Professional Documents
Culture Documents
Lim 012372285
Lim 012372285
Lim 012372285
Author:
Lim, Char Ching
Publication Date:
2000
DOI:
https://doi.org/10.26190/unsworks/8964
License:
https://creativecommons.org/licenses/by-nc-nd/3.0/au/
Link to license to see what you are allowed to do with this resource.
by
September, 2000
CERTIFICATE OF ORIGINALITY
I hereby declare that this submission is my own work and to the best of my knowledge it
contains no materials previously published or written by another person, nor material
which to a substantial extent has been accepted for the award of any other degree or
diploma at UNSW or any other educational institution, except where due
acknowledgement is made in the thesis. Any contribution made to the research by
others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in
the thesis.
I also declare that the intellectual content of this thesis is the product of my own work,
except to the extent that assistance from others in the project's design and conception or
in style, presentation and linguistic expression is acknowledged.
Date: .
" ....... concrete in the good old days seemed to be simple. However, concrete was never
simple. We were! "
H.J. Gilkey
" ...... .it would be very difficult to construct a complete unified theory of everything ... all
in one go. So instead, we have made progress by finding partial theories that describe
a limited range of happenings and by neglecting other effects or approximating them by
certain numbers. "
Stephen W. Hawking
iv
ABSTRACT
This thesis reports an investigation on the influence of cracks on the chloride diffusivity
of OPC concrete and, hence, its service life prediction. Using data from the literature, a
trend is observed between the chloride threshold level in concrete and the surface crack
width/cover ratio (Wcr/C). The significance of using WcrlC as a parameter for durability
performance of a crack reinforced concrete is highlighted. Based on the permissible
crack width and minimum cover thickness allowed in various Codes of Practice, the
WcrfC is computed, compared and critically discussed. A comparison is made between
the chloride diffusion coefficient in the tension and compression zones of reinforced
concrete prisms having a WcJC of 0.01 but different quantity of tensile steel bar.
The main area of interest in this study, however, is the influence of microcracks on the
chloride transport in concrete. The effect of microcracks on the chloride permeability of
concrete under uniaxial compression is investigated. This is necessary to address the
existing conflicting views from previous studies. The non-destructive and destructive
methods of microcrack evaluation are used to characterise the microcracking behaviour
in the concrete during and after the compression tests, respectively. A comparison
between the characteristics of microcracks obtained from the two methods and the
measured electrical charge passed from the rapid chloride permeability test (ASTM
C1202) is made. The occurrence of critical stress is found to affect the charged passed
significantly.
Using the proposed model, the influence of cover variation on the service life prediction
of a structure is investigated. At low w/c, a small variation in the cover can significantly
influence the service life prediction. However, the influence of cover variation on
service life prediction diminishes as w/c is gradually increased.
The present findings show that when concrete is sustained under uniaxial compression
up to 0.5fc. stress level, chloride diffusivity in concrete is not adversely affected by the
development of microcracks. On the other hand, compressive stresses are found to
impede chloride diffusion in concrete. The chloride diffusion coefficient shows a
progressive reduction with increasing sustained stress levels. An approach for a model
to predict the chloride concentration in concrete, taking into consideration the effect of
compressive stresses, is proposed.
VI
ACKNOWLEDGEMENTS
The author would like to express his sincere gratitude to his supervisors, Dr. N.
Gowripalan from the School of Civil and Environmental Engineering, University of
New South Wales (UNSW) and Dr. V. Sirivivatnanon from the Commonwealth
Scientific and Industrial Research Organisation (CSIRO), Division of Building,
Construction and Engineering, Sydney for their invaluable guidance, constructive
comments, challenging ideas and timely advice throughout the course of this work.
Thanks to Dr. R. Khatri for assisting me in some of the experimental works carried out
at the Cement and Concrete Technology Laboratory at CSIRO, Sydney. The assistance
rendered by the staff of the Concrete Laboratory at the University of New South Wales
are gratefully acknowledged.
The author would like to thank the Federal Government of Malaysia for providing the
financial assistance and granting him leave to pursue this study. He is indebted to the
Public Works Department of Malaysia for giving him the opportunity to pursue this
study. He would also like to acknowledge the partial financial support from CSIRO
Building, Construction and Engineering, Sydney for this collaborative research project
between UNSW and CSIRO.
The author would like to dedicate this work specially to his beloved mother, Madam
Lee Lian Tin, who is the driving force behind his academic achievement. She has been
his mentor since childhood days, his motivator and his constant source of inspiration.
The encouragement, the thoughts and the many good wishes from his family members
throughout the duration of this study are gratefully appreciated.
vii
TABLE OF CONTENTS
Page
CHAPTER 1: INTRODUCTION
GLOSSARY
Apparent Poisson' s ratio Refers to any ratio of total transverse strain to total
axial strain.
Chloride binding The ratio of the change in the bound chloride ions
capacity (Cb) to the change in the free chloride ions (C1).
Design service life Service life used in the design of structures to take
into account the time related scatter, and to provide a
required safety against falling below the target
service life.
Initiation period The time taken for chloride ions to reach the steel
reinforcement and to initiate corrosion.
Microcracks Cracks less than 0.01mm wide which are only visible
under microscope.
Service life model Mathematical function for evaluating the service life.
Specific crack area The increase in the area of microcrack per unit cross-
sectional area of the concrete.
xviii
Steady state diffusion When the concentration of the chloride at any point
in the concrete does not change with time, i.e.,
ac
at= 0.
xix
NOMENCLATURE
C concrete cover
k constant
m empirical coefficient
ex. constant
constant
transverse strain
axial strain
diameter of bar
xxii
LIST OF FIGURES
Figure 2.8 Schematic Diagram for the Transport Process in a Sea Wall
Figure 2.16 Flow Diagram for Service Life Calculation on the Basis of
Measurements at Times t0 and tc (Planned Inspection)
Figure 2.17 Flow Diagram for Service Life Calculation on the Basis of
Measurement Once during Lifetime (Inspection Once during
Service Life)
Figure 3.1 Relationship between Chloride Threshold Level and the Inverse
of W cr/C (Linear Relationship)
Figure 3.3 The Effect of Changing W crlC on the Chloride Threshold Level
(a) Crack in Atmosphere
(b) Submerged Crack
Figure 3.10 Chloride Profiles of Concrete Prisms Containing 1R8 Bar after
300 Days in Salt Solution
Figure 3.11 Chloride Profiles of Concrete Prisms Containing 2R8 Bars after
300 Days in Salt Solution
Figure 4.7 Average Specific Crack Area in Concrete During Loading and
Unloading
Figure 4.9 Relationship between Stress-Strength Ratio and Strain at 0.5 fc.
Figure 4.10 Relationship between Stress-Strength Ratio and Strain at 0.85 fc.
(critical stress not exceeded)
XXV
Figure 4.11 Relationship between Stress-Strength Ratio and Strain at 0.85 fe.
Figure 4.12 Relationship between Stress-Strength Ratio and Strain at 0.95 fe.
Figure 5.1 Effect of Initial Curing Time (t;) on the Diffusion Coefficient
against Chloride Exposure Time
Figure 5.7 A Comparison between the Empirical Coefficient (m) and the
Concentration of Chloride Solution
Figure 5.8 A Comparison between the Empirical Coefficient (m) and the
Duration of Moist-Curing
xxvi
Figure 5.9 Analysis Using All Laboratory Chloride Immersion Tests Data
Figure 5.10 Analysis Using Field Exposure Tests from the Literature
Figure 5.11 Analysis Using Data from In-Service Structures from the
Literature
Figure 5.13 Relationship between D2s and w/c for OPC Concrete
Figure 7.2 Relationship between Creep of Specific Crack Area and Axial
Creep
Figure 8.1 Chloride Profiles of Grade 20 Concrete (Control) after 540 Days
of Immersion
xxviii
Figure 8.2 Chloride Profiles of Grade 40 Concrete (Control) after 540 Days
of Immersion
Figure 8.5 Reduction in Dc.cr with Sustained Stresses for Grade 20 Concrete
Figure 8.6 Reduction in Dc,cr with Sustained Stresses for Grade 40 Concrete
LIST OF TABLES
Table 2.2 Number of Drill Holes for Collecting Concrete Drill Powder
Table 8.2 A Comparison between Measured and Predicted Dc,cr for Grade
Table 8.3 A Comparison between Measured and Predicted Dc,cr for Grade
Chapter 1
INTRODUCTION
1.1 GENERAL
Many design codes and specifications regard concrete compressive strength as a primary
indicator of the potential durability of concrete, even though it may not be a valid
criterion. The compressive strength of a standard concrete cube or cylinder constitutes
the mean value of a property of an entire cross-section, whereas concrete durability is
governed primarily by the properties of the concrete cover (Rostam, 1996) and its
soundness, that is, freedom from cracking (Mehta, 1997). At present, the approach to
concrete durability is still strength-oriented (Mehta, 1994).
Even though some codes are beginning to define the concept of service life, there is no
explicit durability design criterion yet developed. Therefore, the design of new concrete
structures for durability still relies on the conventional methods of specifying limiting
values to meet certain concrete "durability" requirements. These limiting values are, for
example, concrete strength, cover, maximum and minimum cement contents, maximum
w/c (or w/b) ratio and allowable surface crack width (Andrade and Alonso, 1996; Kropp
and Hilsdorf, 1995). As a result, many structures, especially those exposed to
aggressive marine environments, are suffering from premature deterioration problems
(Monica et al, 1996; Yokozeki et al, 1997).
directly related to its impermeability, and hence, its durability, needs to be reviewed
critically. As an example, supposing that the concrete strength is increased by
increasing the cement content at a given water content. In this case, the resistance of
concrete to cracking would decrease due to an increase in both the drying shrinkage and
thermal gradient. At the same time, an increase in strength will tend to increase the
elastic modulus and reduce the creep coefficient. This will again adversely affect the
crack resistance of the concrete due to high stress development and less stress relieve.
Thus, a performance-based design criterion should be viewed as the most appropriate
approach to concrete durability (Kropp and Hilsdorf, 1995).
At present, the concept of service life prevalent in most design codes and specifications
does not explicitly address this problem. It is primarily based on prescriptive notional
concepts (Sharp, 1996) which do not provide any rational basis on how to design for a
specific service life from the viewpoint of performance based criteria. A greater
emphasis is now being placed on the importance of durability design by performance
criteria. This is evident from the RILEM TC 130-CSL report on the durability design of
concrete structures (Sarja, 1996). Research works are now focusing more towards the
development of new methodologies in order to be able to predict and quantify the
required service life of a structure (Andrade and Alonso, 1996) especially those exposed
to marine environments.
There is a world-wide trend towards a clear identification of the required service life
during the actual design process (BS7543: 1992; CS 109: 1996). An explicit design for
durability requires analytical models of the various deterioration processes of the
concrete to be predicted with some measures of accuracy. However, at this stage, this
aim has not been reached yet (CS 109: 1996).
The principal causes responsible for deterioration of concrete structures are the
corrosion of reinforcing steel, freezing and thawing, alkali-silica reaction, and sulphate
attack. A well-constituted, properly consolidated, and cured concrete is essentially
Chapter I : Introduction 3
In a marine environment, the service life of a concrete structure depends on how rapid
chloride induced corrosion is initiated on the reinforcing steel. There are many factors
that can influence the rate of chloride ion transport in concrete. Among them are
chloride concentration gradient, temperature, humidity, pore structure of concrete,
degree of saturation of concrete and cracks. Concrete which is cracked has a great
influence on its permeability and subsequently its durability (Mehta, 1994). Even
though reinforced concrete structures are designed to crack, the prediction of chloride
ion transport in concrete does not account for this effect (Pettersson and Sandberg,
1997). Limited information is available on the factual influence of these cracks on the
service life prediction and hence further research in this area is very much needed.
Chloride induced corrosion of reinforcing steel is a major cause of rapid and premature
deterioration of concrete (Mackechnie, 1995; Bamforth, 1996; Maage et al, 1997).
Cracked concrete allows corrosion process to initiate much faster than uncracked
concrete (Lorentz and French, 1995). The initiation of rebar corrosion in cracked
concrete is dependent on the surface crack width (Poston et al, 1987). Wider surface
crack widths have been found to induce corrosion much faster than relatively smaller
ones (Makita et al, 1982; Raharinaivo et al, 1986; Suzuki et al, 1989). On the other
hand, some researchers (Arya and Ofori-Darko, 1996; Schiess} and Raupach, 1997)
have reported lower corrosion initiation in cracked concrete when the cover is increased.
This is because the rate of corrosion depends on the availability of oxygen not at the
crack but in the sound concrete on the cathodic end of the rebar and, hence, on the rate
at which oxygen can diffuse through the cover (Beeby, 1978).
From the discussion above, it appears that for a given cover thickness, the initiation of
corrosion increases with increasing surface crack width. Similarly, for a given surface
crack width, the initiation of corrosion reduces with increasing cover thickness. It shall
be noted that for a given stress level in the rebar, with other variables being constant, the
surface crack width increases proportionally with cover thickness (Beeby, 1978).
Chapter I : Introduction 4
Therefore, it appears that there is a "trade-off' between surface crack width and cover
thickness when the corrosion initiation in the rebar becomes least affected by the
combination of these two parameters. Hence, there is a need to assess the suitability of
using the crack width/cover ratio as a parameter for durability performance of a cracked
reinforced concrete.
Information on the effect of chloride diffusivity in the compression and tension zones of
reinforced concrete beam in flexure is important especially when considering the
flexural crack width which is within the permissible limits stipulated in the codes. For a
given allowable surface crack width, the influence of the quantity of tensile rebar on the
chloride diffusivity is of particular interest. An extensive literature review showed that
this information is particularly scarce and hence needed further investigation.
From the practical viewpoint, most concrete structures are subjected to some magnitude
of stresses at any time during its service life. For example, the compression member of
a marine structure, such as a bridge pier, is normally subjected to stresses which are
predominantly in compression. These stresses may be large depending on the service
loads and the spans of the superstructure. In addition, the various exposure conditions
such as spray, splash, tidal and submerged, are all occurring at this compression
member, making it one of the most vital part of a marine structure. Hence, an
understanding of the chloride ion transport into concrete while under sustained
compression is necessary. A comparison between the chloride diffusivity through a
concrete under short-term uniaxial compression (instantaneous load) and that under
sustained compression is of particular interest. This is because many laboratory-based
chloride tests are carried out on concrete specimens after the completion of the
compression tests, but in reality, chloride transport in concrete occurs under the
influence of stresses.
The effect of microcracking on the chloride ion transport in concrete had been studied
by some researchers in the recent past (Samaha and Hover, 1992; Saito and lshimori,
1995). In the previous studies, microcracks were quantified in terms of crack length by
examining a concrete slice cut from a cylinder after the compression test. The observed
crack length was then interpreted with respect to the measured electrical charge passed
Chapter 1 : Introduction 5
through a concrete specimen using a rapid chloride permeability test (RCPT). In some
instances, crack measurement and RCPT were carried out on a specimen taken from a
different concrete cylinder. Thus, the interpretation of results may be dubious. It was
not known if there is a difference between the crack length in the specimen during
loading and when completely unloaded since crack measurement and RCPT were
conducted after the completion of the compression test. No attempts have been made to
characterise the microcracks during the compression test prior to the chloride
permeability test. In addition, there are conflicting views as to the levels of compressive
stress which affect the chloride permeability of the concrete.
There is evidence showing that the characteristics of microcracks after a concrete has
been unloaded are likely to be different from those while under load (Wang et al, 1997).
A simple and non-destructive method of microcrack evaluation has been developed by
Loo (1992). This method enables the study of the progressive microcracking in the
concrete during the compression test. Using this method, microcracks are quantified in
terms of specific crack area which is defined as the increase in the microcrack area per
unit cross-sectional area of the concrete specimen. A comparison between the
characteristics of microcracks in terms of the specific crack area and crack length is,
thus, necessary to explain the chloride permeability of the concrete in relation to the
stress level.
concrete mix design parameters is, thus, necessary for this purpose. The model can then
be used in conjunction with the solution to Fick's Second Law of pure diffusion for
predicting chloride concentration in concrete. In addition, information on how sustained
compressive stresses affect the chloride diffusivity of concrete with time is lacking and
hence further research is required. The research program is shown in the flow chart in
Figure 1.1.
1.3 OBJECTIVES
In the light of the above discussion, the present study is primarily aimed at developing a
service life prediction model considering the effect of microcracking in concrete
subjected to sustained compressive stresses. In addition, the influence of macrocracks
on the chloride diffusivity of reinforced concrete in flexure is investigated. The main
objectives of the present investigation are as follows :
a) To assess the suitability of the surface crack width/concrete cover ratio (Wc/C) as a
parameter for the durability performance of a cracked reinforced concrete.
b) To determine the effect of tensile steel quantity on the chloride diffusivity in the
compression and tension zones of reinforced concrete prisms having a constant
WcrlC.
c) To study the effect of uniaxial compressive stress levels on the chloride diffusion in
concrete under short-term (instantaneous) and sustained loads.
f) To modify the time-dependent diffusion model in (e) to account for the effect of
sustained compressive stress levels.
g) To propose a modified solution to Fick's Second Law of pure diffusion for chloride
concentration prediction in both stressed and unstressed concrete.
In this study, Grade 20 and 40 concrete mixes were designed using locally available
Nepean crushed gravel coarse aggregate with a maximum size of 19mm and Nepean
sand. Concrete mixing and casting procedures were carried out in accordance with the
Australian Standards, AS 1012 Pt.2. General Purpose cement (Type GP, similar to
ASTM Type 1) complying with AS 3972: 1997 was used throughout the study.
The following parameters were considered in the study of the effect of microcracks on
the chloride diffusivity.
Microcrack evaluation was carried out using the non-destructive method developed by
Loo (1992). In addition, an estimate of the crack length using microscopic observations
was also carried out.
~
Pre-cracked
+
Data from Uniaxial
+ Specimens in
RC prisms literature compression tests chloride
I
I immersion tests
I •I • + +
I Sustained I
Cover
I
Crack
width ,'
!Instantaneous
I I
+
Sustained •
Without
+
Chloride
I Time-dependent
diffusion model
(unstressed concrete)
•
Microcrack
evaluation
,'
Chloride
immersion
+
I RCPT I
stresses stresses
immersion
tests
tests ,,
Time-dependent
diffusion model Determination
~
' . ,,
- Modified solution to
----• Fick' s Second Law of
Diffusion
...
~
Comparison of
results
The chapters are divided in accordance with the major topics. A comprehensive
literature review is presented in Chapter 2.
Chapter 4 addresses the conflicting views from previous studies on how stress levels
affect chloride permeability of concrete which has been subjected to a uniaxial
compression test. Microcrack evaluation is carried out by means of two different
methods, i.e., the non-destructive method during the compression test and the
destructive method at the end of the compression test. Both methods of evaluation are
conducted on the same specimen prior to the rapid chloride permeability test (RCPT).
A comparison between the characteristics of microcracks obtained from the two
methods and the measured electrical charge passed from RCPT is made with respect to
the stress level in which the concrete is tested.
It compares the results obtained from the non-destructive method and the destructive
method.
Chapter 2
LITERATURE REVIEW
2.1 INTRODUCTION
The use of Portland cement concrete in the construction of marine structures dated as far
back as 1849. Reinforced concrete is about 100 years old, but it has been used as a
construction material for port structures from 1897, since when the sea water
environment has served as an accelerated testing ground for the concrete durability
(Sharp, 1996).
The marine environment represents one of the most aggressive environments on the
earth (Mehta and Gerwick, 1982). Concrete has been successfully used for many years
in marine structures and many reports on the long term durability of concrete, have
generally confirmed the excellent behaviour of concrete in the oceans. However, some
cases of serious premature deterioration, associated with the reinforcement corrosion
and subsequent spalling of the concrete also exist.
The salinity of sea water varies between 3.0 g/1 and 37.6 g/1 with lower salinity in the
polar regions than the tropical areas. Table 2.1 shows the chloride content of sea water
in various parts of the world. The pH of sea water is normally between 7 .5 and 8.4 but
may be lower than 7 .5 in sheltered bays due to high concentration of dissolved carbon
dioxide and hydrogen sulphide from marine organisms (Mehta and Monteiro, 1993).
Sea water contains varying types of marine organisms such as barnacles and molluscs
which can damage concrete surfaces by secreting acids and boring holes. For concrete
in submerged zone, a massive growth of marine plants on the concrete surface can serve
as a physical barrier to the ingress of aggressive ions (Roy et al, 1993).
Chapter 2 : Literature Review 12
concrete
reinforcing steel
Atmospheric zone
cracking due to
steel corrosion
cracking due to
freeze-thaw, thermal
& humidity gradients
Tidal zone
physical abrasion
due to wave and
floating objects
Figure 2.2 is a holistic model (Mehta, 1994) which illustrates the physico-chemical
changes occurring simultaneously in the steel and concrete in response to attack by the
penetrating ions. A comprehensive overview of the mechanism of chloride transport in
concrete and the subsequent chemical and physical interactions with the concrete
material that leads to eventual deterioration of concrete is given in section 2.5.
The performance of concrete in a sea water environment has been the subject of
observations and research interests over many years worldwide (Sharp, 1996). In the
past 20 years, there has been a huge upsurge in research and publications owing to the
problems caused by chloride related deterioration of maritime structures.
In spite of the above, knowledge of the durability of concrete in sea water environment
and guidance for the general practitioner remain fragmented mainly because exposure
Chapter 2 : Literature Review 14
conditions vary widely and the deterioration processes take so long to appear that
separation of cause and effect becomes not so easy such as in a situation of cracking-
corrosion interaction.
I Reinforced Concrete I
- I
'
I
I
I A: Expansion of Rust Increases Pressure in Pores
I B: Reduction in Concrete Strength and Stiffness
I
-- - - -1
I
'
Expansion and Cracking of Concrete
I
Figure 2.2 : Model of Reinforced Concrete Damage from Exposure to Sea Water
or Deicing Salts (Mehta, 1994)
The alkalinity of uncarbonated concrete is normally in the range of pH12 to 13.5. This
high alkalinity in concrete is attributed to the high concentrations of soluble alkaline
hydroxides, mainly calcium, sodium and potassium hydroxides, present in the pore
solution.
Steel becomes passivated when embedded in concrete having this range of pH values.
A thin passivating layer, often referred to as gamma ferric oxide, is formed on the steel
surface. This passivating layer is stable and protects the steel against corrosion even in
the presence of moisture and oxygen. However, the passivating environment is not
always maintained. Two processes can break down the passivating layer of steel in
concrete. Firstly, when the alkalinity of the concrete drops to a level where the pH is
between 9 and 10 due to carbonation. Secondly, due to chloride attack even at high
alkalinity (Heiman, 1982; Mehta and Gerwick, 1982; Broomfield, 1997).
Basically, the physical and physico-chemical aspects of protection are related to the
depth and permeability of the concrete cover.
Besides providing a physical barrier to the ingress of aggressive ions to the steel,
concrete cover also serves to contain the expansive forces as a result of steel corrosion.
When high quality concrete is used, engineers begin to question the need for thick cover
on reinforcing steel. Moderately thick cover can be justified by the fact that it would
lead to the development of a few wider cracks under overstress, whereas a thinner cover
results in a more microcracking through dispersal of many small cracks. For a given
quality of concrete, moderately thick cover does provide a better protection against
corrosion (Mehta and Gerwick, 1982).
Chapter 2 : Literature Review 16
Although design codes generally call for increased cover with increasing severity of
exposure, there is no general agreement on safe limits for cover thicknesses. The likely
effectiveness of reduce cover is closely linked to concrete quality (Heiman, 1982).
One of the most important quality of concrete cover as a barrier of steel protection is its
permeability against the ingress of moisture, oxygen and chlorides. Low permeability
will reduce the electrical conductivity within the paste, thereby impeding the galvanic
action.
shrinkage, creep, elastic modulus and tensile strength. It can be seen from Figure 2.3
that the increase in cement content would reduce the crack resistance of the concrete due
to the increase in both drying and thermal effects (Mehta, 1997).
" The attack process can be regarded as a battle between the environment on
one side and the concrete cover on the other side which acts as the defence system for
the steel and for the structure itself. The enemies range from the chlorides, moisture,
oxygen ...... .
In a marine environment, chloride ions are able to diffuse readily through the concrete to
the reinforcement even in high quality concrete (Gjorv and Vennesland, 1979). It has
been found that large amount of sodium chloride penetration has virtually no effect on
the pH of concrete which may remained between 12.4 and 13.0 except when carbonated
(Hime and Erlin, 1981 ). When the chloride concentration at the level of the steel has
reached a certain threshold level, depassivation of the steel begins. The mechanism
through which the gamma ferric oxide (Fe203) film is destroyed is not fully understood.
However, three theories have been proposed (Arya and Newman, 1990), vis-a-vis;
a) Complex formation theory - that chloride ions form a soluble complex with the
ferrous ions which then diffuse from the anode. Decomposition of the complex
molecule precipitates the iron and allows the chloride ions to be recycled (Hoar and
Jacob, 1967).
b) Oxide film theory - that chloride ions penetrate the oxide layer which surrounds the
steel through defects in the film, thereby accelerating the transport of ferrous ions
away from the anodic sites.
Chapter 2 : Literature Review 18
c) Adsorption theory - that chloride ions are adsorbed preferentially on to the metal
surface in competition with dissolved oxygen and hydroxyl ions. The high reaction
rate of metals with chloride helps produce soluble species of iron and chloride
which facilitate anodic dissolution.
The chloride threshold level is dependent on many factors and is discussed in detail in
section 2.5.5. Once the passivating layer has been destroyed, corrosion of steel can
initiate immediately when sufficient moisture and oxygen are available at the steel
surface (Concrete Society TR44, 1995; Schiess! and Raupach, 1997). Chloride induced
corrosion is particularly prone to well defined macrocells formation in which the areas
of pitting corrosion is separated by areas of 'clean' steel. This is partly due to the
mechanism of chloride attack, with pit formation and with small concentrated anodes
being 'fed' by large cathodes. It is also because chloride attack is usually associated
with high levels of moisture (hygroscopic effect) giving low electrical resistance in the
concrete and easy transport of ions so that anodes and cathodes can be separated easily
(Broomfield, 1997).
Corrosion pit results from a local breakdown of the passive film on the steel surface by
chlorides. This localised depassivating area becomes the anode. The area of the steel,
at some distance away from the anode, which is not depassivated becomes the cathode.
The separation of the corroded areas does not necessarily represent the distribution of
chlorides along the steel bar (Broomfield, 1997).
4(0ff) 02
./
---•---- - - .. - ..
- • - current
Typical anodic and cathodic reactions can be represented as follows (Mehta and
Gerwick, 1982; Concrete Society TR44, 1995);
The Off ions generated at the cathodic area will increase the local alkalinity and thus
strengthen the local passive layer warding off the effects of chloride ions at the cathode.
The hydroxyl ions will then seek their way towards the anode. The rate of which is
dependent on the resistance of the path through the concrete between the anode and the
cathode. This physical hindrance to the Off movement can be described as ohmic
resistance which is dependent on the quality and material properties of concrete
(Vesikari, 1988). At the anode, the Off ions neutralise the diluted Fe2+ ions into ferrous
hydroxide;
Chapter 2 : Literature Review 20
The ferrous hydroxide will be further oxidised into hydrated ferric hydroxide (or red
rust) and are generally deposited near the anode (Pettersson and Sandberg, 1997). The
electrochemical reactions are shown below (Broomfield, 1997) ;
Unhydrated ferric oxide, Fe203, has a volume of about twice that of the steel it replaces
when fully dense. When it becomes hydrated, it swells even more and becomes porous.
The volume increase can be many times. In Figure 2.5, the- relative volume of the
corrosion products is compared to a unit volume of the original iron metal (Vesikari,
1988). This expansion may cause cracking and spalling of the concrete cover.
Fe I
Fe304 I
Fe(OH)z I
Fe(OHh I
Fe(OH)J. H20 I
I I I I I I I I
0 1 2 3 4 5 6 Volume (cm3)
The important factors in corrosion of steel in concrete compared with most other
corrosion problems are the volume of oxide and where it is formed. In most aqueous
environments, the excess volume of oxide is transported away and deposited on open
Chapter 2 : Literature Review 21
surfaces within the structure (Andrade and Alonso, 1996). For steel in concrete, two
factors dominate. The main problem is that the pore water is static and there is no
transport mechanism to move the oxide or rust away from the steel surface. This means
that all the corrosion product formed will be deposited on the steel surface. The second
problem is that, the rust is not dense. It is very porous and takes up extremely large
volume than that of the original volume of steel consumed.
Cement matrix
Passivating layer
Steel bar
Chloride ions act as catalysts to corrosion. They are not consumed in the process but
help to breakdown the passive layer on the steel and hence allows corrosion process to
proceed quickly. In the corrosion reactions, chloride may form only intermediate
corrosion products, such as FeC}i. Subsequent hydrolysis liberates the chlorides again
so that it can be recycled elsewhere in the steel (Kropp and Hilsdorf, 1995).
Chapter 2 : Literature Review 22
At the macroscopic level, concrete is a composite material consisting of coarse and fine
aggregates dispersed in a cement paste matrix. The randomness of the concrete
structure is in the order of centimetres (corresponding to the size of aggregate) down to
the order of nanometres (corresponding to the size of CSH structure). At the
microscopic level, the paste is not homogeneous, with variations caused by transition
zones at the aggregate-mortar interface, the pore size and connectivity.
Pores do exist in concrete either as gel pores or capillary pores. In addition, entrained
air bubbles and entrapped air voids may also be present. The hardened cement paste is
considerable more porous than the aggregate which controls the movement of fluid
through concrete. Figure 2.7 shows the pore size distribution in concrete (Mehta, 1982;
Gowripalan et al, 1990).
0
Entrapped Air
Ion Size( A).
s2- Voids - Larger
3.68 0
cr 3.62 than 107 A
o2- 2.70
lO
0 0
The void may vary from the interstitial gel pores of less than one nanometre up to over
one millimetre for the entrapped air voids. Capillary pores are spaces in between cement
particles which are not filled up by cement gel. During hydration, the capillary pore
size, as well as the overall capillary porosity, decreases due to the consumption of water
for hydration process. The hydration products, mainly CSH, will gradually fill the
capillary pores. With prolong hydration, the size of the capillary pores may reduce
down to I O nanometre which is on the lower end scale of the pore size (Young, 1988).
These capillary pores may eventually become air-filled voids when water escapes from
the concrete. On the other hand, the size of gel pores is fixed by the CSH structure and
they remain constant in size. The resulting pore size and connectivity will determine the
resistance of the concrete to the movement of fluid.
The rate of diffusion, through water-filled pores, in the concrete is affect by the
characteristics of the porosity namely the pore volume and pore size. For concrete with
equal pore volume, diffusion will be much slower when the pore volume is made up of
many minute pores than a few larger ones. The smaller the pore size, the greater the
tortuosity of the path an ion has to follow (Kumar et al, 1987).
Immediately after mixing cement and water, the solid phase formed are discontinuous
so that freshly mixed cement paste is a viscous liquid. The solid phase is then built up
through random growth of hydration products and at some point becomes continuous
across the cement paste, mainly due to CSH formation (Chen and Odler, 1992). The
point at which the capillary pore space loses continuity is called percolation threshold.
As hydration products are formed, pieces of the capillary pore space will be trapped and
cut off from the main pore network, thus reducing the fraction of the pores that form a
connected pathway for transport. As this process continues, the capillary pore space
may lose all long-range connectivity.
Ca(OHh causes a reduction in the alkalinity of the concrete. This may lead to the
decomposition of hydration products especially CSH resulting in lost of strength and
durability (Mackechnie, 1995).
The interfacial transition zone is the region at the aggregate-mortar interface which
extend up to 50µm wide from the aggregate surface into the bulk cement paste. The
very weakest part of this interfacial zone lies not right at the physical surface, but 5 to
10µm away from it, within the paste fraction. The interfacial zones are, in general,
profoundly different from the bulk cement paste, in terms of morphology, composition
and density. The microstructure of ITZ is characterised by high porosity, relatively
larger pores and contains large crystals of Ca(OHh (Garboczi and Bentz, 1991). The
ITZ is not only unique to aggregate surface (as in concrete) but is also present at other
interfaces in cement based composites such as steel-cement interface (Monteiro et al,
1985) and steel fibre-cement interface (Bentur et al, 1986).
Garboczi and Bentz (1991) attributed the formation of ITZ to two major causes, namely,
the particle-packing effect and the one-sided growth effect. The particle-packing effect
arises from the fact that particles cannot pack together as well near a flat surface as in
free space. Since aggregate surface is typically many times larger than cement particles,
locally the aggregate edge appears flat to the surrounding cement particles. This
inefficient packing causes high porosity at the ITZ. The one-sided growth effect arises
because, for capillary pore located within the ITZ, hydration products only come from
the cement particle on one side but not from the inert aggregate. While a similar
capillary pore located in the bulk paste may be filled with cement gel coming from all
surrounding directions. Pozzolanic materials such as silica fume can improve the ITZ
significantly by densifying its microstructure.
Chapter 2 : Literature Review 25
Some of the reaction products of sea water attack on concrete have shown to be
beneficial to concrete durability due to their low solubility which helps to block pores
from further sea water ingress. Research works have shown that a protective layer may
form on the concrete surface due to the deposition of magnesium hydroxide (brucite)
and calcium carbonate (aragonite) (Conjeaud, 1980; Pettersson, 1996). The protective
layer of brucite and aragonite on the concrete surface is usually < IOOµm thick and may
be disrupted by wave action and abrasion (Mackechnie, 1995). The mechanisms of
chloride ingress in concrete are discussed more comprehensively in the following
subsections.
Chapter 2 : Literature Review 26
Chlorides may be introduced into concrete from a variety of sources. They can be cast
into the concrete or they can diffuse in from external sources such as sea water.
Chloride cast into concrete can be due to;
Natural aggregates are a common source of chlorides in concrete. For most cases, a
level of 0.003% by weight of concrete (0.07 kg/m 3) has been assumed (West and Hime,
1985) in order to account for traces of chlorides in the mix constituents. However,
natural aggregates may contain as much as 0.03% or more, particularly in some coastal
regions. When polymers made from chlorinated hydrocarbons are incorporated in the
alkaline medium of hardened cement, progressive dehydrochlorination can produce
harmful levels of chloride in several years (West and Hime, 1985).
In contrast, diffusion of chlorides from external sources into concrete constitute a major
problem in most parts of the world today. These external sources are primarily from sea
water exposure or from the use of deicing salts.
Chapter 2 : Literature Review 27
The continuous network of the capillary pore system of the hardened cement paste, the
coarse pore system of the aggregate-paste interface and the presence of microcracks
(bond and mortar cracks) provide the paths along which chloride ions can be transported
into concrete. Different physical and/or chemical mechanisms may govern the transport
of these media into the concrete. These mechanisms are dependent on the substance
flowing and its local concentration, the environmental conditions, the pore structure of
concrete, degree of saturation of the pore system and temperature. Considering the wide
range of pore sizes and the varying moisture condition in the concrete, the mechanism
of chloride transport into concrete, in most cases, is governed by one or a combination
of several mechanisms. The various mechanisms are permeation, capillary suction and
diffusion. Figure 2.8 is a schematic illustration of these transport processes (Concrete
Society, 1996).
+-·······
.
Diffusion of .
salt from sea.. ··
water tl'
..
tl'
Figure 2.8 : Schematic Diagram for the Transport Process in a Sea Wall
(Concrete Society, 1996)
Chapter 2 : Literature Review 28
2.5.2.1 Permeation
Permeation is a transport process whereby the chloride ions are driven into the concrete
under a hydrostatic pressure. Permeation may be a relevant transport mechanism for
chloride ingress only in the submerged zone of marine structures or for the seepage of
solutions through retaining structures. However, permeation does not constitute a major
transport mechanism in marine concrete. It is only sensitive to the large pore fraction
caused by cracking and other defects (Darr and Ludwig, 1973).
Capillary suction is a process in which chloride ions are drawn into concrete by capillary
action of the pore system absorbing a chloride-containing solution. The absorption of
chloride-containing solutions may be considered as the major cause of chloride
contamination of concrete that is subjected to alternating wetting and drying cycles.
During wetting period, the near surface concrete layer readily absorbs the chloride
solution. Subsequently, the water will evaporate and the salt remains in the pore system
of the near surface region. Thus, repeated wetting and drying cycles increase the salt
concentration in the pore system, which may become even higher than the concentration
of the chloride solution at the surface. Depending on the relative humidity of the
environment, the hygroscopicity (ability to retain moisture) of the salt prevents or
lowers the evaporation of water from the concrete, thus increasing also the moisture
concentration in the concrete (Kropp and Hilsdorf, 1995).
2.5.2.3 Diffusion
Diffusion of chloride ions into concrete is caused by the chloride concentration gradient
across the concrete section. Although this transport mechanism does not depend on the
flow of water as a vehicle for chloride ions, water plays an essential role in this transport
process. Diffusion of chloride ions can only take place when there is a significant
moisture content in the concrete. This is essential to provide a continuous liquid path in
Chapter 2 : Literature Review 29
the capillary pore system for the chlorides to be transported. Diffusion mechanism
ceases if the liquid paths become interrupted due to drying. The highest diffusivity can
be expected for a fully saturated concrete while a moisture concentration in the concrete
in equilibrium with approximately 60 - 80% relative humidity may be regarded as the
lower limit for diffusion (Kropp and Hilsdorf, 1995).
Diffusion rates are dependent on the temperature, saturation level of concrete, types of
diffusant and inherent diffusivity of the concrete. Diffusion into concrete is complicated
by chemical interactions, partially saturated condition and defects such as cracks and
voids.
Fick's Second Law of pure diffusion is commonly used to predict the long-term chloride
transport into concrete in a marine environment (Alexander and Streicher, 1997). While
the application of this law may be valid for concrete in the submerged zone, it is
probably inaccurate for the splash zone which is subjected to frequent cycles of wetting
and drying. Chloride transport in the splash zone, may to some extent, be attributed to
capillary suction and permeation. However, for high quality concrete and in most
practical cases, these transport mechanisms result in chloride profiles that are quite
similar to that of diffusion, provided that the concrete is not seriously degraded
(Sandberg, 1995). Further justification was given by Polder (1995). On the time scale
of the service life of a concrete structure and with the relatively high cover depths,
chloride transport is dominated by diffusion (Polder, 1995). It was reported that
capillary suction is only significant for small cover depths and on the short-term
exposure (Bamforth and Chapman, 1994). This is probably because concrete rarely
dries out to more than 20mm from the surface in wet and temperate climates (Sharp,
1996).
Chloride diffusion coefficient is the key parameter used to measure the chloride ingress
into concrete under saturated condition. There are two widely adopted diffusion
coefficients, namely, the effective diffusion coefficient, De and the apparent diffusion
coefficient, D0 • De represents the net chloride flow through the concrete without
chemical reactions taking place (Jones et al, 1996). It depends solely on the
concentration gradient and the concrete porosity such as tortuosity and is associated
Chapter 2 : Literature Review 30
with the measurement of the water-soluble chlorides or the free chlorides (Jones et al,
1996). On the other hand, Da is normally associated with the measurement of the total
or acid-soluble chlorides. It accounts for the chemical reactions with the concrete
(Arsenault et al, 1995; Hooton and McGrath, 1995) such as the chloride binding effect,
which are discussed in section 2.5.3. Though there is no direct relationship between the
acid and water-soluble chlorides, it was reported that the water-soluble chlorides may
range between 70% - 85% of the acid-soluble chlorides (West and Hime, 1985; Swamy
et al, 1994).
It was recognised that the total chloride gradient was not the actual driving force and
thus measurement of the free chloride may yield a more realistic diffusion coefficient
(Johansen et al, 1995; Hooton and McGrath, 1995). With this believe, some researchers
were of the view that diffusion coefficient should be based on the free chloride ions only
(Johansen et al, 1995). However, studies have also indicated that weakly bound
chlorides may unbind when pH of the pore solution decreases (Dhir et al, 1996; Glass
and Buenfeld, 1997). Thus, weakly bound chlorides may also present a significant
corrosion risk. The ability of the bound chlorides to unbind cast some doubts on the
believe that only free chloride contributes to the chloride concentration gradient. Recent
research suggested that chloride binding is, in itself, a diffusion process into the CSH.
Little is known about the possible changes in state between the bound and free chloride
ions.
The determination of the free chloride diffusion is difficult but estimates can be made
from the acid-soluble chloride diffusion coefficient. This apparent diffusion coefficient
appeared to be simpler to measure. Since the complexities of chloride binding are not
fully understood, many design methods adopted this apparent diffusion coefficient,
which is obtained by curve fitting and is dependent on the time at which the profile is
measured. The time dependency of the diffusion coefficient should be taken into
account in the differential equation for diffusion as it can lead to errors of over an order
of magnitude (Concrete Society, 1996).
(the steady state diffusivity, Ds) and the chemical resistance of the concrete. Maage et
al (1996) defines two different types of chloride diffusivity that can be determined from
the chloride profile of drilled cores from the existing marine structures. These diffusion
coefficients are called achieved chloride diffusion coefficient, Dae and the potential
chloride diffusion coefficient, Dp. Dae is dependent on both the material properties (e.g.
maturity) and environmental effects, and is determined from core samples taken close to
the concrete surface. On the other hand, Dp is dependent only on the material
properties and is determined from core samples taken from the virgin concrete from the
inner part of the structure, i.e., not affected by environment.
A steady state diffusion is said to be reached when the concentration of the chloride at
any point in the concrete does not change with time, i.e., ot
ac = 0 (Shewmon, 1981 ).
even after prolonged exposure. A constant Da would only be applicable for a material
with either no or linear chloride binding capacity (Dhir et al, 1998).
Chlorides, whether added during mixing of concrete or are transported into concrete
from external sources, can be bound to the hydrated products in the concrete or exist as
free chlorides in the pore solution. Bound chlorides in the concrete can either be
chemically combined with C3A to form Friedel's salt (calcium chloroaluminate hydrate)
or physically adsorbed on to the cement gel (Tang and Nilsson, 1993; Dhir et al, 1996).
The ratio of the bound chlorides (Cb) to the free chlorides (C1) is called the chloride
binding capacity (Arsenault et al, 1995) that is;
ac
acb = chloride binding capacity
f
Chapter 2 : Literature Review 32
Bound chlorides which are chemically combined with C3A are irreversible while
chlorides that are adsorbed on to the cement gel may unbind when the free chloride
concentration decreases (Tang and Nilsson, 1993). This binding reactions of the
chloride ions with the hydrates slow down the advancement of the diffusion front
(McGrath and Hooton, 1997).
The ability of the concrete to bind chlorides is dependent on the type and quantity of the
binder, water-binder ratio, curing time and temperature. Chloride binding is also
affected by the source of chlorides whether present internally during mixing or
transported externally into the concrete. In the latter case, the degree of hydration and
the rate of chloride ingress into concrete are likely to have a critical influence on the
binding capacity (Mangat and Molloy, 1995). Well cured concrete may result in an
increased amount of cement hydrates which would then lead to an increase in the ability
of the concrete to bind more chlorides (Bamforth and Price, 1993). Arya et al (1990)
observed that externally driven chlorides into concrete showed a similar, though less
pronounced, chloride binding trend than chlorides present in the concrete during mixing.
The source of chlorides (internal or external) can affect both the pH and the degree of
binding and it has been suggested that a higher chloride content can be tolerated when it
is added to the concrete mix. However, it should be noted that a high chloride
concentration at the steel-concrete interface prior to the formation of the cement rich
layer may also have adverse effects on passive film formation (Lambert et al, 1991 ).
With the use of supplementary cementitious materials in concrete, the role of various
binder types (eg. pfa, slag and silica fume) on the chloride binding capacity of concrete
is still not fully understood. One of the reasons is perhaps due to the difficulty of
determining the bound chlorides in the pore solution of the concrete. Tang and Nilsson
(1993) has presented a new approach to evaluate the chloride binding capacity in
concrete.
binding ability. In such a situation, immobilised chlorides are set free which then
contribute to the chloride migration in concrete (Schiessl, 1988).
Very often, the maximum chloride concentration is at some distance away from the
exposed surface of the concrete. This could be attributed to;
Depending on the concrete grade and exposure time, the depth of chloride penetration
into concrete is usually limited to several millimetres or a few centimetres only. For
each sampling location, chloride samples are taken from concrete at different depths
from the surface. Several sampling locations may be done. The powdered samples,
from the various sampling locations and at the same depths, may be combined to give a
test sample. Because concrete is a heterogeneous material, a relatively large test sample
may be necessary to give a representative chloride profile of a given section of the
structure.
Chapter 2 : Literature Review 34
Chloride profiling can be obtained either from thin slices cut from drilled cores or from
drilled powder collected by dry drilling using rotary drill bit. Dry drilling has to be done
slowly because the heat generated may release some of the bound chlorides
(Mackechnie, 1995). The advantage of dry drilling to drilled core is that the risk of
changing chloride concentration in the process of sampling is minimised. However, the
disadvantage of dry drilling is the very small amount of powdered sample available for
analysis. Table 2.2 gives the recommended number of drill holes with respect to the
maximum aggregate size in the concrete and the diameter of drilled hole (Kropp and
Hilsdorf, 1995).
The types of chloride tests to be carried out are dependent on the objective of the
chloride analysis. In most cases, acid-soluble chlorides are carried out for the
determination of chloride diffusion coefficients (Thomas et al, 1990; Bamforth, 1993;
Roy et al, 1993; Maage et al, 1996). Where chloride binding capacity is to be
determined, a distinction between the bound and free chlorides become essential. Free
chlorides in the concrete can be determined from the pore water expression or extraction
with water or ethanol. However, much uncertainty still remains with these methods of
free chloride determination if and to what extent bound chlorides are also extracted
Chapter 2 : Literature Review 35
(Tuutti, 1983; Page and Vennesland, 1983; Glass and Buenfeld, 1997). Figure 2.9 gives
an overview of the general methods used for chloride analysis of concrete.
I Chloride analysis
I
I
I On site I I Laboratory I
I I I
Qualitative
method
Semi-quantitative
method
I Quantitative method I
I
Indicator Types of chlorides -Potentiometric
test titration,
Total chlorides Free chlorides -Photometric
EDAX
-Dissolution in -Pore water
Quantab hot nitric acid, expression,
-Dissolution in -Extraction
deionised water with cold
and boiled water,
-Extraction
with ethanol
Using Mohr titration method, the amount of chloride content expressed in % wt. of
concrete can be determined using the following formula (Berman, 1972);
3.5453VN
chloride content(% wt. of concrete) = (2.1)
w
where V: volume of silver nitrate solution added to the end point (ml)
N : normality of silver nitrate solution
W: weight of the sample (g)
To express the chloride content in % wt. of binder, equation (2.1) shall be multiplied by
a factor and is, thus, given as (Mangat and Molloy, 1995);
Chapter 2 : Literature Review 36
Clearly, these assumptions do not reflect the intrinsic concrete properties and the nature
in which chloride penetrates in concrete. Nevertheless, Fick's Second Law still
provides the only simple and practical model for chloride transport in concrete.
For one-dimensional chloride diffusion into concrete, Fick's Second Law can be written
as (Crank, 197 5);
(2.3)
Applying the following boundary conditions and assuming that De does not vary with
distance, x, from the exposed surface;
Chapter 2 : Literature Review 37
Boundary conditions:
• C (X, t= 0) =0 ; 0 < X < 00
(2.4)
where,
Cx : chloride concentration at depth, x from the exposed surface
C : surface chloride concentration
t : exposure time in seawater
De: chloride diffusion coefficient
erf : error function
The chloride threshold level refers to the chloride concentration that can be tolerated at
the surface of the steel before corrosion can be initiated. This threshold level is not a
constant value for any type of concrete or exposure condition but depends on many
factors such as concrete pH, chloride binding capacity, the state of moisture and oxygen
at the cathodic area of the steel, water-binder ratio, and steel-concrete interface bond.
Therefore, it is not surprising that the literature on the chloride threshold levels are very
conflicting.
Chapter 2 : Literature Review 38
Chloride Threshold
Researcher Level Remarks
(% wt cement)
Many researchers have attempted to evaluate the threshold level for corrosion of steel in
concrete. Threshold levels have been reported in the range between 0.17 and 2.5%
chloride by weight of binder (Glass and Buenfeld, 1995). Some of the published results
are shown in Table 2.3 (Pettersson, 1993). Due to the lack of clear data on how the
various factors affect the threshold level, the variation in threshold level is often
ignored. Instead a fixed value is normally assumed, typically in the range of 0.2 - 0.4%
weight of cement for normal strength concrete (Concrete Society, 1996). Threshold
level is currently expressed in terms of total chloride content (Concrete Society, 1995;
Glass and Buenfeld, 1997).
Glass and Buenfeld ( 1997) suggested that the total chloride representation reflects the
active aggressive ions more accurately than either a cr:OH ratio or a free chloride
Chapter 2 : Literature Review 39
content. They pointed out that weakly bound chlorides in the vicinity of the steel may
also participate in the corrosion process.
Cracks do not only lead to a large spread in chloride profiles, they also cause a large
spread in chloride threshold levels, sometimes more than an order of magnitude. As a
result of the large spread in both the chloride penetration rates and in the chloride
threshold levels for cracked concrete, the initiation time of a concrete structure cannot
be modelled with any accuracy (Pettersson and Sandberg, 1997). For high performance
concrete, the chloride transport rates in uncracked field exposed concrete specimens
have been found to be very low that an initiation time of several hundreds of years can
be expected (Fidjestol and Tuutti, 1995). Long term studies of corrosion rates and
accumulated corrosion damage on reinforcement in cracked high performance concrete
are yet very few. However, for a given concrete quality, available results do indicate a
relationship between corrosion rate and exposure condition, cover thickness and crack
width (Pettersson and Sandberg, 1997).
Cracks are inherently present in concrete. Cracks in concrete are induced by loads and
interaction with the varying environments. Cracks can also developed due to bleeding,
shrinkage and thermal gradients, freeze - thaw and alkali aggregate reaction (Mehta,
1997). In extreme cases, cracks may affect the structural integrity of the concrete
member. However, in most instances, cracks do not affect the load carrying capacity of
the concrete structure, but may adversely affect its durability by providing easy access to
aggressive agents (Mindess and Young, 1981) especially chloride ions in sea water
environments.
Cracks in concrete may exist in the form of macro and/or micro cracks. Macrocracks
are those cracks greater than 0.1mm wide (Pettersson and Sandberg, 1997) and which
are visible to naked eye. On the other hand, microcracks are those less than 0.01mm
wide (Rostam, 1996) which are visible under microscope. The types and behaviour of
microcracks under loads are discussed in detail in section 2. 7.
Surface cracks wider than 0.3mm seldom heal (Mehta, 1997). They tend to enlarge due
to leaching and stress effect, thus providing convenient entry points for aggressive
substances. Shrinkage cracks are often limited through the use of a larger quantity of
steel reinforcement as permitted in most Codes. Mehta (1997) argued that this simply
transforms the wider cracks into many finer cracks and microcracks. These microcracks
may not adversely affect the concrete durability when they are limited in quantity, size
and are discontinuous. However, they have the potential of enlarging and become
interconnected due to applied stress or due to leaching of cement paste (Mehta, 1994).
Thus, affecting the long-term durability of concrete structure. Some of the factors
affecting microcracking of concrete are water-cement ratio, aggregate type and size, bar
size, concrete cover and fatigue of concrete (Mehta and Gerwick, 1982).
The influence of macro and micro cracks on the ability of the concrete cover to provide
a robust protective 'skin' for the steel reinforcement against harmful substances remain
the subject of current research interest.
Chapter 2 : Literature Review 42
Chloride penetration into normal weight reinforced concrete structures situated along
the North Sea coast of Netherlands were inspected by Wiebenga (1980). It was found
that in almost all cases the chloride ions penetrated deeper into concrete near cracks.
Poston et al (1987) found that prestressing had an effect on the chloride ingress near
cracked regions. For the same crack width opening of about 0.38mm (0.015 in), they
observed a much lower average water-soluble chloride content at the crack location of
the prestressed specimens than that of the non-prestressed specimens.
Cracks may undergo self-healing. This effect is more pronounced in sea water exposure
when magnesium and carbonates (from sea water) combine with calcium hydroxide to
form insoluble products of magnesium hydroxide (brucite) and calcium carbonate
(aragonite) and subsequently fill the cracks. However, studies (Raharinaivo et al, 1986)
Chapter 2 : Literature Review 43
and experience (Mehta, 1997) showed that crack widths wider than 0.3mm seldom heal
especially in aggressive environment and subject to a number of cyclic reversing loads.
The question as to whether crack healing may be considered as a phenomenon of
appreciable importance for the durability of concrete elements cannot as yet be
answered and the possibilities of crack healing are not accounted for in the design of
concrete elements.
The effect of microcracking on the chloride transport in concrete has been studied by
some researchers quite recently, for example, Samaha and Hover (1992) and Saito and
Ishimori (1995). In the previous studies, microcracks were generated by subjecting
concrete cylinder under uniaxial compression test. The concrete cylinder was loaded to
different percentages of the 28-day compressive strength and then unloaded
instantaneously. Rapid Chloride Permeability test (ASTM C1202) was then conducted
on a specimen which was cut from the cylinder at the mid-height. Results showed that
chloride permeability was not significantly affected even when the concrete has been
loaded up to 90% of its peak strength. Wang et al (1997) and Loo (1992) pointed out
that the characteristics of microcracks after unloading are likely to be different from
those while under load.
Francois and Maso (1988) compared the chloride concentration profile in the tension
and compression zones of beams in flexure. He found that chloride diffusivity in the
tension zone was more pronounced than in the compression zone. He attributed this to
the increase in porosity and damage at the aggregate-paste interface of the tension zone.
It should be noted that, in his study, tensile macrocracks were also formed under the
flexural loading. Chloride profiling was determined from core samples drilled
horizontally from the side face of the beam at the tension and compression zones. Since
drilled cores in the tension zones were taken between macrocracks, it was not clear if
the chloride diffusivity obtained may be affected by the combined effects of
macrocracks and microcracks.
Microcracks are inherently present in concrete even before the application of load (Hsu
et al, 1963). They are present initially at the aggregate-mortar interface due to several
reasons such as the bleeding characteristics, strength of the transition zone and curing
history of concrete (Mehta and Monteiro, 1993). These microcracks do not pose any
undesirable effect on concrete durability when they are discontinuous. However, they
have the potential of becoming continuous and may grow under the application of load
as discussed in the following sections. Microcracks are usually quantified in terms of
Chapter 2 : Literature Review 45
crack length (Ngab et al, 1981; Najjar and Hover, 1989), crack area (Krishnaswamy,
1968) or specific crack area (Loo, 1992).
Basically, microcracks in concrete can be classified into three types according to their
locations (Hsu et al, 1963). They are;
The existence of bond cracks in concrete even before the application of load was first
observed by the researchers at Cornell University (Hsu et al, 1963; Shah and Slate,
1968). This was done by direct observation of microcracks either by examining a cross-
section of a concrete specimen stained with red dye, under a microscope or by taking X-
ray photograph of a thin slice cut from a concrete cylinder. Hsu et al (1963) has shown
by a mathematical analysis that when concrete shrinks, aggregates are partly debonded
along the interfaces, while some cracks are formed perpendicular to the surface of the
aggregate.
Many researchers have implied that the shape of the stress-strain curve is related to the
microcracking in the concrete (L'Hermite, 1954; Jones, 1960). According to Hsu et al
(1963), bond cracks, which are initially present in the concrete before the application of
load, do not increase in length, width and number before reaching a certain load level.
This load level was observed by Hsu et al (1963) to be about 30% of the concrete
Chapter 2 : Literature Review 46
compressive strength, corresponding to the end of the linear portion of the stress-strain
curve. Therefore, it has been generally implied that concrete behaves elastically up to a
certain stress level, beyond which microcracks begin to propagate. This stress level,
called the initiation stress, marks the onset of microcrack propagation in concrete and
hence the start of the non-linearity of the stress-strain relationship. Values of the
initiation stress ranging from 20-50% of the compressive strength have been reported
for normal strength concrete (Loo, 1992).
Another important physical phenomenon that has been observed to be closely associated
with the initiation stress is the increase in the apparent Poisson' s ratio (Loo, 1992).
The apparent Poisson's ratio refers to any ratio of total transverse strain to total axial
strain. Whereas, the particular value of Poisson's ratio approximately constant within
the lower stress range is called the elastic Poisson's ratio. L'Hermite (1959) has
observed that the onset of microcracking as detected by acoustic emission corresponds
to a marked drop in ultrasonic velocity. In both instances, the load at which the
microcracks begin to propagate also occur at which the Poisson' s ratio starts to increase.
0.9
-
.9
f!
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0-----------.-----
0 0.19 0.38 0.57
Poisson's ratio
Similar observations were also reported by Robinson (1968) and Shah and Chandra
(1968). A typical plot showing the variation of apparent Poisson's ratio with stress-
strength ratio is presented in Figure 2.10.
Bond cracks already exist in the concrete long before the application of load. Bond
cracks developed due to differential shrinkage between the aggregate and the mortar
matrix. This is caused by the incompatibility in their elastic moduli. Studies by Kaplan
(1961), and Kotsovos and Newman (1981) showed that cracks will propagate in a plane
parallel to the direction of compression as this is the only direction that has no
compression component normal to it. However, a penetrating crack may be stopped,
deviated or discontinued due to the presence of large aggregates crossing its path. Such
phenomenon was also observed by McGreath (1969).
According to Mehta and Monteiro (1993), the relationship between the stress level and
microcracking in concrete can be divided into four stages as shown in Figure 2.11.
Stress 100
(% of
ultimate)
50
30
Stage 1
o._______________
Strain
Stage 1 : Below 30% of the ultimate load, bond cracks remain stable, that is, it
does not increase in size, length and width. Therefore, the stress-strain relationship
remains linear which implies that concrete behaves elastically.
Stage 2: Between 30-50% of the ultimate load, bond cracks begin to propagate
with applied stress and the stress-strain curve begins to deviate from its linearity. The
mortar cracking at this stage is negligible. Microcrack propagation is assumed to be
stable because crack lengths rapidly reach its final values if the stress is held constant.
Stage 3: Between 50-75% of the ultimate load, bond cracks begin to grow again
while mortar cracks begin to develop. The rate of microcrack propagation will increase
and becomes unstable at about 75% of the ultimate load, marking the onset of unstable
crack propagation. This unstable crack propagation is called the critical stress which
corresponds to the maximum value of the volumetric strain (Ev = E1 + E2 + E3). In other
words, it is the stress level at which the volume of concrete under increasing
compressive load begins to expand rather than continue to contract. This is illustrated
in Figure 2.12.
··._·: ·.··:·
·. . :· .- ·
~
., .. .
, 1.0 .
Lateral .· · .
Strain --.. .· ·
· E.u
There are many methods used for microcrack detection and observation, amongst
others, X-radiography, acoustic emission, ultrasonic pulse velocity and holographic
interferometry. Recently, Loo (1992) developed a simple and non-destructive method
which provides a quantitative indication of the extent of microcracking in a prismatic
concrete specimen under uniaxial compression. Loo (1992) assumed that the change in
cross-sectional area of a prismatic concrete specimen under uniaxial compression can be
resolved into two components, i.e., the elastic change in cross-sectional area due to
Poisson' s ratio effects and the dilation due to microcracking. Thus,
(2.5)
sectional area due to Poisson's ratio effects and M c = change in cross-sectional area
due to microcracking. Based on this assumption, he has shown that the 5pecific crack
Chapter 2 : Literature Review 50
area (ecr), defined as the increase in the area of microcrack per unit cross-sectional area
(2.6)
where ex and ey are the transverse and axial strains respectively and µe the elastic
Poisson ratio. Ecr has the same unit as the ex and ey which are usually expressed in
microstrain. For strain measurements, usually four electrical strain gauges are required
per specimen, two in each axial and transverse directions. The elastic Poisson's ratio
can be estimated from the expression;
(2.7)
where gx = gradient of the linear portion in the stress-transverse strain curve, and
gy = gradient of the linear portion in the stress-axial strain curve.
0.9
0 0.8
;:
...
ftl
0.7
-...
.c
Cl
C
0.6
-
GI
Cl)
I
Cl)
Cl)
0.5
0.4
Initiation Stress
-...
GI
en
0.3
0.2
0.1
0
0 500 1000 1500 2000
The method does not provide specific infonnation on the individual crack, for example,
the length or area of bond cracks. However, it offers continuous information on the
microcracking behaviour which the conventional slicing methods cannot provide.
Using this method, the initiation stress can also be determined from the plot of stress
against specific crack area. The stress corresponding to the increase in the specific
crack area is the initiation stress and is illustrated in Figure 2.13.
A Sealed 32.6 31 48 -
50- 70%
B 30.8 33 - 78 of le.
A Unsealed 32.6 31 51 -
50- 70%
B 30.8 33 - 81 of fc'
When comparing the crack length between the sealed and the unsealed specimens, only
a marginal increase was observed. For Series A, the increase in crack length of the
unsealed to the sealed specimen is about 6% after 30 days of sustained period.
Similarly, for Series B, an increase of about 4% can be deduced after 42 days of
sustained period.
Ngab et al (1981) carried out a similar studies using normal and high strength concrete.
Sealed and unsealed specimens were used to investigate separately the effects of
shrinkage and creep on the microcracking of the concrete under sustained stress. The
sustained stress levels considered were 45% and 65% of the ultimate compressive
strength. The duration of the sustained loading was 60 days. The compressive strength
of the normal strength concrete was in the range of 33-40 MPa. He reported virtually no
increase in cracking up to 0.45 fc. and a marginal increase was observed at 0.65 fc. for
sealed specimens.
Loo (1995) observed that when the sustained load was below 50% of the compressive
strength, microcrack propagation was negligible irrespective of the initiation stress.
Chapter 2 : Literature Review 53
When the sustained load was from 50% to between 80-90% of the compressive strength,
microcrack propagation generally increased with increasing stress but at a decreasing
rate with sustained time. Beyond 80-90% of the compressive strength, the rate of
microcracking increased rapidly with time. Table 2.6 summarises his observations.
Sustained Stress
Effect of Creep on Microcrack Propagation
Level
0.5 fc' to increases with stress but at a reducing rate with time during
(0.8-0.9 fc') sustained loading.
He also observed a partial closure of cracks both when unloaded immediately from a
peak sustained stress of about 0.83 fc. and after 5 days of creep recovery.
Concrete exhibits creep strain with time under sustained loading and the magnitude of
creep is influenced by several intrinsic and extrinsic factors (Neville, 1970). It is also
known that on unloading a portion of the creep strain is recovered with time. The
remaining irrecoverable creep strain is believed to be partly due to strain associated with
an irreversible process of cracking which occurred under the sustained load. At stress
levels below 30 to 50% of the ultimate strength, microcracking plays a minor role in the
creep of concrete (Neville, 1970).
Under sustained loading, the mechanism by which microcracks develop and propagate
may be explained by a consideration of the internal strain distribution in concrete.
Studies of the micro-mechanical behaviour of concrete using reflective photoelasticity
by Swamy and Sri Ravindrarajah (1982) revealed that under sustained loading there
exists a time-dependent load transfer process from the relatively soft matrix to the rigid
aggregate particles through the interfacial bond. During this process, the bond areas are
subjected to increases in stresses, resulting in a slow growth of the existing bond cracks
and the formation of new bond cracks.
Shah and Chandra ( 1970) considered that crack growth under sustained stress is due to
the phenomenon of stress corrosion which is influenced by the presence of free
moisture. The energy demand for crack growth is reduced because of the reduction in
the surface tension when water is adsorbed to the free surfaces of the solids.
When reinforced concrete was first industrialised some 100 years ago, the pioneers were
convinced to have found a material that can provide a long lasting protection to steel
reinforcement. In spite of the good performance of concrete structures observed in
aggressive environments, there are also evidence of similar structures presenting serious
premature deterioration during its service life. In extreme cases, it may even lead to
structural failure (Hognestad, 1986). One type of deterioration that is widespread is the
corrosion of reinforcement in concrete structures. In the United States, it was estimated
Chapter 2 : Literature Review 55
that the general costs of corrosion deterioration may reach some $126 billion per year
(Jones, 1992).
The concept of service life prediction is, therefore, becoming an area of increasing
interest. It was first proposed in the sixties, followed by the North American initiatives,
which led to the publication of the ASTM E632-81. These proposals were mainly for
building materials and not for marine structures. In addition, it was more of a
philosophical in nature, dealing with definitions and methodologies but does not give
any method of quantifying service life prediction. The JCI ( 1991) publication on the
durability design for reinforced concrete structure was regarded as pioneering methods
for durability design in Japan. However in this document, the concept of durability
design has not been standardised yet. Like ASTM E632-81, BS7543 (1992) was for the
service life prediction of building products and components in United Kingdom. Again,
this document does not give any methods of quantifying service life.
Perhaps the most comprehensive report on durability based design was presented in
RILEM Report TC130 (1995). This report is more relevant to maritime structures. The
report outlines the concept of durability, methods of durability design, stochastic
approach, concept of safety factors and durability models. The concept of structural
durability design was also proposed as shown in Figure 2.14. A recent UK Concrete
Society publication, CS 109 (1996) gives an overview of the current development
towards a performance based design. It also identifies the key areas where further
research is essential.
Chapter 2 : Literature Review 56
ORDINARY STRUCTURAL
DURABILITY DESIGN
DESIGN
Dimensioning of the structure by • Determination of target service life
ordinary design methods and design service life
• dynamic mechanisms
f --. •
Selection of durability models for
degradation mechanisms
Results: • Determination of durability
•
I
., ,. I
I
parameters
According to Clifton (1993), the methods applicable to estimate the service life of
concrete subjected to deterioration processes include;
The two-phase model shown in Figure 2.15, proposed by Tuutti in 1982, is commonly
used for the service life prediction of structures in a marine environment.
Degree of
corrosion
................... :'.\~~~P.~.~~~~.~.i.~}!........................... .
Chloride transport
through concrete
Service Life
1+• '1
The service life consists of the initiation period and the propagation period. For marine
environments, the initiation period is defined as the time taken for the chlorides to reach
the steel reinforcement and to initiate corrosion. The propagation period is the time
Chapter 2 : Literature Review 58
The relative lengths of the initiation and propagation periods, depends on the exposure
conditions, the cover and the characteristics or quality of the concrete (CS 109, 1996).
In some severe exposure conditions, the propagation period could be very short in
relation to its service life. Clifton (1993) cited that the initiation period is over 5 times
longer than the propagation period for a bridge deck. Bamforth (1993) suggested
neglecting the propagation period, to be conservative. Yokozeki et al (1997) indicated
that the initiation time is about 2.3 - 4.0 times the propagation time at the tidal and
splash zones. Lin and Jou (1991) cited propagation time to be relatively short and less
predictable. Christensen (1997) suggested the initiation time to be more rational as
service life from a life cycle viewpoint, since repair of corroded rebars is a major
contributor to life cycle cost. Andrade and Alonso (1996) suggested that no propagation
period should be considered in the case of localised attack due to chlorides.
In addition, when steel reinforcement in the concrete are corroding, the rust produced
may diffuse through the pore network producing brown stains on the concrete surface.
This often happens in wet concrete. In this case, although steel corrosion may be high
but no visible cracks can be observed on the concrete surface (Andrade et al, 1990).
Many researchers would consider the initiation period as the service life of maritime
Chapter 2 : Literature Review 59
structures (Maage et al, 1996; Collins and Grace, 1997). Further more, the propagation
period has received less attention probably due to the scarcity of data available in the
literature on the deterioration rates. The determination of the acceptable level of
corrosion to mark the end of propagation period has been described more
philosophically than quantitatively (Andrade et al, 1990).
The rate of chloride iori penetrating into the concrete cover and subsequently initiating
steel corrosion is the primary factor for determining the initiation period. There are
several transport mechanisms for chloride ions (see section 2.5.2) into concrete.
However, in most practical instances, chloride diffusion is assumed to be the primary
transport mechanism into concrete. The reasons for this are discussed in section 2.5.2.3.
For one-dimensional chloride diffusion into concrete, Fick's Second Law is given as
equation (2.3) and the solution as equation (2.4), both already described in section
2.5.4.3.
Chapter 2 : Literature Review 60
Matsushima et al ( 1998) reported that the one-dimensional diffusion model can be used
to predict the measured chloride profile from field structure with good agreement. It
must be recognised that the assumptions implied in equation (2.3) are (Monica et al,
1996);
a) concrete is homogeneous and isotropic, that is, chloride diffusion coefficient (De)
does not change with depth of concrete,
b) De is not time-dependent, that is, it does not change during the exposure period,
c) the surface chloride concentration, Cs, is constant during the exposure period, and
d) that the medium is non-reactive and non-adsorptive, that is, chlorides do not react
with the concrete.
The surface chloride concentration depends on several factors, among them are the
salinity of the sea water, environmental conditions, exposure type and binder type. It is
usually expressed as the acid-soluble chloride content by weight of binder or concrete.
Results obtained by Bamforth (1996, 1999) from exposure sites in UK indicated Cs can
be reasonably assumed constant after 6 months of exposure. He also found higher Cs in
concrete containing blended cements (OPC with either 30% pfa or 70% slag). He
attributed the higher Cs in blended cement concrete to the improved binding
characteristics. The accumulation of surface chloride was observed to be most severe in
the splash zone. Cs may be influenced by the microclimatic conditions of the structure
and the concrete material properties. The microclimatic conditions are, orientation of
the structural element, direction of prevailing wind, and degree of exposure to rainfall
(Bamforth, 1996). The material properties are, cement content and type, curing regime,
and soptivity when concrete is not saturated (Bamforth, 1994).
Mangat and Molloy (1994) observed that Cs falls within 1% to 2% by weight of binder
for his concrete specimens which were immersed in chloride solution up to 10 months.
He also noted that Cs remained within the same range for blended cement concrete
Chapter 2 : Literature Review 61
specimens exposed to the tidal zone of the North Sea up to 3 years. Takewaka and
Matsumoto (1988) found Cs to be constant at the submerged zone but varies at the
splash and atmospheric zones. Similar observations were also reported by Costa and
Appleton (1999). In the atmospheric zone, the deposits of chlorides occur continuously
with time. In the spray and tidal zones, concrete is subjected to wetting and drying
cycles with salt water. Both situations can lead to the accumulation of chlorides at the
surface which may also give rise to concentration peaks near the surface (usually a few
millimetres below the exposed surface) of the concrete.
> 50MPa
22 kg/m3 (0.92% wt. of concrete)
Costa and Appleton ( 1999) expressed the surface chloride concentration as a power
function with time. Uji et al ( 1990) assumed Cs to be proportional to the square root of
time in service. Table 2.7 presents a summary of the surface chloride concentration
adopted and proposed by various researchers.
Chapter 2 : Literature Review 62
It is common to assume a chloride threshold level of 0.4% wt. of cement for activation
of steel corrosion even though higher values have been reported in practice. In some
instances, this threshold level is also referred to as critical chloride concentration. A
comprehensive review of this is presented in section 2.5.5. Table 2.8 presents some of
the threshold levels that are usually adopted in the prediction of service life of maritime
structures.
As discussed in section 2.8.2, the assumptions made in deriving the solution to Fick's
Second Law, i.e., equation (2.4) may not be realistic for ageing material like concrete.
For a given exposure condition, the chloride diffusion coefficient, De = f ( x, t, C ).
That is, De is dependent on the distance (x) from the exposed surface, the time (t), and
the chloride concentration (C) of the concrete. The effects of distance and chloride
concentration are secondary compared to the influence of time on the chloride ingress
into concrete (Poulsen, 1995; Kanaya et al, 1998). If concrete is sufficiently cured, the
dependency of the diffusion coefficient on distance can be neglected (Poulsen, 1990).
For most practical applications, chloride ingress into concrete exposed to sea water
environments are assumed to be only time-dependent (Gerhard and Walter, 1987;
CS 109, 1996).
k : constant
Chapter 2 : Literature Review 64
D,(t)=D,,.('; J (2.10)
and
Maage et al (1996) proposed the following algorithms for the service life prediction
based on planned impection (Figure 2.16) and inspection once during service life
(Figure 2.17).
Chapter 2 : Literature Review 65
Diffusion Conditions:
coefficient to : exposure starts
(log-scale) le : time of inspection
lLT : service life time
Age (log-scale)
l
l =! 0• Concrete surface inspected by cover meter: c
Jr
t = le. Drilled cores are tested according to APM 207:
Dac,c, Csa, Cj
Jr
ln(Dac,c / D po)
a=
ln(t I tc}
0
Jr
Is a< I ?
NO YES
~r
••
Self- Blocking 1; = 2erfc- 1( Cer
csa
-CJ
_
cj
j
~•
( ]-
2
I-IX
C
tLT =t
0
l;~t 0 Dpo
Figure 2.16: Flow Diagram for Service Life Calculation on the Basis of
Measurements at Times t0 and tc (Planned Impection)
(Maage et al, 1996)
Chapter 2 : Literature Review 66
Diffusion Conditions:
coefficient t0 : exposure starts
(log-scale) le : time of inspection
tir : service life time
Age (log-scale)
tir
l
t = le, Drilled cores tested according to APM 302: Dpc
l = le, •
Drilled cores are tested according to APM 207:
Dac,e , Csa, C;
J,.
ln(Daee f Dpe)
a~ 0.25+ ( · )
ln t I te
0
I
I
•
Is a.< 1 ? I
I
NO YES
~r
'~
,,
2
t LT = te ( C
~~teDae,e
]- I-ex
Figure 2.17: Flow Diagram for Service Life Calculation on the Basis of
Measurement Once during Lifetime (Inspection Once during Service Life)
(Maage et al, 1996)
Chapter 2 : Literature Review 67
Mangat and Molloy (1994) proposed an empirical relationship for chloride diffusion and
time in the form of ;
D C
= D.t-m
I
(2.12)
Equation (2.12) is very similar to equation (2.9) when to= I second. It shall be noted
that equation (2.12) would imply that concrete is exposed to sea water at time equal 1
second. Mangat and Molloy ( 1994) also attempted to relate m with wlc ratio using some
data from his studies and was given as ;
Mackechnie (1995), in his studies, adopted the same empirical relation for De proposed
by Mangat and Molloy (1994). Collins and Grace (1997) considered the early age
effects of moisture movement in unsaturated concrete in addition to the ionic diffusion
once a concentration has been established when the moisture movement becomes
negligible. They proposed the following expression for chloride diffusion coefficient ;
(2.14)
Equation (2.14) above is an interim model. Since many factors can affect the
parameters in that model, Collins and Grace (1997) suggested that much work is needed
before a complete understanding of the processes can be achieved.
Chapter 2 : Literature Review 68
and
(2.16)
Input Data
,,
The effects of cracks on the initiation period may be predicted considering the following
three scenarios (Sandberg, 1995). Cracks are classified as 'dead' and 'dynamic' cracks.
'Dead' cracks do not propagate with time while 'dynamic' cracks grow with time. The
three scenarios are ;
Cracks in submerged concrete are probably harmless so long as they do not propagate.
As long as oxygen is not available to initiate steel corrosion, service life is not likely to
be affected.
The main effect of non-propagating cracks is to reduce the effective cover thickness.
The initiation time may also be predicted from the effective depth cover by subtracting
the measured mean crack depth from the nominal cover thickness as shown in Figure
2.19. The effective cover thickness shall be treated uncracked in evaluating the
initiation time.
These cracks are unstable with time and may propagate in an unpredictable manner. In
this case, the initiation time cannot be predicted without proper knowledge and
prediction of the crack propagation.
Chapter 2 : Literature Review 70
mean
crack
depth effective cover
11111 ~11111 ~I
_________ rebar
.----t- concrete
nominal cover
11111
Chapter 3
CHLORIDE DIFFUSIVITY OF CONCRETE
CRACKED IN FLEXURE
3.1 INTRODUCTION
Cracked concrete allows corrosion process to initiate much faster than uncracked
concrete (Lorentz and French, 1995). The initiation of steel corrosion in cracked
concrete is dependent on the surface crack width (Poston et al, 1987). Wider surface
crack widths have been found to induce corrosion much faster than relatively smaller
ones (Makita et al, 1980; Raharinaivo et al, 1986; Suzuki et al, 1989). Other researchers
(Arya and Ofori-Darko, 1996; Schiess} and Raupach, 1997) reported a slower initiation
of steel corrosion in a cracked concrete when the cover is increased. This is because
corrosion of steel depends on the availability of oxygen, not at the crack, but in the
sound concrete on the cathodic end of the steel and, hence, on the rate at which oxygen
can diffuse through the cover (Beeby, 1978). Thus, it is recognised that both crack
width and cover thickness do affect the initiation of steel corrosion in concrete. From
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 72
the literature review, and comparing only these two variables, the initiation of steel
corrosion can be generally summarised as follows :
i) it occurs faster with increasing surface crack width when the cover is maintained ;
ii) it occurs slower with increasing cover thickness when the surface crack width is
maintained.
For a given stress level in the steel bar and keeping other variables constant, the surface
crack width increases with the cover thickness (Beeby, 1978). In other words, crack
width has a direct relationship with the cover thickness. Hence, the initiation of steel
corrosion in reinforced concrete may be influenced by the combined effect of crack
width and cover thickness, i.e., it is 'directly proportional' to the crack width while at
the same time 'inversely proportional' to the cover thickness. From this point of view,
it is more appropriate to consider the influence of both parameters, crack width (Wer)
and cover thickness (C), simultaneously. The crack width/cover ratio, W c/C can be a
suitable parameter to consider in relation to the durability performance of a cracked
reinforced concrete.
Using published data (Pettersson, 1996), the chloride threshold level when plotted
against the inverse of W c/C (i.e. C/W er) appears to be linear. This is shown in Figures
3.1 (a) and 3.1 (b ). These data were collected from five different high performance
concrete mixes containing silica fume between 5% and 15% replacement. The water-
binder ratio of four of those mixes was 0.3 while the remaining mix was 0.4. The
regression coefficient (R2) of the plot in Figure 3.1 (a) is comparatively low due to a
larger variability in the results. Based on these data on cracked concretes, it appears that
the chloride threshold level can be related to Wc/C by a hyperbolic relationship. This is
shown in Figures 3.2(a) and 3.2(b).
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 73
--
>
CD
-c-
0.3 1R2 =0.761
--
>
CD
-c-
1.4
1.2
I 2
R = 0.851
- C 0.25 - C
0 CD 0 CD
.c E .c E 1
en CD 0.2 en CD
! u.
.c
! u 0.8
;i 0.15 <> =i
.- .g ~ 0.6
:2 # ·.:::!.... 0.4
0
0.1
0
<>
::c
0 0.05
<> ::c 0.2
0
0 0
0 20 40 60 80 0 20 40 60 80
Cover/crack width ratio Cover/crack width ratio
'ii 'ii
--
-
>
CD
-c- C
0.5
0.4
Assumed limit for
uncracked normal
strength concrete
--
~
-c-
-
0 CD
C
2.5
2
Assumed limit for
uncracked concrete
0 CD .c E
.c E
!
CII CD
u 0.3 .
Cl)
CD
=i
CD
u
1.5
=i
.g ~
·.:::~
0
0.2
.-
.g
·-
.2
.c
~
0
::c
0
0.1 0 0.5
<>
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06 0 0.01 0.02 0.03 0.04 0.05 0.06
Many factors affect the threshold level including concrete grade, type and curing regime
(Dhir et al, 1994). Steel corrosion sometimes occurred at a relatively lower chloride
level while sometimes being absent at higher levels (Vassie, 1984). Reinforcement
corrosion has been reported to occur in the splash zone concrete at relatively lower
chloride content than concrete in the submerged zone (Sharp and Pullar-Strecker, 1980).
For uncracked normal strength concrete, the threshold level permitted in many
specifications varies between 0.2% and 0.5% by weight of cement (Glass and Buenfeld,
1997). Sandberg (1995) have reported that the threshold level can be increased as the
water/binder ratio is reduced. In submerged conditions, the threshold level can be as
high as 2.1 % to 2.5% by weight of cement (Pettersson, 1996). In spite of the hyperbolic
trend observed, a cut-off level can, thus, be assumed as WcrfC tends to zero.
The trend observed in Figures 3.2(a) and 3.2(b) implies that the corrosion initiation of
steel in concrete becomes more rapid as the WcrlC ratio is increased. Similar trends are
observed for both the cracked specimens exposed to the atmosphere and the cracked
specimens fully submerged, even though the threshold levels are different for the same
WcrlC . This trend indicates the suitability of W crlC as a parameter for durability
performance of a cracked reinforced concrete. From Figures 3.2(a) and 3.2(b ), it is
observed that the effect of W crlC on the chloride threshold level, and hence the
corrosion initiation, becomes more pronounced as the ratio decreases, i.e., the slope of
the curve becomes steeper as W crlC is reduced. It appears that to enhance the durability
of a cracked reinforced concrete, it is desirable to minimise WcrlC.
A crack width/cover ratio of 0.01 has been reported to represent a transition zone below
which corrosion will be reduced (Vennesland and Gjorv, 1981 ). In a corrosion study,
Schiess! and Raupach (1997) observed no corrosion in one of their specimens (crack
width= 0.1mm, cover= 35mm and w/c = 0.5) after 24 weeks of corrosion test. Using
this information, the corresponding Wcr/C is estimated to be about 0.003. Gerhard and
Walter (1987) carried out an extensive investigation into the effect of cracks on the
chloride induced corrosion of steel in concrete. The range of crack widths considered
was from 0.05 to 0.8mm. A constant cover thickness of 40mm was adopted for their
beams. Chloride concentrations on the crack surface at the level of the steel were
determined after 6, 12, 18, 24 and 48 months immersion in 3% NaCl solution. They
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 75
concluded that the durability of reinforced concrete will not be adversely affected if the
crack widths do not exceed 0.4mm. Based on this information, the corresponding W crlC
is estimated to be 0.01. These provide further evidence on the significance of W crlC as
a parameter for durability performance of a cracked reinforced concrete.
To assess the significance of the W cJC on the chloride threshold level, the slope of the
curves in Figure 3.2 is plotted against the W cJC and is shown in Figure 3.3. The slopes
of Figure 3.2, which represent the change in the chloride threshold level per unit change
in WcrlC, is converted to the change in the chloride threshold level per 1% change in
WcJC in Figure 3.3. A similar trend was observed for both the crack in atmosphere and
the submerged crack. From Figures 3.3(a) and (b), the effect of WcJC on the chloride
threshold level can be categorised as ;
"ii
3.5
-,,i> .:::u
0
0 3:
ii .5
=-i 3
2.5
CD CD E
.,_ c, CD
:SI
C
(V
U
•
2
-O .c
u 't:'.
:s; 1.5
.5 #- #-
CD..- -
D) ...
C CD
(V c. 0.5
.c
0
0 +--..----.-=~==*===<---o--o----4
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Wcr/C
"i
- 3:.
~ ~
't,
0
(,)
Z"
25
.c c C 20
Ill · - CD
CD
._
CD
i,,
E
CD
:$ C u 15
I (II •
-0 .c
(,)
t:'.
:5
.e ~ ~ 10
.,, .
CD.,.. -
C CD 5
1 c.
0
Wcr/C
From the discussion above, the initiation period of a service life of a concrete structure
can be related to the W crlC rather than the crack width or cover thickness, treated as
individual parameters. A survey of the permissible crack width and the minimum
concrete cover given in various Codes for design of reinforced concrete structures in sea
water exposure is shown in Table 3.1. The corresponding WcrlC varies from 0.0025 to
0.0075. Amongst these Codes, and from the viewpoint of WcrlC, ACI Manual (1994)
appears to be more stringent on its durability requirement.
The Australian Standard, AS 3600 (1994) does not give any guidance on the allowable
crack width at serviceability, except for the 'deemed to comply' rules. In AS 3600
(1994), flexural cracking in reinforced concrete beams (at serviceability) shall be
deemed to be controlled when the centre-to-centre spacing of bars near the tension face
of the beam does not exceed 200mm. In addition, the distance from the side or soffit of
a beam to the centre of the nearest longitudinal bar shall not exceed I 00mm. Even if
these 'deemed to comply' rules are satisfied, cracks may develop especially when there
is a tendency now to use higher yield strength steel bars ({y=500 MPa) as reinforcements
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 77
for concrete structures in Australia. Therefore, from the viewpoint of durability, a crack
width limitation in AS 3600 is necessary in addition to the cover thickness, to minimise
the W cJC of a cracked reinforced concrete.
Allowable Minimum
Codes Crack Width, Cover, C We.IC
Wcr (mm) (mm)
CEB/FIP Model Code
(1990) 0.3 40 0.0075
For a given W cJC, it may be of interest to determine the effect of tensile steel quantity
on the chloride diffusivity of concrete in flexure. In the present study, a W cr/C of 0.01 is
chosen while the only variable considered is the quantity of steel bar (same diameter and
type) in the tension zone. A comparison between the chloride diffusion coefficient in
the tension and compression zones of the reinforced concrete prisms is made.
3.3.1 Materials
Normal strength concrete mixes were designed using locally available Nepean crushed
gravel and Nepean sand as coarse and fine aggregates, respectively. Concrete Grade 20
and 40 were used throughout the present study. The coarse and fine aggregate grading
curves comply with AS 2758 and are shown in Figures 3.4 and 3.5, respectively. Figure
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 78
3.6 shows the combined grading used for both concrete mixes. A sand-total aggregate
ratio of 0.45 was adopted.
100
I
- ;
-
,-
90 ;
80
/t,J'
C') 70
//
C 60
/,'
U) f,
U) 50
m
c. 40
/,'
~
0 30 /,' I
I
20
I::;;;~ ~ I
I
10
- --c; ·:,/'
-.
~
0 - -
10 100
Sieve Size (mm)
~ Curn % Passing - a - Upper Unit - -fr-· Lower Unit
100 - ,,.,, - -
,,.,,,,. ,'
90
,,
J/
~
80 I
I
C') 70
i I
C .
-
I
·;
U)
60 ,·
Ii
I
50
m I
c. 40 ./
I
/'
I I
~
-.
I
I,
0 30
20 fa J
10 .,. ~ ....
;
- - -' •
/
0
0.01 0.1 10
100
90
I/
,,,,I
80
V}
70
C')
C 60
It
h
u, 1,
I/~ '/
I
u, 50
(U
D. 40 / [.,' 1-' II
I
~' /
I
I/
~
0 , I./
30 ,- ,
A v -L
/
20
t/ )) /
10 ,
...&, ~
f',;.
v'
l;J
0
0.01 0.1 10 100
A 50-litre capacity pan mixer was used to mix the concrete throughout this study.
Concrete mixes were prepared in accordance with Australian Standard, AS1012 Pt 2.
Both aggregates were pre-soaked in water before batching. Batched aggregates were
then stored in airtight containers to ensure no moisture loss. At the same time, the
moisture contents of the aggregates were determined and adjustments made to the mix
proportions. A water reducing agent was added at a dosage of 400ml/I 00kg of cement
to achieve a required slump of 100± 20mm.
Concrete was placed in two layers in the <I> 100 x 200mm steel moulds. Each layer was
compacted using a vibrating table. After casting, concrete specimens were covered with
plastic sheets. All mixing and casting were carried out in a standard laboratory
condition at 23 ± 2°C and 50 ± 5% RH. The specimens were demoulded the following
day and moist cured in lime saturated water up to 28 days of age.
70
60
-
.c
Cl)
c -
Cl) I'll
.. Q.
50
y = 9.8ox· 2·07
R2 = 0.98
en ==
Cl)-
40
> >-
·-
Ill ,I'll
, 30
1/1 I
Cl)
._
co
N
c._ 20
E ra
0
0 10
0
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Water-Cement Ratio
Trial mixes were designed using the Unit Water Method (Australian Method) by
altering the water-cement ratio to achieve the targeted strength. The water content was
kept constant at 180 kg/m 3 for all the mixes. All mixing, casting and curing were
carried out as described in section 3.3.2. Seven different trial mixes were carried out to
establish the compressive strength to w/c relationship. This is shown in Figure 3.7.
Using a regression analysis, the relationship can be expressed as a power function with a
very good coefficient of determination.
From Figure 3.7, the mix proportions for Grades 20 and 40 concrete were derived. This
is summarised in Table 3.3. These mixes were used throughout this study.
The fc°, given in the last column of Table 3.3, was the 28-day compressive strength
obtained from these mixes. The compressive strength was an average result of three
q, 100 x 200mm concrete cylinders. A hydraulically operated 3000 kN compression
testing machine (Autotest 3000) was used for all the 28-day compressive strength tests.
A constant loading rate of 0.3 MPa/min was adopted.
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 82
To assess the variability of the compressive strength, two batches of Grade 40 concrete
mix were carried out. For each batch, five q>I00 x 200mm cylinders were cast and cured
as described in section 3.3.2. The results are presented in Table 3.4 together with some
results from the literature.
It can be observed that the coefficient of variation calculated from the test results was
the same order of magnitude as that from the literature and was well within the limits
normally anticipated for conventional concrete. To ensure consistency of results, the
same batching and mixing procedures were strictly adhered to for all subsequent
casting.
Reinforced concrete prisms, 90 x 100 x 650mm, were cast using Grade 20 concrete.
The concrete prism was reinforced with one or two mild steel bars, 8mm diameter.
Both ends of the steel bars which projected beyond the concrete were threaded over a
short length so that nuts can be installed. This is to provide anchorage to the bar when
subjected to bending. A uniform concrete cover of 30mm was maintained for all
concrete prisms.
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 83
90mm
I• •I
<j) 12 mm bolt tension
<j)8mm mild steel bars
• •
100
as-cast
face
It shall be noted that the tension and compression zones of the prism, in flexure,
constitute the two vertical sides of the forms during casting. It was intentionally
orientated this way so that an unbiased comparison between the chloride ion diffusion in
the tension and compression zones can be made. For unstressed concrete prisms,
Mangat and Gurusamy ( 1987) found a significant difference in the diffusion rates of
chloride through the as-cast face and bottom face (during casting) of the prism.
However, he reported similar chloride diffusivity through the two vertical faces (during
casting) of the prism. One of the reasons could be the wall effect during casting which
causes more paste or mortar to accumulate at the surface of the concrete specimen
(Andrade et al, 1997).
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 84
All the faces of the prisms were coated with epoxy leaving only the two faces, in tension
and compression (90 x 650mm), uncoated. The paired prisms were immersed in 3%
NaCl solution for 300 days in a controlled room at 23± 2°c and 50± 5% RH.
The positions of sampling for chloride profiling on the prism are shown in Figure 3.9.
At each location, two sample holes were dry drilled using a 20mm diameter rotary
impact drill and at slow speed. Powdered samples from locations, for example, (b)
were combined to give a test sample representing the average chloride diffusivity in the
tension zone 100mm away from the crack. Similarly, powdered samples from (g) were
combined to give a test sample representing the average chloride diffusivity in the
compression zone 100mm away from the midspan. Samples from (a) & (e), and (j) &
(f) were combined to give two test samples, respectively, each representing the
unstressed control. Two test samples are required as a check if there is any significant
difference in their results.
100mm 100mm
r
<tension?>
+
1+-+-+!3om~
~ q>l2 mm bolt
a bi c: id :c ib
j gj h: :i :h jg :f
<corn pression>
The powdered concrete samples obtained were used to extract acid soluble chlorides
(Berman, 1972). Mohr titration was used to determine the chloride concentration in the
solution. The chloride ion transport into concrete is assumed to be one-dimensional in a
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 85
semi-infinite medium complying with Fick's Second Law of pure diffusion (Crank,
1975) given as,
ac = D a c 2
( 3.1 )
at axa 2
( 3.2)
The chloride concentration profiles in the tension and compression zones of the
reinforced concrete prisms in flexure are shown in Figures 3.10 and 3.11. The chloride
concentration profile of the control is also plotted together for comparison. There is no
significant difference between the Da obtained from the two test samples (a) & (e), and
(i) & (f). Thus, the average result of the two test samples was used as a control, to be
more representative of the two sides of the prism. Table 3.5 shows the calculated Da
while the values of the normalised Da (against the control) are given in Table 3.6.
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 86
0.60
C -...
GI
1i
0.50
-.
;0 0.40
ea u
C
C 0
GI u 0.30
go
0 .
03 0.20
0 0
-
..!. ~
0.10
0.00
0 5 10 15 20 25
Distance in mm
- - +- - - C-rridspan - - • - - C-30rrrn - - * --C-1 00rrrn
~ At crack ---0-- T-30rrrn A T-100rrrn
~ - Control
0.6
C -...
GI
0.5
.- 1i
;0
ea u
C
C
0
u
0.4
GI 0.3
go
0 .
03 0.2
..!. ~
0 0
- 0.1
0
0 5 10 15 20 25
Distance in mm
- - +- - - C-rridspan - - • - - C-30rrrn -- * --
C-1 00rrrn
~ At crack ---o--T-30rrrn ---t:r-T-100rrrn
~ -Control
No. of Normalised Da
Bar Compression Zone Tension Zone
M/span C30 Cl00 Crack T30 Tl00
Tables 3.5 and 3.6 clearly show the chloride diffusivity in the tension zone is greater
than that in the compression zone. The chloride diffusivity in the compression zone is
observed to decrease as the steel quantity is increased. Comparing Figures 3.10 and
3.11, a wider difference between the chloride concentration profiles in the tension and
compression zones is observed from the prism containing two bars. This could be
attributed to the increase in stress levels in the tension and compression zones of the
concrete.
For the same W cr/C, higher stress levels can be expected to develop in both the
compression and tension zones of the concrete in flexure when the number of tensile
steel bar is doubled. In the compression zone, the increase in flexural stress appears to
be an advantage as the chloride diffusivity is found to decrease. This is evident by
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 88
comparing the D 0 in the compression zone at midspan and at 100mm away. There is a
progressive reduction in the D 0 towards the midspan of the prism in which maximum
stress is expected. The relatively lower D 0 values in the compression zone of the prism
containing 2R8 bars provide further evidence of the compressive stress effect on
chloride diffusivity in concrete. A similar reduction in the chloride diffusivity was also
observed when concrete specimens were subjected to uniaxial compression (Lim et al,
1999). A possible explanation for this would be the reduction in the porosity of the
concrete in compression which impedes chloride diffusion. The rate of chloride
diffusion in the concrete is affected by the characteristics of the porosity namely the
pore volume and pore size. Diffusion is much slower with smaller pore sizes due to the
greater tortuosity of the path an ion has to follow (Kumar et al, 1987). The
compressive stresses in concrete may partially close some microcracks or capillaries that
exist in the direction of diffusion. Lower chloride contents at the crack location of
prestressed specimens have been reported when compared with that of the non-
prestressed specimens (Poston et al, 1987).
At the crack, salt water may fill the crack and diffusion may occur from the cracked
plane. This is evident from the high chloride content measured along the depth of the
crack. The mechanism of chloride transport through a crack in concrete is, thus,
different from the one-dimensional diffusion in a semi-infinite medium assumed.
Therefore, the 'Da' determined at the crack may not have any significant implication in
terms of the assumed diffusion process. Further study is needed to address the chloride
transport mechanism through a crack.
Chapter 3 : Chloride Diffusivity of Concrete Cracked in Flexure 89
For WcrlC = 0.01 considered in this study, when the number of tensile steel bar is
doubled, the decrease in the Da in the compression zone is 14% at midspan while a
marginal increase is observed in the tension zone of a prism in flexure.
3.5 CONCLUSIONS
c) AS 3600 does not give any guidance on the allowable crack width for reinforced
concrete structures at serviceability, except for the 'deemed to comply' rules.
However, cracks may develop even if the rules are satisfied especially when there is
a tendency now to use higher yield strength steel bars ({y=500MPa) as reinforcement
for concrete structures in Australia. From the viewpoint of durability, a crack width
limitation in AS 3600 is necessary in addition to the cover thickness, to minimise
the W cJC of a cracked reinforced concrete.
e) In specimens reinforced with steel bars in flexure, the Da value in the tension zone
is found to be relatively higher than in the compression zone. This may be attributed
to the damage at the aggregate-paste interface in the tension zone which can
expedite diffusion process. In contrast, the compressive stresses in concrete
impedes chloride diffusion which can be attributed to the reduction in the porosity of
the concrete.
Chapter 4 : Microcracking and Chloride Permeability of Concrete... 90
Chapter4
MICROCRACKING AND CHLORIDE PERMEABILITY
OF CONCRETE UNDER UNIAXIAL COMPRESSION
4.1 INTRODUCTION
Microcracks are present in concrete due to several causes such as bleeding, shrinkage,
thennal gradients, freeze-thaw and alkali-aggregate reaction (Mehta, 1994).
Microcracks in concrete can also be induced by external loadings or as a result of the
interaction of the concrete with the environment. Microcracks, which originally exist in
the concrete, may propagate and become interconnected due to an applied stress (Mehta,
1997). These microcracks may fonn potential flow channels which provide easy access
to aggressive ions such as chloride ions. The importance of microcracks on the
transport properties of concrete has been highlighted recently (Rostam, 1996). The
effect of microcracking on the permeability of concrete under uniaxial compression has
been studied by several researchers (Samaha and Hover, 1992; Saito and lshimori,
1995). However, there are some conflicting views pertaining to their findings.
Samaha and Hover(1992) reported that microcracking in concrete at stress levels below
75% of the compressive strength did not affect the mass transport properties of concrete.
The microcracks were quantified in tenns of crack length by examining a concrete slice
cut from a cylinder after the compression test. The observed crack length was then
compared with the measured electrical charge passed using a rapid chloride penneability
test (RCPT) carried out in accordance with ASTM Cl 202. On the other hand, Saito and
lshimori (1995) found that the chloride penneability of concrete that had been subjected
to 90% of the compressive strength to be nearly equal to that of unloaded control
specimens. However, he did not carry out microcrack evaluation on his specimens. He
compared his chloride penneability results with the microcracking behaviour of
Chapter 4 : Microcracking and Chloride Permeability of Concrete ... 91
concrete reported by other researchers (Hsu et al, 1963; Krishnaswamy, 1968; Shah and
Chandra, 1968). In this case, the interpretation of his results is questionable. Ludirdja
et al ( 1989) reported that compressive loads had little effect on the water permeability of
concrete even though there were some indications of significant microcracking in the
concrete. He did not carry out crack measurements but used an ultrasonic pulse velocity
method to detect microcracking in concrete.
In the previous studies (Ludirdja et al, 1989; Samaha and Hover, 1992; Saito and
lshimori, 1995), no attempts have been made to characterise the microcracks during the
compression test. It is evident that the characteristics of microcracks after a concrete
has been unloaded are different from those while under load. Wang et al (1997), for
example, introduced a feedback controlled splitting test to generate and characterise
tensile cracks in a concrete specimen. He found that the crack opening displacement
reduced after the load had been completely removed from the specimen. Loo (1995)
reported similar observations for concrete cylinders subjected to uniaxial compression
tests.
same cylinder. An estimate of the total crack length from the same specimen was also
carried out to supplement the observed microcracking behaviour from the non-
destructive test. Comparisons between the specific crack area, total crack length and the
electrical charge passed through the specimen are made and discussed.
Grade 40 concrete mix proportions given in Table 3.3 of Chapter 3 was used. The
30 No. of
mixing, casting and curing procedures are described in section 3.3.2. A Concrete
The non-destructive method of evaluation developed recently by Loo (1992) was used
to characterise the microcracks during the uniaxial compression test. The method
provides a quantitative indication of the extent of microcracking in a concrete specimen
during the test. The formula was derived on the assumption that the change in cross-
sectional area of a prismatic concrete specimen under uniaxial compression is equal to
the sum of the elastic change in cross-sectional area due to Poisson's ratio effects and
the dilation due to microcracking, hence,
(4.1)
(4.2)
where Ex and Ey are the transverse and axial strains, respectively, and µe is the elastic
Poisson's ratio. The specific crack area has the same unit as the Ex and Ey, and is
expressed in microstrain. For a circular specimen, the specific crack area is equal to
!l.Ac/rcr2 • Four electrical strain gauges are required per specimen, two in each axial and
transverse directions. The two strain gauges in the transverse direction are positioned at
the mid-height of the specimen. Strain gauges with a gauge length of 30mm and a
gauge factor of 2.13 ± 1% were used. Values of ~Le were detennined as described in section
2.7.5.
After 28 days of casting, concrete cylinders were tested under uniaxial compression
within a stress range of 0.3 fc. and 0.95 fc'. Upon reaching the predetermined stress
level, unloading commenced immediately until the cylinder was completely unloaded.
Strain data were recorded during the loading and unloading process. Recording of the
strain data was continued, at the end of the compression test, until the creep recovery
was negligible. A 2000 kN capacity Baldwin closed-loop servo controlled hydraulic
compression machine was used for this test. A ram displacement rate of 0.09mm/min
for loading and at least 2.0mm/min for unloading were adopted. All strain data from the
electrical strain gauges were logged. For each stress level, microcrack evaluation was
carried out on two concrete cylinders. In addition, several companion cylinders were
also loaded to 90% of the compressive strength and the load was maintained for a period
of 15-25 minutes before commencing unloading.
At the end of the compression test, the strain gauges were removed from the concrete
cylinder and a 50mm thick specimen was cut from the mid-height of the cylinder for
RCPT. This is shown schematically in Figure 4.1. The 50mm thick specimen was air-
dried before coating with epoxy on the circumferential surface. It was then vacuum-
saturated in a vacuum desiccator. RCPT was carried out in accordance with ASTM
Chapter 4 : Microcracking and Chloride Permeability of Concrete... 94
C1202 (1997). The current was recorded every 15 minutes using a data logger. The
chloride permeability of the concrete was evaluated by the amount of electrical charge
passing through the specimen.
<1>100mm
I 0mm thick disc for
100 microcrack examination
The 10mm thick disc, cut just above the 50mm thick specimen shown in Figure 4.1, was
used for microcrack examination under a microscope. It was stained with a red dye on
the cut surface adjacent to the 50mm thick specimen for RCPT. The disc was
subsequently polished with silicon carbide paper and water, on a glass plate, until the
surface became pink in colour. At this stage, some larger bond cracks at the aggregate-
paste interface were stained in deep red and could often be seen by the naked eye.
Enlarged photograph of the polished disc was taken. The enlargement was between 3
and 5 times of the original disc size. The concrete disc was then examined under an
optical microscope using a magnification of 20 to 40 times to observe for microcracks.
The search for bond cracks was initially made using the microscope and the observed
bond cracks were mapped on the enlarged photograph. After this, the same procedure
was followed to detect the mortar and aggregate cracks. The crack length was estimated
from the enlarged photograph by dividing the crack length into several straight
Chapter 4 : Microcracking and Chloride Permeability of Concrete ... 95
mortar
aggregate crack
\
IJ
~
coarse aggregate
Some results from the non-destructive tests are shown in Figures 4.3 to 4.6. The
initiation stress is the stress at which microcracks begin to propagate in the concrete.
This can be determined from the graphs as the point when the curve begins to deviate
from the initial vertical line. In the present study, the initiation stress is found to be
between 0.2 f c. and 0.4 f c.. It is observed that beyond the initiation stress, the specific
crack area increases almost linearly when the stress is below 0.5 f c. and it increases non-
linearly above 0.5 fc.. Figures 4.4 and 4.5 show the characteristics of microcracks in
concrete at 0.85 fc° where the critical stress is not exceeded in one instance and
exceeded in the other, respectively.
Chapter 4 : Microcracking and Chloride Permeability of Concrete... 96
0.9
0
:;:: 0.8
f!
-.
0.7
.s:
en 0.6
C
G)
0.5
1ii
I
0.4
.
1/1
1/1
G) 0.3
en 0.2
0.1
0
0 20 40 60 80
Specific crack area
in microstrain
0.9
-.
.2
m
0.8
0.7
-.
.s:
en
C
0.6
-G)
1/1
I
0.5
0.4
.1/1
1/1
G)
en
0.3
0.2
0.1
0
0 100 200 300 400
0.9
0 0.8
.::
...
ea 0.7
.s= 0.6
'5,
C
-
0.5
...
G>
Ill
I
0.4
Ill
Ill 0.3
-...
G>
(/) 0.2
0.1
recovery
0
0 100 200 300 400 500 600
0.9
0 0.8
.::
...
ea 0.7
-...
.s=
Cl
C
0.6
-G>
I ll
I
Ill
Ill
0.5
0.4
-!
( /)
0.3
0.2
0.1
0
0 200 400 600 800 1000 1200
In this series of tests, upon reaching the predetermined stress level, the specimen was
unloaded immediately. Up to 0.5 fc., the specific crack area recovery was almost 100%
as shown in Figure 4.3. It implies that microcracks close back almost completely upon
complete unloading. However, at 0.7 fc. and above, some residual specific crack areas
are observed after the concrete has been unloaded completely. This implies only a
partial closure of microcracks in the concrete. The residual specific crack area is
determined after 3 days of the creep recovery.
1200-r-----------------,
.m
~
ea 1000
. ·-
u_
ea C 800
(.) ea
u .::
:;::: Ill 600
·- 0
u .. immediate ---1'-----1
Cl) u
c.·- 400 recovery
Ill E
CD-
Cl
f! 200
~
et
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0. 7 0.8 0.9 1.0
Stress--strength ratio
~ A t predeterrrined stress
- 0 - Upon corrplete unloading
The relationship between the specific crack area at predetermined stresses and their
recovery immediately after complete unloading is shown in Figure 4.7. The difference
in the specific crack area between the two curves is the immediately recovery. It
indicates the reduction in the area of microcracks per unit cross-sectional area of the
concrete, when the concrete is unloaded immediately from its predetermined stress
level. This observation shows that the characteristic of microcracks, in terms of the
specific crack area, is indeed different from each other, i.e., when a concrete is loaded
Chapter 4 : Microcracking and Chloride Permeability of Concrete... 99
and when it is completely unloaded. A further reduction in the specific crack area is
observed due to creep recovery.
-Cl)
!II
I
!II
!II
0.5
0.4
0.3
! 0.2
en 0.1
0
0 200 400 600 800 1000
In addition, several companion specimens were also loaded to 90% of its compressive
strength and the load was sustained for a short duration of 15 and 25 minutes before
commencing unloading. The variations of the specific crack area are shown in Figure
4.8. The creep recovery was observed for several days. The residual specific crack area
Chapter 4: Microcracking and Chloride Permeability of Concrete ... 100
is about 350 microstrain when sustained for 15 minutes and 580 microstrain when
sustained for 25 minutes. It shows that microcracks continue to propagate under a high
sustained stress even for a short period of time. The behaviour of microcracks under
sustained compression is further discussed in Chapter 6 and 7.
The critical stress, <>c , represents the onset of unstable microcrack propagation (Loo,
1992). It corresponds to the maximum value of volumetric strain, beyond which the
specimen in compression undergoes a volumetric expansion. Figures 4.9 to 4.12 show
some typical curves of the axial, lateral and volumetric strains for specimens which
have been loaded up to 0.95 fc' in the compression tests.
In the present study, the critical stress is found to be exceeded in specimens that have
been loaded between 0.8 fc. and 0.95 fc .. However, there are some concrete cylinders in
which the critical stress is not exceeded when loaded to 0.80 fc' and 0.85 fc.. Many
concrete cylinders were tested in compression between 0.7 fc. and 0.95 fc.. This is to
determine the stress level where the critical stress is exceeded in some specimens and
not in other specimens. It was carried out intentionally to assess the significance of
critical stress occurrence on the chloride permeability of concrete. The critical stress is
indicated by a horizontal arrow on the volumetric strain curve. Figures 4.10 and 4.11
show the typical strain curves of concrete at 0.85 fc' where the <>c is not exceeded in one
0.9
0 0.8
.:
...
ea 0.7
-
.c
Cl
C
0.6
volumetric
-
I!!
1/1
I
1/1
1/1
0.5
0.4
-...
CD 0.3
en 0.2
0.1
0
-400 -200 0 200 400 600 800 1000
Microstrain
Figure 4.9: Relationship between Stress-Strength Ratio and Strain at 0.5 fc.
0.9 volumetric
0 0.8
.:ea
... 0.7
.c
'5, 0.6
C
-...
CD
1/1
I
1/1
1/1
0.5
0.4
-...
CD
en
0.3
0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
Figure 4.10: Relationship between Stress-Strength Ratio and Strain at 0.85 fc.
0.9 volumetric
0
;:
.
ea
.c
0.8
0.7
'5, 0.6
-.
C
Cl) 0.5
I I)
I
II)
0.4
0
..
II)
Cl) 0.3
0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
Figure 4.11: Relationship between Stress-Strength Ratio and Strain at 0.85 fc.
0.9
0 0.8
.::
f!
-.
0.7
.c
0, 0.6
C
Cl)
0.5
di lateral
I
0.4
.
II)
II)
Cl) 0.3
0 0.2
0.1
0
-1000 0 1000 2000 3000
Microstrain
Figure 4.12: Relationship between Stress-Strength Ratio and Strain at 0.95 fc'
Chapter 4 : Microcracking and Chloride Permeability of Concrete ... 103
The results of crack length measurement from the microscopy examination of the
concrete discs are presented in Table 4.1. Each result is an average of observation made
on two discs.
800
-
E
700
--
E
.c
en
600
500
C
.! 400
~
u
...u
CII 300
jij 200 Present
...
0
100 study
0+-----.--..----.--~--~----
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Stres.strength ratio
The crack length measurement of concrete under short-term (15 and 25 minutes)
sustained stress was not carried out. Figure 4.13 shows the relationship between the
total crack length and the stress-strength ratio.
It is observed that there is virtually no increase in the total crack length below 0.3 f c. • A
marginal increase in the crack length is observed between 0.3 fc. and 0.5 fc .. These are
essentially bond cracks as they occurred at the aggregate-mortar interface. Beyond
0.5 fc., the total crack length increases significantly. On the other hand, mortar cracks
become noticeable only above 0.7 fc .. They are found as isolated cracks in the mortar at
0.7 fc. but become connected with some bond cracks in the specimen which are loaded
to, for example, 0.9 fc' where the critical stress is found to be exceeded. Bond cracks
are found to propagate into the mortar at 0.8 fc.. At any stress level, the bond cracks are
considerably more than the mortar cracks. Cracks through the aggregate are found to be
insignificant at all stages of loading. Although some aggregate cracks were also
observed, it was uncertain if they were true cracks developed during loading or simply
fissures. However, the quantity is very small and was ignored.
Some results of microcrack observations from the previous studies (Ngab et al, 1981;
Smadi and Slate, 1989) are replotted in Figure 4.13 for comparison. They were
obtained from microcrack examination on normal strength concretes (33-41 MPa).
Ngab et al (1981) reported a significant increase in cracking only at stress levels above
0.45 f c.. Smadi and Slate (1989) reported negligible increase in microcrackings below
0.40 fc. but a significant increased in microcracking at stresses above 0.70 fc .. The total
crack lengths are different among the various studies due to different strength of
concrete, mix proportions and the size of specimens used for crack count. The size of
specimens used by Ngab et al (1981), and Smadi and Slate (1989) was 89 x 89mm and
<I> 102mm, respectively. The microcrack observation in the present study is consistent
with the previous works. Some of the cracking maps are shown in Appendix D.
Chapter 4 : Microcracking and Chloride Permeability of Concrete ... 105
The results of rapid chloride permeability tests at different stress-strength ratio are
shown in Figure 4.14. At 0.80 fc. and 0.85 fc., the electrical charge passed through the
concrete where the critical stress is exceeded as well as not exceeded in some cases, are
both shown for comparison. The measured charge passed in all the tests is normalised
against the charge passed through unloaded control specimens and is shown in Figure
4.15. Each result is an average of two specimens.
At stress levels below 0.5 fc., there is no apparent increase in the charge passed when
compared with the unloaded control. This can be attributed partly to the closure of the
microcracks when the specimen was completely unloaded, as depicted in Figure 4.3.
The immediate recovery from 0.3 fc., for example, is found to be about 25 microstrain
with zero residual specific crack area (Figure 4.7). It shows that microcracks close back
completely when the specimen is unloaded from 0.3 fc°. There is a possibility that there
show a difference when concrete is at 0.3 fc. and when unloaded completely. It appears
that the influence of microcracks on the mass transport in concrete cannot be assessed
by the crack length only. Depending on the stress level at which the concrete is
subjected, microcracks can close back partially or completely upon unloading. This is
evident from the immediate recovery in Figure 4.7. The ability of the microcracks to
'open' and 'close' shows that the specific crack area is a more sensitive parameter to
consider than the crack length, when relating permeability to stresses in concrete. The
permeability of concrete can, thus, be influenced by the test condition, i.e., whether the
test is carried out under loading condition or after removal of load.
Chapter 4 : Microcracking and Chloride Permeability of Concrete... 106
3000
-I ll
Jl
E
2500
.2
:, 2000
-- -- - >--
~
>--
~
I-
~
-0
u
"C
G>
Ill
1500 - - >-- >-- I-
- -
Ill
ea
c. 1000 - - I-
G>
...
en
ea
.c
0
500 - - - - I-
0
0 0.3 0.5 0.7 0.8 0.85 0.9 0.95
Stress-strength ratio
D critical stress not exceeded • critical stress exceeded
1.5
"C
G> 1.4 curve 'abd' - O'c not exceeded
Ill C
Ill 1.3
ea
c. curve 'be' - O'c exceeded
Cl) 1.2
...
en
ea 1.1
.c
u 1.0 a
'i 0.9
.!!!
-;
0.8
...0
E
z 0.7
0.6 ~--~--r---r---~--.---.--~,_,,..--,-----1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Stress-strength ratio
At 0.7 fc., the increase in crack length is substantial at 125% when compared with that
of the unloaded control. However, the normalised charge passed (Figure 4.15) appears
marginal at about 1.05. A possible reason for this would be the critical stress is not
exceeded at 0.7 fc.. The inter-connectivity of the microcracks in the concrete matrix
appears to be another factor influencing the charge passed in a RCPT. In the present
study, when the critical stress is exceeded, a comparatively large charge passed is
measured. There are two stress levels, beyond 0.7 fc', where the critical stress is
exceeded in some specimens and not exceeded in other specimens. As shown in Figure
4.15, a comparison between the charge passed at 0.80 fc. and 0.85 fc. clearly shows the
influence of the occurrence of critical stress on the chloride permeability of the concrete.
From the literature, there is conflicting evidence as to how stress level affects chloride
permeability of concrete. Samaha and Hover (1992), for example, reported a low
average charge passed for stress levels below 0.75 fc. despite a considerable increase in
the crack length at 0.75 fc. when compared with the unloaded control. He found the
charge passed to be more pronounced when the concrete was loaded to its full
compressive strength. In contrast, Saito and Ishimori (1995) conducted similar study
but reported negligible increase in the chloride permeability at 0.9 fc.. In both studies,
no attempt was made to determine the critical stress occurrence in their test specimens.
Based on the observation made in the present study, the conflicting evidence can be
explained in terms of the occurrence of the critical stress in the specimen. In the present
study, the increment of stress level considered beyond 0.7 fc. is small to ensure that
useful observations are not eluded. The present study shows that the increase in the
total crack length is notably high at 0.80 f c. and 0.85 fc. although the critical stress is not
exceeded. However, a comparatively higher charge passed is observed only when the
critical stress is exceeded at these stress levels. Therefore, the low charge passed at
0.75 fc. (Samaha and Hover, 1992) and at 0.9 fc. (Saito and Ishimori, 1995) can be
attributed to the critical stress being not exceeded in the specimens. The wide range of
critical stress occurrence in concrete should be considered in addition to the strength
variability.
Chapter 4 : Microcracking and Chloride Permeability of Concrete ... 108
4.4 CONCLUSIONS
microcracks close back completely. However, when unloaded between 0.7 fc. and
0.95 fc. stress levels, some residual specific crack areas are observed immediately
after complete unloading. This implies only a partial closure of the microcracks.
b) It appears that the influence of microcracks on the mass transport in concrete cannot
be assessed by its crack length only. Depending on the stress level at which
concrete is subjected to, microcracks can close back partially or completely upon
unloading. The ability of the microcracks to 'open' and 'close' shows that the
specific crack area is a more sensitive parameter to consider than the crack length,
when relating permeability to microcracking in concrete due to stresses.
c) In the present study, the critical stress is found to be exceeded when the concrete is
loaded to a stress level between 0.80 fc. and 0.95 fc .. However, in some specimens
the critical stresses were not exceeded when tested at 0.80 fc. and 0.85 fc ..
e) The chloride permeability of concrete may be influenced by the test condition, i.e.,
whether the test is carried out on concrete specimen under loading condition or after
removal of load.
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 109
Chapter 5
PREDICTION OF CHLORIDE CONCENTRATION IN
CONCRETE USING A MIX DESIGN PARAMETER
5.1 INTRODUCTION
The rate of chloride transport through the concrete cover is the primary factor that
determines the time to the initiation of corrosion of steel reinforcement in concrete.
External chloride ions may be transported into the concrete by one or a combination of
the various transport mechanisms, namely, capillary suction, permeation or diffusion. In
most practical applications, chloride ion diffusion is assumed to be the primary transport
mechanism into concrete and that Fick' s Second Law of pure diffusion is applicable
(Bamforth, 1994; Mangat and Molloy, 1994; Sandberg, 1995; Maage et al, 1996) vis-a-
vis;
ac
--D-
a2 c (5.1)
ot -
2 C ox
Equation (5.1) represents one-dimensional chloride diffusion into a semi-infinite
medium under a non-steady state condition. Assuming that the initial chloride
concentration, C; = 0, an analytical solution to equation (5.1) is given as;
(5.2)
where Cx is the chloride concentration at distance, x from the exposed surface, Cs is the
surface chloride concentration, De is the chloride diffusion coefficient, tc is the chloride
exposure time (seconds) and erfis the error function (Lin and Jou, 1991; Monica et al,
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 110
1996; Bamforth, 1996). One of the implications of using equation (5.2) is the
assumption that the diffusion process through the concrete is constant with time.
However, concrete is not an 'inert' material. The rate of chloride diffusion into concrete
is influenced by the physical resistance of concrete (depends on maturity) and the
chemical interactions between chloride ions and the hydration products of cement.
Chloride ions can be bound to the hydrated products in the concrete or exist as free
chlorides in the pore solution. Bound chlorides in the concrete can either be chemically
combined with C3A to form Friedel's salt (calcium chloroaluminate hydrate) or
physically adsorbed on to the cement gel (Tang and Nilsson, 1993; Dhir et al, 1996).
Because of these, chloride diffusion coefficient (De) of concrete has been found to be a
time-dependent parameter.
Studies have shown that an empirical relationship between De and t can be expressed
mathematically as a power function (Gerhard and Walter, 1987; Takewaka and
Matsumoto, 1988; Mangat and Molloy, 1994; Maage et al, 1996). Maage et al (1996)
expressed the variation of De with time, typically as,
D =D
C O
(1-)"
t (5.3)
(5.4)
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design ... 111
where tLT = service life time; Dpo = potential chloride diffusion coefficient at t 0 and
ex = parameter depending on material and environment.
Mangat and Molloy (1994) proposed an empirical relationship for De and time as,
D= D.t-m
C I
(5.5)
D C
=D.(!)m
I t (5.6)
Mangat and Molloy (1994) proposed the following equation for the prediction of the
long-term chloride concentration profiles in concrete,
X
Cx = Cs 1- erf - - ; = = = = (5.7)
2 D.I f(l-m)
(1-m)
Andrade et al (1997) reported that D; values obtained from equation (5.7) seem rather
high. Mackechnie (1995) also pointed out that equation (5.7) over-estimates the short-
term chloride concentration considerably while the discrepancy of long-term prediction
is less significant. He attributed the discrepancy to the diffusion coefficient not being an
instantaneous value but an integrated value over the entire time considered.
Mackechnie (1995) adopted the same empirical relationship for De and time, equation
(5.5), in his formulation for a long-term chloride prediction in concrete. It was a
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 112
Equations (5.3) and (5.5) show the time-dependency of De. In this case, De varies with
the age of the concrete (t). On the other hand, the chloride concentration in the concrete
depends on the duration of chloride exposure (te), as indicated by equation (5.2).
However, equations (5.4), (5.7) and (5.8) clearly show no distinction is made between
the concrete age and the chloride exposure time, i.e., it was assumed that t = te .
.
t = t.I + t C
(5.9)
D = D.(t.I +t
C l C
)-m (5.10)
(5.11)
The parameter, m is the slope of a straight line of log De versus log t. This is discussed
in details in section 5.3.
D;fm I D,{t;+tcY'm
3.00 ~ - - - - - - - - - - - - - - - - ,
(a) t; = 90 days
2.50
m=0.1
2.00
1.50
/ m=0.5
m=0.3
1.00
0 6 12 18 24 30 36 42 48 54 60 66
Chloride Exposure Time (Month)
2.50
m=0.1
m=0.5
2.00
m=0.3
1.50
1.00
0 6 12 18 24 30 36 42 48 54 60 66
Chloride Exposure Time (Month)
Figure 5.1 : Effect of Initial Curing Time (t;) on the Diffusion Coefficient
against Chloride Exposure Time.
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 114
The effect of initial curing time, t;, on De is shown in Figure 5.1. From these plots, it
can be seen that t; affects the short-term prediction of chloride concentration
significantly. When t;~6 months, the effect on De is significant when the duration of
chloride exposure is less than 5 years. Furthermore, it is observed that the effect oft; on
De becomes more pronounced for blended cement concrete with increasing replacement
level for pozzolanic materials. This is shown with an increasing value of m. A larger
value of m is associated with a higher replacement level of cement by pozzolanic
materials in concrete (Maage et al, 1996; Bamforth, 1996).
In the present study, the time dependent diffusion coefficient of equation (5.6)
constitutes the basis for developing a diffusion model incorporating the effect oft; and a
concrete mix design parameter. A modification is then proposed to equation (5.2)
which enables prediction of chloride concentration in an unstressed OPC concrete.
(5.12)
where D30 = diffusion coefficient at time equal 30 days (1 month) and t = time in
month. The value of '1' on the numerator of equation (5.12) represents 'I-month'.
When a change in chloride concentration occurs in the concrete due to chloride ingress,
values of m shall not exceed 1.0 (Mackechnie, 1996; Maage et al, 1996). When m~ 1.0,
the concrete is said to be 'self-blocking', that is, there is no change in the chloride
concentration profile with time (Maage et al, 1996).
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 115
It can be seen that the difference between D30 and D2s is insignificant. Thus, equation
(5.12) can also be expressed as,
(5.13)
For short-term prediction, it has been shown in section 5.2 that the effect of initial
curing time (t;) is significant. Thus, substituting equation (5.9) into (5.13) yields,
(5.14)
Note that t; is replaced with t;_J in equation (5.14) to account for the one month of initial
curing time assumed in D2s. The model for D28 is discussed in section 5.3.3.2.
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 116
Mangat and Molloy (1994) proposed an empirical relationship form and w/c as,
Equation (5.14) can be used for both short and long-term chloride concentration
prediction in concrete when the values of D28 and m are known. This is the subject of
discussion in the following sections.
The chloride diffusion coefficient (De) has been found to vary with the maturity of
concrete. A linear relationship is observed between De and time (t) when they are
plotted on a double logarithmic scale (Takewaka and Matsumoto, 1988; Mangat and
Molloy, 1994; Maage et al, 1996; Bamforth, 1996; Cao et al, 1999). Some of the graphs
are reproduced here as Figures 5.2, 5.3, 5.4 and 5.5. These results are obtained from
tests conducted under the same exposure condition.
1.00E-09
MPa, w/c
22, 0.74
u 1.00E-10
CD
25, 0.67
~
E 32, 0.60
.5
u 1.00E-11
C 50, 0.41
1.00E-12 +--....--...---.-....-r-T""....-r--..-.....-..............-..........i
10 100 1000
Time in days
Grade 40
u 1.00E-11
CD
~
E
.5
u 1.00E-12
C
1.00E-13 +----_.......,..-,.-......,.....,..----,--.--..---,--.-l"T"T'"i
1 10 100
Time in days
Grade 50
1.00E-11 . . . . . - - - - - - - - - - - - - - - - - - .
u
CD
~
E 1.00E-12
.5
u
C
w/b = 0.41
(Blended)
1.00E-13 ~---_.......,......,....,...........,......-_.-...-......................-t
10 100
Time in days
10000
l, w'1>-falio
%CSF
ee 1000
.r 0.70
0%
-~
E'
§ 100
C:
J
.
"O
19
C:
s0
10
a.
1
1 10 100 1000
Maturity age of the concrete, days
The empirical constant (m) has been reported to be dependent on the type and grade of
concrete (Mackechnie, 1997). However, Figures 5.2 and 5.5 clearly show that the
water-binder ratio (w/b) does not influence the slope (m) of the straight line although it
affects the individual De values. This finding disputes the relationship between m and
w/c (equation 5.15) proposed by Mangat and Molloy (1994) which they derived based
on very limited data obtained from his studies.
The results in Figures 5.2 to 5.5 indicated that for a given type of concrete and exposure
condition, m is not dependent on the grade of concrete. Environmental factors may
influence the empirical constant (Maage et al, 1996; Bamforth, 1999). The w/b mainly
affects the initial chloride diffusivity, i.e., the value of De at they-intercept. In contrast,
Figures 5.3, 5.4 and 5.5 show a marked influence of binder types and their percentage
replacement on the slopes of the straight line plot. The slope of the straight line
becomes steeper for concrete made from blended cement than OPC alone. From these
figures, it can be concluded that,
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design ... 119
Based on the evident shown here, it is clear that equation (5.15) cannot be used to
predict the empirical constant because it is not dependent on w/c. In the present study, a
typical value of m is derived and a model for D28 is developed, both for unstressed
normal strength OPC concrete ( ~ 50 MPa). It is based on laboratory test results. This is
shown in the following sections.
In the present study, a range of values of m has been determined for OPC concrete ( ~ 50
MPa) based on data collected from the literature. These data consist of results from
laboratory chloride immersion tests, field exposure tests and in-service marine
structures. The results from field exposure tests and in-service marine structures are
considered for the purpose of comparison with the laboratory tests only. The followings
are the analyses carried out on these data. The straight line shown in the graph of log De
versus log t is the 'best-fit' through a set of parallel lines from the same data group.
Results are obtained from 12 sources which include different curing conditions and age,
concentrations of chloride solution and whether acid- or water-soluble chlorides are
used to determine De. In Figure 5.6, the De values are determined from the acid- and
water-soluble chlorides obtained from the same concrete specimen under the same
curing and testing conditions. Thus, a common basis for comparison can be made to
evaluate if the types of chlorides can influence the value of m. It is evident from Figure
5.6 that the slopes of these two lines are similar, indicating that m is not influenced by
the types of chlorides.
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design. .. 120
-10.5 . , . . . . . . - - - - - - - - - - - - - - - - - ,
acid-soluble
-
u
Gl -11
<>
a ~
(m = 0.13)
! <>
~
-
E
u
f ........... ·i.--~ ....... ~
8 ~ a
C
c, -11.5 a / a
.2
water-soluble
(m=0.17)
-12 + - - - - - - , - - - - . . , . . . . - - - , - - - - r - - - -
0.2 0.4 0.6 0.8 1.2
log t (months)
a Water-soluble(Dhir, 1991)
-9
-9.5
2- 5% NaCl
-u
G)
-10
-10.5
<>
<> (m = 0.42)
~
-
u
C
E
C)
-11
-11.5
.2 -12
<>
29% NaCl
-12.5 (m = 0.46)
-13
0 0.5 1 1.5 2
log t (months)
-9.5
7d moist-cured
-10 a
-i
~
N
-10.5
a a
{m = 0.59)
Cl
a a a
-
C
E
u
-11
-11.5
a
-"El-
.2
Cl
rP t:1-G-tJ
28d moist-cured
-12 3d moist-cured
{m = 0.43)
{m = 0.53)
-12.5
0 0.5 1 1.5 2
log t (months)
a 7d rroist-cured: Wee(1996); CSIRO(1998); Cao(1999)
The effect of chloride solution concentration on the value of m is shown in Figure 5.7.
The specimens were moist cured for a duration of between 3 days and 28 days. While
the diffusion coefficient is affected by the concentration of the chloride solution, the
value of m does not appear to be significantly influenced by solution concentration.
Also, the duration of the moist curing does not appear to have any influence, as evident
from Figure 5.8. The value of m shown in Figure 5.6 appears to be relatively small
compared with those in Figures 5.7 and 5.8, probably due to limited data available in
Figure 5.6.
In the following, the empirical coefficient of OPC concrete, based on the laboratory
chloride immersion tests, is determined by analysing all the results from Figures 5.6, 5. 7
and 5.8. This is shown in Figure 5.9. Table 5.2 summarises the various m values
derived from the analyses. The empirical coefficient, m = 0.42 is adopted for the
proposed model given in equation (5.14).
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design ... 122
-7
-8
-
~
~
-9
-10
-. -------
m = 0.42
-
N
E ·11
u
C
···-------
-12
Cl
.2 lower 95%
-13
-14
-15
0 0.5 1 1.5 2
log t (months)
Figure 5.9: Analysis Using All Laboratory Chloride Immersion Tests Data
Empirical Coefficient
Linear Regression Analysis based on
(m)
-7
-8
u
Cl)
-9
-!!! -10
-
N
E
u
-11 a R a a a
C
0, -12
.2
-13
-14
-15
0 0.5 1 1.5 2 2.5
log t (months)
Figure 5.10: Analysis Using Field Exposure Tests from the Literature
For the purpose of comparing them value obtained from the laboratory immersion test,
similar analysis was carried out on data collected from field exposure tests. Results
obtained from 9 sources are shown in Figure 5.10. The data have been published by
researchers in the United Kingdom, Japan, South Africa, Holland, Norway and Sweden.
These data were obtained from laboratory cast OPC concrete specimens which had been
exposed to field conditions, i.e., splash, tidal and submerged zones. No attempt has
been made to differentiate between sources. From the analysis, m = 0.38 is obtained.
The results obtained from 7 sources are shown in Figure 5.11. These data were obtained
from marine structures made from OPC concrete and built in Hong Kong, Taiwan,
Singapore, Norway and Japan. They had been exposed to splash, tidal and submerged
conditions for a duration of between 1.8 years and 68 years. Chloride samplings were
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design ... 124
taken mostly from the piles, piers, wall of storage tanks and beams. Again, no attempt
has been made to differentiate between sources. From the analysis, m = 0.77 is
obtained.
-7
-8
-
u
Cl)
-9
-10
m = 0.77
~
-
E
u
C
C,
-11
-12
.2
-13
-14
-15
1 1.5 2 2.5 3
log t (months)
Figure 5.11: Analysis Using Data from In-Service Structures from the Literature
This value appears larger than those derived from laboratory immersion (m = 0.42) and
field exposure (m = 0.38) tests, described in section 5.3.2.1 and 5.3.2.2 respectively. A
relatively large m implies that the reduction in De is relatively high. This could be
attributed partly to the stresses in the structure in service which impede chloride
diffusivity into the concrete. The evidence of De reducing with increasing compressive
stresses is highlighted and discussed in detail in Chapter 6. The formation of a layer of
aragonite and brucite on the concrete surface is another reason for the reduction in De of
the concrete structures exposed to sea water. The protective layer formed can reduce the
permeability of the concrete. Ionic reactions between chlorides and tricalcium
aluminate to form chloroaluminates, and magnesium and sulphate ions (from sea water)
with cement compounds to form brucite and ettringite can further reduce concrete
porosity. This effect is particularly relevant in the submerged and tidal zones where
concrete is constantly in contact with sea water (Costa and Appleton, 1999).
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design ... 125
5.3.3.1 Introduction
For a given concrete grade and type, the chloride diffusion coefficient (De) of equation
(5.14) can be predicted at any time, t, if D28 and m are known. Using the proposed
model of equation (5.14), the value of m has been shown to be only dependent on the
type of concrete under a given exposure condition. This is discussed in section 5.3.1. It
follows that for a given type of concrete, for example OPC concrete, the progressive
reduction in the diffusion coefficient with time is similar, irrespective of the water-
cement ratio. This is shown in Figure 5.12 using some data from the literature. The
initial moist-curing time is one month, hence, t;.1 = 0. The values of the measured D 2s
corresponding to w/c = 0.67, 0.58 and 0.45 are 70x10· 12 m2/s, 50xlff 12 m2/s and
19.6x 10· 12 m2/s, respectively. The magnitude of the diffusion coefficient at a particular
time, t, depends on the value of D 28 used in the model. As discussed in section 5.3.1,
D2s is influenced by w/c and, hence, the strength of the concrete.
BE-11
u 7E-11
Cl)
Ill
ri:I 6E-11
E
-
C
5E-11
w/c=0.67
C
.!!!
u 4E-11 w/c=0.58
:ECl)
0 3E-11
u
C
0
·;;; 2E-11
:,
:E
C 1 E-11
0
0 10 20 30 40 50 60
Time in months
Weyers and Smith (1989) proposed equation (5.17) for D (in 2/year) as a function of w/c
between 0.35 and 0.50. It was based on the chloride test results conducted in a
laboratory and outdoor exposure conditions for 16 weeks. The concrete specimens were
made from OPC, and some containing silica fume and flyash replacement.
The two models discussed were developed from results obtained from specimens which
were influenced by many factors, for example, environmental loads, stresses (when
sampling from in-service structure), exposure conditions, types of concrete, and other
factors, all 'lumped' together. Therefore, the effect of individual variable such as stress
condition or cracks on D cannot be assessed. The compressive stresses, for example,
have been shown in Chapter 6 to influence the D value. In addition, D is a time-
dependent parameter, and equations (5.16) and (5.17) do not appear to account for this
effect.
In this section, a model for D28 as a function of w/c is developed for OPC concrete
under a laboratory-controlled environment. Data from the literature are reviewed and
included to supplement the results from the present study. The Dis model developed in
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 127
From the literature, data from 11 different sources were obtained. The diffusion results
are from chloride immersion tests conducted in a laboratory-controlled environment at
23± 2°C and 50± 5% RH. The duration of the moist-curing for the concrete specimens
was between 7 days and 27 days. However, most of the specimens were moist-cured for
27 days. The present analysis considers only the results of OPC concrete in sodium
chloride solutions having a concentration between 2% and 5%.
The chloride diffusion coefficients are obtained from concrete after 28 days of
immersion in sodium chloride solutions. However, some are extrapolated to 28 days
based on the known values of De at several chloride exposure durations. This was done
by carrying out linear regression analyses on logDe versus log t.
80.00
70.00 <>
-
~
60.00
-...
E
N
d>
50.00
40.00
>< 30.00
CIO
N
Q 20.00
10.00
0.00
0.3 0.4 0.5 0.6 0.7 0.8
water-cement ratio
A few results are extrapolated to D 28 using the relationship given in equation (5.5) and
adopting m = 0.42. This is due to insufficient data to produce a linear plot of log De
versus log t. The value of m = 0.42 is adopted because it has been shown in section
5.3.2.1 to be the most appropriate empirical coefficient for chloride data derived from
laboratory tests.
From the limited data on D 28 and w/c, it can be generally observed from Figure 5.13 that
D 2 s becomes larger with increasing w/c. The following equation, derived from a
regression analysis, can be used to estimate the value of D28 for OPC concrete having a
w/c between 0.4 and 0.67,
(5.18)
This model can be further improved when more data become available. It shall be noted
that D 2s (in m2/s) refers to the chloride diffusion coefficient of concrete at the age of 28
days. It shall be determined after 28 days of initial moist-curing and a further 28 days of
chloride immersion in a laboratory condition. Using the proposed diffusion model, D 28
is found to vary between 13.2x10· 12 and 69.3x10· 12 m2/s within the range of w/c
considered.
Based on the discussion in this section, the proposed prediction model of equation
(5.11) can be re-expressed as given in equation (5.19). This is carried out by
substituting equation (5.14) into (5.2) and expressing the time in terms of months.
(5.19)
An experimental program was carried out to verify the proposed model, equation (5.19),
for predicting the chloride concentration profile in concrete using laboratory immersion
tests.
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design ... 129
Grade 20 and 40 concrete mixes were designed using locally available coarse and fine
aggregates as described in section 3.3.1. Concrete mixes were prepared in accordance
with section 3.3.2. Details of the mix proportions are given in Table 3.3 of section
3.3.4.
Concrete prisms (75 x 75 x 150mm) were prepared and cast vertically in 2 layers. They
were moist cured in lime saturated water up to 28 days of age. At 28 days, all the
specimens were removed from the moist curing. They were coated with epoxy on all
the surfaces and leaving only one surface, 75 x 150 mm, uncoated. The specimens were
immersed in a 3% NaCl solution at the age of 30 days after casting. The duration of the
chloride immersion are 1, 3, 10 and 18 months. The surface chloride concentration was
determined at the end of each immersion period.
A few Grade 40 concrete specimens moist-cured until they reached the age of 5 months.
At 5 months, the specimens were coated with epoxy, as for earlier specimens, before
dipping into the sodium chloride solution for a duration of 3 and 12 months. The initial
curing time (ti) of these specimens is, thus, 5 months prior to the chloride exposure.
Two concrete specimens were used for chloride concentration profile determination.
For each specimen, two holes were dry drilled using a 20mm diameter rotary impact
drill. These holes were located at the mid-height of the uncoated surface of the concrete
prism. Concrete powder obtained from the same sampling depths from the two prisms,
were combined to give a test sample. At least 5.0g of powdered concrete sample was
used to extract acid-soluble chlorides (Berman, 1972). Mohr titration was used to
determine the chloride concentration in the solution. Chloride diffusion coefficient was
determined from the best-fit curve represented by equation (5.2).
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 130
As the rate of chloride penetration into concrete is determined in part by the surface
chloride concentration (Cs), a reliable value needs to be established. In this section, a
surface chloride concentration based on chloride immersion test in a laboratory-
controlled environment is derived. This is done by carrying out analyses of chloride
concentration profiles of normal strength OPC concretes (:::; 50 MPa ) after immersion in
2-5% NaCl solutions. Relevant data from the literature are also analysed to supplement
the data from the present study so that a more representative value can be obtained.
0.90
0.80
c_ 0.70
.....- c. value
.E CD 0.60
-f! 'i
..
-C u
C 0.50
Cl) 0
u u 0.40
5i
0 0 0.30
~
o-I
0.20
0.10
0.00
0 2 4 6 8 10 12 14 16 18 20 22 24
Distance, x (mm)
It is not likely to determine Cs in practice, for example, using the drilling method. In
many instances, Cs is determined by extrapolation from the best-fit curve of the
measured chloride profile. They-intercept of the best-fit curve of the chloride profile is
the Cs value. A typical chloride profile with the best-fit curve after 3 months of
immersion is shown in Figure 5.14.
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 131
Cs values obtained from the chloride profiles at different periods of immersion in NaCl
solution are plotted in Figure 5.15. Most of the C values tend to fall within a band of
about 0.3-0.8% by weight of concrete. While C varied, the data exhibited no general
trend to increase or decrease with time. It is clear that Cs values remain constant over
the immersion period of up to 18 months. The mean C value of the data is 0.55% by
weight of concrete with a standard deviation of 0.16%. The upper and lower 95% limits
at 0.81 % and 0.29%, respectively, are also shown in the figure. The mean Cs of 0.55%
by weight of concrete seems to be fairly representative for OPC concrete in laboratory
tests and is, thus, adopted for the proposed chloride prediction model of equation (5.19).
0.9 9
a, • Upper 5%
! M~------------------
g
8
0.7
o.6
!a I a
•
a
El •
~ - - - + - ~- ~ - - Mean
0.5 ......
0
0.4
a
a
a t:f Q a•
~
';
M~~-------~---------
0.2
a
a
a
0
Lower 951½
0
0.1
0 +-~--.--..----.--....---..----.--....---.----r---1
0 2 4 6 8 10 12 14 16 18 20 22
Months
• present study
In a review of published data on C values based on field exposure tests, it was reported
that the surface chloride established itself quickly in relation to the expected life of a
structure, and remained approximately constant thereafter (Takewaka and Matsumoto,
1988; Mangat and Molloy, 1994; Bamforth, 1999). This trend was found to occur
primarily for concrete in the submerged zone or in direct exposure to the sea water
(Jaegermann, 1990; Costa and Appleton, 1999).
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 132
Verification of the proposed chloride prediction model was carried out using the
chloride concentration results obtained from the present study and the literature. These
results are not used in any way to formulate the model so as to ensure impartiality of the
data obtained. Results from the present study are initially verified. The initial curing
time of 5 months is chosen in the present study to verify its significance on the predicted
chloride profile. The chloride profile can be predicted using equation (5.19) and using
m=0.42. D28 can be estimated from equation (5.18). The surface chloride concentration
of 0.55% by weight of concrete is adopted.
0.6 . . . - - - - - - - - - - - - - - -
c-
o.! 0.5
~ !
f! CJ 0.4
-C C
0
Cl) CJ
CJ -0
C 0.3
8 .
i
Cl)
0.2
:E >,
... .c
.2 ~
.c
(.)
0
- 0.1
0 +--------~--------''--'
0 5 10 15 20
Depth (mm)
• Measured - - Best-fit curve
- - - - - - Mangat's Model - - Proposed Model
Figure 5.16 shows the comparison between the predicted and the actual chloride profiles
obtained from specimens after 3 months of immersion in 3% NaCl solution. The actual
chloride profile is indicated by a best-fit curve through the measured chloride contents,
by means of the least square fitting. The predicted chloride profile using Mangat' s
model (equation 5.7) is also shown for comparison. The value of m in Mangat's model
is calculated from equation (5.15) and D; from equation (5.6). Mangat and Molloy
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 133
(1994) proposed an average value of 1.5% by weight of binder for Cs. This was based
on his findings that Cs remained constant over the exposure period of up to 5 years. In
the present analysis, the chloride content of 1.5% by weight of binder is converted to
about 0.3% by weight of concrete. This is done by multiplying 1.5% with the ratio of
the weight of binder to the total weight of binder and aggregates used in the mix
proportion (Mangat and Molloy, 1995).
0.6 , - - - - - - - - - - - - - - - - - ,
c-
o .!
:;::: !
f! CJ
-C 0C
Cl) CJ
CJ-
c 0
0
CJ .,;
Cl) 3:
"C
i:
>-
.Q
.2 ~
c3 e...
0 5 10 15 20
Depth (mm)
• Measured - - Best-fit curve
- - - - - - Mangat's Model - - Proposed Model
The predicted chloride profile using the proposed model is found to compare favourably
with the measured data than the predicted profile from Mangat's model. The poorly
predicted profile from Mangat' s model can be due to the values of Cs, m and D; used in
his formulation. In Figure 5.17, the values of Cs = 0.55% and m = 0.42 which are
proposed in the present study, are used in Mangat's model. In addition, D; is also
replaced with D2s. It can be seen that there is an improvement to his predicted profile
although it tends to over-estimate the chloride content. The over-estimation is attributed
to the effect of initial curing time of concrete. The significance of the initial curing
time, especially for short-term prediction, has been shown and discussed in section 5.3.
Chapter 5: Prediction of Chloride Concentration in Concrete Using a Mix Design... 134
0.6
c - 0.5
oS
:;::::;
G)
ea ._
... u
-C C
0
0.4
•
~ u
c -
0 0
0.3
u .
G) i 0.2
:E >-
0 .0
::c ~ 0.1
(J -
0
0 5 10 15 20 25
Depth (mm)
• Measured - - Best-fit curve
- - - - - - Mangat's Model - - Proposed Model
0-
e-
C
.::
G>
0.6
0.5
-
C
f(,)
C 0.4 ~
-
G> 0
"
(,) (,)
C
0 0 0.3
0 .,J ---- -
,,
G> 3 0.2 ---• ~
>-
·c: .0 ~
.c
0 -
.2 -;!!.,
0
0.1
0
--
-- -- -- -----
------- -- -- .
0 5 10 15 20 25 30 35 40
Depth (nm)
• Measured - - Best-fit curve
- - - - - - Mangat's Model - - Proposed Model
Figure 5 .18 shows the measured and predicted profiles of concrete after 12 months of
immersion. Again, the proposed prediction model shows a fairly good correlation with
the measured data.
The validity of the proposed model is further tested using some measured chloride
profiles obtained from the literature. Figure 5.19 shows the measured chloride contents
of a Grade 40 OPC concrete after 12 months of immersion in 3% NaCl solution (Khatri
et al, 1996). The initial curing time was 1 month. The water-cement ratio of the mix
proportion was 0.52. It can be seen that a fairly good prediction of the chloride profile
can be derived using the proposed model.
0.6
C
-
0 -Cl)
·-..
-c,s ..
-C C
Cl)
u
0.5
0 0.4
~ u
c -
O 0 0.3
u .,;
Cl)
0.2
==
:2 >,.
0 .0 0.1
.c~ 0
o-
0
0 5 10 15 20 25 30
Depth (nm)
- - Best-fit curve of measured data
- - - - - - Mangat's Model
- - Proposed Model
Further verification of the prediction model is demonstrated in Figure 5.20. The best-fit
curve shown in the figure is derived from OPC concrete having a w/c = 0.40 (Hooton
and McGrath, 1995). The age of the concrete prior to the chloride immersion test was
about 2 months. Chloride immersion was carried out for 4 months in a 6% NaCl
solution. The reported De after 4 months of immersion was 8.68xl0- 12 m2/s while the
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design... 136
predicted De was 6.71x10· 12 m2/s. The difference in the De value may be partly due to
the higher chloride concentration in which the test was conducted. The proposed model
is formulated based on results of immersion tests in 2% to 5% NaCl. However, the
predicted chloride profile correlates fairly better than the predicted profile from
Mangat's model. Using the proposed model, again, a good prediction is obtained using
data from Weyers and Smith (1989) and this is shown in Figure 5.21.
0.60
c- 0.50
0 CD
-.
;: 1,
ftl ..
u 0.40
C C
CD 0
u u
so 0.30
u
CD
3! >-
i
.
0.20
•
5 .a
:c
o-
';I.
0.10
0.00
0 10 20
•
30 40
Depth (mm)
In conclusion, Figures 5.16 to 5.21 show that the proposed prediction model can
reasonably predict the chloride profile of a normal strength OPC concrete based on
laboratory tests.
Table 5.3 shows the comparison between the measured and predicted chloride diffusion
coefficient from the present study and several other sources. Again, the comparison is
made with results which are not used to formulate the diffusion model. D 28 is estimated
by substituting the value of w/c into equation (5.18). The predicted diffusion coefficient
after chloride exposure time, te, is computed from equation (5.14) and using m = 0.42.
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design ... 137
The measured De values from Weyers and Smith (1989), Hooton and McGrath (1995)
and Khatri et al ( 1996) are based on laboratory immersion tests. As a matter of interest,
several De values from field exposure tests in sea water (Mackechnie and Alexander,
1997; Sandberg et al, 1998; Pedersen and Amtsen, 1998; Thomas, 1998; Costa and
Appleton, 1999) and from in-service marine structures (Browne, 1982; Liam et al, 1992)
are also calculated and compared with the predicted values. The results are from
concrete exposed to tidal and submerged conditions.
It can be seen that a fairly reasonable prediction can be derived using the proposed
diffusion model for concrete in a laboratory test condition. In addition, the predicted De
of concrete from some field exposure tests and in-service structures also appear to be
fairly reasonable although in some cases the predicted De is almost twice of that
measured. The proposed model is developed for laboratory test conditions and the fairly
reasonable prediction for field exposure tests and in-service structures may be just a
coincident.
Very often the service life prediction of a concrete structure is related to the quality of
the concrete. Less emphasis is placed on the effect of concrete cover variation on the
service life prediction. A large variation of cover thickness can be encountered at the
site if a stringent level of supervision is not enforced. A survey had been carried out on
some building sites in Australia in 1979 (Sirivivatnanon and Cao, 1991). It was found
that the levels of confidence for achieving specified minimum concrete cover were poor.
Less than 50% of the sites achieved a 90% levels of confidence. One of the important
aspect in reviewing the current performance based specification for concrete structures
in Australia is the emphasis on the need for concrete cover control at site.
-.
20
u, 18
ea Cover (mm)
-G)
>-
G)
::
16
14
12
60
55
G)
u
'f
G)
10
50
,,
u, 8
-
~
G)
u
f
6
4
45
40
D. 2
0
0.35 0.4 0.45 0.5 0.55 0.6 0.65
water-cement ratio
Using the chloride prediction model (equation 5.19), the influence of cover variation on
the service life prediction of a structure is investigated by making some assumptions to
the values of Cs , m and the chloride threshold level. The objective is to determine the
trend rather than to quantify the service life. Figure 5.22 shows that, at low w/c, a small
variation in the cover can significantly influence the service life prediction (in terms of
absolute value) of a structure. However, the effect of cover variation on service life
prediction diminishes as the w/c is increased. For a marine structure where high
performance concrete (with low w/b) is normally specified, a stringent control on the
cover achieved at site is, therefore, essential.
5.6 CONCLUSIONS
a) The initial curing time of concrete (t;) affects the short-term prediction of chloride
concentration in concrete significantly. The effect of t; on chloride diffusion
coefficient, De becomes more pronounced for concrete made with blended cements
and with increasing replacement for cementitious material.
c) In the present study, a time-dependent diffusion model incorporating the effect oft;
is proposed. This model is developed based on the results of chloride immersion
tests conducted in a laboratory-controlled environment. The chloride diffusion
coefficient at any particular time can be estimated if the w/c of the mix is known.
d) The empirical coefficient (m) derived from results obtained from in-service marine
structures (m = 0. 77) appears to be higher than those derived from laboratory
immersion (m =0.42) and field exposure (m =0.38) tests. A relatively large m value
implies that the reduction in De is relatively high. This could be attributed partly to
the stresses in the structure in service which impede chloride diffusivity into the
concrete.
Chapter 5 : Prediction of Chloride Concentration in Concrete Using a Mix Design ... 140
e) The mean surface chloride concentration (Cs) of 0.55% by weight of concrete seems
to be representative for normal strength OPC concrete ( ~ 50 MPa) in 2-5% NaCl
solution under laboratory test conditions.
g) At low w/c, the variation of concrete cover can significantly influence the service
life prediction of a concrete structure. However, the effect of cover variation on the
service life prediction diminishes as the w/c is increased.
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 141
Chapter 6
CHLORIDE DIFFUSION IN CONCRETE
UNDER UNIAXIAL COMPRESSION
6.1 INTRODUCTION
When reinforced concrete was first industrialised some 100 years ago, the pioneers were
convinced to have found a material that can provide a long lasting protection to steel
reinforcement. In spite of the good performance of concrete structures observed in
aggressive environments, there is also evidence of similar structures presenting serious
premature deterioration during its service life. In extreme cases, it may even lead to
structural failure (Hognestad, 1986). One type of deterioration that is widespread in a
marine environment is the chloride induced corrosion of reinforcement in concrete
structures. The deterioration process depends very much on the rate at which chloride
ions penetrate through the concrete cover and the subsequent chemical and physical
interactions with the concrete material.
It has been recognised that cracks in concrete may also affect the rate of chloride ingress
into concrete, and thus, its service life (Rostam, 1996). However, most of the service
life prediction models developed do not account for these effects, particularly
microcracking. Microcracks exist in concrete even before the application of load. They
are present randomly at the aggregate-mortar interface due to several reasons such as
bleeding characteristics, strength of the transition zone and curing history of the
concrete (Mehta and Monteiro, 1993). These microcracks have the potential of
becoming continuous and may grow under the application of load (Mehta and Gerwick,
1982). Studies have shown that bond cracks begin to propagate at compressive stresses
between 30% to 50% of the concrete compressive strength. Beyond 70% of the
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 142
concrete strength, cracks developed through the mortar and bridging between bond
cracks to form continuous pattern which eventually leads to failure (Ngab et al, 1981).
From an extensive literature review, it is found that information pertaining to the effect
of microcracks on the chloride transport into concrete is very scarce. Francois and Maso
(1988), and Francois et al (1995) investigated the influence of flexural loading and
direct tension on the penetration of chlorides in concrete. The effect of microcracks on
chloride ion intrusion in ultra-high strength composite reinforced concrete in flexure
was investigated by Aarup (1996). Jacobsen et al (1996) studied the effect of
microcracks due to freeze-thaw on chloride penetration using the chloride migration
test. Samaha and Hover (1992) and Saito and Ishimori (1995) observed the chloride
permeability through concrete specimens which had been loaded in compression to
stress levels of between 30% and 100% of the compressive strength. In the latter
studies, chloride permeability tests were conducted on the specimens after the load had
been completely removed. Loo (1992) and Wang et al (1997) pointed out that the
characteristics of the microcracks after unloading, as in the latter studies, were likely to
be different from those while under load.
Instantaneous stresses
r---t (0.20, 0.35 and 0.50 f c.) + microcrack
evaluation by non-destructive method
Chloride
Concrete Sustained stresses profiles
-... pnsms
.- periodically .
(0.20, 0.35 & 0.50 fc•)
Max 540 days
I
Unloaded control II
Grade 20 and 40 concrete mixes were designed using locally available coarse and fine
aggregates as described in section 3.3.1. Concrete mixes were prepared in accordance
with section 3.3.2. Details of the mix proportions are given in Table 3.3 of section
3.3.4.
Concrete cylinders (<1>75 x 150mm) and prisms (75 x 75 x 150mm) were cast. The
cylinders were used for microcrack evaluation and the prisms for chloride concentration
profile determination. A total of 180 specimens were cast.
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 144
In this experiment, specimen size of <j>75 x 150mm and 75 x 75 x 150mm were used.
These sizes were chosen due to the limitation of the load that can be applied using the
steel creep rig described in section 6.2.3. Since all 28-day compressive strengths were
based on <I> 100 x 200mm cylinders, it may be of interest to determine the effect of
specimen shape and size on the compressive strength. Table 6.1 shows a comparison of
the compressive strength results between different shapes and sizes of concrete
specimens. Each result is an average of three specimens. It is clear that the effect of
shape and size does not affect the compressive strength significantly.
Eighteen sets of galvanised steel creep rigs were designed and fabricated to apply
sustained stresses on to the concrete specimens, nine each for the cylinders and prisms.
A schematic sketch of the rig is shown in Figure 6.2.
For the concrete cylinders, eight demountable mechanical (demec) points were glued on
each specimen after removal from the moist-curing chamber. This is to enable the
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 145
monitoring of the progressive deformation of the specimen under sustained load and the
creep recovery at the end of the sustained period. The specimens were subsequently
coated with two layers of silicone rubber and further wrapped with plastic sheets to
prevent moisture loss during the sustained period. All gaps between concrete specimens
and steel plates were sealed with silicone rubber. For each sustained period, three stress
levels at 20%, 35% and 50% of the 28-day compressive strength were considered.
Steel rods
Top plate
Concrete specimen I
Concrete specimen 2
Sustained stresses were applied on to the specimens within 2 days after removal from
the moist-curing. This was done by placing a hydraulic jack on the top plate and jacking
against a temporary reaction plate, placed over the jack. When the required load was
reached, it was maintained by tightening the nuts on the rods. The loaded rigs were then
kept in a controlled room at 23 ± 2°C and 50± 5%RH. Control specimens (unstressed)
were also sealed in the same way and kept in the same controlled room for the same test
periods. These specimens were used for microcrack evaluation as shown in Figure 6.1.
Four additional cylinders were prepared to evaluate the effectiveness of the sealing
method. Two of them were sealed in the same way described above while the
remaining two left unsealed. These unstressed specimens were also kept in the same
controlled room.
On the other hand, the concrete prisms were also stressed in the same way as the
concrete cylinders. To simulate one-dimensional chloride diffusion into the specimen,
three of its vertical faces were coated with epoxy leaving only one vertical face (75 x
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 146
150mm) uncoated. Both top and bottom faces of the prism were also epoxy coated and
mortar capped. The prisms together with the rigs were then immersed in 3% NaCl
solution for 28, 90 and 540 days. Chloride immersion test was conducted in a separate
controlled room at the same controlled temperature and relative humidity levels. For
the control specimens, all except one 75 x 150mm face were coated with epoxy. They
were immersed in the same chloride solution tank as those in the rigs. These specimens
were used for chloride concentration determination as shown in Figure 6.1.
The loads on the concrete specimens were monitored regularly by measuring the
deformation of the dynamometer. A demec gauge with 100mm gauge length was used
for this purpose. The loss in the stress level was compensated by tightening the nuts on
the four rods and checking the readings on the dynamometer. The steel tube was
operating within a low stress range of about 20% of its yield strength, i.e., within the
elastic limit.
Several companion specimens (prisms and cylinders) were also loaded at the age of 28
days to the same stress levels using a compression testing machine. The load on the
specimen was removed immediately when the targeted stress level was reached. These
are called instantaneously loaded specimens. For the prisms, all except one 75 x
150mm face, were epoxy coated after the test. They were left in the same NaCl solution
tanks as the other specimens in the rigs for the same test periods. For the cylinders,
microcrack evaluation was carried during the compression test, using the non-
destructive method described in section 6.2.5. The characteristics of microcracks during
the loading and unloading cycle were determined.
In this study, two methods of microcrack evaluation are used namely, the microscopic
technique and the non-destructive method. The microscopy technique was used to
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 147
quantify the microcracks of the concrete specimens that have been sustained under load.
At the end of each sustained period, a horizontal slice of 10mm thick was cut from the
cylinder at mid-height using a diamond blade masonry saw. Microcrack evaluation of
the concrete disc under a microscope was carried out as described in section 4.2.4.
The non-destructive method developed by Loo (1992) was used to characterise the
microcrack behaviour when a concrete specimen was loaded instantaneously in a
compression test. The method is described in section 4.2.2. The onset of microcrack
propagation under uniaxial compressive load can be evaluated from this method. The
stress level corresponding to the onset of microcrack propagation is called the initiation
stress ( O'i) and it can be obtained as the stress when the Poisson' s ratio starts to increase.
The chloride concentration profile of the concrete specimen was carried out as described
in section 5.4.2.
The concrete cylinders, for microcrack evaluation, are sealed to prevent moisture loss.
It is to simulate as closely as possible the condition of the concrete prisms which are
immersed in sodium chloride solution where there is practically no moisture loss from
the concrete. Figures 6.3 and 6.4 show the shrinkage strain and the total weight loss of
the sealed unstressed cylinders, respectively. The shrinkage strain of the unsealed
unstressed cylinders is also plotted together for comparison.
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 148
-
I ll
.5
800..----------------------,
.......................... ·•· ...... ..
,.... --
700
f!
~ 600
•!:?
-&
E 500
•··* .
ea
400 I
.:.::
C 300
·;::
.c
Ill
Cl
.
C
·;.
C
100
0 50 100 150 200 250 300 350 400 450 500 550
Days
••• • •• Grade 20 ( unsealed) ••• o- .. Grade 20 ( sealed)
• Grade 40 (unsealed) ~ Grade 40 (sealed)
60..---------------------- ,
····· -•· .. . .............
-Ill
E
50
, .....................
•••. ·----------------11-----1
-f!
Cl
40
30
20
10
0 50 100 150 200 250 300 350 400 450 500 550
Days
The shrinkage strain of the sealed specimens is minimal compared to the unsealed
specimens. This observation is consistent with the measured weight loss of the
specimen. It is therefore concluded that the seal was effective in maintaining a virtually
constant moisture content within the specimens.
The axial deformation of the concrete cylinders under sustained load was monitored
throughout the sustained period to determined its creep characteristic. Figures 6.5 and
6.6 show the typical creep behaviour of Grades 20 and 40 sealed concrete after 90 days,
respectively. For comparison purposes, the creep measurement of unsealed concrete
was also carried out on Grade 20 concrete only, and the result is shown in Figure 6.7.
For sealed specimens, both Grades 20 and 40 concrete show an increasing creep
behaviour when the sustained stress is increased from 20% to 50% of the 28-day
compressive strength. The increase in creep is more pronounced in the Grade 20
concrete. Creep is inversely proportional to the strength of concrete at the time of
application of the load (Mindess and Young, 1981; Neville, 1988). The loss of
adsorbed water from the hydrated cement paste, under sustained load, is regarded as the
main cause of creep strains in concrete (Mehta and Monteiro, 1993). Creep has been
found to decrease with an increase in the size of the specimen. The size effect may be
influenced by the surface-volume ratio of the concrete member (Neville, 1988).
For the Grade 20 concrete, the magnitude of creep observed for unsealed specimens
(Figure 6.7) is much higher than that of the sealed specimens (Figure 6.5). This is due
to moisture losses from the unsealed specimens. The resulting drying shrinkage causes
the formation of additional microcracks in the transition zone (bond cracks) which
contribute to a further increase in the creep strains (Mehta and Monteiro, 1993).
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 150
1200
unloading
(I) 1000
C
·;...
-...
( I)
0
()
800
.E 600
.E
c. 400
...al
0 200
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Days
~20% -0-35% ~50%
1200
(I) 1000
·;...C
-...
( I)
0
()
800
.E 600
C
c. 400
Cl)
...
0
Cl)
200
2000
1800 +- unloading
(I)
C 1600
-......
"iii
(I)
0
u
1400
1200
.E 1000
.!: 800
Q. 600
Q)
...
0
Q)
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Days
~20% -0-35% ~50%
The characteristics of microcracks in the concrete during the instantaneous loading cycle
are shown in Figures 6.8 and 6.9. At about 20% stress level, there is virtually negligible
microcrack propagation. However, at 35% and 50% stress levels, an increase in the
specific crack area is observed which implies an increase in the microcrack area within
the concrete specimen during loading. This increase can be attributed to the progressive
enlargement of existing microcracks and the formation of additional bond cracks. A
quantitative distinction between those two is not possible using the non-destructive
method of microcrack evaluation.
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 152
.
=;
J:
0.5
0.4
0.5
0.4
0.5
0.4
Cl
C 0.3 0.3 0.3
-'
!
Ill
Ill
0.2 0.2 0.2
-
! 0.1
U)
0
0.1
0
0.1
0
-10 0 10 20 30 40 50 -10 0 10 20 30 40 50 -10 0 10 20 30 40 50
Sp. crack area Sp. crack area Sp. crack area
(microstrain) (m lcros train) (mlcrostrain)
>
0.2 0.2
From Figure 6.8, the specific crack area (during loading) at 35% stress level is 23
microstrain for Grade 20 concrete. At 50% stress level, it increases to 45 microstrain.
However, when unloaded completely from 35% and 50% stress levels, the recovery is
100% indicating that the microcracks have closed back completely. Similar trends are
also observed for Grade 40 concrete as shown in Figure 6.9 although the recovery from
50% stress level is not 100%. Nevertheless, there is a clear evident of microcracks
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 153
closing up when the load is removed. In the latter, the increase in microcracks at 35%
and 50% stresses is comparatively lower than Grade 20 concrete. The results of
microcrack evaluation in Figures 6.8 and 6.9 clearly show that microcracks, in terms of
specific crack areas, are in fact different during loading and when completely unloaded.
Data obtained from the microcrack examination of concrete slices are presented in
Table 6.2. Each result is obtained as an average of two specimens. The results of Grade
20 concrete after 90 days are not available because an attempt was made to characterise
the microcracks in those specimens using the non-destructive method (Loo, 1992).
However, it was found to be not successful. Subsequently, the microscopic technique of
microcrack evaluation was used for the remaining specimens.
Grade 20 Grade 40
Most of the cracks observed in this study are bond cracks as they are found to occur at
the aggregate-paste interface. Mortar cracks are negligible in most cases. The
characteristics of microcracks with respect to stress level are similar at all test periods.
Figure 6.10 shows a typical result after 90 days of sustained period for Grade 40
concrete.
600
-
E 500
--
E
.&:
Cl
400
Smadi (60d, <l> 102mm)
C
.!! 300
.:,e.
u
...
ea
u 200
,,
C
0
m 100
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Stress-strength ratio
did not observed any increase in microcracks up to 0.45 fc° except at 0.65 fc. for concrete
having a compressive strength between 33 MPa and 40 MPa. Smadi and Slate (1989)
found the bond crack length, in excess of that due to drying shrinkage, after 60 days of
sustained loading below 0.4 fc. to be small. In addition, they reported negligible mortar
cracks below 0.7 fc. sustained stresses. Similar observations were also reported by
Meyers et al (1969). He found significant increased in microcracking in all his sealed
specimens when sustained at stresses between 0.5 fc° and 0.7 fc .. At about 0.57 fc° and
after sustaining for 30 days, he found the crack length to increase by about 55% in
relation to his unloaded control.
Figure 6.11 clearly indicates that there is considerably less cracking in the Grade 40
concrete than in the Grade 20 concrete at all duration considered. The sealed specimens
do not permit moisture loss and, thus, additional microcrack formation due to shrinkage
may not be substantial, if any. The quantity of microcracks observed can be regarded as
caused by the effect of the sustained stresses and over the length of the sustained period.
Comparing the quantity of the microcracks for Grade 40 concrete, there is no evident of
a significant increase at all stress levels considered up to 540 days. The 540 days
considered in the present study is probably the longest sustained period ever
investigated. Ngab et al (1981 ), and Smadi and Slate (1989) carried out similar studies
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 156
for 60 days only. Loo (1995) conducted a series of short-term sustained loading tests
(about 30 minutes) on concrete cylinders in which the extent of microcracking in the
concrete was continuously monitored using a non-destructive method. They reported
negligible microcrack propagation below 0.5 fc.. The present findings show that when
concrete is sustained over a longer period of time from that carried out previously, a
negligible microcrack propagation is observed below 0.5 fc° stress level.
250
E
.§.
200
&===~=~--=~~~~~:::
-
.c
en
C
.!!
150 ~=========-------a
--- - - -x
~
u
f! 100
u
-
m
0
I- 50
0 - - - . . - - - - - . - - - - - . - - -.........- - - . - - - - - - 1
0 100 200 300 400 500 600
Days
- - X· - - G20(0%) - - 0- - - G20(20%) - - ½ - - G20(35%) - - ~ - - G20(50%)
-+- G40(0%) - - - - - G40(20%) A G40(35%) ~ G40(50%)
Table 6.3 shows the calculated apparent chloride diffusion coefficient (Da) of Grade 20
and 40 concretes after immersion in 3% NaCl solution for up to 540 days. These are
specimens which have been previously loaded instantaneously. The values of Da are
also normalised against the unstressed control specimen.
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 157
Apparent Chloride
Diffusion Coefficient, Normalised Da
Grade Stress Da X 10-12 m2/s
Level 28 90 540 28 90 540
days days days days days days
20 0% (control) 23.0 13.8 9.5 1.00 1.00 1.00
20% 19.3 11.8 10.6 0.84 0.85 1.12
35% 26.5 12.7 8.2 1.15 0.92 0.86
50% 25.3 14.3 8.5 1.10 1.04 0.89
40 0% (control) 13.7 5.9 2.6 1.00 1.00 1.00
20% 12.3 6.2 2.5 0.90 1.05 0.96
35% 11.1 6.2 2.8 0.86 1.05 1.08
50% 11.5 5.7 2.1 0.89 0.97 0.81
1.20
1.00
11:1
C 0.80
"C
Cl)
.!!! 0.60
iii
z
..
E
0 0.40
0.20
0.00
0 10 20 30 40 50 60
Instantaneous stress (%)
It can be seen from Figure 6.12 that Da values are not significantly affected when
concrete is instantaneously loaded up to 50% of its compressive strength. This could be
attributed to the closing back of microcracks upon unloading as shown in Figure 6.8 and
6.9. The normalised Da values do not exhibit a general trend to increase or decrease
with the level of stresses. Most of the normalised values are within a band of 0.8 and
1.2. It is clear that the Da of instantaneously loaded concrete is not significantly affected
up to 0.5 fc. stress level, when compared with the control.
Table 6.4 shows the calculated Da of Grades 20 and 40 concrete after immersion in 3%
NaCl solution for up to 540 days under three levels of sustained stresses. The Da values
are normalised against the unstressed control specimens. Figure 6.13 shows the
variation of the normalised Da at various sustained stress levels considered. For both
Grade 20 and 40 concretes, values of Da are found to reduce with increasing levels of
sustained stresses. The reduction in Da is more pronounced in Grade 20 concrete. For
each grade of concrete, the reduction in Da up to 540 days appear to be quite consistent.
The marginal difference between the normalised Da at 28, 90 and 540 days, for each
grade of concrete, can be attributed to the variability in the results. Experimental errors,
for example, chloride dust sampling and chloride content determination may contribute
to the difference. However, all the results show a similar trend, i.e., a gradual reduction
in the Da values with increasing level of sustained stresses. The average normalised Da
value (of 28, 90 and 540 days) for each grade of concrete is plotted against the sustained
stresses in Figure 6.14. At 50% sustained stress level, the average reduction in Da for
Grade 20 and 40 concrete are 47% and 30%, respectively.
Chapter 6 : Chloride Diffusion in Concrete Under Uniaxial Compression 159
Apparent Chloride
Diffusion Coefficient, Normalised Da
Grade Stress Da X 10-12 m2/s
Level 28 90 540 28 90 540
days days days days days days
20 0% (control) 23.0 13.8 9.5 1.00 1.00 1.00
20% 15.7 8.4 6.6 0.68 0.61 0.69
35% 15.8 7.6 6.3 0.68 0.55 0.66
50% 12.8 6.5 5.7 0.56 0.47 0.60
40 0% (control) 13.7 5.9 2.6 1.00 1.00 1.00
20% 10.3 4.7 1.9 0.75 0.80 0.73
35% 9.9 4.3 1.9 0.72 0.73 0.73
50% 9.4 4.5 1.7 0.69 0.76 0.65
1.00
0.80
a,
C
"C
G)
0.60
.!!?
iii
...0
E 0.40
z
0.20
0.00
0 10 20 30 40 50 60
Sustained stress (%)
1.00
I'll
C 0.80
"C
Cl)
.!?
iii 0.60
...0
E
C
Cl)
0.40
C)
...as
Cl)
0.20
~
0.00
0 10 20 30 40 50 60
Sustained stress (%)
Within the stress range considered, the Da of the concrete does not appear to be
influenced by the microcracks development in concrete under uniaxial compression. At
35% sustained stress level, the average Dais found to decrease when microcracks do not
appear to propagate. At 50% sustained stress level, a further reduction in the average Da
is observed although microcracks have been found to propagate. There is an increase in
the quantity of microcracks at 50% than at 30% stress level. On the other hand, the
compressive stress is observed to affect Da of the concrete. There are a few possible
reasons for this observation.
around coarse aggregates. Also, negligible mortar cracks are found at this stress level,
as shown in Table 6.2.
Since diffusion takes place primarily through the mortar, it appears that the chloride
diffusivity is not adversely affected when the crack patterns in the concrete are not
continuous, i.e., bond cracks are not interconnected with mortar cracks. Bridging
between bond and mortar cracks to form continuous crack pattern occurs at compressive
stresses beyond 70% of the ultimate strength (Hsu et al, 1963; Krishnaswamy, 1968).
The stress corresponding to the bridging of cracks is called the critical stress. At this
stress level, concrete undergoes a volumetric expansion while under compression. It
was observed in Chapter 4 that the measured electrical charge passed through a concrete
increased substantially when the critical stress in the concrete was exceeded during the
compression test. The finding here might suggest the influence of critical stress
occurrence on the chloride diffusivity of concrete subjected to uniaxial compression.
However, compressive stress rarely exceeds 0.5 fc. in concrete structures at service loads
(Gilbert, 1988).
On the other hand, the compressive stress is found to impede chloride diffusion into the
concrete. A possible explanation for this would be the volumetric contraction of the
concrete in compression resulting in a reduced porosity. Diffusion in concrete is much
slower the smaller the pore size due to the greater tortuosity of the path an ion has to
follow (Kumar et al, 1987). In addition, the compressive load may partially close some
microcracks that exist in the horizontal plane of the specimen, which are in the direction
of diffusion. Lower chloride contents at the crack location of prestressed specimens
than that of the non-prestressed specimens had also been reported (Poston et al, 1987).
Piasta and Schneider (1992) have carried out experimental studies to determine the
effect of sulphate attack on concrete under sustained uniaxial compression. They found
that sulphate attacked was retarded when the concrete was subjected to sustained
stresses between 20% and 35% of the compressive strength.
Chapter 6: Chloride Diffusion in Concrete Under Uniaxial Compression 162
6.4 CONCLUSIONS
a) When concrete is loaded up to 50% stress level, an increase in the specific crack
area is observed. However, when concrete is completely unloaded, microcracks
appear to have closed back completely. Hence, the characteristics of microcracks
under uniaxial compression are different when a concrete is loaded to a stress level
and when it is completely unloaded.
b) The apparent chloride diffusion coefficient (Da) of a concrete which has been loaded
instantaneously to 50% stress level is not affected by microcrack development. This
is attributed to the complete closure of the microcracks in the concrete when the
load is completely removed.
d) The compressive stress impedes chloride diffusion into the concrete. A possible
explanation for this would be the volumetric contraction of the concrete in
compression resulting in a reduced porosity. In addition, the compressive load may
partially close some microcracks that exist in the horizontal plane of the specimen,
which are in the direction of diffusion.
e) At 50% sustained stress level, the average reduction in Da for Grade 20 and 40
concretes are 47% and 30%, respectively.
Chapter 7: Microcracking of Concrete Under Sustained Compression 163
Chapter7
MICROCRACKING OF CONCRETE UNDER
SUSTAINED COMPRESSION
7.1 INTRODUCTION
However, the increase was significant at stresses above 0.7 fc.. Their results are
consistent with the microcrack observation in the present study using the same method
of characterisation. This has been discussed in Chapter 6.
The present study attempts to augment further information on the long-term behaviour
of microcracks in sealed concrete cylinders using the non-destructive method. The test
is also carried out with the aim to verify the characteristics of microcracks observed in
Chapter 7 : Microcracking of Concrete Under Sustained Compression 164
Chapter 6 which were determined using the destructive method, i.e., by cutting a slice
out from a concrete cylinder after the compression test and examining it under an optical
microscope.
The mix proportion, materials used and the preparation of specimens are described in
section 4.2.1. A total of 18 specimens were cast.
A specially designed steel creep rig was used to apply sustained compressive stresses on
the concrete cylinders. The creep rig was fabricated from high strength boiler plates
(fy = 460 MPa) and Macalloy bars (fy = 1080 MPa). The set-up is different from the one
described in Chapter 6 in that load application on the test specimens including all strain
measurements are fully automated in this set-up. A Riken motor-driven hydraulic pump
with a capacity of 550 kN was used to apply the load. A digital pressure meter (Type
DPS-700S) was installed to monitor the pressure in the pump. The pump can generate
up to 700 kgf/cm 2 of pressure. The pressure meter operates on a semi-conductor strain
gauge type pressure transducer. The load on the concrete test cylinders was sustained at
± 3% of the predetermined stress level. Three different stress levels at 30%, 50% and
65% of the 28-day compressive strength were considered.
A steel ball and a hemispherical seat were incorporated at the top and bottom of the set-
up, respectively, to ensure vertical loading on the test cylinders. All cylinders were
sulphur-capped. They were sealed with two layers of silicon rubber to prevent moisture
loss. Two test cylinders were placed in the set-up and sandwiched between two packers.
On each test cylinder, two transverse electrical resistance strain gauges were installed,
evenly spaced at mid-height. The strain gauges were of the Showa Nl 1-FA-60-120-11
Chapter 7: Microcracking of Concrete Under Sustained Compression 165
type with a gauge length of 60mm and a gauge factor of 2.13 ± 1%. Dow Corning
silicon rubber was applied over the strain gauges. A compressometer consisting of two
vertically arranged displacement transducers, with a sensitivity of 1OOOx10-6 mm/mm,
was attached directly on to the central region of each test cylinder. The compressometer
has a gauge length of 100mm. The electrical resistance strain gauge and the
compressometer are used to measure the lateral and axial deformation of the cylinder,
respectively.
Strain data were continuously logged at every 15 seconds for the first hour and at every
10 minutes for the subsequent hours. The test was carried out in a controlled
environment room at 23 ± 2°C and 50± 5% RH. A complete test set-up is shown in
Figures A12 - Al4 of Appendix A.
The cylinders were loaded progressively to the predetermined stress level. The stress
level was sustained for 7 days. Both the axial and transverse strains were recorded
throughout the test. At the end of the sustained period, the load was gradually reduced
until it was completely removed from the specimen. Recording of the strain data were
continued for several days until the creep recovery was almost negligible.
In addition to the strain measurements of the specimens in the test creep rig, strain data
from an unloaded control specimen was also measured. The control specimen was
prepared in the same way as the test specimens, with strain gauges and a
compressometer attached to it. It was placed on a bench, next to the creep rig, in the
control environment room. A slight drift in the strain measurements is observed from
Chapter 7: Microcracking of Concrete Under Sustained Compression 166
the control specimen. All strain data from the test specimens are adjusted to account for
the drift.
The non-destructive method of crack evaluation requires that the elastic Poisson's ratio
(µe) be determined fairly accurately. The elastic Poisson's ratio is determined in the
lower stress region of the stress-strain curves. It is obtained by calculating the ratio of
the gradients of the initial linear portions of the respective stress-strain curves given as
(Loo, 1992);
(7.1)
where gy = gradient of the linear portion in the stress-axial strain curve, and
Ki = gradient of the linear portion of the stress-transverse strain curve. Values of elastic
Poisson' s ratio obtained from the concrete test cylinders are between 0.15 and 0.21.
-
.E
180
-.
f!
Ill
0
u
160
140
120
-.E.
m
C'II
100
80
60
.
~
u
C'II 40
u
u 20
=i
c.
0
C/J -20
0 10 20 30 40 50 60 70 80
Sustained duration (hour)
The specific crack areas of the concrete under sustained uniaxial compression are
computed from the measured strain data using equation (4.2) in Chapter 4. Figure 7.1
shows the variation of the specific crack area against sustained duration. At each stress
level, the result is an average of two specimens. Although the test was conducted for 7
days, the strain gauge readings started to drop after 3 days of test.
In spite of repeated tests, this was observed in most test cylinders. On the other hand,
the displacement transducers showed stable and consistent readings throughout the test.
As a result, the evaluation of specific crack area was carried out using strain data from
the first 3 days of test.
From Figure 7 .1, it can be seen that there is a small fluctuation in the specific crack area
over the sustained duration. This fluctuation is attributed to the sensitivity of different
strain measuring devices used in the test. Barring this fluctuation, a clear trend in the
variation of specific crack area with time is observed. Within the stress levels
considered, an initial microcrack propagation (in terms of specific crack area) is
observed when the load is initially applied to the concrete. However, when the stresses
are sustained for 3 days, no further crack propagation with time is observed for concrete
sustained at 0.3/c. and 0.5/c·· This observation is consistent with the microcrack
evaluation based on total crack length measurements (destructive method) for sustained
sealed specimens which are shown in Figure 6.11 in Chapter 6. From Figure 7 .1, the
specific crack areas fluctuate within a mean value of about 5 microstrain at 30% stress
level and about 30 microstrain at 50% stress level. However, at 65% stress level, a
gradual increase in the specific crack area is observed with time. The specific crack
area at the start of the sustained period is 126 microstrain but it gradually increases to
156 microstrain after 3 days, an increase of 25%.
Unlike the destructive method, the advantage of using non-destructive method is that
crack evaluation is carried out on the same test cylinder over the length of sustained
period. This can eliminate sample variability which may influence microcrack
evaluation. The results of microcrack evaluation from the non-destructive method
further verify the observations made from the destructive method for sealed concrete
sustained up to 0.5 fc. stress level.
Chapter 7: Microcracking of Concrete Under Sustained Compression 168
have been sustained at 0.30, 0.50 and 0.65 fc' stress levels for 3 days.
60
.I
ea
~
50
(.)
40
ea -C
._
·-.
(.)
(.)
-.
ea
·-:!:: g
(.)
30
20
--.
G) (.)
c.,- 10
Ill E
0 0
c.
G)
G) -10 30%
0
-20
0 50 100 150 200 250 300 350 400 450
Axial creep (microstrain)
The creep of specific crack area is defined as the increase in the specific crack area in
the concrete under a constant stress. In spite of some fluctuation in the values of
specific crack area, a clear trend can be identified. It clearly shows the stress level at
which microcracks propagate under sustained compression. It can be seen that
microcracks do not propagate with time at stress level 0.5 fc. and below, where the lines
are almost horizontal, i.e., lying close to the axial creep axis. However at 0.65 fc., the
gradient of the line generally shows an increase with increasing axial creep, indicating
crack propagation under sustained compression. This observation is consistent with the
Chapter 7: Microcracking of Concrete Under Sustained Compression 169
findings reported by Loo (1995) although his tests were conducted in a much shorter
period than the present study.
The results from the present study and Loo ( 1995) show that the length of sustained
period does not appear to affect the relationship between crack propagation and axial
creep at stress level below 0.5 fc .. Hence, the results of the non-destructive tests support
the observations made from the destructive method (Chapter 6) regarding microcrack
propagation in sealed concrete when sustained up to 0.5 fc ..
7.4 CONCLUSIONS
b) Under sustained compression, microcracks do not propagate with time at stress level
0.5 fc. and below. However, at 0.65 fc', a progressive increase in the specific crack
area is observed with time.
Chapter 7: Microcracking of Concrete Under Sustained Compression 170
c) The results of the non-destructive tests support the observations made from the
destructive method regarding microcrack propagation in sealed concrete when
sustained up to 0.5 fc' stress level.
Chapter 8 : Predicting Chloride Concentration in Concrete... 171
Chapter 8
PREDICTING CHLORIDE CONCENTRATION IN
CONCRETE UNDER A SUSTAINED STRESS
8.1 INTRODUCTION
It has been shown and discussed in Chapter 6 that when a concrete is sustained in
compression up to 0.5 fc. stress level, microcracking in concrete does not affect chloride
diffusivity in spite of the initial microcrack propagation due to the application of load.
The present study concludes that microcracking in concrete under uniaxial compression
(< 0.5 fc.) cannot adversely affect the service life prediction of a structure. On the other
hand, compressive stresses are found to impede chloride diffusion into concrete. In this
chapter, an approach is proposed for predicting a time-dependent De of a concrete under
uniaxial compression. It is a modification to the diffusion model proposed in the
present study for an unstressed concrete, discussed in Chapter 5.
The proposed chloride prediction model, given as equation (5.19) in Chapter 5, is used
again to predict the chloride concentration profiles of the unloaded control specimens
after 540 days of immersion in 3% NaCl solution. The control specimens were
immersed in the same tanks as the specimens under sustained stresses.
Chapter 8: Predicting Chloride Concentration in Concrete ... 172
0.70
-
0.60
c -
0 Cl)
·-...
-
-C'O
C
Cl)
(,)
Cl)
...
(,)
C
0
(,)
0.50
0.40
so.
(,) 0.30
Cl) 3
:2 >,
0 .c 0.20
::c
o-
';I.
0.10
0.00
0 5 10 15 20 25 30 35
Depth (mm)
c_
0
·- -
-C'O
-= (,)
C C
Cl)
Cl)
...
Cl) 0
(,) (,)
c
o
-
0
(,) .
,, 3
Cl)
·-
0 .c>,
::c ';I.
o-
0 5 10 15 20 25 30 35 40
Depth (mm)
Figures 8.1 and 8.2 show that the chloride concentration profiles of the control
specimens can be reasonably predicted using the proposed model. For Grade 20
concrete, the measured and predicted De from equation (5.14) are 9.5x10· 12 m2/s and
12.2xlff 12 m2/s, respectively. For Grade 40 concrete, the measured and predicted De are
2.6x10· 12 m2/s and 4.0x10· 12 m2/s, respectively. The comparison between measured and
predicted chloride profiles shown in Figures 8.1 and 8.2 further validates equation
(5.19) for predicting the chloride profile in an unstressed OPC concrete. In the
following section, the effect of compressive stresses on De is considered and an
approach for predicting the chloride profile in concrete subjected to a compressive stress
is proposed.
coefficient of a concrete under a compressive stress, O', and De O refers to the chloride
fc De.CJ D
- - 1-~
Grade fc' Dco Dc,o
28 d 90d 540 d 28 d 90d 540 d
20 0 1.00 1.00 1.00 0 0 0
0.20 0.68 0.61 0.69 0.32 0.39 0.31
0.35 0.68 0.55 0.66 0.32 0.45 0.34
0.50 0.56 0.47 0.60 0.44 0.53 0.40
40 0 1.00 1.00 1.00 0 0 0
0.20 0.75 0.80 0.73 0.25 0.20 0.27
0.35 0.72 0.73 0.73 0.28 0.27 0.27
0.50 0.69 0.76 0.65 0.31 0.24 0.35
The second last column of the table is the normalised De (against unloaded control) at
28, 90 and 540 days. The last column of the table denotes the amount of reduction in
the chloride diffusion coefficient of a concrete under a stress when compared with
unloaded control specimen of the same age.
It is observed that for a given stress level, the amount of reduction in De.CJ remains
constant with time for both grades of concrete. There is no clear trend of (1- De.CJ )
Dc,o
either increasing or decreasing with time. This is shown in Figures 8.3 and 8.4 for
Grade 20 and 40 concrete, respectively. Hence, it can be concluded that the amount of
reduction in De.CJ is not time-dependent.
D
1-~
Dc,o
0.6
t:.
0.5
t:. a
0.4 0 t:.
a a
0.3 0
0.2
0.1
0-1-----.----.-----.---.......--........- -
0 100 200 300 400 500 600
Days
020% a35% t:. 50%
Grade 20 Concrete
Chapter 8 : Predicting Chloride Concentration in Concrete... 175
D
1-~
Dco
0.4 - - - - - - - - - - - - - - - - - - - ,
6
0.3 6
a a a
<> 6
0.2 <>
0.1
0 +------.---,---,------r----.-------i
0 100 200 300 400 500 600
Days
1 020% a35% 1150%
Grade 40 Concrete
D
1-~
Dc,o R2 = 0.91
0.6 . . . - - - - - - - - - - - - - - - - - - ,
<>
0.5
0.4 <>
0.3
0.2
0.1
0 0------.---,---,------.----.-------1
0 0.1 0.2 0.3 0.4 0.5 0.6
D
1-~
De,o R2 =0.94
0.4 . . - - - - - - - - - - - - - - - - - ,
a
0.3 ---...:.a
a
0.2
0.1
0 ~-~-------,---..-------
0 0.1 0.2 0.3 0.4 0.5 0.6
In contrast, the amount of reduction in De.a is observed to increase as the stress level in
the concrete is increased from 0.2 fe. to 0.5 fc'. Figures 8.5 and 8.6 show the amount of
reduction in De.a with sustained stress levels for Grade 20 and 40 concretes, respectively.
concrete. The regression is carried out through all the results at 28, 90 and 540 days
since it has been shown that the amount of reduction in De.a is not time-dependent.
( 1- ~e,a
e,o
J= -1.97( fefe, ] 2
+ 1.8.J fe,
\.fe
J (8.1)
Chapter 8 : Predicting Chloride Concentration in Concrete ... 177
thus,
(8.2)
where (;:-}s the compressive stress-strength ratio. Similarly, for Grade 40 concrete,
(8.3)
The resulting tenn in brackets [ ] is essentially a reduction factor. The value of De,o in
both equations (8.2) and (8.3) can be estimated from equation (5.14) in Chapter 5.
Equations (8.2) and (8.3) are derived for OPC concrete mixes considered in the present
study having a wle ~ 0. 60 and 0.46, respectively and for ( ;:- ) ,;; 0.5 . It can be seen that
the reduction factor is also dependent on the grade of concrete. This is depicted in
Figure 8.7. The amount of reduction in De,cJor Grade 20 concrete is comparatively
higher than Grade 40 concrete. The difference between the two curves becomes wider
with increasing compressive stress-strength ratio. It shows that when concrete strength
is lowered, the amount of reduction in De.CJ increases with increasing fe / fe. but non-
linearly.
A single model for De.CJ that is applicable to a range of concrete grades, for example up
D
1-~ o Grade 20 (w /c=0.60)
Dc,o • Grade 40 (w /c=0.46)
0.6 ..-------.....::::==============;~
0
0.5
••
•
0.5 0.6
Tables 8.2 and 8.3 show the comparison between the measured Dc,cr and the predicted
Dc,cr for OPC concrete having a w/c=0.60 (Grade 20) and 0.46 (Grade 40), respectively.
Table 8.2 A Comparison between Measured and Predicted Dc,cr for Grade 20
Table 8.3 A Comparison between Measured and Predicted Dc,cr for Grade 40
The models appear to be conservative as the predicted Dc,cr is generally higher than the
measured value. A better Dc.cr prediction is observed for Grade 40 concrete. The poorly
predicted Dc.cr for Grade 20 concrete is due to the large difference between the
measured D2s (23.0 x 10· 12 m2/s) and the predicted D2s (41.2 x 10· 12 m2/s) when
fc / fc' = 0. This discrepancy could be due to some experimental errors. The Grade 20
specimens for 28 days and 90 days chloride immersion tests were cast from the same
batch of concrete. It was the first of the several batches of concrete cast in this study.
Some undue errors were encountered initially, for example, in the batching, casting,
chloride powder sampling and chloride concentration determination. These problems
were ironed out in subsequent works.
The prediction of chloride concentration profile is carried out using the following
equation (8.4 ). It is similar to the proposed model for an unstressed concrete, equation
(5.19), by replacing the terrnD2s(t;.J + term withDC,(j.
Chapter 8: Predicting Chloride Concentration in Concrete... 180
The term Dc.(5 in equation (8.4) can be computed from equation (8.2) for w/c=0.60 and
(8.4)
days (18 months) of chloride immersion tests are shown in Figures 8.8 and 8.9. Other
profiles at 0.2 fc. and 0.35 fc. are shown in Appendix B. The predicted profiles for
both grades of concrete compare reasonably good with the measured best-fits.
0.7
c
0 - Cl)
0.6
-.
:;::
ea ai
._
u 0.5
C C
Cl) 0
u u 0.4
so
u . 0.3
Cl) i
~ >- 0.2
0 Jl
::c ';I. 0.1
o-
0
0 5 10 15 20 25 30 35
Depth (mm)
0.6
c-
0 Cl) 0.5
; Q)
C'II ..
.. u 0.4
-C C
0
~ u
so
u .
0.3 "" ~•
Cl)
3:! >-
5 Jl
::c #-
i 0.2
"' ~ '---..... ....._
o-
0.1
0
--.
0 5 10 15 20 25
Depth (mm)
8.5 CONCLUSIONS
b) The reduction in De.a for Grade 20 concrete is found to be comparatively higher than
c) An approach is proposed for predicting De.a for OPC concrete having a w/c between
0.46 and 0.60. It is a modification to the diffusion model proposed in the present
study for an unstressed OPC concrete.
Chapter 8 : Predicting Chloride Concentration in Concrete... 182
d) The predicted Dc.cr for Grade 40 concrete compares reasonably good with the
measured results. The poorly predicted Dc,cr for Grade 20 concrete, at 28 and 90
days, is due to the large difference between the measured and the predicted D 28
when f c/ fc. = 0. This discrepancy is attributed to some undue errors at the initial
stage of the experimental work.
Chapter 9 : Conclusions and Recommendations 183
Chapter9
CONCLUSIONS AND RECOMMENDATIONS
9.1 CONCLUSIONS
1) The crack width/cover ratio, Wc/C, can be a suitable parameter to assess the
durability performance of a cracked reinforced concrete.
3) AS 3600 does not give any guidance on the allowable crack width for reinforced
concrete structures at serviceability, except for the 'deemed to comply' rules.
However, cracks may develop even if the 'deemed to comply' rules are satisfied
especially when there is a tendency now to use higher yield strength steel bars
(t;,=500 MPa) as reinforcement for concrete structures in Australia. From the
viewpoint of durability, a crack width limitation in AS 3600 is necessary in
addition to the cover thickness, to minimise the W c/C of a cracked reinforced
concrete.
steel bar is doubled in the prism in flexure. However, a marginal increase in the
Da is observed in the tension zone of the prism.
5) In specimens reinforced with steel bars in flexure, the Da value in the tension
zone is found to be relatively higher than in the compression zone. This may be
attributed to the damage at the aggregate-paste interface in the tension zone
which can expedite diffusion process. In contrast, the compressive stresses in
concrete impede chloride diffusion which can be attributed to the reduction in
the porosity of the concrete.
0.5 fc. stress level, the specific crack area recovery is 100% implying that
microcracks close back completely. However, when unloaded between
0.7 fc. and 0.95 fc. stress level, some residual specific crack areas are observed
immediately after complete unloading. This implies only a partial closure of the
microcracks.
3) In the present study, the critical stress is found to be exceeded when the concrete
is loaded to a stress level between 0.80 fc. and 0.95 fc.. However, in some
specimens the critical stresses were not exceeded when tested at 0.80 fc° and
0.85 fc' •
Chapter 9 : Conclusions and Recommendations 185
1) The initial curing time of concrete (t;) affects the short-term prediction of
chloride concentration significantly. The effect of t; on chloride diffusion
coefficient, De becomes more pronounced for concrete made with blended
cements and with increasing replacement for cementitious material.
2) When t;-5: 6 months, the effect on De is significant when the duration of chloride
exposure is less than 5 years.
4) The empirical coefficient (m) derived from results obtained from in-service
marine structures (m = 0.77) appears to be higher than those derived from
laboratory immersion (m = 0.42) and field exposure (m = 0.38) tests. A
relatively large m value implies that the reduction in De is relatively high. This
Chapter 9 : Conclusions and Recommendations 186
could be attributed partly to the stresses in the structure in service which impede
chloride diffusivity into the concrete.
7) At low w/c, the variation of concrete cover will significantly influence the
service life prediction of a concrete structure. However, the effect of cover
variation on the service life prediction diminishes as the w/c is increased.
2) The apparent chloride diffusion coefficient (Da) of a concrete which has been
loaded instantaneously to 50% stress level is not affected by microcrack
development. This is attributed to the complete closure of the microcracks in the
concrete when the load is completely removed.
diffusivity is not adversely affected when the crack patterns in the concrete are
not continuous, i.e., the bond cracks not interconnecting with the mortar cracks.
4) The compressive stress impedes chloride diffusion into the concrete. A possible
explanation for this would be the volumetric contraction of the concrete in
compression resulting in a reduced porosity. In addition, the compressive load
may partially close some microcracks that exist in the horizontal plane of the
specimen, which are in the direction of diffusion.
5) At 50% sustained stress level, the average reduction in Da for Grade 20 and 40
concretes are 47% and 30%, respectively.
3) The results of non-destructive tests support the observations made from the
destructive method regarding microcrack propagation in sealed concrete when
sustained up to 0.5 fc° stress level.
3) An approach is proposed for predicting Dc,a for OPC concretes having a w/c
4) The predicted De.a for Grade 40 concrete compares reasonably good with the
measured results. The poorly predicted De.a for Grade 20 concrete, at 28 and 90
days, is due to the large difference between the measured and the predicted D 2 s
when fc / fc. = 0. This discrepancy is attributed to some undue errors at the
initial stage of the experimental work.
Despite an extensive investigation planned and executed in this study, many other test
parameters which may be relevant and of practical significance could not have been
considered in the present test programme. With the availability of more test data and
test parameters investigated, the accuracy of the present model can be further improved.
It is therefore recommended that further works shall be carried out in the following
areas;
1) More experimental data are needed to improve the relationship between Wc;rlC
and the chloride threshold level.
REFERENCES
ACI Committee 222 (1985), 'Corrosion of Metals in Concrete', AC/ Journal, Proc. 82
(1), pp. 3-32.
ACI 318-89 (Rev 1992), 'Building Code Requirements for Reinforced Concrete',
American Concrete Institute, Detroit.
Alexander, M.G. and Streicher, P.E. (1997), 'Correlations Between Rapid Chloride
Test Results and On-Site Chloride Ingress', Proceedings of the 4th CANMETIACI
International Conference on Durability of Concrete, Sydney, SP170, Vol. 1, pp. 473-
490.
Andrade, C., Alonso, M.C. and Gonzalez, J.A. (1990), 'An Initial Effort to Use
Corrosion Rates Measurements for Estimating Rebar Durability', Corrosion Rates of
Steel in Concrete, ASTM STP 1065, Berke, Chacher and Whiting, pp. 29-37.
Arsenault, J., Bigas, J.P. and Ollivier, J.P. (1995), 'Determination of Chloride
Diffusion Coefficient Using Two Different Steady State Methods: Influence of
Concentration Gradient', Penetration of Chlorides into Concrete - Proceedings of the
International Workshop, St.Remy-les-Chevreuse, France, Edited by L.O. Nilsson and
J.P. Ollivier, pp. 150-160.
References 191
Arya, C., Buenfeld, N.R. and Newman, J.B. (1990), 'Factors Influencing Chloride-
Binding in Concrete', Cement and Concrete Research, Vol. 20, No. 2, pp. 291-300.
ASTM C1202 (1997), 'Standard Test Method for Electrical Indication of Concrete's
Ability to Resist Chloride Ion Penetration', American Society for Testing and
Materials, Philadelphia.
ASTM E632-81 (1981), 'Standard Practice for Developing Accelerated Tests to Aid
Prediction of the Service Life of Building Components and Materials', American
Standards and Testing of Materials, Philadelphia.
Bamforth, P.B. (1999), 'The Derivation of Input Data for Modelling Chloride Ingress
from Eight-Year UK Coastal Exposure Trials', Magazine of Concrete Research, Vol.
51, No. 2, pp. 87-96.
Bamforth, P.B. and Price, W.F. (1993), 'Factors Influencing Chloride Ingress into
Marine Structures', Concrete 2000, Proceedings of the International Conference at the
University of Dundee, Vol. 2, pp. 1105-1118.
References 192
Beeby, A.W. (1978), 'Cracking and Corrosion', Concrete in the Oceans, Technical
Report No. I, Cement and Concrete Association, United Kingdom.
Beeby, A.W. (1978), 'Cracking : What are Crack Width Limits for?', Concrete,
Vol. 12, No. 7, pp. 31-33.
Beeby, A.W. (1978), 'Corrosion of Reinforcing Steel in Concrete and its Relation to
Cracking', The Structural Engineer, Vol. 56A, No. 3, pp. 77-81.
Bentur, A., Gray, RJ. and Mindess, S. (1986), 'Cracking and Pull-Out Process in
Fiber Reinforced Cementitious Materials', Developments in Fiber Reinforced Cement
and Concrete, Vol. 2, Edited by Swamy, R.N., Wagstaffe, R.L. and Oakley, D.R.,
RILEM, Sheffield, England.
Bentz, D.P., Clifton, J.R. and Snyder, K.A. (1996), 'Predicting Service Life of
Chloride-Exposed Steel Reinforced Concrete', Concrete International, Vol. 18, No.
12, pp. 42-47.
Bjegovic, D., Krstic, V., Mikulic, D. and Ukrainczyk, V. (1995), 'Cdct Diagrams
for Practical Design of Concrete Durability Parameters', Cement and Concrete
Research, Vol. 25, No. 1, pp. 187-196.
Browne, R.D. (1982), 'Design Predictions of the Life for Reinforced Concrete in
Marine and Other Chloride Environments', Durability of Building Materials, Vol. I,
No. 2, pp. 113-125.
Buenfeld, N.R. and Newman, J.B. (1987), 'Examination of Three Methods for
Studying Ion Diffusion in Cement Pastes, Mortars and Concretes', Materials and
Structures, Vol. 20, pp. 3-10.
Cao, H.T., Moorehead, D. and Potter, R.J. (1999), 'Prediction of Service Life of
Reinforced Concrete Structures in Marine Environment and AS 3600', Proceedings of
the Concrete Institute of Australia 1 gth Biennial Conference, Sydney, pp. 131-137.
Chen, Y. And Odler, I. (1992), 'On the Origin of Portland Cement Setting', Cement
and Concrete Research, Vol. 22, pp. 1130-1140.
Chin, M.S., Mansur, M.A. and Wee, T.H. (1997), 'Effects of Shape, Size and
Casting Direction of Specimens on Stress-Strain Curves of High-Strength Concrete',
AC/ Materials Journal, Vol. 94, No. 3, pp. 209-219.
Clifton, J.R. (1993), 'Predicting the Service Life of Concrete', AC/ Materials Journal,
Vol. 90, No. 6, pp. 611-617.
Collins, F.G. and Grace, W.R. (1997), 'Specifications and Testing for Corrosion
Durability of Marine Concrete: The Australian Perspective', Proceedings 4th CANMET
I AC/ International Conference on Durability of Concrete, Sydney, SPI 70, Vol. 2, pp.
757-775.
Dhir, R.K., Jones, M.R. and Ahmed, H.E.H. (1990), 'Determination of Total and
Soluble Chlorides in Concrete', Cement and Concrete Research, Vol. 20, No. 4, pp.
579-590.
Dhir, R.K., Jones, M.R. and Ahmed, H.E.H. (1991), 'Concrete Durability:
Estimation of Chloride Concentration During Design Life', Magazine of Concrete
Research, Vol. 43, No. 154, pp. 37-44.
Dhir, R.K., Jones, M.R. and McCarthy, M.J. (1994), 'PFA Concrete: Chloride-
Induced Reinforcement Corrosion', Magazine of Concrete Research, Vol. 46, No. 169,
pp. 269-277.
Dhir, R.K., EI. Mohr, M.A.K. and Dyer, T.D. (1996), 'Chloride Binding in GGBS
Concrete', Cement and Concrete Research, Vol. 26, No. 12, pp. 1767-1773.
Dhir, R.K., Jones, M.R. and Ng, S.L.D. (1998), 'Prediction of Total Chloride
Content Profile and Concentration / Time-Dependent Diffusion Coefficients for
Concrete', Magazine of Concrete Research, Vol. 50, No. 1, pp. 37-48.
Francois, R., Konin, A. and Lasvaladas, I. (1995), 'Influence of the Loading on the
Penetration of Chlorides in Reinforced Concrete', Penetration of Chlorides into
Concrete - Proceedings of the International Workshop, St.Remy-les-Chevreuse, France,
Edited by L.O. Nilsson and J.P. Ollivier, pp. 250-264.
Garboczi, E.J. and Bentz, D.P. (1991), 'Digital Simulation of the Aggregate Paste
Interfacial Zone in Concrete', Journal of Materials Research, Vol. 6, No. 1, pp. 196-
201.
Gjorv, O.E. and Vennesland, 0. (1979), 'Diffusion of Chloride Ions from Seawater
into Concrete', Cement and Concrete Research, Vol. 9, No. 2, pp. 229-238.
Gjorv, O.E., KeFeng Tan and Min-Hong Zhang (1994), 'Diffusivity of Chlorides
from Seawater into High-Strength Lightweight Concrete', AC/ Materials Journal, Vol.
91, No. 5, pp. 447-452.
Glass, G.K. and Buenfeld, N.R. (1995), 'Chloride Threshold Levels for Corrosion
Induced Deterioration of Steel in Concrete', Proceedings Int. RILEM Workshop on
Chloride Penetration into Concrete, Edited by Nilsson and Ollivier, pp.429-440.
Glass, G.K and Buenfeld, N.R. (1997), 'The Presentation of the Chloride Threshold
Level for Corrosion of Steel in Concrete', Corrosion Science, Vol. 39, No. 5, pp. 1001-
1013.
References 196
Gowripalan, N., Cabreira, J.G., Cusens, A.R. and Wainwright, PJ. (1990),
'Effects of Curing on Durability', AC/ Concrete International, Vol. 12, No. 2, pp.47-
54.
Higgins, D.D. (1995), 'The Effect of Some Test Variables on Chloride Profiles',
Penetration of Chlorides into Concrete - Proceedings of the International Workshop,
St.Remy-les-Chevreuse, France, Edited by L.O. Nilsson and J.P. Ollivier, pp. 234-242.
Hoar, T.P. and Jacob, W.R. (1967), 'Breakdown of Passivity of Stainless Steel by
Halide Ions', Nature, No. 216, pp. 1299-1301.
Hooton, R.D. and McGrath, P.F. (1995), 'Issues Related to Recent Developments in
Service Life Specifications for Concrete Structures', Penetration of Chlorides into
Concrete - Proceedings of the International Workshop, St.Remy-les-Chevreuse, France,
Edited by L.O. Nilsson and J.P. Ollivier, pp. 388-397.
Hsu, T.T.C., Slate, F.O., Sturman, G.M. and Winter, G. (1963), 'Microcracking of
Plain Concrete and the Shape of the Stress-Strain Curve', AC/ Journal, Vol. 60, No. 2,
pp. 209-224.
Jacobsen, S., Marchand, J. and Boisvert, L. (1996), 'Effect of Cracking and Healing
on Chloride Transport in OPC Concrete', Cement and Concrete Research, Vol. 26, No.
6, pp. 869-881.
Jones, M.R., Thomas, M.D.A., Dhir, R.K. and Ng, S.L.D. (1996), 'Review of
Estimating of Chloride Ingress into Concrete', Proceedings of the 1h International
Conference on Durability of Building Materials and Components, Edited by Sjostrom,
C., Sweden, pp. 107-116.
Kanaya, M., Masuda, Y., Abe, M. and Nishiyama, N. (1998), 'Diffusion of Chloride
Ions in Concrete Exposed in the Coastal Area', Concrete Under Severe Conditions 2 :
Environment and Loading, Vol. 1, E & FN Spon, pp. 242-249.
Kaplan, M.F. (1961), 'Crack Propagation and the Fracture of Concrete', AC/ Journal,
Vol. 58, No. 5, pp. 591-610.
Khatri, R.P., Sirivivatnanon, V., Kidav, E.U. and Soo, T.P. (1996), 'Durable
Concrete for Marine and Aggressive Sulphate Environments', Proceedings of the Third
Asia-Pacific Conference on Structural Engineering and Construction (APSEC '96),
Johor, Malaysia, pp. 231-244.
Kotsovos, M.D. and Newman, J.B. (1981), 'Fracture Mechanics and Concrete
Behaviour', Magazine of Concrete Research, Vol. 33, No. 115, pp. 103-112.
Kumar, A., Roy, D.M. and Higgins, D.P. (1987), 'Diffusion Through Concrete',
Concrete, pp. 21-31.
L 'Hermite, R., (1959), 'What Do We Know About the Plastic Deformation and Creep
of Concrete', RILEM Bulletin No. 1, pp. 21-51.
Lambert, P., Page, C.L. and Vassie, P.R.W. (1991), 'Investigation of Reinforcement
Corrosion 2: Electrochemical Monitoring of Steel in Chloride Contaminated Concrete',
Materials and Structures, Vol. 24, No. 143, pp. 351-358.
References 198
Liam, K.C., Roy, S.K. and Northwood, D.O. (1992), 'Chloride Ingress
Measurements and Corrosion Potential Mapping Study of a 24-Year Old Reinforced
Concrete Jetty Structure in a Tropical Marine Environment', Magazine of Concrete
Research, Vol. 44, No. 160, pp. 205-215.
Lin, W.M. and Jou, T.W. (1991), 'Corrosion of Reinforced Concrete Structures in
Taiwan', Materials Performance, Vol. 30, No. 3, pp. 38-42.
Loo, Y.H. (1992), 'A New Method for Microcrack Evaluation in Concrete Under
Compression', Materials and Structures, Vol. 25, pp. 573-578.
Ludirdja, D., Berger, R.L. and Young, J.F. (1989), 'Simple Method for Measuring
Water Permeability of Concrete', AC/ Materials Journal, Vol. 86, No. 5, pp. 433-439.
Maage, M., Helland, S., Poulsen, E., Vennesland, 0. and Carlsen, J.E. (1996),
'Service Life Prediction of Existing Concrete Structures Exposed to Marine
Environment', AC/ Material Journal, Vol. 93, No. 6, pp. 602-608.
Maage, M., Helland, S. and Carlsen, J.E. (1997), 'Service Life Prediction of Marine
Structures', Proceedings of the 4th CANMETIACJ International Conference, Sydney,
Australia, SPl 70, pp. 723-743.
Mangat, P.S. and Molloy, B.T. (1994), 'Prediction of Long Term Chloride
Concentration in Concrete', Materials and Structures, Vol. 27, No. 170, pp. 338-346.
Mangat, P.S. and Molloy, B.T. (1995), 'Chloride Binding in Concrete Containing
PFA, GBS or Silica Fume Under Seawater Exposure', Magazine of Concrete Research,
Vol. 47, No. 171, pp. 129-141.
Matsushima, M., Tsutsumi, T., Seki, H. and Matsui, K. (1998), 'A Study of the
Application of Reliability Theory to the Design of Concrete Cover', Magazine of
Concrete Research, Vol. 50, No. 1, pp. 5-16.
McGreath, D.R. (1969), 'The Influence of Aggregate Particles on the Local Strain
Distribution and Fracture Mechanism of Cement Paste During Drying Shrinkage and
Loading to Failure', Materials and Structures, Vol. 2, No. 7, pp. 73-84.
Mehta, P.K. (1986), 'Hardened Cement Paste Microstructure and Its Relationship to
Properties', 8th International Congress on the Chemistry of Cement, Vol. 1, Brazil.
Mehta, P.K. (1997), 'Durability-Critical Issues for the Future', Concrete International,
pp. 27-33.
Mehta, P.K. and Monteiro, P.J.M. (1993), 'Concrete : Structures, Properties and
Materials', Prentice Hall Publication, 548 pp.
Meyers, B.L., Slate, F.O. and Winter, G. (1969), 'Relationship Between Time-
Dependent Deformation and Microcracking of Plain Concrete', AC/ Journal, pp. 60-
68.
References 200
Monteiro, P.J.M., Gjorv, O.E. and Mehta, P.K. (1985), 'Microstructure of the
Steel-Cement Paste Interface in the Presence of Chloride', Cement and Concrete
Research, Vol. 15, No. 5, pp. 781-784.
Najjar, W.S. and Hover, K.C. (1989), 'Neutron Radiography for Microcrack Studies
of Concrete Cylinders Subjected to Concentric and Eccentric Compressive Loads',
AC/ Materials Journal, Vol. 86, No. 4, pp. 354-359.
Neville, A.M. (1970), 'Creep of Concrete: Plain, Reinforced and Prestressed', North-
Holland, Amsterdam, 622 pp.
Neville, A.M. (1988), 'Properties of Concrete', 3rd Edition, Longman, England, 364
pp.
Ngab, A.S., Slate, F.O. and Nilson, A.H. (1981), 'Microcracking and Time-
Dependent Strains in High-Strength Concrete', AC/ Journal, Vol. 78, No. 4, pp. 262-
268.
Page, C.L. and Vennesland, 0. (1983), 'Pore Solution Composition and Chloride
Binding Capacity of Silica Fume Cement Pastes', Materials and Structures, Vol. 16,
pp. 19-25.
Piasta, W.G., Sawicz, Z. and Piasta, J. (1989), 'Sulfate Durability of Concrete Under
Constant Sustained Load', Cement and Concrete Research, Vol. 19, No. 2, pp. 216-227.
Polder, R.B. (1995), 'Chloride Diffusion and Resistivity Testing of Five Concrete
Mixes for Marine Environment', Penetration of Chlorides into Concrete - Proceedings
of the International Workshop, St.Remy-les-Chevreuse, France, Edited by L.O. Nilsson
and J.P. Ollivier, pp. 225-233.
Poston, R.W., Carrasquillo, R.L. and Breen, J.E. (1987), 'Durability of Post-
Tensioned Bridge Decks', AC/ Materials Journal, Vol. 84, No. 4, pp. 315-326.
Price, W.H. (1951), 'Factors Influencing Concrete Strength', AC/ Journal, Vol. 47,
pp. 417-432.
Rodriguez, J., Ortega, L.M., Casal, J. and Diez, J.M. (1996), 'Corrosion of
Reinforcement and Service Life of Concrete Structures', Proceedings of the Seventh
International Conference on Durability of Building Materials and Components,
Sweden, Edited by C. Sjostrom, pp. 117-126.
Rostam, S. (1996), 'High Performance Concrete Cover - Why It is Needed and How to
Achieve It in Practice', Construction and Building Materials, Vol. 10, No. 5, pp. 407-
421.
Rostam, S. and Shiessl, P. (1993), 'Next Generation Design Concepts for Durability
and Performance of Concrete Structures', Proceedings of the 6th International
Conference on Durability of Building Materials and Components, Japan.
Roy, S.K., Liam, K.C. and Northwood, D.O. (1993), 'Chloride Ingress in Concrete
as Measured by Field Exposure Test in the Atmospheric, Tidal and Submerged Zones of
a Tropical Marine Environment', Cement and Concrete Research, Vol. 23, pp. 1289-
1306.
Samaha, H.R. and Hover, K.C. (1992), 'Influence of Microcracking on the Mass
Transport Properties of Concrete', AC/ Materials Journal, Vol. 89, No. 4, pp. 416-424.
Shah, S.P. and Chandra, S. (1968), 'Critical Stress, Volume Change and
Microcracking of Concrete', AC/ Journal, pp. 770-780.
Shah, S.P. and Slate, F.O. (1968), 'Internal Microcracking, Mortar-Aggregate Bond
and the Stress-Strain Curve of Concrete', International Conference on the Structure of
Concrete, Cement and Concrete Association, London, pp. 82-92.
Shah, S.P. and Chandra, S. (1970), 'Fracture of Concrete Subjected to Cyclic and
Sustained Loading', AC/ Journal, Proc. 67 (10), pp. 816-825.
Sherman, M.R., McDonald, D.B. and Pfeifer, D.W. (1996), 'Durability Aspects of
Precast Prestressed Concrete Part 2: Chloride Permeability Study', PC/ Journal, Special
Report, pp. 76-95.
Smadi, M.M. and Slate, F.O. (1989), 'Microcracking of High and Normal Strength
Concretes under Short and Long Term Loadings', AC/ Materials Journal, Vol. 86, No.
2, pp. 117-126.
Suryavanshi, A.K., Swamy, R.N. and McHugh, S. (1998), 'Chloride Penetration into
Reinforced Concrete Slabs', Canadian Journal of Civil Engineering, Vol. 25, No. 1, pp.
87-95.
References 204
Suzuki, K., Ohno, Y., Praparntanatorn, S., Ninomiya, H. and Tamura, H. (1989),
'Influence of Flexural Crack on Corrosion of Steel in Concrete', Technology Reports of
the Osaka University, Vol. 39, No. 1953, pp. 49-57.
Swamy, R.N., Hamada, H. and Laiw, J.C., (1994), 'A Critical Evaluation of Chloride
Penetration into Concrete in Marine Environment', Proceedings of the Int. Conference
on Corrosion and Corrosion Protection of Steel in Concrete, Sheffield Academic Press,
UK, Vol. I, pp. 404-419.
Tang, L. and Nilsson, L.O. (1985), 'Chloride Diffusivity in High Strength Concrete
at Different Ages', Nordic Concrete Research, No. 4, pp. 162-171.
Tang, L. and Nilsson, L. (1993), 'Chloride Binding Capacity and Binding Isotherms
of OPC Paste & Mortars', Cement and Concrete Research, Vol. 23, No. 2, pp. 247-253.
Thomas, M.D.A., Matthews, J.D. and Haynes, C.A. (1990), 'Chloride Diffusion and
Reinforcement Corrosion in Marine Exposed Concretes Containing Pulverised Fuel
Ash', 3rd Int. Symp. on Corrosion of Reinforcement in Concrete Construction, pp. 198-
212.
Tumidajski, P.J., Chan, G.W., Feldman, R.F. and Strathdee, G. (1995), 'A
Boltzmann-Matano Analysis of Chloride Diffusion', Cement and Concrete Research,
Vol. 25, No. 7, pp. 1556-1566.
Tuutti, K. (1982), 'Corrosion of Steel in Concrete', CBI No.4, Swedish Cement and
Concrete Research Institute, Stockholm.
References 205
Tuutti, K. (1983), 'Analysis of Pore Solution Squeezed Out of Cement Paste and
Mortar', Intemationales Kolloquium 'Chlorid Korrosion'.
Vennesland, 0 and Gjorv, O.E., (1981), 'Effect of Cracks in Submerged Concrete Sea
Structures on Steel Corrosion', National Association of Corrosion Engineer, pp. 49-51.
Wang, K., Jansen, D.C. and Shah, S.P. (1997), 'Permeability Study of Cracked
Concrete', Cement and Concrete Research, Vol. 27, No. 3, pp. 381-393.
Wee, T.H., Yong, K.Y., Wong, S.F. and Lim, B.H. (1996), 'Strength Development
and Chloride Ingress in Concrete under Tropical and Temperate Conditions', 21 st
Conference on Our World in Concrete and Structures, Singapore, pp. 251-261.
West, R.E. and Hime, W.G. (1985), · 'Chloride Profiles in Salty Concrete', Materials
Performance, No. 7, pp. 29-36.
Weyers, R.E. and Smith, D.G. (1989), 'Chloride Diffusion Constant for Concrete',
Proceedings of the Sessions Related to Structural Materials at Structures Congress, San
Francisco, pp. 106-115.
Wiebenga, J.G. (1980), 'Durability of Concrete Structures Along the North Sea Coast
of the Netherlands', ACI SP-65, pp.437-452.
Yokozeki, K., Motohashi, K., Okada, K. and Tsutsumi, T. (1997), 'A Rational
Model to Predict the Service Life of RC Structures in Marine Environment',
Proceedings of the 4th CANMETIACI International Conference on Durability of
Concrete, Sydney, SP-170, Vol. 2, pp. 777-799.
Young, J.F. (1988), 'A Review of the Pore Structure of Cement Paste and Concrete
and Its Influence on Permeability', Permeability of Concrete, Edited by Whiting, D.
and W alitt, A, ACI SP-108, Detroit.
Zhang, T. and Gjorv. O.E. (1994), 'An Electrochemical Method for Accelerated
Testing of Chloride Diffusivity in Concrete', Cement and Concrete Research, Vol. 24,
No. 8, pp.1534-1548.
207
LIST OF PUBLICATIONS
The following is the list of publications based on the work carried out in the present
study.
4. Sirivivatnanon, V., Cao, H.T., Lim, C.C. and Gowripalan, N., 'Service Life
Modelling of Crack-freed and Cracked Reinforced Concrete Members',
Proceedings of the Second Asia/Pacific Conference on Durability of Building
Systems: Harmonised Standards and Evaluation, Vol. II, Bandung, Indonesia, 10-12
July 2000, pp.16-1 to 16-13.
APPENDIX A
\ ,_____
\
-
.I ~ .......... ...
... .. .. -
,.____,__ ~
APPENDIX B
2.5
-
- ...
C Cl>
.2 G>
ea u
2
/\ Grade 20
._ C
\
'
1.5
i
- 0
u-
u \
C 0
0
u
.
i
\.,, ~
\
\
\
0
-
..!. -;I!!.
0
0.5
I
I
\
b----~
"'
- .,____ -
---.::.
0
0 5 10 15 20 25
Distance, x (mm)
1--+--control -o -20% ---tr--35% - o -50% I
Figure B1: Instantaneously Loaded and 28 Days
of Salt Immersion
1.1
- 0.3
k\_
I
0~ \
0.2 ',
0.1 '
- ~~ -
0
0 5 10 15 20 25
Distance, x (mm)
1--+-- control - 0 - 20% - - -6- - - 35% - o - 50% I
0.8
C
o-
·-
- Cl)
~
0.7
0.6
Grade 20
-f! C
u 0.5
-; 0
u 0.4
go
0 . 0.3
u3
-
I
0~ 0.2
0.1
0
0 5 10 15 20 25
Distance, x (mm)
1--+- control --a - 20% - - -tr - · 35% - <> - 50% J
0.7
C -
·- ...
0
ti
Cl)
ai
u
0.6
0.5
Grade 20
._ C
cou
Cl)
0.4
u- 0.3
C 0
0 .
u3 0.2
-
I
0~
0.1
0
0 5 10 15 20 25
Distance, x (mm)
j --+- control --a - 20% - - -tr - - 35% - <> - 50% J
0.7
C -
Oa,
·- ...
Cl)
0.6
0.5
Grade 20
-f! u
C
0.4
i
- 0
u
g-s
0 .
0.3
u3
0.2
0
-
..I.-;;!!.
0
0.1
0
0 5 10 15 20 25 30 35
Distance, x (mm)
• control --a - 20% - - -1:r - - 35% - O - 50% I
0.8
0.7
c'i'
o-t!!
:;:
0.6
-...
ffl
C
Cl) u
u-
C
u
C
0
0.4
0
0.5
0 . 0.3
u 'i
(3 al!
- 0.2
0.1
0
0 5 10 15 20 25 30 35
Distance, x (mm)
I--+-- control --o - 20% - - -1:;.- - - 35% - o - 50% I
Figure B6: Under Sustained Load and 540 Days
of Salt Immersion
Appendix B 222
0.8
-
-.
C G>
.2 G)
0.7
0.6
Grade40
ftS CJ 0.5
._ C
- 0 0.4
16 CJ
CJ-
C 0
0 . 0.3
CJ°3
-
0.2
..!. ~
0 0
0.1
0
0 5 10 15 20
Distance, x (mm)
j ----+- control --0 - 20% - - -I!. - - 35% - <> - 50% j
0.8
a__
-
-.
C G>
.2 G) 0.6
0.7
''
\
\'',
Grade40
-.goftS CJ \ a\
C 0.5 \ \
C 0
CJ ~ \\
G)
0 .
0.4 \ \ \
0.3
CJ3
..!. ~
oe... 0.2 ~\6
''
0.1 '
0
0 5 10 15 20
Distance, x (mm)
• control --a -20% --½--35% - <> -50% j
0.9
0.8
CD
C
.
·--0 1i 0.7
~
Grade40
--~
u 0.6
E C \
0.5
i
- 0
u '_,
go
0 .
0.4
\~
u3 0.3 6,
..!. ~
() !?.,.. 0.2 \~,
~~
0.1
0
0 5 10 15 20 25
Distance, x (mm)
• control --a - 20% - - -A- - - 35% - <> - 50% I
0.9
0.8
.- .
C CD 0.7 Grade40
0
;:
1i
I'll u 0.6
C
0
C u 0.5
Cl)
u- 0.4
C 0
0 .
u3 0.3
-
I
0~ 0.2
0.1
0
0 5 10 15 20 25
Distance, x (mm)
I--+-- control --a - 20% - - -tr - - 35% - <> - 50% J
0.7
C
0
-...
G)
1i
0.6
0.5 Grade40
;
-
...ea
C
u
C
0 0.4
\
G)u
u-
C 0 0.3
0 .
uJ 0.2
is
-
~
0.1
0
0 5 10 15 20 25 30 35
Distance, x (mm)
I--+-- control --a - 20% - - -1:r - - 35% - O - 50% j
0.8
C
;:: 1i
0
-...
G)
0.7
0.6 Grade40
-g
~ Cu
'\
0.5
C
G)
0
0
u 0.4 °'
~
0 . 0.3
uJ
'
0~
- 0.2
0.1
0
~' ~
0 5 10 15 20 25 30 35
Distance, x (mm)
I--+-- control --a - 20% - - -1:r - - 35% - o - 50% I
Figure B12: Under Sustained Load and 540 Days
of Salt Immersion
Appendix B 225
0.6
c -
0 G)
·-...
-C'II -
G)
...
u
0.5
Grade40
-C C
0 0.4
~ u
so
u .
'i
0.3
~
~
G) 0.2
:2 >-
0 ,g
:c '#- 0.1 -----.__ ......
o-
0
0 5 10 15 20 25
Depth (mm)
• Measured - - Best-fit curve
- - Proposed Model
0.6
c -
0 G)
-
·-...
~·
0.5
-C'II
u
G)
... Grade40
-C C
0 0.4
~ u ~
so 0.3
~
u .
.g 'i 0.2
·-
0 ,g>- ~
:c
o-
'#- 0.1
0
0 5 10 15 20 25
Depth (mm)
• Measured - - Best-fit curve
- - Proposed 11/bdel
0.7
c -
0 a, 0.6
·--..ea -
._
a,
u 0.5 ~ Grade 20
~
-C C
0
Bu
c-
0.4
0
u
a,
0
i
. 0.3 ~~
3? >- 0.2
5 .c
::c
';I. 0.1
o-
0
0 5 10 15 20 25 30 35
Depth (mm)
• Measured
- - Proposed Wadel
- - Best-fit curve
0.8
.- -
c -
0 a, 0.7
·--ea ._
a, 0.6 Grade 20
u
C C
a, 0 0.5
u u
so
u
0.4
. 0.3
a, i
3? >- 0.2
5 .c
::c ';I. 0.1
o- 0
0 5 10 15 20 25 30 35
Depth (mm)
• Measured - - Best-fit curve
- - - Proposed Wadel
APPENDIX C
--
0.9
0
0.8
f!
.s::. 0.7
-.
OI 0.6
C
Cl)
0.5
I ll
I
.
Ill
Ill
Cl)
0.4
0.3
ci5
0.2 ...
<><>
124%
0.1
0 <>
<
.
0 0.19 0.38
Poisson's ratio
1
0.9
0 0.8
.
::::
ea 0.7
-
.s::.
OI
C
0.6
-!
Ill
I
0.5
0.4
.~ i
l
137%
-.
Ill
Ill
G> 0.3 •
en 0.2
'
0.1 '>
0 <> '
0 0.21 0.42
Poisson's ratio
1
0.9
0 0.8
.
:;::::
tlS 0.7
J
-.
.c
0.6
I..
C,
C
-Cl)
I ll
'
0.5
0.4
~ ....
.
Ill
Ill
Cl) 0.3
j34% I
c'ii 0.2
: I>
0.1
ot>
0
0 0.19 0.38
Poisson's ratio
0.9
0 0.8
.
:;::::
tlS
0.7
-.
.c
C,
C
0.6
- Cl)
I ll
'
Ill
Ill
0.5
0.4
- I!!
( /)
0.3
0.2
0.1
0
0 0.16 0.32 0.48
Poisson's ratio
-.
.2
ea
0.9
0.8
0.7
,,,,~
-.
.c
en
C
0.6
-
Cl)
1/1
I
0.5
0.4 40%
-.
1/1
1/1 0.3
Cl)
en 0.2
0.1
0
0 0.18 0.36
Poisson's ratio
1 .....--------------
0.9
0
0.8
l.c 0.7
'5, 0.6
C
-f
I I)
ii,
II)
0.5
0.4
...
a, 0.3 30%
en 0.2
0.1
0------------
0 0.18 0.36
Poisson's ratio
0.9
0 0.8
.
.::
ftl 0.7
-.
~
0,
C
0.6
-CD
Cl)
I
0.5
0.4
-.
Cl)
Cl)
CD 0.3
u, 0.2
0.1
0
0 0.188 0.376
Poisson's ratio
0.9
-.
.2
~
ftl
0.8
0.7
'5, 0.6
-.
C
CD 0.5
Cl)
I 0.4
-.
Cl)
Cl)
CD 0.3
"' 0.2
0.1
0
0 0.18 0.36
Poisson's ratio
0.9 oO
0
;:
f!
0.8
0.7 r'
-...
.c
en
C
0.6
-QI
1/1
I
1/1
1/1
0.5
0.4
•>
.....- :34% I
...
QI 0.3
ci:i 0.2
I
0.1
0 ~
0
0 0.2 0.4
Poisson's ratio
1
0.9
0 0.8
:;:
m ... 0.7
-...
.c
en
C
0.6
- QI
1/1
I
1/1
1/1
0.5
0.4 37%
-...
U)
QI 0.3
0.2
0.1
0
0 0.2 0.4 0.6
Poisson's ratio
~<><>
0.9
0.8
:;._0 0.7
0.6
0.5
0.4
0.3
0.2
0.1
Poisson's ratio
0.9
0
.
.:
ea
0.8
-.i
J:
en
0.7
0.6
-Cl)
I
0.5
0.4
.
Cl)
Cl)
Cl) 0.3
U) 0.2' 17'"
)
'
0.1' >
o!
0 10 20 30 40 50
Specific crack area
in microstrain
1
0.9
0
.
.:
ea
J:
0.8
0.7
15, 0.6
-.
C
Cl)
0.5
Cl)
I
0.4
-.
Cl)
Cl)
Cl) 0.3
u, 0.2
0.1
1
0.9
0 0.8
~ 0.7
.c
Cl 0.6
C
-!
fCl)f>
Cl)
0.5
0.4
! 0.3
cii 0.2
0.1
0
0 10 20 30 40 50 60 70 80 90
0.9
0 0.8
:;:::
...
CV
-
0.7
.c
Cl 0.6
C
-!
Cl)
I
Cl)
Cl)
0.5
0.4
-...
G)
U)
0.3
0.2
0.1
0
0 100 200 300 400 500
0.9
0 0.8
.::
...
lV 0.7
-...
.c
Cl
C
G>
0.6
0.5
iii I 0.4
Ill
Ill
-...G>
u,
0.3
0.2
0.1
0
0 100 200 300 400
0.9
0 0.8
.::
...
lV 0.7
-...
.c
Cl
C
0.6
-
G> 0.5
Ill
I
0.4
Ill
Ill
-
0.3
!
u, 0.2
0.1
0
0 100 200 300 400 500 600
1
0.9
0 0.8
.
;:
I'll 0.7
-.
.c:
t ,I
C
0.6
-
G>
I ll
I
0.5
0.4
-.
Ill
Ill
G> 0.3
u, 0.2
0.1
0
0 100 200 300 400 500
0.9
0.8
.
0
;:
I'll 0.7
-.
.c:
t ,I
C
0.6
-G>
I ll
I
0.5
0.4
-.
Ill
Ill
G> 0.3
u, 0.2
0.1
0
0 200 400 600 800 1000
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 ~-,----,.----<II0<),0,::,.-~--r---l
0 200 400 600 800 1000 1200 1400
1
0.9
-.
.2
ea
~
0.8
0.7
'5, 0.6
-.
C
G) 0.5
UI
I 0.4
-.
UI
UI
G) 0.3
( /) 0.2
0.1
0
0 200 400 600 800 1000 1200
0.9
0 0.8
.
.::
ea 0.7
-.
.c
en
C
0.6
-G>
U)
I
0.5
0.4
.
U)
U)
G> 0.3
cii 0.2
0.1
0
0 500 1000 1500 2000
0.9
0 0.8
;:
...
ftl
0.7
-...
.c
0)
C
G)
0.6
0.5
1ii
I 0.4
Cl)
Cl) volumetric
-...
G)
en
0.3
0.2
0.1
lateral
0
-400 -200 0 200 400 600 800 1000
Microstrain
0.9
-...
.2
ftl
0.8
0.7
-...
.c
0)
C
0.6
-Cl)
Cl)
I
Cl)
Cl)
0.5
0.4
0.3
...
G)
U) 0.2
0.1
0
-400 -200 0 200 400 600 800 1000
Microstrain
0.9
0 0.8
;: volumetric
...
ea 0.7
-...
.c
en
C
0.6
-Cl)
I ll
I
Ill
Ill
0.5
0.4
0.3
...
Cl)
U) 0.2
0.1
0
-400 -200 0 200 400 600 800 1000
Microstrain
1
0.9
0 0.8
i... 0.7
.c
'5, 0.6
C
-
...
Cl)
I ll
I
Ill
Ill
0.5
0.4
-...
Cl)
en
0.3
0.2
0.1
0
-500 0 500 1000 1500
Microstrain
0.9 volumetric
-.
.2
ea
0.8
0.7
-.
.c
C
en 0.6
-
G)
I ll
I
0.5
0.4
.=
Ill
u5
0.3
0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
0.9 volumetric
0
.
.::
ea
.c
0.8
0.7
"5, 0.6
-.
C
G) 0.5
I ll
I 0.4
-.
Ill
Ill
G) 0.3
"' 0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
1
0.9
0
.
;:
ftl
0.8
-.
.c
C
en
0.7
0.6
-
CD
eIn
en
0.5
0.4
-.
en
CD 0.3
en 0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
0.9
-.
.2
ftl
.c
0.8
0.7
'5, 0.6
-.
C
CD
0.5
eIn
en 0.4
.
en
CD
;;;
0.3
0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
1
0.9
volumetric
-.
.2
C'CI
0.8
-.
0.7
.i=
C) 0.6
C
Cl) 0.5
-;;
I 0.4
Ill
Ill
-
!
en
0.3
0.2
0.1
0
-1000 -500 0 500 1000 1500 2000
Microstrain
1
0.9
0
.
;:
.i=
C'CI
0.8
0.7
'S, 0.6
-.
C
Cl)
0.5
I ll
I
0.4 lateral
-.
Ill
Ill
Cl) 0.3
en 0.2
0.1
0
-1000 0 1000 2000 3000
Microstrain
0.9
0 0.8
.:
...
ffl
.c
0.7
'5, 0.6
C
-...
Cl)
I ll
I
Ill
Ill
0.5
0.4
-...
Cl)
Cl)
0.3
0.2
0.1
0
-1500 -1000 -500 0 500 1000 1500 2000 2500
Microstrain
1
0
0.9
0 0.8 0
.::
...
CV
0.7 0
-...
.=.
C)
C
0.6 0
-Cl) 0
0.5
0
I ll 0
I
Ill
0.4
Ill 0
-...
Cl)
en
0.3
0.2
~ ...
.... : 21% I
0.1 ~
J
0
0 0.19 0.38 0.57 0.76 0.95
Poisson ratio
0.9 ~ - - - - - - - - - - - - -
0
0.8
0 0
;:: 0.7
f!
-!
.=.
C)
0.6
0.5
iii
I 0.4 1111.. ..,•--4: 42.4% I
Ill ~
Ill
! 0.3 ~
en 0.2 >
)
0.1
0 +--------------!
0 0.17 0.34 0.51
Poisson Ratio
V
<>
0.9 <>
<>
0 <>
.
:;::
l'CI
0.8
<>
<>
-.
0.7 <>
.c <>
en 0.6 <>
C <>
-G)
Ill
I
0.5
0.4
<>
<>
L.....
-.
Ill
Ill
G)
0.3 : 32% I
en 0.2
~
0.1
0
!)
0 0.18 0.36
Poisson Ratio
0.9
0
.
:;::
I'll
0.8
-.
.c
Cl
C
0.7
0.6
-
Cl)
I ll
I
Ill
0.5
0.4
..
Ill
Cl)
u5
0.3
0.2
0.1
0
0 1000 2000 3000 4000
0.9
0.8
0
.
:;::
I'll
0.7
-.
.c
Cl
C
0.6
0.5
-Cl)
I ll
I
0.4
-.
Ill
Ill 0.3
CD
en 0.2
0.1
0
0 200 400 600 800 1000
Specific Crack Area
in microstrain
0.9
0
; 0.8
..
f!
.c
c,,
0.7
0.6
.
C
QI
1ii
0.5
I
0.4
.
Cl)
Cl)
QI 0.3
U) 0.2
0.1
0
0 100 200 300 400
Specific Crack Area
in Microstrain
0.9
0 0.8
;:
...
ea 0.7
-...
.c
en 0.6
C
Cl) 0.5
inI 0.4
II)
II)
...
Cl)
U)
0.3
0.2
0.1
0
-2000 -1000 0 1000 2000 3000
Microstrain
0.9
0.8
0 0.7
;:
...ea
-...
.c 0.6
en 0.5
C
-Cl)
I I)
I
II)
0.4
0.3
m
-...
u, 0.2
0.1
0
-400 -200 0 200 400 600 800 1000 1200
Microstrain
1
0.9
0
.
;
«I
0.8
-..
0.7
~
Cl 0.6
C
-
G>
U)
I
0.5
0.4
..,,
U)
U)
,-
G> 0.3
0.2
0.1
0
-500 0 500 1000 1500
Microstrain
APPENDIX D
Grade 40
<
--
• .)
I, ",..,.,.__... . • -~
~~<Y.j. ~
· -· · · :!: ,;,d; -..
'1' '' .,
.,.-~ . . ·,
•'/, f ..· J. .
.-,.- · "" . -· .
~.,
. ~-~.. . ..... ;,;:
...
..
... . ~ ....
~
,: