Professional Documents
Culture Documents
Nguyen Et Al. 2022 - Correlating ASR Induced Exp From Short-Term Lab Testings To Long-Term Field Performance
Nguyen Et Al. 2022 - Correlating ASR Induced Exp From Short-Term Lab Testings To Long-Term Field Performance
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 techniques have been developed over the last decades; although
improvement of existing techniques is always desired, one may say that
Alkali-silica reaction (ASR) is one of the most harmful distress the current procedures used to diagnose (i.e., cause and extent)
mechanisms affecting concrete infrastructure worldwide. ASR is a ASR-induced deterioration are quite reliable; otherwise, there is still
chemical reaction between the alkali hydroxides from the concrete pore very limited knowledge on efficient techniques able to forecast
solution (i.e., Na+, K+, and OH− ) and some unstable mineral phases ASR-induced development of the affected structures/members over
from the aggregates (i.e., fine and coarse) used to make concrete. It time.
generates a secondary product, the so-called alkali-silica gel, that swells To accomplish this requirement, several mathematical models (i.e.,
upon moisture uptake, leading to induced expansion, significant crack empirical, semi-empirical, analytical, and numerical) were developed to
formation and losses in mechanical properties and performance (i.e., provide insightful information on the potential of future deterioration of
durability and serviceability) of affected concrete components [1–3]. ASR-affected concrete. However, most of these models have not been
The best solution for ASR is preventing its occurrence in new con widely used in practice due to either limited applications to laboratory
struction, yet prevention is no longer feasible for current ASR-affected specimens and conditions [4–7], or computational complications
infrastructure. These affected structures require comprehensive diag [8–11]. Hence, developing more practical yet effective empiri
nosis and prognosis protocols for evaluating the current damage degree cal/analytical models to describe ASR-induced expansion and deterio
and forecasting the potential of further deterioration. Such information ration in concrete members exposed to field conditions is still required.
is essential for the efficient selection of rehabilitation strategies for This work proposes a novel semi-empirical approach, based on labora
affected structures. In this context, numerous diagnosis and prognosis 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
2. Background
2
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
Likewise, ongoing research has been focusing on developing alternative parameters, as shown in Eqs. (2)–(3) after Ulm et al. [25].
tests and improving the current test methods to limit or eliminate alkali [ ( )]
1 1
leaching from concrete, such as the miniature concrete prism test τc (θ) = τc (θ0 ) UC − (2)
θ θ0
(MCPT) [23], the concrete cylinder test (CCT) [24] and the RILEM
AAR-10 concrete prism test [25]. On the other hand, with the current [ ( )]
1 1
understanding of the laboratory-field correlations and relatively large τL (θ) = τL (θ0 ) UL − (3)
θ θ0
database available, mathematical modelling based on laboratory ob
servations may be developed, enabling a more reliable assessment of the where, θ is the absolute temperature (θ K = 273 + T ◦ C); θ0 is reference
potential of further ASR-induced development in affected structures absolute temperature, which is typically referred to exposure tempera
along with its likely structural consequences. A number of empirical, ture in laboratory testing; UC and UL are thermal activation constants,
analytical, and numerical models have been proposed to provide pre derived from the Arrhenius physical chemistry concept [4,26]:
dictive information on ASR-induced expansion of affected structures in
the field, such as thermo-chemo-mechanical models [8,26], finite UC = 5400 ± 500K (4)
element inverse analyses [9], multi-scale chemical analyses [7],
UL = 9400 ± 500K (5)
analytical models [27], etc. Yet, to the authors’ best knowledge, the
applicability of the vast majority of the above-proposed approaches is In addition to the temperature dependency of the time constants,
still limited in practice due to either their complexity or even the lack of Saouma and Perotti [8] further developed the model by introducing the
validation with experimental data from various concrete mixtures and effect of relative humidity and stress state (i.e., for structural modelling)
reactive aggregates. Moreover, very few of those models account for the to the expansion behaviour of concrete structures. The authors adopted
effects of alkali leaching in the laboratory specimens as well as the the relative humidity-dependent coefficient of the ASR swelling proposed
laboratory-field differences in exposure conditions. by Capra and Bournazel [5] into the Larive’s model, which directly re
duces the ultimate expansion ε∞:
2.2. Larive’s semi-empirical model for ASR-induced expansion
f(h) = RH8 (6)
A number of empirical and analytical models have been developed to where, RH is the internal relative humidity or relative humidity in the
estimate ASR kinetics and induced expansion in the laboratory [4,5], material pores. This coefficient is normally set to be zero when the
which combined with numerical analyses, were intended to predict the relative humidity within the material is lower than 80% [5,8,30]. This
structural implications (i.e., deformation, stability) of structures, struc updated form of Larive’s model with both temperature and humidity
tural members [8,10,26]. Among these models, the semi-empirical considerations has been thereby utilised in several numerical studies for
model proposed by Larive [4] is probably one of the most accepted ASR-induced expansion and structural implications [8,10,11]. Yet, it is
and widely used by ASR community [8,26–29]. The model for the important to notice that the current form of the relative humidity de
stress-free expansion evolution was developed based on an extensive pendency is highly sensitive, significantly impacting on ASR-induced
experimental database of more than 600 concrete specimens incorpo expansion. For instance, if the internal RH drops from 100% as nor
rating various reactive aggregates and exposed to different laboratory mally expected in the laboratory testing to 95%, ASR-induced expansion
conditions. In this model, ASR kinetics and induced expansion were drops to almost 45%. In this regard, although various sophisticated ASR
constructed as a function of three main model parameters, which are numerical models adopted this humidity-dependent coefficient to ac
ultimate expansion (ε∞), latency (τL), and characteristic (τC) times [4], count for the humidity effect [8,11], relative humidity values as high as
given by 100% have been frequently used in practice, which yields a relative
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) In another study, Comi, Kirchmayr and Pignatelli [10] proposed an
1 + e− τc (θ)
analytical approach to account for the effects of concrete saturation
Typical S-curves formed by the model are shown in Fig. 2. There is no degree on ASR-induced expansion through new parameters coupled to
clear physical meaning for the latency and characteristic times; how Larive’s conventional model (i.e., τC, τL and ε∞). The model was well
ever, evaluating the S-curve one observes that the latency time repre calibrated using experimental data from Larive [4] and yielded efficient
sents the initiation period of ASR-induced expansion while the results in chemo-mechanical or chemo-thermo-damage models to eval
characteristic time defines the expansion rate after the initiation period. uate ASR-induced expansion and deformation in structures [10]. It is
The time constants τL and τC were derived as temperature-dependent 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
Fig. 2. ASR-induced expansion curves obtained from Larive’s model with laboratory and in the field.
considering temperature and relative humidity effects after Saouma and Per
otti [8].
3
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
The ultimate goal of this study is to provide a robust and engineering- The reference model parameters (ε∞,o, τoL and τoC) vary depending on
friendly tool to estimate ASR-induced expansion of unrestrained field the aggregate’s type (fine/coarse) and nature (lithotype). De Grazia,
concrete members based on measurements from accelerated laboratory Goshayeshi, Gorga, Sanchez, Santos and Souza [29] proposed eight
tests such as the concrete prism test (CPT). To accomplish this goal, a different sets of these model parameters for various aggregate’s types
novel semi-empirical model was developed by adopting Larive’s model and reactivity levels. The proposed model parameters that showed to be
to utilise one-year laboratory expansion test results, with additional promising to appraise AAR-induced expansion in the laboratory have
considerations to account for the most important factors affecting ASR- also been utilised for the modelling of expansion and deformation of
induced expansion, such as concrete alkali content, leaching and Robert-Bourassa Charest overpass in Quebec City, Canada [28] and Paulo
releasing of alkalis, temperature, and RH. “Aggregate-related” parame Afonso IV dam in Paulo Afonso, Brazil [33]. Nevertheless, due to the
ters are obtained from the one-year laboratory expansion test results (i. often observed and challenging field conditions (i.e., important varia
e., CPT). The impact of alkali content and alkali leaching was thoroughly tions of temperature, relative humidity, wetting/drying cycles, etc)
evaluated and considered to correlate short and long-term expansions of along with important reactivity variations from one aggregate source to
concrete in the laboratory and in the field. The model was validated another, there are still many challenges to adopt a set of model pa
using a comprehensive laboratory test database available in the litera rameters based on the behaviour of “similar” aggregates gathered in the
ture. The proposed model was then utilised to estimate ASR-induced laboratory. Therefore, the best source of data should be from the same
expansion of concrete members and blocks manufactured over three aggregate used in the field structure under analysis. In this study, the
distinct experimental campaigns: Ontario Ministry of Transportation reference model parameters are obtained through a curve-fitting of
(MTO) project at Kingston (Ontario, Canada) [15], CANMET project Larive’s model using the expansion data measured via the concrete
(Ottawa, Ontario, Canada) [16,17], and the University of Texas project prism test at 38oC and 100% RH.
(Texas, Austin, USA) [14]. Finally, further analyses of the effects of the
key factors (i.e., concrete alkali content, alkali leaching, alkali contri 4.2. Consideration of environmental conditions: temperature and relative
bution from aggregates, and environmental conditions) on ASR-induced humidity
expansion of affected concrete members in the field were performed.
4.2.1. Temperature
4. Model development and calibration The dependency of the time constants τC and τL on temperature is
adopted from Ulm et al. (2000) [26]. These dependent coefficients were
As discussed above, Larive’s model has been widely accepted and developed based on physicochemical parameters and were thoroughly
used to represent ASR-induced expansion. The semi-empirical model calibrated using a significant experimental database from Larive (1997):
proposed in this study is based on Larive’s model and incorporates [ ( )]
1 1
additional factors into the three model parameters, ε∞ , τc , and τL . The kC,T = exp UC − (10)
θ θ0
dependency of the above model parameters to temperature and relative
humidity is adopted from Larive (1997), Ulm et al. (2000) and Hariri- [ ( )]
1 1
Ardebili and Saouma (2018) [4,26,32] as presented in Section 2.2; kL,T = exp UL − (11)
θ θ0
they could relatively account for the effects of environmental condi
tions, i.e., differences between accelerated laboratory and field exposure where, θ0 is reference absolute temperature (θ K = 273 + T ◦ C, T0 =
conditions or between different field conditions. In this study, the alkali 380C); UC and UL are thermal activation constants. This temperature
content and alkali leaching dependencies were proposed to enable dependence model of the τC and τL has been widely accepted and suc
modelling ASR-induced expansion of concrete in the field with different cessfully implemented in forecasting expansion of not only samples
alkali contents along with minimal/or no leaching of alkalis. These manufactured in the laboratory but also field structures such as dams
considerations are presented in detail in the next sections. and bridges, such as in Hariri-Ardebili et al. (2018), Ulm et al. (2000),
In general, the forms of the model parameters are given as hereafter: Saouma and Perotti (2006) [11,26,34]. It is acknowledged that the
ε∞ = ε∞,o ( T0 , RH0 , A0 ) × kε,RH × kε,A (7) 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
1 considered in the scope of this current research, yet it is an important
τC = τoC (T0 , RH0 , A0 ) × kC,T × (8)
kC,LA topic to investigate in future works.
4
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
4.3. Consideration of alkali content and alkali leaching 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
4.3.1. Dependency of the ultimate expansion (ε∞) on alkali content and ideal “no leaching” expansion curve is proposed in this study, which is
alkali leaching calculated using the expansion and leaching data obtained from labo
The dependency of the ultimate expansion on alkali content is rep ratory tests. Considering a concrete mixture containing a reactive
resented by the alkali content coefficient kε,A, as shown in Eq. (7) of the aggregate and an initial alkali content of Ao (kg/m3), a higher alkali
semi-empirical model. Fig. 3 illustrates an example from Fournier and loading of A, kg/m3 is required to compensate for the leaching and to
Bérubé [37], where the authors studied the one-year CPT outcomes of simulate the ideal “no leaching” scenario of the original alkali loading Ao,
concrete mixtures incorporating two reactive aggregates (i.e., Sudbury - which can be calculated as follows:
Su and Spratt - Sp) as a function of the alkali content of the mixtures. It is
Ao
evident that the higher the alkali content, the higher the one-year CPT A= (13)
1 − LA
expansion obtained. Moreover, increasing the alkali content of concrete
beyond 9 kg/m3 showed to have less impact on the final one-year or,
expansion than with lower alkali levels. In addition, the increase of Ao = A × (1 − LA) (14)
expansion was deemed to vary as a function of the aggregate nature and
reactivity. where LA (%) corresponds to the leaching amount measured at one year.
To better understand the alkali content dependency, test results from As shown in Eq. 13 and 14, the use of this increased alkali content A
various concrete mixtures tested in Sinno and Shehata [19], Fournier keeps at one year the same initial amount of alkali Ao of the tested
and Bérubé [37], Costa [38] varying in reactive aggregate types/natures concrete mixture. For instance, for concrete containing 5.25 kg/m3,
and alkali content (i.e., from 1.5 kg/m3 to 12.4 kg/m3, Na2Oe) were Na2Oe alkalis with 20% alkali leaching at one year (i.e., 80% remaining)
collected, and the alkali coefficient was calculated for comparison pur in the concrete prism test (CPT), an increased alkali content of 6.56 kg/
poses. The data is shown in Fig. 4, in which the alkali content level 5.25 m3, Na2Oe is adopted for the ideal “no leaching” expansion calculation,
kg/m3, Na2Oe is assumed to be the control alkali content (kε,A = 1.00), which secures 5.25 kg/m3 (i.e., 80% of 6.56 kg/m3) alkalis after a one
and the alkali content coefficient was calculated accordingly. It is worth year of testing. Next, the alkali content-dependent coefficients kε,A and
noting that most of these works were conducted with alkali contents kε,Ao are derived from the alkali loading of A and Ao, respectively, based
lower than 5.25 kg/m3, Na2Oe to evaluate the alkali threshold for po on the proposed relationship illustrated in Fig. 4. The fractional differ
tential reactivity assessment, as concrete alkali contents beyond 5.25%, ence between kε,A and kε,Ao is utilised to increase the induced expansion
however, are quite rare in practice. Only Fournier and Bérubé (2000) measured on “leaching” CPT samples in order to represent its ideal
[34] tested alkali contents up to 12.4 kg/m3, Na2Oe, where the results expansion curve without leaching:
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 ε∞ ∞
no leaching = εleaching ×
kε,A
(15)
with alkali contents below 2 kg/m3, Na2Oe is also limited. kε,Ao
Apparently, the higher the alkali content, the higher the coefficient
in which, ε∞
no leaching is the ultimate expansion of the “no leaching” sce
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 nario; ε∞
leaching is the ultimate expansion obtained from the best curve-
kg/m3, Na2Oe. Even though kε,A could vary depending on the reactive fitting of the expansion curve measured in the laboratory (i.e., CPT).
aggregate type and nature, there are very limited results in the literature In the case that the original concrete in the laboratory contains 5.25 kg/
to quantify this impact. In this study, kε,A is obtained by averaging these m3, Na2Oe of alkalis, the coefficient kε,Ao is equal to 1, and thus, the
available experimental data to take into account the impact of alkali formula in Eq. (15) becomes the same as in Eq. (7). Based on the cor
content on the ultimate expansion. For instance, the alkali content co relation with this ideal expansion curve, the expansion of different
efficient is 0.58 for the mixture containing 3.75 kg/m3 alkalis, which concrete samples varying in leaching amount could also be correlated.
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 4.3.2. Dependency of the time constants (τC and τL) on alkali leaching
alkalis. The alkali content coefficient increases from 0.08 at the alkali 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
Fig. 3. ASR one-year CPT expansion as a function of the concrete alkali con Lindgård et al. (2013) and Sinno and Shehata (2019). It is worth noting
tent [37]. that in these figures, the value of 0% alkali leaching is linearly projected
5
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
Fig. 4. Dependency of the ultimate expansion (ε∞) to the concrete alkali content.
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
Fig. 5. Effect of alkali leaching on the ASR kinetic (Lindgård et al., 2013). 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.
6
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
4.3.3.2. Model validation. The experimental data (i.e., expansion, Comparing ASR affected concrete in the laboratory and in the field in
leaching of alkalis) of these three series were collected and utilised for Section 2.1 shows two main reasons leading to the difference in their
the calculation of an ideal “no leaching” expansion curve for the concrete expansion. First, more leaching from smaller specimens in “leaching
mixture using the proposed model for alkali content and leaching con preferable” conditions reduces the expansion level of concrete in the
siderations, namely approach 1, approach 2 and approach 3, 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
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. Cement and Concrete Composites 134 (2022) 104817
different expansion rates of laboratory-exposed and field-exposed con attack, etc., on ASR-induced expansion are not accounted for in this
crete, in which laboratory-exposed concrete normally expands at a model.
higher rate under accelerated conditions when compared to concrete Another factor that needs to be considered for field concrete is the
exposed to environmental conditions in the field [15,22]. This is clearly potential release of alkalis from the aggregates, which could signifi
shown in Fig. 1(b), which compares the expansion of concrete con cantly increase the amount of alkalis in the concrete pore-solution in the
taining Spratt limestone tested in the laboratory and in the field. long term. According to Bérubé, Duchesne, Dorion and Rivest [20],
In order to correlate the induced expansion of concrete in the labo reactive aggregates could contribute alkalis in amounts as high as 2.5
ratory to that of field concrete, these two main differences have to be kg/m3, Na2Oe in long-term concrete exposure. This important contri
taken into consideration. In this regard, a two-step procedure is pro bution could significantly increase the induced expansion of field con
posed, as shown in Fig. 8. In general, the first step is to account for the crete in comparison to the short-term exposed concrete in the
leaching of laboratory concrete, while the second step is to take into laboratory; however, there is currently no experimental data available
consideration the differences in exposure conditions between laboratory for this time-dependent alkali release in field concrete. Therefore, the
and field, along with possible changes in field conditions over time. In contribution of alkalis from aggregates (RA, kg/m3) is considered in this
the first step, an ideal expansion curve for the “no leaching” scenario is study as an additional amount to the total concrete alkali content. As
estimated using the procedure presented earlier in Section 4.3 by taking such, after calculating the ideal “no leaching” expansion curve for the
into account the alkali leaching. As previously discussed, the leaching of concrete mixture with the original alkali content of Ao (kg/m3), the al
alkalis from concrete samples/members mainly depends on its size and kali content-dependent coefficients of Ao and Ao + RA alkalis are
the exposure conditions. In this regard, minimal or no leaching is adopted to update the ultimate expansion (i.e., considering the alkali
commonly observed for the field members (i.e., large in size and not in contribution from the aggregates). It is important to note herein that a
the “leaching preferable” environmental conditions). The ideal expansion test procedure for long-term alkali release determination is still not
curve is essential to simulate the possible expansion level of the concrete available, thus the amount of releasing is commonly estimated from
members in the field assuming no leaching occurs. The inputs obtained short-term testing of aggregates.
from the laboratory tests include expansion over time, initial concrete After finalising the ideal expansion curve at the reference laboratory
alkali content and alkali leaching at one year. The implementation exposure conditions (i.e., 38 ◦ C and 100% RH), the effect of environ
procedure of this step is presented in detail in Section 4.3. It is worth mental conditions in the field is considered through temperature-
noting that the ideal expansion curve is simulated at reference labora dependent and RH-dependent coefficients presented in Section 4.2.
tory exposure conditions (i.e., 38 ◦ C and 100% RH); therefore, envi Together with the assumptions made above, Fig. 9 illustrates the
ronmental conditions need to be considered in step 2 to model the time- calculation procedure to capture the effect of the field environmental
dependent expansion curve of field concrete. It is also important to note conditions, which is adopted from Kawabata et al. [27]. As the field
that modelling of field concrete considering long-term leaching of alkalis conditions change over time, consequent changes of temperature and
is implementable using the same procedure if leaching data is available. RH in concrete and their effects on ASR-induced expansion have to be
In field concrete members, ASR-induced development is by far more continuously considered. Each set of temperature and RH produces an
complicated than laboratory specimens subjected to accelerated condi expansion curve of the concrete for a particular environmental condi
tions due to many factors, such as continuous changes in the ambient tion, namely a master curve. To capture the change of temperature and
temperature and RH, effect of rainfall, solar radiation and other damage RH over time, expansion calculation is implemented through the
mechanisms (i.e., shrinkage, creep, freezing and thawing) [28,35,39]. determination of the incremental expansion of every single time step Δt.
For the sake of simplicity, some important assumptions need to be made The incremental expansion at the time ti is calculated based on the
in modelling expansion of field concrete, such as: (1) there is no alkali master curve of the corresponding environmental conditions (i.e., Ti,
leaching from the field concrete members, (2) internal temperature is RHi). Therefore, it is noticed herein that the shorter the time step Δt, the
assumed to be similar to the ambient temperature, and (3) internal RH is more accurate the field conditions changes are captured. In this study,
assumed to be uniformly distributed throughout the concrete specimen. These the monthly average ambient temperature was utilised for the expansion
assumptions have been reasonably used in many other studies, such as in calculation of the concrete members in the field; thus, the time step of 1
Kawabata, Yamada, Ogawa, Martin, Sagawa, Seignol and Toutlemonde month was adopted in this work.
[27] and Gorga, Sanchez and Martín-Pérez [28]. In addition, impacts of The validation of the proposed semi-empirical model has then been
other mechanisms such as shrinkage, freezing and thawing, sulphate conducted using three important experimental campaigns: (1) Kingston
8
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
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].
9
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
applied uniformly in the specimen, remaining the same throughout its exposure. This difference is considered in the proposed model by
exposure duration for the simplification of the expansion calculations. It considering the alkali leaching from the test specimens in the CPT.
is worth noting that the precipitation rate is high in both Ottawa and Furthermore, the concrete beams exposed to lower average (yearly)
Austin. The sensitivity of RH to the field exposed blocks expansion is temperatures in the field (i.e., the maximum temperature of about 27 ◦ C
then considered in a parametric study. – Ottawa data from Fig. 10) show significantly lower but somewhat
In this study, the leaching of alkalis at 1 year of concrete with Spratt more sustained expansion rates over the full testing period when
and Sudbury are assumed to be 21.9% and 35.1% of the total concrete compared to prisms at 38 ◦ C. The temperature-dependent model is
alkali content, respectively, which are estimated at about 85% of the adopted to account for this environmental condition difference.
leaching amount measured 1.5-year in Sinno and Shehata (2019) [19] As stated previously, inputs for the proposed semi-empirical model
from the CPT. In addition, the alkali release from the reactive aggregates are the initial concrete alkali content, the expansion over time and
may also be important in the estimation of the field expansion. As such, leaching of alkalis from the CPT test. The total Na2O equivalent of the
Bérubé, Duchesne, Dorion and Rivest [20] tested the alkali release of the LAPC and HAPC concrete mixtures were 1.92 kg/m3 and 3.33 kg/m3,
Spratt and Sudbury aggregates through crushing (i.e., 1.25–5 mm par Na2Oe (i.e., cements of 0.46% and 0.79% Na2Oe), respectively, which
ticles in size) and immersion of the aggregates particles in different so were all from the cement. Similar to the expansion model of concrete in
lutions (i.e., distilled water, saturated lime and 0.7 N NaOH (or KOH) CANMET and UT exposure sites presented in the previous section, the
solutions) for 578 days at 38 ◦ C; the 0.7 M alkaline solution was the most leaching of alkalis at 1 year of concrete with Spratt is assumed to be
similar to the concrete pore solution bearing a high alkali Portland 21.9% of the total concrete alkali content. The calculation of model
cement. The amount of alkali release in 0.7 M alkaline solution after 578 parameters for the ideal “no leaching” expansion curve of HAPC concrete
days of the Sudbury aggregate was more than 0.166% per aggregate is shown in Table 2. The ideal expansion curve of LAPC is derived from
mass compared to only 0.009% for the Spratt aggregate. This result is the HAPC’s using the proposed relationship in Fig. 4 based on their
reasonable since the total Na2Oe content in the Sudbury and Spratt corresponding alkali contents. Since the internal RH was assumed to be
aggregates measured in the same study was 4.35% and only 0.09% per 100%, only the monthly average temperatures (see Fig. 10) were used to
aggregate mass, respectively. consider the environmental conditions in Kingston, Ontario.
Moreover, the authors stated that the amount of alkalis released in
the long term from concrete incorporating the above coarse aggregate 6. Results
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 For both concrete mixtures HAPC and LAPC, the modelling for field
significant amount of alkalis possibly contributing to ASR-induced beams were carried out with and without considering the alkali leaching
development from the Sudbury aggregate, the expansion calculation from the CPT test for comparison purposes. Fig. 12 shows the predicted
of the Sudbury blocks without and with the alkali contribution of results in comparison to the measured data. As one observes, without
0.166% per aggregate mass (i.e., 1.75 kg/m3, Na2Oe of concrete) were considering the alkali leaching (i.e., directly using the experimental
conducted for comparison purposes. Conversely, the alkalis released results from the CPT), the estimated expansions are significantly smaller
from the Spratt aggregate is assumed negligible. 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
5.3. Modelling expansion of Kingston non-reinforced concrete beams [13–16]. By taking the leaching of the CPT into consideration, the
modelling provided excellent agreement with the measured data. For
Fig. 11 shows expansion data of 27-year field-exposed beams the HAPC (Fig. 12(a)), the ultimate expansion of the modelled expansion
incorporating low and high-alkali cement (LAPC and HAPC), alongside curve is slightly higher than the maximum expansion measured from the
CPT expansion curve of the HAPC concrete. Comparative field/labora blocks. Conversely, the calculated ultimate expansion of the LAPC
tory test results have shown that the ultimate expansion obtained from (Fig. 12(b)) is marginally smaller than the measured data. However, the
the CPT is significantly lower than that of the field beam at 27-year 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
10
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
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
Fig. 12. Model outcomes in comparison to the experimental data: (a) HAPC from model parameters obtained from the measurements, as well as a
and (b) LAPC non-reinforced beams incorporating the reactive Spratt limestone,
need of experimental constants to the model which proposed in Bru
MTO outdoor exposure site [22].
netaud et al. (2004) [41]. More investigations are required in this regard
for an improvement of the model.
keep active the expansion of the HAPC and LAPC. This contribution, For the Sudbury blocks, despite improving the expansion estimation
however, is not considered in the current calculation due to limited data by considering the leaching of alkalis in the laboratory, the predicted
from the test program. expansion is still remarkably lower than the field measurements, i.e.,
In general, the proposed model, which considers the effects of the 0.408% from the model compared to 0.651% from the measurement at
concrete alkali content and alkali leaching, significantly improves the 20 years of exposure (see Fig. 13(c)). Conversely, the model with
correlation between laboratory testing and long-term field performance additional consideration of alkali release, as shown in Fig. 13(d), pro
of the ASR-affected concretes studied. The result also suggests the need vides a significantly better expansion estimation when compared to the
of considering the effect of alkali leaching of laboratory samples in the previous one. By raising the total alkalis from 5.25 kg/m3 to 7.0 kg/m3,
estimation of field performance. However, it also highlights the impor Na2Oe due to alkali contribution from the aggregate, the ultimate
tance of having the data of additional alkalis contributions from ag expansion increases by about 70% compared to the model outcome
gregates or other sources than the cement in forecasting the long term ignoring the releasing effect. This result highlights the significant impact
“sustained” expansion rates of field specimens. of alkali contribution from the aggregates on the induced expansion of
long-term exposed concrete in the field. It also emphasises the validity of
6.1. Modelling expansion of CANMET non-reinforced concrete blocks the alkali release results reported in Bérubé et al. (2002) [20], which are
negligible for the Spratt aggregate and significant for the Sudbury
Test data of concrete incorporating Spratt and Sudbury reactive ag aggregate.
gregates and various concrete alkali contents investigated in CANMET Fig. 14 illustrates the model outcomes for the mixtures made of
and UT sites were collected for validation purposes. Three typical Spratt and Sudbury aggregates varying in concrete alkali level compared
amounts of Na2Oe were used in this experimental campaign: 0.4%, 0.9% to the measured data. For the Spratt concrete blocks, the model captures
and 1.25% (boosted with additional reagent grade NaOH) per cement very well the expansion of concrete blocks with 0.9% and 1.25% alkali
mass, equivalent to approximately 1.7, 3.8 and 5.25 kg/m3, Na2Oe, content (i.e., 3.8 and 5.25 kg/m3, Na2Oe, respectively). At low alkali
respectively. Table 3 presents the calculation of model parameters for content (i.e., 0.4% of alkali, or 1.68 kg/m3, Na2Oe), however, the model
the ideal “no leaching” expansion curve for mixtures made of Spratt and underestimates the expansion of the concrete block. For the Sudbury
Sudbury aggregates bearing 5.25 kg/m3, Na2Oe. The Sudbury concrete concrete blocks, by accounting for the alkali contribution by the
shows lower ultimate expansion from the measurements but higher
11
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
Table 4
Calculation of the ultimate expansion of mixtures varying in concrete alkali content.
Spratt Sudbury
3
A (kg/m ) 5.25 3.78 1.68 5.25 3.78 5.25 + 1.75 (RA)(*) 3.8 + 1.75 (RA) (*)
kε,A 1 0.588 0.108 1 0.588 1.708 1.113
ε∞ (%) 0.399 0.235 0.043 0.430 0.253 0.735 0.479
(*)
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 aggregates at CANMET were utilised in addition to environmental
made of cement with 0.9% of alkalis (i.e., 3.78 kg/m3, Na2Oe). This conditions data in Texas to determine the expansion of the corre
result again shows the reasonability of considering the alkali contribu sponding concrete blocks stored at UT using the proposed procedure in
tion by the aggregates. The model outcomes for both Spratt and Sudbury Section 5.1. Fig. 15 shows the model outcomes of concrete blocks made
concrete blocks also confirms that the alkali content – expansion rela of Spratt and Sudbury at CANMET and UT sites compared to the field
tionship proposed in Section 4.3 is suitable for modelling the induced observations. The model outcomes are relatively close to the block
expansion of concrete mixture varying in alkali contents, yet, the measurements made and kept at UT for both mixtures made of Spratt
improvement of this relationship is necessary for concrete with low al and Sudbury reactive aggregates. Similar to the experimental observa
kali content levels, (i.e., <2 kg/m3, Na2Oe) by obtaining more experi tions of [17], the ASR-induced expansion calculated for blocks kept at
mental data within these levels. 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
6.2. Modelling expansion of UT non-reinforced blocks to transportation of the blocks in the first year. In general, the proposed
model provides a reasonably good estimation of ASR-induced expansion
This section aims to further validate the proposed model at different from the two sites with different environmental conditions by using only
environmental conditions, i.e., warmer climatic conditions in Texas (UT a single set of laboratory testing data.
site), and cooler climatic conditions in Ottawa (CANMET site). The cli
matic conditions in Texas, as shown in Fig. 10, indicate almost the same 7. Discussion
high level of precipitation but significantly higher temperature when
compared to Ontario’s climatic conditions. The field expansion of con 7.1. Consideration of alkali leaching from test samples and alkali
crete blocks in the UT site was thus about 4–5 times faster than the releasing from aggregate: their importance and limitation
blocks with the same concrete mixtures in CANMET [17].
The model parameters established earlier for the ideal “no leaching” It is generally considered that the concrete prism test (CPT) is the
expansion curves of the mixtures made of Spratt and Sudbury reactive
12
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
13
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
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 used to forecast the potential future expansion of the affected structure
temperature and RH. Minimum and maximum temperatures (given in throughout its service life. For instance, Fig. 18 illustrates the model
Fig. 10 as low and high temperatures, respectively) were adopted in the outcomes for Sudbury concrete blocks from UT and CANMET exposure
semi-empirical model for such assessment. The variation of temperature sites as a clear indication of the use of the proposed model for ASR
herein is about ± 5 ◦ C of the average temperature which was used in the prognosis.
previous calculations. Moreover, a lower RH of 95% is adopted in the Assuming that the current expansion level could be estimated using
model with a simplification of the same RH throughout the exposure diagnosis techniques such as DRI, SDT and CI, the concrete blocks at UT,
period. It is worth noting that the assumption of uniform temperature after a 3-year exposure, could potentially keep swelling up to 0.55%
and RH throughout the concrete member is still maintained. within the next 5 years before levelling off. Otherwise, concrete blocks
Fig. 16 illustrates the field expansion calculated at the minimum, stored at CANMET over 20 years would potentially further swell “only”
maximum and average temperatures. It is obvious that the temperature about 0.05% within the same time period. This approach, with addi
significantly influences the rate of ASR-induced expansion. For example, tional considerations on confinement and restraint conditions (i.e., via
for both Spratt and Sudbury concrete blocks, the expansion rate could be numerical assessment), could be extremely helpful for the prognosis of
two times lower if the temperature reduces, for instance, by 5 ◦ C. The ASR-affected concrete structures in the field along with determining
field expansion at the minimum and maximum temperatures thus efficient rehabilitation method(s) and management strategies for
completely cover all the experimental data points for both mixtures. This affected infrastructure.
important sensitivity, however, requires a more comprehensive analysis
of internal temperature in the long-term exposed concrete members to 8. Concluding remarks
provide a better estimation of ASR-induced expansion.
The model results of the field expansion for the internal RH of 95% A semi-empirical model was proposed to correlate the laboratory
are shown in Fig. 17. One can verify a strong effect of the RH on ASR- testing to field concrete expansion, which is essential for the prognosis of
induced ultimate expansion. For example, on Spratt blocks, the ulti ASR affected concrete structures. In addition to the laboratory expansion
mate expansion reduces from 0.391% to only 0.260% when reducing the data, the model took into consideration key factors affecting the
internal RH from 100% to 95%. It is interesting to notice that the field expansion behaviour both from laboratory testing (i.e., concrete alkali
expansion calculated for both Spratt and Sudbury blocks at 100% and content, alkali leaching from laboratory samples) and from the field (i.
95% RH covers very well the measured data. This emphasises the pos e., the temperature and relative humidity of field concrete and alkali
sibility of the variation in the relative humidity during the long-term contribution from the aggregates) in the form of Larive’s model. The
exposure of field concrete. proposed model was able to reproduce the expansion behaviour of
concrete blocks exposed to different field conditions, such as Kingston
7.3. The use of the proposed model for estimating the potential of further site (Ontario, Canada), CANMET site in Ottawa (Ontario, Canada) and
ASR-induced expansion UT site in Texas (Austin, USA). Based on the model development and
results, the following conclusions can be drawn:
In general, the proposed model provides a reasonably good estima
tion of ASR-induced expansion from the two sites with different envi • The leaching of alkalis is significant in laboratory expansion tests on
ronmental conditions using only a single set of laboratory testing data. small samples, affecting both the ASR kinetics and the ultimate
With this capability of the model, a single set of CPT test data can be expansion of the concrete. In contrast, it is commonly considered
used to simulate the free expansion of concrete at different locations minimal in the field concrete blocks/members. By using the expan
varying in environmental conditions. In addition, another significant sion curve and leaching at 1 year measured from laboratory testings,
advantage of the proposed model is for estimating the potential of the proposed model is capable of estimating an ideal expansion curve
further ASR-induced expansion. Nowadays, several state-of-the-art without leaching to reproduce the expansion of field concrete. The
diagnosis techniques have been proposed to access and quantify the model was well-calibrated using the different reliable laboratory
current expansion/or damage level of ASR-affected concrete, such as testing results. Compared to the prediction without considering the
Damage Rating Index (DRI), Stiffness Damage Test (SDT) or Cracking leaching, using the proposed model shows significant improvement
Index (CI) [2,43–45]. However, there are still limited numbers of in forecasting expansion of non-reinforced concrete beam exposed
effective techniques for the prognosis of the ASR-induced development for 27 years at the Kingston site and concrete blocks kept at CANMET
to forecast how ASR will progress in the coming months or years. After for 20 years. Especially, the model results of Spratt concrete are
having reliable quantification of the current damage level using the closely matched the field observations from both Kingston and
abovementioned diagnosis techniques, the proposed model could be CANMET specimens. In this regard, measurement of the alkali
14
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
leaching is suggested to be very important to increase the reliability • The effect of environmental conditions (i.e., ambient temperature
of the CPT in appraising the maximum potential expansion and and relative humidity, precipitation) in this study is considered
damage degree of the field concrete. indirectly through concrete temperature and relative humidity. The
• For the long-term exposed concrete, the total alkali content of the model was able to reproduce the expansion of both Spratt and Sud
concrete is contributed not only from the cement but also possibly bury concrete blocks at CANMET and UT sites (i.e., the same mixture
from the coarse and fine aggregates. A remarkable amount of alkalis composition) using the same set of laboratory test results and their
released from Sudbury reactive aggregate measured in Ref. [20] was corresponding environmental conditions. This is crucially important
adopted to raise the total alkali content of this concrete mixture for in forecasting and correlating the expansion of concrete members
the long-term expansion estimation. The excellent agreement be using the same reactive aggregates and mixture exposed to different
tween the model outcomes and measurement results of the 20-year environmental conditions in a country. It is important to note that
exposed concrete blocks containing Sudbury reactive aggregate at the concrete temperature highly influences the expansion rate,
CANMET site highlighted the significance of alkali contribution from whereas the relative humidity greatly impacts the ultimate expan
aggregates to the long-term expansion of concrete in the field. sion. In addition, the modelling results also show that the assumption
15
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
16
T.N. Nguyen et al. Cement and Concrete Composites 134 (2022) 104817
[33] R. Gorga, L. Sanchez, B. Martín-Pérez, P. Fecteau, A. Cavalcanti, P. Silva, Finite alkali reactions in aggregates: final results and recommendations, Cement Concr.
element assessment of the ASR-affected Paulo Afonso IV dam, J. Perform. Constr. Res. 40 (4) (2010) 611–635.
Facil. 34 (4) (2020), 04020065. [41] X. Brunetaud, L. Divet, D. Damidot, Delayed ettringite formation: suggestion of a
[34] V. Saouma, L. Perotti, Constitutive model for alkali-aggregate reactions, ACI Mater. global mechanism in order to link previous hypotheses, Spec. Publ. 222 (2004)
J. 103 (3) (2006) 194. 63–76.
[35] R.A. Deschenes Jr., E.R. Giannini, T. Drimalas, B. Fournier, W.M. Hale, Effects of [42] C. Drolet, J. Duchesne, B. Fournier, Validation of the alkali contribution by
moisture, temperature, and freezing and thawing on alkali-silica reaction, ACI aggregates to the concrete pore solution, Cement Concr. Res. 98 (2017) 10–23.
Mater. J. 115 (4) (2018) 575–584. [43] F.B. Sanchez Lf, D. Mitchell, J. Bastien, Condition assessment of an ASR-affected
[36] D. Stark, The moisture condition of field concrete exhibiting alkali-silica reactivity, overpass after nearly 50 years in service, Construct. Build. Mater. 236 (2020),
Spec. Publ. 126 (1991) 973–988. 117554.
[37] B. Fournier, M.-A. Bérubé, Alkali-aggregate reaction in concrete: a review of basic [44] L. Sanchez, B. Fournier, M. Jolin, J. Bastien, D. Mitchell, Practical use of the
concepts and engineering implications, Can. J. Civ. Eng. 27 (2) (2000) 167–191. Stiffness Damage Test (SDT) for assessing damage in concrete infrastructure
[38] U. Costa, T. Mangialardi, A.E. Paolini, Minimizing alkali leaching in the concrete affected by alkali-silica reaction, Construct. Build. Mater. 125 (2016) 1178–1188.
prism expansion test at 38◦ C, Construct. Build. Mater. 146 (2017) 547–554. [45] L. Sanchez, B. Fournier, M. Jolin, J. Duchesne, Reliable quantification of AAR
[39] R. Courtier, The assessment of ASR-affected structures, Cement Concr. Compos. 12 damage through assessment of the damage rating Index (DRI), Cement Concr. Res.
(3) (1990) 191–201. 67 (2015) 74–92.
[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