Sunteți pe pagina 1din 75

Concrete Durability Series

Z7/05 Durability Modelling of


Reinforcement Corrosion in
Concrete Structures
Z7/05
Durability Modelling
of Reinforcement Corrosion
in Concrete Structures

i, Durability Modelling of Reinforcement Corrosion in Concrete Structures


ii, Durability Modelling of Reinforcement Corrosion in Concrete Structures

ii, Durability Modelling of Reinforcement Corrosion in Concrete Structures


The principal authors of this Recommended Practice were:
Dr Shengjun Zhou ANCON Beton (Chair of Task Group)
Frank Papworth BCRC (Chair of Durability Committee)

Those actively assisted with the developed of this Recommended Practice were
Warren Green
Joost Gulikers
Rodney Paull

The CIA would like to acknowledge the valuable contribution of members from comments obtained through the peer
review process. Many of the comments have been included directly into this document.

iii, Durability Modelling of Reinforcement Corrosion in Concrete Structures


iii, Durability Modelling of Reinforcement Corrosion in Concrete Structures

iv, Durability Modelling of Reinforcement Corrosion in Concrete Structures


iv, Durability Modelling of Reinforcement Corrosion in Concrete Structures

v, Durability Modelling of Reinforcement Corrosion in Concrete Structures


presently available with general international consensus
for quantitative prediction of service life. A full
probabilistic approach or partial factor approach for
Preface design or service life is therefore not feasible and
deemed to satisfy approaches are the general approach
Durability of a concrete structure is defined in taken. Some preliminary models have been proposed
fib Model Code 2010 (Reference 1) as “the capability of and sometimes used for very specific mechanisms (e.g.
structures, products or materials of continuing to be acid attack in sewers) but their use is not common as
useful after an extended period of time and usage” and avoidance of deterioration measures (e.g. acid resistant
in AS 3600-2009 (Reference 2) as the "ability of the liners for sewers) or deemed to satisfy requirements are
structure and its component members to perform the the more general practice.
functions for which they have been designed, over a In the past, reinforcement corrosion protection
specified period of time, when exposed to their to most concrete structures was designed using a
environment”. It is often measured by the length of time deemed-to-satisfy approach by following code
(design or service life) it can maintain the required requirements, which was predominantly established
structural functionality at a defined reliability level. based on long term field observations. The durability
Benefits of a durable structure include a outcomes using this approach were a mixture of some
sufficient design and service lives as specified, successes and some failures. It was found that the
minimum severe defects or premature failure with durability failures occurred more frequently on the
minimum operational disruption of services and structures in aggressive conditions built since 1970
minimum maintenance cost, improved sustainability while structures built before that performed generally
with minimum consumption of raw materials and energy better.
associated with less rebuilding or repairing cycles, and Although the causes of such a change have not
greater public safety for the service operators and been fully understood, this change has coincided with
users. many changes including the cement characteristics
Durability design of a reinforced concrete (containing more C3S and being finer), climate change
structure mostly involves selecting suitable concrete (higher temperature and more C2O in atmosphere) and
compositions and related durability measures for a construction practices (poor curing and compaction).
specific exposure condition to achieve the specified Furthermore, these changes have not be reflected in
design life. There are mainly four approaches to the durability requirements by various Australia
conducting durability design as defined in fib Bulletin 34 Standards.
- 2006 (Reference 3). These include (1) deemed to Due to lack of long term durability data on the
satisfy design, i.e. complying to the durability new materials characteristics and change of exposure
requirements in various codes, (2) avoidance of conditions, a deemed-to-satisfy approach may not be
deterioration (e.g. use of stainless steel to avoid sufficiently reliable in some cases and overly
potential issues with black steel corrosion), (3) partial conservative in others where higher performing
safety factor design with deterministic modelling, and materials are used. The avoidance of deterioration
(4) full probabilistic design based on stochastic approach can reliably provide a superior durability
modelling. performance in most conditions. However the
A very large proportion of deteriorated concrete associated high cost discourages wide application
structures are related to reinforcement corrosion. except on some critical elements in critical projects.
Therefore, this recommended practice on durability As an alternative durability design method,
modelling deals only with corrosion of reinforcement. durability modelling (of either full probability design or
In addition, no time-dependent model of deterioration partial factor) based on current material characteristics
processes under other physical and chemical attacks is and mathematics has a potential ability to provide a

vi, Durability Modelling of Reinforcement Corrosion in Concrete Structures


much more reliable durability outcome if appropriate Practice will be updated in the future when new
models and parameters are adopted. It is especially understanding and developments justify.
effective to predict long term performance of
reinforcement corrosion in concrete structures for
chloride laden conditions and carbonation conditions.
Dr Shengjun Zhou
The advantage of a modelling approach is that
Z7/05 Task Group Chairman
it is significantly less reliance on long term performance
data of field concrete structures although initial model
calibration may require some such data at the
beginning and it can be adopted for more aggressive
exposure conditions compared those in codes. Such a
modelling approach (for chloride, carbonation and
reinforcement corrosion) has been increasingly applied
in durability design for major infrastructure projects in
Australia and around world. In addition, a modelling
approach is the most effective and essential tool to
determining the remaining service life of existing
concrete structures as part of a condition assessment
process.
Various models (for chloride, carbonation and
reinforcement corrosion) and preferred input
parameters have been established and used in the
past. However they produced significantly different
prediction results and consequently different durability
requirements even for the similar conditions and
materials. Some used models and associated input
values have been incorrect, incomplete, and/or
inappropriate for the prevailing conditions. Therefore,
to achieve accurate and reliable modelling outcomes
without a risk of premature durability failure or being too
conservative at a higher cost, it is critical to select
suitable durability models and input parameters across
the industry. Only by this approach, can durability
designs consistently achieve reliable durability
outcomes.
The key objectives of this Recommended
Practice are to review commonly used models for
prediction of reinforcement corrosion in concrete
structures and input parameters for chloride diffusion,
carbonation and corrosion of reinforcement (including
stressed tendons) to determine the most suitable
models and input parameters with statistical
distributions. Considering the complex nature of these
concrete deterioration processes and future data from
ongoing international research, this Recommended

vii, Durability Modelling of Reinforcement Corrosion in Concrete Structures


Contents
3.8.1 Terms Used to Define Threshold
TERMINOLOGY Chloride Concentration 36
1. INTRODUCTION 1 3.8.2 Defining Corrosion Initiation 37
1.1 General 1 3.8.3 Measuring Threshold Chloride
1.2 Durability Design Approaches 2 Concentration by Varying Chloride
Levels 38
1.2.1 Deemed to Satisfy Design 3
3.8.4 Recommended Threshold
1.2.2 Avoidance of Deterioration 5
Chloride Level 38
1.2.3 Full Probabilistic Modelling 5
3.9 Partial Factors 39
1.2.4 Partial Factor Design 6
1.3 Deterioration Category and Scope of 4. MODELLING CARBONATION IN CONCRETE 40
Durability Modelling 7 4.1 Introduction 40
2. GENERAL DETERIORATION MODEL FOR 4.2 General Carbonation Model 40
4.3 FIB Bulletin 34 Model 41
REINFORCEMENT CORROSION 9 4.4 Validation of fib Input Parameters 44
2.1 Tuutti Model 9 4.4.1 GP Concretes 45
2.2 Limit States 10 4.4.2 Fly Ash Concretes 45
2.3 Design Life 10 4.4.3 GGBS Concretes 46
2.4 Reliability Assessment 11 4.4.4 Silica Fume Concretes 47
3. MODELLING CHLORIDE DIFFUSION IN 4.4.5 Summary 48
4.5 Simplified FIB Model 48
CONCRETE 13 4.6 Test Methods to Determine
3.1 Introduction 13
Carbonation Coefficient 48
3.2 Basic Models of Chloride Diffusion 14
3.2.1 Fick's First Law 14 5. MODELLING REINFORCEMENT CORROSION 49
3.2.2 Diffusion with Free and Bound 5.1 Introduction 49
Chlorides 15 5.2 Critical Criterion for Corrosion
3.2.3 Fick's Second Law 16 Damage 49
3.3 Initial Chloride Concentration (Base 5.3 General Corrosion Modeling 50
Chloride Level) 17 5.3.1 Chloride-Contaminated Concrete 50
3.4 Surface Convection Zone 19 5.3.2 Carbonated Concrete 50
3.5 Decreasing Chloride Diffusivity 21 5.4 Modelling Corrosion under Oxygen
3.5.1 Instantaneous Diffusivity Models 21 Control 51
3.5.2 Analytical Model for 5.4.1 Modelling Method 51
Instantaneous Diffusivity 22 5.4.2 Application to Immersed Concrete 52
3.5.3 Numerical Model for 5.4.3 Application to Buried Concrete 52
Instantaneous Diffusivity 23 5.5 Corrosion Rate with Resistivity
3.5.4 Testing Instantaneous Chloride Control 53
Diffusivity 23 5.5.1 Resistivity and Corrosion Rate 53
3.5.5 Instantaneous Diffusivities 5.5.2 Concrete Resistivity Values 55
Published 26 5.6 Length of Propagation Period 56
3.5.6 Apparent Diffusivity Model 31 5.6.1 General Model 56
3.6 Effect of Temperature 33 5.6.2 Andrade's Model 56
3.7 Surface Chloride Concentration 34
3.7.1 Increasing Surface Chloride 6. SUMMARY AND RECOMMENDATIONS 58
Concentration with Time 34 6.1 Modelling Chloride Diffusion 58
3.7.2 Ultimate Surface Chloride 6.2 Modelling Carbonation 58
Concentration 34 6.3 Modelling Reinforcement Corrosion 59
3.8 Threshold Chloride Concentration 36
REFERENCES 60

viii, Durability Modelling of Reinforcement Corrosion in Concrete Structures


TERMINOLOGY

Avoidance of deterioration design - A durability design process to select durability measures to avoid any
potential deterioration of concrete elements or sections in specific exposure conditions.
Condition assessment - A process of reviewing information gathered about the current condition of a structure
or its components, its service environment and general circumstances, whereby its adequacy for future service
may be established against specified performance requirements for a defined set of loadings and/or
environmental circumstances.
Condition control - The overall through-life process for conserving the condition of a structure involving
condition survey, condition assessment, condition evaluation, decision-making and the execution of any
necessary interventions; performed as a part of the conservation process.
Conservation - Activities and measures taken which seek to ensure that the condition of a structure remains
within satisfactory bounds to meet the performance requirements for a defined time; that is in respect of
structural safety, serviceability and sustainability requirements, which may include considerations such as
aesthetics and heritage overlays.
Corrosion - Deterioration of a metal by a chemical, electrochemical or electrolytic reaction within its service
environment(s).
Deemed to satisfy design - A durability design process assumes that durability is sufficient if the prescriptive
or performance-based durability requirements in international, national or state codes or project specifications
are complied with during design and construction.

Design service life or (specified) design life - The period in which the required performance shall be
achieved and used in the design of new structures. The specified (design) service life is related to the required
service life, as given by the stakeholders (i.e. owners, users, contractors, society) and to the other implications
of service criteria agreement (e.g. with regard to structural analysis, maintenance and quality management).

Deterioration - Worsening of condition with time, or a progressive reduction in the ability of a structure or its
components to perform according to their intended specifications.

Deterioration mechanism - Scientifically describable process of the cause and development of deterioration.

Deterioration model: Mathematical model that describes structural performance as a function of time, taking
deterioration into account.
Diffusion - Movement of ions (e.g. chloride ions) due to a difference in the concentration gradient.

Durability - The capability of structures, products or materials of continuing to be useful after an extended
period of time and usage. In the context of performance-based design of structures, durability refers to the
fulfilment of the performance requirements within the framework of the planned use and the foreseeable actions,
without unforeseen expenditure on maintenance and repair (Reference 1).
Durability checklist - Tabular/matrix form summarising durability issues relating to assets (structures), asset
elements and asset sub-elements of a project. Durability checklists are authored by the durability consultant.
Durability consultant - Person or group who completes the durability assessment and is the author of the
durability assessment report and durability checklists.
Durability limit state (DLS) - A limit state used to define the end of the service life of a structure. The limit state
may be a condition, performance or operational limit state. Most commonly it is a condition limit state. For
example, for reinforced concrete structures subjected to deterioration caused by corrosion of reinforcing steel,
one or more of the following durability limit state levels may be used to define the end of service life:

ix, Durability Modelling of Reinforcement Corrosion in Concrete Structures


 Depassivation of the reinforcing steel (or initiation of corrosion).
 Cracking of the cover concrete.
 Spalling of the cover concrete.
 Loss of section (and reduced structural capacity).
 Collapse of the structure.

Durability Modelling - A systematic and mathematic description of long term deterioration process in concrete
structures exposed to specific exposure conditions to estimate their service lives or to determine the materials'
characteristics required for achieving specified design lives.
Environment/Exposure influences - Physical, chemical and biological actions resulting from the atmospheric
conditions or characteristics of the surroundings to the structure including macro and micro influences.
Full probabilistic modelling - A durability design process to run a mathematical model to predict the
probability of a deterioration failure, using various model input values having a probability distribution
representing variation of materials performances, design parameters and exposure conditions, as well as
uncertainty on quantification.
Ingress - The entry of substances (e.g. gas, liquid, ions) into structural and/or non-structural components of a
structure. Often the term “ingress” is associated with the entry of substances that cause deterioration (e.g.
chlorides into reinforced or prestressed concrete, water, sulphates and carbon-dioxide (CO2) into concretes,
etc.).

Intervention - A general term relating to an action or series of activities taken to modify or preserve the future
performance of a structure or its components.
Maintenance - A set of planned (usually periodic) activities performed during the service life of the structure
intended to either prevent or correct the effects of minor deterioration, degradation or mechanical wear of the
structure or its components in order to keep their future serviceability at the level anticipated by the designer.
Maintenance activities involve recurrent or continuous measures which enable the structure to fulfil the
requirements for reliability. The term ”maintenance” is commonly applied in the context of building fabric
components with a limited life, components associated with water management and rainwater run-off, items
where regular intervention is required to maintain their effective operation, etc. The term “maintenance” is
commonly applied to ancillary items such as gutters, drains, sealants, movement joints, bearings, etc.

Partial factor design - A durability design process to determine the design durability requirements by running
a deterministic model to select the minimum durability requirements and by adding a partial safety factor on top
to account for the uncertain factors.
Permeability - A measure of flow of a liquid (typically water) or gas under pressure through a material.
Permeability can be calculated from measurements of penetration depth at a certain time or flow rate for a
known thickness.
Preventative intervention - A pro-active conservation activity concerned with applying some form of treatment
or taking action prior to a change in a material property (e.g. such as that caused by the influence of carbonation
or chlorides) adversely affecting the ability of the structure, or parts thereof, to meet the required performance
levels because of deterioration.
Rehabilitation - Intervention to restore the performance of a structure or its component parts that are in a
changed, defective, degraded or deteriorated state to the original level of performance, generally without
restriction upon the materials or methods employed.

x, Durability Modelling of Reinforcement Corrosion in Concrete Structures


Reliability - The ability of a structure or a structural member to perform its intended function satisfactorily (from
the viewpoint of the customer) for its intended life under specified environmental and operating conditions.
Reliability is usually expressed in probabilistic terms. In the context of performance-based design of structures,
reliability refers to the ability of a structure or a structural member to fulfil the performance requirements during
the service life for which it has been designed at a required failure probability level corresponding to a specified
reference period.
Repair - Intervention to reinstate to an acceptable level the current and future performance of a structure or its
components which are either defective, deteriorated, degraded or damaged in some way so their performance
level is below that anticipated by the designer; generally without restriction upon the materials or methods
employed.
Residual Service life - The remaining period in which the required performance shall be achieved from current
time until the design service life is achieved.
Risk - The combination of the likelihood of occurrence of a particular hazard and its consequences.

Serviceability limit state (SLS) - State that corresponds to conditions beyond which specified service
requirements for a structure or structural member are no longer met.
Supplementary Cementitious Material (SCM) - Pulverised fuel ash (PFA) or fly ash (FA); ground granulated
blast furnace slag (GGBFS) or slag; silica fume (SF) or amorphous silica (AS); or pozzolans.
Testing - Procedure aiming at obtaining information about the current condition or performance of a structure or
its components. Various types of testing are recognised, their classification being primarily on the basis of the
amount of damage or interference caused to the structure. The main divisions are:

 Non-destructive testing (NDT), which does not cause damage to the structure by the test procedure (e.g.
testing with cover meter, radar, acoustic emission, load testing in the elastic range, etc.),
 Destructive testing, which may cause damage to the structure or marking of the surface finishes (e.g.
pull-out tests, material sampling, load testing beyond the elastic range, etc.).
Ultimate limit state (ULS) - State associated with collapse or with other similar forms of structural failure.
Generally the ultimate limit state corresponds to the maximum load-carrying resistance of a structure or
structural member.

xi, Durability Modelling of Reinforcement Corrosion in Concrete Structures


1. INTRODUCTION

1.1 GENERAL

Durability of a reinforced concrete is often measured by the maximum time period it can maintain the required
structural serviceability or functionality at a defined reliability level. It depends on the intrinsic resistance of the
material or structure against a specific intensity of the deleterious actions applied. Durability design is a design
process to develop the required durability measures to enable a structure to have a sufficient resistance against
the deleterious actions over the entire design life. Durability design of concrete structures mostly involves
selecting the most suitable concrete composition and other related durability measures (e.g. cover thickness or
other protection) for a specific exposure condition to achieve the required durability.

In Australia most durability designs up to now have been based on a deemed to satisfy approach assuming that
the durability will be sufficient if the durability requirements in relevant codes have been complied with.
However, in previous years, a number of premature durability failures of structures, especially those in marine
and other severe exposures, have occurred causing some major concerns regarding the sufficiency of this
durability design approach. Durability design by modelling approaches based on mathematical principles and
measured materials characteristics to predict resistance of structure against specific deterioration actions is
being increasingly used to achieve sufficient durability outcomes as specified. Such methods are finding their
way into some national codes such as NZ 3101-2-2006, Concrete Structures Standards-Commentary
(Reference 4) and are also extensively documented in international codes including fib Bulletin 34-2006, Model
Code for Service Life Design (Reference 3), fib Model Code 2010 (Reference 1), and ISO 16204-2012,
Durability – Service life design of concrete structures (Reference 5). In addition, durability modelling is an
essential tool in condition assessment of existing structures and thus has been more widely used in practice.

A very large proportion of deteriorated concrete structures are related to reinforcement corrosion. Therefore,
this recommended practice on durability modeling deals only with corrosion of reinforcement. In addition, no
time-dependent model of deterioration processes under other physical and chemical attacks is presently
available with general international consensus for quantitative prediction of service lives. A full probabilistic
approach or partial factor approach for design or service life is therefore not feasible and deemed to satisfy
approaches are the general approach taken. Some preliminary models have been proposed and sometimes
used for very specific mechanisms (e.g. acid attack in sewers) but their use is not common as avoidance of
deterioration measures (e.g. acid resistant liners for sewers) or deemed to satisfy requirements are the more
general practice.

As there are no Australian Standards providing guidance to the modelling process of chloride diffusion,
carbonation and reinforcement corrosion, various models and input parameters have been used by durability
consultants giving significantly different durability measures even for similar exposure conditions and design
lives. They can also gives variable predictions of residual service life in condition assessment of similar existing
concrete structures. Apparently, many if not all of the models and input parameter are unsuitable or incorrect.
This is probably because the models and inputs are not well understood and properly used. This issue has led
to severe conflicts among various stakeholders in projects, which overshadow the real technical issues.

This Z7/05 document provides guidance to the durability modelling process of chloride diffusion, carbonation
and reinforcement corrosion, to ensure that consistent and reliable durability outcomes can be achieved across
the Australian construction industry. In addition, it provides guidance to the modelling process for the condition
assessment of the existing structure to determine the residual service life. It should ensure that after design and
construction, reinforced concrete elements will have adequate resistance against deterioration in a given
exposure condition for the entire design life at an appropriate reliability level. This document provides
recommendations for:
A. situations where the use of modelling is appropriate in durability design of new concrete structure and
condition assessment of existing structures;
B. models to be used in Australia;
C. input values/distributions for use in the recommended models; and
D. guidance on specifications of materials performance testing based on model input requirements.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 1


1.2 DURABILITY DESIGN APPROACHES
There are four approaches in the durability design process defined in fib Bulletin 34 - 2006
as shown in

Figure 1, i.e. (1) deemed to satisfy requirements as commonly found in codes, (2) avoidance of
deterioration, e.g. use of stainless steel to avoid potential corrosion risks with black steel corrosion, (3) full
probabilistic design based on modelling and (4) partial factor mostly for deterministic modelling. Their
details are discussed in this section.

assumptions

terms of definition

other administrative provisions

principles of service life

Design criteria

in case of non-conformity with the performance criteria, the


structure becomes either obsolete or subject to a re-design
full probabilistic design partial factor design deemed to satisfy design avoidance of deterioration

probabilistic models design values


-resistance -characteristic values exposure classes exposure classes
-loads/exposure -partial safety factors
-geometry -combination factors

limit states design equations design provisions design provisions

design verification

project specification for material selection and execution


maintenance plan
inspection / monitoring plan

quality plan for execution


(optional)

inspection of execution

condition control
maintenance
during service life

Figure 1, Service life design process showing the four methods of durability design by fib Bulletin
34 - 2006 (Reference 3)

In all these durability design approaches, potentially suitable durability measures (materials and
construction requirements) are selected and tested to see whether they comply with the relevant

Durability Modelling of Reinforcement Corrosion in Concrete Structures 2


durability requirements. These durability requirements are developed based on observed or predicted
durability performance of the concrete in a specific environment.

In the deemed-to-satisfy approach, the requirements are specified by the codes and they are often the
minimum characteristic compressive strengths of the concrete and cover thickness. In the avoidance of
deterioration approach, the avoidance requirements are determined based on anticipated performance of
a structure in an exposure condition. In the durability modelling approach, the durability requirements are
developed based on the predicted performance to achieve the specified design life and these
requirements are often the characteristics of concrete materials (e.g. chloride diffusivity or carbonation
coefficient) and the cover thickness.

1.2.1 Deemed to Satisfy Design

The deemed-to-satisfy design approach assumes that durability is sufficient if the prescriptive or
performance-based durability requirements in international, national or state codes or project
specifications are complied with during design and construction.

In Australia most durability designs are based on the deemed-to-satisfy approach with durability
measures based on the requirements in various codes. Australian national codes having such durability
requirement for concrete structures include:
 AS 3600-2009, Concrete structures (Reference 1)
 AS 5100.5-2004, Bridge design, Part 5: Concrete (Reference 6)
 AS 3735-2001, Concrete structures for retaining liquid (Reference 7)
 AS 2159-2009, Piling-Design and installation (Reference 8)
 AS 4975-2005, Guideline for the design of maritime structures (Reference 9)

Other Australian codes contain deemed to satisfy requirements for concrete elements e.g.:
 AS 1085.14-2012, Railway track materials - Prestressed concrete sleepers (Reference 10)
 AS 1597.2-1997, Precast reinforced concrete box culverts, - Part 2, Large Culverts (Reference 11)
 AS 4058-1996, Precast concrete pipes (Reference 12)

Australian standards also ensure durability through requirements for materials e.g.:
 AS 1379-2007, Specification and supply of concrete (Reference 13)
 AS 2758.1-1998, Aggregates and rock for engineering purposes, Part 1, Concrete aggregates
(Reference 14)
 AS 3972-1997, Portland and blended cement (Reference 15)
 AS 3582.1-1998, SCM, Part 1, Fly ash (Reference 16)
 AS 3582.2-2001, SCM, Part 2, Slag-Ground granulated iron blast-furnace (Reference 17)
 AS 3582.3-2003, SCM Part 3, Amorphous silica (Reference 18)

In many cases, use of these code requirements has been successful. However, in other cases,
especially when the design life is long and exposure conditions are severe, durability of concrete
structures is often not sufficient leading in many cases to premature distress.

Most durability requirements in Australian and international codes are developed based on long term
observations of field performance of in-service concrete structures or concrete samples in various typical
environments. Use of requirements based on in-service concrete structures or large concrete samples
provided reliable durability outcomes for the same conditions, materials compositions and durability
measures. However, small and well prepared samples may not provide a good indication of structure
durability due to different boundary conditions and concrete sample quality. In addition, samples are
often tested in a short period of time with results being extrapolated to predict long term performance.
Such extrapolation may significantly reduce the reliability of predictions of long term performance
(Reference 19).

Durability Modelling of Reinforcement Corrosion in Concrete Structures 3


It is obvious that the long-term observations are only reliable for the same conditions, materials
compositions and durability measures. However, the materials characteristics such as the cement
fineness and C3S content have changed significantly over the past 60 years (Reference 20). Cement
strength has increased greatly and for the same target strength the w/c ratio is much higher leading to
reduced performance. Specifying the same minimum characteristic strength is not achieving the same
durability as 60 years ago.

Some code requirements are also developed based on data from accelerated laboratory tests. Such
accelerated tests often promote the severity of exposure condition artificially to obtained quick results,
such as increasing the concentration of deleterious media, pressure gradient, temperature, humidity,
strength of electrical field, and/or number of wet-dry cycles. Often increased severity changes the
deterioration process significantly and prediction of durability performance in the natural field exposure is
then not reliable. Although a correlation between results in two conditions may be established, such
relationships may change with materials characteristics and compositions and great care should be taken
during the analysis process.

A prime example of this is accelerated carbonation testing where a sample is placed in a carbonation tank
with a high CO2 concentration. In this condition, the increase of carbonation rate with CO2 concentration
becomes non-linear due to the changing process of the penetration and reaction of CO 2 in the concrete
pores. In addition, the effects of a continuous maturing process of concrete cannot be reflected in such a
short term test.

Another change to materials that negates the use of historic durability requirements is the increasing use
of supplementary cementitious materials (SCM) such as fly ash (FA), ground granulated blast furnace
slag (GGBFS) and silica fume (SF). Generally, concretes with SCMs have much better durability in
relatively wet conditions such as marine, water and in-ground. However, such concretes generally have a
faster carbonation rate and longer curing requirement than GP (Portland cement) concretes of the same
strength.

The climate has changed over the past 60 years with temperature rising 1 ºC and CO 2 concentration
increasing by 80 ppm. These changes will lead to a faster deterioration of concrete structures, such as
reinforcement corrosion induced by chloride diffusion and carbonation. Therefore, code requirements
based on the past observations become more unreliable.

The deemed to satisfy durability requirements in the numerous Australian codes (listed above) vary due
to specific exposure, design life, materials and reliability requirements being directly incorporated into the
prescriptive or performance requirements. However, there is not a clear explanation of how the various
requirements were developed and hence developing them to take account of other influences is likely to
lead to errors. A key concern is that different reliabilities appear to have been applied in some codes but
none of the codes state the reliability expected from the requirements given.

For example engineers may be designing a concrete structure and feel that AS 3600 (Reference 1)
provides the appropriate requirements except they want a 100 year life not 40-60 years. AS5100.5
(Reference 6) gives requirements for 100 year design life but it is not clear the extent to which
requirements are influenced by higher reliability required for bridges and what variations are due to
design life difference.

These performance and prescriptive requirements are often based on the observations of materials
durability performance in the past. However, reliable performance characteristics can only be established
after many decades. The changes in materials over the course of time discussed above, and the
consequent changes to durability performance have not been observed sufficiently and reflected in code
requirements. Durability design with a deemed-to-satisfy approach based on observations of past
materials has led to durability failure of many structures in severe exposures.

There are also issues with some codes providing limited guidance on requirements for different
exposures while other codes, or at least their commentaries, provide extensive guidance. For example if
designing a bridge that will have concrete in marsh waters the commentary to AS3735 (Reference 7)
provides some excellent information but the designer has to interpret how to apply the information from

Durability Modelling of Reinforcement Corrosion in Concrete Structures 4


AS 3735 that has a different design life and possibly different reliability to that required for bridges in AS
5100 (Reference 6).

The Australian codes recognise their limitations by noting that the provisions will not always meet the
desired outcomes considering lack of knowledge in deterioration rate in various conditions. For example
exposure conditions are more aggressive at high temperatures in the far north than those in cooler
southern climates. This leaves a large burden on designers to make appropriate allowances for the local
conditions.

Workmanship is another issue not taken into account for concrete using ‘standard’ compaction and
supervision.

Major concrete users such as transport authorities recognise deficiencies in the national codes and
provide specific durability requirements in their specifications. They and other stakeholders also
sometimes require that a durability consultant is used on their projects to define durability requirements
using the national codes as the lowest requirement.

1.2.2 Avoidance of Deterioration

The avoidance of deterioration approach is a design process to select durability measures to avoid any
potential deterioration of concrete elements or sections in specific exposure conditions. A typical
example of such measures is using stainless steel as reinforcement to prevent rebar from corrosion in
chloride or carbonation causing environments. Another example is that durability design to avoid AAR
deterioration is often best approached by avoidance of using reactive aggregate.

Where avoidance of deterioration can be reliably applied at an economic cost it may be the simplest
approach. For example in a very severe marine exposure replacing black steel reinforcement requiring a
cover thickness of 70 mm with stainless steel reinforcement requiring a cover thickness of only 30 mm
with minimum risk of reinforcement corrosion might be a low cost solution for bridge beams where the
reduction in self weight may make the use of stainless steel the preferred solution. The substitution may
eliminate many concrete QA and maintenance requirements and provide a higher reliability level than can
be achieved using black steel.

In most cases avoidance of deterioration may come at a very high cost. Therefore, the application of this
approach is often strictly limited to the critical and high risk elements.

1.2.3 Full Probabilistic Modelling

The general model for durability of concrete structures considers the time to a critical amount of damage,
based on an initiation period where no damage occurs and a subsequent propagation period over which
damage occurs at some rate until the critical damage level is reached. In durability design, the design life
must be longer than the combined initiation and propagation time.

Full probabilistic modelling is a design process to run a mathematical model to predict the probability of a
deterioration process, using various model input values having a probability distribution representing
variation of materials performance, design parameters and exposure conditions, as well as uncertainty on
quantification. Such a modelling process (commonly by Monte Carlo simulation) aims to give an
assessment of the structures’ reliability with exposure time.

Unfortunately the distributions of materials performance and exposures are not well defined or
understood and are quite variable. This can either throw doubt on the models, make them overly
conservative relative to the empirical approach or provide unrealistic expectations if adequate account is
not taken of the variability. There are also few designers with both the software and knowledge (theory of
durability and probability based assessment) to undertake full probability analysis. As a results, the
partial factor approach is more commonly adopted when modelling is used.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 5


In this document, distributions of various parameters are discussed with their mean values and standard
deviation or coefficient of variation recommended. These distribution parameters can be used as a
guideline for full probabilistic modelling.

It is recommended that full probabilistic modelling be used:


 for assessing the condition of existing aged structures;
 for critical and high risk elements or sections in critical projects; and/or
 when the client requests such a design approach.

1.2.4 Partial Factor Design

In the partial factor design approach, the use of input distributions to assess probability of deterioration
failure is replaced by partial safety factors. Using these partial safety factors, the analysis is simplified to
running a single calculation of the mathematical expression that links the degradation mechanism with the
concrete and exposure characteristics. The output of this modelling is generally the required minimum
cover thickness of a given concrete characteristics (chloride diffusivity and cementitious
composition/content) or the required concrete performance at a fixed cover thickness for the specified
design life.

It is recommended that the partial safety factor approach be adopted where


 deemed to satisfy requirements are not provided, e.g. exposure class U
 deemed to satisfy requirements are not sufficient, e.g. in coastal or marine exposures
 deemed to satisfy requirements may not adequately take account of the changing nature of
materials being used, e.g. the nature of GP (OPC) or blended cements
 deemed to satisfy requirements do not take account of the use of high performance materials, e.g.
SCM’s, galvanised reinforcement and inhibitors in marine exposures
 exposures are more severe than would be expected for the deemed to satisfy provisions, e.g. high
carbon dioxide levels in road tunnels or use of SCMs having normally higher carbonation rates
 design life is longer than 50 years
 the measured performance of concrete is lower than might be expected for the deemed to comply
requirements (e.g. the quality of as placed concrete is measured because it is suspect and
performance is found to be poor)
 measures for avoidance of deterioration are uneconomic

What is frequently missing is an understanding of the reliability that the answer provides. Clearly a factor
that accounts for different required reliabilities should be included in the calculations. Without a clear
understanding of what the partial factors mean in terms of the reliability of the life expectancy predicted by
the model the modelling has a limited value.

The reliability assessment for partial factors will be discussed in general in Section 2 for some common
reliabilities required in durability design using any modelling method. Z7/01 (Reference 21) also provides
some fundamentals of reliability-based durability design.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 6


1.3 DETERIORATION CATEGORY AND SCOPE OF DURABILITY MODELLING
Reinforced concrete structures can be deteriorated under various mechanisms in various
conditions and materials characteristics. These deterioration mechanisms of reinforced concrete
structures have been categorised in fib Bulletin 59 - 2006 (Reference 22) as shown in Error!
Reference source not found.

Figure 2. fib Bulletin 59 notes that in Europe, reinforcement corrosion damage is the most common form
of premature deterioration reported and this is mostly due to chloride penetration. The de-icing salt and
seawater salt in marine conditions are the major sources of chlorides. In Australia reinforcement corrosion
is also the predominant form of premature deterioration but here the salts from marine and coastal
exposures are the main issue.

Consequence Type of attack Mechanism

Carbonation induced corrosion


(CO2 atmosphere)
Reinforcement
corrosion Electrochemical
Chloride induced corrosion
(de-icing salts, sea water)

Internal expansion
(sulfate attack)
Chemical
Dissolution of concrete
(acid attack)
Concrete
deterioration
Internal expansion, scaling
(freezing)
Physical
Abrasion
(mechanical attack)

Figure 2, Deteriorations and consequences for reinforced concrete structures


by fib Bulletin 59 - 2005 (Reference 22)

The depassivating mechanisms for initiation of reinforcement corrosion by chlorides and carbonation are
reasonably well understood. The associated mathematical models used to predict the time of the
corrosion initiation period are generally agreed. Chloride ingress into concrete is primarily under four
mass transfer mechanisms, including diffusion, sorption (convection), permeation and wick action.
Carbonation rate is controlled by the rate of ingress carbon dioxide gas and its reaction with calcium
hydroxide.

Many factors control the corrosion rate of depassivated reinforcing steel. Some examples include
electrical resistivity of concrete between the anode and cathode, anode to cathode surface area ratio and

Durability Modelling of Reinforcement Corrosion in Concrete Structures 7


oxygen availability. Although these corrosion factors are understood, the corrosion rate still cannot be
predicted accurately due to very complex materials characteristics. Moreover, on-site, but also in the
laboratory, the actual corrosion rate cannot be measured/determined accurately. As a result, there is no
widespread agreement on how this process should be modelled. Therefore, modelling the propagation
process is still not very reliable in general. However some initial modelling methods are introduced in this
document. Where corrosion rate modelling is adopted high partial factors should be commensurate with
the lack of reliable modelling methods.

There are many other deterioration mechanisms, such as acid attack, sulphate attack and leaching,
associated with the ingress and reaction of contaminants. The models to predict the rate of these
deterioration processes have been proposed in the literature. However, there is not a common agreement
on these models and in many cases the input parameters are not available for the design process.
Therefore, these models are not discussed in this current issue of the durability series.

The mechanisms for internal deterioration processes such as alkali aggregate reaction (AAR) and
Delayed Ettringite Formation (DEF) are also reasonably well understood. However, models to predict the
deterioration rate are not available or not commonly agreed. Durability design is generally based on the
avoidance of deterioration approach by using compatible aggregate and cement systems for ASR control
and limiting peak concrete temperature for control of DEF.

In summary, this document is mainly dealing with modelling of the corrosion of reinforcement and
stressed tendons in concrete structures, including corrosion initiation period by either chloride diffusion or
carbonation and propagation period with corrosion reinforcement. Concrete deterioration mechanisms
other than these will not be covered in this document.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 8


2. GENERAL DETERIORATION MODEL FOR REINFORCEMENT CORROSION

Modelling of reinforcement corrosion in durability design of new concrete structures or in condition assessment
of existing structure requires various input parameters. Materials and exposure inputs are generally chosen by
the durability consultant undertaking the design as a thorough understanding of concrete durability is required in
order to use the mathematical models. These aspects are described in Sections 3, 4 and 5 which refer to
specific mathematical models. These specific models have a common underlying principle of design for service
life adopted by Tuutti (Reference 23). This is discussed in Section 2.1. Other inputs into the mathematical
models are selected specifically for projects to meet client requirements. For example, design life and reliability
are defined by the client or interpreted by the durability consultant based on the client’s general expectations.
This is also discussed in Z7/01 (Reference 20).

2.1 TUUTTI MODEL

The deterioration process can be approximately divided into three stages as shown in Figure 3 based on
the model by Tuutti (Reference 23). The model can be applied to many deterioration processes but Tuutti
applied it originally for reinforcement-corrosion due to chloride penetration or carbonation.
Corrosion Damage Level

Corrosion Severe
Propagation Damage

Corrosion Initiation

0 T0 T1 T2

Time in Service, Years

Figure 3, General deterioration model for reinforcement corrosion in concrete

Stage 1 (T0) is termed as the corrosion initiation period. Its length depends on the specific exposure
environment and the concretes resistance to the environmental loading. For reinforcement corrosion it is
taken as the time from construction until depassivation of the reinforcement (e.g. chloride ions reach the
corrosion activation threshold or carbonation breaks down the reinforcements passivity).

Stage 2 (T1) is termed as the corrosion propagation period. In this period, damage occurs at a certain
rate until a critical level of damage is reached. In the original Tuutti's model this was considered as
cracking or spalling of the concrete and was taken as the end of service life, i.e. the limit of acceptable
damage was defined as cracking and spalling.

Stage 3 (T2) is introduced as a period of accelerating damage that was not included in the original Tuutti's
model. In the case of reinforcement corrosion, cracking and spalling facilitates a more rapid corrosion
rate leading to significantly reduced cross-sectional area of reinforcement so that ultimately the structure
may become unsafe to use.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 9


2.2 LIMIT STATES

Concrete deterioration is generally considered to lead to Serviceability Limit State (SLS) failures although
in some cases it can lead to Ultimate Limit State (ULS) failures if it goes un-observed until the
consequence of failure becomes catastrophic (e.g. collapse). In the original Tuutti model the end of
design life was considered as when cracking and spalling occurred. However, for example concrete with
GP (OPC) cement in severe exposures where corrosion occurs the T1 period was generally very short
and many designers used the T0 period as the design life. On some structures it was recognised that
cracking and spalling were not critical and collapse with an adequate reliability against it occurring could
be used as the design life.

This has led to an understanding that there can be several forms of limit state and that within these forms
different states can be defined as the applicable limit state. In essence a limit state can be set to suit
whoever sets them fib Bulletin 59 - 2006 (Reference 22). The modeller has to interpret the owner’s
requirements into limit states and associated reliabilities that can be applied in models.

fib Bulletin 59 - 2006 (Reference 22) gives the following common limit states:
 A specific condition (e.g. damage such as cracking)
 Timing to an intervention (e.g. time to apply protective coating so that T0 will not be reached)
 A defined performance requirement (e.g. chloride level at a certain depth)

All of these limits would be applied to a design life and reliability requirement, e.g. the risk of failure shall
be less than 5% that cracking will occur within a 50 year design life.

It is recommended that for each durability design undertaken by modelling, the following be provided for
the basis of the design to be undertaken:
 Design service life
 Limit State applicable at the end of the design service life
 Reliability required for that limit state

2.3 DESIGN LIFE

During the durability design process, the design life can be defined based on the stage (or level) of
deterioration process at the end of the design life. In one case, the design life can be defined as T0 with
no reinforcement corrosion allowed during the entire service life. This is a very conservative approach
with a higher reliability and a minimum demand for inspection, maintenance and remediation. The
durability design process will only involve determining the T0 value in this case.

Alternatively, the design life can be defined as the sum of initiation period (T0) and the propagation period
(T1), shown in Equation 1. In this case, both T0 and T1 need to be determined during the durability design
processes. In severe exposures (e.g. marine tidal and splashing zones) using GP cement T1 is relative
shorter than the T0 and determination of T0 has been taken as the most critical activity in the durability
design process in many cases. However, in some other cases (e.g. with carbonation or marine
submerged condition, and internal exposures) T1 can be relatively longer and is included in an even
deemed to satisfy approach.
TL  T0  T1 (1)

Where

TL is the service life, years

T0 is the length of corrosion initiation period, years

T1 is the length of corrosion propagation period, years

Durability Modelling of Reinforcement Corrosion in Concrete Structures 10


For reinforcement corrosion the T0 value can be determined by modelling the chloride
diffusion/carbonation rate and durability design for T0 is common. Modelling reinforcement corrosion in
the corrosion propagation period have not been fully developed and applied due to insufficient accuracy
and reliability of models. However, as it can be a significant aspect of durability design, and does form
part of the design life in dry carbonation environments and for immersed concrete, it is also discussed in
this document.

The most critical modelling process to obtain an accurate prediction of T0 is to select the correct
mathematical formulas for a specific deterioration mechanism (chloride ingress or carbonation), the
suitable values for the input parameters, and the modelling conditions (assumptions). Incorrect models,
input parameters or assumptions will undoubtedly lead to predicting either too short a service life or too
long a service life at high costs.

Many different models have been developed which often give different prediction results for chloride
ingress and carbonation rate. In addition, various input parameters and assumptions are used by
different modellers, which further differs the modelling outcomes. Therefore, there is a need in the
construction industry to have guidance on which models should be used in durability design and what
input parameters/assumptions are suitable for various situations. This Recommended Practice provides
such guidance for durability modelling. Following the recommendations given will improve the accuracy
and reliability of modelling results and will achieve a better consistency throughout the construction
industry.

In general the variations and uncertainty in materials characteristics and achieved durability measures
during construction are such that the models only give and a rough approximation of the design life. This
lack of precision due to variations of input parameters is taken into account by the use of partial factors or
statistical distributions of input parameters used in the modelling.

Normally, structure owners should specify a target design life based on their economic and financial plan
for the asset. If the clients are not specifying design life for their asset, design life can be selected based
on guidance in Z7/01 (Reference 21). A long design life (i.e. over 100years) may be attractive to the client
considering the long term benefits. However, most models currently used may not be very reliable to
predict durability performance with such a long design life and the increased limitations associated with
the modelling should be made clear to the owner.

2.4 RELIABILITY ASSESSMENT

Reliability in the context of service life design is defined as the probability of a structure or an element to
perform its required function or performance satisfactorily over its specified design life under the specific
environmental and operating conditions. The reliability is the direct output when the full probabilistic
modelling approach is applied. However in the partial safety factor modelling approach, reliability can be
indicated by the reliability index (β) calculated using the following method.

As an example, for a normal distribution of the resistance (R) and the load (S), the reliability index can be
calculated using Equation 2 below (Reference 24). For the full probabilistic approach, a common
example of the resistance can be the predicted design life by modelling while that of load can be the
design life required. For the partial safety factor design approach, the resistance can be the selected
concrete cover while the load can be the minimum cover determined by modelling for the given durability
characteristics of concrete.

R  S
 (2)
( R2   S2 ) 0.5

Where

µR is the mean values of the resistance (R),

µS is the mean value of the load (S),

Durability Modelling of Reinforcement Corrosion in Concrete Structures 11


σR is the standard deviation of the resistance,

σS is the standard deviation of the load.

The reliability index (β) is used to assess reliability or failure probability. Some examples of reliability
index values and the associated reliability and failure probability are shown in Table 1.

Table 1, Reliability index and reliability

β Reliability, % Failure Probability,%


0.5 69.1500000 30.850000
1.3 90.3200000 9.6800000
1.5 93.3193000 6.6807000
1.8 96.4070000 3.5930000
2.0 97.7250000 2.2750000
2.3 98.9300000 1.0700000
3.0 99.8650100 0.1349900
3.6 99.9840890 0.0159110
3.8 99.9927652 0.0072348

Different levels of required reliability may apply through the structures life. The reliabilities for different
limit states may be applied and checked by modelling. For example all of the following reliability indices
as shown in Table 2 could apply to one element.

Table 2, Possible limit state requirements for one element


Limit State Type of LS Service Life  (min) Reliability, %
1 Corrosion initiation SLS 25 0.5 69.15
2 Corrosion initiation SLS 50 1.3 90.32
3 Cracking due to corrosion SLS 50 2.3 98.93
4 Structural failure ULS 50 3.6 99.99

These requirements show a low reliability (69.15~90.32%) against corrosion initiation at an earlier period.
The owner may accept a high risk that a protective coating be applied at 25 years as the sunk cost to
cover this would be small. A higher reliability (98.93%) against cracking is required because the cost of
arresting active corrosion is high. The highest reliability (99.99%) is required for ultimate limit state
against structural failure. In reality, this stage is never reached in most elements as remedial measures
are taken long before this type of failure occurs. However it would be appropriate, for example, to apply
such high reliabilities to the design of prestressed beams where cables could fail without visible signs
leading to a catastrophic failure.

The reliability assessment is more accurate in the full probabilistic design approach which considers the
variability of all parameters than that in the partial safety factor design approach which considers only the
variability of one or very few parameters.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 12


3. MODELLING CHLORIDE DIFFUSION IN CONCRETE

3.1 INTRODUCTION

Deterioration related to reinforcement corrosion in concrete due to penetration of chloride ions is the single
biggest cause of damage to reinforced concrete structures in the infrastructures across the world (Reference
22). This is also true for the Australian situation. Major sources of chlorides in Australia are:
 seawater;
 air borne chloride in marine and coastal areas;
 saline soil and groundwater;
 industrial water containing chloride ions (e.g. mine process water); and
 melted snow containing de-icing salt (seldom occurring in Australia).

There are other minor sources of chloride ions that are not be listed here.

Chloride ions can ingress into concrete under various transport mechanisms as listed below:
 Diffusion of chloride through the pore water due to a chloride concentration gradient
 Permeation under a differential hydraulic pressure
 Absorption (sorption or convection) into unsaturated concrete due to a differential moisture concentration
 Wick action with evaporation leaving high salt concentration at drying front

Among them, diffusion of chloride ions into concrete is often considered a dominating mechanism in various
conditions while other transport mechanisms may only occur in some specific situations. Therefore, prediction
of the chloride diffusion process is a key part of durability design for reinforced concrete structures exposed to
chloride. Modelling chloride diffusion is a very effective way to predict the time to corrosion initiation.

Various laws governing the diffusion of chlorides into concrete will be discussed in this section and the
derivation of various chloride ingress models will be outlined. Fick’s First Law (Reference 25) defines the
chloride steady-state diffusion process where the chloride concentration across the depth of thin concrete
samples and the diffusion parameters are constant. This is not applicable to modelling thick concrete exposed
to chloride on one face such that the chloride profile through the sample is increasing.

Fick’s Second Law (Reference 25) defines the chloride concentration change across a sample with time where
there is a differential chloride concentration. Its solutions for semi-infinite exposure condition is more suitable
for chloride modelling in concrete. Early chloride diffusion models for concrete assumed a constant diffusivity
and used the appropriate solution for Fick’s Second Law. However this solution is not correct mathematically
for concrete that has a decreasing diffusivity. To account for decreasing diffusivity two groups of models based
on Fick's Second Law have been developed. One group represented by Tang and Nilsson (Reference 26) and
TM
Life 365 (Reference 27) uses an instantaneous diffusivity approach. The other group represented in TR61
(Reference 28), fib Bulletin 34 (Reference 3), fib Model Code 2010 (Reference 1), ISO 16204 (Reference 5)
uses an apparent diffusivity approach.

The apparent diffusivity approach uses aging factor derived by curve fitting chloride profiles from structures of
different ages and concrete binder compositions in the DuraCrete project. The resultant model is an empirical
one and is not mathematically correct. However, it may still give reasonably reliable modelling results as long
as:

 The period being modelled is not significantly longer than the exposure period covered by the data used
to determine the aging factors

 Material changes since the time of construction of the structures sampled to produce the ageing factors
are similar to the materials being used in the structure being modelled. A question arises as to whether
aging is due to reducing quality of materials or genuine aging improvements.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 13


 Exposure conditions applicable to the location being modelled closely match the exposure of locations
for which ageing factors were derived.

For a decreasing instantaneous diffusivity, analytical and numerical modelling approaches were developed. The
analytical model developed by Tang and Nilsson (Reference 26) uses the correct solution to Fick’s Second Law
for a decreasing chloride diffusivity. This is easy to apply in a spreadsheet and can be simply used for chloride
TM
modelling in concrete. The numerical model, e.g. Life 365 (Reference 27), uses a finite difference approach
and although not easily applied, it is a free program that can be used to provide a solution.

There are various other limitations to the use of the above models, even where they correctly model the
decreasing diffusion coefficients e.g.:
 They are all based on measurement of total chloride concentrations. Although it is only the unbound (or
free) component of the chloride ions that penetrate, the use of total chlorides in diffusion calculations is a
reasonable approach since bound, adsorbed & free chlorides are in equilibrium.
 The chloride diffusion test used to match diffusion mechanisms in the actual structure takes a long time
resulting in test data available being too limited for reliable statistical assessment. A rapid test, called the
rapid chloride migration test, does not give results matching chloride diffusivity due to different governing
laws of test conditions.

The remainder of this section provides details of these models and their suitable inputs for predicting the
diffusion of chloride ions into concrete.

3.2 BASIC MODELS OF CHLORIDE DIFFUSION

3.2.1 Fick's First Law

Chloride ions in an aqueous solution move randomly due to their frequent collisions with the water
molecules. This type of movement is called ‘Brownian motion’ (Reference 29). Such random Brownian
motion statistically drives the chloride ions from a high concentration zones to a low concentration zones
and over time increases the uniformity throughout system. Such a mass transfer of chloride ions from a
high concentration zone to low concentration zone is called ‘diffusion’.

The diffusion flux of chloride ions in an aqueous solution is governed by Fick's First Law (Reference 25),
as described in Equation 3 for the typical one-dimensional diffusion process in a steady-state system, i.e.
where there is no change of chloride concentration at any point along the diffusion path. It can be seen
from Equation 3 that the diffusive flux is proportional to the concentration gradient (spatial derivative) and
the concrete diffusivity.
C
J  D (3)
X

Where
2
J is the diffusion flux (or rate of mass transfer) through unit area in unit time, mol/(m .s)
2
D is the diffusivity or diffusion coefficient of chlorides in the aqueous solution, m /s
3
C is the concentration of chloride ions, mol/m

X is distance, m (or metre)

This equation is considered valid also for concrete with a capillary pore water system having a certain
level of connectivity. Diffusion of chloride ions in solid cement paste is significantly slower than that in the
pore water and thus it can be ignored. Diffusion of chloride ions also cannot occur in dry air and thus in
concrete without continuity of pore water in a dry condition, diffusion of chloride ions will not continue.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 14


The diffusion mechanism in the concrete pore water system is essentially the same as that in bulk water
however the diffusion rate is much slower due to the low cross-section area and the longer diffusion paths
(Reference 30).

3.2.2 Diffusion with Free and Bound Chlorides

The actual chloride diffusion in concrete is not just a simple diffusion process however. Chloride ions can
be chemically bound by some cement hydration products to form a new product such as Friedel's Salt
(e.g. Ca3Al2O6·CaCl2·10H2O). Chloride ions may also be physically adsorbed by van der Waals force to
C-S-H gel. C-S-H gel has a high specific surface area and can hold significant amounts of chloride.

The concentration of bound/adsorbed chloride is in equilibrium with the free chloride concentration in pore
water. The bound and adsorbed chloride content will change with chloride concentration in the pore
water. In normal condition, the bound and adsorbed chloride ions do not diffuse unless the equilibrium
between free and bound chloride ions breaks in special circumstance. As the concentration gradient of
free chloride ions are the driving force for the diffusion process, an increase in bound and adsorbed
chloride will reduce the available free chloride diffusion flux into the concrete.

A linear relationship between concentration of bound/adsorbed chloride and the free chloride in pore
water is assumed in most common chloride diffusion models including fib Bulletin 34 - 2006 (Reference
TM
3), ISO 16204 (Reference 5), and Life 365 (Reference 27). In practice this is known not to be the case.
The amount of bound/adsorbed chlorides increases non-linearly and isotherm curves can be used to
describe the relationship between bound/adsorbed chloride content and the free chloride content in pore
water (Reference 3). Tang (Reference 31) notes different curves appear to fit data over different
3
concentration ranges but over the range 0.01-1 mol/dm it is appropriate to use the Freundlich isotherm
(Reference 32) which has the form:

Cb  K b C n (4)

Where

Cb is the concentration of bound chlorides,

Kb is the chloride binding/adsorption capacity of concrete,

n is the binding intensity parameter.

Spiesz (Reference 33) reports on values for Kb and n for cement pastes and mortars. Values for n
remained relatively constant (0.31 to 0.45) but Kb values were significantly influenced by the type,
composition and content of cementitious materials, w/c ratio and age of sample. Spiesz comments on
Kb[days;w/c] as follows :

 Kb is much smaller for mortar than cement due to the volume of cement hydration products (Kb[42;0.4]
= 2.14 for cement paste and 0.53 for mortar).

 Binding capacity increases with C 3A content of the GP cement (Kb[60;0.5] = 2.97 for T10 cement
paste and 2.40 for T20 cement paste).

 GP cement with 25% slag had higher binding capacity than GP cement alone (Kb[270;0.5] = 2.65 for
25% slag and T20 cement paste and 2.13 for T20 cement paste).

 GP cement with 25% fly ash had a lower binding capacity than GP cement alone (Kb[270;0.5] = 2.00
for 25% Fly Ash with T20 cement paste and 2.13 for T20 cement paste).

 GP cement with 8% silica fume had a lower binding capacity than GP cement alone (K b[270;0.5] =
1.52 for 8% silica fume with T20 cement paste and 2.13 for T20 cement paste).

Durability Modelling of Reinforcement Corrosion in Concrete Structures 15


 Binding capacity reduces with age.

(Type T10 and T20 cement are Canadian Standard cements. They are equivalent to Australia’s GP
cement and a low C3A type sulphate resisting (SR) cement respectively.)

The binding capacity is not constant with time and the period of testing time could account for differences
in opinion as to whether the use of fly ash replacement for GP cement for example increases or
decreases the amount of bound chloride relative to a GP cement only. Arya et al (Reference 34) found
that fly ash and slag increased the bound chloride proportion while silica fume decreased the bound
chloride proportion compared to GP cement.

Spiesz (Reference 33) also notes that the chloride binding is not a single step process. The chlorides
pass from the liquid phase through a transition zone phase at the liquid solid interface and then from the
transition zone to the solid. The resistance to mass transfer at the transition zone causes a concentration
gradient in each phase and this results in a 7-14 day time lag to achieving equilibrium between bound and
free chlorides. Hence, for diffusion tests undertaken over at least 35 days equilibrium is achieved but in
rapid migration tests over 6-96 hrs, it is not. This means that diffusion coefficients from diffusion tests e.g.
NT Build 443 (Reference 35) cannot be the same as the results from migration tests, e.g. NT Build 492
(Reference 36).

Despite non-linear binding Equation 3 can still be used reliably to predict chloride diffusion in concrete
using total chloride concentration as the total chloride flux is broadly proportional to the free chloride flux
since bound, adsorbed and free chlorides are in equilibrium. Analysis results using models developed
based on Equation 3 are considered to be valid within current accuracy limits for concrete although it is
recognised that significant strides are being made to improve predictions using models based on non-
linear chloride binding isotherms, and non-equilibrium conditions between free and bound/adsorbed
chlorides (Reference 33).

The validity of any models based on Equation 3 may be disturbed by carbonation, at least in the near
surface zone. Chlorides bound as Friedel's Salt are released by the carbonation reaction. However,
carbonation is very slow in high moisture content concrete typical of marine structures where the highest
rates of chloride diffusion occurs. Hence, typically for marine structures carbonation has little effect on the
use of such models.

In less common situations, such as lower performance chloride contaminated concrete where carbonation
releases the bound chlorides, the total chloride threshold level is reduced. This begins to occur in front of
the carbonation front indicated by a drop to pH 9 indicated by the phenolphthalein test. Hence, corrosion
may be initiated even in front of the carbonation front, particularly as the concrete’s pH is also reducing in
front of the carbonation front. Under these combined conditions of carbonation plus chlorides, the normal
chloride diffusion modelling methods are of little value and the chloride threshold level needs to be
reduced to a more suitable level.

3.2.3 Fick's Second Law

The change of chloride concentration with time in aqueous solution, and in concrete at various locations,
in the one-dimensional diffusion process can be described by Fick’s Second Law of Diffusion (Reference
25) as shown in Equation 5. It can be seen that the rate of chloride concentration change is proportional
to diffusivity and the second derivative of concentration function with distance.

C  2C
D (5)
t X 2

Where

t is the time, s (or second).

Durability Modelling of Reinforcement Corrosion in Concrete Structures 16


The mathematical solution of Equation 5 for a semi-infinite homogenous medium with the finite face
exposed to a chloride source of constant concentration was provided by Crank (Reference 37) as shown
in Equation 6. It was assumed that initial (or background) chloride concentration in concrete is zero in
this equation. This mathematical expression can determine the chloride concentration change with time at
various distances from the exposed surface. It means that the chloride concentration in the concrete
diffusion zone is proportional to the surface chloride concentration.
C X
 erfc( ) (6)
Cs 2 Dt

Where

erfc is the complement error function,

Cs is surface chloride concentration.

This equation was firstly applied to concrete by Collepardi et al (Reference 38) to model chloride
concentration change with time at various depths from the concrete surface exposed to an external
chloride solution. Since then, many researchers have used it in predicting the diffusion process in
concrete and to design for durability of chloride affected concrete structures.

It has also been used to determine chloride diffusivity based on a chloride profile obtained from a
concrete sample either exposed in the laboratory or in the field with other terms in the equation being
known. This equation is a basis for many test methods, e.g. NT Build 443 (Reference 35), NT Build 492
(Reference 36), and ASTM 1556-2011 (Reference 39) to quantify chloride diffusivity (or diffusion
coefficient).

For concrete, the chloride concentration unit is generally measured as the total chloride content
percentage (%) by mass of total concrete. This is sometimes converted to a percentage by mass of
cement content in the diffusion modelling because in some cases the corrosion activation level has been
presented as percentage of cement content. This conversion requires an assumption or measurement of
the cement content in concrete.

During modelling, it makes no difference if the chloride concentration as a percent of concrete mass is
converted to that as a percent of cement mass or the activation level as a percent of cement mass is
converted to that as a percent by concrete mass so long as both have the same conversion basis.
However there is increasing preference to use percent of concrete mass because such a chloride
concentration is measured on the concrete samples.

An example of a chloride profile calculation is given in Figure 4. The input parameters are shown as the
-13 2
notes in the figure. These include a chloride diffusivity of 3.0x10 m /s, a surface chloride concentration
of 0.6% by concrete mass (not shown in vertical axis title of the figure) and an exposure time of 50 years.

3.3 INITIAL CHLORIDE CONCENTRATION (BASE CHLORIDE LEVEL)

In many cases, raw materials of concrete contain some chloride ions and the initial chloride content in
concrete is not zero. Therefore Equation 6 cannot be used directly to predict the chloride diffusion
process into concrete. In such a condition, chloride diffusion can be predicted by Equation 7 derived by
Crank (Reference 37). This equation indicates that the difference between chloride concentration in
concrete and the background concentration is proportional to the difference between chloride
concentration at the concrete surface and the background concentration.
C  C0 X
 erfc( ) (7)
C s  C0 2 Dt

Where

Durability Modelling of Reinforcement Corrosion in Concrete Structures 17


C0 is the initial chloride concentration in concrete.

Figure 5 shows the chloride profile predicted with Equation 7. The initial chloride concentration is
assumed to be 0.015% by mass of concrete. Other parameters are the same as those in Figure 4. It can
be seen that the chloride profile always starts from the initial chloride concentration.

0.7
Input Parameters
0.6
Chloride Concentration, %

D = 3x10-13 m2/s
Cs = 0.6%
0.5 t = 50 years

0.4

0.3

0.2

0.1

0.0
0 10 20 30 40 50 60 70 80 90 100
Distance from Surface, mm
Figure 4, Chloride profile calculated using Equation 6

0.7
Input Parameters
0.6
Chloride Concentration, %

D = 3x10-13 m2/s
Cs = 0.6%
0.5 t = 50 years
C0 = 0.015%
0.4

0.3

0.2

0.1
0.015%
0.0
0 10 20 30 40 50 60 70 80 90 100
Distance from Surface, mm
Figure 5, Chloride profile calculated using Equation 7

The base chloride level in the concrete essentially shifts the zero location of the vertical axis for the
diffusion analysis and hence can easily be manipulated in the diffusion calculation. Its main significance is
that it reduces the increase in chloride level required to cause activation. The impacts of this initial
chloride content should be assessed both in durability design of new structure and the condition
assessment of existing structures.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 18


In durability design for new structures, initial chloride content in concrete has a significant impact on
3
service life. AS 1379-2007 (Reference 13) stipulates a maximum chloride content of 0.8 kg/m (or 0.03%
3
by concrete mass, assuming a concrete bulk density of 2400 kg/m ). This chloride content is only slightly
lower than common threshold chloride concentration of 0.06% by mass of concrete. A lower initial
3
chloride concentration of 0.4 kg/m is commonly specified where chloride ingress might be an issue
during durability design and this significantly increases the reliability with all other factors remaining the
same.

In the condition assessment of concrete structures to predict the time to corrosion initiation and
consequently the residual service life, the apparent (or average) chloride diffusivity estimated from the
chloride profile is significantly influenced by the initial chloride content, A higher initial content can yield a
lower diffusivity value than a lower initial content. For the same profile, a lower diffusivity would predict a
longer time for corrosion initiation or the residual life. Furthermore, the initial chloride content should be
tested from a location not contaminated by the external chloride. It is common to test such content at a
location 120 mm away from the exposed surface.

The base chloride content is generally treated as a constant with no distribution for full probabilistic
analysis and the base level is taken as the limit applied by the applicable national code or project
specification. This introduces an element of conservatism in the analysis that could be simply overcome
by using data from the concrete supplier on the mean chloride level and its distribution expected for the
mix based on past records.

It is recommended that before undertaking chloride ingress modelling for design purposes a clear
understanding of the expected base chloride level be established and the distribution of that base level be
included where possible in full probabilistic models. For the partial factor approach the expected mean
chloride level can be used to give the 50% reliability case.

3.4 SURFACE CONVECTION ZONE

In the above idealised diffusion situation, the highest chloride concentration should be located at the
concrete surface considering that diffusion is driven by a concentration differential. However in actual
tests on in-situ concrete structures, it was found by Ann et al (Reference 40), Meira et al (Reference 41),
Gjørv & Vennesland (Reference 42), and Farstad et al (Reference 43) that the concentration at the
surface is often lower than the expected chloride level and sometimes even lower than that at a depth.
Figure 6 shows an example of a diffusion curve with low total chlorides in the surface zone than expected.
However, the total chloride concentration in the deeper zone fits well with a typical diffusion curve.
There are complex mechanisms that explain this but they are not mathematically well-defined.
Contributing factors are thought to be:
 Where a chloride solution is repeatedly applied to and absorbed into the surface during each drying
period, the water evaporates leaving an increased chloride content to a certain depth that is not
diffusion based;
 Wetting processes (e.g. tidal, splash, rain) wash out some free chlorides at the surface reducing
the free chlorides and thereby the total chlorides.
 Carbonation of the near surface concrete releases some bound/adsorbed chlorides. This does not
directly affect the total chloride content but will increase the chloride driving force.

This near-surface zone where data does not fit the typical diffusion curve is often called the ‘surface
convection zone’. This surface convection zone can lead to a fast chloride penetration into concrete than
the typical diffusion process due to not only the fast absorption process at the surface zone but also the
relatively higher surface concentration caused by the drying process. fib Bulletin 34 (Reference 3)
adopted a mathematical approach to deal with the effect of convection zone. In this case, the term ‘X’ in
Equation 7 is replaced by the term (X-∆X) to form a general model, i.e. Equation 8.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 19


C  C0 X  X
 erfc( ) (8)
C s  C0 2 Dt

Where,

∆X is the depth of the convection zone, mm.

0.7
Poor f it zone Actual Test Data
0.6 By Equation 7
Chloride Concentration, %

0.5
Input Parameters

0.4 D 56d = 3x10-13 m2/s


C10mm = 0.45%
te = 50 years
0.3
C0 = 0.015%
0.2

0.1 Good f it zone

0.0
0 10 20 30 40 50 60 70 80 90 100
Distance from Surface, mm
Figure 6, Actual chloride profile vs idealised profile

The convection zone depth (∆X) varies with concrete porosity and exposure conditions. According to fib
Bulletin 34, the mean convection zone depth is normally ranging from 6-11 mm in tidal, spray and
splashing marine zones and in coastal conditions with rain-washout action. High performance concrete
with silica fume may have a depth below 5 mm. In ISO 16204 (Reference 5) the ∆X term is not included
on the basis that it is insignificant in high performance marine concrete and is accounted for in the
calculation of apparent diffusion coefficients. However, here it is recommended that consideration always
be given to the potential for a convection zone. In immersed zones, the depth of convection zone is
estimated to be zero. Low performance concrete exposed to a long drying period following a wet process
may have a depth greater than 11 mm.

For full probabilistic modelling, normal distribution can be considered for this parameter. The coefficient
of variation can be considered to be 20% if more reliable values are not available prior to the modelling.

For diffusion beyond the convection zone, Equation 8 can still be valid to describe the chloride profile.
Such a model can be used both in the durability design process for new structures and in the condition
assessment process of existing structures provided a suitable depth is established in the specific
condition.

However, for condition assessment of existing structures, a minimum five test depths beyond the
convection zone (i.e. within the saturated diffusion zone) should be considered. This is to clearly define
the depth of convection zone and to avoid having insufficient data points within the diffusion zone during
the profile analysis. If one or more test points in convection zone are used in the profile analysis, it would
over-estimate the apparent diffusivity and predicted a significantly shorter residual service life.

It is recommended that:

Durability Modelling of Reinforcement Corrosion in Concrete Structures 20


 for concrete in splash and atmospheric zones of marine structures where a 50 MPa concrete using
an SCM is used, the convection zone be taken as zero depth unless there is clear contrary
evidence
 in locations where there are long wet, long dry cycles the depth of the convection zone be
established as part of any modelling
 in immersed zones chloride diffusion modelling is not undertaken as the corrosion is controlled by
oxygen availability and hence any convection zone caused by surface effects is immaterial
 elsewhere an estimate of the convection zone be included in the chloride ingress modelling based
on guidance given above, and
 where chloride profiles are established during condition assessment of exposed concrete, the
following procedures should be used:
1) the profile be established with at least four data points in the diffusion zone beyond convection
zone
2) a base chloride level be established at a 120mm depth where possible (the fifth data point),
and
3) the convection zone depth or carbonation depth (whichever deeper) be established and no
chloride data in these zones be included in the analysis.

3.5 DECREASING CHLORIDE DIFFUSIVITY

Equations 6 to 8 suit the conditions where chloride diffusivity (D) of concrete is constant. However, Tang
and Nilsson (Reference 26) noted that chloride diffusivity of concrete decreases with concrete age,
possibly due to continuous hydration of cement after construction with consequential densification of the
concrete pore structure. Reducing diffusivity will slow down the chloride ingress rate and Equations 6 to 8
would predict a deeper chloride penetration profile than would occur in practice.

There are two basic methods to model chloride diffusion with a decreasing chloride diffusivity in concrete:
 The first one is to consider decreasing the rate of the instantaneous chloride diffusivity in Equations
6 to 8 in either an analytical or numerical method.
 The second one is the empirical method to use an apparent chloride diffusivity and to consider its
decreasing rate with age.

3.5.1 Instantaneous Diffusivity Models

Instantaneous chloride diffusivity is the diffusivity value obtained from concrete at a particular concrete
age and is valid for only that age. Tang and Nilsson (Reference 26) found that the instantaneous chloride
diffusivity decreases with concrete age and the empirical relationship between the diffusivity and age is
given in Equation 9.
tr m
Da  Dr ( ) (9)
ta

Where

ta is the concrete age, year,


2
Da is the instantaneous chloride diffusivity at age ta, m /s,

tr is the reference concrete age at test or measurement, year


2
Dr is the reference instantaneous chloride diffusivity at reference concrete age of tr, m /s,

m is the instantaneous age factor.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 21


The instantaneous age factor of concrete depends on the cementitious materials type or compositions.
Concrete with Supplementary Cementitious Materials (SCMs) including fly ash (FA), slag (GGBSF) and
silica fume (SF) shows a much greater age factor than the concrete with GP cement (or OPC) only. The
environment condition also affects the age factor, with a higher value expected in wet conditions due to
greater late hydration and a lower value for dry conditions due to less late hydration.

Instantaneous age factors for various cementitious materials types and compositions have not been
systematically established. If there is no other more reliable value, the recommendations by ACI Life-
TM
365 model (Reference 27) shown by Equation 10 can be used to estimate the mean instantaneous age
factor for concrete with GP cement, fly ash and GGBS. For (condensed) silica fume, the age factor
estimated for fly ash is also considered suitable.

FA BFS
m  0.2  0.4  (  ) (10)
50 70

Where

FA is the fly ash content, %, (or SF content % or SF+FA content %)

BFS is the GGBSF content, %.

For full probabilistic modelling, the age factor can be considered to have a normal distribution. The
coefficient of variation can be considered to be 15% if there is no more reliable data to show otherwise.

3.5.2 Analytical Model for Instantaneous Diffusivity

Tang and Gulikers (Reference 44) provided a mathematical model to predict chloride diffusion in concrete
with a decreasing instantaneous diffusivity using a transformation method similar to that proposed by
Crank (Reference 37). The model has been modified here to incorporate an initial chloride concentration
and a convection zone depth forming Equation 11 below.
C  C0 X  X
 erfc[ ] (11)
C s  C0 Dr t t t
2 [(1  e 0 )1m  ( e 0 )1m ]( r ) m t e
1 m te te te

Where

te0 is the age of concrete when exposure starts, years,

te is the duration of exposure time, years. It is obvious that concrete age ta equals te0 + te.

Chloride profiles obtained from Equations 11 and 8 are given in Figure 7 for comparison. It can be seen
that the chloride ingress with decreasing instantaneous chloride diffusivity is slower than that with
constant diffusivity, resulting in a shallower chloride profile. At the assumed threshold concentration of
0.06% by mass of concrete, the depassivation depth decreases from 57 mm to 43 mm, i.e. reducing by
13 mm (or 23%) at an exposure time of 50 years. A higher age factor (0.4) gives a shallower diffusion
profile so that the depassivation depth decreases a further 13 mm to 30 mm compared to that with the
age factor of 0.2.

The length of the period with decreasing chloride diffusivity is not fully understood. A value up to 60 years
has been observed in concrete with Portland cement, 50 years with GGBS and 30 years with fly ash
(Reference 45). However, these reduced diffusivities may not be obtained on the same concrete at
various ages but they may be obtained for similar concrete but made in different times. Considering this,
a period of 25 years is considered to be safe in durability design with modern cements.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 22


It should be noted that Equation 11 can be used only to predict the chloride diffusion within the same
exposure period as the period used to develop the age factor. Predicting chloride diffusion beyond such
a period may overestimate durability performance of the concrete structure.

0.6
By Equation 8 Input Parameters
By Equation 11, m=0.2 D 56d = 3x10-13 m2/s
Chloride Concentration, %

0.5
By Equation 11, m=0.4 ΔX = 10 mm
Threshold C10mm = 0.45%
0.4 te = 50 years
ta0 = 28 days
C0 = 0.015%
0.3

0.2

0.1

0.0
30 40 1050 20 60 70 80 90
Distance from Surface, mm
Figure 7, Comparing chloride profiles obtained from Equations 8 and 11

3.5.3 Numerical Model for Instantaneous Diffusivity

Numerical modelling is a finite difference method to solve Fick’s Second Law. It also uses an
instantaneous chloride diffusivity to predict the diffusion process but in this case it is applied in small time
steps and the diffusivity can be adjusted at each step. This is a suitable method to model chloride
diffusion in concrete with a decreasing diffusivity.
TM
One commonly used numerical modelling program is ACI Life-365 model (Reference 27). It is suitable
to predict chloride diffusion in concrete with a decreasing instantaneous diffusivity. It is also flexible for
different periods with a decreasing instantaneous diffusivity. It is considered to be a reliable model under
such conditions. Historically there have been some issues with this model with different versions giving
TM
different results so users should ensure they have the latest version. Unfortunately the current Life-365
makes no provisions for an initial chloride concentration (C0) and for a convection zone (∆X). However,
the latter one can be taken into account by adding a convection zone depth to the required cover
thickness calculated in diffusion zone to obtain the total cover thickness. It is also not suitable directly for
full probabilistic modelling.

It is recommended that diffusion analysis be made using numerical modelling with


 m values to be applied for the first 25 years only for conservativeness; and
 the convection zone be added to the cover thickness where required.

3.5.4 Testing Instantaneous Chloride Diffusivity

There are a three principle methods used to determine the instantaneous chloride diffusivity or chloride
resistance. Each method has its advantages and disadvantages. Their details are discussed in detail in
Z7/07 (Reference 46) and briefly reviewed in this section to show how they are used in the chloride
modelling process.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 23


3.5.4.1 Classical Steady State Diffusion Cell

A concrete slice (25~50 mm thick but at least 1.5 times max aggregate size) is installed between
two chambers, i.e. upstream chamber with high chloride concentration and downstream one with
lime-saturated water (Reference 47). The chloride concentration in the downstream chambers is
monitored regularly (e.g. at 2-4 week interval) to determine the diffusion flux. After the chloride
increase on the down-stream face reaches steady state, the flux is used to determine concrete
diffusivity using Fick’s First Law.

It is an accurate test method. However, the test is time-consuming (up to 9 to 24 months) to


reach steady state across a thick sample while the diffusivity decreases during testing. The test is
primarily used to provide performance data on proprietary products such as repair mortars.

3.5.4.2 Bulk Diffusivity Test (NT Build 443-1995)

During the NT Build 443 (Reference 35) test, a chloride profile through a concrete sample
immersed within a bulk chloride solution after a period of 5 weeks is measured. The profile is
then used to determine the instantaneous chloride diffusivity using the semi-finite solution of
Fick's Second Law (Equation 7).

As NT Build 443 test conditions closely resemble the exposure case in real concrete, it is the
preferred test for assessment of a concrete’s diffusivity when using Equation 11. In addition, this
method is less time-consuming and diffusivity does not change much during testing.

However, the testing time is still very long for some projects with a short lead time. In the case of
high performance concrete, it may take up to 3 months at a high cost. It is suitable for projects
where there is a long lead time or where suppliers can provide data on mixes tested previously. It
is often not considered to be suitable for quality control or small projects.

3.5.4.3 Accelerated Chloride Migration Test (NT Build 492-1999)

The migration of chloride ions through a concrete sample is accelerated by applying an electrical
potential across the samples. The chloride flux can be determined from the chloride
concentration in a downstream chamber and used to determine the chloride penetration index
(Reference 47). This test takes approximately 2 weeks. As such, it is a too long for quality
control purposes.

Alternatively, NT Build 492-1999 (Reference 36) or ASTM C1556-2011 (Reference 39) is often
used to determine quickly the chloride diffusivity. During the test, an electrical potential is applied
to a 50 mm thick sample to drive the chloride ions through the sample during the test. The depth
of a chloride front is used to determine the chloride diffusion coefficient using the Nernst Plank
equation which governs in these test conditions. This is a fast test taking approximately 6-96
hours and is suitable for quality control in terms of time taken.

However, electrical potentials fields drive cations and other anions, which can give an error in the
results. The results obtained are not exactly the same as those obtained in a bulk diffusivity test
(e.g. NT Build 443). However, the results are still reasonably close between the two tests
(Reference 48).

To further improve the accuracy of the NT Build 492 test, Speiesz (Reference 33) proposes an
improved model for the migration test to consider the non-equilibrium effect between bound and
free chloride in short term tests with non-linear binding ratios. A much better correlation of results
is obtained between the two test methods with this model.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 24


Nevertheless, the above improvement has not been confirmed by other researchers. Therefore, it
is recommended that the correlation between results of the two test methods be established
during trial mixes, The results of NT Build 492 can then be used for quality assurance during the
construction.

Figure 8, Key details of chloride diffusion tests

These tests are shown schematically in

Figure 8 based on data from Tang and Truc (Reference 48). The diffusivities shown are Tang’s
results on the same concrete mix and indicate the diffusion cell gives very different results to the
NT Build 443 test.

3.5.4.4 Electrical Charges Passed

Another method often used to assess chloride resistance of concrete is the rapid chloride
permeability test (RCPT), e.g. ASSHTO 277 (Reference 49) and ASTM 1202 (Reference 50). It
measures electrical charges passing through a concrete sample under a given potential and time
duration. Based on the measured values, the concrete resistance to chloride penetration can be
rated, which are shown in Table 3.

Table 3, Rapid chloride permeability based on charge pass

Chloride Permeability Charge Pass (Coulombs)


High >4000
Moderate 2000-4000
Low 1000-2000
Very low 100-1000
Negligible <100

The relationship between electrical charge passed in RCP test and the chloride diffusivity by bulk
test (NT Build 443) has not been established systematically. However as an example, an

Durability Modelling of Reinforcement Corrosion in Concrete Structures 25


empirical relationship based on the test data by Hooton et al (Reference 51) can be established,
which is shown in Figure 9. The results are based on the following test conditions: wet-curing for
28 days in both tests before exposed to chloride solution and samples exposed to chloride
solution for 90 days in bulk diffusion test.

8
y = 0.00000030x2 + 0.00060141x + 0.93750729
7
Chloride Diffusivity, 10-12 m 2/s

R² = 0.98429208
6

0
0 500 1000 1500 2000 2500 3000 3500 4000
RCPT (Charge Passed), Coulums
Figure 9, Empirical relationship between chloride diffusivity and RCPT

3.5.5 Instantaneous Diffusivities Published

If there are no specific test data available at the time of modelling, the mean instantaneous chloride
diffusivity can be selected based on published results available for different cement systems and w/c
ratios as outlined in this section. For full probabilistic modelling, the diffusivity can be considered to have
a normal distribution. The coefficient of variation can be considered to be 20% if there is no more reliable
data available prior to the modelling.

3.5.5.1 GP Cement

For GP cement concretes, instantaneous chloride diffusivity can be determined using Equation
TM
12 based on the Life-365 model (Reference 27).

D28  110( 12.062.40W / C ) (12)

Where
2
D28 is the instantaneous chloride diffusivity of GP concrete at 28 days, m /s,
W/C is the water/cement ratio of concrete.

fib Bulletin 34 (Reference 3) gives also test data using the NT Build 492 method (Reference 36).
Results from these two sources and other sources are plotted in Figure 10. Note that the other
data included in the figure may not have the same test method, curing and exposure conditions.
TM
It can be seen that the values by fib Bulletin 34 and Life-365 are very close in general and they
represent approximately 75% upper limit of the other data. Therefore, both sets of values can be
used for modelling purposes.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 26


60
ACI Lif e 365

Chloride Diffusivity, 1012 m 2/s


50 FIB 34
Others
40

30

20

10

0
0.5 0.2 0.3
0.6 0.7 0.4 0.8
W/C Ratio
Figure 10, Comparison of Chloride Diffusivities of GP Concretes

3.5.5.2 Fly Ash

For concrete with fly ash, as shown in Figure 11, the relative diffusivity tends to decrease with fly
ash content up to 40% based on the data collected by Bamforth et al (Reference 45). The data
are scattered with most results ranging from 2% to 100%. The average values are given by the
trend line, which give a 50% possibility of higher or lower.

A new line in red colour is proposed here targeting the upper 75% limit which gives a 25%
chance of getting higher values but a 75% chance getting lower ones. If there is no more reliable
diffusivity, this upper 75% limit can be used to select relative diffusivity ratio for concrete with fly
ash.

Note that the red line values selected are only the relative values to GP concrete diffusivity. The
actual diffusivity values for fly ash concrete can be determined using the actual values for GP
concrete and the diffusivity ratio between fly ash concrete and GP concrete from the red line.

Alternatively, the diffusivity inputs by fib Bulletin 34 can be also adopted. The values are shown
in Figure 12. The regression equation can be used to determine the values for various w/c ratio.
The diffusivities were tested using NT Build 492 method. They are considered to be
conservative.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 27


Figure 11, Relative diffusivities of fly ash concretes

18

y = 187.57x2 - 116.15x + 23.26


Chloride Diffisivity, 10-12 m 2/s

15 R² = 0.9939

12

3
0.33 0.38 0.43 0.48 0.53 0.58
Water/Binder Ratio
Figure 12, Chloride diffusivity inputs of fly ash concretes from fib Bulletin 34

3.5.5.3 GGBS

For concretes with GGBS, as shown in Figure 13, the relative instantaneous diffusivity tends to
decrease with slag content up to 80% based on the data collected by Bamforth et al (Reference
45). The data are very scattered with most results ranging from 0.3% to 100%. The average
values are given by the trend line, which give a 50% possibility of higher or lower values.

A new line in red colour is proposed here targeting the upper 75% limit which gives a 25%
chance of getting higher values but a 75% chance getting lower ones. If there is no more reliable
diffusivity, this upper 75% limit can be used to select the relative diffusivity ratio for concrete with
GGBS.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 28


Similar to that for the fly ash concrete, the actual diffusivity value can be determined with the
actual values determined for GP concrete and the resistivity ratio between the two concretes.

Similar to fly ash concrete, the alternative diffusivity inputs by fib Bulletin 34 can be also adopted.
The values are shown in Figure 14. They are considered to be conservative.

Figure 13, Relative diffusivities of GGBFS concrete

4.0
y = 10.2x - 2.6
R² = 0.9562
Chloride Diffisivity, 10-12 m 2/s

3.5

3.0

2.5

2.0

1.5

1.0
0.35 0.40 0.45 0.50 0.55 0.60 0.65
Water/Binder Ratio
Figure 14, Chloride diffusivities of GGBS concretes from fib Bulletin 34

Durability Modelling of Reinforcement Corrosion in Concrete Structures 29


3.5.5.4 Silica Fume

For concrete with silica fume, the instantaneous diffusivity can be estimated using Equation 13
TM
recommended in the Life-365 model (Reference 27).

DSF  DPC  e 0.165SF (13)

Where

SF is silica fume content, %,


-12 2
DSF is instantaneous chloride diffusivity of concrete with silica fume, 10 m /s,
-12 2
DPC is instantaneous chloride diffusivity of concrete with GP cement (or OPC cement), 10 m /s.

Similar to fly ash concrete, the alternative diffusivity inputs by fib Bulletin 34 can be also adopted.
The values are shown in Figure 15 for silica fume concretes. They are considered to be
conservative.

6.0
Chloride Diffisivity, 10-12 m 2/s

5.5
y = 4.0293x + 3
R² = 0.9537

5.0

4.5

4.0
0.35 0.40 0.45 0.50 0.55 0.60
Water/Binder Ratio
Figure 15, Chloride diffusivities of silica fume concretes from fib Bulletin 34

3.5.5.5 Diffusivity for Full Probabilistic Analysis

For full probabilistic analysis, if no other more reliable information is available, then the
instantaneous diffusivity is assumed to have a normal distribution with a coefficient of variation of
20 %.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 30


3.5.6 Apparent Diffusivity Model

3.5.6.1 The Model

The apparent chloride diffusivity is determined based on chloride profiles obtained after a certain
exposure period by assuming a constant diffusivity for the period before the test. It is an
approximate approach to obtain an averaged/integrated diffusivity for the period.

Due to continuous densification of the pore structure with increasing concrete age, the apparent
diffusivity also decreases with concrete age. The relationship between apparent diffusivity and
age of concrete is described in Equation 14 (Reference 28). It has a similar form to Equation 9.
However the values are different in the two equations. The terms apparent diffusivity and
apparent age factor are to differentiate them from the instantaneous values.
tr n
Da  Dr ( ) (14)
ta

Where
2
D’a is apparent chloride diffusivity at the exposure age ta, m /s,
2
D’r is reference apparent chloride diffusivity at the reference age tr, m /s,

n is apparent age factor.

The apparent diffusivity model was developed based on the outcomes of Duracrete project
(Reference 52), as shown in Equation 15. It was later adopted by Concrete Society TR61
(Reference 28), fib Bulletin 34 (Reference 3), ISO16204 (Reference 5), NZS3101 (Reference 4),
and fib Model Code 2010 (Reference 1).
C  C0 X  X X  X (15)
 erfc( )  erfc( )
C s  C0 2 Da te tr n
2 Dr ( ) te
ta

Where te is the exposure time, s.

Comparison between chloride profiles by Equation 11 and 15 is demonstrated in Figure 16. It


can be seen that there is a difference in chloride profiles predicted using the two equations with
the same age factor of 0.4. The chloride profile by Equation 15 is slightly lower than that by
Equation 11.

It should be noted that theoretically the ‘m’ and ‘n’ age factor values have a very low chance to be
the same for the same concrete as the 'n' value is estimated from actual chloride profiles at
various ages in the exposed concrete while the 'm' is determined by testing concrete diffusivity at
different ages. Therefore, such a comparison is indicative only and is not relevant for modelling
purpose.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 31


0.7
Input Parameters
By Equation 11
0.6 D 56d = 2x10-12 m2/s

Chot/ride Concentration, %
By Equation 15
ΔX = 10 mm
Threshold
0.5 C10mm = 0.54%
m = 0.4
0.4 n = 0.4
t e = 50 years
t a0 = 28 days
0.3
C0 = 0.015%
0.2

0.1

0.0
10 20 30 40 50 60 70 80 90 100 110
Distance from Surface, mm
Figure 16, Comparing chloride profiles by Equations 11 and 15

3.5.6.2 Inputs of Apparent Diffusivity

For early age instantaneous diffusivity should not be used directly as the apparent diffusivity due
to unknown relationship between the two diffusivities. Only the long-term apparent diffusivity
obtained from the field or laboratory concrete should be used in this model.

The method to determine the long term apparent chloride diffusivity is to obtain a chloride profile
from concrete exposed to a chloride solution in controlled or natural exposure condition. The
semi-finite solution of Fick’s Second Law is used to determine an integrated apparent diffusivity
assuming that diffusivity and surface concentration remain constant for the entire exposure
period until the profile test (Reference 28).

If there is no other more reliable data, Equations 16-18 below (Reference 28) can be used to
estimate the apparent diffusivity at 20 years for PC, PFA, GGBFS and silica fume concrete
respectively.
W
GP Concrete: LogDa 20  12.936  1.995  (16)
C
W
PFA/GGBFS concrete: LogDa 20  13.325  1.409 (17)
C
W
Silica fume concrete: LogDa 20  13.800  3.100  (18)
C

For full probabilistic modelling, the apparent diffusivity can be considered to have a normal
distribution. The coefficient of variation can be considered to be 20% if there is no more reliable
data available prior to the modelling.

3.5.6.3 Inputs of Apparent Age Factor (n)

The apparent age factor can be obtained if the apparent diffusivities at various exposure times
are determined based on the chloride profiles on the same concrete. However, reliable data for

Durability Modelling of Reinforcement Corrosion in Concrete Structures 32


long term exposure takes very time to obtain. Apparent age factor for long term may not be
available in many cases.

If there is no other more reliable data available, apparent age factors for these concrete can be
estimated in accordance with Equations 19-22 below.

GP (OPC) Concrete: n  0.264 (19)


PFA concrete: n  0.699 (20)
GGBFS concrete n  0.621 (21)
W
Silica fume concrete: n  1  1.10  (22)
C

Alternatively, the age factors recommended by fib Bulletin 34 can be adopted. In this case, the
mean age factors are 0.30, 0.60 and 0.45 for concretes with GP (OPC) cement, fly ash and
GGBS respectively. No value is given for silica fume concrete

It should be noted that Equation 15 can be used only to predict the chloride diffusion within the
same exposure period as that the age factors are determined. Predicting chloride diffusion
beyond such a period may lead to significant error regarding the durability performance of the
concrete structure.

For full probabilistic modelling, the apparent age factor can be considered to have a normal
distribution. The coefficient of variation can be considered to be 15% if there is no more reliable
data available prior to the modelling.

3.6 EFFECT OF TEMPERATURE

The chloride diffusion process is influenced by the actual temperature within cover concrete through the
change of the diffusivities. The effect of temperature on chloride diffusivity can be estimated in
TM
accordance with the Arrhenius equation as shown in Equation 23, based on Life-365 (Reference 27).
As the diffusion through the cover zone is of interest, the temperature at concrete surface should be used.
U  1 1 
DT  Dref  exp     (23)
 R  Tref T 

Where

DT is the chloride diffusion coefficient at a temperature T, K

Dref is the chloride diffusion coefficient at reference temperature Tref, K

U is the activation energy of diffusion process, J/mol

R is the gas constant, which equals 8.314 J/K·mol

The term of U/R is equivalent to the ‘be’ adopted by fib Bulletin 34 with a mean value of 4800K being
recommended. However, Page et al (Reference 53) reported that the be value decreases with increasing
w/c ratio as expressed in Equation 24.
be  1000 [52.5  (W / C) 2  41.75  (W / C)  2.3] (24)

Durability Modelling of Reinforcement Corrosion in Concrete Structures 33


For full probabilistic modelling, the be can be considered to have a normal distribution. The coefficient of
variation can be considered to be 15% if there is no more reliable data available prior to the modelling.

It is recommended that Equations 23 and 24 be used in all durability modelling for chloride diffusion. The
prevailing environment can have a significant effect. For example, the temperature of concrete splashed
by cool seawater may only average 15C in cool southern waters. However the temperature of concrete
exposed to the sun in northern climates may exceed 50C during the day time.

3.7 SURFACE CHLORIDE CONCENTRATION

3.7.1 Increasing Surface Chloride Concentration with Time

The surface chloride concentration has been found to be not constant in some cases according to data by
TM
Uji et al (Reference 54), Chrisholm and Lee (Reference 55), and Life-365 (Reference 27). It increases
with age (i.e. with a slow build-up process) especially in the early stage of the exposure period. Such an
increasing surface concentration is common in the atmosphere with airborne chloride, marine tidal/splash
zones and structures exposed to molten snow containing de-icing salt. The length of the period with
increasing surface concentration varies with the exposure conditions from 0 years for the tidal zone to
more than 20 years or even longer for a coastal atmospheric zone with air borne chloride (Reference 27).

Such gradually increasing surface concentration would lead to a relatively slower diffusion process than
that predicted assuming a constant surface concentration if the ultimate surface concentrations are the
same for the two cases.

In such a condition, Equations 11 would predict a faster chloride diffusion with a deeper profile. For
design of new structures this may be considered acceptable as a overestimated surface chloride reflects
a conservative approach.

For Equation 15, the apparent diffusivity is obtained from actual chloride profiles and thus the surface
concentration has been incorporated into the diffusivity results leading to a more reliable prediction.
However, in condition assessment of the existing structures, it would underestimate the diffusivity leading
to overestimation of the residual life of concrete structures.
TM
Numerical modelling such as Life 365 adjusts surface chloride concentration at each time step during
modelling. Therefore, it can predict more accurately the chloride diffusion process for such condition.

3.7.2 Ultimate Surface Chloride Concentration

The ultimate surface chloride concentration varies with exposure conditions. Selection of input values
should be specific to the conditions. If there is no more reliable data available prior to the modelling, the
following values can be considered.

In the marine tidal zone, the mean surface concentration value ranges between total chloride 0.5~0.8%
by mass of concrete for GP concretes. The low values are suitable for dense concrete in humid, low wind
speed and relatively cold conditions. The high values are suitable for porous concrete in dry, high wind
speed and hot conditions.

In the marine splash/spay zones with a longer drying time, a mean value by 10~20% higher than that in
TM
tidal conditions is suitable, based on recommendations by Life-365 (Reference 27).

In submerged conditions, the surface chloride concentration increases with seawater concentration and
porosity of concrete. A mean surface chloride value of 0.3~0.6% is considered suitable, based on data
by Nagataki et al (Reference 56).

Durability Modelling of Reinforcement Corrosion in Concrete Structures 34


For concretes with blended cements (PFA, GGBSF, SF) and the associated high performance concrete,
a value 20~40% higher than GP concretes is suitable based on the recommendations by Bamforth
(Reference 28) and on the test data by Thomas and Bamforth (Reference 57) and Cook et al (Reference
58). The low–end value is suitable for silica fume concrete while the high-end value is suitable for fly ash
and GGBS blended cement concrete.

In offshore concrete well above seawater level and in coastal concrete with surfaces exposed to air borne
chloride deposition, surface chloride concentration increase with decreasing distance to the coast line and
roughness of the seawater. The specific value needs to be selected based on the specific condition
during the durability modelling.

Figure 17, Salt deposition map of Australian land (Reference 62)Cole (Reference 59) notes “A GIS
system has been developed to both define the concentration of marine aerosol at the shoreline arising
from both surf- and ocean-produced aerosol, and then to estimate the transport of this aerosol to a given
point inland. An Australia-wide map of airborne salinity has been derived (Figure 17). The salinity map
highlights the pronounced effect of both ocean state (as defined by whitecap activity) and climate factors
in controlling airborne salinity in Australia. For instance, southern Australian coastal zones, where
whitecap activity is high, have appreciably higher airborne salinity levels than Australia’s northern coast,
where whitecap activity is low.”

Although Cole (Reference 59) provided no way of relating this to surface chloride levels in concrete it is
likely that high deposition rates shown in the map would translate to high ultimate surface chloride levels
in concrete. However, the surface chloride levels are expected to vary significantly with the concrete
surface direction, heights, surface texture, concrete composition and density. Further work is required to
utilise Cole’s research to establish more detailed maps of exposure for concrete similar to those provided
in the New Zealand standard (Reference 4).

For full probabilistic modelling, the surface concentration can be considered to have a normal distribution.
The coefficient of variation can be considered to be 20%.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 35


Figure 17, Salt deposition map of Australian land (Reference 62)

3.8 THRESHOLD CHLORIDE CONCENTRATION

It is generally believed that once the chloride concentration reaches the threshold level at the
reinforcement surface, corrosion of steel reinforcement starts and enters the corrosion propagation
period. However, there are many threshold concentration values reported by various researchers. For
example, values ranging from 0.02% to 3.08% by mass of cement have been reported (Reference 60).
There is no universally accepted threshold concentration. This section discusses how to determine the
threshold level for use in chloride modelling.

3.8.1 Terms Used to Define Threshold Chloride Concentration

Historically, threshold chloride concentration has been expressed in many forms as listed in Table 4
depending on the components of chloride, chloride concentration base and concentration unit. These
expressions of chloride thresholds are interchangeable in most cases. However, each expression has its
advantages and disadvantages, which will be discussed in this section.
The initiation of reinforcement corrosion is directly related to the free (or water soluble) chloride ion
concentration in pore water (tested after extraction) because the free chloride ions in pore water are free
to break the passive layer on the steel surface. Naturally, the threshold chloride concentration is often
expressed as the free chloride concentration as a proportion of pore water volume. However, free
chloride concentration as a proportion of pore water is normally difficult to test. This leads to insufficient
data to determine the condition/time of corrosion initiation. Moreover, the pore water volume may vary
over time with varying exposure conditions, leading to change of chloride concentration and consequently
a poor test accuracy.

Table 4, Various expressions of threshold chloride concentration

Component of Chloride Concentration Base Concentration Unit

Free chloride pore water volume mol/l


(extraction) -
OH concentration in pore water ratio

Free chloride binder mass %


(water soluble) concrete mass or volume % or kg/m
3

Total chloride binder mass %


(acid soluble) concrete mass or volume % or kg/m
3

- -
OH ions are the key in the formation of the passive layer on steel surface and a decrease in OH
concentration would decrease the threshold chloride concentration when other conditions are kept the
same (References 60 and 62). Therefore, the threshold chloride concentration is sometimes expressed
- - -
by the ratio between chloride concentration and OH concentration in pore water, [Cl ]/[OH ] ratio. The
- -
preliminary research results demonstrate that there is a good relationship between [Cl ]/[OH ] ratio and
corrosion initiation. The data have shown the significant effect of C 3A content on the activation level.
Again testing free chloride concentration to determine the ratio is difficult. This again leads to insufficient
data for establishing the value.

The chloride threshold concentration is sometimes expressed by the free chloride concentration as
percent of binder mass, which may come from the convention way of expressing the admixture dosage.
It is noted that binder mass has a poor correlation with total pore volume or pore water volume in

Durability Modelling of Reinforcement Corrosion in Concrete Structures 36


saturated condition. For example, total pore volume change marginally from 18% to 12% when binder
3
content increases significantly from 200 to 600 kg/m . This leads to a significant change of threshold
chloride concentration (Reference 60), making this expression unsuitable as the criterion. In addition,
testing free chloride concentration involves washing concrete sample which would break the equilibrium
between the bound/adsorbed chloride ions leading a poor accuracy. Furthermore, the binder content in
concrete samples is often unknown and difficult to determine. As a result, the test data are insufficient for
establishing the threshold value.

The threshold concentration is sometimes expressed by the free chloride concentration as proportion of
3
concrete mass or volume (% or kg/m ). As discussed above, the total pore volume in concrete varies
only slightly with various binder contents and w/c ratios, such an expression gives a suitable indication of
corrosion initiation. In addition, mass of concrete is easy to determine. Again, testing free chloride is not
easy and accurate, leading to insufficient test data.

The threshold concentration is also expressed by the total (acid soluble) chloride concentration as
percent of binder mass. Chemically bound chloride concentration is normally in equilibrium with the free
chloride concentration. Therefore, total chloride content has a close relationship with free chloride
content and can be used to indicate the corrosion initiation. However, as discussed above, the binder
mass as a base is not directly related to pore water volume and thus this expression is not a suitable
indication of corrosion initiation. As a result, a wide spread of threshold chloride values ranging from
0.02% to 3.08% by mass of cement have been reported (Reference 60), which mean there is no firm
threshold chloride concentration. Considering this, Browne (Reference 63) gave the risk of corrosion as
shown in Table 5, based on the test data only on GP concrete with 0.6 w/c ratio. Again, this expression
requires binder mass which may be not known. Regardless of these shortcomings, this threshold
expression has been used widely.

The most suitable threshold chloride concentration is expressed by the total (acid soluble) chloride
concentration as proportion of concrete mass or volume. As discussed above, the total chloride has a
close correlation to the free chloride and so does the pore water volume to the concrete mass/volume.
This expression thus is closely related to the free chloride content as a proportion of pore water volume
and it is a very good indication of corrosion initiation. In addition, it is easy to test and sufficient data have
been obtained and it is increasingly used to express the threshold concentration.

Table 5, Corrosion risk classification by Browne (Reference 63)


(concrete with GP cement C3A >5% and w/c =0.6)
Chloride Content
Risk of Corrosion
(% cement mass)
<0.4 Negligible
0.4 to 1.0 Possible
1.0 to 2.0 Probable
>2.0 Certain

In summary, it is recommended in this document that total chloride concentration as a proportion of


3
concrete mass/volume (% or kg/m ) be used in Australia considering its advantages and reliability for
indicating the corrosion initiation.

3.8.2 Defining Corrosion Initiation

In order to measure the chloride threshold level, the point in time when the reinforcement becomes
depassivated must be known. There are many methods to define corrosion initiation but, none has been
universally accepted. This section discusses ways of defining corrosion initiation.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 37


Visual inspection of rust staining is often used to determine the initiation of corrosion. It is a destructive
method and tends to identify corrosion initiation a period of time after rust has formed.

Electrical half-cell potential of reinforcement is widely used to determine the corrosion probability and can
thus indicate corrosion initiation. This method is discussed in Z7/07 (Reference 46). A further option is to
measure the corrosion rate of reinforcement by Linear Polarisation Resistance (LPR) or other corrosion
current measuring techniques. If the corrosion rate is greater than a certain value, it means that the
active corrosion starts to present. These methods are also discussed in Z7/07.
2
A suitable corrosion rate value of 0.1 μA/cm (or 1.2 μm/yr) can be adopted to determine the threshold
chloride concentration when corrosion commences (Reference 64). This is the recommended criterion
where the chloride activation level is to be measured either on-site or in laboratory.

3.8.3 Measuring Threshold Chloride Concentration by Varying Chloride Levels

The chloride threshold value can be determined by introducing various concentrations of chloride ions
into the concrete and measuring whether the concentration is sufficient to cause corrosion initiation.
There are generally two generic methods to introduce the chloride into concrete.

One method is to admix chloride salt of various concentrations and determine the minimum dosage that
initiates reinforcement corrosion. It is an easy and fast method to obtain the threshold level and there has
been a significant amount of published data obtained using this method. However, the method by-passes
the formation of the passive film normally formed in uncontaminated concrete. In addition, admixed
chloride salts may participate in cement hydration leading to more bound chloride and less free chloride
than when chloride diffuses in at a later date. For these reasons, this method is likely to determine a
higher threshold level than is really the case.

The other method is to migrate external chloride ions through concrete cover to reach the embedded
reinforcement. It is a much more accurate and reliable method as time is given for the passive film to
form before it is broken down by the penetrating chloride ions. The chloride penetration and chemical
reaction with hydration products also mimics the real situation. However, it is a time-consuming process.
Threshold data obtained using this method are more reliable in chloride modelling.

3.8.4 Recommended Threshold Chloride Level

Overall, the chloride threshold for corrosion initiation is recommended to be expressed in total chloride
2
concentration as proportion of concrete mass/volume. A corrosion rate of 0.1 μA/cm (or 1.2 μm/yr) is
recommended to define corrosion initiation and to determine the chloride threshold value. Migrating
external chloride ions through the concrete cover to reinforcement is recommended in such a
determination.

Before a more reliable threshold value is established, the mean threshold chloride concentration is
selected as total chloride of 0.06% by mass of concrete for non-stressed carbon steel reinforcement
TM
based on the values by Life-365 (Reference 27) and JSCE (Reference 65). For pre-stressed/post-
tensioned steel in GP (Portland) concrete, the threshold chloride concentration is estimated to be 0.040%
based on the value by Stark (Reference 66).

For concrete containing supplementary cementitious materials such as fly ash, ground granulated blast
furnace slag (GGBFS) and silica fume, a lower free chloride threshold level is required considering a
relatively lower pH in such concrete than in GP cement concrete. However, considering these concretes
have a relatively high bound chloride ratio and thus a lower ratio of free chloride, the threshold values of
total chloride content can stay same as those for GP concrete, i.e. 0.06% by mass of concrete for normal
black steel and 0.04% for pre-stressed/post-tensioned steel.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 38


For hot dipped galvanised steel reinforcement, the threshold level can be more than double that of black
steel. For stainless steel reinforcement, the threshold level can be 5-10 times of the black steel
depending on the type (alloy) of the stainless steel.

For full probabilistic modelling, the threshold chloride value is considered to have a normal distribution.
The coefficient of variation is considered to be 10% if more reliable value is not available prior to the
modelling process.

3.9 PARTIAL FACTORS

The inputs given, i.e. surface chloride level, chloride diffusion coefficient and chloride activation level, are
all mean values and modelling using these values in the diffusion equations provided earlier will give the
average design life. That is the probability of achieving the stated design life is 50%. This may be
adequate for some clients as it applies to the end of service life where the current cost of repairs is
negligible.

In a full probabilistic assessment using the distribution of each input given in this section the actual
reliability can be assessed and this can be used to assess partial factors to be applied to allow for
different reliabilities. Appendix 1 of Z7/01 (Reference 21) provides some guidance on different
reliabilities. Care should be taken with this as a high reliability requirement associated with elements that
cannot be inspected probably relate to advanced deterioration rather than corrosion initiation.

When considering corrosion initiation using models, it is more normal that the damage will be easily
observed and repaired and in that case the reliability required is not as high. Table 6 has been extended
here to show how cover requirements can be adjusted to increase reliability. This is only approximate as
the factor varies depending on specific inputs.

Table 6, Recommended reliabilities for time to corrosion initiation


Reliability Probability of Structure or Element Comment Multiplication
Index failure Examples Factor on Cover to
(Approximately) increase from 50%
probability to
proposed reliability
0.5 30% Large element in long lived Possibly suited to wharves 1
structures where with a long design life and no
aesthetics and safety due underside public access
to spalling is not a major
concern.
1.3 10% Domestic Where elements can be 1.1
Non-structural elements easily repaired at a low cost if
Easily accessible parts of corrosion occurs but the cost
marine structures of improving durability at
construction would be
expensive.
2.3 1% Industrial pavements Elements that may have a 1.3
Structural concrete in higher associated repair cost
buildings due to structural support or
Building facades operating costs but where
Bridges damage can be observed
Retaining structures before a major or
catastrophic failure occurs.
3.6 0.01% Piles Elements where failure at 1.5
Prestressed beams earliest observable time
would be after a failure of
moderate to high
consequence.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 39


4. MODELLING CARBONATION IN CONCRETE

4.1 INTRODUCTION

Carbonation is the chemical reaction between mainly calcium hydroxide in concrete and carbon dioxide
gas in the atmosphere. The associated chemical reactions are shown in Equations 25 and 26 below.
Ca(OH)2 + CO2 → CaCO3 + H2O (25)
3CaO·2SiO2·3H2O + 3CO2 → 3CaCO3 + 2SiO2 + 3H2O (26)

The calcium carbonate that forms and depletion of the calcium hydroxide leads to a reduction in the pH of
the concrete from above 12.5 to below 9.5 typically. The carbonation depth is normally determined by
spraying a pH indicator (0.25~1% phenolphthalein in methylated spirits solution) on a freshly fractured
concrete surface. The area showing a pink colour indicates that concrete is not carbonated while the
area showing no colour indicates that concrete is carbonated. The boundary line between the two areas
denotes the carbonation front. It is measured as a depth from the exposed concrete surface.

Carbonation of concrete surrounding the reinforcing steel bars can destroy the passive protection layer on
the steel surface causing depassivation. A depassivated steel will start to corrode. The steel corrosion
products, iron oxide or hydroxide, have a volume 3-7 times greater than the original steel (References 67
and 68). Their formation can cause cracking, spalling or delaminating damages to the surrounding
concrete especially the cover concrete. In addition, the cross-sectional area of the steel bars decreases,
which can decrease the load bearing capacity of the affected concrete elements.

Carbonation is a progressive process with a depth increase with exposure time. The carbonation front
starts from the surface and moves progressively to deeper concrete. The speed of carbonation is related
to concrete carbonation resistance and environmental conditions. Generally, a concrete with a low w/c
ratio has a slower carbonation rate and the use of fly ash, silica fume and GGBS tends to increase
carbonation rate. High CO2 concentrations and low humidity in the environment also cause a faster
carbonation process.

Predicting carbonation of concrete in various environment conditions is frequently required for durability
design of concrete structures. Such prediction is often based on carbonation data in similar condition and
service periods. However, for different conditions and service periods, the prediction is not reliable. The
long term data is very limited for prediction purposes. In addition, concrete materials used currently are
significantly different from those used in the past from which results are obtained. Therefore, for durability
design, carbonation rate modelling based on carbonation characteristics recently obtained becomes a
very good option.

There are many models available and used to predict the concrete carbonation process. They often give
different results even for the same exposure condition, which indicates that some of them may not be
accurate. It is very important to select an 'accurate' model so that reliable predictions can be made. This
section reviews carbonation modelling methods to provide guidance to the selection of model and input
parameters.

4.2 GENERAL CARBONATION MODEL

The depth of the carbonation front of a concrete normally follows a square root relationship with exposure
time for a particular given condition, such as CO2 concentration and humidity, as shown in Equation 27
(References 69 and 70). The constant in the equation is called the carbonation coefficient which is related
to concrete carbonation resistance and the exposure condition.

X K t (27)

Where

X is carbonation depth, mm,

Durability Modelling of Reinforcement Corrosion in Concrete Structures 40


t is exposure time, year,
0.5
K is carbonation coefficient (or rate), mm/year .

For a concrete structure or a concrete sample exposed to a specific condition, carbonation depth
measurements at a given exposure time can be used to determine the carbonation coefficient. Such a
coefficient can then be used to predict carbonation depth at other exposure times in similar conditions via
a simple modelling process.

However, in many durability design cases, the carbonation coefficient is not known. Therefore, the key
objective of carbonation modelling is to determine the carbonation coefficient for a particular concrete and
exposure.

Various models have been developed to assist in selecting the carbonation coefficient or the equivalent
parameters. Among them, the model in fib Bulletin 34 (Reference 3) based on research of DuraCrete
(Reference 52) is used by the construction industry and many durability consultants. In the following
sections, the fib Bulletin 34 model (or simply called ‘fib model’) is introduced and then an evaluation of its
accuracy of various input parameters is provided.

4.3 FIB BULLETIN 34 MODEL

The model in fib Bulletin 34 (Reference 3) predicts the carbonation depth for various concretes and
exposure conditions as shown in Equation 28.

, 0   t )C s
1
X  2k e k c ( k t R ACC t  W (t ) (28)

Where

ke is the environment factor to consider influence of humidity in exposure condition,

kc is the execution transfer parameter to consider curing condition before exposure,

1
RACC , 0 is the inverse carbonation resistance obtained in the DuraCrete standard accelerated test,
0.5 3
(mm/year )/(kg/m ),

kt is the regression parameter to consider the carbonation difference between the accelerated test and
that in natural condition. For full probabilistic modelling, its mean value is 1.25 and its standard deviation
is 0.35,

εt is the regression interception to consider the carbonation difference between two test conditions,
0.5 3
(mm/year )/(kg/m ). For full probabilistic modelling, its mean value is 315.5 and its standard deviation is
48,
3
Cs is the carbon dioxide (CO2) concentration of the exposure environment, kg/m ,

W(t) is the weather function to consider wetting events.

The value of environment factor ke can be estimated using Equation 29 shown below.
ge
 RH real fe 
1  ( ) 
(29)
ke   100 
 1  ( RH refl ) fe 
 100 

Where

Durability Modelling of Reinforcement Corrosion in Concrete Structures 41


RHreal is the actual relative humidity of the carbonated layer, %,

RHref is the reference relative humidity with a value being 65, %.

Fe is the exponent with a value being 2.5,

ge is the exponent with a value being 5.

The value of execution transfer parameter kc can be estimated by Equation 30 shown below.
t
k c  ( c ) bc (30)
7

Where

tc is the period of curing, days,

bc is the exponent of regression. For full probabilistic modelling, its mean value is -0.567 and the
standard deviation is 0.024.

1
Some mean input values of inverse carbonation resistance RACC ,0
are given in Table 7 for various
concretes and cementitious materials. These values were obtained from samples tested in the following
conditions. Concrete prism samples (100x100x500 mm) were cured for 7 days in water having a
temperature of 20 ºC. After that, the concrete samples were pre-conditioned in the laboratory with a
temperature of 20 ºC and a relative humidity of 65% for 21 days. At an age 28 days concrete samples
were placed in a carbonation chamber with a CO2 concentration of 2% by volume, a temperature of 20 ºC
and a relative humidity of 65% for continuous exposure of 28 days. The concrete prisms were split and
the freshly exposed surfaces were sprayed with carbonation indicator solution, which contains
phenolphthalein 1 gram/litre of ethanol solution. The depth of the boundary line between the no-colour
zone and the pink colour zone was measured to give the carbonation depth.

1 0.5 3
Table 7 : Inverse carbonation resistance RACC , 0 , (m/s )/(kg/m )

Equivalent W/Ceqv
Cement Type
Australian Cements 0.35 0.40 0.45 0.50 0.55 0.60
CEM I 42.5R GP -- 3.1 5.2 6.8 9.8 13.4
CEM I 42.5R+FA 22% (k=0.5) GB- fly ash -- 0.3 1.9 2.4 6.5 8.3
CEM I 42.5R+SF 5% (k=2) n/a 3.5 5.5 -- -- 16.5 --
CEM III/B 42.5 GB-slag -- 8.3 16.9 26.6 44.3 80.0

Note: W/Ceqv is equivalent water cement ratio considering an efficiency factor (k) of 0.5 for fly ash and 2.0
for silica fume.

1
The inverse carbonation resistance RACC , 0 can be calculated using Equation 31.

1 xc
R ACC ,0  ( )2 (31)

Where

xc is the carbonation depth, m,

Durability Modelling of Reinforcement Corrosion in Concrete Structures 42


 is the time constant with a value being 420, (s/kg/m 3)0.5.

The actual mean value of Cs in the atmosphere ranges from 350 to 380 ppm with a standard deviation of
3
10 ppm. The above concentrations in ppm can be converted to 0.00057 to 0.00062 kg/m . The CO2
concentration has been increasing 1.5 ppm per year and this trend is expected to continue in future. After
3
80 years, the CO2 concentration is predicted to increase to 0.00082 kg/m . Using this value will give a
conservative prediction in carbonation depth.

The value of W(t) can be estimated with Equation 32 using relevant weather data.
( PSRToW )bw
t
W ( 0) 2 (32)
t

Where

t0 is the time of reference, year with value being 0.0767 (equivalent to 28 days).

t is the exposure time, year

bw is the exponent of regression. For full probabilistic modelling, its mean value is 0.446 and its standard
deviation being 0.163.

PSR is the probability of driving rain; the value to be selected is based on wind direction for vertical
surfaces, PSR is 1 for horizontal surface. For interior condition, PSR is 0.

ToW is the time fraction of wetness, which can be estimated using Equation 33 below, based on the
annual number of rain days.
rain  days
ToW  (33)
365

Where

The 'rain-days' is defined as annual number of days with more than 2.5 mm rain fall.

Some examples of carbonation modelling results using the fib Model and input parameters are shown in
Figure 18. Concrete in the analysis contains 22% fly ash and its w/c ratio is 0.55. Curing condition has a
significant influence on carbonations depth while other conditions are kept constant. Carbonation depths
at 100 year exposure is 43.6, 31.9, 25.1, 17.0 mm for concrete wet-cured for 1, 3, 7 and 28 days
respectively.

The wetness of the concrete surface during exposure also has a significant influence on carbonation
while other conditions are kept constant. With 7 day wet-curing the concrete’s carbonation depth at 100
year exposure is 25.1, 12.6 and 7.7 mm for concrete stored under interior conditions, on a vertical surface
under exterior conditions and on a horizontal surface under exterior conditions, respectively.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 43


50
1d curing, interior
3d curing, interior
7d curing, interior
40 28d curing, interior
Carbonation Depth, mm 7d curing, exterior, vertical PSR 0.3
7d curing, exterior, horizontal
30

20

10

0
30 40 0 5010 20
60 70 80 90 100
Exposure Time, years
Figure 18, Carbonation modelling in fly ash concrete with various curing and exposure conditions

4.4 VALIDATION OF FIB INPUT PARAMETERS

fib Bulletin 34 input parameters need to be validated by the results obtained from other sources for better
predictions. In this section, input parameters by the fib model are compared against available data from
other sources to see whether there is agreement or not.

During data analysis, carbonation coefficient results by the fib's model for various cementitious materials
are obtained by predicting 100 year carbonation depths in its standard indoor natural condition and
calculating the corresponding carbonation coefficients using Equation 27. Mean values of test method
regression constants (kt and εt) were used in the analysis.

An important factor to consider during validation is the sample preparation method and exposure
condition. The carbonation coefficient data from natural indoor exposure conditions are considered more
reliable than those in accelerated conditions or in variable outdoor conditions for validation purposes.
Therefore, carbonation coefficients recommended by the fib model in its standard natural indoor condition
are compared against those obtained from other sources in the similar natural indoor conditions.

In the fib standard natural indoor condition, the concretes are cured in water for 7 days and samples are
exposed to 23 °C temperature and 65% relative humidity. While the test conditions by others in most
cases are slightly different in either temperature, humidity or curing period. The results from other
sources are then adjusted using the fib method for the different curing time and humidity. The results are
also adjusted to consider the effect of temperature on carbonation based on the data by Ishida and Li
(Reference 71). The change ratio was adopted during the adjustment.

For concrete containing fly ash and silica fume in the fib model, the equivalent water/cement ratios
(W/Ceqv) were converted into water/binder ratio so that results from other sources can be compared at an
equal base.

In addition, some test results from the other sources were obtained on concretes with various fly ash
contents (20~30%). Hence to make carbonation coefficients given by the other sources comparable with
fib model results, they were adjusted to give equivalent values for 25% fly ash. This adjustment was
based on the effect of fly ash content on carbonation established by Burden (Reference 72). This is
convenient as 25% fly ash is the common content in Australia concretes.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 44


Similarly, the test date on concretes with slag also have different slag contents (50 to 75%). In this case,
all test data are adjusted to have 60% slag based on the effect of slag content on carbonation established
by Collepardi et al (Reference 73).

4.4.1 GP Concretes

For concrete with GP cement (CEM I 42.5R by fib Bulletin 34), the carbonation coefficient results for
various water/cement ratios from various sources are compared against the results modelled using the fib
model in Figure 19. Other data are from Collepardi et al (Reference 73), Tam et al (Reference 74), Litvan
and Meyer (Reference 75), Burden (Reference 72), He and Jia (Reference 76), and Skjolsvold
(Reference 77).

10
Collepardi, Indoor, Cured 28d
Carbonation Coefficient, mm/y0.5

9 Tam, Indoor, Cured 3d


Litvan, Indoor
8
Burden, Indoor, Cured 7d
7 He, Indoor
Skjolsvold, Indoor, Cured 28d
6
FIB, Indoor, Cured 7d
5

2 GP Cement
1
0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
Water/Cement Ratio
Figure 19, Comparison of carbonation coefficients of GP concretes

The data by Collepardi et al (Reference 73) were obtained on concrete with 28 days curing and 60%
exposure humidity. The data by Skjolsvold (Reference 77) were obtained on concrete with 28 days curing
and 50% exposure humidity. It appears that the results by Litvan and Meyer (Reference 75) were
obtained from the concretes using the similar cement to the fib test. That by Tam et al (Reference 74)
were obtained on concrete exposed in Singapore indoor condition having a humidity range of 65~95%
(with a mean value being 80%) and a temperature range of 25~30 ºC (with a mean value being 27.5 ºC).

It can be seen from Figure 19 that the fib model input values sit far below most other data which can be
fitted with the black trend line. Therefore, the mean carbonation coefficient input values should be
selected using the black trend line for the full range of w/c ratio. For full probabilistic modelling, the
0.5
standard deviation should be 0.5 mm/year .

4.4.2 Fly Ash Concretes

Some carbonation coefficient data on fly ash concretes from various sources are compared in Figure 20.
The data by Collepardi et al (Reference 73) were obtained from concretes with 25% fly ash, 28 day curing
and 60% exposure relative humidity. The data by Burden (Reference 72) were obtained from concretes
with 30% fly ash and 23 ºC exposure temperature. All results are adjusted to the fib test condition.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 45


It can be seen that the carbonation coefficient by fib model input parameters gives a significantly lower
carbonation coefficient than that from other sources. The difference increases with increasing w/c ratio.

8
Carbonation Coefficient, mm/y 0.5 Collepardi, Indoor, Cured 28d
7 Burden, Indoor, Cured 7d
He, Indoor
6 FIB, Indoor, Cured 7d

1
25% Fly Ash
0
0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
Water/Binder Ratio
Figure 20, Comparison of carbonation coefficients of fly ash concretes

Therefore, the mean input values of carbonation coefficient should be selected based on the value of the
black trend line if more reliable values are not available prior to the modelling. For full probabilistic
0.5
modelling, the standard deviation should be 0.5 mm/year .

4.4.3 GGBS Concretes

Carbonation coefficients on concretes with GGBS (slag) from various sources are compared against
those by the fib model in Figure 21. It appears that data by Wierig (Reference 78) and Litvan and Meyer
(Reference 75) were obtained from concretes using the similar cements to the fib test. The data by
Collepardi (Reference 73) were obtained on the concretes with 50% slag, 28 days curing, and 60%
relative humidity during exposure. They are adjusted using the fib method accordingly.

It can be seen from the figure that fib input values are lower than other data at the low w/b ratio range. To
be conservative in durability modelling and design, the mean input values of carbonation coefficient can
be selected from the black trend line if more reliable values are not available prior to the modelling
0.5
process. For full probabilistic modelling, the standard deviation can be considered to be 0.5 mm/year .

Durability Modelling of Reinforcement Corrosion in Concrete Structures 46


12
Collepardi, Indoor, Cured 28d
11

Carbonation Coefficient, mm/y 0.5


Wierig, Indoor, Cured 7d
10 Litvan, Indoor
9 FIB, Indoor, Cured 7d

8
7
6
5
4
3
GGBS
2
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
Water/Binder Ratio
Figure 21, Comparison of carbonation coefficients of slag concretes

4.4.4 Silica Fume Concretes

There is not much data published for silica fume concrete. Figure 22 shows the comparison of available
results from other sources and fib ones. The data by Skjolsvold (Reference 77) were obtained from
concretes with 5% and 10% silica fume, 28 day curing and 50% relative humidity during exposure. The
carbonation coefficients are adjusted to the fib condition using the change ratio.

It can be seen from Figure 22 that the fib results are lower than those by other sources. To be
conservative in durability design, the mean input values of carbonation coefficient can be selected from
the black trend line during the modelling if more reliable data are not available prior to the modelling
0.5
process. For full probabilistic modelling, the standard deviation should be 0.5 mm/year .

9
Skjolsvold, Indoor, SF10%
Carbonation Coefficient, mm/y 0.5

8 Skjolsvold, Indoor, SF5%

7 FIB, Indoor, Cured 7d

2
Silica Fume
1
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
Water/Binder Ratio
Figure 22, Comparison of carbonation coefficients of silica fume concretes

Durability Modelling of Reinforcement Corrosion in Concrete Structures 47


4.4.5 Summary

The comparison of fib model input values of carbonation coefficients against other sources based on the
same conditions indicated that fib carbonation coefficients are mostly lower than data from other sources.
It is recommended that if more reliable values are not available, mean input values be selected from the
0.5
black trend lines shown above and a standard deviation of 0.5 mm/year be used.

4.5 SIMPLIFIED FIB MODEL

Once a carbonation coefficient has been determined by either more reliable values or by the trend lines
,0   t )
1
shown above for controlled indoor conditions, they can replace the following 2( kt R ACC part of
Equation 28 of the fib model. The mathematic relationship is given in Equation 34.

,0   t )
1
k Indoor  2(k t R ACC (34)

Where kindoor is determined in the fib indoor condition (7 day curing, drying for 21 days, exposed to 65%
humidity in the laboratory with CO2 concentration 380 ppm). All other terms in this equation are to remain
the same as those in Equation 28.

For other preparation and exposure conditions including curing time, exposure humidity, CO 2
concentration, and wetness, the fib's model can be modified with Equation 35 below obtained.

X  k Indoor k e k c C s t  W (t ) (35)

Where all terms are the same as those in Equations 28 and 34.

4.6 TEST METHODS TO DETERMINE CARBONATION COEFFICIENT

It is recommended that two standard test conditions be adopted to determine the carbonation coefficient
for modelling purpose. One is the natural indoor condition and the other one is an accelerated test as
detailed in fib Bulletin 34 (Reference 3). The sample preparation and testing methods are summarised in
Z7/07 (Reference 46).

The carbonation results obtained in accelerated test condition ideally should be correlated to those
obtained from the same concrete but in the natural indoor exposure condition if it is possible. The
relationship can be used to convert the data from accelerated tests into those obtained under the natural
indoor exposure condition.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 48


5. MODELLING REINFORCEMENT CORROSION

5.1 INTRODUCTION

Whilst the reinforcement corrosion (or corrosion propagation) period is short for GP cements (OPC) in moist
environments, it can be the dominant factor in the service life of some elements especially for those with
supplementary cementitious materials (SCMs). Durability modelling of the reinforcement corrosion is not well
established. However due to its significance in certain exposures, guidance on modeling methods proposed is
given in this section.

To determine the time for corrosion propagation period, two major parameters need to be determined. One is
the critical criterion for deterioration damage by reinforcement corrosion. The other is the corrosion rate of
reinforcement in various conditions. Once the two parameters are determined, the length of the reinforcement
corrosion (corrosion propagation) period can be determined. In this section, the modeling methods of
reinforcement corrosion are discussed.

5.2 CRITICAL CRITERION FOR CORROSION DAMAGE

The criterion for deterioration failure of a reinforced concrete structural facility depends on the permissible
deterioration level for the limit states specified by the asset owner or any other stakeholders. Regarding the
permissible deterioration level, the loading condition is critical. In most cases, the first sign of severe
deterioration is cracking of cover concrete due to corrosion of reinforcement. Once cover concrete cracks (at
approximately 0.10 mm in width), the subsequent deterioration will be accelerated and the remaining life
becomes very limited. Therefore, regardless of the limit states specified, the time for cracking of cover concrete
is critical and should be determined first. In many cases it can be used as the serviceability limit for most critical
elements.

The initiation of cracks in cover concrete is related to the maximum volume of corrosion product, the concrete
tensile strength, modulus of elasticity, cover thickness, rebar diameter, porosity of concrete around the rebar,
and localisation of the corrosion. The effects of these factors have not been quantitatively established although
great efforts have been made in this regard.

Nevertheless, Webster (Reference 79) proposed a simple criterion of cracking initiation considering combination
of these factors based on test data from various papers including Al-Sulaimani et al (Reference 80), Clark and
Saifullah (Reference 81), Liu (Reference 82), Cabrera (Reference 83), and Alonso et al (Reference 84), which
can be adopted for this document. The criterion is shown in Equation 36 below.

 CR  1.25  C (36)

Where

ᵟCR is the corrosion penetration depth, µm,

C is the concrete cover thickness, mm.

This criterion indicates that concrete cracking initiation due to reinforcement corrosion is mainly dominated by
cover thickness and other factors have a relatively minor influence as a whole on the cracking of concrete due
to reinforcement corrosion. For non-uniform corrosion cracks may initiate earlier due to local stress
concentration. A safety factor should be considered during the modelling in this case.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 49


5.3 GENERAL CORROSION MODELING

The corrosion rate of reinforcement in concrete very much depends on the environment conditions, cementitious
materials type, and the concrete quality or thickness of cover concrete. The corrosion rate also depends on the
type of depassivation, i.e. carbonation or chloride penetration. The corrosion rate therefore should be estimated
for both carbonated concrete and for chloride-contained concrete as shown in the following sections.

5.3.1 Chloride-Contaminated Concrete

There have been significant efforts in research to predict the corrosion rate in chloride contaminated
concretes. However, there is no widely accepted model available for such a prediction due to the
complex electrochemical processes of the corrosion.

DuraCrete (Reference 52) provided a corrosion rate estimation at 20 °C temperature for chloride-
contained concrete in various exposure conditions as shown in Table 8. It can be seen that the corrosion
rate is zero in submerged condition due to lack of oxygen. That in wet conditions is 4 µm/y while in cyclic
wet-dry conditions or in airborne sea water conditions a higher corrosion rate of 30 µm/y is proposed. In
tidal/splash zone, the highest corrosion rate of 70 µm/y is estimated.

Again, a very high level of variation is also presented for each case as shown by the high standard
deviation values. Such a high variation is possibly related to the variation in cover thickness, concrete
w/c ratio, concrete temperature, chloride content and cementitious materials types. High end values of
corrosion rate within the range are expected in conditions with low cover, high w/c ratio, high temperature,
high chloride content near the reinforcement, and a GP concrete without supplementary cementitious
materials (SCMs). With SCMs, a significantly lower corrosion rate is expected due to a higher electrical
resistivity (Reference 85). Therefore, corrosion rates for specific conditions should be selected based on
the specific exposure, concrete cover, binder type, concrete quality etc characteristics.

Table 8, Corrosion rate of reinforcement in chloride-contaminated concrete (Reference 52)

Exposure Type Corrosion Rate, µm/y Standard Deviation, µm/y Distribution


Wet-rarely dry 4 3 Lognormal
Cyclic wet-dry 30 20 Lognormal
Airborne sea water 30 20 Lognormal
Not expected except bad concrete
Submerged - Lognormal
or lower cover
Tidal/Splash Zone 70 40 Lognormal

5.3.2 Carbonated Concrete

Although significant efforts have been made in research to predict the corrosion rate in carbonated
concrete, no widely accepted model is available again for such a prediction. Therefore, a simple method
to determine the corrosion rate is desired and is considered to be more suitable.

DuraCrete (Reference 52) provide a corrosion rate estimation at 20 °C temperature for carbonated
concrete in various moisture conditions as shown in Table 9. It can be seen that corrosion rate is zero in
a dry exposure condition (e.g. <60% relative humidity). It is 2 µm/y in moderate humidity condition, 4
µm/y in wet conditions and 5 µm/y in cyclic wet-dry conditions.

A very high level of variation is also presented as shown by the high standard deviation values. Such a
high variation is possibly related to the variation in cover thickness, concrete w/c ratio, and concrete
temperature. High end values of corrosion rate within the range are expected in conditions with low
cover, high w/c ratio and a high temperature. Therefore, the corrosion rate for specific conditions should

Durability Modelling of Reinforcement Corrosion in Concrete Structures 50


be selected based on the specific exposure, concrete cover, binder type, concrete quality etc
characteristics.

Table 9, Corrosion rate of reinforcement in carbonated concrete (Reference 52)

Exposure Type Corrosion Rate, µm/y Standard Deviation, µm/y Distribution


Dry (<60% humidity) 0 0 Lognormal
Wet-rarely dry (unsheltered) 4 3 Lognormal
Moderate humidity (sheltered) 2 1 Lognormal
Cyclic wet-dry (unsheltered) 5 3 Lognormal

5.4 MODELLING CORROSION UNDER OXYGEN CONTROL

5.4.1 Modelling Method

There are no published general solutions to situations of oxygen starvation (e.g. immersed concrete).
Hence the modelling method here is based on oxygen diffusion theory, Faradays Law and published data
on oxygen diffusion coefficient. The general corrosion situation can be calculated in a series of steps as
shown below.

The oxygen flux at any point can be calculated using Equation 37 shown below assuming the oxygen is
linearly distributed through the depth of cover concrete and the oxygen concentration at the corrosion site
is nearly zero.

DoxCox
J ox  (37)
C

Where
2
Jox is the oxygen flux at the cathode, mole/m ,
2
Dox is the oxygen diffusivity of concrete, m /s
3
Cox is the oxygen concentration at the concrete surface, mole/m

C is the depth of the reinforcement surface (e.g. cover thickness), mm

From the oxygen availability at the cathode, the cathodic current density can be calculated using Equation
38 shown below.

iC  nFJ ox (38)

Where
2,
ic is the corrosion current density at the cathode, A/m

n is the electrons transferred = 4

F is the Faradaic Constant = 96500 C/mole

Durability Modelling of Reinforcement Corrosion in Concrete Structures 51


The anodic current density which dictates the loss of reinforcement section with corrosion can then be
calculated using Equation 39 shown below.

ic
ia  (39)
r

Where
2
ia is the corrosion current density at the anode, A/m ;

r is the ratio of anode area to cathode area.

Finally, using Equation 39, the corrosion rate can then be calculated from the Faradaic constant for steel.
It should be noted that this model is based on the assumption that the corrosion is uniform over the anode
surface, which may be not always true in actual corrosion situations (e.g. pitting corrosion) and thus a
safety factor should be considered in this case.

5.4.2 Application to Immersed Concrete

The corrosion rate of reinforcement in concrete permanently immersed on all faces will be controlled by
oxygen availability at the cathode and the ratio of anode to cathode area. Hence durability modelling can
be undertaken based on oxygen control at the cathode without considering the chloride diffusion process.

Using the process outlined in section the last section, the following parameters can be applied:
 Cox (oxygen concentration of seawater) = 225 µ Mole at 20C.

-6 2
Dox (oxygen diffusivity through saturated concrete) = 5.2 x 10 cm /sec. This is a suitable value for
moderate performance concrete.
 r (ratio of anode area to cathode area) = 100. It is assumed that most steel is anodic.

For the purpose of this example, it has been assumed that the required design life is 100 years. Hence for
a reliability index of 3.1, the target design life when modelling using average inputs should be 261 years.
The corrosion losses over 261 years shown in Table 10 are far lower than that required to cause spalling
and significant loss of reinforcement cross-sectional area.

Table 10, Estimated corrosion rate of reinforcement in concrete permanently immersed on all
faces in seawater

Cover, mm 20 30 40 50 60 70
-6
Corrosion rate, x10 mm/yr 26.1 17.4 13.1 10.4 8.7 7.5
Corrosion over 261 year target life, mm 0.0068 0.0045 0.0034 0.0027 0.0023 0.0019

5.4.3 Application to Buried Concrete

Where concrete element is buried in the soil containing saline groundwater, durability design generally
targets to ensure the reinforcement remains passive. However, if the reinforcement in concrete piles
loses its passivity during service, due to either a low cover or a poor concrete, reinforcement will undergo
corrosion. In such soil condition, oxygen availability to the reinforcement corrosion will also be limited and

Durability Modelling of Reinforcement Corrosion in Concrete Structures 52


will often be in control of the corrosion process (Figure 23). In this condition, the same modelling
approach discussed for immersed concrete can be applied here.

In Case a), if the soil at the full pile depth is saturated with saline groundwater, the oxygen diffusivity of
soil and concrete will be very low and the reinforcement corrosion rate will be similar to the concrete
immersed in seawater.

In Case b), if the top section of the concrete pile is buried in the unsaturated soil above the groundwater
level, the oxygen diffusivity of soil and concrete in this section will increase and a cathode will be formed
here. A longer anode section will be formed in the saturated soil below, resulting in a high anode to
cathode ratio (r) and a low corrosion rate.

In Case c), if the depth of unsaturated soil increases with a very dry section on pile top formed and the
cathode pile section moved downward, the length of anode section will decrease from Case 2. This will
lead to a lower anode to cathode ratio (r) and consequently a higher corrosion rate.

a) b) c)
O2

Dry Part
O2

Cathode

Unsaturated Soil
O2
Unsaturated Soil

O2 ic
ic
O2
ic
O2
ic

Cathode
O2 ic ic
O2 ic
O2 ic ic
Saturated Soil

O2 ic Saturated Soil
ic
Anode

O2

Anode
ic
O2
ic
O2 ic
O2 ic
O2 ic
O2 ic
ic
O2
ic
O2

Saturated
Soil

Figure 23, Anode to cathode ratio comparison in piles

5.5 CORROSION RATE WITH RESISTIVITY CONTROL

5.5.1 Resistivity and Corrosion Rate

In general there is an ample supply of oxygen to the cathode such that corrosion is controlled by the
resistance between anode and cathode rather than oxygen supply. It has long been recognised that as
the electrical resistance increases the corrosion rate decreases to a negligible amount. Browne
(Reference 63) suggested that at resistivities above about 20,000 ohm·cm the corrosion rate would
become negligible. Hunkeler (Reference 86) showed that as the relative humidity in the concrete
decreased to below 65% the resistivity increased to over 20,000 ohm·cm and the corrosion rate became
negligible (Figure 24 and Figure 25). These early studies supported the use of covers limits in European
and Australian codes that might lead to carbonation depths encroaching on reinforcement in dry

Durability Modelling of Reinforcement Corrosion in Concrete Structures 53


environments. The assumption is that even if reinforcement corrosion commences the corrosion rate will
be insignificant.

0.014 7,000

0.012

10,000

Resistivity, Ω·cm
0.010
Conductivity, S/m

0.008

15,000
0.006

0.004 25,000

0.002 50,000
100,000
0-
40 60 80 100
Relative Humidity, %
Figure 24 : Concrete resistivity at different relative humidity (Reference 86)

Resistivity, kΩ·cm
1000 100 10 1
1000
Current Density, µA/cm2

100
Corrosion Rate, mm/year

10

Data from
10 1
Noggerath (1990)
Raupach (1992)
Tondi et al (1993) 0.1
1
0.01

0.1 0.001

0.0001

0.0001 0.001 0.01 0.1


Conductivity, S/m
Figure 25 : Relationship between resistivity and corrosion rate (Reference 86)

Andrade (Reference 87) also shows that the variation of the resistivity with the degree of concrete water
saturation means steel corrosion rate is proportional to the resistivity value, refer Figure 26. This graph
responds to the Equation 40 below:

Durability Modelling of Reinforcement Corrosion in Concrete Structures 54


26
I corr  (40)
ef
Where
Icorr is the corrosion rate, A/cm
2

ef is the resistivity at 28 days in saturated conditions, Kohm.cm

Figure 26, The relationship between Icorr and resistivity of concrete (Reference 87)

5.5.2 Concrete Resistivity Values

Saturated concrete resistivity is influenced by the nature of the cement system and in particular the use of
SCMs, certain admixtures and water/binder ratio. It has also been measured (Figure 27) on various
projects. For some concretes, saturated resistivity exceeds 50,000 ohm·cm. When modelling the
reinforcement corrosion (propagation) period, no factor of safety is used on measured saturated resistivity
as this is a worst case.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 55


40
SCC, 0.3w/b, FA,MS

30 10%MS

Resistivity, kΩ·cm 25 25%FA+10%MS

20
65%Slag+10%MS

15
65% Slag

10

5 OPC (GP) Cement

0
0 10 20 30 40 50 60 70 80 90 100
Time of partial immersion in NaCl solution, days
Figure 27 : Development of resistivity of Australian concretes (Reference 88)

5.6 LENGTH OF PROPAGATION PERIOD

5.6.1 General Model

The length of propagation period can be determined using Equation 41 below:

 CR
T1  (41)
Rcorr

Where
T1 is the length of corrosion propagation period, yr
Rcorr is the rate of reinforcement corrosion, µm/year
ᵟCR is the corrosion penetration depth when crack initiated to 0.1 mm width, µm.
For example, if the concrete cover thickness is 70 mm in a marine tidal/splash condition, the corrosion
depth to initiate cracking in cover concrete is estimated to be 70×1.25 = 87.5 µm. The corrosion rate is
assumed to be 70 µm/year based on DuraCrete recommendation (Reference 52). In this condition, the
length of propagation period is estimated to be 87.5/70 = 1.25 years.

5.6.2 Andrade's Model

Alternatively, the prediction of reinforcement corrosion (propagation) period for service life can be
formulated using the model by Andrade (Reference 87) as shown in Equation 42 below:

t
Pcorr  ef ( ) q 
t0
T1  (42)
K corr 0.00116

Where:
T1 is the length of corrosion propagation period, yrs

Durability Modelling of Reinforcement Corrosion in Concrete Structures 56


Pcorr is the maximum limit of steel cross section at the end of time T1, cm (or mm),
t is the time during the corrosion, yrs,
t0 is the initial age of concrete (normally 28 days or 0.0767 yrs), yrs
q is the aging factor of the resistivity (normally taking value of 0.22 for Cement I, 0.37 for Cement II/A-P
and 0.57 for Cement/A-V),
ξ is the environmental factor of the corrosion rate (it can be of 10±2 for carbonation and 30±5 for
chlorides),
2
Kcorr is the constant with a value of 26 μA/cm ·k·cm (= to 26 mV/cm) relating the resistivity and the
corrosion rate Icorr.

This model can be used to calculate the required 28 day resistivity to achieve a certain propagation period
and to develop mix design parameters to achieve the required resistivity. As the propagation period can be
a very long part of a structures design life in some conditions, this model may be a useful in estimating
propagation period. However as this model does not have a significant amount of support data on real
world structures, a high safety factor of say 3 should be considered to the average propagation period
estimated.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 57


6. SUMMARY AND RECOMMENDATIONS

The following summary and recommendations are provided for modelling of reinforcement corrosion in concrete
structures including chloride diffusion, carbonation and reinforcement corrosion (propagation).

6.1 MODELLING CHLORIDE DIFFUSION

Using the instantaneous diffusivity model (i.e. Equation 11), the chloride diffusion in concrete can be
predicted with determined or assumed instantaneous diffusivity values and aging factors. If more reliable
data are not available at the time of modelling, the mean instantaneous chloride diffusivity can be
estimated using Equation 12 for GP concrete, using the trend line in Figure 11 or Figure 12 for fly ash
concretes, using the trend line in Figure 13 or Figure 14 for GGBF concrete, and using Equation 13 and
the trend line in Figure 15 for silica fume concrete. For full probabilistic modelling, the coefficient of
variation should be 20%.

The mean instantaneous age factor for various cementitious materials can be estimated using Equation
10 and its coefficient variation should be 15% for full probabilistic modelling. It should be noted that this
model is only valid in the period having a decreasing diffusivity while beyond the period, it will
overestimate the service life.
TH
Alternatively, the numerical model such as Life 365 can be used to predict the chloride diffusion
process. This method is accurate for a slow build-up of surface chloride concentration and for a length of
period with a decreasing diffusivity.

Instantaneous chloride diffusivity at early age can be measured using a natural-exposure test such as NT
Build 443 method. An accelerated test such as NT Build 492 can be used, which gives a conservative
estimation.

Using the apparent diffusivity model (i.e. Equation 15), the chloride diffusion process can be predicted
with the input diffusivity determined using Equations 16 -18 for various cementitious materials
respectively. The age factor can be determined using Equations 19 - 22 or using those recommended by
fib bulletin 34. It should be noted that this model is only valid within the same exposure period as the
period the data was collected to develop the model.

Threshold chloride concentration shall be expressed in total chloride as a percent of concrete mass. If
there is no more reliable value established or known prior to modeling, the threshold chloride
concentration shall be 0.06% for black steel reinforcement and 0.04% for stressed steel tendons. The
coefficient variation should be 15% for full probabilistic modeling. A higher threshold value can be
considered for reinforcements of hot dipped galvanised steel and stainless steel.

Other models may also be used, provided that the basic principles formulated by verification by a full
probabilistic method are fulfilled.

6.2 MODELLING CARBONATION

Carbonation should be modelled using the simplified fib model (i.e. Equation 35). The input parameters
of carbonation coefficients should be based in the fib standard natural indoor conditions. If there is no
more reliable carbonation coefficients available prior to modelling, mean input values of carbonation
coefficient on the recommended trend lines in Figure 19 to Figure 22 can be adopted for the standard
0.5
natural indoor conditions. For full probabilistic modelling, the standard deviation should be 0.5 mm/year .

The carbonation coefficient (kindoor) should be tested on concrete sample exposed to indoor condition with
a temperature of 23±2 ºC, a relative humidity of 65±2% and a carbon dioxide concentration of 380±5
ppm. Before carbonation exposure, concrete samples should be wet-cured for 7 days and dried for 21
days under a wind speed of 0.5±0.1 m/s in the same temperature and humidity as for the above exposure
condition. Carbonation shall be measured at 90 days and 1 years of exposure time as the minimum.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 58


Accelerated exposure with 2% volume carbon dioxide may also be used to measure carbonation
coefficient. Results shall be converted/calibrated to those in standard natural indoor condition before
being used in modelling.

Other models may also be used, provided that the basic principles formulated by verification by a partial
probabilistic method are fulfilled.

6.3 MODELLING REINFORCEMENT CORROSION

The general failure criterion by reinforcement corrosion damage can be the corrosion depth causing
cracks of 0.1 mm in width at concrete surface. The corrosion penetration depth to cause such cracks can
be estimated using the general model by Webster (i.e. Equation 36).

The corrosion rate of reinforcement in chloride contaminated concrete can be estimated using the values
given in Table 8. High end values of corrosion rate within the range are expected in conditions of low
cover, high w/c ratio, high temperature, high chloride content at the reinforcement surface, and a GP
concrete without any supplementary cementitious material (SCM).

The corrosion rate of reinforcement in carbonated concrete can be estimated using the values are given
in Table 9. High end values of corrosion rate within the range are expected in conditions of low cover,
high w/c ratio and high temperature.

To determine the reinforcement corrosion propagation period, the general model as shown by Equation
41 can be used with the inputs of the maximum corrosion depth and the corrosion rate determined above.

In special cases where oxygen availability is the rate controlling factor for corrosion, the corrosion rate
can be estimated based on the oxygen diffusive flux by Equation 39. In other cases when resistivity is in
control, the corrosion rate can be estimated by using Andrade's method (i.e. Equation 40). However these
models are not well established and appropriately high safety factors should be applied.

Furthermore, the length of corrosion propagation period can be determined in general by Equation 41
using the corrosion penetration depth to initiate cracking in cover concrete and the corrosion rate
estimated above. Alternatively, it can be calculated using the Andrade's model (i.e. Equation 42).
However, this model is not well supported by real word data and appropriately high safety factors should
be applied.

Durability Modelling of Reinforcement Corrosion in Concrete Structures 59


REFERENCES

1. fib Bulletin 65 and 66, Model Code for Concrete Structures 2010 – Final draft, Volume 1 and 2, Lausanne,
Switzerland, 2012
2. AS 3600-2009, Concrete structures, Standards Australia, Sydney, Australia, 2009
3. fib Bulletin 34, Model Code for Service Life Design, Lausanne, Switzerland, 2006
4. NZS3101:Part 2: 2006, Concrete structures standard - Commentary, New Zealand Standards Council,
Wellington, New Zealand, 2006
5. ISO16204-2012, Durability – Service life design of concrete structures, International Organization for
Standardization (ISO), Switzerland, 2012
6. AS 5100.5-2004, Bridge design, Part 5: Concrete, Standards Australia, Sydney, Australia, 2004
7. AS 3735-2001, Concrete structures for retaining liquid, Standards Australia, Sydney, Australia, 2001
8. AS 2159-2009, Piling-Design and installation, Standards Australia, Sydney, Australia, 2009
9. AS 4975-2005, Guideline for the design of maritime structures, Standards Australia, Sydney, Australia,
2005
10. AS 1085.14-2012, Railway track materials - Prestressed concrete sleepers, Standards Australia, Sydney,
Australia, 2012
11. AS 1597.2-1997, Precast reinforced concrete box culverts, - Part 2, Large Culverts, Standards Australia,
Sydney, Australia, 1997
12. AS 4058-1996, Precast concrete pipes, Standards Australia, Sydney, Australia, 1996
13. AS 1379-2007, Specification and supply of concrete, Standards Australia, Sydney, Australia, 2007
14. AS 2758.1-1998, Aggregates and rock for engineering purposes, Part 1, Concrete aggregates
15. AS 3972-1997, Portland and blended cement
16. AS 3582.1-1998, SCM, Part 1, Fly ash
17. AS 3582.2-2001, SCM, Part 2, Slag-Ground granulated iron blast-furnace
18. AS 3582.3-2003, SCM Part 3, Amorphous silica
19. Chisholm, D.H. and Lee, N.P., Durability prediction for coastal reinforced concrete structures – matching
reality and theory, New Zealand Concrete Society Conference, Taupo, 13-15 Oct, 2000
20. Bentz, D.P., Sant, G., and Weiss, J., Early-age properties of cement-based materials. Part I: Influence of
cement fineness, Journal of Materials in Civil Engineering © ASCE / July 2008
21. Z7/01, Durability Planning, Concrete Institute of Australia, Sydney, 2014
22. fib Bulletin 59, Condition Control and Assessment of Reinforced Concrete Structures, Lausanne,
Switzerland, 2006
23. Tuutti, K., Corrosion of steel in concrete, Swedish Cement and Concrete Research Institute, Report No. 4-
82, pp 469, Stockholm, Sweden,1982
24. Liang, M., Kao, C., Huang, H., and Oung, J., A case study of reliability analysis on the damage state of
existing concrete viaduct structure, Tamkang Journal of Science and Engineering, Vol 12, 2009, No. 4, pp
371-379
25. Fick, A., On Liquid Diffusion, Philosophy Magazine (in English), S.4, Vol.10, 30-39, 1855
26. Tang, L. and Nilsson, L.-O., Chloride diffusivity in high strength concrete at different ages, Nordic Concrete
Research Publication, Vol 11, pp. 162–171, 1992
TM
27. Life 365 Consortium III, Manual for Life-365 Service Life Prediction Model, 2014
28. Bamforth, P., Enhancing Reinforced Concrete Durability, Concrete Society Technical Report 61, The
Concrete Society, Surrey, UK, 2004

Durability Modelling of Reinforcement Corrosion in Concrete Structures 60


29. Brown R., On the general existence of active molecules in organic and inorganic bodies, Philosophy
Magazine, New Series 4, Annual Philosophy, 1828
30. Snyder, K.A., Validation and modification of 4SIGHT computer program, NISTIR 6747, National Institute of
Standards and Technology, Gaithersburg, MD, 2001
31. Tang, L. Chloride transport in concrete – measurement and prediction” PhD thesis, Chalmers University of
Technology, Gothenburg, Sweden 1996
32. Freundlich H.M.F., Über die Adsorption in Lösungen. Z Phys Chem. 1906;57(A): pp 385–470
33. Spiesz, P., Ballari, M.M., Brouwers, H.J.H., “RCM: A new model accounting for the non-linear chloride
binding isotherm and the non-equilibrium conditions between the free- and bound-chloride concentrations.”
Construction and Building Materials 27 (2012) 293-304
34. Arya, C., Buenfeld, N.R. and Newman, J.B., Factors influencing chloride-binding in concrete, Cement and
Concrete Research, Vol 20, pp 291-300, 1990
35. NordTest Build 443, Concrete, Hardened - Accelerated chloride penetration, 1995
36. NordTest Build 492, Concrete, mortar and cement-based repair materials – chloride migration coefficient
from non-steady-state migration experiments, 1999
37. Crank, J., The mathematics of diffusion, Clarendon Press, Oxford, UK, 1975
38. Collepardi, M., Marcialis,A., and Turriziani,R., Penetration of chloride ions into cement pastes and concrete,
Journal of The American Concrete Society, Vol 55, No 10, Oct 1972
39. ASTM C1556 - 11, Standard test method for determining the apparent chloride diffusion coefficient of
cementitious mixtures by bulk diffusion, ASTM International, West Conshohocken, PA, USA, 2011
40. Ann, K.Y. Ahn, J.H., and Ryou, J.R., The importance of chloride content at the concrete surface in
assessing the time to corrosion of steel in concrete structures, Construction and Building Materials, Vol 23,
2009, pp 239-245
41. Meira, G.R., Andrade, C., Alonso, C., Borba Jr., J.C., Padilha Jr. M., Durability of concrete structures in
marine atmosphere zones-the use of chloride deposition on the wet candle as an environmental indicator,
Cement & Concrete Composites, Vol 32, No 6, pp 427-436, 2010
42. Gjørv, O. E. and Vennesland, O., Diffusion of chloride ions from seawater into concrete, Cement and
Concrete Research, Vol 9, pp 229-238, 1979
43. Farstad, T, Measurement of chloride in concrete- an evaluation of two different sampling techniques,
Nordic Concrete Research, Issue 15, 1994
44. Tang, L. and Gulikers, J., On the mathematics of time-dependent apparent chloride diffusion coefficient in
concrete, Cement and Concrete Research, Volume 37, Issue 4, Pages 589-595, April 2007
45. Bamforth, P., Price, W. F. and Emerson, M., An international review of chloride ingress into structural
concrete, Transport Research Laboratory, Berkshire, UK, 1997
46. Z7/07, Performance Tests to Assess Concrete Durability, Concrete Institute of Australia, Sydney, 2014
47. Dhir, R. K. et al, Rapid estimation of chloride diffusion coefficient in concrete, Magazine of Concrete
Research, Vol 42, No 152, pp 177-185, 1990
48. Tang, L. and Truc, O., Effect of exposure solution on chloride penetration test methods, Proceedings of
Second International RILEM Workshop on Testing and Modelling the Chloride Ingress into Concrete,
Edited by Andrade, C. and Kropp, J., 2000
49. AASHTO T 277-2007, Standard method of test for electrical indication of concrete’s ability to resist chloride
ion penetration, American Association of State Highway and Transportation Officials, Washington, DC.
50. ASTM, C1202 – 12, Standard test method for electrical indication of concrete's ability to resist chloride ion
penetration, ASTM International, West Conshohocken, PA, USA, 2012
51. Hooton, R.D., Pun, P., Kojundic, T., and Fidjestol, P., Influence of silica fume on chloride resistance of
concrete, Proceeding International Symposium of High Performance Concrete, Chicago, (P. Johal, Precast
Prestressed Concrete Institute) pp. 245-249, 1997

Durability Modelling of Reinforcement Corrosion in Concrete Structures 61


52. DuraCrete, Modelling of Degradation - Probabilistic Performance Based Durability Design of Concrete
Structures, EU - Brite EuRam III, Contract BRPR-CT95-0132, Project BE95-1347/R4-5, December pp 174,
1998
53. Page, C.L., Shott, N.R. and Tarras, E.L., A diffusion of chloride ions in hardened cement pastes, Cement
and Concrete Research, Vol 11, pp.395-406, 1981
54. Uji, K., Matsuaka, Y. & Marruya, Formulation of an equation for surface chloride content due to permeation
of chloride, Proceedings of the third International Symposium on Corrosion of Reinforcement in Concrete
Construction, Elsevier Applied Science, London, UK, 1990
55. Chisholm, D.H. and Lee, N.P., Durability prediction for coastal reinforced concrete structures – matching
reality and theory, New Zealand Concrete Society Conference, Taupo, 13-15 Oct, 2000
56. Nagataki, S., Otsuki, N., Wee, T.H. and Nakashita, K., Condensation of chloride ions in hardened cement
matrix materials and on embedded steel bars, ACI Materials Journal, Vol 90, No 4, pp 323-333, July-
August, 1993
57. Thomas, M.D.A and Bamforth, P.B., Modelling chloride diffusion in concrete – effect of fly ash and slag,
Cement and Concrete Research, Vol 29, pp.487-495, 1999
58. Cook, D.J., Hinczak, M., Jedy, M. and Cao, H.T., The behaviour of slag cement concretes in marine
environment-chloride ion penetration, Proceeding Third International Conference on the Use of Fly Ash,
Granulated Blast Furnace Slag, Silica Fume and Other Mineral By-Products in Concrete, Ed Malhotra,
Trondhein, pp 1467-1483, 1989
59. Cole, I.S., Muster, T.H., Lau, D. and Ganther, W.D., Some recent trends in corrosion science and their
application to conservation, Proceedings of Metal, Canberra, Australia, 2004
60. Angst, U. and Vennesland, Ø., Critical chloride content in reinforced concrete- state of the art, Concrete
Repair, Rehabilitation and Retrofitting II – Alexander et al Edt, Talyer & Francis Group, London, 2009
61. Hausmann, D.A. Steel corrosion in concrete-How does it occur?, Materials Protection, Vol 6, No 11, pp 19-
23, 1967
62. Gouda, V.K., Corrosion and inhibition of reinforcing steel: 1. Immersed in alkaline solution, British Corrosion
Journals, Vol 5, pp 198-203, 1970
63. Browne R.D. Design prediction of the life of reinforced concrete in marine and other chloride environments,
Durability of Building Materials, Vol. 3, Elsevier Scientific, Amsterdam, Netherlands, 1982
64. Andrade, C. and Alonso, C., Corrosion rate monitoring in the laboratory and on site, Construction and
Building materials, Vol 10, No 5, 1996, pp 315-328.46
65. JSCE (Japan Society of Civil Engineers), Standard Specification for Design and Construction of Concrete
Structures, Tokyo, 1999
66. Stark, D., Determination of permissible chloride levels in prestressed concrete, PCI Journal, pp 106-116,
July-August, 1984
67. Nielsen, A., Durability, Beton Bogen, Aalborg Cement Company, pp 200-243, Aalborg, Portland, 1985
68. Bhargava, K., Ghosh, A.K., Mori, Y., and Ramanujam, S., Analytical model for time to cover cracking in RC
structures due to rebar corrosion, Nuclear Engineering and Design, Vol 236, pp 1123-1139, 2006
69. Guirguis, S., A basis for determining minimum cover requirement for durability, In: Concrete Durability, ACI
Detroit, Michigan, 1987
70. Mallett, G.P., State-of-the-art review: Repair of Concrete Bridges, Thomas Telford, London, 1994
71. Ishida T. and Li, C., Modeling of carbonation based on thermo-hygro physics with strong coupling of mass
transport and equilibrium in micro-pore structure of concrete, Journal of Advanced Concrete Technology,
Vol 6, No 2, pp 303-316, 2008
72. Burden, D., The durability of concrete containing high levels of fly ash, Thesis for MSc Degree, University
of New Brunswick, Canada, 2003
73. Collepardi, M. Collepardi, S., Ogoumah Olagot, J.J., and Simoneli, F., The influence of slag and fly ash on
the carbonation of concrete, Eighth CANMET/ACI Int. Conf. on fly ash, silica fume, slag, and natural

Durability Modelling of Reinforcement Corrosion in Concrete Structures 62


pozzolans in concrete. Proceedings ACI Special Publication, 221-29, pp 483-493. Las Vegas, Nevada,
USA, 2004
74. Tam, C.T., Lim, H.B. and Sisomphon, K., Carbonation of concrete in the tropical environment of Singapore,
The IES Journal, Part A, Civil and Structural Engineering, Vol 1, No 2, pp146-153, May, 2008
75. Litvan, G.G. and Meyer, A., Carbonation of granulated blast furnace slag cement concrete during twenty
years of field exposure, Proceedings of Second International Conference on Fly Ash, Silica Fume, Slag,
and Natural Pozzolans in Concrete, ACI SP91, Vol 2, pp1445-1462, Madrid, Spain, 1986
76. He, R., and Jia, H., Carbonation depth prediction of concrete made with fly ash, Electronic Journal of
Geotechnical Engineering, Vol 16, 2011
77. Skjolsvold, O., Carbonation depths of concrete made with and without condensed silica fume, Proceedings
of 2nd International Conference on Fly ash, Silica Fume Slag and Natural Pozzolans in Concrete, ACI SP-
91, Vol II, pp1031-1043, Madrid, Spain, 1986
78. Wierig, H.J., Long term studies on the carbonation of concrete under normal outdoor exposure,
Proceedings of RILEM Seminar on the Durability of Concrete Structures under Normal Outdoor Exposure,
pp 239-249, Hanover, March, 1984
79. Webster, M.P., The assessment of corrosion-damaged concrete structures, PhD Thesis to University of
Birmingham, Birmingham, UK, 2000
80. Al-Sulaimani, G. J., Kaleemullah, M., Basunbal, I. A. and Rasheeduzzafar, Influence of corrosion and
cracking on bond behaviour and strength of reinforced concrete members, ACI Structural Journal, Vol 87,
No 2, pp 220-231, March-April 1990
81. Clark, L. A. and Saifullah, M., Effect of corrosion on reinforcement bond strength, Structural Faults and
Repairs, Vol 3, pp 113-119, 1993
82. Liu, Y., Modelling the time-to-corrosion cracking of the cover concrete in chloride contaminated reinforced
concrete structures, PhD Thesis to Virginia Tech, pp 117, October, 1996
83. Cabrera, J. G., Deterioration of concrete due to reinforcement steel corrosion, Cement and Concrete
Composites, Vol 18, pp 47-59, 1996
84. Alonso, C., Andrade, C., Rodriguez, J and Diez, J. M., Factors controlling cracking of concrete affected by
reinforcement corrosion, Materials and Structures, Vol 31, pp 435-441, 1998
85. Polder, R.B., The influence of blast furnace slag, fly ash and silica fume on corrosion of reinforced concrete
in marine environment, HERON, Vol 41 (1996), pp 287-300
86. Hunkeler F., Corrosion in reinforced concrete: processes and mechanisms, Corrosion in reinforced
Concrete structures. Ed Hans Bohni. pp1-42. Pub Woodhead Publishing Ltd, Cambridge, England. 2005.
87. Andrade, C. Resistivity test criteria for durability design and quality control of concrete in chloride
exposures. Concrete In Australia, Nov 2014
88. Papworth. F., Durability design for major concrete structures, Malaysian Institute of Engineers, 2008

Durability Modelling of Reinforcement Corrosion in Concrete Structures 63

S-ar putea să vă placă și