where
a constant shrinkage strain εsh is been applied to the segment along both depth and width
of the member and that causes a reduction of length of concrete with an amount of εshLdef
from A-A to A´-A´ in concrete element and this shortening in length would not induce a
stress. Hence any deformation would take place away from A´-A´ induces a stress in the
concrete and therefore, A´-A´ becomes the baseline for the concrete to induce stresses in
concrete element for any further deformation from A´-A´. Hence also from Figure 6.10a,
it can be seen that there is no contraction of reinforcement due to an applied constant
shrinkage strain εsh and so the baseline of reinforcement remain unchanged at A-A. Hence
there will be two neutral axis depth one for concrete and another one for reinforcement
due to have two base line with the effect of shrinkage strain εsh.
Let us now consider a moment Mseg is been applied to the segment and sustained for some
period of time, t to cause a total rotation of θ with the combined effect of shrinkage εsh
144
and applied moment Mseg. This rotation θ induces a deformation δtop from A-A to B-B and
the same analysis outlined above for the case of without shrinkage is now been applied
with shrinkage. Hence the neutral axis of concrete in Figure 6.9a lies in between the
intersection of new base line of concrete A´-A´ and B-B and for the reinforcement the
neutral axis is in between the intersection of A-A and B-B as shown in Figure 6.10a
measured from top or bottom fibre of member.
However, in this case the rotation θ can be varied until the resulting moment Mseg is been
achieved and hence the longitudinal equilibrium in both Figures 6.9d and 6.10d should be
at zero whilst the rotational equilibrium requires a moment of Mseg. An alternative
approach would be to fix the rotation θ and varying the top deformation δtop until there
was longitudinal equilibrium of forces in Figures 6.9d and 6.10d and after that the moment
could be taken for that imposed or fixed rotation θ. The analysis is exactly similar to the
Figures 6.2a and 6.3a until the concrete is uncracked but when the crack tip reaches above
the tension reinforcement partial interaction theory would be applied at the tension
stiffening zone to determine the load developed at the reinforcing bar Prt in Figure 6.3d
based on the total length of the reinforcement LT and where LT in Figure 6.6a can be
determined from simple geometry and also explained in the previous section of this
chapter. Hence, the analysis in Figure 6.9a and 6.10a can be applied for increasing
rotations θ to determine the moment rotation relationship in Figure 6.4a with the effect of
constant longitudinal shrinkage strain along depth and width of the member.
The moment rotation M/θ relationship with the incorporation of constant longitudinal
shrinkage strain along the member achieved from the analysis can be converted to an
equivalent moment curvature M/χ relationship by dividing by the deformation length Ldef
145
as shown in Figure 6.4b and these results can be used to derive the variation of equivalent
flexural rigidity EIequ of the member with moment as represented in Figure 6.4c.
Moreover, the variation of these equivalent flexural rigidities (EIequ) that are been derived
from partial interaction PI moment rotation M/θ approach are different from those that are
achieved from a standard full interaction FI moment curvature M/χ analysis as because
the M/θ approach uses PI theory which incorporates the shrinkage effect to allow for crack
formation and widening when crack occurs and eventually it can represent a mechanics
based solution to describe the behaviour of cracked concrete which also includes the time
effects.
δtop ε σ P
A B C
A´
Pcc
θ
f Pct
Mseg
A´
B A C
εshLdef Ldef
(a) (b) (c) (d)
Figure 6.9: Linear shrinkage strain over width and depth (Concrete element)
δtop ε σ P
A B C
A´ Prc
Mseg
θ LT
f Prt
Δ A´
B A Ldef C
εshLdef
(a) (b) (c) (d)
Figure 6.10: Linear shrinkage strain over width and depth (Reinforcement element)
146
6.8 Variation in longitudinal shrinkage strain along depth
Let us now consider a non-constant shrinkage profile along the depth of the member is
been employed and which shown in Figure 6.11a for concrete element and in Figure 6.12a
for reinforcement element for cracked member specimens. According to Gilbert’s
experimental investigation for the composite slab (Gilbert, Bradford et al. 2012) it can be
found that shrinkage strain is not constant along the depth of the member, it does vary
along the depth and shows nonlinear and non-uniform shape as shown in Figure 6.11a to
Figure 6.15a. Furthermore, numerical diffusion model also can predict the nonlinear
shrinkage strain profile along the depth and width of the member which is been used in
this report for analysis.
As it can be seen from the Figure 6.11a, due to a non-uniform shrinkage strain concrete
baseline moved from A-A to A´-A´ with an amount of εshLdef where Ldef is the half of the
crack spacing but the reinforcement baseline remains stationary at A-A shown in Figure
6.12a. Hence, for concrete any deformation away from new baseline A´-A´ will induce a
stain dividing by length of deformation Ldef will produce a stress in the concrete and for
reinforcement any movement occurs from A-A will cause a strain to stress in the
reinforcement bars and so therefore, two strain profiles exist with the application of
shrinkage strain in the member one for concrete as shown in Figure 6.11b and another one
for reinforcement as shown in Figure 6.12b. Therefore, two neutral axis exist in the
analysis one for concrete element (intersection in between A´-A´ and B-B) and other one
for reinforcement element (intersection in between A-A and B-B) which can be seen in
Figure 6.11a and 6.12a respectively.
147
An iterative approach would be used for performing the sophisticated segmental analysis
by fixing the rotation θ and varying the top deformation δtop until there was longitudinal
equilibrium in both Figures 6.11d and 6.12d after which the moment could be taken for
that fixed or imposed rotation θ. Therefore, by repeating the analysis for increasing
rotations moment rotation relationship shown in Figure 6.4a will be established bearing
in mind to apply the partial interaction theory as soon as the crack tip just crosses the
tensile reinforcement bars to calculate the load developed at the reinforcement bars P rt as
shown in Figure 6.12d. Hence therefore, M/θ can be converted to equivalent M/χ relation
as shown in Figure 6.4b by dividing by the deformation length Ldef and hence can be
achieved the M/EI as shown in Figure 6.4c and finally be able to quantify the deflection
of a member.
δtop
ε σ P
A B C
A´
Pcc
θ
Mseg f Pct
A´
B A C
εshLdef Ldef
(a) (b) (c) (d)
Figure 6.11: Non-linear shrinkage strain over depth (Concrete element)
148
δtop
ε σ P
A B C
A´ Prc
θ
Mseg f
Prt
A´
B A C
εshLdef Ldef
(a) (b) (c) (d)
Figure 6.12: Non-linear shrinkage strain over depth (Reinforcement element)
6.9 Variation in longitudinal shrinkage strain along depth and width
Let us now consider the non-constant shrinkage strain were applied along the depth and
width of the member as shown in Figure 6.13a to Figure 6.15a. This segmental analysis
will also be continued by breaking down into components along the width of the member
shown in Figure 6.5a, bearing in mind with applying same rotation and depth of neutral
axis to all of the components. The concrete member is sliced up to four components along
the width of the specimen having identical shrinkage profile for slices width of b1 and b2
due to symmetry so therefore only half of the member analysis could be performed and is
been displayed in Figure 6.13a and 6.14a.
Similar procedure has been discussed earlier for the segmental M/θ approach would be
applied for this case, by fixing the rotation θ, guessing the top deformation δtop initially
then varying δtop until longitudinal equilibrium is been achieved that is sum up all the
forces in Figures 6.13d, 6.14d and 6.15d to zero and after that moment could be taken for
that particular imposed or fixed rotation θ. Therefore, by repeating the analysis for
increasing rotations, the moment rotation M/θ relationship is established for both cracked
149
and uncracked conditions but when the concrete is cracked and the crack tip reaches above
the tension reinforcement, partial interaction theory is been used to determine the forces
developed in the reinforcement bars Prt as shown in Figure 6.15d.
Moreover, it should also be noted that in order to allow for creep a reduced elastic modulus
must be used where the magnitude of this reduction can be determined by using any
convenient method at the time of calculating stress in the concrete.
Hence the M/θ approach can be converted to an equivalent standard M/χ approach by
dividing the deformation length Ldef and the results according to Figure 6.4b can be used
to determine the variation of flexural rigidity EIequ with moment along the member. Both
M/θ and M/χ approaches are identical up to uncracked portion so therefore EIequ is similar
value of EIuncracked but after cracked it is not the same as because M/θ approach uses partial
interaction theory which includes the effect of nonlinear shrinkage strains to allow for
crack formation and widening and hence provide a mechanics based solution to explain
the behaviour of cracked concrete members including time effects due to shrinkage and
creep.
δtop ε σ P
A B C
A´ Pcc1
Pct1
Mseg
θ
f
A´
B A C
εshLdef
Ldef
(a) (b) (c) (d)
150
δtop
ε σ P
A B C
A´ Pcc2
θ
Mseg f Pct2
B
A´
A C
εshLdef
Ldef
(a) (b) (c) (d)
δtop
ε σ P
A B C
A´ Prc
θ
Mseg f
Prt
A´
B A C
εshLdef
Ldef
(a) (b) (c) (d)
Figure 6.15: Non-linear shrinkage strain over width and depth (Reinforcement element)
6.10 Parametric study
Some material properties need to be defined to apply the moment rotation M/θ segmental
analysis and these are not the critical component of the segmental approach and that is
why any material models can provide better results could be replaced. The following
material models have been used in the parametric study to illuminate the time dependent
151
behaviour of cracked reinforced concrete members where the effect of both shrinkage and
creep need to be considered simultaneously but for this study only shrinkage effect has
taken into account. Hence at later stage to simulate with experimental test results (Gilbert
and Nejadi 2004), both shrinkage and creep effect have been considered to quantify the
deflection of RC beams under sustained loading conditions for a period of 394 days.
The elastic modulus of concrete in a time t after first loading at time t0 at any point can be
defined as Ec (t, t0) and using effective modulus method can be written as
𝐸 (𝑡, 𝑡 )
𝐸𝐶 (𝑡, 𝑡0 ) = 1 +𝑐𝜑(𝑡,0𝑡 Equation 6.5
0)
Where t0 is the time at initial loading and φ is the creep coefficient of concrete at time t
after initial loading at time t0. According to Gilbert and Ranzi (2011), the change in elastic
modulus in concrete due to creep would be applied to both in compression and tension
zone of concrete at the segmental analysis for time dependent deformations of the RC
members.
A linear elastic stress-strain relationship has been assumed for concrete in tension and for
concrete in compression the following parabolic stress distribution by Popovics (1973)
𝜀
( )𝑟
𝜀𝑝𝑘
𝜎 = 𝑓𝑐 𝑟 Equation 6.6
𝜀
𝑟−1+( )
𝜀𝑝𝑘
Where ε is the concrete strain to cause a stress r is a factor that controls the ductility of the
concrete and it is given by [Ec/ (Ec – fc/ εpk)]. εpk is the strain corresponds to the peak stress,
fc and that has been taken as defined by Chen, Visintin et al. (2013)
𝜀𝑝𝑘 = 4.76 × 10−6 𝑓𝑐 + 2.13 × 10−3 Equation 6.7
Where fc is in MPa and the peak strain εpk must need to be increased according to the
reduction of elastic modulus due to creep as defined by Equation 6.5.

152
Let us investigate the analysis of a simply supported beam B1a having depth = 340 mm,
width = 250mm and length = 3500 mm reinforced with 2 No. 16 mm bars with 48 mm
cover tested by Gilbert and Nejadi (2004) under sustained loaded with 2 point loading of
18.6 kN for a period of 394 days. The partial interaction PI moment rotation M/θ approach
have been used to determine the variations in moment curvature M/χ and moment
equivalent flexural rigidities M/EIequ for this beam B1a with consideration of nonlinear
shrinkage profile as shown in Chapter 4 of this thesis. Due to the application of sustained
point loading, the distribution of moments in the beam can be calculated and hence from
the M/EIequ relationship variation in flexural rigidities EI also known. The variation in
curvature χ can also be determined by dividing moment M with flexural rigidities EI and
then these variations in curvature can be integrated to achieve the member deflection at
some point in time and the analysis can be repeated with a new M/EIequ relationship at
various points in time. These section properties described above will now be used to
analyse the beam B1a by dividing the section along its width into 10, 4 and single slices
to show the influence of slices on member deflection with accounting the effect of
shrinkage over time. It can be found in Figure 6.16, deflection of the beam increases over
time in which 10 slices and 4 slices shown negligible variations in member deflections
over time whereas single slice provides the higher estimation with almost constant values
after 28 days of loading. Therefore 4 slices would be appropriate consideration for the
analysis need to be performed to simulate with experimentally tested six beams by Gilbert
and Nejadi (2004) under sustained loading for a period of 394 days.
153
Figure 6.16: Influence of slice number along the width of beam on member deflection
6.11 Application to test specimens
These methods have been applied to a series of six beams tested by Gilbert and Nejadi
(2004) in order to validate the partial interaction PI moment rotation M/θ segmental
approach on quantifying the deflection of RC beams. There were six simply supported
beams having span of 3500 mm with a fixed width of 250 mm and depth varying from
325 to 340 mm under two points at third span point sustained loading conditions varying
from 11.8 to 27.0 kN and reinforced with 2 No. 16 to 3 No. 16 mm diameter bars been
tested for a period of 394 days. Beams B1a and B1b with having reinforcement of 2 No.
16 mm diameter bars with clear cover of 48 mm, beams B2a and B2b with having
154
reinforcement of 2 No. 16 mm diameter bars with clear cover of 33 mm and beams B3a
and B3b with having reinforcement of 3 No. 16 mm diameter bars with clear cover of 33
mm. All beam specimens were subjected to two concentrated loads located at third span
point at an age of 14 days. Beam B1a and B2a were loaded with two point loads of 18.6
kN, Beams B1b and B2b were loaded with two point loads of 11.8 kN, Beam b3a sustained
loaded with 27.0 kN and Beam B3b was loaded with 15.2 kN located at third point of the
span of the beam. Concrete strength was increased to 24.8 MPa at 28 days after been
loaded having strength of 18.3 MPa at an age of 14 days. Creep coefficient was determined
periodically through test over the testing period by Gilbert and Nejadi (2004) and concrete
strength calculations over time been performed using CEB-FIP (’90) model code. Non-
linear shrinkage strain profile shown in Chapter 4 also been determined numerically for
each specimen by using finite difference method from diffusion model and that already
discussed at earlier stages.
The partial interaction PI moment rotation M/θ approach was used to determine the
variations in moment curvature M/χ and moment flexural rigidity M/EI as shown in Figure
6.4 for each of the beam specimens which is then used to quantify the deflections of the
experimentally tested six beams. The predicted beam deflections have been simulated with
experimental test results and which are represented in Figure 6.17 for beams B1a to B3b.
Generally, it can be found that the partial interaction PI moment rotation M/θ segmental
approach is able to predict the beams deflections over time reasonably well. Beams B1b,
B2b and B3b have shown close agreement with the experimental test results whereas
beams B1a, B2a and B3a given some variations in a range below 15% with recorded and
predicted results. These could be due to the variations in loading conditions with same
cross section of beams in which primary crack spacing considered in the analysis may be
155
reduced due to its higher loading conditions compare to others. Therefore, it may mean
that the stiffness of the cross section reduces as the reduction of primary crack spacing
which in turn can provide higher deflections over time.
Figure 6.17: Predicted deflection of six beams tested by Gilbert and Nejadi (2004)
In order to show the influence of exposed surfaces on member deflection, the partial
interaction PI moment rotation M/θ has also been applied to three way and one way
flow in a beam as shown in Chapter 4, to predict the deflections of the beams over
time. According to Figure 6.18, it can be found that member deflection is reduced
significantly over time as the minimization of exposed surfaces due to the effect of
shrinkage strain profile and which is decreasing over time in effect of reduced exposed
surface areas.
156
Figure 6.18: Influence of exposed surfaces on member deflection
157
Chapter 7 Conclusions and Recommendations
The research presented in this thesis covers the development of a complete numerical
procedure to quantify the long term deflection of RC members. Significantly, the proposed
model simulates the observed non-linear shrinkage strain distribution by considering
diffusion of moisture throughout the cross section. Moreover the approach directly
simulates the partial interaction mechanism, which is responsible for the formation and
widening of cracks as well as tension stiffening and which in this thesis is shown to be
highly dependent upon concrete shrinkage.
The major outcomes of this research are:
і) the development of a moisture diffusion model to predict the non-linear variation
in shrinkage strain throughout a reinforced concrete
іі) the proposal of a new form of long term experiment to measure the non-linear
shrinkage profile within a concrete block.
ііі) the development of a sophisticated numerical partial interaction segmental
approach to predict the long term serviceability deflection of RC members.
Based on the development of the models above several specific conclusions can be drawn:
1) Current design code approaches suggest a uniform shrinkage profile for RC
members determined through small scale test on prismatic specimens with all
surfaces are exposed to the environment. Based on moisture diffusion modelling
this has been shown to be inappropriate for a full size RC beams in which the
shrinkage profile is highly non-linear.
2) To address the issues in # 1) above a test methodology for predicting the non-linear
shrinkage strain distribution has been developed. Importantly this approach can
158
allow for concrete with different exposed surfaces and different member sizes in
terms of volume over exposed surface areas (V/S) of the member.
3) It has been found that moisture diffusion model predicts well the non-linear
shrinkage variations in a full size RC members and where it shows the non-linear
variations along both its width and depth of the member for four way and three
way flow in a beam. It has been shown that shrinkage strains are symmetrical about
mid-depth and are of maximum at top and bottom sections of the beam for four
way flow condition. In three way flow condition, the shrinkage strain has the
greatest variation and is maximum at the bottom of the beam and in the remaining
region the shrinkage strain can be considered to be uniform.
4) It has been observed that existing models for predicting long term deflection of
RC members are based on the constant shrinkage strain profile. These do not
simulate the non-linear shrinkage distribution observed in practice. This is
significant as it has been shown that the non-linearity of the shrinkage strain has a
significant influence on member deflection. It has also been shown that the greatest
variation in shrinkage occurs in the region where tensile reinforcement is located.
It is therefore suggested that the consideration of non-linear shrinkage may be
particularly important in the design of pre-stressed concrete members where
shrinkage leads to a loss of prestress.
5) The wider variations of deflection of a RC beam have been observed due to the
effect of exposed surface conditions and it shows reduction in deflections over
time with respect to the minimized surface exposers.
159
Although a lot of research has been conducted on the shrinkage behaviour of concrete;
further research is required to better understand the effect of shrinkage on material and
member level in other types of concrete as well. Based on the development of numerical
procedure allowing the non-linear shrinkage strains for quantifying the long term
serviceability deflection of RC members, the following recommendations for future
research can be suggested:
1) The moisture diffusion model developed in this thesis can be used to reanalyse all
existing shrinkage test data in order to develop a material model for describing the
non-linear distribution of shrinkage strain.
2) A simplified hand calculation procedure should be developed from the model in #
1) as well as a simplified material model for the ‘best’ constant shrinkage strain in
members.
3) A series of experiments should be undertaken using the proposed work to further
understand the influence of size effect in RC members.
4) The developed segmental model should be extended to allow for prestressed
concrete members where an accurate understanding of shrinkage strain is required.
According to Knight, Visintin et al. 2013 and Knight, Visintin et al. 2015, this can
be done by changing the partial interaction tension stiffening model.
5) The developed moisture diffusion model may be extended to various types of new
concretes such as Ultra high performance, Self-consolidating and Geopolymer
concrete members to extract the shrinkage variations and then segmental model
can be applied to quantify the long term deflections of RC members.
160
References
AASHTO (2010). LRFD Bridge Design Specifications, 5th Edition. Washington D. C.:
American Association of State Highway and Transportation Officials.
ACI209R-92 (1992). Prediction of Creep, Shrinkage and Temperature Effects in Concrete

Structures. USA, American Concrete Institute: 47.
ACI209R-92 (1997). Prediction of Creep, Shrinkage and Temperature Effects in Concrete

Structures. USA, American Concrete Institute: 47.
ACI Committee 209 (1982). Prediction of creep, shrinkage and temperature effects in
concrete structures. Farmington Hills, Michigan: American Concrete Institute
Akita, H., T. Fujiwara and Y. Ozaka (1997). "A practical procedure for the analysis of
moisture transfer within concrete due to drying." Magazine of Concrete Research 49(179):
129-137.
Al-Saleh, S. A. (2014). "Comparison of theoretical and experimental shrinkage in

concrete." Construction and Building Materials 72: 326-332.
Al-Saleh, S. A. and R. Z. Al-Zaid (2006). "Effects of drying conditions, admixtures and

specimen size on shrinkage strains." Cement and concrete research 36(10): 1985-1991.
Almudaiheem, J. A. and W. Hansen (1987). "Effect of specimen size and shape on drying
shrinkage of concrete." ACI MATERIALS JOURNAL 84(2).
American Concrete Institute (2008). Building code requirements for structural concrete,
ACI 318-08. Detroit, USA, American Concrete Institute.
Asad, M. (1995). "Computational Modelling of shrinkage in repaired concrete." Civil

Engineering.
Asad, M., M. Baluch and A. AI-Gadhib (1997). "Drying shrinkage stresses in concrete
patch repair systems." Magazine of Concrete Research 49(181): 283-293.
AS 1012.8.4 (2015). Methods of testing concrete; Method 8.4: Method for making and
curing concrete - Drying shrinkage specimens prepared in the field or in the laboratory.
Sydney (Australia). Australian Standard.
Bazant, Z. P., Baweja S (1995). "Creep and shrinkage prediction model for analysis and
design of concrete structures – Model B3". Materials and Structures 28: 357–365.
Bazant, Z. and G. Li (2008). Database on concrete creep and shrinkage. Infrastructure

Technology Institute (ITI), Northwestern University. Evanston, IL,USA.
161
Bažant, Z. and L. Najjar (1971). "Drying of concrete as a nonlinear diffusion problem."
Cement and Concrete Research 1(5): 461-473.
Bazant, Z. P. and L. J. Najjar (1972). "Nonlinear water diffusion in nonsaturated

concrete." Materials and Structures 5(1): 3-20.
Bazant, Z. P. and B. H. Oh (1984). "Deformation of progressively cracking reinforced

concrete beams." Computer-Aided Design 16(6): 338-338.
Bazant, Z. P., S. Sener and J.-K. Kim (1987). "Effect of cracking on drying permeability
and diffusivity of concrete." ACI Materials Journal 84(5).
Bažant, Z. P. and X. Yunping (1994). "Drying creep of concrete: constitutive model and
new experiments separating its mechanisms." Materials and structures 27(1): 3-14.
Bissonnette, B. t., P. Pierre and M. Pigeon (1999). "Influence of key parameters on drying
shrinkage of cementitious materials." Cement and Concrete Research 29(10): 1655-1662.
CEB-FIP (1990). CEB-FIP model code 1990: design code, Telford.
Chen, Y., P. Visintin, D. Oehlers and U. Alengaram (2013). "Size-Dependent Stress-

Strain Model for Unconfined Concrete." Journal of Structural Engineering 140(4):
04013088.
Faber, O. (1928). Plastic Yield, Shrinkage, and Other Problems of Concrete, and their
Effect on Design. Minutes of the Proceedings, Thomas Telford.
Fantilli, A. P., D. Ferretti, I. Iori and P. Vallini (1998). "Flexural deformability of

reinforced concrete beams." Journal of Structural Engineering 124(9): 1041-1049.
fib Bulletin 65. (2012) "Model code 2010 Vol 1 & 2”, International Federation for
Structural Concrete Lausanne Switzerland.
Gardner, N. J., Lockman M. J (2001). "Design provisions for drying shrinkage and creep
of normal-strength concrete." ACI Materials Journal 98:159–167.
Gilbert, R. and S. Nejadi (2004). "An experimental study of flexural cracking in reinforced
concrete members under sustained loads, UNICIV Report No." R-435, School of Civil
and Environmental Engineering, University of New South Wales, Sydney, Australia.
Gilbert, R. I. (1988). Time effects in concrete structures. Amsterdam, The Netherlands,

Elsevier Science Publishers B.V.
Gilbert, R. I. (1999). "Deflection calculation for reinforced concrete structures—why we

sometimes get it wrong." Structural Journal 96(6): 1027-1032.
162
Gilbert, R. I., M. A. Bradford, A. Gholamhoseini and Z. T. Chang (2012). "Effects of
shrinkage on the long-term stresses and deformations of composite concrete slabs."
Engineering Structures 40: 9-19.
Gilbert, R. I. and G. Ranzi (2011). Time-dependent behaviour of concrete structures. UK,

Spon Press.
Granger, L., J.-M. Torrenti and P. Acker (1997). "Thoughts about drying shrinkage:
experimental results and quantification of structural drying creep." Materials and
Structures 30(10): 588-598.
Gravina, R. J. (2002). "Non-linear overload behaviour and ductility of reinforced concrete

flexural members containing 500MPa grade steel reinforcement/by Rebecca Jane
Gravina."
Gravina, R. J. and S. T. Smith (2008). "Flexural behaviour of indeterminate concrete

beams reinforced with FRP bars." Engineering Structures 30(9): 2370-2380.
Gravina, R. J. and R. Warner (2003). "Local deformation model for reinforced concrete
members in flexure." Australian Journal of structural engineering 5(1): 29-36.
Gribniak, V. (2009). Shrinkage influence on tension-stiffening of concrete structures.

PhD, Vilniaus Gedimino technikos universitetas.
Gribniak, V., G. Kaklauskas and D. Bacinskas (2008). "Shrinkage in reinforced concrete

structures: A computational aspect." Journal of Civil Engineering and Management 14(1):
49-60.
Hansen, T. C. and A. H. Mattock (1966). Influence of Size and Shape of Member on the
Shrinkage and Creep of Concrete. ACI Journal Proceedings, ACI.
Haskett, M., D. J. Oehlers and M. M. Ali (2008). "Local and global bond characteristics
of steel reinforcing bars." Engineering Structures 30(2): 376-383.
He, W. (2013). Creep and Shrinkage of High Performance Concrete, and Prediction of
Long-Term Camber of Prestressed Bridge Girders. Masters, Iowa State University.
Holt, E. and M. Leivo (2004). "Cracking risks associated with early age shrinkage."
Cement and Concrete Composites 26(5): 521-530.
Huo, X. S.-O (2001). "Creep, shrinkage and modulus of elasticity of high performance
concrete." ACI Materials Journal 98(6), 440-449.
Jafarifar, N. (2012). Shrinkage Behaviour of Steel-Fibre-Reinforced-Concrete. PhD, The

University of Sheffield.
Jayasinghe, T. (2011). Prediction of time-dependent deformations in post-tensioned

concrete suspended beams and slabs in tall buildings. PhD, RMIT University Australia.
163
Jun, Z., H. Dongwei and C. Haoyu (2010). "Experimental and theoretical studies on
autogenous shrinkage of concrete at early ages." Journal of Materials in Civil Engineering
23(3): 312-320.
Kang, S. T., J. S. Kim, Y. Lee, Y. D. Park and J. K. Kim (2011). "Moisture Diffusivity of
Early Age Concrete Considering Temperature and Porosity." KSCE Journal of Civil
Engineering 16(1): 179-188.
Kim, J. K. and C. S. Lee (1998). "Prediction of differential drying shrinkage in concrete."

Cement and Concrete Research 28(7): 985-994.
Kim, J. K. and C. S. Lee (1999). "Moisture diffusion of concrete considering self-

desiccation at early ages." Cement and Concrete Research 29: 1921-1927.
Knight, D., P. Visintin, D. Oehlers and M. Jumaat (2013). "Incorporating residual strains
in the flexural rigidity of RC members with varying degrees of prestress and cracking."
Advances in Structural Engineering 16(10): 1701-1718.
Knight, D., P. Visintin and D. J. Oehlers (2015). "Displacement‐based simulation of time‐

dependent behaviour of RC beams with prestressed FRP or steel tendons." Structural
Concrete 16(3): 406-417.
Lam, J. P. (2002). Evaluation of concrete shrinkage and creep prediction models. Masters,
San Jose State University.
Lewis, R. W., B. A. Schrefler and N. Abd Rahman (1998). "A finite element analysis of
multiphase immiscible flow in deforming porous media for subsurface systems."
Communications in numerical methods in engineering 14(2): 135-149.
Mu, R. and J. Forth (2009). "Modelling shrinkage of concrete from moisture lost using
moisture diffusion theory." Magazine of Concrete Research 61(7): 491-497.
Mu, R. and J. P. Forth (2009). "Moisture shrinkage of concrete from moisture lost using
moisture diffusion theory." Magazine of Concrete Research 61(7): 491-497.
Muhamad, R., M. M. Ali, D. Oehlers and A. H. Sheikh (2011). "Load-slip relationship of

tension reinforcement in reinforced concrete members." Engineering Structures 33(4):
1098-1106.
Oehlers, D. J., I. Liu and R. Seracino (2005). "The Gradual Formation of Hinges
Throughout Reinforced Concrete Beams#." Mechanics based design of structures and
machines 33(3-4): 373-398.
Oehlers, D. J., M. Mohamed Ali, M. Haskett, W. Lucas, R. Muhamad and P. Visintin

(2011). "FRP-Reinforced concrete beams: unified approach based on IC theory." Journal
of Composites for Construction 15(3): 293-303.
164
Ong, K. C. G., L. R. Chandra and K. M. Lay (2010). "Early-Age Shrinkage Strains Versus
Depth of Low Water-Cement Ratio Mortar Prisms." ACI Matetrials Journal 107(3): 213-
221.
Pickett, G. (1946). "Shrinkage stresses in concrete." Journal of the American Concrete

Institute 17(3): 165-203.
Popovics, S. (1973). "A numerical approach to the complete stress-strain curve of

concrete." Cement and concrete research 3(5): 583-599.
Powers, T. C. and T. L. Brownyard (1946). Studies of the physical properties of hardened

Portland cement paste. ACI Journal Proceedings, ACI.
Prasada Rao, A., R. Jayaraman, V. Vimalanandam and S. S. Babu (1994). "Predicting

creep and shrinkage effects in cracked concrete elements." Journal of Structural
Engineering 120(9): 2784-2792.
Pulmano, V. A. and Y. S. Shin (1987). "Simplified finite-element analysis of deflections

of reinforced concrete beams." Structural Journal 84(4): 342-348.
Sakata, K. (1982). "A Study on moisture diffusion in drying and drying shrinkage of
concrete " Cement and Concrete Research 13: 216-224.
Sakata, K (1993). "Prediction of concrete creep and shrinkage of concrete." In: 5th
international RILEM Symposium, Barcelona, Spain.
Sakata, K. and T. Shimomura (2004). "Recent progress in research on and code evaluation
of concrete creep and shrinkage in Japan." Journal of Advanced Concrete Technology
2(2): 133-140.
Silliman, K. R. and C. M. Newtson (2006). Effect of misting rate on concrete shrinkage.

HPC: Build Fast, Build to Last. The 2006 Concrete Bridge Conference.
StandardsAustralia (2009). Australian Standard for Concrete Structures. AS3600-2009.

Sydney (Australia): Standards Australia. Sydney, AS 3600-2009.
Suwito, A., A. Ababneh, Y. Xi and K. William (2006). "The coupling effect of drying
shrinkage and moisture diffusion in concrete." Computers and Concrete 3(2/3): 103-122.
Terrill, J., M. Richardson and A. Selby (1986). "Non-linear moisture profiles and
shrinkage in concrete members." Magazine of Concrete Research 38(137): 220-225.
Visintin, P., D. Oehlers and M. Haskett (2013). "Partial-interaction time dependent

behaviour of reinforced concrete beams." Engineering Structures 49: 408-420.
Visintin, P., D. Oehlers, R. Muhamad and C. Wu (2013). "Partial-interaction short term

serviceability deflection of RC beams." Engineering Structures 56: 993-1006.
165
Visintin, P., D. Oehlers, C. Wu and M. Haskett (2012). "A mechanics solution for hinges
in RC beams with multiple cracks." Engineering Structures 36: 61-69.
Walker, D., M. Leonard, A. Metcalfe and M. Lambert (2009). Engineering Modelling and
Analysis, Taylor and Francis.
Zhang, J., H. Dongwei and S. Wei (2010). "Experimental study on the relationship
between shrinkage and interior humidity of concrete at early age." Magazine of Concrete
Research 62(3): 191-199.
Zhang, J., D. Hou and Y. Gao (2012). "Integrative studies on autogenous and drying
shrinkages of concrete at early-age." Advances in Structural Engineering 15(7): 1041-
1052.
Zhang, J., D. Hou, Y. Gao and S. Wei (2011). "Determination of Moisture Diffusion
Coefficient of Concrete at Early Age from Interior Humidity Measurements." Drying
Technology 29(6): 689-696.
Zhang, J., D. Hou and Y. Han (2012). "Micromechanical modeling on autogenous and
drying shrinkages of concrete." Construction and Building Materials 29: 230-240.
Zhang, J., K. Qi and Y. Huang (2009). "Calculation of moisture distribution in early-age

concrete." Journal of engineering mechanics 135(8): 871-880.
166

02whole PDF

Uploaded by

Copyright:

Available Formats

02whole PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

02whole PDF

Uploaded by

Copyright:

Available Formats

Time Dependent Flexural Analysis of

Reinforced Concrete Members

Noor Md. Sadiqul Hasan

A dissertation submitted for the degree of

Noor Md. Sadiqul Hasan Date

At first I would like to express my deepest gratitude and sincere appreciation to my

Proceedings of the Institution of Civil Engineers – Structures and Buildings.

shrinkage” (Submitted to Proceedings of the Institution of Civil Engineers – Structures

1.2 Scope of the Research ......................................................................................... 6

1.3 Aims and Objectives of the Research ................................................................. 6

1.4 Structure of the Thesis ........................................................................................ 7

Chapter 2 Literature Review ...................................................................... 9

2.1 Shrinkage and its types ....................................................................................... 9

2.2 Effect of Shrinkage and Creep on Structures ................................................... 12

2.3 Prediction of Shrinkage Strains ....................................................................... 14

2.4 Using Moisture Diffusion to quantify Shrinkage ............................................. 17

2.4.1 Moisture and Humidity Diffusion in Concrete.......................................... 18

2.5.1 Models to predict long term Deflections ................................................... 37

Chapter 4 Simulating Shrinkage Strain using Moisture Diffusion ....... 84

4.2 Moisture diffusion equation .............................................................................. 84

4.3 Moisture diffusion coefficient .......................................................................... 86

4.4 Finite difference method on moisture diffusion analysis ................................. 87

4.6 Quantification of moisture diffusion coefficient .............................................. 90

4.7 Four way flow in rectangular beam .................................................................. 92

4.8 Three way flow in a beam ................................................................................ 94

4.9 Two way flow in a beam .................................................................................. 95

4.10 One way flow in a beam ................................................................................... 96

4.11 First simulation with Asad, Baluch et al. (1997) .............................................. 98

4.12 Second simulation with Jafarifar (2012)......................................................... 100

Chapter 5 Proposed Experimental Work .............................................. 114

5.1 Introduction ................................................................................................. 114

6.1 Introduction..................................................................................................... 131

6.2 Partial interaction segmental analysis ............................................................. 132

6.3 Prior to cracking segmental analysis .............................................................. 133

6.4 Accommodation of cracking in the segmental approach ................................ 135

6.5 Partial interaction tension stiffening model .................................................... 137

6.6 Partial-interaction segmental model ............................................................... 143

6.7 Constant longitudinal shrinkage along depth and width................................. 144

6.8 Variation in longitudinal shrinkage strain along depth................................... 147

6.10 Parametric study ............................................................................................. 151

6.11 Application to test specimens ......................................................................... 154

Chapter 7 Conclusions and Recommendations..................................... 158

References ................................................................................................................. 161

Figure 1.2: Thesis layout of this research .......................................................................... 7

Figure 2.1: Shrinkage stages and types of concrete ........................................................ 10

Figure 2.10: Comparisons between experimental and numerical results (ambient

Figure 2.16: Experimental results of different concrete specimens drying shrinkage as a

Figure 6.1: A standard multi-crack segmental analysis ................................................ 133

Figure 6.2: Separating elements of RC beam (Concrete element) ................................ 135

Figure 6.3: Separating elements of RC beam (Reinforcement element) ....................... 135

Figure 6.5: Cracked segmental analysis ........................................................................ 137

Figure 6.6: Tension stiffening prism ............................................................................. 138

Figure 6.7: Local deformation of nth segment in prism ................................................. 140

Figure 6.8: Tension stiffening analysis ......................................................................... 141

Figure 6.18: Influence of exposed surfaces on member deflection ............................... 157

Table 5.2: Test details for material properties ............................................................... 116

adversely affect the durability of RC structures.

water contained within the porous structure of concrete.

 Chemical shrinkage, which is also known as autogenous shrinkage occurs due to