Cement and Concrete Composites 134 (2022) 104817

Contents lists available at ScienceDirect

Cement and Concrete Composites

journal homepage: www.elsevier.com/locate/cemconcomp

Correlating alkali-silica reaction (ASR) induced expansion from short-term

laboratory testings to long-term field performance: A semi-empirical model
Thuc Nhu Nguyen a, *, Leandro F.M. Sanchez b, Jianchun Li a, Benoit Fournier c,
Vute Sirivivatnanon a
a
School of Civil and Environmental Engineering, University of Technology Sydney, Australia
b
Department of Civil Engineering, University of Ottawa, Canada
c
Laval University, Quebec, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Correlating short-term expansion of concrete specimens in the laboratory and long-term expansion of concrete in
Alkali-silica reaction (ASR) the field is crucial to evaluate the reliability of laboratory test methods and essential for the prognosis of alkali-
Alkalis silica reaction (ASR) in concrete infrastructures. In this study, a novel semi-empirical approach is proposed for
Expansion
forecasting ASR-induced expansion of unrestrained concrete in the field using laboratory measurements data. In
Field exposure
addition to the use of short-term laboratory expansion data, the model accounts for the effects of alkali leaching,
Semi-empirical model
alkali contribution from aggregates, and environmental conditions (i.e., temperature and relative humidity). A
comprehensive database from the literature was gathered for the development and calibration of the proposed
model. Finally, the model was used for various concrete blocks incorporating different reactive aggregates and
exposed to three outdoor conditions in Canada and the USA. Model outcomes show that it is highly promising for
forecasting the induced expansion of concrete in the field from the accelerated laboratory tests data. Analysing
the modelling results also highlights the importance of alkali leaching and environmental conditions on the
correlation between laboratory and field performance.

1. Introduction
Alkali-silica reaction (ASR) is one of the most harmful distress
mechanisms affecting concrete infrastructure worldwide. ASR is a
chemical reaction between the alkali hydroxides from the concrete pore
solution (i.e., Na+, K+, and OH− ) and some unstable mineral phases
from the aggregates (i.e., fine and coarse) used to make concrete. It
generates a secondary product, the so-called alkali-silica gel, that swells
upon moisture uptake, leading to induced expansion, significant crack
formation and losses in mechanical properties and performance (i.e.,
durability and serviceability) of affected concrete components [1–3].
The best solution for ASR is preventing its occurrence in new con­
struction, yet prevention is no longer feasible for current ASR-affected
infrastructure. These affected structures require comprehensive diag­
nosis and prognosis protocols for evaluating the current damage degree
and forecasting the potential of further deterioration. Such information
is essential for the efficient selection of rehabilitation strategies for
affected structures. In this context, numerous diagnosis and prognosis
techniques have been developed over the last decades; although
improvement of existing techniques is always desired, one may say that
the current procedures used to diagnose (i.e., cause and extent)
ASR-induced deterioration are quite reliable; otherwise, there is still
very limited knowledge on efficient techniques able to forecast
ASR-induced development of the affected structures/members over
time.
To accomplish this requirement, several mathematical models (i.e.,
empirical, semi-empirical, analytical, and numerical) were developed to
provide insightful information on the potential of future deterioration of
ASR-affected concrete. However, most of these models have not been
widely used in practice due to either limited applications to laboratory
specimens and conditions [4–7], or computational complications
[8–11]. Hence, developing more practical yet effective empiri­
cal/analytical models to describe ASR-induced expansion and deterio­
ration in concrete members exposed to field conditions is still required.
This work proposes a novel semi-empirical approach, based on labora­
tory test results, to forecast the ASR-induced expansion of concrete

* Corresponding author. School of Civil and Environmental Engineering, University of Technology Sydney, Australia.
E-mail address: thuc.nguyen@uts.edu.au (T.N. Nguyen).

https://doi.org/10.1016/j.cemconcomp.2022.104817
Received 25 February 2022; Received in revised form 14 October 2022; Accepted 16 October 2022
Available online 19 October 2022
0958-9465/© 2022 Elsevier Ltd. All rights reserved.
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817

exposed to field conditions.

2. Background

2.1. ASR-induced expansion: laboratory versus field performance

Several laboratory test procedures such as the accelerated mortar bar

test (AMBT), concrete prism test (CPT) and accelerated concrete prism
test (ACPT), i.e., from ASTM International, Canadian Standards Asso­
ciation CSA, RILEM, and Australian Standard test methods, have been
developed to assess the potential alkali-reactivity of concrete aggregates
and the efficiency of preventive measures before their use in the field.
Moreover, several comparative field exposure studies have been devel­
oped to establish reliable benchmarking and examine the reliability of
the laboratory standard tests for the applications mentioned above [12].
Despite the robustness and applicability of these standards/protocols
in identifying potential reactivity of a large number of aggregates, the
majority of experts agree that some limitations still exist in classifying
borderline reactive aggregates along with using the above tests to
determine the level of preventive measures required (i.e., control of the
concrete alkali content, use of supplementary cementitious materials
(SCMs) and lithium-based admixtures) to suppress ASR [13–16].
Furthermore, observations from comparative laboratory-field studies
show that the behaviour of concrete mixtures incorporating similar
reactive aggregates vary greatly in the laboratory and in the field; it is
often reported that field specimens such as exposed concrete blocks
display significantly higher expansion than laboratory-made specimens
[14–16], especially when preventive measures are used. An example
from Thomas et al. (2006) is shown in Fig. 1(a) [13], where the authors
tested different specimens (i.e., mortar bars, concrete prisms and con­
crete blocks) containing high-alkali cement (1.15% Na2Oe per cement
mass) and incorporating reactive flint sand (25% of the total aggregate)
under laboratory and field conditions. Analysing the plot, one observes
that the concrete blocks exposed to field conditions display two times
greater ASR-induced expansions than laboratory specimens at two
years. The lack of a proper laboratory-field correlation brings important
concerns on the use of laboratory testing protocols to forecast the Fig. 1. Differences in expansion of concrete in the laboratory and in the field,
as reported by: (a) Thomas, Fournier, Folliard, Ideker and Shehata [13], and (b)
long-term performance of ASR-affected concrete structures in the field.
MTO [22].
Moreover, it raises a critical question for ASR community: what is the
reason for the currently poor laboratory-field performance correlation?
Thomas et al. (2006), who after conducting several comparative alkali contribution from aggregates in long-term exposed field members
laboratory-field performance studies, concluded that the main reason might be amongst the factors causing variations between laboratory and
for this difference is the significant leaching of alkalis from laboratory field performance of concrete specimens. For instance, the authors re­
specimens, which are often much smaller in size and exposed to a more ported that some aggregates may release up to 3.4 kg/m3 of Na2O
“leaching-preferable” condition when compared to the field members equivalent (Na2Oe) to the pore solution of concrete, which could be the
[13]. The same conclusion was reported in Ideker, Bentivegna, Folliard reason why several dams built from concrete mixtures with an original
and Juenger [14] and Fournier, Ideker, Folliard, Thomas, Nkinamubanzi alkali loading of less than 2 kg/m3 of Na2Oe (i.e. from the cement)
and Chevrier [17]. Lindgård and co-workers, when measuring present significant induced expansion in the field while presenting
ASR-induced expansion and alkali leaching over the concrete prism test negligible expansion in laboratory testing [20,21].
(CPT), indeed verified that concrete specimens might lose between 3 and Another obvious distinction in the induced expansion of concrete in
20% of alkalis in the first 4 weeks and up to 50% after one year [18]. the laboratory and in the field is on the expansion rate due to differences
Recently, Sinno and Shehata [19] measured ASR-induced expansion and in exposure conditions. In this regard, concrete in the field normally
leaching on concrete specimens made of the reactive Spratt limestone swells at a much lower rate due to lower (average) annual temperature
and Sudbury greywacke-argillite aggregates using different sample sizes when compared to constant laboratory conditions (i.e., 38 ◦ C or 60 ◦ C
(i.e., 75 × 75 × 285 mm -standard prisms and 100 × 300 mm cylinders). from both CPT or ACPT, respectively). This observation can be found
The results showed over 25% and 40% of alkali leaching at 1.5 years for from test data of MTO [22] for concrete specimens made of Spratt
the Spratt and Sudbury samples, respectively. Furthermore, for both limestone and exposed at the Kingston site in Ontario, Canada (Fig. 1
reactive aggregates, smaller cross-section samples (i.e., the prisms) (b)). The laboratory-made concrete specimens stored at 38 ◦ C start
resulted in more prominent leaching and lower expansion when swelling rapidly after the beginning of the test, while concrete blocks in
compared to larger samples (i.e., the cylinders). Hence, it is very likely the field took about 2–3 years before initiating the expansion due to low
that alkali leaching from large blocks (i.e., 400 × 400 mm cross-sections (average) annual temperature at the exposure site. In order to correlate
as reported in Thomas et al. [13]) exposed to field conditions could be the induced expansion measured from the laboratory to the expansion of
significantly lower than that of small laboratory-made samples, result­ field concrete, these distinct environmental conditions should be taken
ing in the currently poor laboratory-field performance correlation. into consideration.
Besides the alkali leaching frequently observed in the laboratory, The limitations mentioned above still require extensive studies to
Bérubé, Duchesne, Dorion and Rivest [20] also observed that a potential improve laboratory test procedures and laboratory-field correlations.

2
T.N. Nguyen et al.
Likewise, ongoing research has been focusing on developing alternative
tests and improving the current test methods to limit or eliminate alkali
leaching from concrete, such as the miniature concrete prism test
(MCPT) [23], the concrete cylinder test (CCT) [24] and the RILEM
AAR-10 concrete prism test [25]. On the other hand, with the current
understanding of the laboratory-field correlations and relatively large
database available, mathematical modelling based on laboratory ob­
servations may be developed, enabling a more reliable assessment of the
potential of further ASR-induced development in affected structures
along with its likely structural consequences. A number of empirical,
analytical, and numerical models have been proposed to provide pre­
dictive information on ASR-induced expansion of affected structures in
the field, such as thermo-chemo-mechanical models [8,26], finite
element inverse analyses [9], multi-scale chemical analyses [7],
analytical models [27], etc. Yet, to the authors’ best knowledge, the
applicability of the vast majority of the above-proposed approaches is
still limited in practice due to either their complexity or even the lack of
validation with experimental data from various concrete mixtures and
reactive aggregates. Moreover, very few of those models account for the
effects of alkali leaching in the laboratory specimens as well as the
laboratory-field differences in exposure conditions.
2.2. Larive’s semi-empirical model for ASR-induced expansion
A number of empirical and analytical models have been developed to
estimate ASR kinetics and induced expansion in the laboratory [4,5],
which combined with numerical analyses, were intended to predict the
structural implications (i.e., deformation, stability) of structures, struc­
tural members [8,10,26]. Among these models, the semi-empirical
model proposed by Larive [4] is probably one of the most accepted
and widely used by ASR community [8,26–29]. The model for the
stress-free expansion evolution was developed based on an extensive
experimental database of more than 600 concrete specimens incorpo­
rating various reactive aggregates and exposed to different laboratory
conditions. In this model, ASR kinetics and induced expansion were
constructed as a function of three main model parameters, which are
ultimate expansion (ε∞), latency (τL), and characteristic (τC) times [4],
given by
t humidity coefficient f(h) equals to 1 (i.e., for dam structures).
1 − e− τc (θ)
ε(t, θ) = t− τL (θ)
× ε∞ (1)
1 + e− τc (θ)
Typical S-curves formed by the model are shown in Fig. 2. There is no
clear physical meaning for the latency and characteristic times; how­
ever, evaluating the S-curve one observes that the latency time repre­
sents the initiation period of ASR-induced expansion while the
characteristic time defines the expansion rate after the initiation period.
The time constants τL and τC were derived as temperature-dependent
Fig. 2. ASR-induced expansion curves obtained from Larive’s model with
considering temperature and relative humidity effects after Saouma and Per­
otti [8].
3
Cement and Concrete Composites 134 (2022) 104817
parameters, as shown in Eqs. (2)–(3) after Ulm et al. [25].
[ ( )]
1 1
τc (θ) = τc (θ0 ) UC − (2)
θ θ0
[ ( )]
1 1
τL (θ) = τL (θ0 ) UL − (3)
θ θ0
where, θ is the absolute temperature (θ K = 273 + T ◦ C); θ0 is reference
absolute temperature, which is typically referred to exposure tempera­
ture in laboratory testing; UC and UL are thermal activation constants,
derived from the Arrhenius physical chemistry concept [4,26]:
UC = 5400 ± 500K (4)
UL = 9400 ± 500K (5)
In addition to the temperature dependency of the time constants,
Saouma and Perotti [8] further developed the model by introducing the
effect of relative humidity and stress state (i.e., for structural modelling)
to the expansion behaviour of concrete structures. The authors adopted
the relative humidity-dependent coefficient of the ASR swelling proposed
by Capra and Bournazel [5] into the Larive’s model, which directly re­
duces the ultimate expansion ε∞:
f(h) = RH8 (6)
where, RH is the internal relative humidity or relative humidity in the
material pores. This coefficient is normally set to be zero when the
relative humidity within the material is lower than 80% [5,8,30]. This
updated form of Larive’s model with both temperature and humidity
considerations has been thereby utilised in several numerical studies for
ASR-induced expansion and structural implications [8,10,11]. Yet, it is
important to notice that the current form of the relative humidity de­
pendency is highly sensitive, significantly impacting on ASR-induced
expansion. For instance, if the internal RH drops from 100% as nor­
mally expected in the laboratory testing to 95%, ASR-induced expansion
drops to almost 45%. In this regard, although various sophisticated ASR
numerical models adopted this humidity-dependent coefficient to ac­
count for the humidity effect [8,11], relative humidity values as high as
100% have been frequently used in practice, which yields a relative
In another study, Comi, Kirchmayr and Pignatelli [10] proposed an
analytical approach to account for the effects of concrete saturation
degree on ASR-induced expansion through new parameters coupled to
Larive’s conventional model (i.e., τC, τL and ε∞). The model was well
calibrated using experimental data from Larive [4] and yielded efficient
results in chemo-mechanical or chemo-thermo-damage models to eval­
uate ASR-induced expansion and deformation in structures [10]. It is
worth mentioning that measuring the internal RH and degree of capil­
lary saturation in practical applications for field structures is not always
convenient for long-term measurements due to its destructive nature. In
addition, the relation between the internal RH and the DCS varies
depending on several factors, the water/binder-ratio in particular is the
most important one as per Lindgård et al. (2006) [31].
Finally, concrete mixtures with various reactive aggregate types and
reactivity levels have distinct sets of Larive’s model parameters [3,4].
Therefore, De Grazia, Goshayeshi, Gorga, Sanchez, Santos and Souza
[29] proposed several sets of the model parameters for both ASR and
ACR-induced expansion according to the reactive aggregate’s type and
nature/reactivity. The model also accounts for temperature, RH, and
concrete alkali content; however, it is applicable only for accelerated
tests performed in the laboratory. Hence, further analyses are still
required to assess the suitability of the proposed approach for structures
in the field by considering the leaching of alkalis from accelerated tests
and modelling differences in exposed environmental conditions in the
laboratory and in the field.
T.N. Nguyen et al.
3. Scope of work
The ultimate goal of this study is to provide a robust and engineering-
friendly tool to estimate ASR-induced expansion of unrestrained field
concrete members based on measurements from accelerated laboratory
tests such as the concrete prism test (CPT). To accomplish this goal, a
novel semi-empirical model was developed by adopting Larive’s model
to utilise one-year laboratory expansion test results, with additional
considerations to account for the most important factors affecting ASR-
induced expansion, such as concrete alkali content, leaching and
releasing of alkalis, temperature, and RH. “Aggregate-related” parame­
ters are obtained from the one-year laboratory expansion test results (i.
e., CPT). The impact of alkali content and alkali leaching was thoroughly
evaluated and considered to correlate short and long-term expansions of
concrete in the laboratory and in the field. The model was validated
using a comprehensive laboratory test database available in the litera­
ture. The proposed model was then utilised to estimate ASR-induced
expansion of concrete members and blocks manufactured over three
distinct experimental campaigns: Ontario Ministry of Transportation
(MTO) project at Kingston (Ontario, Canada) [15], CANMET project
(Ottawa, Ontario, Canada) [16,17], and the University of Texas project
(Texas, Austin, USA) [14]. Finally, further analyses of the effects of the
key factors (i.e., concrete alkali content, alkali leaching, alkali contri­
bution from aggregates, and environmental conditions) on ASR-induced
expansion of affected concrete members in the field were performed.
4. Model development and calibration
As discussed above, Larive’s model has been widely accepted and
used to represent ASR-induced expansion. The semi-empirical model
proposed in this study is based on Larive’s model and incorporates
additional factors into the three model parameters, ε∞ , τc , and τL . The
dependency of the above model parameters to temperature and relative
humidity is adopted from Larive (1997), Ulm et al. (2000) and Hariri-
Ardebili and Saouma (2018) [4,26,32] as presented in Section 2.2;
they could relatively account for the effects of environmental condi­
tions, i.e., differences between accelerated laboratory and field exposure
conditions or between different field conditions. In this study, the alkali
content and alkali leaching dependencies were proposed to enable
modelling ASR-induced expansion of concrete in the field with different
alkali contents along with minimal/or no leaching of alkalis. These
considerations are presented in detail in the next sections.
In general, the forms of the model parameters are given as hereafter:
ε∞ = ε∞,o ( T0 , RH0 , A0 ) × kε,RH × kε,A (7)
1 considered in the scope of this current research, yet it is an important
τC = τoC (T0 , RH0 , A0 ) × kC,T × (8)
kC,LA
1 4.2.2. Relative humidity
τL = τoL (T0 , RH0 , A0 ) × kL,T × (9)
kL,LA
where, RH, T, A and LA denote relative humidity (%), temperature (oC),
concrete alkali content (kg/m3), and alkali leaching at 1-year (%),
respectively. ε∞,o , τoc and τoL are reference model parameters, which are
ultimate expansion and ASR time constants from laboratory testing
conditions, i.e., at T = 38oC, 100% RH and alkali loading A = Ao (i.e.
5.25 kg/m3 in CPT). Dependent coefficients kε,RH and kε,A account for the
effect of relative humidity and alkali content, respectively, to the ulti­
mate expansion, while kC,T and kC,LA (kL,T and kL,LA ) are for considering
the impacts of temperature, relative humidity, and leaching of alkalis
from laboratory samples, respectively, on the characteristic time (la­
tency time). Therefore, laboratory test procedures will also have their
impacts on the dependent coefficients.
4
Cement and Concrete Composites 134 (2022) 104817
4.1. Reference model parameters
The reference model parameters (ε∞,o, τoL and τoC) vary depending on
the aggregate’s type (fine/coarse) and nature (lithotype). De Grazia,
Goshayeshi, Gorga, Sanchez, Santos and Souza [29] proposed eight
different sets of these model parameters for various aggregate’s types
and reactivity levels. The proposed model parameters that showed to be
promising to appraise AAR-induced expansion in the laboratory have
also been utilised for the modelling of expansion and deformation of
Robert-Bourassa Charest overpass in Quebec City, Canada [28] and Paulo
Afonso IV dam in Paulo Afonso, Brazil [33]. Nevertheless, due to the
often observed and challenging field conditions (i.e., important varia­
tions of temperature, relative humidity, wetting/drying cycles, etc)
along with important reactivity variations from one aggregate source to
another, there are still many challenges to adopt a set of model pa­
rameters based on the behaviour of “similar” aggregates gathered in the
laboratory. Therefore, the best source of data should be from the same
aggregate used in the field structure under analysis. In this study, the
reference model parameters are obtained through a curve-fitting of
Larive’s model using the expansion data measured via the concrete
prism test at 38oC and 100% RH.
4.2. Consideration of environmental conditions: temperature and relative
humidity
4.2.1. Temperature
The dependency of the time constants τC and τL on temperature is
adopted from Ulm et al. (2000) [26]. These dependent coefficients were
developed based on physicochemical parameters and were thoroughly
calibrated using a significant experimental database from Larive (1997):
[ ( )]
1 1
kC,T = exp UC − (10)
θ θ0
[ ( )]
1 1
kL,T = exp UL − (11)
θ θ0
where, θ0 is reference absolute temperature (θ K = 273 + T ◦ C, T0 =
380C); UC and UL are thermal activation constants. This temperature
dependence model of the τC and τL has been widely accepted and suc­
cessfully implemented in forecasting expansion of not only samples
manufactured in the laboratory but also field structures such as dams
and bridges, such as in Hariri-Ardebili et al. (2018), Ulm et al. (2000),
Saouma and Perotti (2006) [11,26,34]. It is acknowledged that the
impact of temperature on the expansion rate and extent is not unique
from one reactive aggregate to another [10,17,18,35]. This is not
topic to investigate in future works.
A wide range of experimental studies has been conducted to inves­
tigate the effect of RH on the ASR-induced expansion. By controlling/
measuring RH and the corresponding induced expansion on concrete
specimens at 21 ◦ C, Stark (1991) [36] and Bérubé et al. (2002) [20]
stated that an internal RH of 80% was required for ASR to occur in concrete.
In terms of the modelling, the relative humidity dependency model
proposed by Capra and Bournazel [5], as aforementioned in Section 2.2,
is adopted in this study. This internal RH dependency is written in the
form of the dependent coefficient kε,RH :
kε,RH = RH 8 (12)
where, RH herein is the internal relative humidity or relative humidity
in the material pores. The model was utilised for developing a consti­
tutive model for ASR in Saouma and Perotti (2006) and is widely
accepted and used in modelling ASR-induced expansion [8,11].
T.N. Nguyen et al.
4.3. Consideration of alkali content and alkali leaching
4.3.1. Dependency of the ultimate expansion (ε∞) on alkali content and
alkali leaching
The dependency of the ultimate expansion on alkali content is rep­
resented by the alkali content coefficient kε,A, as shown in Eq. (7) of the
semi-empirical model. Fig. 3 illustrates an example from Fournier and
Bérubé [37], where the authors studied the one-year CPT outcomes of
concrete mixtures incorporating two reactive aggregates (i.e., Sudbury -
Su and Spratt - Sp) as a function of the alkali content of the mixtures. It is
evident that the higher the alkali content, the higher the one-year CPT
expansion obtained. Moreover, increasing the alkali content of concrete
beyond 9 kg/m3 showed to have less impact on the final one-year
expansion than with lower alkali levels. In addition, the increase of
expansion was deemed to vary as a function of the aggregate nature and
reactivity.
To better understand the alkali content dependency, test results from
various concrete mixtures tested in Sinno and Shehata [19], Fournier
and Bérubé [37], Costa [38] varying in reactive aggregate types/natures
and alkali content (i.e., from 1.5 kg/m3 to 12.4 kg/m3, Na2Oe) were
collected, and the alkali coefficient was calculated for comparison pur­
poses. The data is shown in Fig. 4, in which the alkali content level 5.25
kg/m3, Na2Oe is assumed to be the control alkali content (kε,A = 1.00),
and the alkali content coefficient was calculated accordingly. It is worth
noting that most of these works were conducted with alkali contents
lower than 5.25 kg/m3, Na2Oe to evaluate the alkali threshold for po­
tential reactivity assessment, as concrete alkali contents beyond 5.25%,
however, are quite rare in practice. Only Fournier and Bérubé (2000)
[34] tested alkali contents up to 12.4 kg/m3, Na2Oe, where the results
demonstrate a huge variation of the coefficient kε,A for alkali contents
higher than 5.25 kg/m3. Also, test data for the expansion of concrete
with alkali contents below 2 kg/m3, Na2Oe is also limited.
Apparently, the higher the alkali content, the higher the coefficient
kε,A; yet, the induced expansion starts increasing significantly from al­
kali contents of about 3 kg/m3 before starting to be more stable after 9
kg/m3, Na2Oe. Even though kε,A could vary depending on the reactive
aggregate type and nature, there are very limited results in the literature
to quantify this impact. In this study, kε,A is obtained by averaging these
available experimental data to take into account the impact of alkali
content on the ultimate expansion. For instance, the alkali content co­
efficient is 0.58 for the mixture containing 3.75 kg/m3 alkalis, which
means that the ultimate expansion of this mixture can be approximated
58% of the value of the CPT standard mixture with 5.25 kg/m3, Na2Oe
alkalis. The alkali content coefficient increases from 0.08 at the alkali
Fig. 3. ASR one-year CPT expansion as a function of the concrete alkali con­
tent [37].
5
Cement and Concrete Composites 134 (2022) 104817
content of 1.5 kg/m3 to 3.08 at the alkali content of 12.4 kg/m3, Na2Oe.
In order to simulate the no-leaching scenario of the field concrete, an
ideal “no leaching” expansion curve is proposed in this study, which is
calculated using the expansion and leaching data obtained from labo­
ratory tests. Considering a concrete mixture containing a reactive
aggregate and an initial alkali content of Ao (kg/m3), a higher alkali
loading of A, kg/m3 is required to compensate for the leaching and to
simulate the ideal “no leaching” scenario of the original alkali loading Ao,
which can be calculated as follows:
Ao
A= (13)
1 − LA
or,
Ao = A × (1 − LA) (14)
where LA (%) corresponds to the leaching amount measured at one year.
As shown in Eq. 13 and 14, the use of this increased alkali content A
keeps at one year the same initial amount of alkali Ao of the tested
concrete mixture. For instance, for concrete containing 5.25 kg/m3,
Na2Oe alkalis with 20% alkali leaching at one year (i.e., 80% remaining)
in the concrete prism test (CPT), an increased alkali content of 6.56 kg/
m3, Na2Oe is adopted for the ideal “no leaching” expansion calculation,
which secures 5.25 kg/m3 (i.e., 80% of 6.56 kg/m3) alkalis after a one
year of testing. Next, the alkali content-dependent coefficients kε,A and
kε,Ao are derived from the alkali loading of A and Ao, respectively, based
on the proposed relationship illustrated in Fig. 4. The fractional differ­
ence between kε,A and kε,Ao is utilised to increase the induced expansion
measured on “leaching” CPT samples in order to represent its ideal
expansion curve without leaching:
ε∞ ∞
no leaching = εleaching ×
kε,A
(15)
kε,Ao
in which, ε∞
no leaching is the ultimate expansion of the “no leaching” sce­
nario; ε∞
leaching is the ultimate expansion obtained from the best curve-
fitting of the expansion curve measured in the laboratory (i.e., CPT).
In the case that the original concrete in the laboratory contains 5.25 kg/
m3, Na2Oe of alkalis, the coefficient kε,Ao is equal to 1, and thus, the
formula in Eq. (15) becomes the same as in Eq. (7). Based on the cor­
relation with this ideal expansion curve, the expansion of different
concrete samples varying in leaching amount could also be correlated.
4.3.2. Dependency of the time constants (τC and τL) on alkali leaching
The dependency of the time constants (τC and τL) on alkali leaching is
represented by alkali leaching coefficients kC,LA and kL,LA in Eq. (8) and
Eq. (9) of the semi-empirical model. According to Lindgård, Thomas,
Sellevold, Pedersen, Andiç-Çakır, Justnes and Rønning [18] and Sinno
and Shehata [19], the more the alkalis leach out, the earlier the
expansion levels off. Lindgård, Thomas, Sellevold, Pedersen, Andiç-Ça­
kır, Justnes and Rønning [18] reported that the expansion of small
concrete prisms (i.e., 70 × 70 × 280 mm) with 33% alkali leaching at
one year is likely to level off after 1.5 years of exposure, while larger
Norwegian prisms (i.e., 100 × 100 × 450 mm, 16.5% leaching at one
year) continue to expand throughout two years of exposure to about
40% higher than the smaller samples [18]. In terms of Larive’s expan­
sion model parameters, this observation means that an increase in the
leaching amount results in a decrease of the characteristic and latency
times, as shown in Fig. 5. The same experimental observation was also
reported in Sinno and Shehata [19]. This phenomenon could be
explained by the plateau of the expansion curve after the alkali content
of the concrete mixture drops under the alkali threshold required to
maintain ASR-induced development.
Fig. 6 illustrates the reduction of τC and τL with respect to one-year
alkali leaching, which is obtained from experimental data of both
Lindgård et al. (2013) and Sinno and Shehata (2019). It is worth noting
that in these figures, the value of 0% alkali leaching is linearly projected
T.N. Nguyen et al.
Fig. 4. Dependency of the ultimate expansion (ε∞) to the concrete alkali content.
Fig. 5. Effect of alkali leaching on the ASR kinetic (Lindgård et al., 2013).
Fig. 6. ASR kinetics parameters with respect to alkali leaching at one year.
from different leaching values from the experimental data. The char­
acteristic time thus displays a higher reduction rate due to leaching than
the latency time. Linear functions were calculated and adopted for the
determination of kC,LA and kL, LA, given as:
kC,LA = − 1.05 × LA + 1 (16)
6
Cement and Concrete Composites 134 (2022) 104817
kL,LA = − 0.7 × LA + 1 (17)
The characteristic and latency time constants of the ideal expansion
curve for the “no leaching” scenario could thus be given as:
τC,leaching
τC,no leaching = (18)
kC,LA
τL,leaching
τL,no leaching = (19)
kL,LA
in which, τC,no leaching and τL,no leaching are the ideal characteristic and la­
tency times of the “no leaching” scenario, respectively; τC,leaching and
τL,leaching are the characteristic and latency times obtained from the best
curve-fitting of the expansion curve measured from the laboratory test
(i.e. the CPT). Also, it should be noted that these two alkali leaching-
dependent coefficients on the time constants (i.e., kC,LA and kL,LA) are
to take into consideration leaching only.
4.3.3. Validation of modelling the ideal expansion curve for “no leaching”
scenario
As previously discussed in Section 2.1, alkali leaching is commonly
not likely to occur, or at a very minimal rate, for concrete members in
the field (some exceptions would be small size members or members
immersed in freshwater). Therefore, in order to model the expansion of
concrete in the field, it is important to estimate how much a concrete
member could swell if no leaching occurs, namely, the ideal expansion
curve. In this study, the calculation of the ideal expansion curve is based
on the model proposed above by quantifying the impact of alkali content
and alkali leaching on the induced expansion. Model validation was
conducted using a reliable laboratory dataset reported in Lindgård et al.
[18] to examine the reliability of the proposed model in estimating the
ideal expansion curve prior to modelling the expansion of field concrete.
4.3.3.1. Lindgård et al. experimental data [18]. Lindgård and co-workers
conducted three series of CPTs varying specimen dimensions and
exposure conditions: (1) 100 × 100 × 450 mm prisms exposed to 38 ◦ C
and 100% RH, (2) 70 × 70 × 280 mm prisms exposed to 38 ◦ C and 100%
RH, and (3) 70 × 70 × 280 mm prisms submerged in de-ionised water at
38 ◦ C where the water was replaced regularly to promote leaching of
alkalis [18]. An identical concrete mixture incorporating a reactive
coarse aggregate and 3.75 kg/m3 of alkalis was used in the study for all
three series. In addition to the induced expansion, total alkali leaching
was measured over time from both water in the bottom of storage
containers and in the lining inside the containers. The leaching amounts
at one year for these three series were about 16.5%, 33.2% and 79.5%,
respectively. Analysing expansion results in Fig. 7(a) together with the
leaching amounts, once again one verifies that for the same concrete
mixture, the higher the alkalis leached out, the lower the induced
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817

respectively, as shown in Table 1. Details on the experimental data and

model parameters are shown in Table 1 and Fig. 7.
The model outcomes from approach 1 and 2 (i.e., based on test re­
sults of 70 × 70 mm and 100 × 100 mm prism kept in 100% RH) show
very good agreement for the ideal expansion curves. Calculated ultimate
expansions from these two approaches are closely matched, i.e., 0.603
and 0.599 for approach 1 and approach 2, respectively, while only the
characteristic time τC is slightly different. Approach 3 (based on samples
soaked in de-ionised water) provides a divergent ideal expansion curve
compared to the first two approaches. The extremely high leaching of
alkalis due to the harsh exposure condition (i.e., about 45% and 80%
alkalis leached out at 8 weeks and 1 year, respectively) led to an
increased alkali content of 18.5 kg/m3, Na2Oe after the compensation,
which is out of the alkali range proposed to build this model (i.e., 12.4
kg/m3, Na2Oe). This observation also indicates that the test results from
the samples in 100% RH should be preferable for the calculation of the
ideal expansion curve.
Multon and Sellier [7] developed a multi-scale analysis on the same
set of experimental data from Lindgård’s work to obtain a theoretical
expansion curve of the concrete without leaching. The result is shown in
Fig. 7(b). It shows that the ideal expansion curves obtained from the first
two approaches in this study are comparable to the multi-scale model­
ling results reported in Multon and Sellier [7], thus suggesting that the
semi-empirical and multi-scale models could produce similar results
whether no leaching occurs. In addition, it highlights a great impact of
leaching on ASR-induced expansion, which is important for evaluating
the expansion of the field exposure members with no or minimal
leaching.

5. Forecasting ASR-induced expansion in the field

In this section, the proposed semi-empirical model is implemented to

Fig. 7. Model validation on Lindgård’s data [18] for the ideal expansion curve.
expansion observed. In an idealised scenario, however, there might be
an identical expansion curve for this particular concrete mixture
regardless of the specimen dimensions and exposure conditions if there
was no alkali leaching.
model ASR-induced expansion of field blocks incorporating different
reactive aggregates, alkali contents, and exposed to outdoor conditions
to appraise its applicability for field members.
5.1. Implementation of the semi-empirical model for forecasting
expansion of field exposed concrete blocks

4.3.3.2. Model validation. The experimental data (i.e., expansion,
leaching of alkalis) of these three series were collected and utilised for
the calculation of an ideal “no leaching” expansion curve for the concrete
mixture using the proposed model for alkali content and leaching con­
siderations, namely approach 1, approach 2 and approach 3,
Comparing ASR affected concrete in the laboratory and in the field in
Section 2.1 shows two main reasons leading to the difference in their
expansion. First, more leaching from smaller specimens in “leaching
preferable” conditions reduces the expansion level of concrete in the
laboratory [13–15]. Second, differences in exposure conditions lead to

Table 1
Experimental data and semi-empirical model parameters for the concrete tested in Lindgård et al. (2013).
Lindgård et al. (2013) [18] Approach 1 Approach 2 Approach 3 Reference for the calculation

Test series 100 × 100 mm 70 × 70 mm Prism 70 × 70 mm Prism (In

Prism (100% RH) (100% RH) de-ionised water)

Experimental and fitting Initial alkali content (kg/m3, 3.75 Test data [18]
data (leaching cases) Na2Oe)
kε,A0 0.575 Fig. 4
τC 130 85 40 Curve-fitting of the measured
expansion curve (Fig. 7(a))
τL 205 185 145
ε∞ 0.435 0.300 0.045
one-year leaching (%) 16.5 33.2 79.5 Test data [18]
Model parameters (non- Increased alkali content (kg/m3, 4.49 5.61 18.29 Eq. (13)
leaching cases) Na2Oe) (After compensation)
kC,LA 0.865 0.728 0.348 Eq. (16)
kL,LA 0.921 0.841 0.618 Eq. (17)
kε,A 0.798 1.147 3.866 Fig. 4
kε,A 1.387 1.995 6.723
kε,A (3.75)
τC 150.3 116.8 114.9 Eq. (15)
τL 222.6 220.1 234.5 Eq. (18)
ε∞ 0.603 0.599 0.303 Eq. (19)

7
T.N. Nguyen et al.
different expansion rates of laboratory-exposed and field-exposed con­
crete, in which laboratory-exposed concrete normally expands at a
higher rate under accelerated conditions when compared to concrete
exposed to environmental conditions in the field [15,22]. This is clearly
shown in Fig. 1(b), which compares the expansion of concrete con­
taining Spratt limestone tested in the laboratory and in the field.
In order to correlate the induced expansion of concrete in the labo­
ratory to that of field concrete, these two main differences have to be
taken into consideration. In this regard, a two-step procedure is pro­
posed, as shown in Fig. 8. In general, the first step is to account for the
leaching of laboratory concrete, while the second step is to take into
consideration the differences in exposure conditions between laboratory
and field, along with possible changes in field conditions over time. In
the first step, an ideal expansion curve for the “no leaching” scenario is
estimated using the procedure presented earlier in Section 4.3 by taking
into account the alkali leaching. As previously discussed, the leaching of
alkalis from concrete samples/members mainly depends on its size and
the exposure conditions. In this regard, minimal or no leaching is
commonly observed for the field members (i.e., large in size and not in
the “leaching preferable” environmental conditions). The ideal expansion
curve is essential to simulate the possible expansion level of the concrete
members in the field assuming no leaching occurs. The inputs obtained
from the laboratory tests include expansion over time, initial concrete
alkali content and alkali leaching at one year. The implementation
procedure of this step is presented in detail in Section 4.3. It is worth
noting that the ideal expansion curve is simulated at reference labora­
tory exposure conditions (i.e., 38 ◦ C and 100% RH); therefore, envi­
ronmental conditions need to be considered in step 2 to model the time-
dependent expansion curve of field concrete. It is also important to note
that modelling of field concrete considering long-term leaching of alkalis
is implementable using the same procedure if leaching data is available.
In field concrete members, ASR-induced development is by far more
complicated than laboratory specimens subjected to accelerated condi­
tions due to many factors, such as continuous changes in the ambient
temperature and RH, effect of rainfall, solar radiation and other damage
mechanisms (i.e., shrinkage, creep, freezing and thawing) [28,35,39].
For the sake of simplicity, some important assumptions need to be made
in modelling expansion of field concrete, such as: (1) there is no alkali
leaching from the field concrete members, (2) internal temperature is
assumed to be similar to the ambient temperature, and (3) internal RH is
assumed to be uniformly distributed throughout the concrete specimen. These
assumptions have been reasonably used in many other studies, such as in
Kawabata, Yamada, Ogawa, Martin, Sagawa, Seignol and Toutlemonde
[27] and Gorga, Sanchez and Martín-Pérez [28]. In addition, impacts of
other mechanisms such as shrinkage, freezing and thawing, sulphate
Fig. 8. Overall procedure for modelling expansion of concrete in the field.
8
Cement and Concrete Composites 134 (2022) 104817
attack, etc., on ASR-induced expansion are not accounted for in this
model.
Another factor that needs to be considered for field concrete is the
potential release of alkalis from the aggregates, which could signifi­
cantly increase the amount of alkalis in the concrete pore-solution in the
long term. According to Bérubé, Duchesne, Dorion and Rivest [20],
reactive aggregates could contribute alkalis in amounts as high as 2.5
kg/m3, Na2Oe in long-term concrete exposure. This important contri­
bution could significantly increase the induced expansion of field con­
crete in comparison to the short-term exposed concrete in the
laboratory; however, there is currently no experimental data available
for this time-dependent alkali release in field concrete. Therefore, the
contribution of alkalis from aggregates (RA, kg/m3) is considered in this
study as an additional amount to the total concrete alkali content. As
such, after calculating the ideal “no leaching” expansion curve for the
concrete mixture with the original alkali content of Ao (kg/m3), the al­
kali content-dependent coefficients of Ao and Ao + RA alkalis are
adopted to update the ultimate expansion (i.e., considering the alkali
contribution from the aggregates). It is important to note herein that a
test procedure for long-term alkali release determination is still not
available, thus the amount of releasing is commonly estimated from
short-term testing of aggregates.
After finalising the ideal expansion curve at the reference laboratory
exposure conditions (i.e., 38 ◦ C and 100% RH), the effect of environ­
mental conditions in the field is considered through temperature-
dependent and RH-dependent coefficients presented in Section 4.2.
Together with the assumptions made above, Fig. 9 illustrates the
calculation procedure to capture the effect of the field environmental
conditions, which is adopted from Kawabata et al. [27]. As the field
conditions change over time, consequent changes of temperature and
RH in concrete and their effects on ASR-induced expansion have to be
continuously considered. Each set of temperature and RH produces an
expansion curve of the concrete for a particular environmental condi­
tion, namely a master curve. To capture the change of temperature and
RH over time, expansion calculation is implemented through the
determination of the incremental expansion of every single time step Δt.
The incremental expansion at the time ti is calculated based on the
master curve of the corresponding environmental conditions (i.e., Ti,
RHi). Therefore, it is noticed herein that the shorter the time step Δt, the
more accurate the field conditions changes are captured. In this study,
the monthly average ambient temperature was utilised for the expansion
calculation of the concrete members in the field; thus, the time step of 1
month was adopted in this work.
The validation of the proposed semi-empirical model has then been
conducted using three important experimental campaigns: (1) Kingston
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817

5.2.2. CANMET and UT exposure sites

The CANMET and UT exposure sites were established as a part of a
comparative field and laboratory study using a wide variety of reactive
aggregate types and natures across North America to study the efficiency
of laboratory testing in representing field performance in different cli­
matic conditions as well as to investigate the effectiveness of numerous
preventive measures [14,17]. Concrete mixtures incorporating several
reactive aggregates and different concrete alkali contents were tested for
expansion in both the laboratory (i.e., CPT) and in the field (i.e., 0.40 ×
0.40 × 0.70 m blocks manufactured in Ottawa - CANMET and Texas
-UT). There are two main series of concrete blocks tested in these two
sites: (1) an “exchange series” where concrete blocks were made at
CANMET to be tested at both CANMET and UT sites, and (2) blocks with
almost identical mixtures both made and tested at CANMET and UT. The
CPT expansion test results and corresponding field data of the second
series were reported in Refs. [12,16,17] up to 20-year exposure, while
tested data of the exchange series is available only up to 3 years. In this
study, the data of two concrete mixtures (i.e., containing Spratt lime­
stone and Sudbury gravel) are collected for the model validation. The
mix proportioning used for the concrete mixtures incorporating these
two reactive aggregates are almost identical from the two sites; there­
fore, the results are preferable to evaluate the effect of environmental
conditions. Fig. 10 shows the information on weather conditions,
including yearly average temperature, relative humidity and precipita­
tion. The environmental conditions in Kingston are almost the same as in
Ottawa (CANMET); therefore, the same temperature was used to
calculate concrete expansion from both sites.
In terms of precipitation effect, according to the field study of
“PARTNER” project [40], expansions of concrete blocks exposed to
ambient rainfall and partially immersed in water are about the same. In
addition, Kagimoto and Kawamura (2011) [29] measured the RH dis­
tribution in large concrete samples and suggested that the ambient RH
affects only the cover of the members. As per the authors, it takes years
to balance the internal RH in the concrete cover and the ambient RH,
while the concrete core most likely remains saturated. In the context of
the semi-empirical model proposed in this study, the internal RH of large
Fig. 9. Schematic procedure for the consideration of environmental conditions.
concrete specimens exposed in the field is first assumed to be 100% and

exposure site, ON, Canada [15,22]; (2) CANMET exposure site, ON,
Canada [12,16,17]; and (3) The University of Texas at Austin, USA (UT)
[14,17].

5.2. Overview of exposure sites/samples and climatic conditions

5.2.1. Kingston exposure site

This outdoor exposure site was established by the Ontario Ministry of
Transportation (MTO) in Kingston, Ontario, Canada in 1991, which
aimed at providing correlation between short-term laboratory tests and
long-term performance, as well as evaluating the efficiency of SCMs in
mitigating ASR development [15,22]. Different concrete mixtures
incorporating the Spratt reactive aggregate, both high- and low-alkali
cements, and several types and proportions of SCM have been used in
this experimental campaign. The low- and high-alkali cements (LAPC
and HAPC) contained 0.46% and 0.79% Na2O equivalent, respectively,
which correlates with 3.33 kg/m3 and 1.91 kg/m3, Na2Oe in the con­
crete, respectively. The field study was conducted on both pavement
slabs (0.2 × 1.2 × 4 m) and beams (0.6 × 0.6 × 2 m), both reinforced and
non-reinforced, while three prisms (75 × 75 × 400 mm) were manu­
factured from the same concrete mixtures for laboratory testing. In Doug
Hooton, Rogers, MacDonald and Ramlochan [15], the authors reported
for up to 20-year data of all these field-exposed elements and CPT
samples. A recent report from the MTO in 2018 presented the updated
27-year data of this field exposure site [22]. In this study, only the data
from the CPT samples and non-reinforced field beam elements (with
both low- and high-alkali cements) were utilised for the purpose of Fig. 10. Average yearly temperatures (A) and precipitation (B) at Austin and
validation of the proposed model. Ottawa & Kingston, adapted from Ref. [17].

9
T.N. Nguyen et al.
applied uniformly in the specimen, remaining the same throughout its
exposure duration for the simplification of the expansion calculations. It
is worth noting that the precipitation rate is high in both Ottawa and
Austin. The sensitivity of RH to the field exposed blocks expansion is
then considered in a parametric study.
In this study, the leaching of alkalis at 1 year of concrete with Spratt
and Sudbury are assumed to be 21.9% and 35.1% of the total concrete
alkali content, respectively, which are estimated at about 85% of the
leaching amount measured 1.5-year in Sinno and Shehata (2019) [19]
from the CPT. In addition, the alkali release from the reactive aggregates
may also be important in the estimation of the field expansion. As such,
Bérubé, Duchesne, Dorion and Rivest [20] tested the alkali release of the
Spratt and Sudbury aggregates through crushing (i.e., 1.25–5 mm par­
ticles in size) and immersion of the aggregates particles in different so­
lutions (i.e., distilled water, saturated lime and 0.7 N NaOH (or KOH)
solutions) for 578 days at 38 ◦ C; the 0.7 M alkaline solution was the most
similar to the concrete pore solution bearing a high alkali Portland
cement. The amount of alkali release in 0.7 M alkaline solution after 578
days of the Sudbury aggregate was more than 0.166% per aggregate
mass compared to only 0.009% for the Spratt aggregate. This result is
reasonable since the total Na2Oe content in the Sudbury and Spratt
aggregates measured in the same study was 4.35% and only 0.09% per
aggregate mass, respectively.
Moreover, the authors stated that the amount of alkalis released in
the long term from concrete incorporating the above coarse aggregate
particles and exposed in the field could be as high as from the crushed
particles measured after 578 days in this aggressive condition. With the
significant amount of alkalis possibly contributing to ASR-induced
development from the Sudbury aggregate, the expansion calculation
of the Sudbury blocks without and with the alkali contribution of
0.166% per aggregate mass (i.e., 1.75 kg/m3, Na2Oe of concrete) were
conducted for comparison purposes. Conversely, the alkalis released
from the Spratt aggregate is assumed negligible.
5.3. Modelling expansion of Kingston non-reinforced concrete beams
Fig. 11 shows expansion data of 27-year field-exposed beams
incorporating low and high-alkali cement (LAPC and HAPC), alongside
CPT expansion curve of the HAPC concrete. Comparative field/labora­
tory test results have shown that the ultimate expansion obtained from
the CPT is significantly lower than that of the field beam at 27-year
Fig. 11. Experimental results of field beams incorporating the reactive Spratt
aggregate and exposed in Kingston exposure site, along with the corresponding
CPT [22].
10
Cement and Concrete Composites 134 (2022) 104817
exposure. This difference is considered in the proposed model by
considering the alkali leaching from the test specimens in the CPT.
Furthermore, the concrete beams exposed to lower average (yearly)
temperatures in the field (i.e., the maximum temperature of about 27 ◦ C
– Ottawa data from Fig. 10) show significantly lower but somewhat
more sustained expansion rates over the full testing period when
compared to prisms at 38 ◦ C. The temperature-dependent model is
adopted to account for this environmental condition difference.
As stated previously, inputs for the proposed semi-empirical model
are the initial concrete alkali content, the expansion over time and
leaching of alkalis from the CPT test. The total Na2O equivalent of the
LAPC and HAPC concrete mixtures were 1.92 kg/m3 and 3.33 kg/m3,
Na2Oe (i.e., cements of 0.46% and 0.79% Na2Oe), respectively, which
were all from the cement. Similar to the expansion model of concrete in
CANMET and UT exposure sites presented in the previous section, the
leaching of alkalis at 1 year of concrete with Spratt is assumed to be
21.9% of the total concrete alkali content. The calculation of model
parameters for the ideal “no leaching” expansion curve of HAPC concrete
is shown in Table 2. The ideal expansion curve of LAPC is derived from
the HAPC’s using the proposed relationship in Fig. 4 based on their
corresponding alkali contents. Since the internal RH was assumed to be
100%, only the monthly average temperatures (see Fig. 10) were used to
consider the environmental conditions in Kingston, Ontario.
6. Results
For both concrete mixtures HAPC and LAPC, the modelling for field
beams were carried out with and without considering the alkali leaching
from the CPT test for comparison purposes. Fig. 12 shows the predicted
results in comparison to the measured data. As one observes, without
considering the alkali leaching (i.e., directly using the experimental
results from the CPT), the estimated expansions are significantly smaller
than the measured data from both HAPC and LAPC concrete mixtures.
Such finding is consistent with the experimental observations from
many other studies when comparing laboratory and field expansion
[13–16]. By taking the leaching of the CPT into consideration, the
modelling provided excellent agreement with the measured data. For
the HAPC (Fig. 12(a)), the ultimate expansion of the modelled expansion
curve is slightly higher than the maximum expansion measured from the
blocks. Conversely, the calculated ultimate expansion of the LAPC
(Fig. 12(b)) is marginally smaller than the measured data. However, the
expansion curve of the LAPC beam shows no sign of flattening; there­
fore, the estimated expansion could significantly underestimate the
expansion of the LAPC beam after 27 years. These differences could
come from the difference in one-year alkali leaching amount of HAPC
and LAPC, which is assumed the same for concrete mixtures of different
alkali contents. Furthermore, the proposed unique relationship between
alkali content and ultimate expansion may also contribute to the dif­
ferences. It is also possible that there might be alkali contribution from
the fine aggregate used in the concrete (as alkali releasing from the
Spratt limestone is insignificant according to Bérubé et al. [20]), which
Table 2
Model parameters for ideal expansion curve of LAPC and HAPC concretes.
Test data and model parameters Model parameters for the ideal “no-leaching”
derived from test data expansion curve
Initial concrete alkali 3.33 Increased alkali content (kg/m3, 4.27
content (kg/m3, Na2Oe) (After compensating the
Na2Oe) leaching)
kε,Ao 0.423 kC,LA 0.781
τC 120 kL,LA 0.846
τL 120 kε,A 0.721
ε∞ 0.165 kε,A /kε,Ao 1.702
1-year leaching (%) 21.9 τC 153.7
τL 141.8
ε∞ 0.281
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817

Table 3
Model parameters for ideal expansion curves of Spratt and Sudbury blocks tested
at CANMET.
CANMET and UT sites Spratt Sudbury

Test data and model Initial alkali content (kg/m3, 5.25

parameters derived from Na2Oe)
test data kε,Ao 1
τC 100 85
τL 130 215
ε∞ 0.250 0.200
1-year leaching (%) 21.9 35.1
Model parameters for the Updated alkali content (A, 6.72 8.09
ideal “no-leaching” kg/m3) (After compensated)
expansion curve kC,LA 0.781 0.649
kL,LA 0.846 0.754
kε,A 1.597 2.150
kε,A /kε,Ao 1.597 2.150
τC 128.1 131.0
τL 153.6 285.0
ε∞ 0.399 0.430

Fig. 12. Model outcomes in comparison to the experimental data: (a) HAPC
and (b) LAPC non-reinforced beams incorporating the reactive Spratt limestone,
MTO outdoor exposure site [22].
keep active the expansion of the HAPC and LAPC. This contribution,
however, is not considered in the current calculation due to limited data
from the test program.
In general, the proposed model, which considers the effects of the
concrete alkali content and alkali leaching, significantly improves the
correlation between laboratory testing and long-term field performance
of the ASR-affected concretes studied. The result also suggests the need
of considering the effect of alkali leaching of laboratory samples in the
estimation of field performance. However, it also highlights the impor­
tance of having the data of additional alkalis contributions from ag­
gregates or other sources than the cement in forecasting the long term
“sustained” expansion rates of field specimens.
6.1. Modelling expansion of CANMET non-reinforced concrete blocks
Test data of concrete incorporating Spratt and Sudbury reactive ag­
gregates and various concrete alkali contents investigated in CANMET
and UT sites were collected for validation purposes. Three typical
amounts of Na2Oe were used in this experimental campaign: 0.4%, 0.9%
and 1.25% (boosted with additional reagent grade NaOH) per cement
mass, equivalent to approximately 1.7, 3.8 and 5.25 kg/m3, Na2Oe,
respectively. Table 3 presents the calculation of model parameters for
the ideal “no leaching” expansion curve for mixtures made of Spratt and
Sudbury aggregates bearing 5.25 kg/m3, Na2Oe. The Sudbury concrete
shows lower ultimate expansion from the measurements but higher
ultimate expansion for the ideal expansion curve when compared to the
Spratt. This observation could be explained by the significant alkalis
leaching from Sudbury concrete prisms in the laboratory [19] and strong
possibility of alkali release from the Sudbury gravel material in the field
blocks [20].
Table 4 shows the ultimate expansion calculated for other mixtures
varying in concrete alkali content. Fig. 13 shows the predicted results of
the mixtures containing Spratt and Sudbury aggregates bearing 5.25 kg/
m3, Na2Oe, with and without considering the alkalis leaching. Similar to
the observation from Kingston’s beams, it is evident that without
considering the leaching effect, the model significantly underestimates
field expansions of blocks made of both Spratt and Sudbury aggregates.
In this regard, the model that accounted for the leaching provides a
much better estimation of the induced expansion of Spratt blocks. The
model results are closely matched with experimental observations for
induced expansion up until 0.1%, and slightly higher than the measured
one at higher expansion levels (i.e., above and beyond 0.15%, from 5
years to 20 years of exposure). The rationales for this difference could be
from model parameters obtained from the measurements, as well as a
need of experimental constants to the model which proposed in Bru­
netaud et al. (2004) [41]. More investigations are required in this regard
for an improvement of the model.
For the Sudbury blocks, despite improving the expansion estimation
by considering the leaching of alkalis in the laboratory, the predicted
expansion is still remarkably lower than the field measurements, i.e.,
0.408% from the model compared to 0.651% from the measurement at
20 years of exposure (see Fig. 13(c)). Conversely, the model with
additional consideration of alkali release, as shown in Fig. 13(d), pro­
vides a significantly better expansion estimation when compared to the
previous one. By raising the total alkalis from 5.25 kg/m3 to 7.0 kg/m3,
Na2Oe due to alkali contribution from the aggregate, the ultimate
expansion increases by about 70% compared to the model outcome
ignoring the releasing effect. This result highlights the significant impact
of alkali contribution from the aggregates on the induced expansion of
long-term exposed concrete in the field. It also emphasises the validity of
the alkali release results reported in Bérubé et al. (2002) [20], which are
negligible for the Spratt aggregate and significant for the Sudbury
aggregate.
Fig. 14 illustrates the model outcomes for the mixtures made of
Spratt and Sudbury aggregates varying in concrete alkali level compared
to the measured data. For the Spratt concrete blocks, the model captures
very well the expansion of concrete blocks with 0.9% and 1.25% alkali
content (i.e., 3.8 and 5.25 kg/m3, Na2Oe, respectively). At low alkali
content (i.e., 0.4% of alkali, or 1.68 kg/m3, Na2Oe), however, the model
underestimates the expansion of the concrete block. For the Sudbury
concrete blocks, by accounting for the alkali contribution by the

11
T.N. Nguyen et al.
Table 4
Calculation of the ultimate expansion of mixtures varying in concrete alkali content.
Spratt
3
A (kg/m ) 5.25 3.78 1.68
kε,A 1 0.588 0.108
ε∞ (%) 0.399 0.235 0.043
(*)
Considering additional long-term alkali contribution from the aggregate.
Fig. 13. Model outcomes in comparison to the experimental data of concrete blocks stored on the CANMET outdoor exposure site (experimental data from Ref. [12]).
aggregate, the model also provides excellent results for the concrete
made of cement with 0.9% of alkalis (i.e., 3.78 kg/m3, Na2Oe). This
result again shows the reasonability of considering the alkali contribu­
tion by the aggregates. The model outcomes for both Spratt and Sudbury
concrete blocks also confirms that the alkali content – expansion rela­
tionship proposed in Section 4.3 is suitable for modelling the induced
expansion of concrete mixture varying in alkali contents, yet, the
improvement of this relationship is necessary for concrete with low al­
kali content levels, (i.e., <2 kg/m3, Na2Oe) by obtaining more experi­
mental data within these levels.
6.2. Modelling expansion of UT non-reinforced blocks
This section aims to further validate the proposed model at different
environmental conditions, i.e., warmer climatic conditions in Texas (UT
site), and cooler climatic conditions in Ottawa (CANMET site). The cli­
matic conditions in Texas, as shown in Fig. 10, indicate almost the same
high level of precipitation but significantly higher temperature when
compared to Ontario’s climatic conditions. The field expansion of con­
crete blocks in the UT site was thus about 4–5 times faster than the
blocks with the same concrete mixtures in CANMET [17].
The model parameters established earlier for the ideal “no leaching”
expansion curves of the mixtures made of Spratt and Sudbury reactive
Cement and Concrete Composites 134 (2022) 104817
Sudbury
5.25 3.78 5.25 + 1.75 (RA)(*) 3.8 + 1.75 (RA) (*)
1 0.588 1.708 1.113
0.430 0.253 0.735 0.479
aggregates at CANMET were utilised in addition to environmental
conditions data in Texas to determine the expansion of the corre­
sponding concrete blocks stored at UT using the proposed procedure in
Section 5.1. Fig. 15 shows the model outcomes of concrete blocks made
of Spratt and Sudbury at CANMET and UT sites compared to the field
observations. The model outcomes are relatively close to the block
measurements made and kept at UT for both mixtures made of Spratt
and Sudbury reactive aggregates. Similar to the experimental observa­
tions of [17], the ASR-induced expansion calculated for blocks kept at
UT was obviously faster than CANMET’s, i.e., about 4–5 times. For the
exchange series, the measured ones are notably smaller than the
calculated ones. This difference could be from a delay of expansion due
to transportation of the blocks in the first year. In general, the proposed
model provides a reasonably good estimation of ASR-induced expansion
from the two sites with different environmental conditions by using only
a single set of laboratory testing data.
7. Discussion
7.1. Consideration of alkali leaching from test samples and alkali
releasing from aggregate: their importance and limitation
It is generally considered that the concrete prism test (CPT) is the
12
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817

considering impacts of long-term alkali leaching of the field concrete on

the induced expansion is feasible using the proposed model if the
leaching data is available; for instance, for small concrete members
exposed to “leaching-preferable” field conditions, i.e, immerged in fresh
water.
The procedure presented in Section 5.1 shows its capability to ac­
count for the effects of the alkali content, alkali leaching from laboratory
small samples and alkali release from aggregates in long-term exposed
field concrete to predict ASR-induced expansion in field concrete. The
validity of the proposed procedure is shown by the good agreement
between modelling and field results for unreinforced concrete specimens
incroporating Spratt reactive aggregate and stored in three different
exposure sites (i.e., Kingston, CANMET and UT); whereas, the model
without considering the leaching of alkalis obviously underestimates
ASR-induced expansion of affected concrete in all these three exposure
sites.
On the effect of alkali leaching, the proposed alkali content coeffi­
cient (i.e., the relationship between the ultimate expansion (ε∞) and
alkali content) was based on laboratory data of one-year CPT from
various studies. It is important to note that the leaching levels of the
concrete may vary from one study to another, which leads to significant
variations of the coefficient and cause difficulties in precisely evaluating
the impact of the concrete alkali content. Hence, data of the ultimate
expansion measured from concrete at different alkali contents, either
from the field exposed blocks or from laboratory testing with limited
leaching such as the concrete cylinder test (CCT) [24], is very useful to
improve the alkali content–induced expansion relationship. Moreover,
the current practice of using CPT is mainly for assessing the potential
reactivity of aggregates; therefore, attention is normally not paid to
measuring the alkalis leaching. The success of the proposed
semi-empirical model emphasised the great significance of alkali
leaching measurements in laboratory expansion testing for predicting
the potential of future expansion and deterioration (i.e., prognosis) of
ASR-affected concrete in the field.
Furthermore, alkali release was reported in several studies to be
significant for some particular aggregates, such as Sudbury gravel [19,
20], which is thought to contribute to the expansion of long-term
exposed field concrete as previously observed by Ref. [12]. Therefore,
the proposed model still underestimates the expansion of Sudbury field
concrete blocks despite considering the leaching of alkalis. By addi­
tionally considering the alkali contribution from Sudbury aggregate
using the data from Ref. [20], the model provides a better estimation of
the field blocks expansion. However, it is worth noting that the reli­
ability of test methods for assessing the alkali release from aggregates is
still under investigation, and therefore, more attention should be paid to
obtain better measurements of the potential alkali contribution from
aggregates [42].

Fig. 14. Model outcomes in comparison to the experimental data of specimens

7.2. Effect of environmental conditions: temperature and RH
incorporating different alkali contents and stored on the CANMET outdoor
exposure site (experimental data from Ref. [12]).
In the previous sections, the average monthly concrete temperature
and 100% internal RH were reasonably assumed for forecasting ASR-
most reliable standard test method for assessing the potential alkali-
induced expansion of field concrete specimens. The temperature in
reactivity of aggregates in the laboratory; however, the use of CPT test
concrete, however, could vary due to the continuous change of the
results for prediction of ASR-ultimate induced expansion (i.e., maximum
ambient temperature and effects of solar radiation and snow coverage.
expansion) in the field is not always straight forward [13,17]. Specif­
For instance, the concrete temperature in the skin portions could
ically, CPT expansion results were found to underestimate the long-term
significantly increase under solar radiation [28,43]. The relative hu­
expansion of field concrete members with or without incorporating
midity in concrete normally remains very high (i.e., saturated) in
various types of supplementary cementitious materials [13]. As afore­
high-humidity climates, plus the frequent water supply from rain as the
mentioned, the main reason for this lack of laboratory-field correlation
conditions in Ottawa and Texas in the above calculations. Yet, it must be
is thought to be the alkalis leaching of concrete, which is significantly
noted that the RH in the skin portions of concrete blocks could reduce
higher from laboratory test specimens (and minimal in field members
over time to balance with the ambient RH, especially when cracking
commonly due to larger size samples and less “leaching-preferable”
occurs. Therefore, it is important to evaluate the effect of these tem­
exposed conditions). Therefore, consideration of the alkalis leaching is
perature and RH variations on concrete expansion in the field. As such, a
crucial in forecasting ASR-induced expansion in field concrete members
parametric study was conducted on concrete mixtures made of Spratt
based on laboratory observations. It is also worth noting that
and Sudbury reactive aggregates from the CANMET exposure site to

13
T.N. Nguyen et al.
Fig. 15. Model outcomes in comparison to the experimental data of concrete blocks kept at CANMET and UT sites.
appraise the sensitivity of the field concrete expansion to variations in
temperature and RH. Minimum and maximum temperatures (given in
Fig. 10 as low and high temperatures, respectively) were adopted in the
semi-empirical model for such assessment. The variation of temperature
herein is about ± 5 ◦ C of the average temperature which was used in the
previous calculations. Moreover, a lower RH of 95% is adopted in the
model with a simplification of the same RH throughout the exposure
period. It is worth noting that the assumption of uniform temperature
and RH throughout the concrete member is still maintained.
Fig. 16 illustrates the field expansion calculated at the minimum,
maximum and average temperatures. It is obvious that the temperature
significantly influences the rate of ASR-induced expansion. For example,
for both Spratt and Sudbury concrete blocks, the expansion rate could be
two times lower if the temperature reduces, for instance, by 5 ◦ C. The
field expansion at the minimum and maximum temperatures thus
completely cover all the experimental data points for both mixtures. This
important sensitivity, however, requires a more comprehensive analysis
of internal temperature in the long-term exposed concrete members to
provide a better estimation of ASR-induced expansion.
The model results of the field expansion for the internal RH of 95%
are shown in Fig. 17. One can verify a strong effect of the RH on ASR-
induced ultimate expansion. For example, on Spratt blocks, the ulti­
mate expansion reduces from 0.391% to only 0.260% when reducing the
internal RH from 100% to 95%. It is interesting to notice that the field
expansion calculated for both Spratt and Sudbury blocks at 100% and
95% RH covers very well the measured data. This emphasises the pos­
sibility of the variation in the relative humidity during the long-term
exposure of field concrete.
7.3. The use of the proposed model for estimating the potential of further
ASR-induced expansion
In general, the proposed model provides a reasonably good estima­
tion of ASR-induced expansion from the two sites with different envi­
ronmental conditions using only a single set of laboratory testing data.
With this capability of the model, a single set of CPT test data can be
used to simulate the free expansion of concrete at different locations
varying in environmental conditions. In addition, another significant
advantage of the proposed model is for estimating the potential of
further ASR-induced expansion. Nowadays, several state-of-the-art
diagnosis techniques have been proposed to access and quantify the
current expansion/or damage level of ASR-affected concrete, such as
Damage Rating Index (DRI), Stiffness Damage Test (SDT) or Cracking
Index (CI) [2,43–45]. However, there are still limited numbers of
effective techniques for the prognosis of the ASR-induced development
to forecast how ASR will progress in the coming months or years. After
having reliable quantification of the current damage level using the
abovementioned diagnosis techniques, the proposed model could be
14
Cement and Concrete Composites 134 (2022) 104817
used to forecast the potential future expansion of the affected structure
throughout its service life. For instance, Fig. 18 illustrates the model
outcomes for Sudbury concrete blocks from UT and CANMET exposure
sites as a clear indication of the use of the proposed model for ASR
prognosis.
Assuming that the current expansion level could be estimated using
diagnosis techniques such as DRI, SDT and CI, the concrete blocks at UT,
after a 3-year exposure, could potentially keep swelling up to 0.55%
within the next 5 years before levelling off. Otherwise, concrete blocks
stored at CANMET over 20 years would potentially further swell “only”
about 0.05% within the same time period. This approach, with addi­
tional considerations on confinement and restraint conditions (i.e., via
numerical assessment), could be extremely helpful for the prognosis of
ASR-affected concrete structures in the field along with determining
efficient rehabilitation method(s) and management strategies for
affected infrastructure.
8. Concluding remarks
A semi-empirical model was proposed to correlate the laboratory
testing to field concrete expansion, which is essential for the prognosis of
ASR affected concrete structures. In addition to the laboratory expansion
data, the model took into consideration key factors affecting the
expansion behaviour both from laboratory testing (i.e., concrete alkali
content, alkali leaching from laboratory samples) and from the field (i.
e., the temperature and relative humidity of field concrete and alkali
contribution from the aggregates) in the form of Larive’s model. The
proposed model was able to reproduce the expansion behaviour of
concrete blocks exposed to different field conditions, such as Kingston
site (Ontario, Canada), CANMET site in Ottawa (Ontario, Canada) and
UT site in Texas (Austin, USA). Based on the model development and
results, the following conclusions can be drawn:
• The leaching of alkalis is significant in laboratory expansion tests on
small samples, affecting both the ASR kinetics and the ultimate
expansion of the concrete. In contrast, it is commonly considered
minimal in the field concrete blocks/members. By using the expan­
sion curve and leaching at 1 year measured from laboratory testings,
the proposed model is capable of estimating an ideal expansion curve
without leaching to reproduce the expansion of field concrete. The
model was well-calibrated using the different reliable laboratory
testing results. Compared to the prediction without considering the
leaching, using the proposed model shows significant improvement
in forecasting expansion of non-reinforced concrete beam exposed
for 27 years at the Kingston site and concrete blocks kept at CANMET
for 20 years. Especially, the model results of Spratt concrete are
closely matched the field observations from both Kingston and
CANMET specimens. In this regard, measurement of the alkali
T.N. Nguyen et al.
Fig. 16. Field expansion of concrete blocks at CAMENT at various tempera­
tures. (Note: Taverage, Tmax and Tmin are obtained from Fig. 10a as average,
high and low temperatures, respectively).
leaching is suggested to be very important to increase the reliability
of the CPT in appraising the maximum potential expansion and
damage degree of the field concrete.
• For the long-term exposed concrete, the total alkali content of the
concrete is contributed not only from the cement but also possibly
from the coarse and fine aggregates. A remarkable amount of alkalis
released from Sudbury reactive aggregate measured in Ref. [20] was
adopted to raise the total alkali content of this concrete mixture for
the long-term expansion estimation. The excellent agreement be­
tween the model outcomes and measurement results of the 20-year
exposed concrete blocks containing Sudbury reactive aggregate at
CANMET site highlighted the significance of alkali contribution from
aggregates to the long-term expansion of concrete in the field.
15
Cement and Concrete Composites 134 (2022) 104817
Fig. 17. Field expansion of concrete blocks at CAMENT at various RH inputs.
• The effect of environmental conditions (i.e., ambient temperature
and relative humidity, precipitation) in this study is considered
indirectly through concrete temperature and relative humidity. The
model was able to reproduce the expansion of both Spratt and Sud­
bury concrete blocks at CANMET and UT sites (i.e., the same mixture
composition) using the same set of laboratory test results and their
corresponding environmental conditions. This is crucially important
in forecasting and correlating the expansion of concrete members
using the same reactive aggregates and mixture exposed to different
environmental conditions in a country. It is important to note that
the concrete temperature highly influences the expansion rate,
whereas the relative humidity greatly impacts the ultimate expan­
sion. In addition, the modelling results also show that the assumption
T.N. Nguyen et al.
Fig. 18. Forecasting potential future expansion using the proposed model.
of high internal RH such as 95–100% is reasonable for the concrete
blocks kept in relatively high ambient RH and precipitation envi­
ronmental conditions such as at CANMET-Ottawa and UT-Texas.
The proposed simplified empirical model provides a practical yet
effective tool for forecasting the ASR induced expansion of non-
reinforced concrete in the field based on laboratory testing results.
Yet, to be able to use for the prognosis of ASR affected concrete struc­
tures, a model developed for reinforced and prestressed concrete with
consideration of restraints is necessary and currently in progress of
development. Finally, additional reliable experimental data on ASR
expansion, alkali leaching and alkali releasing from aggregates from
both the laboratory and the field are essential to enhance the proposed
model.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Data availability
No data was used for the research described in the article.
Acknowledgements
This work is supported by the Australian Research Council Research
Hub (IH150100006) for Nanoscience Based Construction Materials
Manufacturing (NANOCOMM) and the industry partner Transport for
New South Wales (TfNSW).
References
[1] ISE, Structural Effects of Alkali-Aggregate Reaction: Technical Guidance on the
Appraisal of Existing Structures, 1992.
[2] B. Fournier, M.-A. Berube, K.J. Folliard, M. Thomas, Report on the Diagnosis,
Prognosis, and Mitigation of Alkali-Silica Reaction (ASR) in Transportation
Structures, 2010.
[3] L.F.M. Sanchez, B. Fournier, M. Jolin, D. Mitchell, J. Bastien, Overall assessment of
Alkali-Aggregate Reaction (AAR) in concretes presenting different strengths and
incorporating a wide range of reactive aggregate types and natures, Cement Concr.
Res. 93 (2017) 17–31.
16
Cement and Concrete Composites 134 (2022) 104817
[4] C. Larive, Apports combinés de l’expérimentation et de la modélisation à la
compréhension de l’alcali-réaction et de ses effets mécaniques, École Nationale des
Ponts et Chaussèes, Paris, 1997.
[5] B. Capra, J.-P. Bournazel, Modeling of induced mechanical effects of alkali-
aggregate reactions, Cement Concr. Res. 28 (2) (1998) 251–260.
[6] E. Grimal, A. Sellier, S. Multon, Y. Le Pape, E. Bourdarot, Concrete modelling for
expertise of structures affected by alkali aggregate reaction, Cement Concr. Res. 40
(4) (2010) 502–507.
[7] S. Multon, A. Sellier, Multi-scale analysis of alkali–silica reaction (ASR): impact of
alkali leaching on scale effects affecting expansion tests, Cement Concr. Res. 81
(2016) 122–133.
[8] V. Saouma, L. Perotti, Constitutive model for alkali-aggregate reactions, Materials
Journal 103 (3) (2006) 194–202.
[9] A. Sellier, E. Bourdarot, S. Multon, M. Cyr, E. Grimal, Combination of structural
monitoring and laboratory tests for assessment of alkali-aggregate reaction
swelling: application to gate structure dam, ACI Mater. J. 106 (3) (2009) 281–290.
[10] C. Comi, B. Kirchmayr, R. Pignatelli, Two-phase damage modeling of concrete
affected by alkali–silica reaction under variable temperature and humidity
conditions, Int. J. Solid Struct. 49 (23) (2012) 3367–3380.
[11] M.A. Hariri-Ardebili, V.E. Saouma, C. Merz, Risk-informed condition assessment of
a bridge with alkali-aggregate reaction, ACI Struct. J. 115 (2) (2018).
[12] B. Fournier, J. Lindgård, B.J. Wigum, I. Borchers, Outdoor Exposure Site Testing
for Preventing Alkali-Aggregate Reactivity in Concrete–A Review, 2018.
[13] M. Thomas, B. Fournier, K. Folliard, J. Ideker, M. Shehata, Test methods for
evaluating preventive measures for controlling expansion due to alkali–silica
reaction in concrete, Cement Concr. Res. 36 (10) (2006) 1842–1856.
[14] J.H. Ideker, A.F. Bentivegna, K.J. Folliard, M.C. Juenger, Do current laboratory test
methods accurately predict alkali-silica reactivity? ACI Mater. J. 109 (4) (2012)
395.
[15] R. Doug Hooton, C. Rogers, C.A. MacDonald, T. Ramlochan, Twenty-year field
evaluation of alkali-silica reaction mitigation, ACI Mater. J. 110 (5) (2013).
[16] B. Fournier, A. Bilodeau, N. Bouzoubaa, P.-C. Nkinamubanzi, Field and Laboratory
Investigations on the Use of Fly Ash and Li-Based Admixtures to Prevent ASR in
Concrete, International Conference on Durability of Concrete Structures, 2019.
[17] B. Fournier, J.H. Ideker, K.J. Folliard, M.D. Thomas, P.-C. Nkinamubanzi,
R. Chevrier, Effect of environmental conditions on expansion in concrete due to
alkali–silica reaction (ASR), Mater. Char. 60 (7) (2009) 669–679.
[18] J. Lindgård, M.D. Thomas, E.J. Sellevold, B. Pedersen, Ö. Andiç-Çakır, H. Justnes,
T.F. Rønning, Alkali–silica reaction (ASR)—performance testing: influence of
specimen pre-treatment, exposure conditions and prism size on alkali leaching and
prism expansion, Cement Concr. Res. 53 (2013) 68–90.
[19] N. Sinno, M.H. Shehata, Effect of sample geometry and aggregate type on
expansion due to alkali-silica reaction, Construct. Build. Mater. 209 (2019)
738–747.
[20] M.-A. Bérubé, J. Duchesne, J. Dorion, M. Rivest, Laboratory assessment of alkali
contribution by aggregates to concrete and application to concrete structures
affected by alkali–silica reactivity, Cement Concr. Res. 32 (8) (2002) 1215–1227.
[21] M.P.M.-A. Bérubé, N. Dupont, J. Frenette, M. Langlois, Expansion Test Methods for
Mass Concrete Exposed to Alkali-Aggregate Reaction, Canadian Electrical
Association, Montreal, 1992 (99pp. plus Appendices; Project CEA No.715 G687).
[22] MTO, Kingston, Outdoor Exposure Site for ASR – 27 Year Update, Soils and
Aggregates Section, Materials Engineering and Research Office, Ontario Ministry of
Transportation, 2018.
[23] T. Aashto, Standard Method of Test for Potential Alkali Reactivity of Aggregates
and Effectiveness of ASR Mitigation Measures (Miniature Concrete Prism Test,
MCPT), 380, American Association of State and Highway Transportation Officials,
2019.
[24] K. Chopperla, T. Drimalas, J. Tanesi, A. Ardani, M. Laskey, M. Thomas, K. Folliard,
J. Ideker, Benchmarking the miniature concrete prism test and the concrete
cylinder tests to field exposure blocks for mitigation of alkali-silica reaction, in:
Proc., 15th International Congress on the Chemistry of Cement, 2019. Prague,
Czech Republic.
[25] T.F. Rønning, B.J. Wigum, J. Lindgård, Recommendation of RILEM TC 258-AAA:
RILEM AAR-10: determination of binder combinations for non-reactive mix design
using concrete prisms–38◦ C test method, Mater. Struct. 54 (6) (2021) 1–30.
[26] F.-J. Ulm, O. Coussy, L. Kefei, C. Larive, Thermo-chemo-mechanics of ASR
expansion in concrete structures, J. Eng. Mech. 126 (3) (2000) 233–242.
[27] Y. Kawabata, K. Yamada, S. Ogawa, R.P. Martin, Y. Sagawa, J.F. Seignol,
F. Toutlemonde, Correlation between laboratory expansion and field expansion of
concrete: prediction based on modified concrete expansion test, ICAAR 2016-15th
international conference on alkali aggregate reaction (2016).
[28] R. Gorga, L. Sanchez, B. Martín-Pérez, FE approach to perform the condition
assessment of a concrete overpass damaged by ASR after 50 years in service, Eng.
Struct. 177 (2018) 133–146.
[29] M.T. De Grazia, N. Goshayeshi, R. Gorga, L.F.M. Sanchez, A.C. Santos, D.J. Souza,
Comprehensive semi-empirical approach to describe alkali aggregate reaction
(AAR) induced expansion in the laboratory, J. Build. Eng. 40 (2021), 102298.
[30] H. Kagimoto, M. Kawamura, Measurements of strain and humidity within massive
concrete cylinders related to the formation of ASR surface cracks, Cement Concr.
Res. 41 (8) (2011) 808–816.
[31] J. Lindgard, E. Rodum, B. Pedersen, Alkali-silica reactions in concrete-Relationship
between water content and observed damage on structures, Spec. Publ. 234 (2006)
147–166.
[32] M.A. Hariri-Ardebili, V.E. Saouma, Sensitivity and uncertainty analysis of AAR
affected reinforced concrete shear walls, Eng. Struct. 172 (2018) 334–345.
T.N. Nguyen et al.
[33] R. Gorga, L. Sanchez, B. Martín-Pérez, P. Fecteau, A. Cavalcanti, P. Silva, Finite
element assessment of the ASR-affected Paulo Afonso IV dam, J. Perform. Constr.
Facil. 34 (4) (2020), 04020065.
[34] V. Saouma, L. Perotti, Constitutive model for alkali-aggregate reactions, ACI Mater.
J. 103 (3) (2006) 194.
[35] R.A. Deschenes Jr., E.R. Giannini, T. Drimalas, B. Fournier, W.M. Hale, Effects of
moisture, temperature, and freezing and thawing on alkali-silica reaction, ACI
Mater. J. 115 (4) (2018) 575–584.
[36] D. Stark, The moisture condition of field concrete exhibiting alkali-silica reactivity,
Spec. Publ. 126 (1991) 973–988.
[37] B. Fournier, M.-A. Bérubé, Alkali-aggregate reaction in concrete: a review of basic
concepts and engineering implications, Can. J. Civ. Eng. 27 (2) (2000) 167–191.
[38] U. Costa, T. Mangialardi, A.E. Paolini, Minimizing alkali leaching in the concrete
prism expansion test at 38◦ C, Construct. Build. Mater. 146 (2017) 547–554.
[39] R. Courtier, The assessment of ASR-affected structures, Cement Concr. Compos. 12
(3) (1990) 191–201.
[40] J. Lindgård, P.J. Nixon, I. Borchers, B. Schouenborg, B.J. Wigum, M. Haugen,
U. Åkesson, The EU “PARTNER” Project — European standard tests to prevent
17
Cement and Concrete Composites 134 (2022) 104817
alkali reactions in aggregates: final results and recommendations, Cement Concr.
Res. 40 (4) (2010) 611–635.
[41] X. Brunetaud, L. Divet, D. Damidot, Delayed ettringite formation: suggestion of a
global mechanism in order to link previous hypotheses, Spec. Publ. 222 (2004)
63–76.
[42] C. Drolet, J. Duchesne, B. Fournier, Validation of the alkali contribution by
aggregates to the concrete pore solution, Cement Concr. Res. 98 (2017) 10–23.
[43] F.B. Sanchez Lf, D. Mitchell, J. Bastien, Condition assessment of an ASR-affected
overpass after nearly 50 years in service, Construct. Build. Mater. 236 (2020),
117554.
[44] L. Sanchez, B. Fournier, M. Jolin, J. Bastien, D. Mitchell, Practical use of the
Stiffness Damage Test (SDT) for assessing damage in concrete infrastructure
affected by alkali-silica reaction, Construct. Build. Mater. 125 (2016) 1178–1188.
[45] L. Sanchez, B. Fournier, M. Jolin, J. Duchesne, Reliable quantification of AAR
damage through assessment of the damage rating Index (DRI), Cement Concr. Res.
67 (2015) 74–92.