ACI SP 308 Chloride Thresholds and Limits For New Construction Tepke

An ACI Technical Publication
SYMPOSIUM VOLUME
Chloride Thresholds and Limits for

New Construction
SP-308
Editors:
David Tepke, David Trejo, and O. Burkan Isgor
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
Chloride Thresholds
and Limits for
New Construction
Editors:
David Tepke, David Trejo,
and O. Burkan Isgor
SP-308
@Seismicisolation
@Seismicisolation
First printing, June 2016
Discussion is welcomed for all materials published in this issue and will appear ten months from this journal’s date
if the discussion is received within four months of the paper’s print publication. Discussion of material received
after specified dates will be considered individually for publication or private response. ACI Standards published
in ACI Journals for public comment have discussion due dates printed with the Standard.
The Institute is not responsible for the statements or opinions expressed in its publications. Institute publications
are not able to, nor intended to, supplant individual training, responsibility, or judgment of the user, or the supplier,
of the information presented.
The papers in this volume have been reviewed under Institute publication procedures by individuals expert in
the subject areas of the papers.
Copyright © 2016
AMERICAN CONCRETE INSTITUTE
38800 Country Club Dr.
Farmington Hills, Michigan 48331
All rights reserved, including rights of reproduction and use in any form or by any means, including the making of
copies by any photo process, or by any electronic or mechanical device, printed or written or oral, or recording
for sound or visual reproduction or for use in any knowledge or retrieval system or device, unless permission in
writing is obtained from the copyright proprietors.
Printed in the United States of America
Editorial production: Gail Tatum
ISBN-13: 978-1-942727-94-1
@Seismicisolation
@Seismicisolation
Preface
The detrimental influence of chlorides on the corrosion of reinforcing steel in

concrete has been widely documented. The literature clearly shows that chloride
concentration at the steel level must exceed a critical chloride threshold to
initiate active corrosion of reinforcement embedded in concrete. It is now well
accepted that this critical chloride threshold is not a unique value, but rather a
range that depends on several factors. Regardless, placing concrete with chloride
concentrations above the critical chloride threshold for a particular situation
would result in active corrosion of the reinforcement and is therefore undesirable.
Unnecessarily restrictive limits, however, can lead to preclusion of some otherwise
acceptable materials or require use of supplemental materials or alternative mixture
designs that may increase costs or impact sustainability. Thus, there is a need from
a practical standpoint to establish conservative, yet reasonable, limits so that the
effects of corrosion can be managed without undue restrictions. ACI documents
place limits on the amount of chlorides that can be incorporated into new concrete
– these limits are referred to as the allowable admixed chloride limits.
Documents published by ACI Committees 201 and 222 currently recommend

limiting admixed chlorides based on a mass percentage of the portland cement
in the concrete mixture. Other documents, such as ACI 318, limit the admixed
chlorides based on weight percentage of cement. With the movement of the
industry towards greener systems, the inclusion of supplementary cementitious
materials (SCMs) as part of the cement could be beneficial. SCMs, however,
when used in large quantities, have been reported to decrease the pH of the
pore solution, which may lower the critical chloride threshold values. If the critical
chloride threshold values for concrete systems containing only portland cement
are different than the critical chloride threshold values for systems containing
portland cement and SCMs, the published allowable admixed chloride limits may
not be applicable. A further complication in establishing values exists due to
performance-based cements in which the specific amounts of SCMs might not be
known to the specifier.
This special publication (SP), based on two technical sessions held during the
Fall 2015 Concrete Convention and Exposition in Denver, CO, November 8-11,
2015, addresses challenges associated with allowable admixed chloride limits,
critical chloride thresholds, testing for the critical chloride threshold, binding of
chlorides in different systems, and how admixed chlorides influence service life.
Authors and presenters from North America and Europe provided a variety of
perspectives, experiences, and opinions. Based on the presentations, the open
discussion that followed the presentations, and the papers in this SP, evidence
indicates that allowable chloride limits should be based on cementitious materials
content including both portland cement and SCMs. However, because research
on the amount of chlorides required to initiate corrosion in systems containing
high SCM replacement levels suggests that there may be upper limits at which
the inclusion is appropriate, it was suggested that it may be appropriate to place
limits on the replacement percentages of SCMs used for calculations of cement
content when determining allowable admixed chloride limits. Although the
Denver sessions and the papers in this SP provide a significant move forward on
better defining allowable chloride limits and likely allow for refinement of current
recommendations in ACI documents, more research is needed.
@Seismicisolation
@Seismicisolation
On behalf of ACI Committees 201 and 222, the editors sincerely thank all authors and
presenters for their efforts and contributions to the presentations, open forum, and this
SP volume. Special thanks are extended to the peer reviewers of the manuscripts for their
constructive comments and recommendations. The editors are also indebted to the ACI staff
for their assistance in organizing the sessions, organizing the open forum, and in preparing this
volume. The editors earnestly hope that this symposium and SP volume will serve as a valuable
resource to those searching for data, guidance, and better clarity on allowable admixed
chloride limits in concrete.
Editors
David Tepke
Consultant
SKA Consulting Engineers Inc.
David Trejo
Professor & Hal Pritchett Endowed Chair
Oregon State University
O. Burkan Isgor
Associate Professor
Oregon State University
@Seismicisolation
@Seismicisolation
TABLE OF CONTENTS
SP-308—1
Chloride Threshold Values in Concrete – A Look Back and Ahead............................ 1.1
Authors: Ueli M. Angst and Bernhard Elsener
SP-308—2
Confusion on Chloride Limits in Specifications that Challenge the Industry............ 2.1
Author: Colin L. Lobo
SP-308—3
Effect of Temperature on the Chloride Binding of Portland Cement Exposed to
CaCl2..............................................................................................................................................3.1
Authors: Jianqiang Wei, Bernard Tao, and W. Jason Weiss
SP-308—4
Probabilistic Treatment of Chloride Threshold...................................................................4.1
Authors: Carmen Andrade, Fabiano Tavares, Nuria Rebolledo, and David Izquierdo
SP-308—5
Accelerated Mortar Test Method to Determine Chloride Threshold Values............. 5.1
Authors: Neal S. Berke, Matthew A. Miltenberger, Lianfang Li, Brian Miller, and
Ralf Carvajal
SP-308—6
Impact on Anticipated Service Life of Chloride Thresholds.......................................... 6.1
Author: Kyle Stanish
SP-308—7
The Influence of SCM Type and Quantity on the Critical Chloride Corrosion
Threshold..................................................................................................................................... 7.1
Authors: David Trejo and Cody Tibbits
SP-308—8
A Thermodynamic Perspective on Admixed Chloride Limits of Concrete
Produced With SCMs............................................................................................................... 8.1
Authors: Vahid Jafari Azad and O. Burkan Isgor
SP-308—9
Effect of Fly Ash and Silica Fume on Time to Corrosion Initiation for
Specimens Exposed Long Term to Seawater ................................................................... 9.1
Authors: Francisco J. Presuel-Moreno and Eric I. Moreno
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
SP-308—01
CHLORIDE THRESHOLD VALUES IN CONCRETE – A LOOK BACK AND AHEAD
Ueli M. Angst and Bernhard Elsener
Synopsis: Over the last 60 years, extensive research efforts aimed at determining the so-called chloride threshold
value in reinforced concrete. The belief that such a threshold exists is the root of all efforts to measure and model
chloride ingress into concrete. This paper addresses the usefulness of this established concept by evaluating the
experience available for portland cement systems. Additionally, it is critically discussed whether the concept can be
applied to modern materials, particularly SCMs. Finally, suggestions for future research are made.
It is concluded that the pronounced stochastic nature of the chloride threshold currently permits only corrosion
prognoses with large uncertainties. It is shown that even the most sophisticated chloride transport model in concrete
will not significantly improve this. Instead of refining mass transport models, future research should thus aim at finally
understanding the relevant parameters governing corrosion initiation in concrete. There is strong indication that a
number of such parameters are overlooked in the current concept of the chloride threshold value. We believe that as
long as initiation of chloride-induced corrosion is not fully understood, it does not make sense to continue applying
the (unsuccessful) concept of the chloride threshold value to modern materials.
Keywords: Chloride threshold value; Chloride induced corrosion; Critical chloride content; Durability,
Reinforcement corrosion; Service life modelling; Supplementary cementitious materials; Sustainabilit
1.1
@Seismicisolation
@Seismicisolation
Ueli M. Angst is a lecturer at the Institute for Building Materials at ETH Zurich, Switzerland, as well as a consultant
for the Swiss Society for Corrosion Protection (SGK). He studied civil engineering at ETH Zurich and received his
PhD from the Norwegian University of Science and Technology in Norway in 2011. His research interests include
corrosion of metals, methods and sensors for condition assessment and laboratory measurements, and durability of
reinforced and prestressed concrete.
Bernhard Elsener is professor at ETH Zurich, Institute for Building Materials, Zurich, Switzerland. He studied
materials science at the Department of Chemistry at ETH Zurich and received his PhD from ETH Zurich, Switzerland
in 1982. His research interests include corrosion of metals in general, electrochemical techniques, sensors, condition
assessment and durability of reinforced and post-tensioned structures, and corrosion of steel in concrete.
INTRODUCTION
In many countries around the world, portland cement has for decades been the most used type of cement to build
reinforced concrete infrastructure. In recent years, however, portland cement has increasingly been substituted with
supplementary cementitious materials (SCMs). Table 1 illustrates this by means of the example of Switzerland. While
for some of these materials there may be positive track records from certain countries – for instance for ground
granulated blast furnace slag that has been used in the Netherlands for roughly a century – the diversity of cementitious
materials and mineral binders to be used in the future is expected to increase, owing to, amongst others, reasons of
local availability of raw materials (Angst et al. 2012). The main motivation for reducing the “clinker factor”, i.e. to
increasingly substitute portland cement clinker partially with SCMs, is on the one hand to reduce the environmental
impact, particularly by reducing the carbon footprint, and on the other hand to improve the durability. Achieving
sustainability clearly not only requires decreasing the environmental footprint of the materials at the time of their
production, but to combine this with long and maintenance free service lives of the structures in their actual exposure
environments.
Due to the lack of long-term experience on the field performance of modern materials, we are forced to make
predictions. This is done based on theoretical reasoning and on accelerated laboratory (or field) testing, believed to
provide a basis to extrapolate the short-term behaviour to long-term and field conditions.
Essentially, these efforts consist in adapting concepts established with experience from portland cement and applying
them to the new materials. The conceptual understanding of degradation due to chloride-induced corrosion of
reinforcing steel in concrete evolved in the second half of the last century and has since then essentially been
unchanged. The widespread hypothesis is that the chloride concentration at the surface of the embedded steel is the
by far most important parameter, and hence that initiation of corrosion can be predicted by reducing the entire problem
to a question of chloride concentrations only. This is illustrated by the extensive research efforts performed over the
last 60 years, seeking to determine the so-called chloride threshold value or critical chloride content (Angst et al.
2009), i.e. to specify a threshold concentration below which, conceptually, there is no corrosion and above which
corrosion occurs.
The belief that such a threshold exists is also the root of all research efforts to measure and model chloride ingress
into concrete. This issue received considerable research attention over the last decades, and a high number of test
methods have been devised internationally for assessing the ability of concrete to resist chloride penetration, as e.g.
reviewed in Tang et al. (2012).
Table 1 — Types of cement sold in Switzerland in the years 1996, 2006, and 2015, illustrating the sharp transition
from Portland cement to blended cements; data from (CemSuisse).
Cement type Designation (acc. to EN 197-1) 1996 2006 2015

Portland cement CEM I 87% 37% 13%
Blended Portland cement CEM II 11% 60% 83%
Slag cement CEM III <1% <2% <1%
Other <2% <2% <4%
1.2
@Seismicisolation
@Seismicisolation
Chloride Threshold Values in Concrete – A Look Back and Ahead
The consequence of the concept of the chloride threshold value is that materials such as cement types or concrete
formulations are commonly regarded as parameters, i.e. that chloride ingress properties and chloride threshold values
can be determined by test methods for these different materials in order to predict the time to corrosion initiation.
This paper addresses first the usefulness of this established concept by evaluating the experience available for portland
cement systems. Subsequently, it is critically discussed whether the concept can be applied to modern materials,
particularly SCMs, and which factors are currently seen to potentially handicap this application. Finally, suggestions
for future research are made.
EXPERIENCE FROM PORTLAND CEMENT SYSTEMS
Literature data
Laboratory and field data—In a literature review performed by Angst et al. (2009), the variability of published
chloride threshold values was found to be extremely high. Values determined in laboratory studies scattered from
almost 0 to more than 8% chloride by mass of binder if all cement types are considered; if only studies with portland
cement (excluding low C3A cements) are considered, the results span still from almost 0 to more than 3% chloride by
mass of cement. No systematic trends could be found such as with respect to water/cement ratios, methods of
accelerating chloride ingress, storage conditions (e.g. humidity), and corrosion detection methods.
It may also be worth mentioning that within this huge span of reported laboratory chloride threshold values, the results
of most individual studies covered only a small fraction of the entire span. In other words, some studies tended to
give results in the lower range of the literature span, some were on an intermediate level, and some studies yielded
results in the higher end of the literature span. Thus, the entire variability in the literature seems to arise from the
combination of different experimental parameters used in each study, i.e. from the measurement method.
Considering data from field exposure sites or from measurements on real structures shows a better agreement between
individual reports, but still a large variability. In the 1970s and 1980s, a number of studies reported that chloride
threshold values scattered in the range 0.1 to 2% chloride by weight of cement (Stratfull et al. 1975, Vassie 1984,
Treadaway et al. 1989). It can be assumed that all of these studies dealt with portland cement.
Figure 1 shows a compilation of literature data providing information on the statistical distribution of the chloride
threshold value. Note that data on the probability of corrosion initiation at certain chloride concentrations is scarce,
and that not many more such references can be found in the literature. The data shown in Figure 1 as triangles stems
from laboratory studies. In the work of Zimmermann (2000) ribbed carbon steel rebars (diameter 8 mm (0.315 inches),
length 65 mm (2.56 inches)) were embedded in mortar (w/c=0.6). The specimens were immersed in a chloride
solution; initiation of corrosion was determined by monitoring the steel potential. In another laboratory study (Breit
2001), similar specimens were used (rebar diameter 10 mm (0.39 inches), length 55 mm (2.17 inches), smooth steel
bars; w/c=0.5–0.6), but they were not left at their open circuit potential during immersion in chloride solution, but
polarized to relatively positive potentials (+0.5 V vs. SHE). Both experimental setups can only mimic reality to a
certain extent; differences in this regard have been discussed in (Angst et al. 2011b). Nevertheless, both setups permit
to accurately detect corrosion onset and thus, the determined chloride content at the time of corrosion initiation, is
considered to represent well the chloride threshold value (under the given conditions). In contrast, the well-known
data from field experience reported by Vassie (1984) (bridge structures in the UK) may include larger uncertainties
since the time of corrosion initiation was not accurately known. Nevertheless, the results stem from real structures
and thus, they are free from artefacts that typically arise from the way specimens are produced and exposed in
accelerated laboratory testing (Angst et al. 2009). Figure 1 also shows the distribution of the chloride threshold value
indicated in the fib model code for service life design (International Federation for Structural Concrete (fib) 2006),
which is based on the data by Breit (2001).
1.3
@Seismicisolation
@Seismicisolation
Recent field data from Switzerland—In a recent study by ETH Zurich, a more than 40 years old tunnel in the Swiss
Alps was investigated. Cores of 150 mm (5.9 inches) were drilled from within a concrete area of less than 2 m2 (10.8
ft2) in size, thus presumably from the same batch of concrete. Each core centrally contained a segment of reinforcing
steel. The location of the sampling area was selected based on condition assessment of the structure in order to ensure
that the reinforcement was still in passive condition. Additional measurements also indicated a negligible level of
chloride concentrations at the steel depth (in this area) and that the carbonation front was still far from the steel depth.
The samples were transferred to the laboratory and underwent a procedure of sample preparation. This included
establishing a cable connection to the rebars, special actions to prevent the cut faces of the rebars from initiating
corrosion (by applying a dense mortar), and subsequent coating of the lateral faces of the core with an epoxy-resin.
The samples were then subjected to wetting/drying exposure in the laboratory (with stepwise increasing chloride
concentration in the exposure solution) and the potential of the rebars was monitored vs. a reference electrode.
Detection of corrosion initiation was based on a criterion related to a drop in potential as established in previous work
(RILEM , Angst et al. 2011c). At the time of corrosion initiation, the chloride content at the steel surface was
measured, thus yielding the chloride threshold value. A more detailed description of the procedure can be found in
(Wagner et al. submitted to advisory commission). The advantage of the used approach lies in ensuring realistic
specimens thanks to taking them from real structures (mature concrete, realistic conditions at the steel/concrete
interface, etc.) and combining them with the accuracy of laboratory measurement methods for detecting corrosion
initiation. The results are shown in Figure 1, denominated “Swiss field data”.
It is evident from Figure 1 that the field data (Vassie’s results and Swiss data) scatters somewhat more than the
laboratory data (Zimmermann, Breit). This may be explained by the higher variability in material properties in real
structures compared with laboratory samples that are usually cast, compacted, cured, etc. under more controlled
conditions. One can also see that the field data scatters more than what is used in the fib model code for service life
design. Nevertheless, one may argue that all the data in Figure 1 indicates a probability of corrosion initiation of
approx. 20% at chloride contents in the range 0.3–0.5 % by cement weight. This agrees also rather well with the
famous threshold value of 0.4% chloride by weight of cement suggested in the 1960s by Richartz (1969).
A second observation from Figure 1 is that the chloride content associated with corrosion initiation in reinforced
concrete is not a unique threshold value, but must be described with a statistical distribution. The field data suggests
that steel may be corroding or still passive at chloride contents ranging from 0.2% chloride by weight of cement to
more than 1.5% chloride by weight of cement. Towards the lower end of this range, the probability for corrosion is
Figure 1 — Literature and own data on the statistical distribution of chloride threshold values in portland
cement systems. Circles = field data, triangles = laboratory data; solid line = distribution suggested in the fib
model code for service life design; see text for explanations.
1.4
@Seismicisolation
@Seismicisolation
Figure 2 — Example of forecasting the time to corrosion by a simple chloride ingress model (see text for
explanations) for two different cover depths and as a function of the assumed chloride threshold value (deterministic
model).
low, with increasing chloride concentrations, the probability for corrosion increases. Even at values as high as 1%
chloride by cement weight, the field data suggests that the probability for corrosion is not much higher than 50% (in
portland cement systems).
Implications for service life modelling

The fact that the distribution of the chloride threshold value spans over such large ranges has major implications for
projecting the time to corrosion initiation in service life modelling. This may be illustrated with a simple model of
chloride ingress, based on Fick’s second law of diffusion and the well-known error function solution, as described
e.g. in the fib model code for service life design (International Federation for Structural Concrete (fib) 2006). The
chloride concentration in the concrete c(x,t) as a function of depth x and time t can be expressed as follows:
, 1 (1)
2 ∙
Assuming a surface chloride concentration cs of 3% chloride by weight of cement, a negligible initial chloride content
co in the concrete, and a constant apparent diffusion coefficient Dapp of 1 x 10–12 m2/s, the chloride content in the
concrete can be predicted for different cover depths. For a given chloride threshold value, it is then possible to predict
the time to corrosion initiation. Note that this example is a simple deterministic model, where any variability in the
input parameters, e.g. in the chloride diffusion coefficient, the surface chloride concentration, the cover depth, etc. is
neglected.
Figure 2 illustrates the wide range in predicted time to corrosion that results from different chloride threshold values.
Note that the chosen range in chloride threshold values corresponds to the one of the field data shown in Figure 1.
Depending on the actual value assumed for the chloride threshold, for a cover depth of 50 mm, the projected time to
corrosion may be as low as 15 years or higher than 80 years.
It is important to recognize that even small differences in the assumed chloride threshold value have a strong impact
on the resulting time to corrosion (and thus on the predicted service life). This is due to the fact that particularly at
higher cover depths and towards later stages (typically after a few decades), the increase in chloride concentration vs.
time becomes moderate. Note that this marked sensitivity of the prediction to the assumed chloride threshold becomes
even more pronounced the higher the level of chloride threshold values.
1.5
@Seismicisolation
@Seismicisolation
Probabilistic considerations—By taking into account the probability distribution of the chloride threshold (as shown
in Figure 1) the modelled time to corrosion may also be presented differently, namely by plotting the probability at
which corrosion has initiated after a certain numbers of years. This is shown in Figure 3.
For instance, at a cover depth of 50 mm, there is a 10% probability that corrosion has initiated after approx. 14 years,
after only 5 more years (i.e. after 19 years), this probability has already increased to 20%. However, after a service
life of 100 years, there is still a probability of approx. 20% that no corrosion has initiated. Note that the probability
for corrosion increases rather drastically at early ages (such as a jump in 10% probability from 14 to 19 years for 50
mm cover depth). This is the direct result of the relatively steep increase in chloride concentrations at young exposure
times combined with the fact that the lower tail of the probability distribution of the chloride threshold value spreads
down to relatively low chloride concentrations (in the range of 0.2% chloride by cement weight). This presents a
potential problem, since the acceptable probability for corrosion, is usually chosen in the lower end, such as 5%, 10%,
or eventually 30%. Obviously, the model uncertainties are largest in this range.
Note that in this example, the projection of the chloride content in the concrete at the steel surface was done
deterministically, i.e. with fixed input parameters. Recognizing that also these parameters are variable and thus
describing them with help of a statistical distribution, which is common practice, would add even further uncertainties
to the model predicting the time to corrosion initiation.
Large uncertainties in current modelling—In any case, the calculations plotted in Figure 2 and Figure 3 – based on
realistic input parameters for chloride ingress modelling and on field data for the chloride threshold value – clearly
reveal the large uncertainties inherent to current predictive modelling. From an engineering perspective, one may
question the benefit of a prognosis stating that corrosion will initiate sometime between 15 and 100 years after
construction. It is time to admit that the established, state-of-the-art approach to forecast the time to corrosion has a
very weak predictive power. As a result of the high inherent uncertainties of this approach, engineers, owners, or
infrastructure managers are forced to take highly conservative decisions, which in turn implies high costs to our
society.
Considering corrosion initiation in concrete as a mere question of chlorides does and will only permit corrosion
prognosis with a very low level of confidence. Having in mind the vast statistical dispersion of the chloride threshold
value (Figure 1), it may also be important to recognize that even the most sophisticated model to predict chloride
ingress into concrete will not significantly improve this. In our opinion, the concept of the chloride threshold value
needs to be replaced or enriched with a more refined approach capable of taking into account all other dominant and
relevant influencing factors.
Figure 3 — Example of forecasting the time to corrosion by a simple chloride ingress model (see text for
explanations) for two different cover depths. The plot details the probability at which corrosion has initiated after a
certain numbers of years, assuming a distribution of the chloride threshold value according to the “Swiss field data”
shown in Figure 1.
1.6
@Seismicisolation
@Seismicisolation
APPLYING THE ESTABLISHED CONCEPT TO MODERN MATERIALS
The concept of the chloride threshold value implies that any material characteristics – such as the type of cement,
concrete mix proportions, the steel grade, etc. – can be regarded as a test parameter that influence:
1) The kinetics of chloride ingress into the concrete

2) The chloride threshold value
Ideally, these influences can be tested and quantified in order to provide the input data needed in predictive modelling.
Concerning the kinetics of chloride penetration, there seems to be some literature agreement about the effect of certain
concrete mix related parameters. It is well established that decreasing the w/c ratio and/or using supplementary
cementitious materials (at least some of them) can create a denser, more tortuous concrete pore system and thus
improve the resistance against chloride penetration (Bertolini et al. 2013). This can also reasonably be quantified by
experimentally determining the chloride diffusion coefficient and the ageing coefficient of the material under study
if a Fick’s second law of diffusion type of chloride ingress model is used (Tang et al. 2012). Alternatively, for more
sophisticated mass transport models, different parameters need to be quantified (e.g. (Marchand and Samson 2009,
Tang et al. 2012)). Nevertheless, the current state-of-the-art permits relatively well predicting chloride penetration
through concrete for different systems.
Concerning the second point named above, the chloride threshold value, the literature documents a considerable
number of attempts to experimentally determine the chloride threshold value for fly ash concrete, slag concrete, etc.
or for different w/c ratios (Angst et al. 2009). Despite this, the current state-of-the-art offers very little convincing
agreement on these parameters, particularly concerning the type of binder. This may be illustrated with help of the
example of fly ash: While some authors reported decreasing chloride threshold values with increasing fly ash contents
(Thomas 1996, Oh et al. 2003), others found increasing chloride threshold values for concrete with fly ash (Schiessl
and Breit 1996) or no influence at all (Alonso et al. 2002). This literature disagreement was in (Angst et al. 2009)
explained by the huge differences in experimental approaches used by the different authors, and the unfortunate fact
that some of the parameters inherent to experimental setups have a more pronounced influence on the resulting
chloride threshold value than the variables under test.
Potential problems with the established concept

A variety of parameters – other than the chloride concentration – have been observed to play a role in corrosion
initiation in chloride contaminated concrete. This includes the pH of the concrete pore solution, the reinforcing steel
potential, and the conditions at the steel/concrete interface. The following sections outline a few potential issues
arising from the use of SCMs that may complicate the application of the concept of the chloride threshold value to
these systems.
Alkalinity of the concrete pore solution—The introduction of non-portlandite binders has strongly increased the
diversity of the pore solution chemistry in concrete. One important factor is that the pH of systems with SCMs is
typically lower than the pH in portland cement systems. Additionally, also the pH buffer capacity is generally
depressed as a result of the reduction (or elimination) of the calcium hydroxide reserve. This calcium hydroxide
reserve is one of the main reasons for the corrosion inhibiting nature of portland cement systems (Bäumel 1959, Page
1975). From experiments in aqueous solutions, it is well known that the chloride concentration needed to initiate
corrosion increases with increasing pH (Venu et al. 1965, Hausmann 1967, Gouda 1970). This is reflected in the
suggestion of expressing the chloride threshold value as the ratio chloride/hydroxyl ions (Hausmann 1967).
Nevertheless, it is still common practice to express the chloride threshold value as percentage of chloride by weight
of binder also for blended cements, thus neither the pH nor the pH buffer capacity are directly taken into account.
Chloride binding capacity—SCM containing concretes typically have a different ability to bind chlorides than
ordinary portland cement concrete (Larsen 1998). Thus, when measuring (or predicting) total chloride contents in the
concrete, the relative fractions of bound and free chlorides are different in modern systems as compared to traditional
portland cement concrete. This directly affects the ratio chloride/hydroxyl ions in the concrete pore solution and thus
initiation of corrosion.
Establishing stable pitting corrosion—The reduction in pH buffer capacity in the presence of SCMs and the
differences in chloride binding capacity likely play a role in establishing stable pitting corrosion. It is well known that
1.7
@Seismicisolation
@Seismicisolation
initiation of pitting corrosion occurs as a sequence from pit nucleation to metastable pitting to stable pit growth
(Burstein et al. 1993). For steel in concrete, stable pit growth may in many cases not directly establish; due to
insufficient supply of (free) chloride ions to the nucleated pit, the steel may undergo several repassivation events
before stable pitting is established (Angst et al. 2011a, Angst et al. 2011c). The factors governing this in concrete are
the availability of free, thus mobile hydroxyl and chloride ions in vicinity of the nucleated pit. Both these species
migrate and diffuse into the pit in a competitive manner. For stable pit growth to be achieved, it is needed that transport
of chloride becomes dominant over transport of hydroxyl ions (Angst et al. 2011a). After a certain time, these species
will be transported from more remote areas; additionally they will also be supplied by dissolving solid phases of the
cement paste (e.g. calcium hydroxide) and by releasing bound chlorides from the concrete at the pit mouth. Although
it appears that the pH buffer capacity and the chloride binding capacity of the concrete are decisive parameters in this
regard, their influence is currently not well understood and can at present not yet be taken into account in predictive
models.
Sulphides and other species in the concrete pore solution—From experience with slag cements, it is known that the
presence of sulphides can strongly depress the potential of the embedded reinforcing steel. In (Garcia et al. 2014) for
instance, potentials more negative than –500 mV SCE were reported for slag containing concrete (in the absence of
chlorides). It is known that more negative steel potentials can largely increase the chloride concentration needed to
initiate corrosion (Alonso et al. 2002, Angst et al. 2009). Different hypotheses have been suggested to explain the
effect of the sulphides (from the slag) on the steel potential. Macphee and Cao suggested that oxidation of the
sulphides consumes the oxygen in the concrete pore solution, and thus, the lack of oxygen is seen as the cause of the
negative steel potential (Macphee and Cao 1993). Other authors suggested that different oxidation/reduction couples
involving sulphides determine the steel potential, e.g. (Holloway and Sykes 2005). In any case, there is a lack of
thorough understanding of the processes occurring in the presence of species such as sulphides in the concrete pore
solution and how they affect corrosion initiation.
Steel concrete interface—It is recognized that voids, bleed water zones, cracks, pre-existing rust layers, mill scales,
etc. can act as preferable corrosion initiation sites. However, the accurate mode of action of these characteristics in
promoting corrosion initiation is currently not well understood. Nevertheless, there is indication that some
characteristics at the steel/concrete interface play a highly dominant role in corrosion initiation (Angst et al. 2009). It
is the aim of the recently established RILEM technical committee SCI (RILEM) to compile in detail the available
literature and knowledge in this field and to identify the areas, where more research work is most urgently needed.
Gaining fundamental and mechanistic understanding concerning the influence of individual characteristics at the
steel/concrete interface is important also in the context of SCMs as it is well known that blended cements can greatly
modify some of the properties of the steel/concrete interface.
Corrosion propagation stage—Finally, it is common practice to understand the service life of a structure as the time
to corrosion initiation. This is primarily due to the fact that chloride-induced and thus localized corrosion is often
associated with relatively high rates of loss of steel cross sectional area, which may give rise to critical conditions
from the viewpoint of serviceability or structural safety within relatively short time (Bertolini et al. 2013). Thus, the
propagation stage may indeed in many practical cases be short in comparison with the initiation stage. Another reason
for conservatively understanding the end of the service life as the time to corrosion initiation can also be found simply
in the lack of reliable models for the corrosion propagation stage (Raupach 2006, Otieno et al. 2011, Hornbostel et al.
2013).
Nevertheless, in non-portlandite systems, corrosion may in principle initiate relatively early and at lower chloride
concentrations than in portland cement systems (due to the lower pore solution pH), but the rate of steel mass loss
may still be lower. One possible reason for this is the decreased mass transport in many of the modern systems, which
can potentially slow down all the different reaction steps of the corrosion process (e.g. the cathodic reaction by
limiting oxygen access (Hassan et al. 2000), the anodic reaction by limiting mass transport at the anode (Page and
Havdahl 1985), the macro-cell current flow in the bulk concrete (Raupach 1992), etc.). Thus, for concretes containing
SCMs it may be even more important than for portland cement concrete to include also the propagation stage in
service life models. Continuing to neglect this may not only lead to highly conservative predictions, but also be
misleading when comparing the durability (and sustainability) of different modern and traditional binders.
1.8
@Seismicisolation
@Seismicisolation
SUGGESTIONS FOR FUTURE RESEARCH
From the previous sections, it is apparent that many questions regarding the mechanism of corrosion initiation of
reinforcing steel in concrete are still open. We believe that as long as we do not understand the governing factors for
corrosion initiation in chloride contaminated concrete it does not make sense to continue applying the (unsuccessful)
concept of the chloride threshold value. Instead, research is urgently needed to clarify some fundamental issues. Only
this will allow for a real breakthrough in forecasting the durability of concrete structures.
From the current state-of-the-art it is known that the pH, the pH buffer capacity, and the chloride binding capacity are
important parameters. Models to forecast corrosion could thus be improved when they are capable of taking into
account the free chloride ion concentrations and the pH of the concrete pore solution rather than only the total chloride
content in the concrete. However, particularly for condition assessment of existing structures, there are no feasible
methods to accurately and reliably determine the free chloride ion concentrations and the pH. Also for durability
predictions of new structures, to be designed with novel materials, accurately modelling the pore solution chemistry
during long-term field exposure is a complicated task, also because the experimental validation is not straightforward.
Future research should thus address the following issues:
 Models capable of predicting the pore solution chemistry (rather than only the chloride concentration) during
long-term field exposure, and capable of taking into account the release of bound chlorides and the changes
in pH at the pit mouth during early, metastable stages of pitting corrosion are expected to significantly
improve the accuracy of prognoses of the time to reach stable pitting corrosion (thus the corrosion initiation
stage).
 Reliable methods for the determination of free chloride and hydroxyl concentrations in the concrete pore
solution, particularly also methods suitable for measurements on-site. This will permit to validate predictive
models.
 The effect of sulphides on corrosion initiation needs to be understood in more detail, as there is indication
that these cannot be neglected in corrosion prognoses. The same applies to other potential differences in pore
solution chemistry of concretes made with SCMs.
 There is an urgent need to understand the role of individual characteristics at the steel/concrete interface.
Their influence needs to be clarified and mechanistically understood as only this can lay the basis to develop
novel predictive models taking into account the relevant parameters.
 Mechanistic models of corrosion propagation in concrete in order to replace to current, largely empirical
models. This will permit taking into account the corrosion propagation stage that may form a substantial part
of the service life of concretes made with SCMs.
CONCLUDING REMARKS
This paper discussed the established concept of the chloride threshold value by evaluating the experience available
for portland cement systems. Additionally, potential issues arising from the use of SCMs that may complicate
application of this concept to these systems were highlighted. The following major conclusions are drawn:
 The chloride threshold inherently exhibits stochastic variability and cannot be treated as a deterministic
threshold value. Both field and laboratory data indicate that the probability distribution for corrosion
initiation covers a wide range of chloride concentrations – even for a specific concrete or structure and when
determined with one unique test method.
 The implication of the pronounced stochastic nature of the chloride threshold is that corrosion prognoses can
currently only be made with large uncertainties. This was in this work illustrated with help of an example,
considering state-of-the-art data on the distribution of the chloride threshold value.
1.9
@Seismicisolation
@Seismicisolation
 From this example, it became also apparent that even the most sophisticated model to predict chloride ingress
into concrete will not significantly improve this. Instead of focusing on developing more advanced mass
transport models, we thus suggest that future research efforts should aim at finally understanding the relevant
parameters governing corrosion initiation in concrete.
 There is strong indication that corrosion initiation in concrete can be significantly affected by a number of
parameters that are overlooked in the current concept of the chloride threshold value. Thus, considering
corrosion initiation in concrete as a mere question of chlorides does and will only permit corrosion prognoses
with a very low level of confidence. Only if the concept of the chloride threshold value can be replaced (or
at least enriched) with a more refined approach capable of taking into account all relevant factors influencing
corrosion initiation, can the current large uncertainties in predictive modelling be reduced.
 Finally, corrosion initiation and propagation are likely to be mechanistically different in concrete made with
SCMs compared to portland cement. This handicaps applying the established (largely unsuccessful) concept
of the chloride threshold value to modern binders. There is thus an urgent need for fundamental and
mechanistic understanding of the relevant influences.
ACKNOWLEDGEMENTS
The authors acknowledge the financial support of the Institute for Building Materials, ETH Zurich, Switzerland.
REFERENCES
Alonso, C., Castellote, M. and Andrade, C. (2002). "Chloride threshold dependence of pitting potential of
reinforcements." Electrochimica Acta 47(21): 3469-3481.
Angst, U., Elsener, B., Larsen, C. K. and Vennesland, Ø. (2009). "Critical chloride content in reinforced concrete - A
review." Cement and Concrete Research 39(12): 1122-1138.
Angst, U., Elsener, B., Larsen, C. K. and Vennesland, Ø. (2011a). "Chloride induced reinforcement corrosion: Rate
limiting step of early pitting corrosion." Electrochimica Acta 56(17): 5877-5889.
Angst, U., Rønnquist, A., Elsener, B., Larsen, C. K. and Vennesland, Ø. (2011b). "Probabilistic considerations on the
effect of specimen size on the critical chloride content in reinforced concrete." Corrosion Science 53(1): 177-
187.
Angst, U. M., Elsener, B., Larsen, C. K. and Vennesland, Ø. (2011c). "Chloride induced reinforcement corrosion:
Electrochemical monitoring of initiation stage and chloride threshold values." Corrosion Science 53(4): 1451-
1464.
Angst, U. M., Hooton, R. D., Marchand, J., Page, C. L., Flatt, R. J., Elsener, B., Gehlen, C. and Gulikers, J. (2012).
"Present and future durability challenges for reinforced concrete structures." Materials and Corrosion 63(12):
1047-1051.
Bäumel, A. (1959). "Die Auswirkung von Betonzusatzmitteln auf das Korrosionsverhalten von Stahl in Beton (in
German)." Zement-Kalk-Gips 7: 294-305.
Bertolini, L., Elsener, B., Pedeferri, P., Redaelli, E. and Polder, R. B. (2013). Corrosion of Steel in Concrete:
Prevention, Diagnosis, Repair, WILEY-VCH.
Breit, W. (2001). Critical corrosion inducing chloride content – State of the art and new investigation results.
Betontechnische Berichte 1998-2000. Düsseldorf, Germany, Verein Deutscher Zementwerke e.V. (VDZ).
Burstein, G. T., Pistorius, P. C. and Mattin, S. P. (1993). "The nucleation and growth of corrosion pits on stainless
steel." Corrosion Science 35(1-4): 57-62.
CemSuisse. Webpage of the Association of the Swiss Cement Industry. Retrieved Oct. 20, 2015, from
http://www.cemsuisse.ch/.
Garcia, V., Francois, R., Carcasses, M. and Gegout, P. (2014). "Potential measurement to determine the chloride
threshold concentration that initiates corrosion of reinforcing steel bar in slag concretes." Materials and
Structures 47(9): 1483-1499.
Gouda, V. K. (1970). "Corrosion and corrosion inhibition of reinforcing steel. I. Immersed in alkaline solutions,."
British Corrosion Journal 5: 198-203.
Hassan, K. E., Cabrera, J. G. and Maliehe, R. S. (2000). "The effect of mineral admixtures on the properties of high-
performance concrete." Cement & Concrete Composites 22(4): 267-271.
1.10
@Seismicisolation
@Seismicisolation
Hausmann, D. A. (1967). "Steel corrosion in concrete. How does it occur? ." Materials Protection 6: 19-23.
Holloway, M. and Sykes, J. M. (2005). "Studies of the corrosion of mild steel in alkali-activated slag cement mortars
with sodium chloride admixtures by a galvanostatic pulse method." Corrosion Science 47(12): 3097-3110.
Hornbostel, K., Larsen, C. K. and Geiker, M. R. (2013). "Relationship between concrete resistivity and corrosion rate
- A literature review." Cement & Concrete Composites 39: 60-72.
International Federation for Structural Concrete (fib) (2006). fib bulletin No. 34: Model code for service life design.
Lausanne, Switzerland.
Larsen, C. K. (1998). Chloride binding in concrete. Dissertation, No. 1998:101, Norwegian University of Science
and Technology, NTNU.
Macphee, D. E. and Cao, H. T. (1993). "Theoretical description of impact of blast-furnace slag (BFS) on steel
passivation in concrete." Magazine of Concrete Research 45(162): 63-69.
Marchand, J. and Samson, E. (2009). "Predicting the service-life of concrete structures - Limitations of simplified
models." Cement & Concrete Composites 31(8): 515-521.
Oh, B. H., Jang, S. Y. and Shin, Y. S. (2003). "Experimental investigation of the threshold chloride concentration for
corrosion initiation in reinforced concrete structures." Magazine of Concrete Research 55(2): 117-124.
Otieno, M. B., Beushausen, H. D. and Alexander, M. G. (2011). "Modelling corrosion propagation in reinforced
concrete structures - A critical review." Cement & Concrete Composites 33(2): 240-245.
Page, C. L. (1975). "Mechanism of corrosion protection in reinforced concrete marine structures." Nature 258: 514-
515.
Page, C. L. and Havdahl, J. (1985). "Electrochemical monitoring of corrosion of steel in microsilica pastes." Materials
and Structures 18: 41-47.
Raupach, M. (1992). Zur chloridinduzierten Makroelementkorrosion von Stahl in Beton (in German) Doctoral thesis,
Deutscher Ausschuss für Stahlbeton (Heft 433), Aachen University, Germany.
Raupach, M. (2006). "Models for the propagation phase of reinforcement corrosion - an overview." Materials and
Corrosion 57(8): 605-613.
Richartz, W. (1969). "Die Bindung von Chlorid bei der Zementerhärtung (in German)." Zement-Kalk-Gips 10: 447-
456.
RILEM Technical committee 235-CTC "Corrosion initiating chloride threshold concentrations in concrete", chair:
Dr. Luping Tang, started in 2009.
RILEM Technical committee 262-SCI "Characteristics of the steel/concrete interface and their effect on initiation of
chloride-induced reinforcement corrosion", chair: Dr. Ueli Angst, started in 2014.
Schiessl, P. and Breit, W. (1996). Local repair measures at concrete structures damaged by reinforcement corrosion -
aspects of durability. Proc. 4th Int. Symp. "Corrosion of Reinforcement in Concrete Construction", The Royal
Society of Chemistry: 525-534.
Stratfull, R. F., Jurkovich, W. J. and Spellman, D. L. (1975). "Corrosion testing of bridge decks." Transportation
Research Records 539: 50-59.
Tang, L., Nilsson, L.-O. and Basheer, M. P. A. (2012). Resistance of concrete to chloride ingress: testing and
modelling. Boca Raton, CRC Press, Taylor & Francis Group.
Thomas, M. (1996). "Chloride threshold in marine concrete." Cement and Concrete Research 26: 513-519.
Treadaway, K. W. J., Cox, R. N. and Brown, B. L. (1989). "Durability of corrosion resisting steels in concrete."
Proceedings of the Institution of Civil Engineers Part 1 86: 305-331.
Vassie, P. (1984). "Reinforcement corrosion and the durability of concrete bridges." Proceedings of the Institution of
Civil Engineers Part 1 76: 713-723.
Venu, K., Balakrishnan, K. and Rajagopalan, K. S. (1965). "A potentiokinetic polarization study of the behaviour of
steel in NaOH–NaCl system." Corrosion Science 5: 59-69.
Wagner, M., Angst, U. and Elsener, B. (submitted to advisory commission). Research report "Methode zur
Bestimmung des kritischen Chloridgehalts an bestehenden Stahlbetonbauwerken" (in German), Swiss Federal
Roads Office.
Zimmermann, L. (2000). Korrosionsinitiierender Chloridgehalt von Stahl in Beton (in German). Dissertation, ETH
Nr. 13870, ETH Zurich, Switzerland.
1.11
@Seismicisolation
@Seismicisolation
1.12
@Seismicisolation
@Seismicisolation
SP-308—02
CONFUSION ON CHLORIDE LIMITS IN SPECIFICATIONS THAT CHALLENGE THE INDUSTRY
Colin L. Lobo
Synopsis: Chloride limits in concrete are intended to ensure that composition of concrete mixtures do not cause
corrosion of reinforcement. This paper provides a brief overview of the requirements in industry standards and
discusses some of the test methods used for chloride limits. Estimation of total chloride content in concrete mixtures
calculated from the materials and mixture proportions is outlined. It is suggested that the calculated chloride content
is a conservative estimate of the available chlorides for corrosion and can be used as an alternative to testing for water
soluble chlorides in concrete. The results of a survey of ready mixed concrete producers indicate that chloride limits
in concrete are not often required or enforced. Requirements in specifications for concrete vary considerably, are not
consistent with Code requirements and often result in contradictory interpretation and requirements.
Keywords: Building code; Chlorides; Corrosion; Durability; Exposure classes; Specification; Specification limits.
2.1
@Seismicisolation
@Seismicisolation
Colin L. Lobo
Colin L. Lobo is the senior vice president of the Engineering Division at the National Ready Mixed Concrete
Association (NRMCA). He is active in representation of the concrete industry in several technical committees of ACI
and ASTM. He is a member of ACI 132, ACI 301, ACI 318, ACI 329, and several others. He is active on several
ASTM subcommittees that maintain standards for cement, aggregates and concrete. He manages NRMCA technical
education programs and is responsible for certification programs for industry personnel and production facilities. He
is involved in the research activities of the NRMCA. Colin Lobo has a Ph.D. on concrete materials from Purdue
University in West Lafayette, Indiana. He is a licensed professional engineer in the state of Maryland.
INTRODUCTION
Corrosion of reinforcing steel in concrete is a significant contributor to the deterioration of infrastructure that includes
bridges, buildings, pavements and other structures in a wide range of environments. Reinforcing steel is naturally
passivated from corrosion when embedded in the highly alkaline environment in concrete. Corrosion initiates when a
threshold chloride concentration is exceeded at the reinforcing steel or the concrete cover has carbonated to a level
that the alkalinity is reduced to negate the passivation. Industry standards attempt to minimize corrosion by ensuring
good quality concrete with adequate cover to extend the time it takes for the transport of chloride ions to the reinforcing
steel, minimize cracking, use of corrosion inhibiting admixtures, and to set limits on internal chlorides in concrete
from the materials used to make it. In general, good quality concrete is resistant to carbonation and the emphasis in
industry standards is on chlorides.
The Building Code (ACI 318-14) defines exposure classes for corrosion protection of reinforcing steel and establishes
requirements for concrete used in reinforced concrete members assigned to these exposure classes. These requirements
are incorporated in specification format in ACI’s reference specification (ACI 301-10). For structural members
exposed to an external source of chlorides, the concrete mixture has to be developed with a low w/cm to reduce the
rate at which external chlorides migrate to the reinforcing steel. Typically these impact structural members in marine
construction or exposed to the application of deicing chemicals that are typically chloride-based.
For all structural members, there are limits on internal chlorides for the concrete in the Building Code (ACI 318-14).
The limits vary depending on the exposure class assigned to the member. A lower conservative limit is set for
prestressed concrete regardless of the exposure condition because of the dire consequences to structural safety if
prestressing strands corrode. The chloride limits are determined on the basis of tests performed on concrete at an age
between 28 and 42 days. The water-soluble chloride limits are stated on the basis of percent by mass of cement. With
the evolution of more sustainable concrete mixtures with higher quantities of pozzolans and slag, the chloride limits
stated on the basis of cement poses a problem for the industry as the limits are considerably lower for these mixtures.
While it is recognized that concrete materials are the source of internal chlorides, industry standards do not impose
chloride limits on the materials. One exception is an optional limit on mixing water if a non-potable source is used
(ASTM C1602).
It is recognized that some of the chlorides introduced from the materials are chemically bound by the cement hydration
products and will not be available in the concrete pore solution to initiate corrosion. Carbonation releases some of the
bound chlorides to make it available for corrosion (ACI 222R-01). If the chloride content of the concrete materials is
known and the total chloride content of concrete is calculated based on the mixture proportions, it can be assumed that
the water-soluble chlorides measured on the concrete will be somewhat lower than that calculated. Thereby, if the
calculated chloride content of concrete is less than the Code limits, it is reasonable to assume that the concrete will
comply with the water-soluble chloride limits when tested. Internal chlorides in concrete mixtures also have other
effects on the rate of corrosion (ACI 222R-01).
2.2
@Seismicisolation
@Seismicisolation
Confusion on Chloride Limits in Specifications that Challenge the Industry
NRMCA provides simplified guidance to concrete producers on chloride limits and much of the discussion in this
paper is based on that reference (NRMCA 2014, Gaynor 1983).
MEASUREMENT OF CHLORIDES
As indicated earlier, not all of the total chlorides in hardened concrete contribute to corrosion. Some of the total
chlorides in hardened concrete are chemically bound in hydration products and only a portion of the total chlorides,
perhaps 50 to 75%, continue to exist in the concrete pore solution and thereby available to initiate corrosion of
reinforcement (ACI 222R-01). For that reason, the Building Code (ACI 318-14) establishes limits on the basis of
measurements using the water-soluble test. The acid-soluble test measures the total chlorides. This method may be
used to measure the chlorides in solid ingredients used in concrete mixtures.
ASTM C1218 - Water-Soluble Chloride in Mortar and Concrete

For determining the water soluble chloride in concrete for new construction, the test involves making concrete
cylindrical specimens, curing them for at least 28 days and then obtaining powder samples from the concrete specimen
before an age of 42 days. Some duration of time is required to allow for adequate hydration of cementitious materials
that can encapsulate some of the chlorides in the hydration products. A composite powder sample should be prepared
from different portions of the specimen so that it is representative. The powder sample tested is about 10 g (0.02 lb)
and it is finer than the 850-µm (No. 20 sieve). The powder sample is boiled in water for 5 minutes and allowed to
stand for 24 hours. The water is then tested for dissolved chlorides and this is reported as a percent of the cement or
of concrete if the cement content is not known. Generally, the water-soluble chloride content is measured on concrete
and not on concrete materials.
ASTM C1152 – Acid-Soluble Chloride in Mortar and Concrete

This test involves digesting the same type of powder sample of hardened concrete in dilute nitric acid. The solution is
then tested for chloride content. This chloride content is reported as a percent of the weight of the material being
analyzed. This method can be used for concrete materials even though the method does not specifically state that.
ASTM C1524 – Water Extractable Chloride in Aggregate (Soxhlet Method)

This test was developed because some aggregates have a higher quantity of background chlorides when measured by
the acid soluble method on powder samples. Since these chlorides are bound within the aggregate minerals they are
not available in concrete to contribute to corrosion. The method extracts chlorides from non-pulverized material using
the Soxhlet apparatus. The method involves boiling water that condenses and drips on to the sample to saturate it and
extract the available water-soluble chlorides. The chloride content of the water is measured. The method has been
used to measure the chlorides in non-pulverized concrete samples with low reproducibility. This test was developed
primarily for dolomitic limestone found in southern Ontario, New York, northern Michigan and the Chicago area
(Whiting, 1997). The method is not intended for marine aggregates or those that contain absorbed chloride in the
aggregate pores.
SOURCES OF INTERNAL CHLORIDE
Internal chlorides in concrete originate from the materials used. There are no limits on chlorides for the ingredients
used in concrete. Mixing water is one exception; there is an optional limit for non-potable water if used. In most cases,
the chloride content of materials used to make concrete are not measured or reported by the supplier. Project
specifications often include a chloride limit on admixtures. Eventually it is the chloride limits in concrete that govern.
a. Cementitious Materials - Generally, the total chlorides in portland cement will be less than 50 to 100 ppm.
2.3
@Seismicisolation
@Seismicisolation
Colin L. Lobo
Chlorides in slag cement and fly ash should be similarly low. The chloride content of cementitious materials
is typically available from the supplier. It is being proposed that chloride content of portland cement be
reported on the mill test report, but with no specification limit. The chloride content of portland or blended
cement will typically be measured by X-ray fluorescence; a rapid method that is typically used to determine
the chemical composition of cement.
b. Aggregates - Total chloride content of aggregates will vary greatly depending on the aggregate source.
Marine aggregates, especially sand, that have not been properly washed can contribute a significant quantity
of water soluble chloride if used in concrete. Total chlorides in certain dolomitic aggregates from quarries in
Chicago and Ontario are as high as 0.1 to 0.15 percent by weight of aggregate. This results from the nature
of its formation in shallow oceanic environments. Aggregate suppliers from these sources generally provide
a statement indicating the successful service history and use of these sources. Most aggregates will have total
chlorides less than 0.025 percent by weight of the aggregate. Chlorides in aggregates tend to be bound within
the minerals and are not available as water soluble chlorides. However, when aggregates or concrete are
crushed to a fine powder for the purpose of testing, these bound chlorides will be measured. It may also be
more appropriate to measure the water-soluble chloride content of aggregates in their as-used form, rather
than crushing to powder and measuring the acid-soluble chloride content. This would better represent the
chloride contribution of aggregate as used in concrete mixtures.
c. Water - Drinking water standards typically set a maximum chloride ion limit of about 250 ppm. Chloride
content in city water is generally less than 50 ppm. The chloride content of other non-potable sources of
water should be measured. ASTM C1602 establishes an optional limit on chlorides in mixing water (ASTM
C1602-13). The limits are 500 ppm for water used in prestressed concrete and 1000 ppm for reinforced
concrete or concrete in contact with aluminum embedments or galvanized metal forms. At 1000 ppm, the
chloride content from mixing water in concrete amounts to about 0.05% by weight of cement. The chloride
content in wash water in concrete plants is generally much less than 1000 ppm, unless concrete containing
chlorides is washed out into wash water basins or reclaimers.
d. Admixtures – Use of calcium chloride-based set accelerating admixtures is the primary source of chlorides
from admixtures. Calcium chloride is a very effective accelerating admixture but the contribution of chlorides
requires its use to be limited in reinforced concrete and prohibited from use in prestressed concrete. The
chloride content in non-chloride water-reducing admixtures will generally be between 100 and 800 ppm.
Where calcium chloride is used to offset retardation in some water-reducing admixtures to make it set neutral
the admixture might contain as much as 200,000 ppm chloride ion. Since the quantity of admixtures used is
rather small, the chloride contribution in a concrete mixture is also small. At this concentration an admixture
dosage of 3 mL/kg (5 fl.oz./100 lb) of cement results in a total chloride ion concentration of approximately
0.06 percent by weight of cement.
Calculation of Total Chloride in Concrete

Table 1 illustrates the calculation of chloride content in concrete when the chloride content of the materials is known
(NRMCA 2014). The chloride contents of the materials are assumed and should not be considered as typical.
As indicated earlier, it can be assumed that the water-soluble chloride in concrete will be about 25 to 50% lower than
the calculated value of total chlorides because of being chemically bound in hydration products.
2.4
@Seismicisolation
@Seismicisolation
Table 1. Example Calculation of Total Chloride Content in Concrete

Ingredient Mixture Total Cl⁻ Calculation Total Cl⁻,
proportion*, lbs.*
lb/yd3
Cement 600 0.005% (0.005 ÷ 100) × 600 0.03
SSD Sand 1150 0.01% (0.01 ÷ 100) × 1150 0.115
SSD Coarse Agg 1800 0.106% (0.106 ÷ 100) × 1800 1.908
6
Water 280 250 ppm (250 ÷ 10 ) × 280 0.07
Admixture 5 oz* 800 ppm (800 ÷ 106) × 5 × 6 × (1/16) 0.0015
Total Cl⁻ in 1 cubic yard 2.1245
.
Total Chloride, % by weight of cement 100 = 0.354 %
*
1 lb/yd3 = 0.59 kg/m3; 1 lb=0.4536 kg; 1fl.oz = 29.6 mL
CHLORIDE LIMITS FOR CONCRETE
The Building Code sets requirements for chloride limits in concrete depending on the exposure to the environment.
Chloride limits for concrete were first established in the 1983 Building Code. Currently, ACI 318 defines durability
exposure classes (EC) that the design engineer has to assign to concrete structural members. Assigned exposure classes
for structural members establish requirements for concrete. These requirements are also reflected in ACI Specification
(ACI 301-10).
The chloride limits in ACI 318 and ACI 301 are provided in Table 2. ACI Guide to Durable Concrete (201.2R-08)
and corrosion document (ACI 222R-10) provide guidance (not requirements) on chloride limits in concrete. ACI
Committee 201 refers to the guidance of ACI Committee 222. The chloride limits suggested in ACI 222 are based on
research on the threshold chloride concentration to initiate corrosion and are more conservative than those stated in
ACI 318 and ACI 301. The reasoning for the more conservative recommended limits is discussed (ACI 222R-01).
The recommendations on chloride limits by ACI Committee 222 are provided in Table 3. If the limits of ACI
Committee 222 are invoked in a project specification, these will essentially prohibit the use of admixtures that contain
any significant amount of chlorides.
The Code for Environmental Structures uses the same limit as ACI 318 for prestressed concrete but a more
conservative single limit of 0.10 percent water soluble chloride by weight of cement for reinforced concrete. ACI
Committee 350 states that this more conservative limit is required for environmental engineering structures due to
their greater susceptibility to corrosion of metals, when experiencing prolonged exposures to chloride-saturated
solutions (ACI 350-06).
Note that chloride limits in ACI standards and guides are stated on the basis of percent by weight of portland cement
and not cementitious materials. Thereby, concrete mixtures that contain fly ash, slag cement, and silica fume are
penalized on this basis. Efforts continue to change the chloride limits to the basis of cementitious materials as they are
in the Canadian and European standards.
In the past, the Building Code permitted the calculation of chlorides from the concrete mixture proportions to
determine whether concrete met the chloride limits, as illustrated in Table 1. Currently ACI 318 requires the water
soluble chloride content in concrete mixtures be measured by ASTM C1218 at an age between 28 and 42 days. The
2.5
@Seismicisolation
@Seismicisolation
Colin L. Lobo
commentary (ACI 318R-14) discusses the option of calculating chloride, but the Code requirement for testing governs.
Calculating the chloride content can provide a concrete producer a conservative estimate of the water soluble chloride
that might be measured by testing and to decide whether changes to the materials and mixture proportions are needed
prior to a submittal. Considering the wide variety of concrete mixtures developed for various applications by ready
mixed concrete producers, if testing for chlorides on concrete is enforced, this represents a considerable expense. The
alternative to document calculated chloride contents would make this less onerous, especially when the chloride
contents are considerably lower than the specified or Code limits. A code change proposal to permit calculation of
chlorides, in lieu of testing, has been submitted for letter ballot to Committee 318.
Table 2: Chloride Limits in ACI 318 & ACI 301

Max water-soluble Cl⁻, % wt of cement
EC Description
Reinforced Prestressed
C0 Concrete dry or protected from moisture 1.00 0.06

Concrete exposed to moisture but not to external source of
C1 0.30 0.06
chlorides
Concrete exposed to moisture and an external source of
C2 chlorides from deicing chemicals, salt, brackish water, 0.15 0.06
seawater, or spray from these sources
Table 3. Chloride Limits for new construction in ACI 222R and ACI 201.2R
Chloride (Cl⁻) limit, % wt of cement
Category Acid-soluble Water-soluble
ASTM C1152 ASTM C1218
Prestressed concrete 0.08 0.06

Reinforced concrete in wet condition 0.10 0.08
Reinforced concrete in dry condition 0.20 0.15
INDUSTRY SURVEY
NRMCA conducted an informal survey of ready mixed industry companies on chloride limits and testing requirements
they see in project specifications (NRMCA 2015). Responses were received from about 50 small and large companies
supplying concrete in most regions of the US for different types of structures or applications. The questions are listed
below and responses are summarized. There were considerable differences in the requirements and enforcement of
chloride limits in specifications.
Do you have to do testing for water-soluble chlorides in concrete as required by specifications? How often do you do
these tests?
Responses varied from never to very rarely to several projects. Many producers perform tests at least bi-annually to
maintain documentation when requested. Requirements for test results on water-soluble chlorides in accordance with
ASTM C1218 are common for high-rise buildings in larger metropolitan areas, parking and environmental engineering
structures. Testing is required by some design firms involved in these types of projects. In smaller rural markets, the
requirement to test is not common.
2.6
@Seismicisolation
@Seismicisolation
Producers also reported a range of chloride limits in specifications. The limit for prestressed concrete (0.06%) is
sometimes invoked for cast-in-place reinforced concrete and for post-tensioned construction. While post-tensioned
concrete is considered prestressed concrete, the relevant question is whether the chloride limits should apply if the
concrete does not come in contact with the tendons. ACI 318 establishes limits for post-tensioning grout. Many limits
on measured water-soluble chlorides seen in specifications are not consistent with those stated in ACI 318 or based
on assigned exposure classes as in the Building Code. The ACI 350 limit of 0.10% is sometimes seen in specifications
for buildings. Some producers indicate that their tested (and calculated) values on chlorides are very low and
considerably less than the Code limits. Many engineers accept this as a basis to waive testing for specific projects. It
was also reported that many specifications only state limits on concrete materials, most commonly on chemical
admixtures. This is easy to document since this information is available from the admixture suppliers on all products.
Test reports on chloride tests often report results on the basis of percent by mass of cement, percent by mass of total
cementitious materials, and percent by mass of concrete.
Roughly what percentage of specifications requires this testing?

Responses varied by region. Chloride limits were seen in anywhere from 5 to 50% of the specifications, but the
requirement to test concrete is seldom enforced. As mentioned earlier, testing is required by some design firms and
for certain types of structures.
Do you do a calculation on the chloride ion content? Have you included calculated chloride in a submittal? Has
calculated chlorides been accepted by the engineer?
Many producers indicated that they do calculate the chlorides and maintain that documentation. In many cases they
compare the calculation to measured chlorides. Most producers who do this indicate that the values are considerably
less that limits in specifications. Some indicate that they do not submit calculated chlorides. Others indicate that when
calculated chlorides are submitted, some engineers accept this, while some design firms will require testing to be
performed. Calculated chlorides are often accepted when the project schedule does not permit time for obtaining a test
result. Some engineers will conditionally accept calculated chlorides to allow the start of a project but require testing
that will govern.
Do you have information on chloride content on materials you use – cementitious, aggregates, admixtures, etc?
Most producers who are impacted by chloride limits in specifications indicate they have information on chemical
admixtures and cementitious materials, but not many have this information on aggregates. Some indicate they measure
the chloride content of their concrete materials, including mixing water, so that they can calculate the chloride content
in concrete.
Have you had chlorides reported on the basis of cementitious materials accepted?
A few producers indicated that they had seen specifications that stated chlorides on the basis of total cementitious
materials. Some indicated that concrete mixture submittals reporting chlorides on the basis of cementitious materials
have been accepted. As indicated earlier, test reports often state results on different basis.
Many producers responding to the survey are not impacted by chloride limits in specifications and have never had to
perform tests on chloride content in concrete. Chloride limits are stated in many specifications but the requirement to
test for chlorides is most often not enforced. One commented that some engineers are unsure why it is in their
specification or understand its significance. The producers that calculate or test for chlorides indicate that the values
they obtain are significantly less than the specification or Code limits. Some mentioned overly conservative limits in
some specifications – such as the limit for pre-stressed concrete for reinforced concrete; in one case 0.03%; or the
ACI 350 limit of 0.10% for commercial building structures. There were no responses indicating that specifications
required the acid-soluble test on concrete or limits in Table 3 as recommended in ACI 222R.
2.7
@Seismicisolation
@Seismicisolation
Colin L. Lobo
Producers who are impacted by chloride limits suggested that the Code should permit the use of calculated chloride
contents and that the limits be revised to be stated on the basis of total cementitious materials.
CONCLUSIONS
Corrosion of reinforcing steel is a significant reason for deterioration of concrete structures. This is primarily impacted
by external sources of chlorides that can be addressed by the use of good quality concrete, adequate cover, and the use
of corrosion inhibiting admixtures.
Regarding internal chlorides, ACI standards and guides have varying requirements and recommendations on chloride
limits for concrete. This can likely cause some confusion to the practicing design professional. Considering the large
number of concrete mixtures developed for projects by concrete producers and the potential for different specification
requirements on chloride limits, the cost for testing can be significant. However, the specification and enforcement of
chloride limits is not extensive and generally limited to some types of structures and by some design firms.
It would be beneficial if internal chloride limits for concrete are based on clearly established threshold concentration
levels of chlorides that result in the initiation of corrosion. It is recognized that establishing threshold chloride
concentration for wide range of materials and exposure conditions is difficult. Recognizing that calculated chlorides
is a conservative estimate of the water-soluble chlorides for the initiation of corrosion, this option should be permitted.
The process of measuring the chlorides on concrete materials (test methods) and determining the calculated chloride
content should be clearly defined in an ACI document. Lastly, ACI standards should be revised to state chloride limits
on the basis of total cementitious materials as most of the concrete mixtures in use include supplementary cementitious
materials that are essential for superior performance and durability. It is recognized that these revisions can result in
different numerical values for the limits.
REFERENCES
1. ACI Committee 201, “Guide to Durable Concrete,” ACI 201R-08, American Concrete Institute, Farmington
Hills, MI, 2010, 49 pp.
2. ACI Committee 222, “Protection of Metals in Concrete Against Corrosion,” ACI 222R-01, American
Concrete Institute, Farmington Hills, MI, 2010, 41 pp.
3. ACI Committee 301, “Specifications for Structural Concrete,” ACI 301-10, American Concrete Institute,
Farmington Hills, MI, 2010, 77 pp.
4. ACI Committee 318, “Building Code Requirements for Structural Concrete,” ACI 318-14, “and
Commentary”, ACI 318R-14, American Concrete Institute, 2014, 519 pp.
5. ACI Committee 350, “Code Requirements for Environmental Engineering Structures and Commentary,”
ACI 350-06, American Concrete Institute, 2014, 485 pp.
6. ASTM C1152, C1218, C1524, C1602, Annual Book of ASTM Standards, Volume 04.02, ASTM
International, West Conshohocken, PA, 2014.
7. Gaynor R.D., “Understanding Chloride Percentages,” NRMCA Publication 173, National Ready Mixed
Concrete Association, Silver Spring, MD, 1983, 5 pp.
8. NRMCA, “TIP 13 – Chloride Limits in Concrete,” Technology in Practice Series, National Ready Mixed
Concrete Association, Silver Spring, MD, 2014, 6 pp.
9. NRMCA, Informal industry survey – personal communication with Colin Lobo, National Ready Mixed
Concrete Association, Silver Spring, Maryland, 2015.
10. Whiting, D.A., “Origin of Chloride Limits for Reinforced Concrete,” PCA R&D Serial No. 2153, Portland
Cement Association, Skokie, IL, 1997, 16 pp.
2.8
@Seismicisolation
@Seismicisolation
SP-308—3
EFFECT OF TEMPERATURE ON THE CHLORIDE BINDING OF PORTLAND CEMENT EXPOSED

TO CaCl2
By Jianqiang Wei, Bernard Tao and W. Jason Weiss
Synopsis: When concrete is exposed to calcium chloride (CaCl2) deicing salts the concrete may deteriorate or the
reinforcing steel may corrode due to the transport of ionic species through it’s pore structure. As the chloride ions
enter the concrete pore structure the ions will either react or be bound by hydration products or will remain free in
the pore solution. This work examines the potential reaction that may occur between the CaCl2 and the calcium
hydroxide (Ca(OH) 2) from the paste resulting in the formation of calcium oxychloride. To better understand the role
of calcium oxychloride on binding this study investigates the effect of temperature when CaCl2 solutions are present
in Type I and Type V cements at -5, 10, 23 and 40 ˚C (23, 50, 73.4 and 104 ºF). The results indicate that the chloride
binding due to calcium oxychloride increases with decreasing temperature. A two part Freundlich isotherm was
developed to fit the chloride binding behavior of hydrating cement paste with the first portion considering the
common forms of binding (absorption, Friedel’s and Kuzel’s salts) while the second portion considering binding due
to calcium oxychloride formation.
Keywords: binding, calcium oxychloride, chloride, chloride binding isotherm, CaCl2, diffusion, Freundlich, ingress,
temperature
3.1
@Seismicisolation
@Seismicisolation
Wei et al.
Jianqiang Wei is a postdoctoral research associate in the Lyles School of Civil Engineering at Purdue University
from October 2014 to present. He received his Ph.D. from the Department of Civil Engineering and Engineering
Mechanics at Columbia University in 2014. His research interests include chemistry of cement, fiber-reinforced
cementitious materials, natural materials for construction, durability and microstructure analysis.
Bernard Tao is a Professor in Agricultural and Biological Engineering at Purdue University. He recently was
elected Soybean Board Professor in Soybean Utilization. He has served as the chair of Biochemical, Pharmaceutical
and Foods Division of the American Institute of Chemical Engineers, an executive board member of the ACS
carbohydrate division. His research interests include biological engineering, bioenergy, and application of soy based
sealer in construction.
FACI Jason Weiss is the Edwards Distinguished Professor of Engineering and the head of the School of Civil and
Construction Engineering at Oregon State University.
INTRODUCTION
It is understood that solutions containing chloride based salts can be transported through the pore structure of
concrete. Numerous publications describe this transport. The chloride ions can either move freely in the pore space
or may be bound as a part of the hydrated cement. While the vast majority of reported studies focus on chloride
binding isotherms obtained at room temperature; temperature can influence the chloride binding capacity of the
hydrated cement paste.
Some studies have investigated the influence of temperature on chloride binding. Roberts1 reported that the amount
of bound chloride decreased with elevated temperature. Dousti and Shekarchi2 observed that the highest binding
capacity was observed at room temperature in studying samples between -4 ˚C and 70 ˚C (23 to 104 ˚F). Panesar and
Chidiac3 obtained a greater chloride binding for cold temperature which they reported as due to the decreased
thermal vibration of chloride ions and the formation of ice in the chloride solution. Hussain4 and Maslehuddin5, 6
reported a reduction of chloride binding in cement as temperature increased and this is in agreement with the results
reported by Larsson7. This increased binding was attributed to the decreased solubility of Friedel’s salt and calcium
chloro-aluminate phase, which results in more free chlorides at equilibrium1, 8. Zibara9 found that a decreased
temperature caused a greater chloride binding at low chloride concentration; while a reverse was observed at high
chloride concentration (3.0 M). Enhanced chloride binding was observed by Yuan10 as temperature increases from 5
to 40 ˚C (41 to 104 ºF) in the presence of NaCl. It can be noticed that although the trends in reports are not
consistent, most of the results indicate a decreased chloride binding of cement at an elevated ambient temperature
and there appears to be a dependence on the concentration of the chloride as well. More importantly, except Roberts
1
, all the results were obtained based on sodium chloride (NaCl) and salts like calcium chloride (CaCl2) and
magnesium chloride (MgCl2) have been studied much less frequently.
It is well known that chloride ions can react with both unhydrated and hydrated aluminate phases to form Friedel's
salt and Kuzel’s salt11-13. However, a recent investigation shows that a calcium oxychloride (3CaO·CaCl2·15H2O)
phase can be formed through the reaction between calcium chloride (CaCl2) and calcium hydroxide (Ca(OH)2) in the
cementitious matrix14, 15 for a higher concentration of CaCl2·and/or MgCl2. Monosi and Collepardi16 reported the
presence of oxychloride identified by X-ray diffraction of the wet hardened cement paste samples that have been
immersed in 30 wt.% CaCl2 solution under 5 ºC (41 ºF) for 2 months. In addition, they also found that this paste
disappeared if the samples were ground, washed with methyl alcohol and dried at a low relative humidity (≤1%).
This formation and unstable nature of oxychloride was observed by Shi17, Birnin-Yauri and Glasser18 based on X-
ray diffraction analysis. Petrographic evidence of oxychloride formation was presented by Sutter et al19 based on
optical microscopy, scanning electron microscopy and microanalysis. They reported that these oxychloride phases
form through consumption of portlandite and precipitation in voids in cracks, where calcium oxychloride
pseudomorphs can be observed after portlandite crystals. Recently, the rate of formation for calcium oxychloride in
cement in the presence of CaCl2 deicing salt was determined by Farnam and Weiss15 using Low Temperature
Differential Scanning Calorimetry and Isothermal Micro-Calorimetry. Equation (1) describes the reaction that can
result between CaCl2 and Ca(OH)2 to form calcium oxychloride:
3.2
@Seismicisolation
@Seismicisolation
Effect of Temperature on the Chloride Binding of Portland Cement Exposed to CaCl2
3Ca(OH)2 + CaCl2 + 12H2O → 3CaO·CaCl2·15H2O (1)
Figure 1 illustrates a phase isopleth that describes the temperature concentration dependence that is exhibited when
CaCl2 is added to a solution containing calcium hydroxide (Ca(OH)2) (like that exhibited in a cement paste system)
to form calcium-oxychloride.
60 140
120
C*)
40 (C 0/
8 ln 100
Ca(OH)2 + solution 27.8
T=
80
Temperature (C)
Temperature (F)
20
Oxychloride + solution 60
Rm = 3 40
0
20
-20 0
R
m =1
Ice + Oxychloride + solution -20
-40 -40
-60
Ice + Oxychloride + CaCl2  6H2O
-60
0 5 10 15 20 25 30
CaCl2 concentration (wt. %)

Figure 1 — Phase diagram for Ca(OH)2-CaCl2-H2O system15, 20 (Rm = Ca(OH)2/CaCl2 molar ratios)
Figure 1 describes the behavior of the ternary Ca(OH)2-CaCl2-H2O system. Farnam and Weiss15 approximated the
phase boundary at which calcium oxychloride forms as Equation (2):
T = 27.88 ln (C0 / C*) (2)
where T (°C) is sample temperature, C0 is the initial CaCl2 concentration associated with the formation of calcium
oxychloride, and C* (=4.97 wt. %) is the intersection between Ca(OH)2 and calcium oxychloride phase boundary
(the liquidus line) and the phase boundary for ice formation, so it demonstrates the theoretical minimum CaCl2
concentration solution at which the calcium oxychloride forms. From this model, it can be seen that the formation of
calcium-oxychloride and the corresponding chloride binding should have a strong temperature dependence.
In addition to being influenced by temperature, carbonation is another factor that should be taken into consideration
for chloride binding. By exposing a cementitious matrix to gaseous CO2, the calcium hydroxide and calcium silicate
hydrates (C-S-H) may carbonate, however the calcium hydroxide will preferably react especially at low
concentration. The carbonation process can be described as a chemical reaction of calcium hydroxide with carbon
dioxide resulting in the formation of calcium carbonate and water21 as follow:
Ca(OH)2 + H2CO3 → CaCO3 + 2H2O (3)
Kayyafi el at. 22, 23 investigated accelerated carbonation on chloride binding of cement paste and observed a decrease
in chloride binding capacity. Zibara9 investigated the effect of carbonation on chloride binding capacity of cement
paste in the presence of external sodium chloride. The reduction of calcium hydroxide though accelerated
carbonation has been proven as an effective way to mitigate the formation of calcium oxychloride24, 25.
3.3
@Seismicisolation
@Seismicisolation
Wei et al.
RESEARCH SIGNIFICANCE
Chloride binding plays an important role in the transport of chloride ions in concrete. While binding has been
studied for many decades, this paper suggests that the potential reaction between the hydrated cement matrix and
CaCl2 deicing salt solutions may require further study. Specifically, the CaCl2-H2O-Ca(OH)2 phase isopleth suggests
the formation of calcium oxychloride at high CaCl2 concentration which can substantially influence the chloride
binding. This work examines the effect of temperature on chloride binding capacity of hydrated portland cement
paste in the presence of CaCl2 deicer. Two types of cement (type I and V) and four temperatures, -5 ºC, 10 ºC, 23 ºC
and 40 ºC (23, 50, 73.4 and 104 ºF), were used. The influence of carbonation was also studied. An equation was
developed to describe chloride-binding isotherms that accounts for calcium oxychloride.
EXPERIMENTAL PROCEDURE
Materials used in making cement pastes

Type I and Type V ordinary portland cements were used in this study with a reported Blaine fineness of 375 m2/kg
(1830 ft2/lb) and 316 m2/kg (1542 ft2/lb), respectively. The chemical compositions of the two types of cement are
summarized in Table 1. Type V cement contains less C3A than Type I. As a result, it would be expected to form less
Friedel and Kuzel salt (however this may also be impacted by the increase in C4AF).
Table 1 — Chemical and mineralogical composition of Type I and Type V cements.

Chemical and phase composition
Chemical composition in equivalent oxide [wt. %] Phase composition of cement [wt. %]
Oxide Type I Type V Type I Type V
CaO 63.5 63.2 a
SiO2 19.4 21.3
C3S 60 64
Fe2O3 3.18 4.20
Al2O3 5.39 2.60
C2Sb 10 13
SO3 3.38 2.80
MgO 2.97 4.50
C3Ac 9 0
K2O 0.77 -
Na2O 0.35 -
C4AFd 10 13
Free lime - 0.38 Readily soluble equivalent alkalis as Na2O [wt%]
Loss on ignition 0.88 1.20
0.86 0.21
Insoluble residue 0.25 0.18
a
Tricalcium silicate; b Dicalcium silicate; c Tricalcium aluminate; d Tetra-calcium Aluminoferrite
Preparation of specimens
A water to cement ratio (w/c) of 0.42 was used throughout this study. The cement pastes were mixed following a
procedure similar to ASTM C305-14; however a vacuum mixer was used a speed of 400 rpm15, 26. After mixing,
samples were cast in 76.2 cm (diameter) x 152.4 cm (height) (3 x 6 in) cylindrical molds. The samples were sealed
and cured at 23+/-2 °C for over 3 months. The well hydrated cement pastes were then ground layer by layer from the
middle in 2 mm (0.079 in) lifts using a milling machine30. The cement powder was then collected and passed
through a 45 μm (0.0018 in) sieve.
Materials used in making CaCl2 solutions

Granular reagent calcium chloride dehydrate was used. De-ionized (DI) water was used as the solvent.
Accelerated carbonation
A carbon dioxide (CO2) chamber was employed to accelerate the carbonation process at room temperature. A
relative humidity of 55% was used.22, 27, 28 The relative humidity was maintained using saturated sodium bromide
solution. A CO2 concentration of 1.5% was used. In order to fully carbonate the hydrated cement, powders were
placed in the CO2 chambers for 10 days until the mass of samples reach equilibrium. This was shown to result in
nearly complete calcium hydroxide consumption (and/or encapsulation).22
3.4
@Seismicisolation
@Seismicisolation
Equilibrium method for chloride binding

Before mixing with the CaCl2 solutions, both the plain and carbonated cement samples were conditioned to a
temperature of 23 ± 1 °C (73.4 ± 1.8 ºF) and a relative humidity of 11% for 2 weeks in a sealed desiccator
containing saturated lithium chloride solution. Following the procedure developed by Tang and Nilsson28, after the
mass of samples reached equilibrium, both the cement powder and solutions were separately sealed using plastic
bottles and conditioned at the four temperatures for three days. Then 4 g of cement powder was uniformly mixed
and immersed in 10 ml CaCl2 solutions with different concentrations as shown in Figure 2. This cement powder to
solution ratio was selected according to the recommendation reported by Delagrave et al29, who mixed
approximately 20 g powder with 50 ml solution in their study. Thirteen CaCl2 solutions were prepared at different
CaCl2 concentrations (0.5, 1.5, 2.5, 3.5, 5, 7.5, 10, 12.5, 15, 17.5, 20, 25, and 30 wt. %) for all the four temperatures.
60 140
120
40 100
Ca(OH)2 + solution
80
Temperature (C)
Temperature (F)
20
Oxychloride + solution 60
Rm = 3 40
0
20
-20 0
R
m =1
Ice + Oxychloride + solution -20
-40 -40
-60
Ice + Oxychloride + CaCl2  6H2O
-60
0 5 10 15 20 25 30
CaCl2 concentration (wt. %)

Figure 2 — CaCl2 solution design for chloride binding under different temperatures
The critical chloride concentration for the formation of calcium oxychloride is a function of temperature. Equation
2 can be rearranged to calculate this concentration as shown in equation 4:
.
∙ % exp (4)
.
where CI-Oxy % is the critical chloride concentration at which calcium oxychloride begins to form and T is the
temperature in ºC. This equation describes the development of phase boundary between Ca(OH)2 and calcium
oxychloride (the red liquidus line in Figure 2) in a calcium chloride range from 5 wt. % to 35 wt. %.
According to Eq. (4), additional CaCl2 solutions with concentrations in close proximity to the phase boundary were
prepared as shown in Figure 2, such as 6.5 wt. % and 8.5 wt. % for 10 ºC (50 ºF), 21.25 wt. % and 22.5 wt. % for 40
ºC (104 ºF), and 11.5 wt. % for 23 ºC (73.4 ºF).
The chloride binding samples (cement powder + solutions) were placed in vials, which were sealed to minimize the
potential for the evaporation of water. The vials were placed in four chambers with temperatures of -5 ± 1 ºC, 10 ± 1
ºC, 23 ± 1 ºC, and 40 ±1 ºC (23, 50, 73.4 and 104 ºF) for 2 months to ensure that it is sufficient to reach the
equilibrium between the solid and the solution. This time was confirmed by performing additional measurements on
selected samples at all the four temperatures. After the reaction between the solutions and powders reached steady
state, the exact chloride concentrations of the initial solutions and the equilibrium free chloride concentrations of the
extracted solutions were determined by using an automated titration unit through potentiometric titrations with a
0.0102 M AgNO3 electrode. The silver nitrate was added in 0.2 ml increments (this addition rate however decreases
while approaching the saturation point), while simultaneously the electric potential of the solution was monitored
with a silver reference electrode. The chloride content of the solution is determined using the dosage of silver nitrate
3.5
@Seismicisolation
@Seismicisolation
Wei et al.
once the maxima of the first derivative curve of the potential reached30. The amount of bound chlorides (Cb) (mg / g
sample, 0.001 lb / lb sample) was then calculated by using equation 531:
.
(5)
where V is the volume of CaCl2 solution (10 ml) (0.34 fl oz), C0 is the initial concentration of chloride solution (mol
/ L, 0.0296 mol / fl oz), C1 is the equilibrium concentration of CaCl2 solution (mol / L, 0.0296 mol / fl oz), and W is
the mass of cement powder (4 g) (0.0088 lb).
RESULTS AND DISCUSSION
Effect of temperature on chloride binding

Figure 3 shows the chloride binding isotherms of Type I and Type V hydrated cement paste in the presence of CaCl2
solutions at -5, 10, 23 and 40 ºC (23, 50, 73.4 and 104 ºF). Each data point represents the average of two
measurements and the corresponding coefficients of variation are less than 6.0%. It can be seen that for both the
Type I and Type V cement pastes, the amount of bound chloride increases significantly with increased free chloride
at low concentrations (≤5 wt. %) regardless of temperature. This has been conventionally attributed to physical
chloride binding by the cement hydration products, such as C-S-H, C-A-H, C-A-S-H and ettringite, and chemical
binding between chloride ions and monosulfate and aluminate phases. Due to the lower C3A content in the samples
made using Type V cement, less chloride was bound in the Type V cement paste than that bound in the Type I
cement paste under each temperature. Although the chloride binding behavior of cements may show a slight
temperature dependence at low concentrations, this temperature dependence is relatively small compared with that at
high CaCl2 concentrations (and the influence of temperature at low concentrations is not considered in this study).
An inflection point was observed in the chloride binding isotherms for both the Type I and Type V hydrated cement
pastes at each temperature. The inflection points are located approximately at CaCl2 concentrations of 5 wt. %, 7.5
wt. %, 11.5 wt. % and 20 wt. %, for -5, 10, 23 and 40 ºC (23, 50, 73.4 and 104 ºF), respectively. Beyond these
inflection points, the amount of bound chloride increases more substantially. According to the liquidous line
showing the boundary of phase formation in Figures 1 and 2, this inflection point is attributed to the formation of
calcium oxychloride, which binds chloride ions. From Figure 3, it can be seen that as the temperature that the
sample was stored at was decreased, the CaCl2 concentration at which the inflection point occurred decreased and
more chloride was bound.
150 0.15 150 0.15

140 Type I 0.14 140 0.14
Type V
130 0.13 130 0.13
Bound chloride (lb Cl / lb sample)

Bound chloride (mg Cl / g sample)
120 -5 C (23 F) 0.12 120 0.12

-5 C (23 F)
110 10 C (50 F) 0.11 110 10 C (50 F) 0.11
23 C (73.4 F)
100 0.10 100 23 C (73.4 F) 0.10
40 C (104 F)
90 0.09 90 40 C (104 F) 0.09
80 0.08 80 0.08
70 0.07 70 0.07
60 0.06 60 0.06
50 0.05 50 0.05
40 0.04 40 0.04
30 0.03 30 0.03
20 0.02 20 0.02
10 0.01 10 0.01
0 0.00 0 0.00
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
CaCl2 Concentration (wt. %) CaCl2 Concentration (wt. %)

Figure 3 — Chloride binding isotherms of Type I and Type V cement under different temperatures.
3.6
@Seismicisolation
@Seismicisolation
Chloride-binding isotherms for low CaCl2 concentrations

Figure 4 shows the chloride binding isotherms for Type I and V cement paste at –5 and 40 ºC (23 and 104 ºF) using
a single curve which is typical of the approach commonly used by many currently. (Note a single Freundlich curve
is commonly used to describe the entire range of CaCl2 concentrations). It can be seen that, simply fitting the data by
using a single isotherm may fail to capture specific features of the curve. For example, the single isotherm does not
appear to fit the data well at lower CaCl2 concentrations (e.g., 3 to 10% for the 5 ºC sample and from 3 to 20% for
the 40 ºC sample).
150 0.15 150 0.15

140 0.14 140 0.14
Freundlich isotherms Freundlich isotherms Type V

130 0.13

130 0.13

Type I
120 0.12 120 -5 C (23 F) 0.12

110 -5 C (23 F) 0.11 110 40 C (104 F) 0.11
100 40 C (104 F) 0.10 100 0.10
90 0.09 90 0.09
80 0.08 80 0.08
70 0.07 70 0.07
60 0.06 60 0.06
50 0.05 50 0.05
40 0.04 40 0.04
30 0.03 30 0.03
20 0.02 20 0.02
10 0.01 10 0.01
0 0.00 0 0.00
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
CaCl2 Concentration (wt. %) CaCl2 Concentration(wt. %)

Figure 4 — Oversimple Freundlich isotherms for Type I and V cement under -5 and 40 ºC.
Several mathematical models have been employed in the literature to fit the chloride-binding isotherms, such as
linear isotherm32, Langmuir non-linear isotherm33, Freundlich isotherms29, and the combination of Langmuir (low
concentrations) and Freundlich (high concentrations) isotherms31. Equation (6) (the Freundlich isotherm) was used
in this study due to its ability to best reflect the trends in data. Equation (6) describes the chloride binding:
(6)
where α and β are constants, Cf is the initial free chloride concentration (wt %), and Cb is bound chloride (mg
chloride / g sample).
The relationship between the bound chloride (Cb) and free chloride (Cf) for concentrations lower than that associated
with the than inflection points for all the four temperatures was fitted by using a single isotherm which was assumed
to be independent of temperature (note strictly speaking a temperature dependence likely exists however a single
curve appeared to fit the data reasonably well). Figure 5 shows the isotherms obtained by using the Freundlich
fitting curves. The isotherm regression parameters for each isotherm are summarized in Table 2. The Freundlich
isotherms fit the data well for Type I and Type V, respectively.
Chloride-binding isotherms of cement in the presence of CaCl2

When the concentration of CaCl2 increases to a concentration greater than the inflection point that is associated with
the formation of calcium hydroxide a considerable increase in the amount of bound chloride can occur. In order to
fit the chloride binding to the data more accurately a two-part Freundlich curve was used. At low concentration a
term like that shown in Equation (6) was used to account for the bound chloride physical binding and the formation
of Friedel’s and Kuzel’s salts. At higher concentration the bound chloride, due to the formation of calcium
oxychloride, was also taken into account. A two-part Freundlich isotherm is described in Equation (7a and 7b):
when (7a)
when (7b)
3.7
@Seismicisolation
@Seismicisolation
Wei et al.
where Cb is the total amount of bound chloride, α and β (as shown in Table 2) are the regression parameters for low
CaCl2 concentration, and αoxy and βoxy describe the chloride binding due to calcium oxychloride formation. Using Eq.
(4), the concentrations of CaCl2 (CI-Oxy ) associated with the formation of calcium oxychloride could be calculated for
the four temperatures and these results are summarized in Table 3. Figure 6 shows the chloride binding isotherms of
Type I and Type V cement for the four temperatures fitted by the two-part Freundlich curves. Using the regression
parameters shown in Table 3, it can be seen that the two part Freundlich isotherms fit the data well.
50 0.050 50 0.050
Type I Type V
-5 C (23 F) 0.045 -5 C (23 F) 0.045

Bound chloride (mg CL / g sample)
10 C (50 F)

40 10 C (50 F) 0.040 40 0.040
23 C (73.4 F) 23 C (73.4 F)
40 C (104 F) 0.035 40 C (104 F) 0.035
30 0.030 30 0.030
0.025 0.025
20 0.020 20 0.020
0.015 0.015
10 0.010 10 0.010
0.005 0.005
0 0.000 0 0.000
0 5 10 15 20 25 0 5 10 15 20 25

Figure 5 — Mathematical models to fit chloride binding isotherms of Type I and V cements for low concentrations
that no calcium oxychloride forms.
Table 2 — Freundlich isotherm regression parameters for Type I and V cements at low salt concentrations.
Cement α β R2
Type I 9.30 0.42 0.89
Type V 5.32 0.56 0.96
Table 3 — Parameters of the Freundlich isotherm for chloride binding of Type I and Type V cement due to calcium
oxychloride formation.
CI-Oxy Freundlich
Cement Temperature (ºC) (wt. %) αOxy βOxy R2
-5 (23 ºF) 4.15 16.75 0.54 0.98
10 (50 ºF) 7.11 5.27 0.82 0.97
Type I
23 (73.4 ºF) 11.33 5.91 0.49 0.99
40 (104 ºF) 20.85 10.96 0.20 0.99
-5 (23 ºF) 4.15 23.78 0.35 0.94
10 (50 ºF) 7.11 8.99 0.51 0.98
Type V
23 (73.4 ºF) 11.33 3.03 0.62 0.99
40 (104 ºF) 20.85 11.56 0.25 0.97
3.8
@Seismicisolation
@Seismicisolation
150 0.15 150 0.15

140 0.14 140 0.14
Type I Type V
130 0.13 130 0.13


120 -5 C (23 F) 0.12 120 -5 C (23 F) 0.12
110 10 C (50 F) 0.11 110 10 C (50 F) 0.11
100 23 C (73.4 F) 0.10 100 23 C (73.4 F) 0.10
90 40 C (104 F) 0.09 40 C (104 F)
90 0.09
80 0.08 80 0.08
70 0.07 70 0.07
60 0.06 60 0.06
50 0.05 50 0.05
40 0.04 40 0.04
30 0.03 30 0.03
20 0.02 20 0.02
10 0.01 10 0.01
0 0.00 0 0.00
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35

Figure 6 — Chloride binding isotherms for Type I and Type V cement by using piecewise Freundlich isotherms.
Contribution of calcium oxychloride

A substantial increase in the chloride binding capacity of the hydrated cement paste at high CaCl2 concentrations is
attributed to the formation of calcium oxychloride. In this study, as shown in Figure 7, the amount of bound chloride
due to the formation of calcium oxychloride under each temperature was simply determined using the second part of
equation (7), which indicates the differences between Freundlich isotherms at high CaCl2 concentrations and the
extending line of the Freundlich isotherm at low concentrations.
150 0.15
140 0.14
130 -5 C (23 F) 0.13

120 0.12
110 0.11
100 0.10
90 0.09
80 0.08
70 0.07
60 0.06
50 0.05
40 0.04
30 0.03
20 0.02
10 0.01
0 0.00
0 5 10 15 20 25 30 35
CaCl2 Concentration (wt. %)

Figure 7 — Difference between isotherms due to formation of calcium oxychloride.
Figure 8 shows only the contribution of the binding curve associated with calcium oxychloride formation (i.e., the
second portion of Equation (7b)). It can be seen that, the amount of calcium oxychloride decreases with an
increasing temperature, and consequently less chloride was bound. It can be noticed (when comparing Figures 8a
and 8b) that less chloride was bound by Type V cement than Type I cement at high CaCl2 concentrations. This can
be attributed to a reduction in the formation of calcium hydroxide in Type V cement pastes which results in less
calcium oxychloride formation.
3.9
@Seismicisolation
@Seismicisolation
Wei et al.
110 0.11 110 0.11
Type V
Bound chloride due to Ca-Oxychloride

100 Type I 0.10 100 0.10

90 0.09 90 0.09
-5 C (23 F) -5 C (23 F)
80 10 C (50 F) 0.08 80 10 C (50 F) 0.08
(lb Cl / lb sample)
(mg Cl / g sample)
(73.4 F)
(lb Cl / lb sample)
23 C
(mg Cl / g sample)
70 0.07 70 23 C (73.4 F) 0.07

40 C (104 F) 40 C (104 F)
60 0.06 60 0.06
50 0.05 50 0.05
40 0.04 40 0.04
30 0.03 30 0.03
20 0.02 20 0.02
10 0.01 10 0.01
0 0.00 0 0.00
5 10 15 20 25 30 5 10 15 20 25 30
CaCl2 concentration (wt. %) CaCl2 concentration (wt. %)

(a) (b)
Figure 8 — Bound chloride due the formation of calcium oxychloride for Type I (a) and Type V (b) cement.
Influence of carbonation on chloride binding

Figure 9 shows the chloride binding behavior of hydrated pastes made using Type I and Type V cements before and
after carbonation. The carbonation of the hydrated cement paste reduces the amount of accessible calcium
hydroxide9, 24. As a result, carbonation can reduce the chloride binding capacity at both low and high CaCl2
concentrations resulting in less Friedel’s salt formation and less calcium oxychloride formation. From Figure 10, it
can be seen that at 23 ºC (73.4 ºF) the amount of chloride binding decreases. In addition, Figure 10 shows that the
inflection point on the chloride binding isotherm associated with calcium oxychloride formation disappears. This
reduction in calcium oxychloride is likely attributed to the inability to form calcium oxychloride due to the removal
(or encapsulation) of calcium hydroxide after carbonation25.
70 0.070 70 0.070
0.065 0.065
23 C (73.4 F) 23 C (73.4 F)
60 0.060

60 0.060
Bound chloride (mg CL / g sample)
Type I
0.055 Type V 0.055

Carbonated Type I Carbonated Type V
50 0.050 50 0.050
0.045 0.045
40 0.040 40 0.040
0.035 0.035
30 0.030 30 0.030
0.025 0.025
20 0.020 20 0.020
0.015 0.015
10 0.010 10 0.010
0.005 0.005
0 0.000 0 0.000
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
(a) (b)
Figure 9 — Chloride binding behavior of Type I (a) and Type V (b) cement with and without carbonation.
CONCLUSIONS
This paper examined the influence of temperature on the chloride binding behavior of hydrated cement pastes made
using Type I and Type V cement. Specifically, this work examined the binding that occurs between the ionic species
in CaCl2 salt solutions and the hydrated cement paste at varying temperatures. The results indicate that the chloride
binding capacity at high CaCl2 concentrations decreases with the increasing temperature. This temperature
3.10
@Seismicisolation
@Seismicisolation
dependency is proposed to be due to the formation of calcium oxychloride which occurs specifically at high CaCl2
concentrations. It was shown that the concentration at which calcium oxychloride forms was likely to be changed
with sample/solution temperature which was consistent with the phase isopleth for the CaCl2-H2O-Ca(OH)2 system.
The formation of calcium oxychloride resulted in an inflection point (that was linked to the liquidus curve on the
phase isopleth) in the chloride binding isotherm. This demonstrates that more chloride ions were bound by the
hydrating cement paste, and the amount of free chloride in the pore solutions decrease. As a result a two part
Freundlich isotherm was developed to fit the chloride binding behavior of both Type I and Type V cements at both
low and high salt concentrations. As expected the carbonation of the cement paste resulted in significant reductions
in their chloride binding capacity. Carbonation resulted in a significant reduction (or encapsulation) of calcium
hydroxide, which is responsible for chloride binding (through the formation of calcium oxychloride) at high CaCl2
concentrations.
ACKNOWLEDGMENTS
The authors would like to acknowledge the financial support of the Indiana Soybean Alliance through the Grant
titled “Testing to Quantify the Performance of Soy Based Sealer to Reduce Fluid Ingress and Reduce Deicing Salt
Damage” which was awarded to Purdue University to the second and third author. The work was performed in the
Pankow Materials Laboratory and the authors gratefully acknowledge the support which made this possible.
REFERENCES
1. Roberts MH, (1962), "Effect of calcium chloride on the durability of pre-tensioned wire in prestressed
concrete," Magazine of Concrete Research, V. 14, No. 42, pp. 143-54.
2. Dousti A, Shekarchi M, (2015), "Effect of exposure temperature on chloride-binding capacity of cementing
materials," Magazine of Concrete Research, V. 67, No. 15, pp. 821-32.
3. Panesar DK, Chidiac SE, (2011), "Effect of Cold Temperature on the Chloride-Binding Capacity of Cement,"
Journal of Cold Regions Engineering, V. 25, No. 4, pp. 133-44.
4. Hussain SE, Rasheeduzzafar, (1993), "Effect of temperature on pore solution composition in plain cements,"
Cement and Concrete Research, V. 23, No. 6, pp. 1357-68.
5. Maslehuddin M, Paget CL, Rasheeduzzafar, (1997), "Temperature effect on the pore solution chemistry in
contaminated cements," Magazine of Concrete Research, V. 49, No. 178, pp. 5-14.
6. Maslehuddin M, Rasheeduzzafar, Pace CL, Al-Mana I, A-Tayyib AJ, (1993), “Effect of temperature and sulfate
contamination on the chloride-binding capacity of cements,” 4th International Conference on Deterioration and
Repair of Concrete in the Arabian Gulf, Bahrain Society of Engineers: Manama, pp. 735-50.
7. Larsson J, (1995), "The enrichment of chlorides in expressed concrete pore solution submerged in saline
solution," Proceedings of the Nordic seminar on field studies of chloride initiated reinforcement corrosion in
concrete, Lund University of Technology, pp. 171-6.
8. Yuan Q, Shi C, De Schutter G, Audenaert K, Deng D, (2009), "Chloride binding of cement-based materials
subjected to external chloride environment – A review," Construction and Building Materials, V. 23, No. 1, pp.
1-13.
9. Zibara H, (2001), "Binding of External Chlorides by Cement Pastes," Ph.D. Thesis, Department of Building
Materials, University of Toronto, Toronto, Canada.
10. Yuan Q, Deng D, Shi C, De Schutter G, (2013), "Chloride binding isotherm from migration and diffusion tests,"
J Wuhan Univ Technol-Mat Sci Edit, V. 28, No. 3, pp. 548-56.
11. Nagataki S, Otsuki N, Wee TH, Nakashita K, "Condensation of Chloride Ion in Hardened Cement Matrix
Materials and on Embedded Steel Bars," ACI Materials Journal, V. 90, No. 4, pp. 323-32.
12. Glasser FP, (1999), "Role of Chernical Binding in Diffusion and Mass Transport," Intemtional Conference on
Ion and Mass Transport in Cernent-Based Materials, Toronto, Canada.
13. Ben-Yair M, (1974), "The effect of chlorides on concrete in hot and arid regions," Cement and Concrete
Research, V. 4, No. 3, pp. 405-16.
14. Collepardi M. Coppola L, Pistolesi C, (1994), "Durability of Concrete Structures Exposed to CaCl2 Based
Deicing Salts," ACI Special Publication, V. 145, pp. 107-15.
15. Farnam Y, Dick S, Wiese A, Davis J, Bentz D, Weiss J, (2015), "The influence of calcium chloride deicing salt
on phase changes and damage development in cementitious materials," Cement and Concrete Composites, V.
64, pp. 1-15.
3.11
@Seismicisolation
@Seismicisolation
Wei et al.
16. Monosi S, Collepardi M, (1990), "Research on identification in concretes damaged by CaCl2 attack," Il
Cemento, V. 87, No. 1, pp. 3-8.
17. Shi C, (2001), "Formation and stability of 3CaO·CaCl2·12H2O," Cement and Concrete Research, V. 31, No. 9,
pp. 1373-5.
18. Birnin-Yaui UA, Glasser FP, (1991), "Chlorides in cement: Phase studies of the Ca(OH)2-CaCl2-H2O system,"
Cemento, V. 88, No. 3, pp. 151-7.
19. Sutter L, Peterson K, Touton S, Van Dam T, Johnston D, (2006), "Petrographic evidence of calcium
oxychloride formation in mortars exposed to magnesium chloride solution," Cement and Concrete Research, V.
36, No. 8, pp. 1533-41.
20. Makarov SZ, Vol'nov II, (1964),"Phase Diagrams Ceram. (Vol. 1)." Westerville, OH: American Ceramic
Society.
21. Saetta AV, Schrefler BA, Vitaliani RV, (1993), "The carbonation of concrete and the mechanism of moisture,
heat and carbon dioxide flow through porous materials," Cement and Concrete Research, V. 23, No. 4, pp. 761-
72.
22. Kayyali O, Qasrawi M, (1992), "Chloride Binding Capacity in Cement‐Fly‐Ash Pastes," Journal of Materials in
Civil Engineering, V. 4, No. 1, pp. 16-26.
23. Kayyali OA, Haque MN, (1988), "Effect of carbonation on the chloride concentration in pore solution of
mortars with and without flyash," Cement and Concrete Research, V. 18, No. 4, pp. 636-48.
24. Villani C, Farnam Y, Washington T, Jain J, Weiss J, (2015), " Conventional Portland Cement and Carbonated
Calcium Silicate–Based Cement Systems: Performance During Freezing and Thawing in Presence of Calcium
Chloride Deicing Salts," Transportation Research Record: Journal of the Transportation Research Board, V.
2508, pp 48-54.
25. Ghantous RM, Unal E, Farnam Y, Weiss WJ, (2015), "The Influence of Carbonation on the Formation of
Calcium Oxychloride, Unpublished manuscript," Book The Influence of Carbonation on the Formation of
Calcium Oxychloride, Unpublished manuscript.
26. Lura P, Couch J, Jensen OM, Weiss J, (2009), "Early-age acoustic emission measurements in hydrating cement
paste: Evidence for cavitation during solidification due to self-desiccation," Cement and Concrete Research, V.
39, No. 10, pp. 861-7.
27. Verbeck G, (1956), "Carbonation of hydrated portland cement," Portland Cement Assoc R & D Lab Bull, V.
205, pp. 20.
28. A.M. N, (1990),"Properties of Concrete, 3rd Edition." London: Pitman Publishing.
29. Delagrave A, Marchand J, Ollivier J-P, Julien S, Hazrati K, (1997), "Chloride binding capacity of various
hydrated cement paste systems," Advanced Cement Based Materials, V. 6, No. 1, pp. 28-35.
30. Di Bella C, Villani C, Weiss J, "Chloride Transport Measurements for a Plain and Internally Cured Concrete
Mixture," ACI Special Publication, V. 290, pp. 1-16.
31. Luping T, Nilsson L-O, (1993), "Chloride binding capacity and binding isotherms of OPC pastes and mortars,"
Cement and Concrete Research, V. 23, No. 2, pp. 247-53.
32. Mohammed TU, Hamada H, (2003), "Relationship between free chloride and total chloride contents in
concrete," Cement and Concrete Research, V. 33, No. 9, pp. 1487-90.
33. Sergi G, Yu SW, Page CL, (1992), "Diffusion of chloride and hydroxyl ions in cementitious materials exposed
to a saline environment," Magazine of Concrete Research, V. 44, No. 158, pp. 63-9.
3.12
@Seismicisolation
@Seismicisolation
SP-308—4
PROBABILISTIC TREATMENT OF CHLORIDE THRESHOLD
Carmen Andrade, Fabiano Tavares, Nuria Rebolledo, David Izquierdo
Synopsis: Chloride threshold is known to be a variable value depending on many parameters related to the cement
chemistry, concrete characteristics and external environment. Several studies have been performed trying to find a
general law that can predict the threshold in a particular concrete. Although this goal is the most rigorous approach to
solving the problem, it seems very difficult to measure all influencing parameters in real work. Another approach
consists in analysing the variability in a rational way. This is done in the present work following a recent method
suggested in the new Model Code 2010, which considers deterministic to probabilistic models for service life
prediction. The depassivation event is analysed first from its progressive nature which aims to determine chloride
threshold variability, whose statistical distribution has been measured in the laboratory and which has shown to agree
with observations in real structures. Also, comments are given on the meaning of the probability of depassivation and
its consideration from a structural point of view. Finally an accelerated test method is mentioned which enables the
determination of the chloride threshold in testing times shorter than 4 to 8 weeks. Testing of samples containing
portland cement only and portland cement with slag are assessed with the accelerated test.
Keywords: Chloride; Concrete; Corrosion; Statistics; Testing; Threshold.
4.1
@Seismicisolation
@Seismicisolation
Andrade et al.
Carmen Andrade is Dr. in Industrial Chemistry, Research Professor of the IETcc-CSIC. Her work has been of
on durability with particular attention at the phenomenon of corrosion of steel in reinforced concrete. She has
been President of UEAtc, RILEM, WFTAO and Liaison Committee. She is Doctor Honoris Causa by the
Trondheim University (Norway) and by the University of Alicante (Spain). She has been General Director of
Technological Policy of the Ministry of Education and Science.
Fabiano Tavares is Mechanical Engineer by UF of Espiritu Santo and Institut Supérieur des Matériaux et de la
Construction Mécanique-Paris-France. He passed the MBE – Master in Business Engineering – UFF –
Universidad Federal Fluminense – Civil Engineering Department – Brazil and he is Dr. in Civil Eng by
Politechnical University of Madrid in 2013. He has been lecturer at UNIEST UNIVERSITY (Electrical
Engineering Department) from January 2003 to January 2005. He has been working 6 months at PETROBRAS
and at ANP – Petroleum National Agency of Brasil. Also has visited for 3 months EDF – Electricité de France
(PARIS) and RENAULT (PARIS. From 2005 is working at the Institute of Construction Sciences “Eduardo
Torroja” of the CSIC of Spain were he did the PhD.
Nuria Rebolledo is a Chemical Engineer from 2008 by the University Rey Juan Carlos of Madrid. She is currently
performing PhD studies at the Institute of Construction Sciences “Eduardo Torroja” of the CSIC of Spain on
chloride ingress in concrete. She has participated in numerous research activities related to service life prediction,
in particular, that related to the building of the new locks of Panama Canal.
David Izquierdo is Dr. in Civil Engineering in 2003 by the Polytechnical University of Madrid. He joined the
Institute of Construction Sciences “Eduardo Torroja” of the CSIC of Spain in 1998 until 2004. He had been
structural designer at Dragados, INTECSA, and at present is working for INITEC. He has been a lecturer at
Univertsity Alfonso X el sabio of Madrid and at present is at the Polytechnical University of Madrid.
INTRODUCTION
In reinforced concrete the steel passivates and remains stable if the surrounding conditions do not change.
Corrosion initiates when certain amount of chlorides, known as chloride threshold, reach the bar surface. In the
standards the chloride threshold is more often defined as 0.4% by cement weight. This value resulted from the
calculation of the bound and free chlorides in an ordinary Portland cement1. The chloride threshold results is one
of the important input parameters in the prediction of service life,2 which from the incorporation of the models
into the fib Model Code MC20103 has become a crucial aspect in the modelling in chloride-bearing environments.
The chloride threshold has attracted much attention4-15 in research due to its impact in the standards and in the
evaluation of concrete properties, protection methods and environmental conditions. Several reviews have been
more recently published16-18 which summarized very well the present situation and it is not to be repeated here.
The chloride threshold can be studied from different perspectives:
- One is the research level in which basic laws should be developed as a function of different controlling
parameters. This is the case of the previous studies made by the authors on the influence of the corrosion
potential on the chloride threshold16 or the influence of the chloride concentration into the corrosion rate
developed at the rebar19, 20. When the potential is used for the threshold identification, the value enabling
passivity disruption is referred as the pitting or breakdown potential21.
- Another perspective, which is being developed in this present communication, is the engineering need for the
chloride ingress modelling. In this respect the chloride threshold has to be understood as an input parameter
that has to be accurate enough to reasonably decrease the uncertainty of service life predictions. The most
used equation for prediction of the time to corrosion is the known as “error function equation” (equation 1),
now incorporated into MC2010 and more and more used worldwide (Cs is the surface concentration of
chlorides, Dap is the Apparent diffusion coefficient, t is the time, x is the distance of a certain chloride
concentration, erf is the error function, and Cx is the concentration inside the concrete and can be the chloride
threshold at the surface of the bar).
x
C ( x, t )  C s (1  erf ) (1)
2 Dap t
It is beyond the scope of present paper to detail the limitations of this equation, but we will concentrate on the
chloride threshold term. From this engineering approach this needs a clear definition, a value which could be
considered as an averaged one for deterministic calculations and a statistical distribution for probabilistic
calculations. Finally, a reasonably rapid test method should be suggested for the experimental verifications. All
@Seismicisolation
4.2
@Seismicisolation
Probabilistic Treatment of Chloride Threshold
these aspects will be addressed in the present paper and contributions to a better appraisal of chloride threshold
are offered.
DEFINITION OF THE CHLORIDE THRESHOLD
As mentioned in the previous papers16-18, 20 one of the sources of disagreement among the authors has been the
different manners and variables to define the threshold or to determine the pitting potential. Also, there is
discrepancy in the conversion of concentration in the pore solution and by weight of binder or concrete mass11, as
well as the cation accompanying the chloride8 and the roughness of the steel surface13.
For the definition it has to be considered on one hand, that once a pit develops it may repassivate and then the
depassivation has to be definitive which usually corresponds to a time period and not to a single event. From the
corrosion point of view, depassivation is considered when in a delimited surface of steel (let’s say of 1 cm2) the
corrosion current density is higher than 0.1 A/cm2, 19. In these references the measurement of the Polarization
Resistance was used to determine the corrosion rate, Icorr, in order to monitor the evolution of solutions simulating
the pore solution with different quantities of chlorides, or mortars also with several amounts of chlorides. The
results found for smooth and ribbed rebars are given in Table 1.
With respect to the definition as a function of the potential, Alonso et al.16 performed potentiostatic tests with six
different types of binders in order to determine the relation between free potential and chloride threshold. The
dependence found is shown in figure 1 and the expressions of the minimum averaged thresholds values are given
in Table 2.
Another interesting study has been made by G. Meira22 comparing the chloride thresholds in twin specimens
submitted to accelerated wet-dry cycling in 1M of sodium chloride solution with the natural exposure to the
airborne atmosphere at 10 m of sea border in a beach in Brazil. In the paper concretes are tested with three different
w/c ratios. The technique for detecting chloride depassivation was the monitoring of the Polarization Resistance
and associated Corrosion Potential. The author reported the following.
- The values of the threshold are smaller in natural conditions than with the accelerated wet-dry cycling, which
can be related to the fact that in the natural exposure the concrete is partially saturated while in the cycling
the conditions are more similar to wet and submerged conditions. The relation found was:
Cl (field) = 0.786 Cl (lab) – 0.373
- Values of potentials more negative than -275 mVSCE were not registered, which, following figure 1, indicates
that chloride threshold lower than 1.6% cement weight should have been recorded. However, in Meira’s study
in the laboratory conditions of wet-dry cycles, values of 1.8-2.45% by cement weight of chlorides were found,
which indicates that the potential frontier of -200 mVSCE of figure 1 for chloride threshold lower than 1.2%
has to be taken as indicative. Other variables, such as moisture content, bar surface roughness or steel
composition, should have some influence as well.
- According with other authors, in Meira’s study the chloride threshold results are lower with higher w/c ratios.
A final study on chloride threshold determination to be commented is that by Gil-de-Viedma et al23 who tested
the same type of concrete by three different methods: 1) lab natural diffusion (ponding), 2) potentiostatic and 3)
migration (integral test)24. For the sake of comparison, the specimens from natural diffusion and potentiostatic
tests will be compared (those from the integral test will be commented in the testing section later). Concrete
specimens have been fabricated with cement of the type CEM I-42.5R/SR, with a w/c ratio of 0.45. For the
diffusion tests, 7x7x7 cm (2.75x2.75x2.75 in.) in size specimens were used while for the potentiostatic tests
4x4x16 cm (1.6x1.6x6.3 in.) specimens were fabricated. The specimens were cast with a centred corrugated steel
rebar. The technique to detect depassivation was for the natural diffusion, the monitoring of the Polarization
Resistance and associated Corrosion Potential and the increase in current for the potentiostatic tests. Table 3
presents the results obtained.
The main deductions that can be made from these results are:
- The natural diffusion tests which last very long time have registered an averaged value of chloride thresholds
around 2%, which is similar to those found in Ref. 16 by wet-dry cycles. The test moisture conditions,
although not identical, are similar in both cases with longer periods in the saturated conditions.
- The value of 2% for the chloride threshold is again higher than in atmospheric conditions (see figure 2).
- With respect to the potentiostatic tests the averaged value is higher than in the tests shown in figure 1
(neglecting the case of -250 mVSCE the average is 1.165%) but are lower than those in wet or wet-dry
conditions.
@Seismicisolation
4.3
@Seismicisolation
Andrade et al.
A general observation, although still preliminary, from all these references is that in concretes that are very wet
such as permanent immersion or wet-dry cycles, the chloride threshold is higher than in concretes only partially
saturated and then, more oxygenated. The comparison between references is not straight forward due to the
different testing conditions.
Depassivation process and service life

The values of thresholds found and their scatter are in agreement with the ranges identified in numerous
publications1, 2, 4. The problem is that the use in predictive models of one value or other may represent a significant
difference in the application of the equation 1. Table 4 shows results from a calculation when considering two
critical chloride concentrations: 0.4 and 0.7%.
On the other hand, an additional source of scatter is that the depassivation step is not a single event, due to the
size of the bar as figure 3 depicts. The depassivation starts at the external surface of the bar closer to the concrete
surface and it progresses around the bar perimeter as the aggressive front moves inwards the concrete interior. It
may take decades until the back bar surface corrodes if the penetration rate of the aggressive front is slow.
Additionally, depending on the bar size, the depassivated front face may or not polarize the back bar face and this
may influence its chloride threshold, which may be different in the initial steps than during the further process.
All these considerations call for the need to link the depassivation event to the particular model or parameter
which is used for its determination. There is not the need for a single general definition of the chloride threshold,
but when selected the model or the controlling characteristic, the definition has to be univocal and clear. For
instance, if the shift in the corrosion potential is taken as the controlling parameter, then the level and amplitude
of the shift should be univocally defined. In the case of the corrosion rate, it is clearly established that if the value
increases to values above 0.1-0.2 A/cm2 active corrosion develops, but in order to account for possible
repassivation, this increase should be maintained during a time period of several weeks/months.
From an engineering point of view the demand that the definition should be related to the model used for the
service life prediction is even more evident and, if this model is equation 1, then the chloride threshold has to be
defined in terms of the critical chloride content at the level of the external bar surface. However, as this value is
not constant because of the influence of numerous parameters (state of bar surface, type of binder, concrete
humidity content, concrete porosity, pH of the pore solution, etc.) one possible approach to overcome this
difficulty is to make a probabilistic treatment.
PROBABILISTIC TREATMENT OF CHLORIDE THRESHOLD
If the chloride threshold at rebar level follows a statistical distribution, it would help the mathematical treatment
of the critical concentration value in the model of equation 1, because the establishment of a mean value and a
standard deviation would introduce a framework of engineering evaluation of the selected threshold value as will
be commented later.
The finding of a possible statistical distribution for chloride thresholds tested in the lab was undertaken by
Izquierdo et al.21 In that study, the values of 10 identical specimens tested (from figure 1) were treated with
standard statistical tools and it was concluded that the values for the range of potentials more positive than -200
mVSCE (atmospheric conditions) follow a normal distribution with a mean of 0.70 ± 0.20% by cement weight
(figure 2). As the distribution came from tests in mortar specimens, it was considered a reference but with no
practical application.
However Markeset published in 200425 the results of a survey made in bridges in Norway where cores were drilled
in order to study their state of chloride contamination and the values of chloride at the rebar level when the
corrosion was “incipient.” These chloride threshold values with incipient corrosion were considered to be the “site
chloride threshold.” The results were also statistically treated by Markeset and her distribution presented a mean
value of 0.77 ± 0.24% of chloride ion by cement weight (figure 4). Both distributions seem coherent from the
point of view that in real conditions and real partially saturated concrete, the amount of chloride needed for
producing corrosion is a bit higher than in the laboratory conditions, and then, it seems that there is a unique
chloride threshold statistical distribution. Results from the laboratory are more conservative and can be taken as
reference and considered a general one for atmospheric conditions in spite of the very different sources of the
data. Regarding the value the 0.4% with respect to cement weight of current codes, it corresponds to a corrosion
probability of around 7% in the distribution of figure 4.
@Seismicisolation
4.4
@Seismicisolation
With respect to the range of potentials more cathodic than -200 mVSCE (permanently wet or submerged conditions)
in Ref. 21 it was found that a log-normal distribution with a mean value of 1.53% 0.53 of chlorides by cement
weight (figure 5) fits. The fact that with more cathodic potentials the threshold is higher, for chloride thresholds
around 3% Cl by cement weight, explains the high values reported by different authors. This distribution with
higher average chloride threshold at more cathodic potentials agrees better with the thresholds found in wet-dry
testing or permanently wet conditions (in spite that in the natural conditions the potentials before depassivation
did not show so cathodic).
Then, it can be summarized that for service life modelling purposes there are reference statistical distributions of
the chloride threshold that can be used if a probabilistic treatment is made. For deterministic calculations these
distributions have the advantage to give the idea that the 0.4% by cement weight represents a conservative value,
being the average value for atmospheric conditions of 0.7% and of 1.53% by cement weight when the concrete is
very water saturated.
Depassivation limit state

Regarding which probability should be applied to the limit state of depassivation in regard to service life
calculation, in fib Model Code 2010 (see Table 3.3-5),3 it is ascribed for a life time of 50 years (without more
explanation) a Reliability Index =0 for “service life” and a =1.5 (a probability of around 7%) for irreversible
Serviceability Limit State (SLS). Later, only a =1.3 or 1.5 is suggested for the depassivation always identified
with a SLS. This identification between depassivation and SLS has two main limitations:
- The first is the difficulty to define when “corrosion depassivation” occurs, as, unless monitoring of the
electrochemical corrosion parameters is made, no external sign can be detected until concrete cover cracks or
corrosion spots appear on the surface. Even monitoring, the identification of the probability of 10% (=1.3)
would need certain corrosion progression which at present is not feasible to be accurately detected.
- The second is that when the bar starts to corrode, nothing is happening to the structure and then, the classical
definition of the SLS is not fulfilled (a state which corresponds to conditions beyond which specified
serviceability requirements for a structure or its components are no longer satisfied). Just after corrosion
initiation the structure satisfies all the design requirements.
This second aspect and the attribution of a probability of 10% to the depassivation is commented by Markeset 25
and Gulikers 26 as not based in sound arguments. Our proposal consists in instead of considering depassivation as
a SLS, to consider it a Limit State of “initiation of deterioration” as defined in the ISO Standard 1328327: initiation
limit state, ILS is a state which corresponds to the initiation of significant deterioration of a component of the
structure. This will aim into quantifying a probability of failure for depassivation differently than the 10%
indicated in MC 2010, that is, with a lower  value than 1.3. Our proposal does not consider depassivation as SLS
but as ILS with a lower  value than 1.3 and does however consider other structural consequences (as cover
cracking) as the SLS.
Taking the case of SLS of corrosion-induced cover cracking, we suggest calculating the time to depassivation and
the further corrosion propagation until reaching a corrosion attack, Pcorr, which could provoke a surface crack
width due to corrosion of around 0.3 mm (0.01 in). In figure 6 the accumulated corrosion, Pcorr (homogeneous
corrosion) is represented schematically for different stages after the initial corrosion onset. The progressive loss
in cross section starts from the outer bar surface. With only 10 m (0.39 mil) of corrosion penetration Pcorr a crack
of 0.05 mm ( 2 mil) in width may appear28 and for a crack of 0.3 mm (11.8 mil) in width, Pcorr = 100 m (3.9 mil)
is a reasonable value. The probability of reaching this corrosion level will depend on the rate of chloride ingress
and the further corrosion rate. This is shown in table 5 as an example using Monte Carlo simulation. For the values
of diffusion coefficients taken as examples in the table, values of corrosion rates were assumed according to the
concrete quality. The results indicate that the probability of achieving 100 m (3.9 mil) of diameter loss is higher
as faster is corrosion process. Then, the probability of failure to be ascribed to deterioration processes cannot be
a single value, but would depend on the rate of the deterioration itself. More calculations should be made to find
some general recommendation.
For the case of the verification of ultimate limit state (ULS), it should be considered all the damage types induced
by the corrosion process in addition to the decrease in cross section due to corrosion. These additional damages
are: the decrease in steel ductility, the reduction of concrete cross section load-bearing capacity due to the cover
cracking and the steel/concrete bond deterioration. The ULS verification should make several scenarios to identify
which failure mode is reached first.
@Seismicisolation
4.5
@Seismicisolation
Andrade et al.
CHLORIDE THRESHOLD TESTING
It has been deduced before that the chloride threshold depends on the test and the concrete moisture conditions.
Partial and fully saturated concrete have presented different averaged chloride threshold values. Then, the testing
should take into account these two concrete moisture conditions.
On the other hand, although natural testing seems the most realistic situation, the testing duration is very long due
the need to have a cover depth thick enough to represent real concrete surface finishing and needs the monitoring
of the corrosion state. Acceleration through wet-dry cycling may be also very long and has shown to represent
only wet concrete conditions.
Our proposal about testing can be summarized by: a) the use of potentiostatic tests to reproduce atmospheric
conditions as this test has presented chloride values close in those environment conditions to the real ones, as
shown by the work of Markeset25 and b) the use of another test type (integral test)24 in which all service life
parameters can be calculated. In both cases acceleration of chloride ingress is made through the application of an
electrical potential drop.
Potentiostatic tests9, 16, 21 consist in that a potential is applied to the bar and the current between an anode and
cathode is monitored. Mortar or concrete specimens can be suitable with a bar embedded with the real cover
thickness (figure 7). The specimen is introduced in a solution (better of 1M NaCl to ensure on the chloride binding)
and the bar is made to act as an anode with respect to a cathode that is placed in the external solution. When the
bar starts to corrode a dramatic increase of the current is noticed and then the test is stopped and the specimen
broken to take samples from near the concrete/steel interface, just where corrosion spots are detected. As shown
in figure 1, with potentials more positive than -200 mVSCE the chloride threshold is more or less constant, then a
potential of +100 MVSC seems a reasonable value to accelerate the process.
The integral test also uses concrete specimens with an embedded bar (figure 8).24. In this case the bar is not directly
polarized. The cover thickness should also reproduce reality. A ponding dam is glued to the surface of one of the
faces of the specimen, in which a 1.2 M Cl- solution was introduced. It is concentrated enough in order to not
exhaust the chlorides during the testing. A potential drop of 12 to 30 V is applied between the cathode located
inside the pond and the anode placed on the bottom surface of the specimen. The onset of corrosion is detected
by periodical monitoring of the corrosion potential of the bar (the voltage drop is periodically disconnected during
30 to 180 minutes to monitor the bar state). The depassivation can be detected also by measuring the Polarization
Resistance. The depassivation is detected when the potential shifts towards around -300SCE or the Icorr is higher
than 0.2 A/cm2. After depassivation is noticed, the specimen is broken to find the chloride threshold. A small
sample (about 2 g (0.004 lb)) is collected from the concrete/steel interface only where corrosion spots are detected.
In order to measure the corrosion rate after depassivation, other specimen is left to corrode naturally by switching
off the potential applied during around 15 to 30 days. If the Icorr in very corroded conditions is of interest, then the
potential can be left applied until the bar is supposed very uniform corroded and the Polarization Resistance is
measured.
Then the four main parameters of service life can be calculated through this integral test: the apparent diffusion
coefficient, the chloride threshold, the surface concentration and the corrosion rate after depassivation. A typical
plot of the corrosion potential evolution is shown in figure 9. 24 In table 6 are details of concrete composition and
results from the integral test of the example shown in figure 9. The effect of the slag is mainly noticed in the delay
of depassivation attributed to their higher binding capacity because the chloride threshold was relatively similar
in both concretes. Regarding the comparison with natural diffusion testing, figure 10 shows the similar values of
the Dap obtained from both test types.
It is worth noticing that the chloride thresholds values measured through this test are in the normal low range of
partial saturation conditions (figure 5). Therefore, it is an important tool for the testing because of the relative
short times needed and the possibility to test many variables, including different concrete compositions,
inhibitors25, coatings and other bar treatments.
CONCLUSIONS
In the 1970’s the chloride content of the constituent materials of the concrete was controlled for the avoidance of
reinforcement corrosion. The limit of 0.4% by weight of cement was considered a threshold values for corrosion
initiation. This threshold however was identified to be variable and with the need of the service life prediction it
@Seismicisolation
4.6
@Seismicisolation
has gained importance due to the impact in the calculations. In present work the main contributions to the
discussion on the optimal threshold value are:
- The chloride threshold depends on the definition and test method used for its identification.
- Moisture condition of the concrete is also a critical characteristic. Two main saturation degrees appear to give
different average chloride threshold values: partial and total moisture saturation.
- Corrosion initiation cannot be considered a Serviceability Limit State because it does not fit into the definition
by ISO standard 13283. Instead it can be considered and Initiation Limit State, ILS.
- Consequently, the attribution to depassivation of a general probability of 10% (Reliability Index of around
=1.3) is not correct. Additionally, the depassivation probability depends on the rate of
chloride/carbonation/corrosion.
- To calculate a Serviceability Limit State from the corrosion process needs to identify a “corrosion
consequence” as for instance is cover cracking induced corrosion. The calculation of the SLS needs the use
of a corrosion propagation model.
- From the different testing methods to determine the chloride threshold, the “integral corrosion method” has
shown to be very suitable for service life prediction as gives the four main needed parameters: diffusion
coefficient, chloride threshold, surface chloride concentration and corrosion rate.
ACKNOWLEDGMENTS
The authors are grateful to the Spanish Ministry of Economy and Competitiveness, MINECO, through the project
“Reinforcement corrosion with low oxygen or absence content” and to the Spanish Agency of Radioactive Waste
Storage: ENRESA for the financing of som of tehs tudies of this reasearch.
REFERENCES
1. Everett L.H., Treadaway K.J.W.: Deterioration due to Corrosion in Reinforced Concrete, BRE Information
Paper 12/80, Building Research Establishment, Garston, 1988
2. Tuutti, K., “Corrosion of Steel in Concrete,” Doctoral Thesis, Swedish Cement and Concrete Research
Institute (CBI), Stockholm, 1982, pp. 263-278.
3. MC 2010 Model Code 2010. fib (2012).
4. Byfors K., “Chloride-initiated reinforcement corrosion (chloride binding),” CBI Report 1:90, Swedish
Cement and Concrete Research Institute, Stockholm, 1990.
5. Hausmann D.A., “A probability model for steel corrosion in concrete,” Materials Performance V. 37, No.
10, 1996, pp. 64-68.
6. Hope B.B., A.K.C. Ip, “Chloride corrosion threshold in concrete,” ACI. Mater J, July-August, 1987, pp. 306-
314.
7. Gouda V.K., “Corrosion and corrosion inhibition of reinforcing steel,” Br Corros J, V. 5, 1970, pp. 198-203.
8. Andrade C., Page C.L., “Pore solution chemistry and corrosion inhydrated cement systems containing
chloride salts. A study of cation specific effects,” Cem Concr Res, V. 21, 1986, pp. 49-53.
9. Hansson C.M., Sorensen B., “The threshold concentration of chloride in concrete for initiation of
reinforcement corrosion, in: N. Berke, V.Chacker, D. Whiting (Eds.),” Corrosion Rates of Steel in Concrete,
ASTM Spec Tech Publ, V. 1065, 1988, pp. 3- 16.
10. Pettersson K., “Chloride threshold value and the corrosion rate in reinforced concrete, in: R.N. Swamy (Ed.),”
Proceedings of the International Conference on Corrosion and Protection of Steel in Concrete, Academic
Press, Sheffield, 1994, pp. 461.
11. Glass G.K., Buenfeld N.R., “The presentation of the chloride threshold level for corrosion of steel in
concrete,” Corrosion Science ,Vol 39, No. 5, 1997, pp. 1001-1013.
12. Breit W., Schiesl P., “Investigations on the threshold value of the critical chloride content,” Int. RILEM
Workshop on Chloride penetration into concrete. Edt. L.O. Nilsson and J.P. Olivier. France, 1995, pp. 441.
13. Mammoliti L.T., Brown L.C., Hansson C.M., Hope B.B., “The influence of surface finish of reinforcing steel
and pH of the test solution on the chloride threshold concentration for corrosion initiation in synthetic pore
solutions,” Cement& Concrete Research, V. 26, No. 4, 1996, pp. 545-550.
14. Hussain S.E., Rasheeduzzafar S.E., Al-Musallam A., Al-Gahtani A.S., “Factors affecting threshold chloride
for reinforcement corrosion in concrete,” Cement& Concrete Research, Vol. 25, 1995, pp. 1543- 1555.
15. Lambert P., Page C.L., Vassie P.R.W., “Investigation of reinforcement corrosion. Electrochemical
monitoring of steel in chloride contaminated concrete,” Mater Struct, V. 24, 1991, pp. 351- 358.
16. Alonso C., Castellote M., Andrade C., “Chloride threshold dependence of pitting potential of
reinforcements,” Electrochemical Acta, V. 47, 2002, pp. 3469-3481.
@Seismicisolation
4.7
@Seismicisolation
Andrade et al.
17. Ann K. Y., Song H-W., “Chloride threshold level for corrosion of steel in concrete Corrosion
Science,” Corrosion Science, V. 49, No. 11, 2007, pp. 4113-4133.
18. Angst U., Elsener B., Larsen C.K., Vennesland Ø., “Critical chloride content in reinforced concrete – A
review,” Cement and Concrete Research, V. 39, 2009, pp. 1122–1138.
19. Alonso C., Andrade C., Castellote M., Castro P., “Chloride threshold values to depassivate reinforcing bars
embedded in a standardized OPC mortar,” Cem. and Conc. Rs, V. 30, 2000, pp. 1047-1055.
20. Bertolini L., Bolzoni F., Pastore T., Pedeferri P., “New experiences on cathodic prevention of reinforced
concrete structures,” Corrosion of reinforcements in concrete construction, Edt. C.L. Page, P.B. Bamforth
and J.W. Figg, SCI Cambridge, 1996, pp. 389.
21. Izquierdo, D., Alonso C., Andrade C., Castellote M., “Potentiostatic determination of chloride threshold
values for rebar depassivation Experimental and statistical study,” Electrochimica Acta, V. 49, No. 17-18,
2004, pp. 2731-2739.
22. Meira G.R., Andrade C., Vilar E.O., Nery K.D., “Analysis of chloride threshold from laboratory and field
experiments in marine atmosphere zone,” Construction and Building Materials, V. 55, 2014, pp. 289–298.
23. de Viedma P.G., Castellote M., Andrade C., “Comparison between several methods for determining the
depassivation threshold value for corrosion onset,” J. Physics IV France, V. 136, 2006, pp. 79-88.
24. Andrade C., Rebolledo N., “Accelerated evaluation Accelerated evaluation of chloride corrosion by means
of the integral corrosion test,” Structural Magazine, Italy, Nº. 186, March-April 2014, paper 09.
25. Markeset G., “Critical chloride content and its influence on service life predictions Critical chloride content
and its influence on service life predictions,” Materials and Corrosion, V. 60, No. 8, 2009, pp. 593-596.
26. Gulikers J., “A simplified and practical approach regarding design for durability of reinforced concrete
structures based on probabilistic modelling of chloride ingress,” Concrete Repair, Rehabilitation and
Retrofitting II, Alexander et al (Eds.) 2009 Taylor & Francis Group, London.
27. ISO 13283- General principles on the design of structures for durability.
28. Andrade, C., Alonso, C., Molina, F.J., “Cover cracking as a function of rebar corrosion: Part I – Experimental
test,” Materials and Structures, V. 26, 1993, pp. 453-464.
@Seismicisolation
4.8
@Seismicisolation
APPENDIX
TABLES AND FIGURES
Table 1- Ranges of chloride threshold and trends found with respect to the averaged corrosion rate Imean
(averaged the whole testing time). The results are indicative of the scatter mentioned19.
Type of expression Ranges of chloride Trends found between Imean (A/cm2) and
threshold chloride contents
% Total Cl- (cement Wt.) 1.24 - 3.08 log Imean= -1.07 + 0.76 log (%Cl-)
% FREE CL- CEMENT WT 0.39 - 1.16 Log Imean= -0.74 + 0.64 log (%Cl-)
Cl-/OH- (Pore solution) - 1.17-3.98 log Imean = -1.04 + 0.57 log (Cl-/OH-)
- 0.66 - 1.45* log Imean = -0.84 + 0.90 log (Cl-/OH-)*
*Pore solutions and mortars together
Table 2- Values of the chloride threshold expressed as acid soluble (total), water soluble (free) and Cl-/OH-
ratio21.
Range of applied potentials E > -200  50 mV(ECS) E <//200mV50 (SCE)
Minimum mean chloride % Cl- total = 0.705  0.197 E= -/497x/log %Cltotal-/223

threshold values
E   2,533 10
3 8
Cl total  289 ,16 10 5 3 Cl total  1, 49554 10 8
% Cl- free = 0.50  0.03 E=-/615x/log % Clfree
Cl-/OH- = 1.76  0.3. E=-/465x/log Cl-/OH--/24
Table 3- Results of chloride thresholds obtained in a lab natural diffusion test (ponding) and potentiostatic tests.
Natural diffusion tests
Potentiotatic tests
(1M NaCl)
Chloride
Testing time Concentration Potential- % total Cl
thresholds % total Cl
(days) in the external mV(SCE) (cement
(cement wt.)
solution mVSCE wt.)
2.70 336 0.5M -250 2.34
1.63 440 0.5M -250 0.72
3.42 700 0.5M -100 0.74
Individual 1.95 463 0.5M -100 0.82
Values 0.75 222 0.5M -100 0.53
3.0M -100 2.27
3.0M +100 1.43
3.0M +100 1.20
Mean value 2.09 432 1.26 0.93
Stand.dev 1.02 178 0.76 0.48
CoV (%) 48.84 41.09 60.44 51.04
Table 4- Calculation through equation 1 with two critical chloride concentrations: 0.4 and 0.7%.
Parameter Values
Surface concentration, Cs (%cement wt.) 4 4
Apparent Diffusion Coef. (cm2/s) 1E-8 1E-8
Time life (years) 50 50
Critical Cl conc. (% cement wt.) 0.4 0.7
Depth of Ccrit (cm) 9,24 7,61
@Seismicisolation
4.9
@Seismicisolation
Andrade et al.
Table 5- Values of the parameters and time to depasivation for two examples of chloride ingress using equation
1 and 2 and Monte Carlo simulation for the calculation of the random distributions.
Case Parameter Average Standard Time to reach Failure
value deviation 100 m probability for
(years) 100 m (%)
1 Dap (cm2/s) 3E-8 3E-9
Surface concentration 0.5 0.05 11  8.6%
(wt.% binder)
Aging factor 0.3 0.05
Concrete cover (cm) 5 1
Initial concentration (wt.% 0.01 0.005
binder)
Critical Chloride 0.05 0.005
concentration (wt.%
binder)
Corrosion rate (m/year) 13 1
2 Dap (cm2/s) 1E-8 3E-9
Surface concentration 0,5 0.05 28  2.7%
(wt.% binder)
Aging factor 0.3 0.05
Concrete cover (cm) 5 1
Initial concentration (wt.% 0.01 0.005
binder)
Critical Chloride 0.05 0.005
concentration (wt.%
binder)
Corrosion rate (m/year) 10 1
Table 6- Concrete composition and results of the integral test.

Concrete types Without slag With slag
Type CEM I 42,5R CEM I 42,5R
Cement
kg/m3 350 175
Slag kg/m3 - 175
w/c 0,45 0,45
aggregate (5/16) kg/m3 943 943
aggregate (0/4) kg/m3 899 899
Time to depassivation (days) 23,46 181,7
Dap (m2/s) 27,355E-12 5,136E-12
(% by weight of
Ccrit 0,535 0,635
cement)
@Seismicisolation
4.10
@Seismicisolation
Figure 1- Values of chloride thresholds (by weight of cement) for the different types of binders and the
potentials applied. The values at each potential indicate the individual values of 10 twin specimens tested for the
same potential and binder type21.
Figure 2- Chloride thresholds reported by Meira et al.16 in natural conditions and in laboratory testing by wet-
dry cycles. The technique to detect depassivation was the monitoring of the Polarization resistance and
associated corrosion potential.
@Seismicisolation
4.11
@Seismicisolation
Andrade et al.
Figure 3- Progressive events of depassivation during carbonation and chloride penetration.
Figure 4- Statistical distributions of the chloride threshold21 made in laboratory conditions and Markeset25 taken
from bridges in the field.
Figure 5- Statistical distributions of the chloride threshold (Izquierdo, et al. 2004) made in laboratory conditions
for two ranges of potential: a) more anodic than -200 mV (Ag/AgCl) typical atmospheric conditions and b) more
cathodic than -200 mV (Ag/AgCl) typical of submerged conditions.
@Seismicisolation
4.12
@Seismicisolation
Figure 6- Progressive loss in cross section with the advance of the aggressive front.
Figure 7- Arrangement for potentiostatic test.
Cathode -
ZONES OF EXTRACTION
OF SAMPLES FOR NaCll- 1.2 M
CHLORIDE ANALYSIS
dV
30 Specimen
Isolating tape Anode
Figure 8- Test setup of the integral test24
@Seismicisolation
4.13
@Seismicisolation
Andrade et al.
Figure 9- From the evolution of the corrosion potential after disconnecting the potential drop in the integral test,
is possible to detect the depassivation event. In the figure the concrete with slags depassivated much later than
that made with OPC. There were prepared twin specimens that corroded almost simultaneously.
Figure 10- Values of Dap for the two concretes of table 6 obtained through the integral test and natural diffusion
(ponding test).
@Seismicisolation
4.14
@Seismicisolation
SP-308—5
ACCELERATED MORTAR TEST METHOD TO DETERMINE CHLORIDE

THRESHOLD VALUES
Neal S. Berke, Matthew A. Miltenberger, Lianfang Li, Brian Miller and Ralf Carvajal
Synopsis: Chloride-induced corrosion of steel in concrete adversely affects the service life of concrete in marine or
deicing salt environments. Service life models determine failure based upon the time to reach a critical chloride
concentration at the reinforcing bar known as the chloride threshold. Unfortunately, there is no universally accepted
value or test method for determining this critical chloride threshold value. In this paper a new test method
developed in ASTM subcommittee G1.14, Corrosion of Metals in Cement, Mortar or Concrete is discussed. This
paper presents the preliminary screening research conducted in several laboratories that was used to develop the
method.
The proposed test method uses reinforcing steel embedded in a standard mortar. Macrocell current and polarization
resistance measurements are used to detect corrosion initiation. The data show that the mean critical chloride
threshold value is close to what is commonly accepted, but is quite variable. The data suggests that the chloride
threshold values follow a normal probability distribution. The time to reach the threshold value is a function of
curing time and w/cm. Addition of corrosion inhibitors significantly increases time to corrosion initiation, due to
increasing the critical chloride threshold value, reducing chloride ingress, or both.
Keywords: Chloride; Chloride threshold; Concrete; Corrosion; Corrosion inhibitors; Reinforcement; Service
life
@Seismicisolation
@Seismicisolation
5.1
Berke et al.
Dr. Neal S. Berke is Vice President, Research at Tourney Consulting Group, Kalamazoo, MI. He is a member of
ACI Committees 212, Chemical Admixtures; 222, Corrosion; 224, Cracking; and 365, Life Prediction. He has
extensive experience in the development and application of concrete admixtures and materials as well as corrosion
and the durability of concrete. He is a recipient of the J.C. Roumain Innovation in Concrete Award. He has a Ph.D.
from the University of Illinois, Urbana, and a BA from the University of Chicago.
Matthew A. Miltenberger is Vice President of Vector Corrosion Services, Inc. a corrosion consulting firm in Wesley
Chapel, FL. He has over 30 years of experience in concrete construction, concrete materials engineering, repair and
restoration of infrastructure, service life modeling, and research on durability aspects of concrete materials. Matt
earned a Master of Science in Civil Engineering from the University of Maryland, and a Bachelor of Business
Administration in Construction Management from University of Miami, FL
Lianfang Li, Member ACI, is currently the director of product engineering with SILPRO LLC in Ayer,
Massachusetts. He formerly worked as a principal scientist for Grace Construction Chemicals. He holds PhD in civil
engineering and has expertise in corrosion and cementitious materials.
Brian Miller is the Global Marketing Director, Precast Concrete for W.R. Grace, and has more than 28 years of
experience in the precast concrete and concrete materials industries. He holds degrees in civil engineering and a
MBA. His technical experience includes concrete materials and mix design, durability, corrosion and service-life
prediction, construction and project management, architectural precast, and enclosure systems. Miller is an ACI
member, and past Chairman of the Committee 533, Precast Panels.
Ralf Carvajal is a Chemist with an MSc. in Physical Chemistry from the University of Texas A&M. He has been
involved in the Construction Chemicals industry for 27+ years with main activities in concrete admixtures, cement
chemistry, cementitious mortars, corrosion protection, epoxy resins and polyurethane resins for coatings and floors,
acrylic coatings, and bituminous products. His activities have been carried out in Colombia, the United States, and
various countries in Europe.
INTRODUCTION
Corrosion of steel in reinforced concrete is a major cause of the premature failure of concrete structures exposed to
chlorides from deicing salt or marine exposures. Solution experiments have indicated that there is a critical chloride
threshold value for corrosion initiation [1-4]. Numerous laboratories have reported threshold values for corrosion
initiation based upon chloride content relative to the amount of cement or concrete, typically ranging from 0.17 to
2.5% 5-11. The primary reason for such wide variability is that there is no standard test method for determining and
reporting the chloride concentration needed for corrosion initiation. The lack of a standard test limits useful
comparisons between work performed by different laboratories and hinders modeling of the time to corrosion
initiation.
This paper describes work conducted at four laboratories to develop an ASTM standard test method to determine the
chloride threshold value as a function of cement content. The test program results indicate that the threshold value
is not a fixed value, but follows a normal probability distribution. Curing time, w/cm, and corrosion inhibitors were
shown to affect time to corrosion. In addition, reducing w/cm or adding corrosion inhibitors significantly increases
the chloride threshold. The authors at this time are not associated with the laboratories that conducted the testing.
The laboratories were Grace Construction Products (Laboratory 1), BASF (Laboratory 2), Axim Concrete
Technologies (Laboratory 3), and Sika Corporation (Laboratory 4).
EXPERIMENTAL PROCEDURES
Mortar Mix Designs
Four different laboratories participated in this work. Three of the laboratories (Laboratories 1, 2, and 3) used a
cement meeting the requirements of Type I and II according to ASTM C 150. The other laboratory (Laboratory 4)
used a Type I cement. The basic mixture design was mortar with a w/cm of 0.485 as suggested in ASTM C 109.
The cement factor was 843 lb/yd3 (500 kg/m3) and the sand content was 2318 lb/yd3 (1375 kg/m3).
@Seismicisolation
@Seismicisolation
5.2
Accelerated Mortar Test Method to Determine Chloride Threshold Values
Two of the laboratories produced additional mixtures to determine the effects of fly ash, silica fume, lower w/c
(Laboratory 1) and corrosion inhibitors (Laboratory 2). Table 1 shows the modified mortar mixture designs used by
Laboratories 1 and 2.
Table 1 — Mixture Variations in Laboratories 1 and 2

Laboratory Mix # CF, lb/yd3 Pozzolan Pozz. Dosage, Admixture Dose,
(kg/m3) lb/yd3 (kg/m3) fl.oz./cwt (L/100kg)
1 1 843 (500) - - - -
1 2 942.5 (559)* - - PC** 8 (0.52)
1 3 661 (392) Type F Fly 165 (98) - -
Ash
1 4 769 (456) Silica Fume 67.4 (40) PC** 4 (0.26)
2 8 843 (500) - - OCI** 14.7 (0.96)
2 9 843 (500) - - CNI** 59 (3.85)
2 10 843 (500) - - - -
2 11 843 (500) - - OCI** 20.7 (1.35)
2 12 843 (500) - - CNI** 83 (5.42)
Note: C109 sand was held constant at 1375 kg/m3 for all mixtures and w/cm was 0.485.
*w/c=0.40, cement volume increased to keep batch volume constant
**PC is polycarboxylate high range water reducer, OCI is an amine-ester based organic corrosion inhibitor, and
CNI is 30% calcium nitrite corrosion inhibitor.
Compressive strengths for the mortars produced in Laboratories 1 and 3 are given in Table 2. Differences are
expected as different materials were used.
Table 2 — Compressive Strength Data

Laboratory Mix # fc @ 1 day fc @ 3 day fc @ 7 day fc @ 14 day
psi (MPa) psi (MPa) psi (MPa) psi (MPa)
1 1 1639 (11.3) - 4075 (28.1) 5626 (38.8)
1 2 943 (6.5) - 5409 (37.3) 6366 (43.9)
1 3 1044 (7.2) - 3509 (24.2) 4596 (31.7)
1 4 1218 (8.4) - 3654 (25.2) 4785 (33.0)
3 1 2146 (14.8) 3625 (25.0) 4423 (30.5) -
Specimens
Cylindrical corrosion specimens were produced with a height of 175 mm and a 150 mm diameter. Two No. 4 (13
mm) reinforcing bars were placed four inches (100 mm) from the top surface and one inch (25 mm) from the bottom
surface. The ends of the bars were protected as in ASTM G 109 and had an exposed length of four inches (100
mm).
Laboratory 1 cured specimens for 14 days in a fog room and dried them for 14 days at 50% RH at room temperature
of approximately 72 °F (22 oC) for two weeks before testing. Laboratories 2-4 moist cured for a total of 28 days and
subjected specimens to ponding after 14 days of drying under similar conditions to Laboratory 1.
After curing the top 3.125 inch (80 mm) of concrete was removed to provide ¾ inch (20 mm) of cover over the top
bar. The outside was painted with epoxy as described in ASTM G 109. A dam was placed on top of the specimens
and sealed with silicone caulk. A 10-ohm resistor was attached to the specimens to use in macrocell corrosion
measurements. Figure 1 is a schematic of the specimens. Note that Laboratory 4 only produced specimens for
chloride ponding.
The number of corrosion specimens per mixture varied between Laboratories 1-3. Laboratory 1 cast fifteen (15)
specimens per mixture. Laboratory 1 determined the chloride content of ten (10) specimens per mixture at corrosion
initiation, and determined the chloride content of the remaining five (5) at the end of the test program. Laboratory 2
tested six (6) specimens, where three (3) were removed for chloride analysis after initiation and the remainder were
@Seismicisolation
@Seismicisolation
5.3
Berke et al.
tested at set time periods for chloride analysis. Laboratory 3 had ten (10) specimens that were tested for chloride
analysis after corrosion initiated.
100 mm
75 mm
3% NaCl solution
20 mm cover
10 ohm
Figure 1 — Schematic cross-section of test specimen
Corrosion Testing
Corrosion testing consisted of cyclic ponding with 3% NaCl solution. In addition, Laboratory 4 examined the effect
of chloride ingress as a function of solution concentration using 3, 6, 9 and 12% NaCl. The specimens were ponded
for four days at which the solution was removed and three days of drying occurred. Macrocell current was
measured on the first and last day of ponding (though some labs measured it more often) as described in ASTM G
109. In addition, polarization resistance at a scan rate of 0.1 mV/s using IR interruption was performed by some of
the laboratories. Corrosion potentials were measured as described in ASTM C 876. Specimens were considered to
be corroding when the macrocell current exceeded 1 µA for two cycles.
Chloride Analysis
Specimens were removed for chloride analysis when they had a macrocell current over 1 µA for two cycles in
Laboratories 1, 2, and 3 and, and at fixed times in Laboratories 2 and 4. Chloride was then analyzed by either taking
specimens at various depths to have a profile that included the reinforcing bar depth, or by taking a 0.5 inch (12.5
mm) wide sample centered at the midpoint of the upper bar 0.75-inch to 1.25-inch (20-32.5 mm depth). The total
acid soluble chloride was determined as in ASTM C 1152.
RESULTS
Macrocell current and corrosion potential curves, for the control mixture without admixtures, versus time, from
Laboratory 1 are shown in Figures 2 and 3. A sharp increase in macrocell current was associated with a significant
drop in the corrosion potential. Similar behavior was observed in the other laboratories, and their time to corrosion
data is shown in Figures 4 and 5.
@Seismicisolation
@Seismicisolation
5.4
Figure 2 — Macrocell Current Versus Time for Laboratory 1 Control Mix #1
Figure 3 — Corrosion Potential Versus Time for Laboratory 1 Control Mix #1
@Seismicisolation
@Seismicisolation
5.5
Berke et al.
Figure 4 — Macrocell Corrosion for Control Mixture in Laboratory 3
Figure 5 — Typical Polarization Resistance and Corrosion Potential Data Laboratory 2 Control Mixture #10.
Decrease in Rp, and corrosion potential are indicative of corrosion initiation.
The addition of corrosion inhibitors significantly increased times to corrosion, with none of the inhibitor specimens
in Laboratory 2 showing any corrosion up to 62 weeks of testing.
Figures 6-8 show the macrocell data versus time for the lower w/c-0.4, fly ash, and silica fume mixtures for mortars
produced by Laboratory 1. As in the case of the control mixes there is a wide range of initiation times. The fly ash
mixture, Figure 8 has what looks like a bimodal distribution as some specimens fail early and others failed quite late.
The silica fume mixture gave a slight improvement in performance, and the lower w/c mixture a more significant
increase. The averages for the fly ash mixture are somewhat misleading given the distribution.
@Seismicisolation
@Seismicisolation
5.6
Figure 6 — Macrocell Versus Time for 0.4 w/c Mortars in Laboratory
Figure 7 — Macrocell Versus Time for Fly Ash Mortars in Laboratory 1
@Seismicisolation
@Seismicisolation
5.7
Berke et al.
Figure 8 — Macrocell Versus Time for Silica Fume Mortars in Laboratory 1
There is a fairly good correlation between corrosion rate (1/Rp) and corrosion potential as seen in Figure 10 where
there is a linear region on the semi-log plot between about –200 to –450 mV vs. SCE. The large change in slope at
about –500 mV vs. SCE indicates that a limiting current density is being approached. Above –200 mV the steel is
passive so corrosion potential is not a good indicator of corrosion rate. This relationship has been noted in the past
for reinforced steel corroding in concrete subjected to chloride [12].
Figure 9 — Corrosion Rate (1/Rp) Vs. Corrosion Potential from Laboratory 1 Control Mix #1
@Seismicisolation
@Seismicisolation
5.8
Chloride analyses showed that as in the time to corrosion, there is a wide spread in the chloride level needed to
initiate corrosion. Table 3 has chloride at initiation for Laboratory 2 control specimens, and highest chloride
reached for inhibitor specimens that were not corroding. Table 4 gives the chloride threshold data for Laboratory 3.
The chloride threshold data for Laboratory 1 are represented on a normal probability plots in Figure 10. The fly ash
mortar has a bimodal distribution, but the other mixtures follow normal probability statistics. The significance of
these results will be discussed later.
Table 3 — Chloride Threshold Data from Laboratory 2

Mix Specimen Time to Corrosion Chloride as %Cementitious
(weeks or cycles)
Control 10A >36 0.466
Control 10B 22 0.54
Control 10C 22 0.31
Control 10D 22 0.50
Control 10E >62 1.23
Control 10F 21 1.18 (41 weeks)
OCI 9A >36 0.057
OCI 9B >62 0.35
CNI 8A >36 0.57
CNI 8B >62 0.87
Table 4 — Threshold Chloride Values Obtained in Laboratory 3

Specimen Corrosion Initiation Chloride as %Cementitious
(weeks or cycles)
1 4 0.82
2 7 1.04
3 2 0.77
4 2 0.46
5 5 0.53
6 >7 -
7 1 0.34
8 2 0.64
9 5 1.65
10 6 1.92
@Seismicisolation
@Seismicisolation
5.9
Berke et al.
Figure 10 — Normal Probability Plots for Chloride Threshold wt% cementitious for Laboratory 1
Comparing the time to corrosion data in Tables 3 and 4 and Figures 3 and 4, it appears that overall Laboratory 3
specimens corroded earlier. However, the chloride threshold data are comparable between the three laboratories.
The effect of chloride concentration on chloride ingress can be seen in Figure 11. Chloride content in the top one
inch (25 mm) is increased at the NaCl levels above 3%, however, the values at two inches (50 mm) are essentially
the same, and the chloride contents for the 3% NaCl at one inch (25 mm) are still substantial.
Percent chloride
Figure 11 — Percent Chloride on Cementitious as a Function of Time and Chloride Concentration in the Ponding
Solution at 25 and 50 mm (Laboratory 4).
@Seismicisolation
@Seismicisolation
5.10
DISCUSSION
The data strongly indicate that the chloride threshold value for corrosion initiation is not a fixed value, but follows a
normal probability distribution as seen in Figure 10. This would be consistent with the wide range of chloride
threshold values mentioned in the literature [1-11]. For a single threshold value for chloride induced corrosion
initiation to be used in modeling, the engineer would need to determine an acceptable probability of failure below
this threshold value. Given the ability of this test method to provide the probability distribution, a better approach is
to use probabilistic models based on Monte Carlo simulation or other techniques that make use of the concrete
cover, chloride threshold value, and chloride ingress distribution parameters.
The addition of silica fume had a minor effect at the w/cm level used in this test method and the short curing period.
This doesn’t mean that silica fume is not effective, as ample data exist showing that it substantially improves time to
corrosion in lower w/cm systems with higher cover [13]. What this data does show is that silica fume is not a
corrosion inhibitor. Or more simply put, silica fume addition does not significantly increase the chloride threshold.
The fly ash results produced a bimodal distribution. Given the overall statistical distribution of some specimens
failing earlier than others, this is not unexpected. As time increased, the fly ash mortar specimens that didn’t
corrode were still hydrating, and thus their permeability reduced. This is seen in Figure 13 based upon fitting the
chloride profiles from Laboratory 1 data to Fick’s Law. The top 0.1 inch (2 mm) were not included in the analyses
to minimize sorption effects. These data indicate that in field exposure, extended moist curing of concretes
containing fly ash is important.
Figure 10 shows that there was a significant improvement in chloride initiation concentration by lowering the w/c.
This is consistent with an increase in pH that would occur with the same cement at a lower w/c [8,14]. However, the
improvements were not as substantial as noted below for the case of inhibitors.
Figure 12 — Diffusion Coefficient Probability Plots, Laboratory 1
Finally, the data from Laboratory 2 clearly show that inhibitors substantially improve time to corrosion initiation as
none of the specimens failed within 62 cycles. In the case of calcium nitrite the 36 and 62 week chloride levels
exceeded those of the reference mortar at 22 weeks where corrosion initiated. The OCI mortars at 62 weeks had
similar chloride concentrations at the reinforcement level as the reference had at 22 weeks, but still were not
corroding. Longer testing times would be required to determine the chloride threshold values for the inhibitor
@Seismicisolation
@Seismicisolation
5.11
Berke et al.
mixtures. Though not addressed in this study, the combination of lower w/c and inhibitor could bring about a
synergy in increasing chloride threshold levels [15-16], and future work with this test method could help to quantify
the improvements.
The chloride distributions as a function of chloride concentration in the ponding water indicate that the 3% NaCl
concentration will provide sufficient chloride to initiate corrosion in a reasonable timeframe. The slightly slower
buildup at the reinforcing bar level at earlier times gives more time for corrosion initiation at a lower chloride
content which is more consistent with reinforcement in good quantity concretes with more cover.
Subsequent testing has also indicated that it is very important to pickle the reinforcing or to retain the original mill
scale on the reinforcing to obtain realistic test results in a reasonable time period. The mill scale on reinforcing steel
is a very good cathode surface and helps facilitate corrosion initiation. Removal of this cathode surface by wire
wheel or sand-blasting, or polishing significantly extends the time to corrosion initiation, resulting in unrealistic
elevated chloride threshold values.
CONCLUSIONS
Based upon the results obtained by the laboratories participating in this work, the chloride threshold value is best
represented as a normal probability distribution. This accounts for the wide range of values in the literature. It
appears that the method discussed herein provides results consistent with those reported in the field, and helps to
explain the wide range of threshold values stated with control specimens showing failure within six months.
Future work includes standardizing the method to conduct an interlaboratory test as a means of having a standard
test method to determine the chloride threshold value for corrosion of steel in cementitious systems. Though the
standard test method will be at the 0.485 w/c with a Type II cement, the method as shown, can be used to evaluate
corrosion inhibitors, pozzolans, and changes in w/cm. As different cements could have significantly different
chloride binding and diffusion properties at the same w/cm, this method could be useful in quantifying differences
on chloride threshold values as a function of cement type and work is planned in this area.
ACKNOWLEDGEMENTS
The authors wish to thank the management of their former companies for allowing them to participate in and publish
this work as a joint effort. The authors also wish to acknowledge the valuable contributions by Violetta Munteanu,
formerly of Axim, and Maria Hicks, formerly of W.R. Grace, conducted at their former employer’s research
laboratories. Also, we thank Dr. A. Rosenberg of the Concrete Corrosion Inhibitor Association for helping to
facilitate this intercompany testing program.
REFERENCES
1. Hausmann, D. A., “Steel Corrosion in Concrete. How Does It Occur?,” J. Materials Protection (1967) 19-23.
2. Gouda, V. K., Corrosion and Corrosion Inhibition of Reinforcing Steel, Br. Corrosion Journal, 5, (1970) 198-
203.
3. Goni, S. and Andrade, C., “Synthetic Concrete Pore Solution Chemistry and Rebar Corrosion Rate in the
Presence of Chloride,” Cemenet and Concrete Research, 20 (1990) 525-539.
4. Berke, N. S., "The Use of Anodic Polarization to Determine the Effectiveness of Calcium Nitrite as an Anodic
Inhibitor," Corrosion Effect of Stray Currents and the Techniques for Evaluating Corrosion of Rebars in
Concrete, ASTM STP 906, V. Chaker, Ed., American Society for Testing and Materials, West Conshohocken,
PA (1986), pp. 78-91.
@Seismicisolation
@Seismicisolation
5.12
5. Rasheduzzafar, M., Al-Sandoun, S. S., Al Gahtani, A. S. and Dakhil, F. H., “Effects of Tricalcium Aluminate
Content of Cement on Corrosion of Reinforcing Steel in Concrete,” Cement and Concrete Research, 20 (1990)
723.
6. Slater, J. E., Corrosion of Metals in Association with Concrete, ASTM STP 818, ASTM International, West
Conshohocken, PA (1983).
7. Kayyali, O. A. and Haque, M. N., “The Ratio of Cl-/OH- in Chloride Contaminated Concrete: A Most
Important Criterion,” Mag. Concrete Res. 47 (1995) 235-242.
8. Tuutti, K., Corrosion of Steel in Concrete—CBI Research Report 4:82, Swedish Cement and Concrete Research
Institute, Stockholm (1982).
9. Hansson, C. M. and Sorensen, B., “Threshold Concentration of Chloride in concrete for Initiation of
Reinforcement Corrosion,” Corrosion Rates of Steel in Concrete, ASTM STP 1065, N. S. Berke, V. Chaker, and
D. Whiting, Eds., American Society for Testing and Materials, West Conshohocken, PA (1988) 3-16.
10. Bamforth, P. B. and Chapman-Andrews, J. F., “Long Term Performance of Reinforced Concrete Elements
Under UK Coastal Conditions,” Corrosion and Corrosion Protection of Steel in Concrete—I, R. N. Swamy,
Ed., Sheffield Academic Press, Sheffield (1994) 139-156.
11. Lukas, W., “Relationship Between Chloride Content in Concrete and Corrosion in Untensioned Reinforcement
in Austrian Bridges and Concrete Road Surfacings,” Betonwerk Fertigeil-Tech 51 (1985) 730.
12. Berke, N. S., "The Effects of Calcium Nitrite and Mix Design on the Corrosion Resistance of Steel in Concrete
(Part 2, Long-Term Results)," Corrosion 87, National Association of Corrosion Engineers, Houston (1987),
Paper Number 132.
13. Berke, N. S., “Resistance of Microsilica Concrete Concrete to Steel Corrosion, Erosion and Chemical Attack,”
in Fly Ash, Silica Fume, Slag, and Natural Pozzolans in Concrete, Proceedings Third International Conference,
Trondheim, Norway, SP-114, ed. V. M. Malhotra, American Concrete Institute, Detroit, p. 861, 1989.
14. Tritthart, J. and Banfill, P. E. G., Cement and Concrete Research, 31, (2001) 1093-1100.
15. Schiessel, P., “Effectiveness and Harmlessness of Calcium Nitrite as a Corrosion Inhibitor”, RILEM
International Symposium on the Role of Admixtures in High Performance Concrete, Monterrey, Mexico, 1999.
16. Berke, N. S., “Corrosion Inhibitors in Concrete,” Concrete International, Vol.13, No. 7, 1991, pp. 24-27.
@Seismicisolation
@Seismicisolation
5.13
Berke et al.
@Seismicisolation
@Seismicisolation
5.14
SP-308—6
IMPACT ON ANTICIPATED SERVICE LIFE OF CHLORIDE THRESHOLDS
Kyle Stanish
Synopsis: The corrosion-initiation threshold for corrosion is a key parameter that controls the expected service life of
a concrete structure. Different methods of establishing threshold values that consider different components of concrete
mixtures can have a significant impact on anticipated service life. This impact is evaluated using typical concrete
mixtures that are used in structures exposed to chloride-laden environments across the United States. The often
synergistic impact of including supplementary cementing materials (SCMs) in a concrete mixture and corrosion-
initiation thresholds is examined, where the reduction of diffusion values with time can lead to a greater impact that
would be expected from variation in corrosion-initiation thresholds alone.
Keywords: Concrete; Corrosion; Service Life Modelling; Corrosion-Initiation Thresholds; Supplementary

Cementitious Materials
6.1
@Seismicisolation
@Seismicisolation
Kyle Stanish
Kyle Stanish, Ph.D., S.E., P.E., ACI member, is a Senior Restoration Consultant at Walker Restoration Consultants
in Chicago, IL. He completed his Ph.D. in 2002 at the University of Toronto, Toronto, Canada on the migration of
chloride ions in concrete under an electrical potential gradient. He currently evaluates deteriorated concrete structures
and designs appropriate rehabilitation approaches. He is currently the Secretary of ACI 365: Service Life Prediction.
INTRODUCTION
The service life of a concrete structure is determined by a number of mechanisms, but chloride-induced corrosion
frequently controls for concrete structures exposed to either deicing salts or salt water. For a conventionally reinforced
concrete structure, the time to first repair depends on the rate at which chlorides enter the concrete (the diffusion
coefficient), the depth of the surface steel layer and the corrosion-initiation threshold.
Supplementary cementing materials (SCMs) have become ubiquitous in concrete mixtures. While this is a ‘green’
exercise as it diverts materials from disposal, it also improves the properties of the concrete mixture. Among other
effects, the use of SCMs has been shown to reduce the long term diffusion coefficients of concrete. The current ACI
building code for concrete [ACI 318-14] bases chloride limits solely on the Portland cement content of the mixture,
however. Inclusion of large proportions of SCMs can have the effect of reducing the corrosion-initiation threshold
when expressed in terms of the mass of the concrete. The impact that this has on the expected service life of a concrete
exposed to a chloride-laden environment is discussed.
BACKGROUND
In reinforced concrete structures, the steel embedded in the concrete is initially passivated. In the high pH environment
that is present in the concrete pores, a thin dense oxide layer is formed which is tightly adhered to the steel. This
tightly adhered layer then prevents further corrosion of the reinforcing steel. Until this layer is compromised, the steel
remains protected.
The expected service life of a concrete structure exposed to a chloride-contaminated environment is usually controlled
by the onset of corrosion. Corrosion of the reinforcing steel is initiated by chloride ions penetrating through the
concrete and reaching to the level of the steel. When sufficient concentration of chloride ions is reached, the protection
provided by the passivating layer is overcome, and corrosion will initiate. The chloride concentration that is necessary
for corrosion to initiate is referred to as the corrosion-initiation threshold.
The corrosion-initiation threshold has been studied extensively [ACI 222; Virmani, 2012; Mohammed and Hamada,
2006; Pradham, 2007, among others], and there has been a wide scatter in the reported results [Angst, 2011]. The
concentration value varies depending upon a number of factors, including whether it is acid-soluble or water-soluble
chlorides, if the chlorides are admixed or are entering mature concrete, the composition of the concrete and the
moisture content of the concrete, among other factors. Typical values range from 0.2 % to 0.4 % by mass of cement.
The amount of time it takes for chloride to reach the corrosion-initiation threshold is a significant portion of the
expected service life of the concrete structure. The time after corrosion initiates until damage occurs is relatively short
for conventional reinforcing steel in the typical conditions it is used. To predict the time to corrosion, there are a
number of service life modelling programs available [ACI 365; Ehlen et al., 2009; Samson and Marchand, 2007;
Concrete Society, 2004; Scheissl et al., 2004 among others]. They are all typically based on non-steady state diffusion
of chloride ions through the concrete pore structure as modelled by Fick’s Second Law:
(1)
Different models can vary in how they account for variations in concrete properties with time, whether they explicitly
or implicitly account for chloride binding, or if they use extended versions of the basic equation that accounts for
charge separations and ion-ion interaction. The service life model selected for a specific project will depend upon the
project objectives.
6.2
@Seismicisolation
@Seismicisolation
Impact on Anticipated Service Life of Chloride Thresholds
For the purposes of this study, the service life modelling program used was Life-365 [Ehlen et al, 2009]. Life-365 is
a freely available program that was developed under the auspices of an industry consortium. This study could be
performed with different service life programs, but the trend of the results would be similar. Diffusion is modelled in
Life-365 by a finite element application of Fick’s Second Law, using effective diffusion coefficient values that
implicitly include the influence of chloride binding. The development of diffusion coefficients with time is accounted
for by the equation:
28
(2)
where D(t) is the diffusion coefficient as a function of time, D28 is the diffusion value at 28 days, t is the time in days
and m is a parameter that depends on the cementitious materials used in the concrete.
This approach follows the work of a number of researchers [Bamforth, 1998; Thomas and Bamforth, 1999; Tang and
Nilsson, 1992; Mangat and Molloy, 1994; Maage et al, 1995]. Suggested values of these parameters are provided by
the program, but they can be adjusted by the user. External temperature history and surface concentrations are also
provided by the program for different locations.
Once the chloride concentration reaches the corrosion-initiation threshold, the program considers that corrosion has
begun, but this is not considered the end of the service life of the concrete structure. There is also a propagation period
considered, during which damage to the steel bars due to corrosion is occurring but the damage has not reached
sufficient magnitude to be visible or to require repair. The duration of this period varies depending upon the type of
steel used (black or epoxy coated) but is modelled by Life-365 as fixed duration independent of the concrete mixture.
For the purposes of this study, black steel was considered and the standard propagation period of 6 years suggested
by the program was used. The combination of the time-to-initiation and the propagation period is the time to first
repair and is the parameter of interest. Although Life-365 does model the remaining life of the structure after the
predicted time to first repair, this was not considered in this study.
Chloride Concentration Units

There are three common units that chloride concentrations, and corrosion-initiation threshold values are typically
expressed: (i) as the ratio of chloride concentration to hydroxide concentration in the pore solution ([Cl-]/[OH-]), (ii)
as the percentage by mass of cement content, or (iii) as percentage by mass of concrete.
The chloride:hydroxide ratio is frequently used in laboratory studies, where the concentration in the pore solution
around the bar of these two species can be directly measured. The corrosion-initiation threshold occurs when the steel
depassivates and its protective oxide layer is compromised [Broomfield, 1997]. In the presence of chlorides, a simple
model of depassivation is that the competition between chloride ions forming FeCl2 and hydroxide ions forming
Fe(OH)2 favors the production of ferrous chloride. Thus the relative concentration of these two ions is the parameter
of interest, although, as with all things that have to do with concrete, it is in reality more complicated than that simple
model. Although this has been successful in laboratory studies of corrosion-initiation threshold [ACI 222R-01; CEB,
1992], it is difficult to implement in practice. Predicting the hydroxide concentration at the level of the steel is not a
simple exercise, depends on a number of unknown parameters, and can vary with time. It is particularly difficult
during the schematic design phase of the structure prior to it being built.
As an alternative to using the chloride:hydroxide ratio for expressing chloride concentrations, the chloride as the mass
percent of the cement in the concrete is often used. The cement content becomes a proxy for the hydroxide
concentration as the majority of hydroxide in the concrete pore solution originates from the cement and is liberated
during the cementing reactions [Neville, 2002]. The cement contents is easily determined from the concrete mix
designs, if they are known. This is the approach that is used in the ACI building code [318-14]. This approach is
straight-forward to implement when the source of chlorides are internal, that is they come from the concrete mixture
components. It does have some limitations when the source of chlorides is external to the concrete, and they are
penetrating by diffusion, permeation, sorption or some other transport mechanism.
The final alternative is expressing the chloride concentration in terms of the mass percent of the concrete. This has
the advantage of being able to be directly measured for existing structures. It is also how concrete experiences
chlorides penetrating from the outside. Chloride-contaminated water penetrates the pore structure over the entire
6.3
@Seismicisolation
@Seismicisolation
Kyle Stanish
surface, and is thus related to the concrete as a whole, and not the cement content alone, although it will be affected
by the porosity of the concrete. It is believed that this measurement technique is more directly applicable for an
external source of chlorides, and it what is frequently used in service life modelling to characterize the external
environment.
SERVICE LIFE MODELLING
To evaluate the impact of corrosion-initiation threshold on predicted life, two typical structures that are exposed to
chloride ions were considered. In order to capture the two extremes of behavior within the continental United States,
two models were run as the base line. To represent cold weather climates with heavy use of deicing salts, a parking
structure in Chicago, Illinois was considered. To represent a warm climate exposed to salt water, a jetty or pier in
Miami, Florida was also evaluated. The effective diffusion coefficient is dependent on the temperature, accelerating
at higher temperature. This is modelled in Life-365 using an Arrhenius relationship, with an activation energy of 3500
J/mol [Ehlen et al, 2009]. The environmental exposure conditions for these two locations, as recommended by Life-
365, are shown in Table 1. These consist of the average monthly temperature (based on historic records) and the
chloride surface concentration by mass of concrete. The chloride surface concentration is assumed to start at 0 % and
build linearly to some maximum over a number of years. Both the maximum chloride surface concentration and the
number of years it takes to reach that maximum vary depending on location and structure type. Both structures were
modelled with a 2 in. (50 mm) cover over the reinforcing steel.
Table 1 – Environmental Conditions

Annual Temperature Profile [ºF (ºC)] Surface [Cl-]
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Max Time
Chicago 21 25 37 49 59 69 73 72 64 53 40 27 1.0 % 7.1 yrs
(-6) (-4) (3) (9) (15) (20) (23) (22) (18) (12) (4) (-3)
Miami 67 69 72 75 79 81 83 83 82 78 74 69 1.0 % 10 yrs
(20) (20) (22) (24) (26) (27) (28) (28) (28) (26) (23) (21)
In accordance with ACI 318-14, the corrosion-initiation threshold concentration is considered to be dependent on the
amount of cement in the concrete mixture. For a concrete with 760 lb/yd3 (450 kg/m3) of cement and a density of
3800 lb/yd3 (2255 kg/m3), a chloride concentration of 0.4 % by mass of cement translates into a threshold value of
0.08 % by mass of concrete, while 0.2 % by mass of cement is 0.04% by mass of concrete.
Influence of Corrosion-Initiation Threshold Concentration on Portland Cement Concrete

The impact of the corrosion-initiation threshold concentration is shown for a typical concrete that is made with 100 %
Portland cement and no supplementary cementing materials is shown in Figure 1. This is based on a concrete with a
w/cm ratio of 0.40. The diffusion parameters predicted by Life-365 for this concrete are a D28 = 1.23 x 10-8 in.2/s
(7.94 x 10-12 m2/s) and an m-value of 0.20.
Threshold Chicago Miami

0.08 % (conc) 13.3 yr 14.2 yr
0.06 % (conc) 12.5 yr 13.2 yr
0.05 % (conc) 12.0 yr 12.6 yr
0.04 % (conc) 11.5 yr 12.0 yr
Figure 1 – Effect of Threshold Concentration on

Time to First Repair, Typical Portland Cement
Concrete.
6.4
@Seismicisolation
@Seismicisolation
Influence of Corrosion-Initiation Threshold Concentration on a Fly Ash-containing Concrete

The impact of the corrosion-initiation threshold concentration is shown for a concrete that contains fly ash is shown
in Figure 2. This is based on a concrete with a w/cm ratio of 0.35 and a fly ash content of 25 %. The diffusion
parameters predicted by Life-365 for this concrete are a D28 = 1.10 x 10-8 in.2/s (7.10 x 10-12 m2/s) and an m-value of
0.40.

0.08 % (conc) 21.9 yr 23.1 yr
0.06 % (conc) 19.7 yr 20.8 yr
0.05 % (conc) 18.5 yr 19.5 yr
0.04 % (conc) 17.2 yr 18.2 yr

Time to First Repair, Fly Ash-containing
concrete.
Influence of Corrosion-Initiation Threshold Concentration on a Slag-containing Concrete

The impact of the corrosion-initiation threshold concentration is shown for a typical concrete that contains slag is
shown in Figure 3. This is based on a concrete with a w/cm ratio of 0.35 and a slag content of 50 %. The diffusion
parameters predicted by Life-365 for this concrete are a D28 = 9.34 x 10-9 in.2/s (6.02 x 10-12 m2/s) and an m-value of
0.49.

0.08 % (conc) 37.5 yr 38.0 yr
0.06 % (conc) 32.2 yr 32.9 yr
0.05 % (conc) 29.4 yr 30.3 yr
0.04 % (conc) 26.7 yr 27.6 yr

Time to First Repair, Slag-Containing Concrete.
DISCUSSION
The corrosion-initiation threshold concentration did affect the predicted time to first repair for all of the concrete
mixtures studied for both locations. Figure 4 and 5 display the time to first repair normalized by value for a threshold
concentration of 0.04 % by mass of concrete for both the Chicago and Miami structures. The impact of the threshold
concentration on time to first repair does not have a direct one-to-one correlation. Doubling the threshold
concentration did not double the time to first repair. This is due to the non-linear nature of diffusion, particularly at
the relatively low concentrations (compared to the surface concentration) that are of interest. There is a large portion
of the time between first exposure and the time the threshold concentration is reached that there are no chlorides at
the level of the steel. This time does not vary when different threshold values are considered.
6.5
@Seismicisolation
@Seismicisolation
Kyle Stanish
Figure 4 – Normalized Time to First Repair – Chicago. Figure 5 – Normalized Time to First Repair – Miami.
The impact of varying the threshold concentrations was more significant for the mixtures containing supplementary
cementing materials than for when there are no supplementary cementing materials, as can be seen in Figures 4 and
5. This increased relative impact of the corrosion-initiation threshold concentration for the modelled concrete mixtures
is due to the evolution of the diffusion value with time. With the concrete mixtures containing supplementary
cementing materials, the time required for the chloride concentration to reach the threshold value is longer. This
provides more time for the diffusion value to reduce, which has a synergistic effect. Supplementary cementing
materials have a larger m-value, or the rate at which the diffusion coefficient reduces with time than concrete mixture
containing only Portland cement. [Tang and Nilsson 1992; Mangat and Molly, 1994; Maage et al, 1995; Bamforth
1998; Thomas and Bamforth, 1999; Stanish and Thomas, 2003] Thus variations in assumed threshold concentrations
have a greater impact on the time to first repair for concretes containing fly ash or slag than they would with a plain
Portland cement mixture.
The appropriate threshold value for different concrete mixtures will vary, however. According to ACI 318 [2014],
the corrosion-initiation threshold concentration is depending only upon the cement content of the concrete mixture.
When a portion of the cement is replaced by supplementary cementing materials, the amount of chloride that is
permitted to be in the concrete as a whole is reduced. The use of 25 % fly ash, for example, would reduce the threshold
concentration by mass of concrete by 25 % as compared to an identical concrete mixture that is only contains Portland
cement. Supplementary cementing materials are not an inert substance. They do react and influence the composition
of the pore solution, and thus affect the corrosion-initiation threshold concentration [Fagerlund, 2011]. However, the
influence that they have will not be the same as if they were an inert, non-reacting material. Their contribution will
likely be different that the contribution of Portland cement and will likely vary between different supplementary
cementing materials. Determining the appropriate influence is important for predicting the expected time to first
corrosion of a concrete mixture. This is particularly important for concrete that has a long expected service life as is
typically the case when supplementary cementing materials are used.
CONCLUSIONS
The time to first repair for two different structures that would be typically exposed to chlorides were modelled with
three different typical concrete mixtures, varying the corrosion-initiation threshold using a freely available service life
program, Life-365. Based upon this investigation, it could be seen that:
1. The time to first repair does not have a one-to-one correlation with the assumed chloride-initiation threshold
value. Doubling the corrosion-initiation threshold value does not double the time to first repair.
2. The variation of the time to first repair with chloride initiation threshold values is more significant for
concrete that has a longer expected service life. This is attributable to the greater decrease in chloride
diffusion value that occurs with longer time spans of interest.
3. The synergistic effect of longer time spans allowing reduced chloride diffusion coefficients is more
pronounced when the concrete mixture contains supplementary cementing materials such as fly ash and slag.
This is due to their greater rate of reduction of diffusion coefficient with time as compared to a Portland
cement mixture.
6.6
@Seismicisolation
@Seismicisolation
REFERENCES
ACI 222R-01 – Protection of Metals in Concrete against Corrosion

ACI 318-14 – Building Code Requirements for Structure Concrete
Angst, U., 2011, Chloride Induced Reinforcement Corrosion in Concrete: Concept of Critical Chloride Content -
Methods and Mechanisms, Thesis for the degree of Philosophiae Doctor, Norwegian University of Science and
Technology, Trondheim, Norway, 73 pp.
Bamforth, P.B., 1998, “Spreadsheet Model for Reinforcement Corrosion in Structures Exposed in Chlorides,”
Concrete Under Severe Conditions 2, O.E. Gjørv, K. Sakai, and N. Banthia, ed., E&FN Spon, London, pp. 64-75.
Broomfield, J.P., 1997, Corrosion of Steel in Concrete: Understanding, Investigation and Repair, E&FN Spon, New
York, 240 pp.
CEB Design Guide for Durable Concrete Structure, 2nd Edition, 1992, Thomas Telford Publishers.
Concrete Society, The, 2004, “Enhancing Reinforced Concrete Durability, P.M. Bamforth,” 208 pp.
Ehlen, M.A., Thomas, M.D.A., and Bentz, E.C., 2009 “Life-365 Service Life Prediction ModelTM Version 2.0,”
Concrete International, V. 31, No. 5, May, pp. 41-46.
Fagerlund, G., 2011, The Threshold Chloride Level for Initiation of Reinforcement Corrosion in Concrete: Some
Theoretical Considerations, Report TVBM-3159, Lund, 47 pp.
Maage, M., Helland, S., and Carlsen, J.E., 1995, “Practical Non-Steady State Chloride Transport as a Part of a Model
for Predicting the Initiation Period,” Chloride Penetration into Concrete, L.-O. Nilsson and J. Olliver, ed., pp 338-
46.
Mangat, P.S., and Molloy, B.T., 1994, “Prediction of Long Term Chloride Concentrations in Concrete,” Materials
and Structures, Vol. 27, pp. 338-46.
Mohammed, T.U., and Hamada, H., 2006, “Corrosion of Steel Bars in Concrete with Various Steel Surface
Conditions,” ACI Materials Journal, V. 103, No. 4, pp. 233-42.
Neville, A.M., 2002, Properties of Concrete, 4th Edition, 844 pp.
Pradham, B., and Bhattacharjee, B, 2007, “Role of Steel and Cement Type on Chloride-Induced Corrosion in
Concrete,” ACI Materials Journal, Vol. 104, No. 6, pp. 612-19.
Samson E., and Marchand J., 2007, “Modeling the effect of temperature on ionic transport in cementitious materials”,
Cement and Concrete Research, V. 37, p. 455-468.
Schiessl, P., Gehlen, C., and Kapteina, G., 2004, “Assessment and Service Life Updating of Existing Tunnels,” Safe &
Reliable Tunnels - Innovative European Achievements, First International Symposium, Prague, Austria.
Stanish, K., and Thomas, M.D.A., 2003, “The Use of Bulk-Diffusion Tests to Establish Time-Dependent Chloride
Diffusion Coefficients,” Cement and Concrete Research, Vol 33, pp. 55-62.
Tang, L. and Nilsson, L.-O., 1992, “Chloride Diffusivity in High Strength Concrete at Early Ages,” Nordic Concrete
Research, pp. 162-171.
Thomas, M.D.A., and Bamforth, P.B., 1999, “Modelling Chloride Diffusion in Concrete: Effect of Fly Ash and Slag,”
Cement and Concrete Research, Vol 29, pp. 487-95.
Virmani, P., 2012, Literature Review of Chloride Threshold Values for Grouted Post-tensioned Tendons, FHWA
Publication No. FHWA-HRT-12-067, 16 pp.
6.7
@Seismicisolation
@Seismicisolation
Kyle Stanish
6.8
@Seismicisolation
@Seismicisolation
SP-308—7
THE INFLUENCE OF SCM TYPE AND QUANTITY ON THE CRITICAL CHLORIDE

CORROSION THRESHOLD
David Trejo and Cody Tibbits
Synopsis: Chloride–induced corrosion of embedded metals in reinforced concrete structures is dependent on the
quantity of chlorides in the concrete material. Because of this, most ACI documents limit the amount of chlorides in
the concrete for new reinforced concrete structures. Most documents in the United States limit the chlorides as a
function of cement content, generally accepted to be the portland cement content. Significant changes have occurred
in the cement and concrete industries, such as performance based specifications for cement (e.g. ASTM C1157) and
the ubiquitous use of SCMs. The allowable chloride limits published by ACI Committee 222 specifically states that
the allowable limits are based on portland cement content. ACI 318 specifies limits based on weight of cement but is
not clear on what determines cement (portland cement only or all cementitious materials?). At issue is whether the
allowable chloride limits reported by ACI and other organizations are applicable for mixtures containing SCMs and
whether these limits should be based on portland cement content or total cementitious materials (cement + SCM)
content. Significant research has been performed to assess the influence of SCMs on transport properties but limited
research has been performed on assessing the influence of SCMs on the critical chloride corrosion threshold of steel
in concrete. This paper reports on research investigating the influence of SCMs replacement types and quantities on
the diffusivity and critical chloride corrosion threshold. Mortar mixtures containing fly ash had three levels of fly ash
(0, 20, and 40%) and mortar mixtures containing slag had three levels of slag (0, 30, and 60%). The reinforcing steel
for all specimens was conventional steel reinforcement meeting ASTM A615 specifications.
Results indicate that the diffusivity of the mixtures containing SCMs was significantly lower than the mixture with
portland cement as the only binder. The results also indicate that the critical chloride threshold levels for mixtures
containing SCMs is significantly lower than the critical chloride threshold of mixtures containing only portland cement
as the binder. An analysis of time to corrosion indicates that although the mean time to corrosion can be lower for
systems containing SCMs, t-tests indicate that the null hypothesis (that the means of the time to corrosion of the
systems are equal) cannot be rejected. This indicates that although there is a significant reduction (positive influence)
in diffusivity when using SCMs, there is also a significant reduction (negative influence) in critical chloride threshold.
The benefits of the reduction in the rate of diffusivity seems to be offset by the detrimental effects of SCMs on critical
chloride threshold. Limits on allowable chlorides in concrete for new construction published by ACI 222 were based
on concretes containing only portland cement as the binder. This research indicates that the ACI published limits may
not be applicable to systems containing SCMs and that the limits are less conservative when basing the limits on total
cementitious materials content.
Keywords: Corrosion; Chloride transport; Diffusion coefficient; Diffusivity; Critical chloride corrosion threshold;
Chloride; Fly ash; Slag; Portland cement
7.1
@Seismicisolation
@Seismicisolation
The Influence of SCM Type and Quantity on the Critical Chloride Corrosion Threshold
FACI David Trejo is Professor and Hal Pritchett Endowed Chair in the School of Civil and Construction Engineering
at Oregon State University. He is Chair of ACI committee 222 and a member of committees 201, Durability; 236,
Materials Science of Concrete; and 365, Service Life. His research interests include construction, durability, and
performance issues associated with reinforced concrete systems, corrosion of steel in cementitious materials, service-
life analyses, innovative concrete materials for improved construction, and modeling deterioration mechanisms.
ACI Student member, Cody Tibbits, is an undergraduate student in the School of Civil and Construction Engineering
at Oregon State University. He will begin his graduate studies in Infrastructure Materials and Structures in January
2016, where his research interests will focus on time-variant performance of materials and structural systems.
INTRODUCTION
The detrimental influence of chlorides on the corrosion of steel reinforcement in concrete has been well documented.
Although well documented, there are still several challenges that have not been addressed in the scientific community.
A recent concern is whether allowable chloride limits published by ACI (222R and 318) should be based on portland
cement content or the overall cementitious materials content. ACI 222R specifically notes that “Based on the present
state of knowledge, the chloride limits in Table 3.1 [this is the table the list allowable chloride limits] for concrete
used in new construction, expressed as a percentage by mass of portland cement [bold and italics added for emphasis],
are recommended to minimize the risk of chloride-induced corrosion.” ACI 318 is silent on the whether the allowable
chloride limits are based on portland cement content or overall cementitious materials content (portland
cement+supplementary cementing materials [SCMs]). The ACI 222R document indicates that allowable chlorides
limits are based on studies from concrete with portland cement as the binder (i.e., no SCMs). Yet as the concrete
industry strives to become greener and more environmentally responsible, the increased use of SCMs is expected.
However, constructing reinforced concrete products that experience early deterioration and reduced service lives
provides limited environmental benefits. Two issues need investigation: first, are the limits published by ACI 222 and
318 appropriate for systems containing SCMs, and second, if appropriate for systems with SCMs should these
allowable chloride limits be based on portland cement content only or overall cementitious materials content (portland
cement+SCMs).
The allowable chloride limits published by ACI are chloride limits for new, unexposed concrete. These limits are
published to minimize the risk of corrosion but with the understanding that some constituent materials (e.g.,
aggregates, cement, SCMs, chemical admixtures, water) could likely contain chlorides. A concrete with zero chlorides
would minimize the risk of chloride-induced corrosion but would also likely be uneconomical in many cases. These
allowable chloride limits are published considering both risk of corrosion and economics and were mainly based on
studies funded by the Federal Highway Administration (FHWA). These studies focused on identifying how much
chlorides are required to initiate corrosion (i.e., CT) and reported CT values of about 0.2% by mass of portland cement
(tested using acid soluble testing). ACI 222R limits the allowable chlorides for new construction exposed to wet
conditions to half of the 0.2% value reported in the FHWA studies. Limiting the allowable chlorides for new
construction as a percentage of CT is a reasonable approach to minimizing the risk of corrosion. However, it is
unknown whether these limits are applicable to systems containing SCMs. To better understand allowable chloride
limits for mixtures containing SCM, a review of the history of chlorides and corrosion in concrete and how the ACI
documents have been developed would be beneficial. This is provided next.
THE HISTORY OF ADMIXED CHLORIDES AND CODE/GUIDE DEVELOPMENT
Nearly 100 years ago Abrams (1924) reported on the potential detrimental influence of chlorides on corrosion of
reinforcing steel embedded in concrete. Although Abrams warned against the potential consequences of chlorides in
concrete, the codes were relatively slow in adopting chloride limits in concrete. ACI 318-47 (1949) made no reference
to limiting chlorides in concrete and in fact, ACI 604-48―Recommended Practice for Winter Concreting Methods
(1949) allowed the use “of small amounts of additional cement or accelerators such as calcium chloride to accelerate
the hardening of the concrete at low temperatures (above 32 F) …” The 318-47 requirement for water noted that water
“shall be clean, and free from injurious amounts of oils, acids, alkalis, organic materials, or other deleterious
substances,” but had no specific mention of salts. Sixteen years later, in 1963, ACI 318 modified the requirement for
water to the following: “water used for mixing concrete shall be clean and free from injurious amounts of oils, acids,
7.2
@Seismicisolation
@Seismicisolation
alkalis, salts (emphasis added), organic materials, or other substances that may be deleterious to concrete or steel.”
The commentary for ACI 318-63 specifically noted that for prestressed structures, “calcium chloride or an admixture
containing calcium chloride shall not be used” for concrete or grout. The document also specified that “seawater shall
not be used” for prestressed concrete. Interestingly, no limits on the use of seawater in reinforced concrete was
specified.
ACI 604-56 (1964) still allowed “calcium chloride to accelerate hardening of the concrete in cold weather” allowing
up to 2% calcium chloride, except when sulfate-resisting concrete was required. Sulfate-resisting cements have lower
C3A contents and Hope and Ip (1987) reported that cements and SCMs that contain higher C3A contents have higher
chloride binding capacities, thus reducing the potential for corrosion. The 604-56 document also had a footnote added
to the section on accelerators. The footnote noted that “another exception to use calcium chloride may be when there
is reason to suspect that steel will be corroded by stray currents. In this event, an extra sack of cement should be used
in preference to calcium chloride.” The footnote went on to discuss results from the U.S. Bureau of Standards
Technical Paper No. 18 (1913) reported in the Highway Research Board Bibliography No. 13, where “calcium
chloride, even to the small extent of 0.03 percent, has been found to accentuate the effect of stray electric currents in
causing corrosion of iron.” The footnote also noted that “calcium chloride should not be used in prestressed concrete
because of possible stress corrosion of the highly stressed special wire steel, but ordinary reinforcement is not
affected.” The cause and effect of chloride and corrosion of conventional reinforcement had not yet made it into the
code in 1964. ACI 604-56 also noted that 1 percent calcium chloride was the “minimum requirements for jobs taking
maximum risk.”
The potential issues associated with chlorides and their influence on corrosion of steel in concrete had been recognized,
although the degree of influence was not yet clear. ACI Committees were struggling with how or whether limits on
admixed chlorides were needed for conventionally reinforced structures. The issue with limits on chlorides in
prestressed structures was clear as several failures had occurred with prestressed structures that had chlorides admixed
into the concrete. Because of this, ACI 201 (1970) reported that “calcium chloride and other soluble chlorides should
not be used as admixtures in prestressed concrete.” The document also discussed the potential challenges of adding
chlorides to concrete for structures exposed to marine conditions or conditions where “water-soluble salts” were
present. Although caution was expressed, no limits on chlorides were presented for conventionally reinforced concrete
structures. However, ACI 201 (1970) did place limits on the water-cement ratio and these limits were based on
exposure condition and member type. For thin sections exposed to seawater, a maximum water-cement ratio of 0.4
(4.5 gallons per sack of cement)―for moderate and thick sections the water-cement ratio could be increased to 0.45
(5.0 gallons per sack). Recommended water-cement ratios ranged from 0.4 to 0.53 (4.5 to 6 gallons per sack).
ACI 318-71 underwent significant changes from the ACI 318-64 document. The later document prohibited the use of
calcium chloride in both grout (for bonded tendons) and concrete and noted that “attention is called here to the possible
adverse effects of excessive chloride ions in the presence of aluminum and in prestressed concrete.” The commentary
also noted that “a new provision has been added concerning chloride ion content of water (including the portion of the
mixing water contributed from the free moisture of aggregates) to be used in prestressed concrete or in concrete with
aluminum embedment.” The commentary went on to note that “no numerical quantities are stipulated” but “it is
suggested that chloride ion concentrations exceeding 400 or 500 ppm might be considered dangerous…” The
commentary also noted that ACI 222 recommended levels well below these values and recommended that chlorides
contained in the aggregate and admixtures should be considered in evaluating the acceptability of total chloride content
in the mixing water.
ACI 318-77 included only minor revisions to the chloride limits established in ACI 318-71. Interestingly, the
commentary noted that “high (emphasis added) concentrations of chloride ion may be harmful in prestressed concrete,
because of possible corrosion, and in concrete containing aluminum…” The suggested levels of 400 to 500 ppm were
replaced by “high.” Clearly, the Committees were still struggling with the issue of chloride limits and corrosion.
ACI 201.2R-77 was one of the first ACI documents to detail challenges associated with chlorides and specifically
addressed the issue of limiting admixed chlorides in the concrete mixture. The document noted the potential hazards
of chlorides but noted that “specifying a zero chloride content for the mix, however, is impossible to realize in
practice.” The 1977 document noted that “the threshold value for a chloride content in concrete necessary for the
corrosion of embedded steel can be as low as 0.15 percent by weight of cement” but cautioned that this universal limit
did not consider availability of chlorides, oxygen, and moisture. The 201.2R-77 document referenced three references
7.3
@Seismicisolation
@Seismicisolation
to justify this “threshold value.” Two references were from studies on bridge decks and one was from a reference on
general corrosion of steel in concrete. The 201.2R-77 document also provided a discussion regarding “soluble” and
“combined” chlorides, challenges with different tests methods, and how these could potentially influence corrosion.
More specifically, the 201.2R-77 document provided specific limits for chloride in concrete prior to service exposure
based on free and combined chlorides. These limits were based on weight percent of cement and were reported as 0.06
for prestressed concrete, 0.10 for reinforced concrete in moist environments and exposed to chlorides in service, 0.15
for reinforced concrete in a moist environment but not exposed to chlorides, and no limits for above ground building
construction that would stay dry. The document noted that these limits were in agreement with ACI Committee 222,
even though the 222 Committee would not publish its first document for another eight years. The 201.2R-77 document
also noted that calcium chloride should not be “intentionally added to the mix in prestressed concrete or conventionally
reinforced concrete which will be exposed to moisture and chlorides in service.” This was a significant change from
earlier ACI documents as now the influence of chlorides on conventionally reinforced concrete was recognized.
ACI 318-83 revised the requirements for chlorides to state that “calcium chloride or admixtures containing chloride
from other than impurities from admixture ingredients shall not be used in prestressed concrete, in concrete containing
embedded aluminum, or in concrete cast against stay-in-place galvanized metal forms.” This did not agree with the
1977 201.2R document that limited calcium chloride use to conventionally reinforced structures exposed to a dry
environment and to moist environments containing no chlorides. The 1983 318 code also published specific limits
(based on weight of portland cement) on the chlorides based on structure type or exposure conditions―this was the
beginning of a long dispute. The 1983 318 code specifically noted that “for corrosion protection, maximum water
soluble chloride ion concentrations in hardened concrete at an age of 28 days contributed from the ingredients” shall
not exceed a specific weight percent of the cement. The published limits were 0.06 for prestressed concrete, 0.15 for
reinforced concrete exposed to chlorides in service, 1.00 for reinforced concrete that will remain dry in service, and
0.30 for other reinforced concrete. The 1977 201 guide and the 1983 318 Code agreed on the 0.06 limit for prestressed
concrete. However, the 1977 201 Guide recommended lower limits than the 1983 318 Code for reinforced concrete
exposed to chlorides in service (0.10 versus 0.15) and for reinforced concrete exposed to moisture (0.15 versus 0.30).
The 1983 318 Code was more conservative on chloride limits for dry conditions, requiring 1% maximum by weight
of cement versus no limit for the 1977 201 Guide. The 1983 318 Code also placed a maximum water-cement ratio
(0.40), a minimum strength (4750 psi [33 MPa]), and a minimum cover for corrosion protection of reinforced concrete
exposed to deicing salts, brackish water, seawater or spray from these sources. Although progress was being made on
limiting chlorides in new concrete, variations in limits were leading to confusion.
In 1985 ACI Committee 222 published its first document on corrosion of metals in concrete and noted in the document
that “placing limits on the allowable amounts of chloride ion in concrete is an issue still under active debate.” The
document suggested acid soluble limits based on weight of portland cement of 0.08 for prestressed concrete and 0.20
for reinforced concrete. The Committee opted to specify chloride limits based on acid soluble testing because it was
believed that acid soluble testing provided more consistent results than water soluble testing and because carbonation
could lead to the release of the chlorides (Tutti 1982). In general, the 0.08 limit was considered to be similar to the
0.06 limits reported by Committees 201 and 318 as water-soluble chloride contents were widely reported to be about
75 to 85% of acid soluble chloride contents (Hope and Poland 1987, Enevoldsen et al. 1994). The 222 chloride limits
reported for reinforced concrete were similar to those reported by 201 (75% of 0.20 is 0.15 and 0.15 was the published
limit in the 201 document). The 1985 222 document noted that “there are some exposure conditions where the chloride
levels may exceed the recommended values and corrosion will not occur.” These conditions included submerged
conditions where oxygen would not be available and cases where “concrete is continuously dry.” However, no limits
were suggested for these cases. The 1985 222 document noted that its chloride limits differed from the values reported
in the documents by Committees 201 and 318 and noted that “committee 222 has taken a more conservative approach
because of the serious consequences of corrosion, the conflicting data on corrosion threshold values, and the difficulty
in defining the service environment throughout the life of a structure.” In fact, the 222 limits could be considered to
be less conservative than the limits published in the 201 document, but more conservative than the limits published in
the 318 document.
ACI 318-86 attempted to provide some clarification on the admixed chloride limits in concrete. First, the 1986
document specified testing for chlorides be performed on the hardened concrete of ages between 28 and 42 days (ACI
318-83 required testing at an age of 28 days). This document also defined that testing conform to test procedures
described in Federal Highway Administration Report No. FHWA-RD-77-85, Sampling and Testing for Chloride Ion
in Concrete (Clear and Harrigan 1977). The commentary for ACI 318-86 also noted that the chloride limits were
7.4
@Seismicisolation
@Seismicisolation
different than those published in ACI 222 and ACI 201. Error! Reference source not found. shows a comparison of
the limits published by the different ACI documents in 1986. The 318 code tended to take a less conservative approach
to limiting chlorides than recommendations from Committees 201 and 222.
Table 1 — Chloride limits for concrete as published by different ACI Committees (Trejo & Weyers 2012).
Max Chloride (% weight of cement) Allowed in New Concrete as Published in
Exposure/Structure Type 1986 ACI Committee Documents
201 (water-soluble) 222 (acid-soluble) 318 (water-soluble)
Prestressed Concrete 0.06 0.08 0.06
Reinforced Concrete Exposed
0.10 0.15
to Chlorides in Service
0.20
Reinforced Concrete Exposed
0.15 0.3
to Damp Exposure in Service
Reinforced Concrete that will
1.0 No limit 1.0
Remain Dry in Service
The 1995 and 2002 318 Codes required the same limits published in Table 1. One significant difference between the
ACI 318-86 and 318-95 codes were that allowable water-cement ratio (w/c) limits were changed to allowable water-
cementitious materials ratio (w/cm) limits. Although all cementitious materials could be used to conform to the 318
limits, the chloride limits were still based on cement content, not cementitious materials content. The limits published
in the 1996, 2001, and 2011 222 documents specifically noted that limits were “expressed as a percentage by weight
of portland cement.” The 222 and 318 documents seemed to agree that the amount of admixed chlorides should be
based on portland cement content and not total cementitious materials content, although this was not clearly stated.
This was interesting as significant efforts were underway to make portland cement concrete more environmentally
friendly by replacing the portland cement, the component that made a significant CO2 contribution during production,
with SCMs.
In 2008, ACI 318 published a significant revision to the chloride limits. The document defined exposure classes
for different exposure conditions: class C0 consisted of concrete that is dry or protected from moisture; class C1
(moderate exposure) consisted of concrete exposed to moisture but not to external sources of chlorides, and; class
C2 (severe exposure) included concrete exposed to moisture and an external source of chlorides. Class C0 was to be
assigned when exposure conditions do not require additional protection against the initiation of corrosion; C1 and
C2 are assigned when additional protection is required, making the degree of protection dependent on the severity of
the exposure condition. The 318-08 document specified maximum w/cm, minimum design strength ( ), minimum
cover, and maximum chloride content for new construction. The limits from 318-08 are shown in
2. These limits are essentially the same limits from the earliest 318 document, but with defined exposure conditions.
Although the 318-08 document specifies a maximum w/cm, the document limits the water soluble chlorides based on
“weight of cement,” which results in confusion in the industry. ACI CT13 (2013) defines cement as “any of a number
of materials that are capable of binding aggregate particles together,” and defines cementitious materials as “pozzolans
and hydraulic cements.” The term “by weight of cement” is vague and is likely limiting the use of SCMs and more
environmentally friendly concrete systems.
A recent discussion in the 222 Committee meeting addressed whether the term “cement” in the 201 and 318 documents
meant “portland cement” or “portland cement plus supplementary cementitious materials.” The Committee reached
no agreement and concluded that the documents require clarification. The Committee also noted that although there
is significant literature on how SCMs can reduce the chloride transport characteristics of concrete mixtures, there is
limited research results in the literature on how SCMs influence the critical chloride threshold value, CT. The
Committee concluded that the chloride limits in the 222 were intentionally expressed as a percentage by weight of
portland cement due to the lack of research results in the literature on critical chloride threshold values of systems
containing SCMs.
7.5
@Seismicisolation
@Seismicisolation
Table 2 — Chloride limits published in ACI 318-08.

Maximum water-soluble chloride
Minimum specified strength, content in concrete, percent by weight of
Exposure cement
Maximum w/cm ′
Class , psi [MPa]
Reinforced Prestressed
Concrete Concrete
C0 N/A 2500 (17) 1.00 0.06
C1 N/A 2500 (17) 0.30 0.06
C2 0.40 5000 (34) 0.15 0.06
Two recent papers by Alonso and Sanchez (2009) and Angst et al. (2009) reported on variability and factors
influencing the amounts of chlorides in concrete required to initiate corrosion (i.e., CT) of the reinforcing steel. Both
papers reported that the CT values published in the literature varied significantly. The authors also reported that factors
influencing CT included concrete cement type and content, degree of cement hydration, environmental temperature,
reinforcement type, electrical potential of the steel surface, the presence of air voids and cracks at the concrete-
reinforcement interface, moisture content of the concrete, oxygen content of the concrete pore water at the
reinforcement-concrete interface, concentration of hydroxyl ions in the concrete pore solution at the reinforcement
surface, and the compound associated with the chloride. Of these variables, the authors reported that the condition of
the reinforcement-concrete interface, the concrete pore solution pH (hydroxyl ion content), and the steel reinforcing
bar surface potential were factors that significantly influence the CT. The large number of influencing variables can
result in significant variations in reported CT values.
Although the literature indicates large variations in reported CT values, ACI reports a single CT value for specific
exposure conditions. The ACI 222 indicates that this limit is based on portland cement content (i.e., mixtures
containing no SCMs) because most of the early literature included only results of steel reinforcement embedded in
concrete containing portland cement only. Yet there is a push to make concrete more environmentally friendly and
the use of SCMs can reduce the portland cement content, which in turn reduces the environmental impact of the
concrete (mainly from the reduction of CO2 emissions during cement production). Frankly, the use of SCMs is and
should be encouraged in the concrete industry. It has been well-established that binary or ternary cement systems
(portland cement+SCMs) can reduce the transport rate of chlorides in concrete, and therefore could be beneficial for
corrosion resistance (Byfors 1987, Jones et al. 1997, Thomas and Bamforth 1999, Papdakis 2000, Boddy et al. 2001,
Bouteiller et al. 2012, Andrade and Bujak 2013). However, it has also been reported that SCMs reduce the pH of the
concrete pore solution (Byfors 1987, Hussain and Rasheeduzzafar 1994), a factor identified by Alonso and Sanchez
(2009) and Angst et al. (2009) that influences CT. Lower hydroxyl ion contents can result in lower CT values which
could make concrete systems containing SCMs more susceptible to early corrosion.
By limiting the allowable chlorides for new construction as a function of portland cement content only creates specific
challenges. In the US, cement is specified using ASTM standards. Portland cement-based systems can be specified
using ASTM C150, Standard Specification for Portland Cement, C595, Standard Specification for Blended Hydraulic
Cements, or C1157, Standard Performance Specification for Hydraulic Cement. ASTM C150 provides standard and
optional minimum and maximum requirements for composition and physical characteristics of cement Types 1-V,
with and without air entrainment. ASTM C595 provides specifications for binary and ternary blended cements (slag,
pozzolan, and limestone). Although section 4 of the specification provides a naming convention for these different
blended cements based on the weights of the portland cement and SCMs, the standard also allows for simplified
naming conventions. These simplified naming conventions only require that the actual maximum weight percent of
all non-portland cement content be listed. If the requirement for allowable chlorides were 1% of the portland cement
weight content and a cement meeting C595 was designated as having less than 60% slag and pozzolan, the maximum
allowable chloride could only be estimated (for this case it would have to be assumed that 40 percent of the overall
weight was portland cement, although it could be significantly higher). However, the largest challenge associated with
allowable chloride limits arises from ASTM C1157. This standard is a performance standard and specifies cement
types by performance needs (e.g., high-early strength, sulfate resistance). This standard does not require that portland
cement content be reported. If the amount of portland cement content is unknown and the allowable admixed chloride
content is dependent only on the portland cement content, engineers and contractors cannot determine whether the
concrete product being placed meets the project requirements. This is a challenge when ACI documents require
maximum allowable chlorides in new construction to be based on mass of portland cement content.
7.6
@Seismicisolation
@Seismicisolation
The published limits on allowable admixed chlorides, that is chlorides that are usually from the constituent materials,
are vague and confusing for the concrete industry. Allowable chlorides should be a percentage of the CT, which now
could be determined based on the mass of the portland cement content or the mass of the total cementitious materials
content (portland cement+SCMs), depending on which ACI document is being referenced. It is clear in ACI 222 that
these allowable limits were based on studies from concrete containing only portland cement as the binder. ACI
documents and documents from other organizations are contradictory. To shed light on the influence of SCMs and
allowable admixed chlorides for new construction, this paper will first provide a review of the effects of SCMs on
chloride transport rate and pore solution pH and will then present experimental results from a research study
investigating the influence of SCMs on critical chloride threshold levels for systems containing SCMs. The research
reported in this paper specifically addresses the influence of transport rate and critical chloride threshold for portland
cement systems containing different levels of fly ash (0, 20, and 40%) and different levels of slag (0, 30, and 60%).
These results, with the results from the accompanying papers in this special publication, should provide additional
information to initiate discussions such that reasonable, engineering-based decisions can be made on whether
allowable chlorides limits are applicable to systems containing SCMs and whether these limits should be based on
amount of portland cement content or overall cementitious materials content (portland cement plus SCMs).
EFFECTS OF SCMs ON CONCRETE CHARACTERISTICS
Concrete is the most widely used man-made material in the world (Worrel et al. 2003, Mehta and Monteiro 2006).
The production of portland cement, a major constituent of concrete, generates significant amounts of greenhouse gases
(GHG) and has been reported to contribute 5 to 7% of the total man-made GHG emissions on earth (Mehta and
Monteiro 2006). Bonden and Blasing (2010) estimated that the cement industry generated about 33 billion tons of
GHG in 2010. Although the cement and concrete industries are becoming more sustainable, the GHG emissions from
cement production are significant and more efforts for improved sustainability are needed. The replacement of
portland cement with SCMs in concrete reduces the amount of portland cement in the concrete mixture, thereby
reducing GHG emissions (e.g., a 20% replacement could reduce the portland cement content by 20% and up to 20%
of GHG from the production process). In addition, most SCMs are by-product and waste materials and using these
materials in concrete reduces waste and landfilling. Although seemingly environmentally beneficial, if the SCM
reduces the durability and service life of a concrete structure, these benefits could be lost. It is important to understand
the influence of SCMs on the characteristics that influence time to corrosion and service life.
This research is investigating the effects of SCMs on the transport characteristics and critical chloride threshold of
steel reinforcement embedded in cementitious materials. The time to corrosion of steel reinforcement embedded in
cementitious materials subjected to chlorides is dependent on the rate at which the chlorides are transported through
the materials, the distance the chlorides need to travel (i.e., the cover depth), and the amount of chlorides needed to
initiate corrosion of the reinforcement. The amount of chlorides required to initiate corrosion is dependent on the steel
reinforcement type, the pH of the pore solution at the steel reinforcement-concrete interface, and the physical condition
of the interface. Significant research has been performed to assess the influence of SCMs on the transport rates of
chlorides in these materials. A brief review follows. However, limited research has been performed on how SCMs
influence the pore solution pH and critical chloride threshold of steel-concrete systems. A brief review of the literature
on the influence of SCMs on pore solution pH and critical chloride threshold, although limited, also follows.
Significant research has been performed to assess the influence of SCMs on chloride transport rate. Jones et al. (1997)
reported that the chloride resistance was increased (decreased diffusivity) for concretes containing ternary blended
binders. The authors also reported that the chloride resistance for the ternary binder systems was higher than the than
binary systems. Li et al (1999) also reported significant improvements in chloride resistance of ternary systems (fly
ash and silica fume). Thomas and Bamforth (1999) reported that fly ash and slag have little impact on the early-age
transport properties but at later ages concrete with these SCMs exhibit significantly lower chloride transport rates.
Thomas et al. (1999) also reported that ternary systems (silica fume and fly ash) exhibit a “very high” resistance to
chloride penetration. Papadakis (2000) reported that when SCMs are used as both a replacement for aggregate and as
a replacement for portland cement, the concrete exhibits significantly lower chloride levels (i.e., lower apparent
diffusion coefficients). Leng et al. (2000) also reported that the apparent diffusion coefficient of concrete containing
fly ash or slag is reduced when compared with systems without the SCMs. Bouteiller et al. (2012) reported that
concrete containing granulated blast furnace slag (GGBS) exhibited a delay in chloride ingress. Burris and Riding
(2014) reported that ternary systems exhibit reduced diffusivity and, similar to the findings of Thomas et al. (1999),
7.7
@Seismicisolation
@Seismicisolation
ternary systems can exhibit better resistance to chloride ingress than binary systems. There is significant information
in the literature indicating that the inclusion of SCMs in concrete systems can reduce the chloride transport, possibly
increasing the service-life of the system.
However, the rate at which chlorides are transported through the concrete cover is only one variable that influences
the time to corrosion and service life of reinforced concrete structures. Jones et al. (1997) reported that the chloride
ingress is reduced for mixtures containing SCMs but the rate of carbonation is increased for these systems. The
researchers reported that the degree of increase was dependent on the amount of SCM in the mixture. Andrade and
Buják (2013) also reported that binary systems can decrease the chloride transport rate yet the rate of carbonation is
increased. The authors attributed this increase in carbonation rate to the lower portlandite content. Work by Byfors
(1987) also reported significant reductions in chloride transport rates but also reported that SCMs (silica fume and fly
ash) reduce the pore solution pH, which the author reported “has a negative effect on the capacity of the cement paste
to resist reinforcement corrosion.” In addition to the lower pore solution pH and the susceptibility to carbonation,
Bouteiller et al. (2012) showed that specimens containing GGBS exhibited highly negative half-cell potential values
(indicating active corrosion) and higher current densities than specimens containing only portland cement at early
ages. Although the literature is clear about the benefits of SCMS in reducing chloride transport, the literature also
indicates that SCMs can reduce the pore solution pH, which does influence the time to corrosion. Additional research
is needed to assess the influence of SCMs on pore solution pH and more importantly on critical chloride threshold.
EXPERIMENTAL PROGRAM
The objective of this research is to assess both the chloride transport characteristics and the critical chloride threshold
level of mortar mixtures containing different replacement levels of fly ash and slag. Testing included assessing the
pore solution chemistry, apparent diffusivity (at 28 and 56 days and at the time of corrosion initiation), corrosion
activity, and the amount of chlorides required to initiate corrosion (the critical chloride threshold, CT). Table 3 shows
the mixture identifications and proportions assessed in the research. A control (C) mixture and two mixtures containing
fly ash (F) and two mixtures containing slag (S) were evaluated. The replacement levels for the fly ash mixtures were
20 and 40% and the replacement levels for the slag were 30 and 60%. Mixtures are designated first by the SCM type
(or C for control) following by the percent replacement. The chemical composition of the cement, fly ash, and slag
are shown in Table 4. The sand has a specific gravity in the saturated surface dry (SSD) state of 2.59 and a fineness
modulus of 3.1. The steel reinforcement met ASTM A615 specifications and the chemical composition is shown in
Table 5.
Mixing of the mortar followed the requirements in ASTM C305, Standard Practice for Mechanical Mixing of
Hydraulic Cement Pastes and Mortars of Plastic Consistency. After mixing, the mortar was paced into cylinder molds
in three lifts; each lift being consolidated using a metal rod. A schematic of the mold is shown in Figure 1. The
specimens were cured in a 100% humidity curing room at a constant temperature of 72 oF (23 oC) until testing. Note
that the specimens used to assess the apparent diffusivity were cured for 28 and 56 days and then exposed to a chloride
solution for 56 days. Table 6 shows the overall testing plan. Pore extraction for the chemical analysis of the pore
solution and pH measurements followed the method reported by Cyr et al. (2008). Inductively coupled plasma - optical
emission spectroscopy (ICP-OES) was used to assess the sodium (Na), calcium (Ca), potassium (K), silicon (Si), and
Aluminum (Al) in the pore solutions at 28 and 56 days after casting. The target strength for all mixtures was 6000 psi
(41.4 MPa) and all mixtures exceeded this target strength at 56 days.
Table 3 — Mixture proportions of mixtures evaluated in study.

Portland
Fly Ash, lbs./cy Slag, lbs./cy Water, lbs./cy Sand, lbs./cy
Mixture ID Cement, lbs./cy 3 3
(kg/m ) (kg/m ) (kg/m3) (kg/m3)
(kg/m3)
C 893 (530) 0 0 402 (238) 2457 (1458)
F20 709 (421) 177 (105) 0 399 (237) 2438 (1446)
F40 528 (313) 352 (209) 0 396 (235) 2419 (1435)
S30 618 (367) 0 265 (157) 397 (236) 2428 (1441)
S60 349 (207) 0 524 (311) 393 (233) 2401 (1424)
7.8
@Seismicisolation
@Seismicisolation
Table 4 — Chemical compositions (percent) of cementitious materials.

Material LOI SO3 SiO2 Fe2O3 MgO Al2O3 Na2O CaO C3S C3A K2O TiO2 IA
Portland
Cement 2.6 3.1 20.2 3.5 0.7 4.8 0.52 64 52 7 n/a n/a n/a
Type I/II
Class F
0.25 0.7 51.5 6.2 4.1 16.9 1.1 11.7 n/a n/a n/a n/a n/a
Fly Ash
Grade
2.2 4.8 31.6 0.8 4.2 12.4 43.5 n/a n/a n/a 0.6 0.5 9
100 Slag
LOI: loss on ignition; IA: inorganic addition; n/a: not available
Table 5 — Chemical composition of steel reinforcement used in corrosion specimens.

Steel C Mn Si S P Cu Cr Ni Mo V Fe
ASTM A615 0.32 0.91 0.23 0.039 0.034 0.51 0.55 0.30 0.09 0.014 Remainder
Figure 1 — Schematic of specimen used to assess corrosion activity and critical chloride threshold.
7.9
@Seismicisolation
@Seismicisolation
Table 6 — Test type, test times, and number of samples tested.

Test Time (days after casting)
Test at corrosion
28 days 56 days 84 days 112 days
initiation
Pore Solution pH 2 2 NT NT NT
Pore Solution Chemistry 4 4 NT NT NT
Apparent Diffusivity, Da,
NT NT 3 3 5
ASTM C1556
Corrosion Activity, Modified
bi-weekly test until corrosion activation
ASTM G109
Critical Chloride Threshold,
NA NA NA NA 5
CT, ASTM C1152
NT: not tested; NA: not applicable
RESEARCH RESULTS
Pore Solution pH and Chemistry

The pH of the pore solution has been reported to be a significant factor in corrosion initiation of steel reinforcement
embedded in cementitious materials. This research evaluated the pH of the pore solution at 28 and 56 days after
casting. Fig. 2 shows the results from the testing. The results indicate that the pH of the F20 mixture is similar to the
pH of the control mixture at both 28 and 56 days after casting. The pH of the F40, S30, and S60 mixtures are
significantly lower than the control mixture.
The chemistry of the pore solution was also assessed at 28 and 56 days after casting. Figs. 3 through 5 show the results
from this testing (note the different scales on the ordinates). Only results from calcium, sodium, and potassium are
shown. Results from the ICP-OES analysis for the Al and Si showed concentrations of less than 20 ppm for all times
evaluated. The results show that the calcium concentration for the control mixture increased between 28 and 56 days
and was higher than the mixtures containing SCM. Also, the calcium concentration for the mixtures containing SCM
was similar at 28 and 56 days. The concentration of potassium in the pore solution increased for the mixtures
containing slag and either decreased or showed a slight increase for the control mixture and mixtures containing fly
ash. The sodium concentration of the pore solution decreased from 28 to 56 days for the control specimen while this
same concentration either showed a slight decrease or increase for the other mixtures.
13.6 500
Control S30
13.4 400 F20 S60
Concentration of Ca (ppm)
F40
13.2
300
pH
13
200
12.8
100
12.6
0
12.4
28d 56d 28d 56d 28d 56d 28d 56d 28d 56d 25 30 35 40 45 50 55 60
Time After Casting (days)
Control F20 F40 S30 S60
Figure 2 — Pore solution pH for different mixtures. Figure 3 — Concentration of Ca vs time for all
mixtures.
7.10
@Seismicisolation
@Seismicisolation
4000 4500
Control S60
3500 4000
Concentration of Na (ppm)
F20 S30
Concentration of K (ppm)
3500 F40
3000
3000
2500
2500
2000
2000
1500 Control S30 1500
F20 S60
1000 1000
F40
500 500
25 30 35 40 45 50 55 60 25 30 35 40 45 50 55 60
Time After Casting (days) Time After Casting (days)
Figure 4 — Concentration of K vs time for all mixtures. Figure 5 — Concentration of Na vs time for all
mixtures.
Apparent Chloride Diffusivity on Mixtures

The rate at which chlorides are transported through concrete is often measured via apparent diffusion. ASTM C1556,
Standard Test Method for Determining the Apparent Chloride Diffusion Coefficient of Cementitious Mixtures by Bulk
Diffusion, can provide a measure of resistance to chloride transport. However, this method uses a specific exposure
condition and to fairly compare the performance of mixtures all specimens must be exposed under the same conditions.
This research assesses the apparent chloride diffusion coefficient of the control, F20, F40, S30, and S60 mixtures after
28 and 56 days of curing followed by 56 days of exposure (continuous exposure). In addition, the apparent chloride
diffusion coefficient was determined for the specimens used to assess corrosion activity and critical chloride threshold
(wet-dry exposure conditions). Fig. 5 shows the results from the specimens cured for 28 days and exposed to chloride
solution for 56 days and Fig. 6 shows the results from the specimens cured for 56 days and exposed to chloride solution
for 56 days.
10-10 10-10
D (m /sec)
D (m /sec)
10-11 10-11
2
2
a
-12 -12
10 10
C F20 F40 S30 S60 C F20 F40 S30 S60
Figure 6 — Measured diffusion coefficients for Figure 7— Measured diffusion coefficients for
specimens cured for 28 days and exposed for specimens cured for 56 days and exposed for
56 days. 56 days.
7.11
@Seismicisolation
@Seismicisolation
A t-test analysis of the diffusion coefficient data for the 28-day cure specimens indicates that the null hypothesis (μ1
= μ2) is rejected (p-value < 0.05) for all comparisons with the control mixture. That is, the mean diffusion coefficient
of specimens cured for 28 days containing SCMs are all significantly lower than the control mixture cured for the
same period. This was also the case for the specimens cured for 56 days. In addition, statistical t-tests indicate that the
null hypothesis is not rejected for differences in curing
times (e.g., control specimens exhibit similar diffusion 10-11
coefficients for control specimens cured for 56 days) for all
mixture types except S30. The null hypothesis is rejected
for the S30 specimens with different curing times and this
is likely a result of the small scatter associated with the S30
D (m /sec)
specimens cured for 56 days. 10-12
2
a
In addition to the specimens subjected to continuous
ponding, all corrosion monitoring specimens were assessed
for apparent diffusion coefficients at the time of corrosion
initiation. These specimens were ponded with a 10% -13
10
chloride solution for one week followed by one week of
C F20 F40 S30 S60
drying. This was continued until the coulombs passed in
the specimens reached 5 coulombs, which is consistent Figure 8 — Measured diffusion coefficients for
with ASTM G109, where activation is assumed to occur corrosion monitoring specimens.
when the average current over a 6 month period is 0.8
μA/in2 (0.12 μA/cm2). The diffusion coefficients from the corrosion monitoring specimens are all lower than the same
mixtures exposed to continuous ponding (note the different scales on the ordinates).
Critical Chloride Threshold Testing

Figure 1 showed the design of the specimens used to assess the corrosion activity and critical chloride threshold. The
specimens had a cathode to anode surface area ratio of 2.3. In addition, initiation of corrosion was assumed to occur
when the average of all specimens achieved 5 coulombs and at least half of the specimens had exhibited a charge
transfer of 5 coulombs. This 5 coulombs correlates to an average current over a 6 month period of 0.8 μA/in2 (0.12
μA/cm2) and is similar to initiation requirements defined in ASTM G109. Figure 9 shows the average values of the
corrosion tests. Note that S30 specimens activated first, followed by the S60 samples, F40 samples, and then the F20
and control samples.
After initiation, the activated samples were prepared for acid soluble chloride testing. The perimeter of the samples
were cut approximately at the center of the anode (approximately ¾ inch (19 mm) below the ponded surface). Grinding
then initiated at the level of the reinforcement (lower chloride concentrations) in 0.08 inch (2 mm) increments until
the exposed surface was reached (these data were used for determining apparent diffusion coefficients reported
earlier). The chloride concentration at the top reinforcement surface is reported as the critical chloride threshold.
Figure 10 shows the critical chloride threshold values as a percentage of the mass of mortar. T-tests indicates that the
null hypothesis (μ1 = μ2) is rejected (p-value < 0.05) for all comparisons with the control mixture. That is, the mean
critical chloride threshold value of the specimens containing SCMs are all significantly lower than the control mixture.
In fact, the mean critical chloride threshold of the specimens containing SCMs as a percent of mortar mass is
approximately 7.7% of the critical chloride threshold for the control samples (no SCMs).
ACI documents specify allowable chloride limits based on the percent of portland cement. To assess the influence of
cement versus cementitious materials content on critical chloride threshold, the critical chloride threshold is shown as
a percentage of portland cement and overall cementitious material content. Figures 11 and 12 show these plots. The
mean critical chloride threshold of the specimens containing SCMs as a percent of portland cement content is
approximately 7.7% of the critical chloride threshold for the control samples (no SCMs) and 12.7% when measured
as a percent of mass of cementitious materials.
The results from the mortar specimens and t-tests indicate that the critical chloride threshold of specimens containing
SCMs is significantly lower than specimens containing portland cement only (all p-values <0.00002). This is likely a
result of the reduced pH of the pore solution. These critical chloride threshold values are important because the
allowable chloride limits should be based (possibly as a percentage) of these critical values. This research indicates
7.12
@Seismicisolation
@Seismicisolation
that if allowable chlorides are a function of critical chloride threshold, the allowable values published by ACI for
mixtures containing SCMs may be unconservative. However, because the research shows there are benefits in reducing
the transport rate for mixtures containing SCMs, this does not mean that early corrosion would occur. A more
comprehensive analysis is needed. This analysis should assess time to corrosion, which considers both transport rates
and critical chloride threshold values, which would provide a better measure of the potential effectiveness of SCMs.
This will be assessed in the following section.
6 0.6
5
Average Coulombs (A/sec)
0.5
C (% by mass of mortar)
4
0.4
3
0.3
2
Control
0.2
1 F20
F40
T
0 S30 0.1
S60
-1 0
0 50 100 150 200 250 300 350 400 C F20 F40 S30 S60
Exposure Time (days)
Figure 9 — Average corrosion data from each mixture Figure 10 — Critical chloride threshold values for each
type. mixture type (% of mortar mass).
C (% by mass of cementitious materials)
2.5 2.5
C (% by mass of portland cement)
2 2
1.5 1.5
1 1
0.5 0.5
T
0 0
T
C F20 F40 S30 S60 C F20 F40 S30 S60
Figure 11 — Critical chloride threshold values for each Figure 12 — Critical chloride threshold values for each
mixture type (% by mass of portland mixture type (% by mass of cementitious
cement). materials).
Assessing Time to Corrosion

This research indicates that the apparent diffusion coefficients of mixtures containing SCMs are lower than the
apparent diffusion coefficient of mixtures containing only portland cement as the binder. This indicates that SCMs are
beneficial for reducing the transport rate of chlorides into concrete. The results indicate that for 28 day curing when
exposed to continuous ponding the apparent diffusion coefficient could be reduced by 75%, but on average is about a
40% reduction. For specimens cured for 56 days with continuous chloride solution exposure, the apparent diffusion
coefficient could be reduced by almost 90% but on average is about a 75% reduction. For the specimens cured for 56
days and subjected to cyclic wet-dry exposure the apparent diffusion coefficient was reduced by about 90% on
average.
This research also indicates that the critical chloride threshold, that is the amount of chlorides that are needed to initiate
corrosion, is lower for mortar mixtures containing SCMs. The critical chloride threshold from this research indicates
that the threshold value for systems containing SCMs is about 8% of the values of the samples containing only portland
7.13
@Seismicisolation
@Seismicisolation
cement as the binder when reported by percent mass of mortar or cementitious materials. When reported as a mass
percent of portland cement only, the critical chloride threshold of systems containing SCMs is about 13% of the system
containing portland cement only. This indicates that SCMs are not beneficial in resisting corrosion when chlorides are
at the steel-concrete interface. This research indicates that SCMs will slow the transport rate of chlorides through a
cementitious material but lower quantities are needed to initiate corrosion. However, knowing only these values
(diffusivity and threshold), it is difficult to assess whether SCMs are beneficial in resisting the corrosion unless a more
comprehensive study is performed. To better understand the overall effects of SCMs on time to corrosion, a parametric
study will be performed.
To assess the potential overall influence of SCMs on time to corrosion, a time to corrosion study will be performed.
This time to corrosion study will assume Fick’s second law, as follows:
(1)
where CT = critical chloride threshold, % by mass of portland cement, cementitious materials, or mortar
CS = surface chloride concentration, % by mass of portland cement, cementitious materials, or mortar
Ci = background chloride concentration, % by mass of portland cement, cementitious materials, or
mortar
x = depth of reinforcement or concrete cover, meters
Da = apparent diffusion coefficient, m2/sec
t = time to corrosion, sec
Equation 1 can be rewritten so that the time to corrosion can be determined as follows:
(2)
To assess the effect of the diffusion coefficient (Da) and critical chloride threshold value (CT) on the time to corrosion,
the low, average, and high Da and CT values determined from the testing can be used. These values are shown in Table
7. To assess the time to corrosion, a cover depth, x, of 0.05 meters (2 inches), a background chloride concentration
(Ci) of 0%, and a surface chloride concentration, CS, of 0.6% by mass of mortar will be used for a parametric analysis.
Note that other conditions could change the outcome and these will be assessed in further studies.
Figure 13 shows the comparison of time to corrosion for each mixture. Note that the each box in the box plot includes
9 data points; 3 apparent diffusion coefficients (low average and high Da values) and 3 critical chloride threshold
values (low, average, and high CT values) are shown. The upper and lower whiskers represent the high and low values,
the edges of the box represent the 25th and 75th percentile, and the line within the box represents the median value.
The figure shows the potential distribution of time to corrosion values for each mixture and indicates that the system
without SCMs (the control) can provide longer times to corrosion than the systems containing SCMs. The figure also
shows that average time to corrosion for the control mixture is higher than the mixtures containing fly ash and slag.
T-tests at a 95% confidence interval indicate that the null hypothesis (μ1 = μ2) cannot be rejected (p-value > 0.05) for
the mixtures containing fly ash and slag when compared with the control. However, at a 90% confidence interval, the
null hypothesis is rejected (μ1 ≠ μ2) for the control and SCM mixtures. The p-values are 0.07, 0.09, 0.06, and 0.07 for
the F20, F40, S30, S60 and control mixtures time to corrosion comparison. The large variation of time to corrosion
values of the control group influences these outcomes.
Figure 14 shows the 3-dimensional plot with apparent coefficient of diffusion (Da) and critical chloride threshold (CT)
as a function of mortar mass versus time to corrosion using the data developed in this research. The figure shows that
the time to corrosion is very sensitive to CT. Fig. 15 shows time to corrosion as a function of Da and CT for the limits
7.14
@Seismicisolation
@Seismicisolation
of data found in this research. The additional data points clearly show the significant influence of CT and that in the
range of Da values shown, time to corrosion is less sensitive to Da.
This study evaluated the influence of SCM replacements on diffusivity and critical chloride threshold, assuming no
background chlorides. If ACI recommendations on allowable chlorides for reinforced concrete structures in wet
conditions (0.10% by weight of portland cement) were allowed for these mixtures, the research indicates that several
of the mixtures would initiate corrosion immediately after concrete placement. That is, the limit set by ACI is greater
than the values determined in this research. Although the value that defines corrosion activation in ASTM G109 and
used in this research (0.8 μA/in2 [0.12 μA/cm2]) is arbitrary, this is a reasonable value reported in the literature
(Broomfield 1994). Using the ACI limit for reinforced concrete exposed to wet condition, none of the control
specimens would exhibit corrosion initiation immediately after concrete placement, 2 of the F20 specimens would
have activated, 2 of the F40 specimens would have activated, 4 of the S30 specimens would have activated, and 1 of
the S60 specimens would have activated (9 of 23 would corrode immediately or 39% of the samples). If the allowable
chloride levels were based on total cementitious materials content, the number of activated samples would be 0, 2, 2,
4, and 3 for the control, F20, F40, S30, and S60 samples (11 of 23 would corrode immediately or 48% of the samples).
This indicates two potential issues: that the chloride limits reported in the ACI documents may be too low for mixtures
containing SCMs and that basing these limits on total cementitious materials content seems to be even less
conservative. However, more research is needed.
Table 7 — Da and CT values for all mixtures tested.

CT, % by mass of
CT, % by mass of CT, % by mass of
Mixture Value Da, m2/s cementitious
mortar portland cement
materials
Min 3.29E-12 0.239 1.002 1.002
Control Avg 6.59E-12 0.416 1.748 1.748
Max 9.91E-12 0.513 2.153 2.153
Min 7.97E-13 0.011 0.044 0.055
F20 Avg 9.20E-13 0.049 0.204 0.256
Max 2.24E-12 0.095 0.400 0.499
Min 5.446E-13 0.017* 0.071* 0.117*
F40 Avg 6.673E-13 0.037 0.157 0.261
Max 8.449E-13 0.089 0.374 0.624
Min 8.021E-13 0.005 0.019 0.028
S30 Avg 9.635E-13 0.015 0.062 0.088
Max 1.111E-12 0.025 0.106 0.151
Min 7.316E-13 0.018 0.074 0.185
S60 Avg 7.964E-13 0.027 0.114 0.286
Max 8.675E-13 0.042 0.175 0.436
*Indicates values that were not detected; in these cases 45% of the average value (the mean value for the other data
sets) is used.
7.15
@Seismicisolation
@Seismicisolation
200
x = 0.05 m (2 inches)
C = 0.6% by mass of mortar
s
Time to Corrosion (years)
C=0
150 i
100
50
0
C F20 F40 S30 S60
Figure 13 — Time to corrosion assessment for mixtures with and without SCMs.
Figure 14 — Time to corrosion as a function of Da and Figure 15 — Time to corrosion as a function of Da and
CT for data generated in this research. CT for data ranges determined from
research.
SUMMARY, DISCUSSION, AND FUTURE NEEDS
This research evaluated the apparent diffusion and critical chloride threshold values of standard black reinforcement
embedded in mortar (made in the laboratory) containing portland cement with and without SCM replacements. The
research indicates that the replacement of portland cement with SCMs can reduce the apparent diffusivity of the
cementitious mixture for lab-made samples. This research indicates that SCMs can reduce the apparent diffusion
coefficient by up to 90%., again for lab-made samples Significant reductions in diffusivity for mixtures containing
SCMs has been widely reported in the literature.
his research also indicates that the critical chloride threshold of mixtures containing SCMs is lower than the critical
chloride threshold of mixtures containing SCMs. The results from this research indicate that the average critical
threshold value for mixtures containing fly ash is about 12% of the critical chloride threshold of mixtures containing
7.16
@Seismicisolation
@Seismicisolation
no fly ash. There was a small decrease in average critical chloride threshold when the replacement level increased
from 20 to 40%. The results from this research also indicate that the critical chloride threshold of the mixtures
containing slag is significantly lower than the critical chloride threshold of mortar mixtures containing portland cement
only as the binder. The average critical chloride threshold of the mixtures containing slag was determined to be
approximately 6% of the average critical chloride threshold of mortar mixtures containing portland cement only for
the binder. For the mixtures containing slag a small increase in average critical chloride threshold values was observed
with increased replacement levels. These finding have not been widely reported in the literature but some work by
Byfors (1987) and Bouteiller et al. (2012) indicated the addition of SCMs to concrete mixtures could reduce the critical
chloride threshold value.
This research indicates that the addition of SCMs results in a decrease in diffusivity and a decrease in critical chloride
threshold for specimens made in the laboratory. The decrease in diffusivity is positive and alone would increase the
time to corrosion. The decrease in critical chloride threshold is negative and alone would decrease the time to
corrosion. To better assess the overall impact of SCMs on longer-term corrosion performance, time to corrosion
analyses, which are dependent on both apparent diffusion coefficient and critical chloride threshold, were performed.
Assuming specific values for surface chloride concentration, background chloride concentration, and cover, the results
indicate that time to corrosion values for mixtures containing no SCMs can be longer than mixtures containing SCMs.
However, statistical testing the 95% confidence interval indicates that insufficient information is available to conclude
that the mean values of the time to corrosion for the mixtures containing SCMs and no SCMs are different.
The assessment reported here assumed the background chlorides in the mixture was zero. ACI limits the amount of
chlorides in new concrete. The results indicate that had the limit on allowable chlorides been added to the tested
specimens, 39% of the specimens containing SCMs would have initiated corrosion at or soon after placement when
the limit is based on portland cement content (0.1% of mass of portland cement). If this chloride limit was based on
overall cementitious material content (0.1 % of ~893 lbs. [530 kg]), 48% of the samples would have initiated corrosion
at or near the time of placement. This research indicates that the allowable limits published in ACI 222 and 318 seem
to be less than the critical chloride threshold value, although high, for systems containing only portland cement as the
binder. However, this research (based on specimens made in the laboratory) indicates that the limits listed in ACI 222
and 318 may be unconservative for mixtures containing SCMs when based on portland cement content and even less
conservative when based on overall cementitious materials content.
As with all research, limitations exists. Additional research is needed to validate the results from this research. This
research only evaluated reinforcement meeting ASTM A615 specifications embedded in mortar (not concrete) at one
water-cementitious materials ratio with one type and source of cement, one type and source of fly ash, and one type
and source of slag. Concrete placed in the field may exhibit different performance than the laboratory produced
specimens. It is recommended that concrete specimens containing various sources and types of cement, fly ash, and
slag be evaluated. Also, additional SCMs should be evaluated.
7.17
@Seismicisolation
@Seismicisolation
REFERENCES
Abrams, D. A., “Tests of Impure Waters for Mixing Concrete,” Journal of the American Concrete Institute, Vol. 20,
Feb. 1924, pp. 442-486.
ACI 222R, Protection of Metals in Concrete Against Corrosion, American Concrete Institute, 2001 (Re-approved
2010), 41 pgs.
Andrade, C. and Buják, R., “Effects of Some Mineral Additions to Portland Cement on Reinforcement Corrosion,"
Cement and Concrete Research, Vol. 53, November 2013, pp. 59-67.
Angst, U., Elsener, B., Larsen, C. K., and Vennseland, O., “Critical Chloride in Reinforced Concrete,” Review,
Cement and Concrete Research, V. 39, 2009, p. 1122-1138.
Alonso, M. C. and Sanchez, M., “Analysis of Variability of Chloride Threshold Values in the Literature,” Materials
and Corrosion, V. 60, N. 8, 2009, p. 631-637.
Boddy, A., Hooton, R.D., Gruber, K.A., "Long-term Testing of the Chloride Penetration Resistance of Concrete
Containing High-reactivity Metkaolin,” Cement and Concrete Research, Vol. 31, 2001, pp. 759-765.
Bonden, T., Blasing, T.J., “Record High 2010 Global Carbon Dioxide Emissions from Fossil-fuel Combustion and
Cement Manufacture,” posted on CDIAC site, in: U.S. Department of Energy (Ed.), Carbon Dioxide Information
Analysis Center Oak Ridge, TN, 2012.
Burris, L.E., and Riding, K.A., “Diffusivity of Binary and Ternary Concrete Mixture Blends,” ACI Materials Journal,
Vol. 111, No. 4, July-August 2014, pp. 373-382.
Byfors, K. (1987). “Influence of Silica Fume and Fly Ash on Chloride Diffusion and pH Values in Cement Paste”
Cement and Concrete Research. Vol. 17. No. 1. pp. 115-129.
Bouteiller, V., Cremona, C., Baroghel-Bouny, V., Maloula, A., “Corrosion Initiation of Reinforced Concretes based
on Portland or GGBS Cements: Chloride Contents and Electrochemical Characterizations Versus Time,” Cement
and Concrete Research, Vol. 42, 2012, pp. 1456-1467.
Broomfield, J.P., "Assessing Corrosion Damage in Reinforced Concrete Structures," Corrosion and Corrosion
Protection of Steel in Concrete, Ed. R.N. Swamy, Sheffield Academic Press, 1994, pp. 1-25.
Clear, K. C., and Harrigan, E. T., “Sampling and Testing for Chloride Ion in Concrete,” Federal Highway
Administration Report No. FHWA-RD-77-85, August 1977, pp. 25.
Cyr, M., Rivard, P., Labrecque, R., Daidie, A. (2008), “High-pressure Device for Fluid Extraction from Porous
Materials: Application to Cement-based Materials,” Journal of the American Ceramic Society, Vol. 91, No. 8.
pp. 2653-2658.
Hussain, S.E. and Rasheeduzzafar, “Corrosion Resistance Performance of Fly Ash Blended Cement Concrete,” ACI
Materials Journal, Vol. 91, No. 3, 1994, pp. 264-272.
Jones, M.R., Dhir, R.K., and Magee, B.J., “Concrete Containing Ternary Blended Binders: Resistance to Chloride
Ingress and Carbonation,” Cement and Concrete Research, V. 27, No. 6, 1987, pp. 825-831.
Leng, F., Feng, N., Lu, X., “An Experimental Study on the Properties of Resistance to Diffusion of Chloride ions of
Fly Ash and Blast Furnace Slag Concrete,” Cement and Concrete Research, Vol. 30, No. 6, June 2000, pp. 989-
992
Li, Z., Peng, J., Ma, B., “Investigation of Chloride Diffusion for High-Performance Concrete Containing Fly Ash,
Microsilica and Chemical Admixtures,” ACI Materials Journal, Vol. 96, No. 3, 1999, pp. 391-396.
Mehta, P.K. , Monteiro, P.J.M., Concrete: Structure, Properties, and Materials, 3rd ed., McGraw-Hill, New York, 2006.
Papadakis, V.G., “Effect of Supplementary Cementing Materials on Concrete Resistance Against Carbonation and
Chloride Ingress,” Cement and Concrete Research, V. 30, 2000, pp. 291-299.
Thomas, M.D.A., and Bamforth, P.B., 1999, “Modelling Chloride Diffusion in Concrete: Effect of Fly Ash and Slag,”
Cement and Concrete Research, V. 29, pp. 487-495.
7.18
@Seismicisolation
@Seismicisolation
Thomas, M.D.A., Shehata, M.H., Shashiprakash, S.D., Hopkins, D.S., and Cail, K., “Use of Ternary Cementitious
Systems Containing Silica Fume and Fly Ash Concrete,” Cement and Concrete Research, Vol. 29, 1999, pp.
1207-1214.
Worrell, E. Price, L., Martin, R., Hendricks, C.A. , and Ozawa Meida, L., “Annual Review of Energy and
Environment,” Annual Reviews of Energy and the Environment, Vol. 26, 2001, pp. 303-329.
7.19
@Seismicisolation
@Seismicisolation
7.20
@Seismicisolation
@Seismicisolation
SP-308—08
A THERMODYNAMIC PERSPECTIVE ON ADMIXED CHLORIDE LIMITS OF CONCRETE

PRODUCED WITH SCMs
Vahid Jafari Azad and O. Burkan Isgor
Synopsis: A thermodynamic modeling investigation was conducted to quantify free and bound chloride
concentrations, and pore solution pH, in mixtures produced with different types of blended cements and admixed
chlorides. Specifically, the validity of using total cementitious materials content, instead of cement content, as the
basis for allowable admixed chloride limits in new construction was evaluated. The effects of replacing OPC with two
types of fly ash (class C and F) and one type of slag at different replacement levels were investigated. Assuming
chlorides are only chemically bound, the water soluble Cl-/OH- remained rather stable up to 20% class F fly ash, 30%
class C fly ash, and 40% slag replacements. Beyond these replacement levels Cl-/OH- were shown to increase, in some
cases rather sharply, for the mixtures that were studied in this investigation. The results of this investigation support
the ACI committee recommendations for allowable chloride limits to be reported in terms of total cementitious
materials content up to 20% class F fly ash, 30% class C fly ash, and 40% slag replacements. Since SCM replacements
affect the binding capacity, water soluble chlorides, and the pH of pore solution simultaneously, it is recommended
that allowable chloride limits are made in terms of Cl-/OH- rather than chloride contents alone.
Keywords: Chloride binding; Admixed chlorides; Chloride limits; Cement; Blended cements; Supplementary
cementitious materials; Thermodynamic reaction modelling.
8.1
@Seismicisolation
@Seismicisolation
Vahid Jafari Azad is currently a Postdoctoral Scholar at Oregon State University. He obtained his PhD degree in
Civil Engineering form University of Tehran, Iran. His research background and interests include multi-physics
modelling of cement-based materials and reinforcement in concrete as well as their associated deterioration
mechanisms.
O. Burkan Isgor is an Associate Professor at Oregon State University. He is the secretary of the ACI committee 222
on Corrosion of Metals in Concrete and a member of committees 236 on Materials Science of Concrete and 365 on
Service Life Prediction. His research interests include analysis, design and durability of reinforced concrete structures,
corrosion of steel in concrete, and the development of non-destructive and model-assisted test methods.
INTRODUCTION
The role of chlorides on the service life of reinforced concrete structures is well established [1-8]. Limiting initial
chloride content of mixtures in new construction and reducing chloride ingress during service are critical in mitigating
chloride-induced corrosion issues. Currently, ACI uses cement content alone as the basis for allowable admixed
chloride limits for new constructions. In this context, “cement” is interpreted as ordinary portland cement (OPC).
Table 1 shows the allowable water soluble chloride limits by different ACI committees, represented as a percentage
by weight of cement content. With the increased use of supplementary cementitious materials (SCMs) and
performance-based specifications for producing concrete, there is need to evaluate the validity of using total
cementitious materials content, instead of cement content, as the basis for allowable chloride limits in new
construction. In the current state, the ACI allowable chloride limits impose stricter restriction on blended cements than
OPC since the former will have a smaller OPC content than the latter. This restriction makes it more difficult to
incorporate SCMs into concrete mixtures for new construction.
Table 1 — Allowable chloride limits for new construction proposed by different ACI committees.
Maximum water soluble chloride content in concrete
Category
(% by weight of cement)
ACI 201 [9] ACI 222 [10] ACI 318 [11]
Prestressed concrete 0.06 0.06 0.06
Reinforced concrete
0.10 0.15(1)
(chloride exposure)
0.08
Reinforced concrete
0.15 0.3(2)
(wet conditions)
Reinforced concrete
1.0 0.15 1.0(3)
(dry conditions)
(1) ACI 318-11 defines this exposure condition C2, where concrete is exposed to both moisture and an external
source of chlorides. Additional restrictions exist on maximum w/c, minimum f’c and concrete cover.
(2) ACI 318-11 defines this exposure condition C1, where concrete is exposed to moisture but not to an external
source of chlorides.
(3) ACI 318-11 defines this exposure condition C0, where the concrete will be dry or protected from moisture.
Replacing a percentage of OPC with SCMs has two main implications on chloride-induced corrosion in concrete
structures. The first one is the effect of SCM replacement on the chloride binding capacity of concrete, which is
important because water soluble (free) chlorides, not bound chlorides, are prominently involved in the corrosion
initiation process. The second one is the effect of SCM incorporation on the pH of the concrete pore solution. Since
the latter is also related to chloride binding capacity, which is reduced with decreasing pH of the pore solution [12], the
two effects are interconnected.
Chloride ions can be bound in cementitious materials both physically and chemically. The physical binding is mainly
due to the adsorption of chloride ions on the C-S-H phase and is the prominent binding mechanism at low chloride
concentrations [13-16]. Chemical binding is strongly associated with the reactions of the free chloride ions with
tricalcium aluminate (C3A) and tetracalcium aluminoferrite (C4AF) phases in cement [17, 18] to produce complex
chloride salts such as Friedel’s salt (3CaO·Al2O3·CaCl2·10H2O) or other similar complex chloride compounds (e.g.
(3CaO·Fe2O3·CaCl2.10H2O) [19]. The C3A phase is the main binding component of cement with a greater binding
8.2
@Seismicisolation
@Seismicisolation
A Thermodynamic Perspective on Admixed Chloride Limits of Concrete Produced with SCMs
capacity than that of the C4AF phase [18]. The chemical binding capacity of cementitious materials are also influenced
by other factors such as alkali content, water-to-binder ratio (w/b) and sulfite (SO3) content [13]. The incorporation of
SCM in concrete mixtures affect some of these factors; therefore, it is important to study allowable chloride limits in
blended cements.
In this study, a thermodynamic modeling investigation was conducted to quantify free and bound chloride
concentrations, as well as pH, in mixtures produced with different types of blended cements and admixed chlorides.
The objective of the work is to quantify the effect of SCMs on the state of chloride in concrete, and if warranted, to
provide supporting theoretical data for revising allowable chloride limits in concrete that are typically specified in
terms of cement content alone. Specifically, the validity of using total cementitious materials content, instead of
cement content, as the basis for allowable chloride limits in new construction was evaluated. The effects of replacing
OPC with two types of fly ash (class C and F) and one type of (ground granulated blast furnace) slag at different
replacement levels were investigated. In addition, parametric studies were performed to investigate the effects of w/b,
alkali content and SO3 content on the free and bound chloride concentrations, as well as pH, in mixtures produced
with different types of blended cements.
THERMODYNAMIC MODELING
Background
The thermodynamic modeling of cementitious systems was performed using the open source GEMS3K software [20,
21]
, which is based on Gibbs free energy minimization method [22]. GEMS3K is able to compute the molar amounts of
dependent components (e.g. ions and molecules) and their activities, as well as chemical potentials of the analyzed
system simultaneously. In order to model cement-based blended systems, their hydration process, and pore solution
chemistry, a thermodynamic database for cement-based materials, CEMDATA (version 14.01) [23-35], was used in
GEMS3K. This database allows the modeling of the evolution of the hydration products as well as the precipitation
and dissolution of complex chloride compounds such as Friedel’s salt [36-38]. Calculations can be performed at a
temperature range between 0oC and 100oC and at different pressures; however, all calculations in this research were
performed at room temperature (25oC) and atmospheric pressure.
Several validation studies of the GEMS3K-CEMDATA thermodynamic modeling framework exist [27, 38-42] therefore,
additional validation work will not be presented here. In order to demonstrate the capabilities of the GEMS3K-
CEMDATA modeling platform as they relate to the objectives of the current investigation, an example analysis is
presented here to show the hydration mechanism of OPC with mixed chlorides at a w/b of 0.5. The chemical
composition of the OPC used in the analysis is provided in Table 2 [43]. The dissolved amounts of main cement phases
(i.e. alite (C3S), belite (C2S), aluminate (C3A) and ferrite (C4AF)) were calculated using an empirical approach for
cement hydration modelling, developed by Lothenbach et al. [25, 27]. An alkali uptake model, developed by Hong and
Glasser [44, 45], was used in all simulations to model for physical adsorption of alkali on hydration products. GEMS
uses these dissolved amounts as input data during thermodynamic modeling calculations. Chlorides were assumed to
be added into the mixture in the form of 1% NaCl. The temperature and pressure were also assumed as average room
temperature and atmospheric pressure; i.e., 25oC and 0.1 MPa (14.50 psi), respectively. The effect of aggregates on
chemical and physical properties of the mixture was ignored. The total modeled sample weight was 0.142 kg (0.313
lb). Figure 1 illustrates the thermodynamic modeling results over 1000 days from initial mixing. The hydration
process, the pH evolution in the pore solution, and the relationship between available aluminates (C3A) and Friedel’s
salt formation are clearly shown in the figure.
Table 2 — Cement chemical composition for typical Portland cement sample [43].
Cement C3S C2S C3A C4AF Na2Oeq* MgO SO3 Available
sample (%) (%) (%) (%) (%) (%) (%) Na2Oeq (%)
OPC
52.2 21.6 8.1 7.6 0.2 2.3 3.0 -
average
*
: Sodium equivalent available alkali content
8.3
@Seismicisolation
@Seismicisolation
Figure 1 —GEMS3K analysis showing the hydration mechanism of OPC with mixed chlorides at a w/c of 0.5.
Incorporation of SCMs
Using the thermodynamic modeling framework that is induced in the previous section, the effect of fractional SCM
replacement of OPC (with the same composition as given in Table 1) on free and bound chloride concentrations, as
well as pH, in blended cements were investigated at full hydration [46]. The following three types of SCMs were
considered: (1) class F fly ash, (2) class C fly ash, and (3) slag. Table 3 provides the assumed chemical compositions
of each type of SCMs used in this numerical investigation. Increasing amounts of SCM replacement (0-50%) for
mixtures with three water-to-binder ratios (w/b: 0.4, 0.5 and 0.6) were analyzed. Figure 2 illustrates the modeling
steps used to conduct the thermodynamic analysis for all these cases. In total over 1500 cases were analyzed. In
addition, the effects of alkali and SO3 contents were investigated through a parametric investigation of mixtures with
w/b of 0.5. Table 4 shows the parametric analysis grid.
Table 3 — Chemical composition of the SCMs that are used in this modeling study.
SCM type SiO2 Al2O3 Fe2O3 CaO SO3 MgO Available* Na2Oeq
(%) (%) (%) (%) (%) (%) (%)
Class F Fly ash [43]** 50.3 20.8 13.1 6.9 0.9 1.2 0.6
Class C Fly ash [43] ** 38.3 18.8 5.9 22.3 2.3 4.6 1.8
Slag [47] 35 12 1 40 9 - 0.6
*
: Available alkali content as per ASTM C311 [48]
**: Average of three Class F and 15 Class C fly ash samples
Table 4 — Parametric analysis grid alkali and SO3 contents.

SCM type Equivalent sodium (%) SO3 (%)
min base max min base max
Fly ash, class F [43] 0.3 0.6 1.1 0.2 0.9 1.4
Fly ash, class C [43] 0.8 1.8 4.8 0.2 2.3 4.8
Ground slag [47] 0.3 0.6 0.9 4.5 9 13.5
As shown in Table 1, the allowable water soluble chloride concentrations for new concrete mixtures vary based on
the recommending committee, structural element type and exposure conditions. In this investigation, the following
8.4
@Seismicisolation
@Seismicisolation
approach is taken to simplify this complication. It is assumed that chloride, in the form of NaCl, is mixed with OPC
concrete, with no SCM replacement, to produce the water soluble chloride concertation given by different ACI
committees for reinforced concrete elements exposed to chlorides; specifically three allowable levels were considered:
0.08% (ACI 222), 0.1% (ACI 201), 0.15% (ACI 318). The amount of NaCl necessary to achieve the allowable
concentration is determined through an iterative scheme as illustrated in Figure 2 (Algorithm A). Thermodynamic
analysis of the mixture with this amount of added NaCl provides the baseline calculations for water soluble and bound
chloride concentrations and the pH of the pore solution. The effect of SCMs are presented with respect to this baseline.
Figure 2 — Modeling steps used to conduct the thermodynamic analysis.

(*: Alkali uptake model from Hong and Glasser [44, 45]).
8.5
@Seismicisolation
@Seismicisolation
The effect of SCM replacement

In this section, the effect of fractional SCM replacement of OPC on free and bound chloride concentrations, as well
as pH of the pore solution, in mixtures produced with blended cements and admixed chlorides are presented at full
hydration. The results are shown for the assumption that the allowable water soluble chloride concentration in new
concrete mixtures is 0.15% by weight of cement for the SCM replacement case (OPC alone), coinciding with the ACI
318 recommendations; and the changes in free and bound chloride concentrations, pH and water soluble Cl-/OH- ratio
were shown for cases with SCM replacements at this chloride addition (in the form of NaCl). It should be noted that
these results do not include any physical chloride binding by the hydration products. In other words, in reality, the
chloride binding is expected to be higher (and the water soluble chlorides are expected to be lower) than the reported
values.
Fly ash  Figure 3 shows the changes in free and bound chlorides in concrete at different class F fly ash replacement
levels for different w/b. It can be observed from Figures 3(a)-(c) that the w/b plays an important role in the binding
capacity of the mixtures: the binding capacity decreases with increasing w/b. Another important observation is that
class F fly ash replacement level increases the free chloride levels in concrete, albeit rather modestly until about 20%
replacement, but rather more significantly afterwards. The decrease in chemical binding capacity with class F fly ash
replacement can be related to two main factors: (1) the available alkali content is larger in class F fly ash than in OPC
cement; and (2) the C3A content of the blended cement is lowered by class F fly ash replacement.
The changes in pH and water soluble Cl-/OH- with Class F fly ash replacement are illustrated in Figure 3(d). It can be
seen in this figure that the pH does not change significantly up until 20% class F fly ash replacement, but beyond this
level, it decreases to about 10.5 at 50% replacement. It can also be observed in this figure that the water soluble Cl-
/OH- remains rather stable up until 20% class F fly ash replacement, but increases rather rapidly beyond this level.
The changes in both quantities are depended to the w/b ratio, as shown in Figure 3(d); the lower w/b leads to larger
decrease in the pH and larger increase in the water soluble Cl-/OH-. From these figures in can be stated that up to 20%
class F fly ash replacement, the allowable water soluble chloride limits recommended by ACI 318 can be used;
however, these limits are not applicable for larger replacement levels. Additional analysis assuming lower allowable
water soluble chloride limits as shown in Table 1 (0.08% by ACI 2222 and 0.1% by ACI 201) did not change this
general observation. For brevity, these cases are not presented here.
Figure 4 shows the changes in free and bound chlorides in concrete at different class C fly ash replacement levels for
different w/b. Although general trends are similar to the cases with class F fly ash replacement (as shown in Figure
3), in general, class C fly ash replacement results in lower chemical binding than class F fly ash replacement, mainly
because of the high alkali content of class C fly ash. The available sodium equivalent content in class C fly ash is
approximately three times larger than that of class F, as shown in Table 3. Class C fly ash replacement level increases
the free chloride levels in concrete; this increase is relatively linear with class C fly ash replacement up until 30%.
However, the increase in free chloride content in the 0-30% replacement region is steeper than that of the cases with
0-20% class F replacement levels. Binding capacity decreases with increasing w/b. At around 30% class C fly ash
replacement levels, almost all bound chloride is released for cases with w/b of 0.5 and 0.6.
The changes in pH and water soluble Cl-/OH- with class F fly ash replacement are illustrated in Figure 4(d). It can be
seen in this figure that the pH increases slightly up until 30% class F fly ash replacement for w/c levels of 0.4 and 0.5,
and up to 40% for w/c of 0.6, but beyond this level, it decreases to about 12.5 for w/b of 0.6 and to about 11 for w/b
levels of 0.4 and 0.5 at 50% replacement. It can also be observed in this figure that the water soluble Cl-/OH- remains
rather stable up until 30% class C fly ash replacement, but increases rather rapidly beyond this level. From these
figures in can be stated that up to about 30% class C fly ash replacement for w/c levels of 0.4 and 0.5, and up to 40-
45% for w/c of 0.6, the allowable water soluble chloride limits recommended by ACI 318 can be used in as a weight
percentage of total cementitious materials content; however, the representation of limits as percentage of total
cementitious material content is not applicable for larger replacement levels. Additional analysis assuming lower
allowable water soluble chloride limits as shown in Table 1 (0.08% by ACI 2222 and 0.1% by ACI 201) did not
change this general observation. For brevity, these cases are not presented here.
8.6
@Seismicisolation
@Seismicisolation
(a) (b)
(c) (d)
Figure 3 — Effects of class F fly ash replacement on internal chloride chemical binding for a fully hydrated cement
for (a) w/b = 0.4; (b) w/b = 0.5, (c) w/b = 0.6; and (d) pH and water soluble Cl-/OH- for different w/b at different
replacement levels.
8.7
@Seismicisolation
@Seismicisolation
(a) (b)
(c) (d)
Figure 4 — Effects of class C fly ash replacement on internal chloride chemical binding for a fully hydrated cement
for (a) w/b = 0.4; (b) w/b = 0.5, (c) w/b = 0.6; and (d) pH and water soluble Cl-/OH- for different w/b at different
replacement levels.
Figure 5 provides additional explanation as to why the effect of fly ash addition becomes problematic after 20%
replacement for class F fly ash and after 30% for class C fly ash. This figure illustrates the C-S-H production and
portlandite consumption levels due to pozzolanic reactions with increasing class C and class F fly ash replacement
percentages. It is clear in this figure that increasing fly ash replacement percentages increases C-S-H levels in concrete
up to about 20% replacement level for class F fly ash and up to about 30% replacement level for class C fly ash. These
levels correspond to the depletion of the porlandite phase from the concrete. The difference in C-S-H and portlandite
levels in concrete with class F and class C fly ashes can be attributed to the difference in their calcium oxide (CaO)
content as shown in Table 3 (6.9% in class F; 22.3% in class C). It should also be noted that since C-S-H content in
concrete increases with increasing fly ash content up to 20% fly ash replacement, compared to an OPC mixture,
physical binding of chlorides will be larger in mixtures with fly ash replacement. Many researchers indicate that the
physical binding is strongly related to C-S-H amounts [14-16]. This additional binding, which is not studied in this
research, will further reduce the water soluble (free) chlorides in the pore solution. As a result, the results presented
here with respect to water soluble Cl-/OH- values can be considered to be on the conservative side.
8.8
@Seismicisolation
@Seismicisolation
(a)
(b)
Figure 5 — Changes in C-S-H and portlandite phases with fly ash replacements: (a) class F and (b) class C.
Slag  Figure 6 shows the changes in free and bound chlorides in concrete at different slag replacement levels for
different w/b. It can be observed from Figures 6(a)-(c) that the w/b plays an important role in the binding capacity of
the mixtures: the binding capacity decreases with increasing w/b. However, in general, the level of water soluble
chlorides and chloride binding are relatively insensitive to slag addition up to 40% replacement levels. This is mainly
due to the stable and slightly increasing pH levels in the concrete pore solution up to 40% slag replacement, as shown
in Figure 6(d). The stability of pH can be attributed to its high CaO content of slag (40% as shown in Table 3) and
relatively low alkali content (0.6% as shown in Table 3). Beyond 40% slag replacement, water soluble chloride content
increases; the increase is more prominent in mixtures with larger w/b.
The changes in pH and water soluble Cl-/OH- with slag replacement are illustrated in Figure 6(d). It can be seen in
this figure that the pH does not change significantly up until 40% slag replacement, but beyond this level, it decreases
8.9
@Seismicisolation
@Seismicisolation
to below 10 at 40% replacement. It can also be observed in this figure that the water soluble Cl-/OH- remains rather
stable up until 40% slag replacement, but increases rather rapidly beyond this level. The changes in both quantities
depend on the w/b ratio, as shown in Figure 6(d); the lower w/b leads to larger decrease in the pH and larger increase
in the water soluble Cl-/OH-. From these figures in can be stated that up to 40% slag replacement, the allowable water
soluble chloride limits recommended by ACI 318 can be used as a percentage of total cementitious materials; however,
the representation of limits as percentage of total cementitious material content is not applicable for larger replacement
levels. Additional analysis assuming lower allowable water soluble chloride limits as shown in Table 1 (0.08% by
ACI 2222 and 0.1% by ACI 201) did not change this general observation. For brevity, these cases are not presented.
(a) (b)
(c) (d)
Figure 6 — Effects of slag replacement on internal chloride chemical binding for a fully hydrated cement for (a) w/b
= 0.4; (b) w/b = 0.5, (c) w/b = 0.6; and (d) pH and water soluble Cl-/OH- for different w/b at different replacement
levels.
Figure 7 provides additional explanation as to why larger slag replacement levels can be tolerated when compared to
fly ash. This figure illustrates the C-S-H production and portlandite consumption levels with increasing slag
replacement percentages. It is clear in this figure that increasing slag replacement percentages increases C-S-H levels
in concrete up to about 40% replacement level, which corresponds to the depletion of porlandite from the mixture. It
should also be noted that since C-S-H content in concrete increases with increasing slag (up to 40% replacement),
compared to an OPC mixture, physical binding of chlorides will be larger in mixtures with slag replacement [14-16].
This additional binding, which is not studied in this research, will further reduce the water soluble (free) chlorides in
the pore solution. As a result, the results presented here with respect to water soluble Cl-/OH- values ca be considered
to be in the conservative side.
8.10
@Seismicisolation
@Seismicisolation
Figure 7 — C-S-H production with ground slag replacements.
Parametric investigation
The effects of alkali and SO3 contents on the bound and water soluble chloride levels, pH of the pore solution and
water soluble Cl-/OH- are presented in this section. All results shown here are for mixtures with w/b of 0.5 and are
based on the analysis grid given in Table 4.
Alkali content  Figures 8 and 9 illustrate the effect of alkali content on the bound and water soluble chloride levels,
the pH of the pore solution, and water soluble Cl-/OH- for different SCM replacement ratios. It is clear from Figure 8
that a higher alkali content generally reduces the binding capacity of the mixture and leads to higher water soluble
chloride amounts. One of the possible reasons behind this effect might be that high concentrations of dissolved sodium
oxide and potassium oxide in the pore solution cause a decrease in bound chlorides to maintain a high enough
concentration of free chlorides to satisfy electroneutrality in the solution. The SCMs that have relatively low alkali
contents, such as class F fly ash and slag, showed low sensitivity to the alkali content change, as shown in Figures
8(a) and 8(c), respectively. However, class C showed a large sensitivity to alkali content changes, as shown in Figure
8(b). This apparent sensitivity is mainly due to the fact that the range of analysis for alkali content was rather large for
class C fly ash (i.e., 0.8-4.8%). As shown in literature, the chemical composition of class C fly ash from different
producers show a high degree of variability [43], particularly with respect to alkali content. It is possible to see
equivalent alkali contents as high as 5% in class C fly ashes; therefore, for these materials, the free chloride
concentrations in concrete can be rather high. It can be observed in Figure 8(b) that the difference in free chloride
concentration between high (4.8%) and average (1.8%) equivalent alkali contents can as large as 50%.
However, it should also be noted that alkali contents in the mixture also affect the pH of the pore solution, hence the
water soluble Cl-/OH-. As shown in Figure 9, although the water soluble chloride connotation increases with increasing
alkali content, the pH also increases with the same; therefore, the effect of the alkali content on water soluble Cl-/OH-
is not very sensitive to the parameter. The observations made for the base case, as presented in the previous section,
are still valid for the range of alkali in this parametric investigation. In other words, the water soluble Cl-/OH- remains
rather stable up until 20% class F fly ash replacement, 30% class C fly ash replacement, and 40% slag replacement.
Beyond these replacement levels Cl-/OH- have been shown to increase for the mixtures that are studied in this
investigation. These observations also support that chloride content determinations and allowable chloride limits in
mixtures with SCMs should be made in terms of Cl-/OH-, rather than chloride contents alone.
8.11
@Seismicisolation
@Seismicisolation
(a) (b)
(c)
Figure 8 — Parametric study of the effect of alkali content on free and bound chloride levels for different SCM types:
(a) class F fly ash, (b) class C fly ash, (c) slag.
8.12
@Seismicisolation
@Seismicisolation
(a) (b)
(c)
Figure 9 — Parametric study of the effect of alkali content on pH and water soluble Cl-/OH- for different SCM types:
SO3 content  Figures 10 and 11 illustrate the effect of SO3 content on the bound and water soluble chloride levels,
the pH of the pore solution, and water soluble Cl-/OH- for different SCM replacement ratios. It is clear from these
figures that the SO3 content does not affect the free and bound chloride concertation, the pore solution pH, and water
soluble Cl-/OH- for both types of fly ash in the range studied in this investigation. However, there is a minor effect of
the parameter on the mixtures with slag replacement, as shown in Figures 10(c) and 11(c), particularly after 40% slag
replacement. The observations made for the base case, as presented in the previous section, are still valid for the range
of SO3 in this parametric investigation. In other words, the water soluble Cl-/OH- remains rather stable up until 20%
class F fly ash replacement, 30% class C fly ash replacement, and 40% slag replacement. Beyond these replacement
levels Cl-/OH- have been shown to increase for the mixtures that are studied in this investigation.
8.13
@Seismicisolation
@Seismicisolation
(a) (b)
(c)
Figure 10 — Parametric study of the effect of SO3 content on free and bound chloride levels for different SCM types:
8.14
@Seismicisolation
@Seismicisolation
(a) (b)
(c)
Figure 11 — Parametric study of the effect of SO3 content on pH and water soluble Cl-/OH- for different SCM types:
CONCLUSIONS
A thermodynamic modeling investigation was conducted to quantify free and bound chloride concentrations, as well
as the pH of pore solution, in mixtures produced with different types of blended cements and admixed chlorides. The
effects of replacing OPC with two types of fly ash (class C and F) and one type of (ground granulated blast furnace)
slag at different replacement levels were investigated. In addition, parametric studies were performed to investigate
the effects of w/b, alkali content and SO3 content on free and bound chloride concentrations, and the pH of the pore
solution. The following conclusions were drawn from this thermodynamic investigation for the specific compositions
of the OPC and SCMs that were used in this study. Cementitious materials, particularly SCMs, can show a high degree
of variability in chemical composition; therefore, extrapolating these conclusions to all types of OPC, fly ash, and slag
might require additional investigations.
1) Ignoring physical chloride binding mechanisms and only considering chemical binding, the water soluble Cl-/OH-
remained rather stable up to 20% class F fly ash, 30% class C fly ash, and 40% slag replacements. Beyond these
replacement levels Cl-/OH- were shown to increase, in some cases rather sharply, for the mixtures that were
studied in this investigation.
8.15
@Seismicisolation
@Seismicisolation
2) Although it is not studied in this research, physical binding is expected to reduce the water soluble (free) chlorides
in the pore solution; therefore, the range within which the water soluble Cl-/OH- remained stable, as described in
conclusion 1, can be considered to be on the conservative side.
3) The results of this investigation support the ACI committee recommendations for allowable chloride limits up to
20% class F fly ash, 30% class C fly ash, and 40% slag replacements. Beyond these replacement levels Cl-/OH-
were shown to increase, in some cases rather sharply, for the mixtures that were studied in this investigation.
4) Since SCM replacements affect the binding capacity, water soluble chlorides, and the pH of pore solution
simultaneously, chloride content determinations and allowable chloride limits in mixtures with SCMs should be
made in terms of Cl-/OH-, rather than chloride contents alone.
5) The w/b plays an important role in the binding capacity of the mixtures: the binding capacity decreases with
increasing w/b.
6) The SCMs that have relatively low alkali contents, such as class F fly ash and slag, showed low sensitivity to the
alkali content change. Class C fly ash, on the other hand, showed larger sensitivity to alkali content, particularly
after 30% slag replacement, mainly due to the fact that the range of analysis for available alkali content was rather
large for class C fly ash as shown in the literature.
7) The SO3 content did not significantly affect the free and bound chloride concertation, the pore solution pH, and
water soluble Cl-/OH- for both types of fly ash in the range studied in this investigation. The variability for slag
was larger, particularly after 40% slag replacement.
REFERENCES
1. Angst, U., et al., Critical chloride content in reinforced concrete—a review. Cement and Concrete Research,
2009. 39(12): p. 1122-1138.
2. Li, L. and A. Sagues, Chloride corrosion threshold of reinforcing steel in alkaline solutions-Open-circuit
immersion tests. Corrosion, 2001. 57(1): p. 19-28.
3. Martın-Pérez, B., et al., A study of the effect of chloride binding on service life predictions. Cement and
Concrete Research, 2000. 30(8): p. 1215-1223.
4. Ghods, P., et al., Electrochemical investigation of chloride-induced depassivation of black steel rebar under
simulated service conditions. Corrosion Science, 2010. 52(5): p. 1649-1659.
5. de Viedma, P.G., M. Castellote, and C. Andrade, Comparison between several methods for determining the
depassivation threshold value for corrosion onset. Journal De Physique Iv, 2006. 136: p. 79-88.
6. Liu, T. and R.W. Weyers, Modeling the dynamic corrosion process in chloride contaminated concrete
structures. Cement and Concrete Research, 1998. 28(3): p. 365-379.
7. Trejo, D. and P.J. Monteiro, Corrosion performance of conventional (ASTM A615) and low-alloy (ASTM
A706) reinforcing bars embedded in concrete and exposed to chloride environments. Cement and Concrete
Research, 2005. 35(3): p. 562-571.
8. Poursaee, A., A. Laurent, and C.M. Hansson, Corrosion of steel bars in OPC mortar exposed to NaCl, MgCl2
and CaCl2: Macro- and micro-cell corrosion perspective. Cement and Concrete Research, 2010. 40(3): p.
426-430.
9. 201.2R-08, Guide to durable concrete. 2008, American Concrete Institute. p. 49.
10. ACI-222R, Protection of Metals in Concrete Against Corrosion. 2001 (Re-approved 2010), American
Concrete Institute. p. 41.
11. ACI-318-11, Building code requirements for structural concrete and commentary. 2010, American Concrete
Institute.
12. Suryavanshi, A. and R.N. Swamy, Stability of Friedel's salt in carbonated concrete structural elements.
Cement and Concrete Research, 1996. 26(5): p. 729-741.
13. Yuan, Q., et al., Chloride binding of cement-based materials subjected to external chloride environment–a
review. Construction and Building Materials, 2009. 23(1): p. 1-13.
14. Ramachandran, V.S., Possible states of chloride in the hydration of tricalcium silicate in the presence of
calcium chloride. Matériaux et Construction, 1971. 4(1): p. 3-12.
15. Luping, T. and L.-O. Nilsson, Chloride binding capacity and binding isotherms of OPC pastes and mortars.
Cement and concrete research, 1993. 23(2): p. 247-253.
16. Beaudoin, J.J., V.S. Ramachandran, and R.F. Feldman, Interaction of chloride and C-S-H. Cement and
17. Rasheeduzzafar, Influence of cement composition on concrete durability. ACI Materials Journal, 1992. 89(6).
8.16
@Seismicisolation
@Seismicisolation
18. Zibara, H., Binding of external chlorides by cement pastes. 2001.

19. Suryavanshi, A., J. Scantlebury, and S. Lyon, Mechanism of Friedel's salt formation in cements rich in tri-
calcium aluminate. Cement and concrete research, 1996. 26(5): p. 717-727.
20. Kulik, D.A., et al., GEM-Selektor geochemical modeling package: revised algorithm and GEMS3K
numerical kernel for coupled simulation codes. Computational Geosciences, 2013. 17(1): p. 1-24.
21. Wagner, T., et al., GEM-Selektor geochemical modeling package: TSolMod library and data interface for
multicomponent phase models. Canadian Mineralogist, 2012. 50(5): p. 1173-1195.
22. Kosakowski, G. and N. Watanabe, OpenGeoSys-Gem: A numerical tool for calculating geochemical and
porosity changes in saturated and partially saturated media. Physics and Chemistry of the Earth, Parts
A/B/C, 2014. 70: p. 138-149.
23. Kulik, D.A. and M. Kersten, Aqueous Solubility Diagrams for Cementitious Waste Stabilization Systems: II,
End‐Member Stoichiometries of Ideal Calcium Silicate Hydrate Solid Solutions. Journal of the American
Ceramic Society, 2001. 84(12): p. 3017-3026.
24. Kulik, D.A. and M. Kersten, Aqueous solubility diagrams for cementitious waste stabilization systems. 4. A
carbonation model for Zn-doped calcium silicate hydrate by Gibbs energy minimization. Environmental
science & technology, 2002. 36(13): p. 2926-2931.
25. Lothenbach, B. and F. Winnefeld, Thermodynamic modelling of the hydration of Portland cement. Cement
and Concrete Research, 2006. 36(2): p. 209-226.
26. Matschei, T., B. Lothenbach, and F.P. Glasser, Thermodynamic properties of Portland cement hydrates in
the system CaO-Al2O3-SiO2-CaSO4-CaCO3-H2O. Cement and Concrete Research, 2007. 37(10): p. 1379-
1410.
27. Lothenbach, B., et al., Thermodynamic modelling of the effect of temperature on the hydration and porosity
of Portland cement. Cement and Concrete Research, 2008. 38(1): p. 1-18.
28. Moschner, G., et al., Solubility of Fe-ettringite (Ca6[Fe(OH)(6)](2)(SO4)(3).26H(2)O). Geochimica Et
Cosmochimica Acta, 2008. 72(1): p. 1-18.
29. Schmidt, T., et al., A thermodynamic and experimental study of the conditions of thaumasite formation.
Cement and Concrete Research, 2008. 38(3): p. 337-349.
30. Moschner, G., et al., Solid solution between Al-ettringite and Fe-ettringite (Ca-6[Al1-
xFex(OH)(6)] (2)(SO4)(3).26H(2)O). Cement and Concrete Research, 2009. 39(6): p. 482-489.
31. Kulik, D.A., Improving the structural consistency of C-S-H solid solution thermodynamic models. Cement
32. Dilnesa, B., et al., Iron in carbonate containing AFm phases. Cement and Concrete Research, 2011. 41(3):
p. 311-323.
33. Lothenbach, B., L. Pelletier-Chaignat, and F. Winnefeld, Stability in the system CaO-Al2O3-H2O. Cement
34. Dilnesa, B.Z., et al., Stability of monosulfate in the presence of iron. Journal of the American Ceramic
Society, 2012. 95(10): p. 3305-3316.
35. Dilnesa, B.Z., et al., Synthesis and characterization of hydrogarnet Ca 3 (Alx Fe1− x)2 (SiO4)y(OH)4(3− y).
Cement and Concrete Research, 2014. 59: p. 96-111.
36. Balonis, M. and F.P. Glasser, The density of cement phases. Cement and Concrete Research, 2009. 39(9): p.
733-739.
37. Balonis, M., The Influence of Inorganic Chemical Accelerators and Corrosion Inhibitors on the Mineralogy
of Hydrated Portland Cement Systmes. 2010, University of Aberdeen Aberdeen, UK.
38. Balonis, M., et al., Impact of chloride on the mineralogy of hydrated Portland cement systems. Cement and
39. Matschei, T., B. Lothenbach, and F. Glasser, The AFm phase in Portland cement. Cement and Concrete
Research, 2007. 37(2): p. 118-130.
40. Deschner, F., et al., Hydration of Portland cement with high replacement by siliceous fly ash. Cement and
41. Lothenbach, B., et al., Effect of temperature on the pore solution, microstructure and hydration products of
Portland cement pastes. Cement and Concrete Research, 2007. 37(4): p. 483-491.
42. Deschner, F., et al., Effect of temperature on the hydration of Portland cement blended with siliceous fly ash.
Cement and Concrete Research, 2013. 52: p. 169-181.
43. Shehata, M.H. and M.D. Thomas, The effect of fly ash composition on the expansion of concrete due to
alkali–silica reaction. Cement and Concrete Research, 2000. 30(7): p. 1063-1072.
8.17
@Seismicisolation
@Seismicisolation
44. Hong, S.-Y. and F. Glasser, Alkali sorption by C-S-H and C-A-S-H gels: Part II. Role of alumina. Cement
45. Hong, S.-Y. and F. Glasser, Alkali binding in cement pastes: Part I. The CSH phase. Cement and Concrete
Research, 1999. 29(12): p. 1893-1903.
46. Wang, X.-Y. and H.-S. Lee, Modeling the hydration of concrete incorporating fly ash or slag. Cement and
47. Kosmatka, S.H. and M.L. Wilson, Design and control of concrete mixtures. 15th Edition ed. 2011: Portland
Cement Association (PCA).
48. ASTM-C311, Standard test methods for sampling and testing fly ash or natural pozzolans for use in portland-
cement concrete. 2013.
8.18
@Seismicisolation
@Seismicisolation
SP-308—9
EFFECT OF FLY ASH AND SILICA FUME ON TIME TO CORROSION INITIATION FOR SPECIMENS
EXPOSED LONG TERM TO SEAWATER
Francisco J. Presuel-Moreno, Eric I. Moreno
Synopsis: The aim of this study was to determine the chloride threshold concentration of carbon steel rebar embedded
in high performance concrete under exposure conditions relevant to the substructure of coastal bridge in Florida. The
experiments were based upon a series of reinforced and non-reinforced concrete specimens that contained 1) 20, 35
and 50 percent cement replacement by fly ash, 2) 6, 15 and 27 percent cement replacement by silica fume, and 3)
control specimens (no pozzolanic admixture). All specimens had a target w/cm ratio of 0.37. The specimens have been
exposed to one week wet - one week dry ponding cycle with natural seawater since January, 1995. Rebar potential
values were monitored with time in order to determine when corrosion initiated. The rebar of several specimens
activated after 15.6 to 17 years of exposure, and selected specimens were terminated for forensic examination and
extent of corrosion. Cores were obtained to determine the extent of chloride ingress, the apparent diffusion coefficient
(Dapp) and concrete resistivity. The chloride concentration above the rebar trace was also measured on most of the
terminated specimens. The value for Dapp was correlated against the corresponding measured resistivity. Specimens
with 50 percent FA had the lowest Dapp but also the lowest chloride threshold.
Keywords: Chloride Diffusivity, Chloride Threshold, Fly Ash, Resistivity, Silica Fume, Time to Corrosion Initiation,
9.1
@Seismicisolation
@Seismicisolation
Francisco J. Presuel-Moreno and Eric I. Moreno
ACI member Francisco J. Presuel-Moreno is an Associate Professor at Florida Atlantic University, Department of
Ocean and Mechanical Engineering. He received his Ph.D. in Engineering Science from University of South Florida.
His research interests include metallic corrosion, corrosion of reinforcing steel in concrete, durability of reinforced
concrete structures, non-destructive testing, and computational modeling.
ACI member Eric I. Moreno is a Professor at Universidad Autonoma of Yucatan, Civil Engineering Department. He
received his Ph.D. in Civil Engineering from University of South Florida. His research interests include durability of
reinforced concrete structures, carbonation, and corrosion of reinforcing steel in concrete.
INTRODUCTION
Steel reinforced concrete is one of the most used materials in construction. However, deterioration of reinforced
concrete structures has been a concern for years; and corrosion of reinforcing steel is the main cause of this
deterioration. Once corrosion has initiated, cracks in concrete form and propagate due to tensile stress forces produced
by the accumulated solid corrosion products during the propagation stage. These products occupy a larger volume
compared to that of non-corroding rebar. A critical area in a partially-immersed reinforced concrete bridge structure
exposed to marine weather conditions (i.e., direct exposure to sea water) is the splash zone, where the most severe
corrosion conditions usually develops with time.
Many publications have reported the chloride resistance of concrete made with binary or ternary concrete mixes. In
the present effort an additional set of results is reported based upon measurements of corrosion potential, time to
corrosion, chloride concentration above the rebar trace, and resistivity of mature (>15 year) wet concrete. Based upon
the chloride measurements, an apparent chloride diffusivity (Dapp) was calculated. Also, a correlation between wet
concrete resistivity (ρwet) and Dapp was obtained. Both ρwet and Dapp values were determined from tests performed on
cores obtained from terminated specimens that were exposed outdoors for over 15 years. Currently, the remaining
specimens have been subjected to more than 20 years of exposure to natural seawater.
Most modern structures are built with concrete that has a low water-to-cementitious ratio (w/cm). However, many of
the published experimental results use relatively high water-to-cement ratio as a way to reduce the time to corrosion
initiation. This paper reports results obtained from concrete specimens prepared with a relatively low water-to-
cementitious ratio of ~0.37.
EXPERIMENTAL
Specimens
A set of 72 specimens was made based upon concrete mix designs with different fly ash, silica fume and calcium
nitrite contents, as listed in Table 1. Fly ash and silica fume were added as cement replacement, see formula below
Table 1. All mixes were made with Holnam Type H Portland cement (sulfate resistant cement) and a water-to-
cementitious ratio of 0.37. The coarse aggregate was crushed limestone and the fine aggregate was Florida silica sand.
The slabs were cast at the FDOT Materials Laboratory in Gainesville, Florida, in October, 1994 and were subsequently
delivered to the Marine Corrosion Laboratory at Florida Atlantic University. The different mix designs and concrete
properties are listed in Tables 2 through 4. Silica fume slurry was used; and part of the slurry mass on each mix counted
towards the water.
The specimens were of two types — reinforced and non-reinforced, where the former contained five number three
(9.5 mm diameter) steel bars and the latter contained no embedded steel. The reinforced specimens comprised the
corrosion test program, per se. Most of the reinforced specimens were 12.7 cm (5 in) high with 1.9 cm (0.75 in) cover
over the top bars, while a limited number of specimens were 15.2 cm (6 in) thick with 3.2 cm (1.25 in) cover. Table 5
lists the number of reinforced specimens cast for each mix and corresponding height/cover. The reinforced slab
geometry is as illustrated in Figure 1.
9.2
@Seismicisolation
@Seismicisolation
Effect of Fly Ash and Silica Fume on Time to Corrosion Initiation
for Specimens Exposed Long Term to Seawater
Table 1 — Mix design and admixture content*

Mix Design Fly Ash Silica Fume Calcium Nitrate
Designation percent Percent l/m3
AO 0 0 0
CO 0 0 19.8
FA20 20 0 0
FA35 35 0 0
FA50 50 0 0
FA35-N 35 0 19.8
SF06 0 6 0
SF15 0 15 0
SF27 0 27 0
SF15-N 0 15 9.9
* %FA and %SF content are % of FA/(c+FA) and SF/(c+SF) respectively
Natural Seawater
Pond
Reinforcing Steel
Bars (5)
1.9 cm (0.75 in)/

12.5 cm (5 in)/ 3.2 cm (1.25 in)
15.2 cm (6 in)
45.7 cm (18 in)
35.6 cm (14 in)
Electrical Connection
Figure 1 — Schematic illustration of the specimen geometry
Prior to exposure testing, the specimens were inverted (as-cast face down); and a polycarbonate bath was mounted on
what became the top surface. The exposure took place outdoors most of the time, but early the specimens were exposed
indoors for a couple of years. A one week wet — one week dry ponding cycle was instituted in January, 1995 using
fresh natural sea water. The typical conductivity of seawater measured at the FAU laboratory is ~ 40 mS/cm (101.6
mS/in). A rectangular plastic plate was placed on top of the specimens to prevent rainfall or minimize evaporation
during the rainy/dry seasons, respectively. As shown in Figure 1, each of the bottom bars was electrically connected
to one of the top bars so that a macro-cell resulted between the two. An electrical switch (not shown) between these
connected bars permitted measurement of macro-cell current. The top-middle bar was left electrically-isolated to
obtain data independent of any macro-cell action. Macrocell measurements were performed up to day 3000, and results
from these measurements will not be presented in here.
9.3
@Seismicisolation
@Seismicisolation
Table 2 — Mix design AO and CO

AO CO
Cement, kgs 113.5 113.4
Calcium Nitrite, kgs 0 5.2
Water, kgs 33.3 28.6
Coarse Aggregates, kgs 286.8 286.8
Coarse Aggregates, % excess moisture 2.06 2.5
Fine Aggregates, kgs 215.6 215.6
Fine Aggregates, % excess moisture 1.3 1.5
Unit Weight, kgs/m3 2,292.5 2,276.4
w/cm ratio 0.37 0.367
RCP Average at 91 Days, C 4896 (High) 6285 (High)
Strength Avg. 28 days (MPa) 44.2 48.1
Cementious per unit volume, kgs/m3 399 394.4
Note: 1 kg/m3 = 1.6842 lb/yd3 , 1 kg = 2.205 lbs, 1MPa – 145.03 psi
Table 3 — Mix design FA20 FA35, FA50, FA35N

FA20 FA35 FA50 FA35N
Cement, kgs 90.8 73.8 56.7 73.8
Fly Ash, kgs 22.7 39.7 56.7 39.8
Calcium Nitrite, kgs 0 0 0 5.2
Water, kgs 27.6 27.8 28.0 24.4
Coarse Aggregates, kgs 287.4 287.4 287.4 288.4
Coarse Aggregates, % excess moisture 2.96 2.3 2.3 2.9
Fine Aggregates, kgs 212.4 206.0 199.6 203.8
Fine Aggregates, % excess moisture 2.68 3.68 3.68 3.0
Unit Weight, kgs/m3 2,263.6 2,247.6 2,231.6 2,231.6
w/cm ratio 0.367 0.37 0.37 0.363
RCP Avg. 91 Days, C 989 713 731 NA
Strength Avg. 28 days (MPa) 45.5 42.7 36.3 34.2
Cementious per unit volume, kgs/m3 399 400.4 401 396
Note: 1 kg/m3 = 1.6842 lb/yd3 , 1 kg = 2.205 lbs, 1MPa – 145.03 psi. Type F fly ash was used.
Table 4 — Mix design SF06, SF15, SF27, SF15N

SF06 SF15 SF27 SF15N
Cement, kgs 110.2 104.4 96.6 104.4
Silica Fume, kgs 7.2 19 35.6 19
Calcium Nitrite, kgs 0 0 0 2.68
Water, kgs 25 18.8 10.2 16.6
Coarse Aggregates, kgs 288.4 288.4 288.4 288.4
Coarse Aggregates, % excess moisture 2.65 2.7 2.9 2.9
Fine Aggregates, kgs 217.4 215.2 212.2 215.2
Fine Aggregates, % excess moisture 2.680 2.68 2.68 2.7
Unit Weight, kgs/m3 2,279.6 2,273.2 2,262.0 2,265.2
w/cm Ratio 0.37 0.367 0.368 0.365
RCP Avg. 91 Days, C 2061 720 598 868
Cementious per unit volume, kgs/m3 397 397.6 399 396
Note: 1 kg/m3 = 1.6842 lb/yd3 , 1 kg = 2.205 lbs, 1MPa – 145.03 psi. Silica Fume Slurry was used, part of the slurry
mass counted towards the water.
9.4
@Seismicisolation
@Seismicisolation
Table 5 — Number of samples with reinforcements per mix and height/cover

Height/ Cover AO CO FA20 FA35 FA50 FA35N SF06 SF15 SF27 SF15N
12.7/1.9 cm (5/0.75 in) 5 4 4 4 4 3 3 3 3 3
15.2 /3.2 cm (6/1.25 in) 3 3 3 3
Potential measurements
Corrosion state of the reinforcement, i.e., determination whether the embedded steel was passively or actively
corroding, was assessed as a function of time by macro-cell current (first 3000 days) and corrosion potential (half-cell
potential) measurements; the latter being measured in general accord with standard practice during the wet cycle. In
the present experiments, potentials were recorded using a high resistance voltmeter that measured the rebar potential
against that of a saturated calomel electrode (SCE) with a pre-wetted sponge being used as an electrical junction
between the concrete surface and the reference electrode. The electrode was placed above the rebar at the center of
the rebar length. There are 77 mV difference between a saturated copper sulfate electrode and a saturated calomel
electrode. Thus, -350 mVcse and -200 mVcse correspond to -273 mVsce and -123 mVsce respectively. These are the
potential values typically used to determine corrosion probability.
pH
Selected non-reinforced concrete specimens were cored and the pH measured via a leaching technique. These
measurements were performed when the concrete age was ~1100 days. The pH was measured during an early stage
of this project. A description of the method used and additional details for the testing conditions can be found in
reference [1]. The pH was measured both under laboratory conditions and also under a nitrogen gas atmosphere to
prevent carbonation of the leached solution.
Terminated specimens: diffusivity, resistivity and visual examination

Some of the non-reinforced specimens served as companion specimens from which cores were extracted at various
times so that the ingress of chlorides could be monitored without destructively altering the reinforced specimens.
Concrete cores from non-reinforced specimens were obtained after 1200 days and chloride profiles obtained and Dapp
determined1. After 5700, 6000, and 6250 days of exposure it was decided to terminate selected reinforced specimens.
On day 5700, at least one specimen per mix type was terminated even if the rebars were not necessarily undergoing
corrosion. Before cutting the specimens open, these specimens were also cored. A core drill-bit of 5 cm (2 in.) nominal
diameter was used. Four cores were typically drilled from each specimen, with two cores sliced for chloride profiles
and the other two cores were used to measure the concrete resistivity. After segmenting the concrete sample along the
rebar length, the rebars at the top row were exposed and the surface condition documented. The pieces of concrete
above the rebar were milled to collect concrete powder (1 to 2 mm depth) along the rebar trace. Caution was used to
prevent milling over rust spots. These concrete samplings were then used to measure the chloride concentration
representative of the concentration above the rebar trace. The chloride concentration determination was performed
using the FDOT method2 or a modified approach that used a smaller amount of concrete powder when necessary. The
FDOT method measures the total acid soluble chlorides.
As indicated in the previous paragraph, two (or one) concrete cores per terminated specimen were selected to assess
the resistivity (ρwet) of the concrete (when the concrete had reached a high moisture content). The resistivity
measurements follow a FDOT method3. This ρwet was achieved by placing the cores in a high humidity chamber (~95%
RH) for several months. Two different measurements were taken on these 4.5 cm (1.77 in) diameter cores: 1) resistivity
measurements along the longitudinal direction of the concrete core ρ vs. time (eight readings 90 degrees apart two
times around), and 2) weight percent change vs. time (from initial recorded weight). Both measurements were
performed at laboratory conditions with temperature ~21 ± 2 ºC. These two sets of measurements started about a week
after coring which might have allowed the cores to become drier when compared to outdoor exposure conditions;
however, the cores were obtained using a wet-coring procedure. The values reported correspond to the minimum
values observed after the mass of the concrete core stop changing. The resistivity values reported include a cell
constant correction.
9.5
@Seismicisolation
@Seismicisolation
Potential vs. time

Figures 2 to 7 show measured potential plots vs. time. Each plot has 3 series. Series ‘E1 on’ and ‘E3 on’ correspond
to the measured potential of the coupled rebars on the left and on the right on each specimen, respectively. E2
corresponds to the measured potential on the center rebar. Figure 2 shows the potential evolution on specimens AO2
and AO3. Both specimens were terminated around day 5700. It can be observed that corrosion initiated on the three
rebars on specimen AO3. But based on the terminal potential values, none of the rebars on specimen AO2 appeared
active. However, at around day 4500, there was a momentary potential drop; the potential of all rebars recovered
(shifted to more positive potential values) soon after this event. Upon opening specimen AO2, the three rebars showed
some red rust spots (rather than black/dark green corrosion products characteristic of on-going corrosion under moist
conditions). For AO specimens with 1.9 cm (0.75 in) concrete cover, seven out of 12 rebars became active of those
terminated, and 1 of 3 rebars from the non-terminated specimen appeared active.
100 E1 on
100 E1 on
E2 AO2 E2 AO3
E3 on E3 on
0 0
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1000 2000 3000 4000 5000 6000 7000
Time (days) Time (days)
Figure 2 — Typical potential evolution on AO specimens (1.9 cm (0.75 in) cover)
Figure 3 shows the potential evolution on specimen AO7, which is one of the specimens with a cover of 3.2-cm (1.25
in). A potential transient toward more negative potential values started around day 4200 (with a momentary drop and
recover). Rebar E1 experienced a potential drop on day 6800 and has remained at values < -200 mVsce, suggesting
that corrosion has initiated. The potential values for rebars E2 and E3 were close to -50 mVsce. The potential values
of the rebars in the other two specimens showed similar behavior to what was observed for E2 and E3 on specimen
AO7. Excursions to values more negative than -100 mVsce have been periodically observed, but usually return to a
trend of gradual potential decay. The appendix of a recent report contains additional potential vs. time plots for all
specimens up to day 7000.
Figure 4 shows the potential evolution on specimens CO-2 and CO-3 from the CO group. In specimen CO-2, two
rebars (‘E1-on’ and E2) have become active, currently showing potential values of -400 mVsce. Specimen CO-2 has
not been terminated. With respect to rebars in CO-3 specimen, ‘E1 on’, i.e., the potential of the left-coupled rebars
reached ~-500 mVsce. The other two (‘E2’ and ‘E3 on’) showed a more modest potential drop, ‘E3 on’ approached -
200 mVsce, whereas E2 reached a value of -110 mVsce. CO-3 specimen was terminated and all rebars showed
corrosion spots. One other CO specimen was terminated but no rebars were active, and the fourth CO specimen has
not been terminated and the rebars potential have experienced a modest potential decay similar to that described above
for AO-5 specimen.
9.6
@Seismicisolation
@Seismicisolation
100
E1 on
E2
AO7
E3 on
0
E (mVsce) -100
-200
-300
-400
-500
0 1500 3000 4500 6000 7500
Time (days)
Figure 3 — Typical potential evolution on AO specimen with 3.2 cm (1.25 in) concrete cover
100 100
E1 on
E1 on
CO-2 E2 CO-3
E2
E3 on
E3 on
0 0
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500
-500
0 1000 2000 3000 4000 5000 6000
0 1500 3000 4500 6000 7500
Figure 4 — Typical potential evolution on CO specimens. (1.9 cm (0.75 in) Cover)
Figure 5 shows typical potential evolution on FA specimens with 1.9 cm (0.75 in) cover and Figure 6 shows typical
potential evolution on FA specimens with 3.2 cm (1.25 in) cover. Figure 5 shows the potential evolution for FA20-3,
FA35-6, FA50-5 and FA35N-1 specimens. The potential of the three rebars in FA20-3 have shifted towards more
negative values. The three rebars on FA35-6 (right rebar E3 section terminated ~ day 6250) had initiated corrosion
based on the measured potential and empirical observation. The potential of all rebars from specimen FA50-5 had
shifted to more negative values, indicating that corrosion likely initiated. On FA35N-1, all rebars also experience a
negative potential shift. Moreover, the potential of the left and right rebar pairs monotonically decayed to values
approx. -200 mVsce and then a more abrupt potential drop took place. No significant potential shift was observed on
specimens with 3.2 cm (1.25 in) cover for groups FA20 (e.g. FA20-5 in Figure 6) and FA35 (e.g. FA35-3 in Figure
6). The gradual potential decay is on-going with a few momentary excursions to more negative values. Using the
potential criteria, several rebars from the FA50 group have become active, three out of nine rebars with 3.2 cm (1.25
in) concrete cover (e.g., FA50-1 in Figure 6) and ten out of twelve rebars with 1.9 cm (0.75 in) concrete cover.
Similarly, eight out of nine rebars prepared with FA35N mix have become active. For the FA35N group no specimen
was prepared with 3.2 concrete cover.
9.7
@Seismicisolation
@Seismicisolation
100 E1 on
100 E1 on
E2 FA20-3 E2 FA35-6
E3 on E3 on
0 0
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1500 3000 4500 6000 7500
100 E1 on
100 E1 on
E2
FA50-5
E2
E3 on E3 on
0 0
FA35N-1
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1000 2000 3000 4000 5000 6000 7000
Figure 5 — Typical potential evolution on specimens with FA and 1.9 cm (0.75 in) cover
Figure 7 shows the rebar potential evolution on selected SF specimens (one per concrete mix). Five rebars out of nine
became active from the SF06 specimens by day 7500. Six rebars out of nine became active by day 7500 for the
specimens of group SF15, two are from specimens not yet terminated. The non-corroding rebars corresponded to a
specimen that was terminated on day 5700 before corrosion initiated. Although the last set of potential values
measured on specimen SF15-1 ranged between -150 and -180 mVsce, the top rebars on the left and right side showed
black/dark green rust spots upon forensic visual inspection. One out of nine rebars on SF27 specimens and no rebar
from those in the SF15 group have become active based on potential transients. All specimens for SF06, SF15, SF27
and SF15N have a concrete cover of 1.9 cm (0.75 in).
9.8
@Seismicisolation
@Seismicisolation
100 E1 on
100
E1 on
E2 E2
E3 on E3 on
0 0
FA20-5 FA35-3
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1500 3000 4500 6000 7500 0 1500 3000 4500 6000 7500
100 100
E1 on E1 on
E2 E2
E3 on E3 on
0 0
FA50-1
FA50-3
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1500 3000 4500 6000 7500
Figure 6 — Typical potential evolution on specimens with FA and 3.2 cm (1.25 in) cover
In the passive state, the potential ranged from 50 to -100 mVsce for AO and CO groups, and tended to range between
100 and -50 mVsce; i.e., even nobler (more positive potential) for those with supplementary cementitious material
admixtures. These range of values correspond to readings taken after 2000 days of exposure. Angst4, Hartt1 and others
have reported that an asymptotic potential increase occurs with time, in here during the first 400 to 600 days and is
most likely due to passive layer growth and the associated reduction in passive current but could be also influenced
by the change in resistivity as the concrete hydrates. The lower pH (see section below) for concrete with admixtures
(FA or SF) might have contributed to an additional potential increase. In the plots shown above, a saturation value
was usually observed by day 3000. In several instances the potential remained at the saturation value for some
additional time, but in many a gradual potential decay was observed (with the slope being as little as 20 mV in 1000
days). After this initial decay the rebars experienced a potential drop. In some cases the rebar potential returned to
more positive values (repassivation). Angst4 suggested that this might occur in instances in which the chloride
concentration at the rebar can be sufficiently high to initiate pitting but might not necessarily be able to sustain pit
growth. Such events can be observed on Figures 2, 3, 5 and 7. Interestingly, the duration of the repassivation lasted
from a few months (e.g., AO3, CO3, SF06-1 in Figures 2, 4 and 7, respectively) to a few years (e.g., FA20-3 and
FA35-6 in Figure 5). The potential of the rebars then continue to decay, usually with a steeper slope. Eventually, a
significant and abrupt potential drop is observed with the rebar potential values remaining or shifting to even more
negative values. This large potential drop is typically associated with corrosion initiation. Angst4 suggests identifying:
the time to corrosion initiation and the time to stable corrosion to determine when corrosion is propagating at a stable
rate. In this investigation, a gradual potential decay, was observed from a relatively stable passive potential value, then
followed by the more abrupt potential drop. In several specimens only one rebar (out of three) experienced a stable
potential drop that can be identified as indicative of corrosion initiation and with maintenance of this condition. In
several instances it was decided to terminate the specimen upon one of the rebars reaching potential values approx. -
160 to -180 mVsce and with potential oscillation (this applied for some of the specimens terminated at day 6000 or
9.9
@Seismicisolation
@Seismicisolation
6250) and corrosion was confirmed via visual inspection. Not all rebars that experience corrosion had a single
transition from passive to active. In several cases momentary potential drops were observed with the potential
subsequently recovering (i.e. shifted to more positive values) after some time. The reason for the recovery to more
positive potential values is likely a combination of environmental (Temperature, seasonal changes, e.g., rain events)
and electrochemical conditions (e.g., not being able to sustain the incipient corrosion).
100 E1 on
100
E1 on
E2 E2
E3 on E3 on
0 0
SF06-1 SF15-1
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1000 2000 3000 4000 5000 6000 7000
100 E1 on
100 E1 on
E2 E2
E3 on E3 on
0 0
SF27-2 SF15N-3
-100 -100
E (mVsce)
E (mVsce)
-200 -200
-300 -300
-400 -400
-500 -500
0 1000 2000 3000 4000 5000 6000 7000 0 1000 2000 3000 4000 5000 6000 7000
Figure 7 — Typical potential evolution on specimens with SF
Visual inspection on terminated samples

All rebars from the terminated specimens were subjected to visual inspection. Figure 8 shows typical photos for six
instances in which corrosion actively took place. The pictures correspond to rebars from samples AO-3, CO-3, FA50-
4, FA35N-3, SF06-3 and SF15-1, respectively, and showcase worst case scenarios (i.e., some of the larger observed
corrosion spots). Only one of the terminated specimens suffered from delamination, AO3, although three of the
remaining specimens now show delamination. A recent report5 contains photographs for all terminated specimens
grouped according to mix composition. In some cases, either no corrosion was found or only red-rust and in others
dark gray spots (but different to those associated with on-going corrosion).
9.10
@Seismicisolation
@Seismicisolation
Figure 8 — Example of terminated rebars undergoing corrosion
Four AO specimens were terminated and showed different amounts of corrosion. Rebars from specimens AO3 and
AO8 showed significantly larger corrosion spots as shown for AO3 in Figure 8. Corrosion spots on rebars from AO1
and AO2 specimens were mainly red rust of various sizes. Terminated specimens from FA20 and FA35 groups only
showed red-rust or dark gray spots, as is not typically associated with on-going corrosion.
Chloride profiles and concentration above the rebar trace

This section presents selected chloride concentration profiles obtained from slicing the cores chosen for chloride
analysis. Results are also presented for chloride concentration just above the rebar trace. The chloride concentration
shown is the average concentration from triplicate measurements per slice or the average of two measurements for the
concentrations corresponding to locations above the rebar trace. The build-up effect due to rebar presence has been
reported before by Sagues6 and Hartt7 from modeling and experimental results, respectively. The rebar interrupts the
pathway for chloride transport. The rebar is impervious to the passage of chlorides, which in turn, causes the chloride
concentration to build-up faster in the region adjacent to the rebar closest to the exposed surface. Figures 9, 10 and 11
show a primary and a secondary y-axis: on the left the scale values are shown in kg/m3 and on the right y-axis the
chloride amounts are shown in %cm (assuming a target cementitious content of 400 kg/m3). Figure 9 shows the
chloride concentration profiles for AO and CO specimens. Chloride concentrations corresponding to concrete
locations just above the rebar trace ranged from 5 to 11.5 kg/m3 and are inside a drawn ellipse. Figures 10 and 11
show the chloride profiles obtained from FA and SF specimens, respectively. The results shown provide further
evidence that for some geometries the rebar presence effect on chloride build up should not be neglected, particularly
for thinner concrete covers. CT/CS is the ratio between the chloride threshold (CT) and the chloride surface
concentration (CS). The derating factor for the tested geometry ranges from 0.57 (assuming CT/CS=0.5) to 0.76
(assuming CT/CS=0.1) for a ~0.5 ratio corresponding to the rebar diameter to cover of the smaller cover. The ratio of
the rebar diameter to concrete cover for the 3.2 cm (1.25 in) concrete cover is 0.3 and the corresponding derating
factors are 0.7 (CT/CS=0.5) and 0.79 (CT/CS=0.1), thus a more modest reduction in time to reach CT is predicted but
non-negligible. Two of the FA specimens terminated shown in Figure 10 were 15.2 cm (6 in) thick with 3.2 cm (1.25
in) cover; all other terminated specimens had 1.9 cm (0.75 in) concrete cover.
9.11
@Seismicisolation
@Seismicisolation
Figure 9 — Chloride profiles and chloride concentration at the rebar trace, AO and CO specimens. (Obtained after
5700 days of exposure)
20 5
FA20-1-C1
FA20-1-C3
FA35-1-C1
FA35-1-C3
16 FA50-2-C1
4
FA50-2-C3
FA35N-2-C1
FA35N-2-C3
12 3
%cm
kg/m3
8 2
4 1
0 0
0 1 2 3 4
distance (cm)
Figure10 — Chloride profiles and chloride concentration at the rebar trace, FA specimens. (Obtained after 5700
days of exposure)
9.12
@Seismicisolation
@Seismicisolation
Figure 11 — Chloride profiles and chloride concentration at the rebar trace, SF specimens. (Obtained after 5700
days of exposure)
Table 6 — Chloride concentrations at the rebar trace (kg/m3)

Specimen Left Center Right Specimen Left Center Right
AO1 NA Y NA NA NA Y FA20-1 4.3 N 3.4 N 4.7 N
AO2 8.0 Y 7.9 Y 10.0 Y FA20-3 6.4 Y 2.9 Y 3.3 Y
AO3 6.9 Y 7.9 Y 6.7 Y FA35-1 0.3 N 0.3 N 0.2 N
AO8 5.8 Y 7.4 N 8.9 Y FA35-6 C NA C NA 5.5 Y
CO1 6.8 N 4.8 Y 5.4 N FA50-1 1.2 Y NA Y 1.4 Y
C03 6.4 Y 8.4 Y 8.4 Y FA50-2 0.4 N 0.4 N 0.3 N
SF06-1 7.2 Y 8.1 N 6.4 N FA50-4 3.1 Y C NA C NA
SF06-3 14.8 Y 14.9 Y 11.6 Y FA50-5 2.2 Y 1.6 Y 1.9 Y
SF15-1 8.2 Y 7.9 Y 6.8 Y FA50-7 3.5 Y 1.3 Y 3.3 Y
SF15-2 6.1 Y 5.0 Y 5.2 Y FA35N-1 8.9 Y 7.6 Y 6.9 Y
SF27-2 2.6 N 2.4 Y 2.5 N FA35N-2 9.0 Y 8.9 Y 9.4 Y
SF15N-3 4.1 N 5.7 N NA N FA35N-3 9.9 Y C NA C NA
Y- Corrosion was confirmed. N – No corrosion observed. NA – Not Available,
C – Continues to be exposed.
Note: 1 kg/m3 = 1.6842 lb/yd3
Table 7— Minimum and average [Cl-] kg/m3 and in [%cm] at the rebar trace on those corroding
Mix ID Minimum Average Mix ID Minimum Average
AO 5.8 [1.45%] 7.7 [1.92%] CO 6.4 [1.62%] 7.7 [1.95%]
FA20 2.9 [0.75%] 4.2 [1.05%] SF06 7.2 [1.81%] 12.1 [3.04%]
FA35 5.5 [1.37%] 5.5 [1.37%] SF15 5.0 [1.26%] 6.5 [1.63%]
FA50 1.4 [0.29%] 2.2 [0.54%] SF27 2.4 [0.60%] 2.4 [0.60%]
FA35N 6.9 [1.74%] 8.7 [2.2%] SF15N >5.7 [>1.46%] >5.7 [>1.46%]
Note: 1 kg/m3 = 1.6842 lb/yd3
9.13
@Seismicisolation
@Seismicisolation
Table 6 shows the chloride concentration measured at the rebar trace for those forensically analyzed. Table 7 shows
the minimum and average chloride concentration at the rebar trace grouped per concrete mix, expressed in kg/m3 and
also inside brackets in %cm. The cementitious amount in kg/m3 per mix type are included in Tables 2 to 4, these
values were used to calculate the %cm values shown in Table 7. The average values shown on Table 7 were obtained
from those where corrosion was visually confirmed. The chloride concentrations displayed in Table 6 and Table 7 are
not chloride thresholds (CT), per se, as some specimens were terminated sometime after the negative potential shift
occurred. However, these values are used here to compare between groups. For cases in which the specimen was
terminated shortly after activation, the measured values could be considered representative of CT. Eleven rebars from
AO group showed corrosion with 5.8 kg/m3 (1.45 %cm) as the minimum chloride measured at the rebar trace ([Cl-]tmin).
For the CO group, [Cl-]tmin= 6.4 kg/m3 (1.62 %cm) and was just slightly higher than for the AO group. Hence, not
much benefit was achieved by the calcium nitrate admixture on the specimens in which corrosion initiated. [Cl-]tmin
values suggest that the chloride threshold for rebars embedded in FA50 is lower than for those in FA20 and FA35.
Actually, the [Cl-]tmin values measured on FA20 and FA50 specimens are lower than those observed on AO or CO
specimens. Rebars from each FA-group showed corrosion: ten rebars from FA50; seven rebars from FA35N; three
from FA20; and one from group FA35. (Some other specimens were terminated, but the rebars did not display
corrosion. See Table 6 for details.) The chloride concentration that caused corrosion for rebar embedded in FA35N
seems to be higher than for the other FA groups; then again, a higher concentration was reached at the rebar trace in
the same amount of time. The [Cl-]tmin was ~7.2 kg/m3 (1.81 %cm) for the SF06 group, and four of the terminated
rebars have displayed corrosion. The [Cl-]tmin was 5 kg/m3 (1.26 %cm) for SF15, and six rebars have shown corrosion.
For SF27, the [Cl-]tmin was ~2.4 kg/m3(0.6 %cm), and only one rebar displayed corrosion. Finally, for SF15N, the
[Cl-]tmin was 5.7 kg/m3 (1.46 %cm), and no rebar displayed corrosion. Thomas8,9 has previously reported a reduction
in the chloride threshold for concrete containing fly ash as the fly ash amount is increased. However, in Thomas study
the chloride concentrations were not obtained above the rebar trace.
Diffusion coefficient
Table 8 shows the diffusion coefficients calculated for each sliced core grouped per mix-design. The column named
Fit indicates if any values were removed to obtain a better fit (due to skin effect). The table also includes the age at
which the specimen was terminated, and this was the age used to calculate Dapp. The calculated surface concentration
(Cs) from this fit is also included. The average Dapp for mixes AO and CO were slightly greater than 110-12 m2/s. The
average Dapp for SF mixes ranged between ~210-13 (SF27) and 710-13 (SF06) m2/s. The average Dapp were even
smaller for FA mixes, ranging from 910-14 (FA50) and 2.910-13 (FA35N) m2/s, i.e. the mix with 50% FA (FA50)
had the lower Dapp. The relatively low Dapp are in part due to the fact that seawater was used as the solution. Justnes10
and others have reported that seawater components (e.g. magnesium) can interact with the pore solution and form
other compounds which further slows down the chloride penetration.
Resistivity vs. diffusion coefficient

Table 9 shows the average chloride diffusivity calculated from the profiles after 1200 and after more than 5700 days
of exposure. Dapp values after 1200 days were reported earlier by Hartt1. For the profiles obtained after more than 5700
days (see Figures 10 and 11) of exposure there were cases in which the maximum concentration did not correspond
to the slice closest to the surface. For those cases the first layer (lower values) was omitted when performing the fit.
The values shown in the table are the average Dapp obtained from at least two cores. No Dapp was measured at 1200
days on CO specimens. At 5700 days, the largest Dapp was measured on AO and CO specimens followed by Dapp
values measured on SF06 specimens. FA20 and FA35 showed very similar Dapp. The smallest Dapp value corresponded
to the FA50 concrete mixture. Table 9 also includes the average ρwet measured (average of the values measured on at
least two cores) after more than 15 years of outdoor exposure and with the moisture increased in the lab.
9.14
@Seismicisolation
@Seismicisolation
Table 8 — Cs and Dapp calculated for each sliced core

Dapp Dapp Dapp Dapp
Age Cs m2/s Ave. Age Cs m2/s Ave.
ID Fit yrs kg/m3 ×10-12 ×10-12 ID Fit yrs kg/m3 ×10-12 ×10-12
AO2-A 1R 15.6 15.0 1.3 FA20-1-A 1R 15.6 42.6 0.15
AO2-C 1R 15.6 13.7 1.5 FA20-1-C 1R 15.6 37.1 0.15
AO3-A 1R 15.6 15.8 1.5 FA20-3-A 1R 17 22.3 0.11 0.12
1.5
AO3-B 1R 15.6 17.1 1.3 FA20-3-B 1R 17 23.0 0.099
AO8-A NR 16.5 15.3 1.5 FA20-3-C 1R 17 23.1 0.10
AO8-B NR 16.5 16.4 1.7 FA35-1-A 1R 15.6 37.1 0.15
0.13
CO1-A 2R 15.6 9.3 0.84 FA35-1-C 1R 15.6 44.1 0.10
CO1-C 2R 15.6 10.8 0.83 FA35-1-A 1R 17 30.9 0.074
1.4
CO3-A 2R 15.6 10.3 2.0 FA50-1-B 1R 17 35.2 0.063
CO3-B NR 15.6 11.0 2.0 FA50-1-C 1R 17 36.9 0.056
SF06-3-A NR 15.6 21.8 0.73 FA50-2-A 1R 15.6 50.2 0.10
SF06-3-D 1R 15.6 21.7 0.69 FA50-2-B 1R 15.6 41.5 0.095
0.71 0.09
SF06-1-A NR 16.5 21.8 0.69 FA50-5-B 1R 16.5 31.3 0.16
SF06-1-B NR 16.5 20.6 0.72 FA50-5-A NR 16.5 20.3 0.16
SF15-2-C 1R 15.6 23.9 0.26 FA50-7-A 1R 17 36.0 0.065
SF15-2-D 1R 15.6 21.4 0.27 FA50-7-B 1R 17 31.9 0.078
0.24
SF15-1-A 1R 16.5 26.5 0.22 FA50-7-C 1R 17 39.6 0.049
FA35N-2-
SF15-1-B 1R 16.5 23.9 0.21 A 2R 15.6 35.8 0.23
FA25N-2-
SF27-2-A 1R 15.6 31.6 0.21 C 1R 15.6 23.0 0.33
0.19 0.29
FA35N-1-
SF27-2-B 1R 15.6 13.5 0.17 A 1R 16.5 34.0 0.32
FA35N-1-
SF15N-3-C NR 15.6 17.3 0.27 0.26 B 1R 16.5 29.1 0.28
SF15N-3-D NR 15.6 17.1 0.24
Note: 1 kg/m3 = 1.6842 lb/yd3, 1×10-12 m2/s = 4.89 ×10-2 inch2/year
Table 9 — Apparent Diffusivity after 3.3 and >15 years and ρwet after 15 years
Mix Design Dapp (m2/s × 10-12) ρwet >15 yrs
Designation 3.3 years > 15 years kohm-cm
AO 3.48 1.5 4.6
CO n/a 1.4 4.8
FA20 0.80 0.12 19.5
FA35 0.74 0.13 36.0
FA50 0.57 0.09 62.0
FA35N 0.74 0.29 18.0
SF06 1.96 0.71 8.8
SF15 2.29 0.24 16.0
SF25 1.01 0.19 20.0
SF15N 0.99 0.26 13.5
Note: 1×10-12 m2/s = 4.89 ×10-2 inch2/year, 1 kohm-cm = 0.39 kohm-in
9.15
@Seismicisolation
@Seismicisolation
Figure 12 shows two series of data that relate ρwet (wet resistivity) vs. Dapp. Both series are shown with the same
resistivity values, although it is likely that the resistivity values at three years were somewhat smaller than those
recorded after more than 15.6 years. All Dapp values were lower after 15 years of exposure when compared to those
measured after 3.3 years. Adding calcium nitrate appears to lower the resistivity and increased the Dapp for both SF
and FA when compared to corresponding SF or FA concrete mixtures that contained comparable SF or FA amounts
in the concrete mix. There appears to be an outlier at 3.3 years, but not at 15 years. For the series after 3.3 years, the
outlier corresponds to SF2. After 3.3 years the Dapp is lower for FA concretes. After more than 15.6 years of exposure
the Dapp for SF3 was as low as the average Dapp for FA1. The observed Dapp values at 15 years are likely influenced
by not fully saturated concrete and that the exposed surface did not contain solution all the time. Additionally, Justnes10
and others have reported that seawater components (e.g. magnesium) can interact with the pore solution and form
other compounds which further slows down the chloride penetration.
1.E-11
3.3 yrs
>15.6 yrs
AO
SF06
AO SF15
CO
SF15N
Dapp (m2/s)
SF27
1.E-12
FA20 FA35
SF06 FA35N FA50
FA35N
SF15N
SF15 SF27
FA35
FA20
1.E-13
FA50
1.E-14
1 10 100
Resistivity (kohm-cm)
Figure 12 — Resistivity vs. Dapp
Dapp, pH and Cl- at the rebar trace

Table 10 shows pH values measured on cored specimens from each concrete mix at 1200 days. The values are
somewhat higher than those reported by others. Comparing only how much the pH was reduced by different amounts
of admixtures and remembering that a 0.1 change in pH corresponds to a factor of two variations in OH- activity. This
might explain why corrosion initiated at a significantly lower concentration on specimens with 50% FA; even though
these specimens have the lower chloride diffusivity. The small concrete cover and rebar presence might have influence
when corrosion initiated. Thus, the lower CT needs to be kept in mind when estimating service life of structures that
contain high admixture replacement ratios even though these mixes usually have a lower Dapp.
9.16
@Seismicisolation
@Seismicisolation
Table 10 — pH values measured at day 1200 [1]

Laboratory Admixture N2 glove box Admixture
Specimen pH pH reduction pH pH reduction
AO 13.7 13.84
FA20 13.46 0.24 13.57 0.27
FA35 13.27 0.43 13.39 0.45
FA50 12.57 1.13 13.18 0.66
FA35N 13.27 0.43 13.36 0.48
SF06 13.3 0.4 13.7 0.14
SF15 13.51 0.19 13.51 0.33
SF25 13.27 0.43 13.47 0.37
SF15N 13.42 0.28 13.59 0.25
Effect of CT, Dapp, CS on time to corrosion initiation for more traditional covers and rebar diameter
The time to corrosion initiation was calculated for a structure having rebars of 1.6 cm (#5) diameter with a concrete
covers of 5 cm (2 in) or 7.5 cm (3 in) using the calculated Dapp and Cs values, for selected cases. The CT was assumed
to be slightly larger than the minimum observed chloride concentration above the rebar trace. The model assumed a
1-D semi-infinite slab. Table 11 shows the assumed values used in the computations, and Table 12 shows the derating
values estimated from6 to account for the rebar presence, rebar diameter to cover ratio, and assumed CT/CS from Table
11.
Table 11 — Assumed parameter based on measured and fitted values

CT Dapp × 10-12 Cs
3
Mix Kg/m m2/s Kg/m3 CT/CS
AO 6.2 1.5 15.6 0.40
FA20 3.2 0.13 30 0.11
FA50 1.6 0.09 30 0.05
Note: 1 kg/m3 = 1.6842 lb/yd3, 1×10-12 m2/s = 4.89 ×10-2 inch2/year
Table 12 — Derating factors for the geometry and CT/CS of interest

Cover Diameter/cover OPC FA20 FA50
5 cm (2.95 in) 0.32 0.71 0.79 0.82
7.5 cm (2.95 in) 0.21 0.79 0.82 0.84
Table 13 shows the time to corrosion initiation in years for the considered cases taking into account the rebar presence.
There is a distinctive difference in the time to corrosion for those specimens with only OPC (similar to AO mix) and
those with FA when compared to the experimental results described above. Note that now, the time to corrosion for a
mix with 20% FA is comparable to the time corrosion initiation for a mix with 50% FA, whereas the time to corrosion
initiation for the mix with no fly ash was significantly shorter. The time to corrosion initiation was 3.1 to 3.5 times for
those with FA when compared with those containing only OPC. The time to corrosion was slightly longer (by 1% to
5%) for the mixture with 20% FA than with 50% FA, the lower CT for the latter countered the corresponding smaller
Dapp. Thus, when calculating service life it might be relevant to include not just the smaller Dapp due to the presence
and amount of fly ash, but also adjust downwards the CT value as the FA percent is increased. It is important to recall
that the average Dapp values were obtained from specimens prepared in the laboratory (albeit exposed outdoors and to
seawater) and might be somewhat smaller than those observed in field structures. Thus the computed values in this
section might be upper bound values. These results suggest that a thick enough concrete cover is needed to observe
the benefits of including fly ash in the mixture, and that similar Ti might be observed for concrete mixtures containing
20 to 50 percent fly ash.
9.17
@Seismicisolation
@Seismicisolation
Table 13 — Time to corrosion initiation in years including rebar presence effect

Cover OPC FA20 FA50
5 cm (2.95 in) 25.9 90.9 90.2
7.5 cm (2.95 in) 64.8 211.6 201.6
A similar 1-D diffusion analysis was done using the geometry of this study with the smaller cover; however’ the CS
values used in these computations were smaller and similar to those measured experimentally. Table 14 shows the
parameters used for Dapp, CT, CS, and CT/CS, and the corresponding derating factors. It was found that the calculated
Ti values for specimens with OPC only, FA20, and FA50 would have been 11, 21 and 20 years, respectively. These
times to corrosion initiation include the rebar presence effect. This is different from what was observed experimentally
for specimens with covers of 1.9 cm (0.75 in); the time to corrosion initiation for the specimens with 50% fly ash and
no fly ash being of the same order of magnitude (15 to 17 years). The 21/20 years - Ti results for concrete with FA -
is somewhat higher than observed experimentally. One possible explanation for this difference could be due to higher
CT for AO specimens than that assumed in the computations. Additionally, the chloride transport might be also
influenced by the skin effect (i.e., the max chloride concentration is at a certain distance from the surface), interaction
of concrete closer to the surface with seawater, and to a lesser extent the coarse aggregate presence.
Table 14 — Parameters used to calculate Ti for specimens with 1.9 cm (0.75 in) concrete cover
CT CS 1st layer CT/CS Dapp Derating
3 3 2 -12
kg/m kg/m m /s ×10 Factor
7.4 12 0.58 1.50 0.53
3.4 15 0.25 0.13 0.67
1.6 11 0.15 0.09 0.73
-12
Note: 1×10 m /s = 4.89 ×10 inch /year, 1 kg/m = 0.062 lb/ft3
2 -2 2 3
CONCLUSIONS
1. The Dapp of concrete mixture prepared with FA decreased close to one order of magnitude (at least 5X) from 1200
days to 5700 days, with the largest reduction observed on those specimens with higher FA. A more modest reduction
in Dapp was observed on samples with SF. The observed Dapp is likely influenced by the concrete not being fully
saturated all the time and that the exposed surface did not contain solution all the time.
2. The [Cl-]average (measured on rebars corroding) suggest that the chloride threshold for specimens of concrete mixtures
with 50% FA (~ 0.5 %cm) appear to be lower than those observed with the other FA admixture amounts (1 to
1.3 %cm). Corrosion initiated on more rebars from specimens with 50% FA, than with OPC only (1.9 %cm). This
could in part be to the lower chloride threshold.
3. A good correlation was observed between ρwet (wet resistivity) and Dapp.
4. Further evidence was found that the rebar presence increases the chloride concentration compared to the
concentration measured at the same depth.
5. A reduction in CT as FA percent is increased is recommended to be included for service life computations, in

addition to the lower Dapp that is associated with FA presence. Rebar presence effect is recommended to be included,
particularly for cases with small covers (lab specimens).
ACKNOWLEDGMENTS
The authors are indebted to the Florida Department of Transportation (FDOT) for financial support of this research
and to Mr. Mario Paredes and Mr Ron Simmons of the FDOT State Materials Office. In addition, the authors
appreciated the assistance with laboratory work and measurements provided by Mr. A. Sajban, Mr. J. Zielske, Mr. B.
9.18
@Seismicisolation
@Seismicisolation
Seo, Mr. F. Gutierrez and several undergraduate students of the Marine Materials Lab at FAU. The authors also
acknowledge Dr. W. Hartt and Dr. S.K. Lee whom were in charge of phase I of this project. The opinions expressed
in this paper are those of the authors and not necessarily of the FDOT.
REFERENCES
[1] Hartt W.H.; Charvin S.C.; Lee S.K., “Influence of Permeability Reducing and Corrosion Inhibiting Admixtures in
Concrete upon Initiation of Salt Induced Embedded Metal Corrosion” Final Report WPI 0510716 Tallahassee, FL:
Florida Department of Transportation Research Center, 1999.
[2] FM5-516 (2005). “Florida Method of Test For Determining Low-Levels of Chloride in Concrete and Raw
Materials,” Florida Department of Transportation, Tallahassee, FL, 2005, X pp.
[3] FM5-578 (2004). “Florida Method of Test for Concrete Resistivity as an Electrical Indicator of its Permeability,”
Florida Department of Transportation Tallahassee, FL, 2004, 4 pp.
[4] Angst U.M.; Elsener B.; Larsen C.K.; Vennesland O., (2011), “Chloride induced reinforcement corrosion:
Electrochemical monitoring of initiation stage and chloride threshold values”, Corrosion Science, Vol 53, 2011, pp
1451-1464.
[5] Presuel-Moreno, F., Wu, Y.Y., Arias, W., and Liu, Y., “Analysis and estimation of service life of corrosion
prevention materials using diffusion, resistivity and accelerated curing for new bridge structures,” Final Report
BDK79-977-02 (Volume 1); Boca Raton, FL: Florida Atlantic University, 2013.
[6] Kranc, S.C., Sagüés, A.A., and Presuel-Moreno, F.J., “Decreased Corrosion Initiation Time of Steel in Concrete
Due to Reinforcing Bar Obstruction of Diffusional Flow”, ACI Materials Journal, Vol.99, No.1, 2002, pp 51-53.
[7] Yu H. and Hartt W.H., Corrosion, “Effect of reinforcement and coarse aggregates on chloride ingress into concrete
and time-to-corrosion: Part I – spatial chloride distribution and implications”, Vol. 63, No.9, 2007 pp. 843-849.
[8] Thomas M.D.A. and Matthews J.D., “Chloride penetration and reinforcement corrosion in marine-exposed fly ash
concretes”. In: Malhotra VM, editor, Third CANMET/ACI International Conference on Concrete in a Marine
Environment, ACI SP-164, Detroit: American Concrete Institute; 1996, p. 317-38.
[9] Thomas M.D.A., “Chloride thresholds in marine concrete”, Cement Concrete Research, Vol 26, No. 4, 1996, pp.
513-9.
[10] Justnes, H.; Weerdt, K.D.; Geiker M., “Chloride binding in concrete exposed to seawater and salt solutions”, in
Concrete under Severe Conditions, Eds. Li Z.J.; Sun W, et al., RILEM Proceeding PRO84, 2013, pp. 647-659.
9.19
@Seismicisolation
@Seismicisolation
9.20
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation

ACI SP 308 Chloride Thresholds and Limits For New Construction Tepke

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ACI SP 308 Chloride Thresholds and Limits For New Construction Tepke

Uploaded by

Copyright:

Available Formats

An ACI Technical Publication

Chloride Thresholds and Limits for

Printed in the United States of America

Editorial production: Gail Tatum

The detrimental influence of chlorides on the corrosion of reinforcing steel in

Documents published by ACI Committees 201 and 222 currently recommend

CHLORIDE THRESHOLD VALUES IN CONCRETE – A LOOK BACK AND AHEAD

Ueli M. Angst and Bernhard Elsener

Cement type Designation (acc. to EN 197-1) 1996 2006 2015

EXPERIENCE FROM PORTLAND CEMENT SYSTEMS

Implications for service life modelling

APPLYING THE ESTABLISHED CONCEPT TO MODERN MATERIALS

1) The kinetics of chloride ingress into the concrete

Potential problems with the established concept

SUGGESTIONS FOR FUTURE RESEARCH

Future research should thus address the following issues:

CONFUSION ON CHLORIDE LIMITS IN SPECIFICATIONS THAT CHALLENGE THE INDUSTRY

ASTM C1218 - Water-Soluble Chloride in Mortar and Concrete

ASTM C1152 – Acid-Soluble Chloride in Mortar and Concrete

ASTM C1524 – Water Extractable Chloride in Aggregate (Soxhlet Method)

SOURCES OF INTERNAL CHLORIDE

Calculation of Total Chloride in Concrete

Table 1. Example Calculation of Total Chloride Content in Concrete

CHLORIDE LIMITS FOR CONCRETE

Table 2: Chloride Limits in ACI 318 & ACI 301

C0 Concrete dry or protected from moisture 1.00 0.06

Prestressed concrete 0.08 0.06

Reinforced concrete in dry condition 0.20 0.15

Roughly what percentage of specifications requires this testing?

EFFECT OF TEMPERATURE ON THE CHLORIDE BINDING OF PORTLAND CEMENT EXPOSED

By Jianqiang Wei, Bernard Tao and W. Jason Weiss

3Ca(OH)2 + CaCl2 + 12H2O → 3CaO·CaCl2·15H2O (1)

T = 27.88 ln (C0 / C*) (2)

Ca(OH)2 + H2CO3 → CaCO3 + 2H2O (3)

Materials used in making cement pastes

Table 1 — Chemical and mineralogical composition of Type I and Type V cements.

Materials used in making CaCl2 solutions

Equilibrium method for chloride binding

RESULTS AND DISCUSSION

Effect of temperature on chloride binding

150 0.15 150 0.15

Bound chloride (lb Cl / lb sample)

Bound chloride (mg Cl / g sample)

120 -5 C (23 F) 0.12 120 0.12

Chloride-binding isotherms for low CaCl2 concentrations

150 0.15 150 0.15

Bound chloride (mg Cl / g sample)

Bound chloride (lb Cl / lb sample)

Bound chloride (lb Cl / lb sample)

120 0.12 120 -5 C (23 F) 0.12

Chloride-binding isotherms of cement in the presence of CaCl2

Bound chloride (lb Cl / lb sample)

Bound chloride (lb Cl / lb sample)

Bound chloride (mg Cl / g sample)

150 0.15 150 0.15

Bound chloride (lb Cl / lb sample)

Bound chloride (lb Cl / lb sample)

Bound chloride (mg Cl / g sample)

Contribution of calcium oxychloride

Bound chloride (lb Cl / lb sample)

110 0.11 110 0.11

Bound chloride due to Ca-Oxychloride

Bound chloride due to Ca-Oxychloride