ML17307A317

La - (External_Sender) FW: Seabrook

ML17307A317
Person / Time
Site:	Seabrook
Issue date:	10/24/2017
From:	- No Known Affiliation
To:	Division of Operating Reactor Licensing
References
17-953-02-LA-BD01
Download: ML17307A317 (19)

Text

SeabrookLANPEm Resource From: Morante, Richard J <morante@bnl.gov>

Sent: Tuesday, October 24, 2017 3:28 PM To: Lehman, Bryce

Subject:

[External_Sender] FW: Seabrook Attachments: Alkali-Silica Reaction 5571e6f008ae7536374c6022.pdf From: Morante, Richard J Sent: Monday, July 18, 2016 11:47 AM To: Braverman, Joseph I

Subject:

Seabrook

Joe, For your information.

Rich 1

Hearing Identifier: Seabrook_LA_NonPublic Email Number: 162 Mail Envelope Properties (4909515901062646BE47A0C16E45C8673755A297)

Subject:

[External_Sender] FW: Seabrook Sent Date: 10/24/2017 3:27:35 PM Received Date: 10/24/2017 3:28:12 PM From: Morante, Richard J Created By: morante@bnl.gov Recipients:

"Lehman, Bryce" <Bryce.Lehman@nrc.gov>

Tracking Status: None Post Office: EXMB3.bnl.gov Files Size Date & Time MESSAGE 175 10/24/2017 3:28:12 PM Alkali-Silica Reaction 5571e6f008ae7536374c6022.pdf 2107918 Options Priority: Standard Return Notification: No Reply Requested: No Sensitivity: Normal Expiration Date:

Recipients Received:

Cement and Concrete Research 76 (2015) 130-146 Contents lists available at ScienceDirect Cement and Concrete Research journal homepage: http://ees.elsevier.com/CEMCON/default.asp Alkali-silica reaction: Current understanding of the reaction mechanisms and the knowledge gaps Farshad Rajabipour a,, Eric Giannini b, Cyrille Dunant c, Jason H. Ideker d, Michael D.A. Thomas e a

Dept. of Civil and Environmental Engineering, Pennsylvania State University, University Park, PA 16802, USA b

Dept. of Civil, Construction, and Environmental Engineering, The University of Alabama, Tuscaloosa, AL 35487, USA c

Laboratory of Construction Materials, École Polytechnique Fédérale de Lausanne, EPFL STI IMX LMC, CH-1015 Lausanne, Switzerland d

School of Civil and Construction Engineering, Oregon State University, Corvallis, OR 97331, USA e

Department of Civil Engineering, University of New Brunswick, Fredericton, NB E3B 5A3, Canada a r t i c l e i n f o a b s t r a c t Article history: Alkali-silica reaction (ASR) is a major concrete durability problem, resulting in signi"cant maintenance and Received 15 October 2014 reconstruction costs to concrete infrastructures all over the world. Despite decades of study, the underlying Accepted 19 May 2015 chemical and physical reaction mechanisms remain poorly understood, especially at molecular to micro-scale Available online xxxx levels, and this has resulted in the inability to ef"ciently assess the risk, predict the service life, and mitigate deterioration in ASR-susceptible structures. This paper intends to summarize the current state of understanding Keywords:

Alkali-aggregate reaction (C) and the existing knowledge gaps with respect to reaction mechanisms and the roles of aggregate properties Pore solution (B) (e.g., composition, mineralogy, size, and surface characteristics), pore solution composition (e.g., pH, alkalis, Modeling (E) calcium, aluminum), and exposure conditions (e.g., temperature, humidity) on the rate and magnitude of ASR.

ASR gel (-) In addition, the current state of computer modeling as an alternative or supplement to physical testing for Thermodynamic calculations (B) prediction of ASR performance is discussed.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction (b) developing rapid and reliable test methods to identify reactive minerals and aggregates [19-21], (c) predicting the long-term perfor-Alkali-silica reaction (ASR) is a deleterious reaction between the mance of concrete mixtures in service [22-24], and (d) proposing and alkaline pore solution of concrete and various metastable forms of silica evaluating effective methods to mitigate ASR in new or existing struc-contained in many natural and synthetic aggregates [1,2]. The silica tures [8,25,26].

structure is dissolved by the nucleophilic attack of OH ions, and the This paper further explores items (a) and (d) above, by summarizing highly degraded silica structure behaves as a hygroscopic silica gel. the current state of knowledge on the ASR mechanisms and mitigation, Alternatively, the dissolved silica can cross-link (e.g., in presence of Ca and presenting important questions related to the reaction mechanisms ions), coagulate, and form ASR gel. Swelling of this gel leads to stress and the roles of various species (e.g., OH, Na/K, Li, Ca, and Al) and development and potentially, cracking of concrete. ASR is a major exposure conditions (e.g., temperature, pressure, humidity) on the durability problem of concrete, and continues to damage important rate and magnitude of ASR. In addition, the current state of computer structures, including dams and hydraulic structures, pavements, modeling as an alternative or supplement to physical testing for predic-bridges, walls, barriers, and nuclear/power plant structures [3-13]. tion of ASR performance will be discussed.

Much has been learned about ASR over the last 70 years, starting from the pioneering work of Stanton [14], who correctly recognized 2. Reaction mechanisms the signi"cance of cement alkali content, availability of moisture, and the effect of pozzolans on the severity of ASR in concrete. Stanton also ASR damage is a result of a number of sequential reactions, including:

proposed a test method, which later evolved into the mortar bar test (1) dissolution of metastable silica, (2) formation of nano-colloidal (ASTM C227 [15]), to evaluate the potential reactivity of cement- silica sol, (3) gelation of the sol, and (4) swelling of the gel:

aggregate combinations. Since the 1940s, research has been primarily focused on (a) understanding the reaction mechanisms [1,16-18], 1 2 3 4 SiO2 solid SiO2 aqueous SiO2 sol SiO2 gel swelling of gel: 1 Corresponding author. Tel.: +1 814 863 0601.

E-mail addresses: farshad@psu.edu (F. Rajabipour), ergiannini@eng.ua.edu Alternatively, a highly degraded (SiO2)solid may transform directly (E. Giannini), cyrille.dunant@gmail.com (C. Dunant), jason.ideker@oregonstate.edu into (SiO2)gel, as long as suf"cient but not excessive cross-linking (J.H. Ideker), mdat@unb.ca (M.D.A. Thomas). is maintained between silica chains. Among these reactions, silica http://dx.doi.org/10.1016/j.cemconres.2015.05.024 0008-8846/© 2015 Elsevier Ltd. All rights reserved.

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 131 dissolution is often the slowest [27], and as such, primarily controls the environment, hydroxyl (OH) ions progressively attack the (Si-O-)

rate of ASR in concrete. Factors that accelerate silica dissolution bonds, resulting in network dissolution of silica:

(e.g., higher alkalinity and temperature) are often used to speed up ASR in concrete, for example in accelerated ASR tests. In addition to

Network dissolution : Si-OHs 3OH aq SiOH4 aq 2a silica dissolution, swelling of the gel depends on availability of moisture and mass transport properties of concrete, and could limit the rate of ASR damage in dry and dense concrete. where ( Si-OH) represents the silanol groups that are present at the silica-water interface [17]. The dissolving Si(OH)4 leaves three Si-O groups on the solid surface (not shown on the right side of Eq. (2a)),

2.1. Dissolution of metastable silica which charge balance the equation. In addition to the mono-silicic Si(OH)4, dissolution products can be oligomers of the form Silicates are the most abundant class of rock-forming minerals on SinOa(OH)b, where 2a + b = 4n [30]. Eq. (2a) can be alternatively writ-the earth's surface. They are primarily composed of a 3D network of tet- ten as Eq. (2b), which is easier to describe thermodynamically, but does rahedral silica (SiO2) units, with one Si atom in the center, surrounded not convey the actual dissolution mechanism at high pH:

by four oxygen atoms [28]. The tetrahedral units are connected through oxygen vertices (known as bridging oxygens), which form siloxane

Network dissolution : SiO2 s 2H2 O SiOH4 2b

( Si-O-Si) bonds. Here, suggests that each Si atom is connected aq to 3 other oxygen atoms. In addition to pure silica (SiO2), silica minerals often contain hydrogen or metallic elements (e.g., Na, K, Ca) that are The solubility product of reaction (2b) has been reported by Walther connected to oxygen atoms. The O-Si-O bond angle in silica is "xed at and Helgeson [37] for different polymorphs of (SiO2)solid and at different 109°, but the Si-O-Si bond between SiO2 tetrahedra can vary from temperatures. For example, at 25 °C, logKsp = 3.999 and 2.714 has 100 to 170° [28]. As a result, a large number of silica structures exist been reported for -quartz (crystalline and stable) and amorphous that could be macro-crystalline (Fig. 1a), micro/nano-crystalline, or silica, respectively. This data can be used to calculate the equilibrium amorphous (Fig. 1b) [29]. solubility limit of SiO2 in neutral water as presented in Table 1. It is ap-The phenomenon of silica dissolution in water has been studied for parent that the solubility increases with temperature and reduction in over a century, with several hundred articles published, e.g., [30-36]. silica crystallinity [33].

Most of the existing literature, however, focuses on pure silica and mod- Table 1 shows that the solubility limit of silica in neutral water is erate levels of alkalinity (pH b 11). It is well known that in an alkaline small. While the solubility of non-ionic (Si(OH)4)aq is not signi"cantly (a) (b)

(c)

Fig. 1. (a) 2D ball and stick representation of quartz (a common form of crystalline solid SiO2) composed of uniformly sized silica rings where all oxygens are bridging silicon atoms; (b) amorphous solid SiO2, showing non-uniform rings and the contribution of alkalis and Ca in forming non-bridging oxygens; (c) alkali-silica (ASR) gel composed of clustering of colloidal silica particles surrounded by the gel's pore solution; the silica structure is highly degraded with lots of non-bridging oxygens.

These models were constructed based on information from Refs. [28,52].

132 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 Table 1 1.E+04

{Si(OH)4}aq {H3SiO4-}aq Thermodynamic solubility limit (mM) of various SiO2 polymorphs in neutral water.

1.E+03 Calculated using data from Ref. [37].

{H2SiO4=}aq Total {Si}aq 1.E+02 Calculated solubility limit of Si [M]

T (°C) -Quartz Chalcedony -Cristobalite -Cristobalite Amorphous silica 0 0.03 0.06 0.13 0.42 1.01 1.E+01 25 0.10 0.19 0.36 0.99 1.93 pH 50 0.24 0.42 0.74 1.81 3.12 1.E+00 75 0.46 0.79 1.32 2.88 4.57 7 8 9 10 11 12 13 14 100 0.80 1.32 2.11 4.22 6.30 1.E-01 Approximate solubility 1.E-02 limit of sodium silicate 1.E-03 affected by the pH of the solution [30] (as also suggested by Eq. (2b)), at high pH, ionization of Si(OH)4 occurs, resulting in conversion to highly 1.E-04 soluble ions [35]: 1.E-05

Ionization : SiOH4 aq H3 SiO4 aq H LogKI 1/4 9:473 3a 1.E-06

 

Fig. 2. The calculated solubility limit of amorphous SiO 2 in water as a function of pH SiOH4 aq H2 SiO2 4 2H LogKII 1/4 22:12: 3b (T = 25 °C) (note that to be accurate, y-axis represents Si activity, and the actual molar aq solubility limit at a given pH is likely to be higher assuming activity coef"cient b1).

From Ref. [41].

Using the solubility products of reactions (3a) and (3b), the equilib-rium solubility of silica species in an aqueous solution can be calculated as a function of pH, as shown in Fig. 2. It is observed that at high pH, the and amorphous silica in deionized water [42,43]. For soda-lime glass apparent solubility of silica increases by several orders of magnitude. In at pH = 14, activation energy was measured as 86.2 kJ/mol [41].

addition, the dominant silica species in the solution are H3SiO 4 and H2SiO24 . Sjberg [35] reported that at high pH, in addition to mono-2.2. Formation and gelation of colloidal silica mers, di-, tri-, tetra-, and other poly-meric silicate ions are also present at lower concentrations. It is also important to note that at high pH, As long as the pH and temperature are maintained to avoid supersat-reactions (2a), (3a) and (3b) proceed in the direction to reduce the uration with respect to aqueous silica, and in the absence of Ca ions, the pH. In addition, ion exchange reactions could occur to further decrease dissolved species remain in solution [44]. At high pH, aqueous silica the pH; for example:

species are negatively charged (Eqs. (3a) and (3b) and Fig. 2), and the

resulting electrostatic repulsion prevents gelation. In such a case, aggre-Ion exchange : SiOH4 aq Na aq HO3 Si-O Na aq gate dissolution slows and eventually stops when [Si]aq approaches its H aq 4 solubility limit [45]. This, however, almost never happens in portland cement concrete, since Ca2+ (or other polyvalent metal ions; e.g., Al) where ONa+ indicates that the bond is more likely a strong Van der present in the pore solution, can link silica ions to form poly-metal-Waals type. These reactions suggest that the pH of concrete pore silicates [30]; e.g.,:

solution should generally decrease as ASR progresses. As the dissolved silica gels (gelation mechanisms are discussed below), however, some

Condensation : 2 HO3 Si-O aq of the alkalis incorporated in silica gel may be replaced by calcium

Ca2 aq HO3 Si-OCaO-SiOH3  : 6 ions, which recycled these alkalis back into concrete pore solution sol (Eq. (5a)). The resulting calcium de"ciency of the pore solution pro-motes further dissolution of solid portlandite (Eq. (5b)), which results The compound on the right hand side of Eq. (6) can be alternatively in increasing the pH: written as H6CaSi2O8 or CS2H3 in cement chemistry notation. Gaboriaud et al. [46] suggested that Ca may only serve as a catalyst and could be

Alkali recycling : 2 HO3 Si-ONa aq Ca2 2Na further ejected, leading to formation of siloxane bonds; although this

5a reaction is not thermodynamically favorable at high pH:

HO3 Si-OCaO-SiOH3 Si-OCaO-Si H2 OSi-O-Si Ca2 2OH : 7 2

CaOH2 Ca 2OH : 5b Once a nucleus of critical size is formed, it grows to nano-colloidal This is further discussed in Section 3.3. silica sol through further condensation [27]. Aggregation of colloidal The above discussion is focused on the thermodynamics of silica particles in 2 or 3 dimensions forms larger metal silicate structures, dissolution. The kinetics are equally important because the rate at which create either a continuous and space-"lling silicate (ASR) gel, or which silica dissolves in pore solution can largely control the overall more compact macro-scale precipitates [46-48]. It has been argued rate of ASR. Literature on this subject is limited and signi"cant knowl- that the presence of portlandite (or other soluble sources of calcium) edge gaps exist with respect to quantifying the dissolution rate as a is essential to form ASR gel [18,49-51]; otherwise, silica species remain function of aggregate composition, mineralogy, and surface properties, in the solution (more on this in Section 3.3). Fig. 1c is a schematic pore solution composition, temperature and pressure. The few available representation of the structure of alkali-silica (ASR) gels, which are studies have used batch reactors (e.g., ASTM C1285 [38]) to measure the composed of clustering of colloidal silica particles surrounded by the dissolution rate of silica (e.g., quartz or fused silica glass) in water [39, gel's pore solution. For simplicity and due to space limitation, the "gure 40]. A recent study by Maraghechi [41] using soda-lime silica glass shows only 2 small colloidal particles. In reality, particles are 10 to shows that the dissolution rate is highly related to the pH and temper- 30 nm in size [52]. Further, the distance between the particles is a func-ature of the system. Up to pH = 14 and at 60 °C, the dissolution rate tion of the water content of the gel. Recent work on rheology of ASR gels of glass was found to be a function of [OH]0.2. However, higher pH [53] suggests that they behave as viscous "uids with considerable yield resulted in smaller dissolution rate. Activation energies in the range strength up to 85 kPa. This indicates that the particles need to be very 60.9 to 76.6 kJ/mol have been reported for dissolution of crystalline close to each other and possibly weakly inter-connected.

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 133 2.3. Swelling of the gel attention should be focused on the mineral constituents of the rock. The thermodynamic stability of silica minerals decreases as the degree of A number of phenomena can be accountable for the signi"cant microstructural disorder increases. Amorphous SiO2 (e.g., opal, natural sorptivity and swelling of ASR gels. The silica gel is porous with a high sur- or synthetic glass) is known to be the most reactive, followed by face area, bearing many hydrophilic groups (e.g., -OH, -ONa, and -O) meta-stable crystals (e.g., cristobalite and trydimite), micro-crystalline (Fig. 1c) [31]. This results in osmosis, adsorption of water, and swelling silica, and other crystalline forms containing many lattice defects, resid-of the gel [17]. Additional osmosis could be generated by the Gibbs- ual strains, or internal microcracks. Amorphous and crypto-crystalline Donnan effect [54], where the gel acts as a semi-permeable membrane, silica forms could be due to rapid cooling from a molten state; residual allowing small alkali and hydroxyl ions to diffuse into the gel, while the stresses and microcracks could be due to rock deformation under geo-counter diffusion of larger silica ions (e.g., H3SiO 2 4 and H2SiO4 ) to con- logical forces [66] and also during aggregate processing and crushing crete pore solution is prevented [55]. This leads to a higher osmotic [65]. Some natural materials that have been identi"ed as ASR reactive pressure and movement of water from the surrounding concrete into are listed in Table 2. It should be noted that aggregates often contain the gel. Chatterji et al. [56] argued that the swelling is caused by a differ- more than one mineral and the volume of the reactive mineral needed ential kinetics of fast inward diffusion of deleterious ions (i.e., Na+, to produce deleterious ASR can be as small as 1% [1,14]. The rock matrix Ca2+ and OH) into the reactive sites and slow outward diffusion of sil- porosity (e.g., in chert) or presence of cracks in aggregates allow access ica ions from these sites. Swelling has also been attributed to the electri- of concrete pore solution to the reactive silica that could be embedded cal double-layer repulsive forces that form on the charged surface of the in the interior of aggregates. Silica micro-porosity resulting in a large gel's solid particles [57,58]. Garcia-Diaz et al. [59] attributed the ASR ex- surface area also promotes ASR.

pansion to the increase in the aggregate pore volume during the Q4 Q3 de-polymerization of silica by hydroxyl attack, where they 3.2. Aggregate size found a "ve-fold increase in the aggregate pore volume during this tran-sition. The swelling magnitude is also related to the gel stiffness, which The link between aggregate size/gradation and expansion has been is a function of the degree of crosslinking between the colloidal particles and still is the subject of study. In general, ASR expansions are expected (e.g., higher Ca gels are generally stiffer and more viscous [51,53]). to increase with reduced particle size, because of the increased surface area of reactive aggregates [14,69]. However, experimental observa-

3. Effect of aggregate properties tions and "eld reports have showed that the greatest observed expan-sion varies non-monotonically as a function of aggregate size, and the Understanding the behavior of the aggregate, as the main reactive most deleterious effects may occur for aggregates of some intermedi-phase, is fundamental to improving our scienti"c understanding of ate/pessimum size. For example, Stanton [14] found that a siliceous ASR and the way ASR is mitigated. Currently, a common practice is to magnesian limestone shows the highest reactivity for particles in the rely on the results of standardized tests (e.g., the accelerated mortar range 0.18 to 0.6 mm, while smaller particles showed no expansion.

bar test, AMBT (ASTM C1260 [60]) or the concrete prism test, CPT Poyet and colleagues found that for "int, the smallest size fraction had (ASTM C1293 [61])), to classify aggregates as: non-reactive, moderately an ASR mitigating effect [70]. Using an Alpine chloritic schist, Dunant reactive, highly reactive or very highly reactive [62,63]. These reactivity and Scrivener found that the size fraction yielding the most expansion levels are then often used to determine appropriate levels of mitigation was 2 to 4 mm [71]. Multon and colleagues [72], in a study which also using supplementary cementitious materials (SCMs) and/or chemical investigated the effect of alkalinity, found similarly for a siliceous admixtures. While this approach may be practical, it does little to cap- limestone that the fraction that yielded the most expansion was 1.25 ture the in"uence of aggregate composition and physical properties to 3.25 mm. In both of these studies, the different size fractions of aggre-on the magnitude and rate of ASR. Speci"cally, the effect of non- gates had the same mineralogy. Recycled soda-lime glass aggregates siliceous components of aggregates (e.g., Na, K, Ca, Al, and Fe-bearing show the highest expansions in the size range 1.18 to 2.36 mm (again phases) on ASR is poorly understood. In addition, it is unclear why the while maintaining the same composition and mineralogy) [64,73].

same aggregate may exhibit very different ASR expansion depending This pessimum size effect has also been observed in the "eld and has on its particle size (e.g., larger particles may show larger ASR expansion) led to recommendations like that of French [74] who noted the most or why some aggregates react at their surface while others only react deleterious aggregates to have a size from 4 to 10 mm. Interestingly, within internal micro cracks and defects [64,65]. silica fume, when well dispersed, is an effective ASR mitigator; however, poorly dispersed silica fume agglomerates have been reported to result 3.1. Aggregate composition and mineralogy in severe ASR [75,76].

Different hypotheses have been put forward to explain the Poole [1] has suggested that it is incorrect to consider rock type pessimum aggregate size effect. Very "ne reactive aggregates may re-(e.g., basalt) as a criterion for an aggregate's ASR potential; and instead, duce or even suppress the expansion, probably due to a pozzolanic Table 2 Some natural silicates susceptible to ASR [1,19,17,67,68].

Name Description Example rock types Opal Amorphous silica; a form of highly condensed silica gel Occurs as primary or secondary mineral in rocks such (contains 4-9% water); always highly reactive as chert, "int, shale, limonite, sandstone, rhyolite, marl and basalt Volcanic glass (e.g., Obsidian) Reactive in some cases Common minor constituent of "ne-grain volcanic rocks such as basalt, rhyolite, andesite Cristobalite and trydimite High-temperature polymorphs of SiO2; Are stable only Can occur in volcanic rocks at high temperature, but can crystallize and persist meta-stably at lower temperatures Cryptocrystalline ("ne grain) quartz and other Moderately reactive Constituents in rocks such as cherts and "ints and as "brous silica (e.g., chalcedony) cementing material in greywacke Mylonite Highly crystalline silica, embedded in other minerals Common Norwegian reactive material Strained quartz Only mildly reactive but degree of reactivity increases with strain Quartzite

134 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 reaction [77,78], but this does not explain the observation with larger there is believed to be a surplus of reactive silica, which may consume aggregates. In natural sands, it is likely that certain reactive minerals or neutralize the available alkalis before the concrete has even hardened dominate certain sieve sizes. As such, reactivity does not necessarily in- [80,81]. Several authors have suggested that the pessimum proportion crease for smaller particles. A related hypothesis is that the amount of will occur when the ratio of reactive silica/available alkalis is equal to reactive material in contact with pore solution differs as a function of 6 [2,45,77].

the aggregate size. In this view, the observed differences in reactivity The pessimum aggregate content effect holds some interesting of different size particles is due to the time needed to diffuse OH- to implications:

the reactive sites, which may be in the interior of aggregates [72]. A re-lated model proposed by Bazant and Steffens [79] suggests that the

• Mitigation may not be necessary for certain aggregates displaying swelling pressure of the ASR gel strongly depends on the size of reactive pessimum behavior, provided they are not used in combination with aggregates and is maximum at some intermediate particle size. Another non-reactive aggregates. Engineers can essentially use the reaction theory is that the effects observed can be explained by fracture mechan- against itself to prevent expansion with pessimum aggregates, and ics. The stress magnitude that causes splitting of an aggregate, depends this is an accepted practice in several European countries [82]. How-on the aggregate size; as such the overall macroscopic ASR expansion ever, not all aggregates displaying pessimum behavior will contain and its rate are aggregate size dependent. Detailed microstructural suf"cient reactive silica to self-mitigate [83].

modeling showed results that are consistent with this theory, although

• Pessimum behavior may lead to false negatives in accelerated labora-instead of a single crack, the model considered development of a crack tory tests such as the AMBT [84]. Such aggregates may only be correct-network in each aggregate [71]. Maraghechi et al. [65] observed that ly identi"ed as reactive by the CPT where they are paired with a non-the reactivity of recycled glass aggregates depends signi"cantly on the reactive aggregate or by replacing a fraction of the aggregate in the size/width of residual cracks that exist in the interior of aggregates, AMBT with non-reactive aggregates.

which are formed during aggregate crushing. They found that larger ag-gregates (N1.2 mm) contain wider cracks and show higher ASR reactiv-ity. They also showed that annealing glass aggregates to heal the While the mechanism of the pessimum behavior with respect to the residual cracks before mixing in concrete can signi"cantly reduce the proportion of reactive silica appears to be well-explained, testing magnitude of ASR expansions. methods need to be updated to deal with pessimum aggregates. The ap-Overall, the following conclusions can be drawn. (a) Aggregate reac- proaches given by Brouard [82] and Drimalas and co-workers [84] show tivity depends on its size, and the trend of this dependence may not be promise. It is also possible that the so-called pessimum size may be in-predictable (e.g., smaller aggregates are not necessarily more reactive). terrelated with the pessimum proportion, and more work is needed to (b) The size dependence of reactivity could be due to different reasons fully describe this relationship, because tests such as the AMBT and for different aggregates. (c) Because of such size dependence, aggre- CPT use speci"c aggregate gradations [82,85].

gates need to be tested at the same size that they will be ultimately used in "eld concrete. For example, crushing potentially reactive coarse 3.4. Surface vs. intra-particle reaction and its impact on ASR rate aggregates to allow testing their reactivity in a mortar bar test (e.g., ASTM C227 or ASTM C1260) could be erroneous [23]. At the crux of the issue of aggregate reactivity is the accessibility of the alkaline pore solution to reactive silica. If the aggregate particle is 3.3. Reactive aggregate content composed entirely of reactive silica (as is the case with many volcanic and synthetic glasses, which have uniform composition throughout Certain aggregates (e.g., "int and chert) containing highly-reactive their volume), the aggregate surface may react quickly with the alkaline forms of silica have been documented to exhibit a pessimum aggregate pore solution to form ASR gel at the surface (Fig. 3a). However, if the re-content. As summarized by Hobbs [2], for these aggregates, ASR expan- active silica is absent at the surface and is contained as "ne reactive par-sion increases with the proportion of reactive aggregate until a maxi- ticles distributed within a non-reactive matrix of the aggregate (as is the mum is reached (i.e., the pessimum content or proportion). For higher case with many greywackes and siliceous limestones), it takes time for contents of reactive aggregate, expansions decrease, and in some in- the pore solution to come into contact with the reactive silica and as a stances, reach a point where no expansion is observed at a suf"ciently result, the reaction proceeds more slowly and the gel forms within the high proportion of reactive aggregates. The pessimum content varies aggregate. In addition, the presence of pre-existing micro-cracks in the depending on the alkali loading of concrete, with higher alkalis resulting aggregate's interior will play an important role in the alkali-silica reac-in a larger pessimum content. Above the pessimum aggregate content, tion. As more surface area is exposed, the reaction rate should increase.

(a) (b)

CH Glass aggregate ASR gel Fig. 3. (a) ASR occurring at aggregate surface for granite gneiss (right) and chert particles [88], (b) despite homogenous composition of soda-lime glass particle, ASR occurs in the interior of aggregate (through pre-existing microcracks) while the interface with cement paste is protected, possibly due to presence of solid portlandite [86].

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 135 However, the local chemistry at the aggregate surface (in contact with from this source [100]. Goguel [92] showed that in selective leaching cement paste) could be different that the chemistry at the surface of tests on rapidly-cooled basalts that contained poorly crystallized feld-intra-aggregate cracks. Rajabipour and coworkers [64,65,86] observed spars, the alkali release was approximately 1.0% Na2Oeq of cement that for soda-lime glass aggregates (Fig. 3b), the presence of solid mass, whereas a well-crystallized basalt only produced about 0.1%

portlandite at the aggregate surface favors formation of innocuous poz- Na2Oeq of cement mass after leaching. Thermodynamic analysis by zolanic C-S-H [87], which forms a physical barrier against further disso- Wang and co-workers [101] showed that in alkaline solutions with a lution of glass and progression of ASR [41]. However, due to absence of pH N 13, the greatest release of alkalis from alkaline minerals in aggre-portlandite in the interior of the aggregate, ASR can proceed freely gates followed the order of: dawsonite N leucite N nepheline N dehydra-through the intra-aggregate microcracks, resulting in further cracking tion analcime N paragonite N muscovite N jadeite N Na-feldspar N and deterioration of the aggregate. K-feldspar. They also showed that a maximum contribution of alkalis was mostly affected by the pore solution/aggregate ratio rather than

4. Alkalis the level of the pH of the pore solution.

Batic and co-workers [102] conducted accelerated mortar bar Suf"cient alkalis are required to raise the pore solution hydroxyl tests on basalts containing montmorillonite and volcanic glass from concentration to a level (e.g., 0.2 to 0.25 M [89,90]) at which the Argentina that were suspected of causing ASR in structures. Examina-aggregates are attacked at a signi"cant rate, and to sustain the attack tion of thin sections showed that both volcanic glass and montmorillon-over time. While the majority of alkali contribution is linked to the ce- ite participated in ASR by supplying soluble silica. In addition, the alkalis ment (e.g., alkali sulfates), aggregates and SCMs can also supply alkalis in montmorillonite could interchange with Ca2+ from the concrete pore to further exacerbate the reaction. In addition, external sources, such solution, liberating Na+ or K+. A zeolite (sodic, potassic) was found to as seawater or deicing salts, could also increase the alkalinity of the con- be the principal component observed in the ASR reaction products.

crete. This section discusses studies to evaluate alkali contributions from In contrast to previous research which relied on indirect measure-aggregates, SCMs, and external sources, as well as differences in the ments of alkali leaching potential from aggregates, Constantiner and mechanisms of ASR as it relates to the type of alkali cation (i.e., Na+, Diamond [103], measured the pore solution of concretes containing ag-K+, and Li+). gregates with varying alkali-bearing feldspar minerals. They con"rmed that these minerals did contribute alkalis to the pore solution of 4.1. Alkali contribution by aggregates concrete. They further demonstrated that some of these alkalis were tied up in the alkali-silica reaction products. Work by Ideker and co-Alkalis can be found in both reactive and non-reactive aggregates. workers [98] also showed that alkali release from the so-called non-Reactive aggregates are those that contain a metastable silica structure reactive natural granitic "ne aggregate fraction of a concrete mixture that is attacked and dissolve by OH- ions, leading to formation and swell- in the concrete prism test (CPT) could contribute alkalis to the pore ing of ASR gel and ultimately cracking of concrete. When alkalis are solution in the 38 °C (standard) version of the test and in a non-present within this reactive silica structure (e.g., see Fig. 1b), they can standard accelerated version run at 60 °C. This granitic sand and seven be gradually released into the pore solution as the silica dissolves. For ex- other "ne aggregates in the study, were classi"ed as non-reactive ample, Shafaatian [91] showed that when soda-lime glass aggregates or based on a 14-day expansion of less than 0.10% in ASTM C1260 soda-lime glass powder/SCM is used in concrete, Na is released into pore (AMBT) [60] testing. In addition to alkali release, the non-reactive "ne solution as a result of glass dissolution, and this maintains a high pH level aggregates, when combined with the benchmark reactive coarse aggre-in the pore solution of the concrete over a long period of time. gate Spratt (a Canadian siliceous limestone), were shown to produce a Furthermore, even some alkali-containing aggregates that are classi- high variance in the CPT results (both at 38 °C and 60 °C).

"ed as non-reactive, have been shown to contribute to the alkalinity of More recent work by Giannini [104] suggests that a highly-reactive concrete pore solution. Even though these aggregates are not consid- "ne aggregate from El Paso, Texas, commonly known as Jobe, may ered alkali-silica reactive (e.g., because of their low expansion when release signi"cant quantities of alkalis. This aggregate has been the tested according to the accelerated mortar bar test, ASTM C1260 [60]), subject of numerous ASR studies [105-107], particularly when rapid they may contain releasable alkalis in certain mineral phases. Minerals and severe expansion from ASR is desired. Pulverized samples of the known to release alkalis include alkali feldspars, micas, zeolites, and aggregate subjected to hot-water extraction with a soak time of 24 h re-clay minerals [92-97]. The exact mechanism of alkali release by non- leased an average Na2Oeq of 0.24 kg/m3 (by mass of concrete), assuming reactive minerals is not completely understood but it could be through the aggregate would constitute 30% by mass of concrete. Fig. 4 shows some ion exchange reactions (e.g., replacement of Na from siliceous aggregate with Ca from pore solution; Eqs. (5a) and (5b) above).

Water-soluble alkalis, Na2Oeq (kg/m3)

It has been suggested [98] that alkali release from non-reactive 7.50 aggregates can be large enough to in"uence the concrete prism test 7.00 (CPT, ASTM C1293 [61]) results. Field studies [94,99] of dams have 6.50 found cases where the total concrete alkalis measured by hot-water extraction exceeded the estimated alkali contribution by portland 6.00 cement, in many cases by more than 1.0 kg/m3 of concrete (or approx-5.50 imately 0.25 to 0.30% Na2Oeq by mass of cement, assuming concrete with a typical cement content). In at least one case [99], deleteriously 5.00 expansive ASR was attributed to alkalis released by the aggregates. 4.50 Hobbs [2] summarized that certain aggregate types including 4.00 arti"cial glasses, volcanic glass and Witwatersrand quartzite could 0 0.5 1 1.5 2 contribute signi"cant quantities of alkalis to concrete. He also pointed Average Expansion, Transverse Direction (%)

out that in the southwestern United States, the combination of low-Beam 1 Beam 2 Cement Alkalis + NaOH alkali cements with certain rhyolites and andesites resulted in cracking of concrete pavements by ASR. He attributed this to alkali contribution Fig. 4. Water-soluble alkali content (kg/m3 of concrete) for two full-scale beams after two from these aggregate types. In South Africa, the Witwatersrand quartz- years of outdoor exposure, plotted against expansion. Dashed line represents total alkalis ite was shown to contain between 0.1 and 0.7% Na2Oeq, and ASR was from cement and NaOH admixture at mixing.

implicated in concrete containing both the "ne and coarse aggregate Adapted from Ref. [104].

136 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 water-soluble alkali data for cores taken from full-scale beams fabricat- effective at increasing the alkali-binding capacity of C-S-H than high-ed in the laboratory using this aggregate and left in outdoor exposure Ca ashes. This greatly increases the availability of alkalis for ASR where site for two years. Water-soluble alkali contents measured via hot high-alkali + high-Ca "y ashes are used [115,116]. Efforts to develop water extraction ranged from 0.42 to 1.60 kg/m3 greater than the a reliable standardized test method or methods should attempt to take known contribution from portland cement and NaOH addition. this into account, and given the widespread use of "y ashes in mitigating This may help explain the extraordinary capacity of this aggregate to ASR in new concrete, such a test method is of critical importance.

contribute to expansive ASR, even at very low alkali loadings from non-aggregate sources. However, based on these data, researchers 4.4. External alkalis would be wise to exercise caution with this aggregate, and avoid using it as the sole reactive aggregate in phenomenological studies. External sources of alkalis include seawater, deicers, and anti-icers.

Seawater comprises approximately 3.5% of salts, with NaCl being the 4.2. Testing for alkalis in aggregates primary constituent. Sources of alkalis in deicing and anti-icing chemicals include NaCl, alkali acetates, and alkali formates. The latter Considerable research efforts have been placed on evaluating the two classes of chemicals do not contain chlorides, and are typically alkali contribution of aggregates by performing solution-based leaching used in corrosion-sensitive air"eld and bridge deck applications.

tests. Typically aggregates (or concrete) are processed to a certain "ne- Potassium acetate (CH3COOK), in particular, has attracted signi"cant ness and exposed to water (distilled or deionized), alkaline solution attention in the last decade for potentially accelerating cases of ASR in (e.g., 1 N NaOH) and/or an alkaline solution plus Ca(OH)2. The aggre- air"eld pavements containing reactive aggregates [117-121]. Laborato-gates are allowed to react (or leach alkalis) at different temperatures, ry studies by Giebson et al. [122] showed that potassium acetate can typically 38-80 °C and for varying time periods. The solution is analyzed react with portlandite in the cement paste to form calcium acetate and the alkali release is quanti"ed. Pore solution expressed from cement and increase the hydroxyl concentration of the pore solution. Other pastes, mortars and concrete has also provided useful information on laboratory studies suggest that this mechanism can lead to ASR in con-the contribution of alkalis from various sources. Since it is dif"cult to crete containing reactive aggregates, even if normally effective mitigat-obtain pore solution from "eld concrete, destructive techniques involv- ing measures such as low alkali loadings, "y ash, or slag are used [123, ing the aforementioned leaching tests are generally used. In testing of 124]. Outdoor exposure testing and observations from suspected "eld 17 Canadian aggregates, Bérubé et al. [94] determined that an alkaline cases seem to indicate that the potential for potassium acetate to lead soak solution of 0.7 N NaOH or KOH was the most representative of to distress in the "eld may be signi"cantly overestimated by laboratory conditions inside concrete. Alkalis released in these tests ranged from tests, which are commonly conducted using elevated temperatures, 0.1 to 12.7 kg/m3 (by volume of a typical concrete mixture) after small specimens, and immersion in highly concentrated solutions 578 days several times greater than that obtained by soaking in (e.g., 50% potassium acetate). Field applications of deicers and anti-lime-saturated or pure water. A task group from RILEM committee TC icers occur during or prior to cold-weather precipitation events. As a re-219-ACS is currently working to develop and standardize a reliable sult, they are diluted by melting snow and ice, while the lower temper-method for measuring alkalis released by aggregates to be designated atures will impede any diffusion into and reaction with the concrete as RILEM AAR-8 method. This is expected to be completed in the near [124].

future, but additional work will be needed to determine criteria for Cores from a pavement exposed to ten years of potassium acetate interpreting test results [108]. applications showed an increased potassium content to a depth of only 5 mm, while distress from ASR was detected uniformly through 4.3. Alkali contribution by SCMs the depth of the pavement [124,125]. Hayman et al. [124] reported that slabs containing reactive aggregates placed in outdoor exposure Certain SCMs (primarily some moderate- to high-Ca "y ashes) and periodically treated with potassium acetate in Texas, USA (mean include signi"cant contents of alkalis (e.g., Na2Oeq N 6% by mass [109, annual temperature = 20.8 °C) experienced an increase in expansion 110]). Monitoring of outdoor exposure concrete blocks [111] show over untreated control slabs. Companion specimens in New Brunswick, that some mixtures with high-alkali SCMs show acceptable ASR perfor- Canada (mean annual temperature = 5.6 °C) treated with 100 applica-mance only up to 10 years, but after this time cracking began to appear. tions each winter for three years exhibited no increase in distress com-This was potentially due to slow release of alkalis from SCMs. All such pared to controls. At the time of this writing, there are no reports in the mixtures had expansions less than 0.04% after 2 years in the concrete literature that describe the performance of concrete exposed to potassi-prism test (CPT), and therefore would not have been expected to exhibit um acetate over a period of time equivalent to the expected service life expansive behavior in outdoor exposure. This further points out a of a bridge or air"eld pavement. Therefore, it is not possible to state that limitation in the CPT that the users should be aware of. potassium acetate will never lead to worsening ASR in "eld structures.

Currently, no standard test method is available to quantify the total The ingress of NaCl into concrete, either from marine exposure or its alkalis contributed by SCMs to concrete pore solution over the service use as a deicing salt, has also been thought to exacerbate certain cases life of a structure, or the rate at which alkalis are released. This and laboratory studies of ASR [126-129], while having no effect on signi"cantly impacts the ability to predict the service life of structures, expansion in others [130,131]. Although the overall effect on the pore especially based on accelerated ASR tests. ASTM C114 [112] provides solution pH caused by the addition of NaCl should be neutral [131],

test methods for determining the total and water-soluble alkalis in Cl ions can react with aluminate phases to form Friedel's salt SCMs, and a test method for determining available alkalis in pozzolans (calcium chloroaluminate) and can be replaced in the pore solution is provided in ASTM C311 [113]. The ASTM C311 procedure involves with OH ions released to maintain charge neutrality [132]. However, leaching alkalis from a paste sample immersed in a saturated limewater Suryavanshi et al. [133] noted that Freidel's salt forms primarily by solution for 28 days, and has been the subject of criticism for both two mechanisms: one that binds Cl ions and releases OH ions from under- and over-estimating available alkalis [114,115]. Recommenda- portlandite or AFm phases to the pore solutions, and another that tions to improve this method include extending the test to allow more involves adsorption of chlorides to AFm phases and binding of Na+

alkalis to leach out [114], or using a more alkaline soak solution with ions to the C-S-H. Both mechanisms may occur simultaneously, and an [OH] in the region of 0.25 mol/L [115]. binding is primarily limited by the C3A and C4AF contents of the cement.

There remains little agreement internationally regarding the The adsorption mechanism does not increase the pore solution alkalin-percentage of total alkalis in SCMs (particularly in "y ashes) that can ity because no OH ions are released; therefore, the formation of be considered available for ASR. Low-Ca "y ashes are much more Friedel's salt does not guarantee an increase in pore solution pH. Yuan

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 137 et al. [134] state that it may be possible to predict chloride binding limit of the concrete prism test (CPT). They also showed that the capacity of concrete from the cement composition if only portland response of lithium to aggregate reactivity level and mineralogy could cement is present. However, for binder systems involving one or more not be easily correlated. Therefore, they recommended performance SCMs, more work is needed before chloride binding capacity can be testing to determine the safe level of lithium based on the speci"c adequately modeled. This has signi"cant implications not only for ASR type of potentially reactive aggregate being investigated [153].

in concrete exposed to chloride-based deicers and marine environ- This was con"rmed by Drimalas and co-workers [154] where long-ments, but also for service-life prediction related to corrosion of steel term exposure of concrete blocks incorporating different aggregate reinforcement. types and various dosages of lithium-based salts exposed for up to 16 years showed a varying response to lithium based on aggregate 4.5. Lithium type. In their study, Drimalas, et al. demonstrated that either 30%

class C "y ash (27.7% CaO) or a LiNO3 dosage providing molar ratio The use of lithium compounds to control ASR was "rst reported in [Li]/[Na + K] = 0.56 (equivalent to 75% of the standard dosage of 0.74 1951 by McCoy and Caldwell [135]. In this study, over 100 different molar ratio) were both effective in reducing the expansion of highly re-chemical compounds for controlling ASR were investigated using the active "ne aggregate from Texas, with the 75% LiNO3 showing no expan-ASTM C227 [15] test method. A wide range of lithium compounds that sion observed after 3500 days in outdoor exposure. However, a mixture nearly eliminated expansion at 8 weeks of exposure at 38 °C were with 30% class C "y ash and 75% LiNO3 dosage only showed a 50% reduc-deemed the most promising. After this, research on lithium compounds tion in expansion compared to 30% class C "y ash alone, reaching an ex-was nearly stagnant until the late 1980s. Much of the focus of research pansion of 0.20% at 3500 days. This demonstrates that there may not be since that time has been on understanding the mechanisms by which synergistic effects in terms of expansion reduction when lithium is com-lithium compounds control or reduce ASR [136-145]. A thorough sum- bined with "y ash of certain composition. Speci"cally interactions be-mary can be found in several review documents [146-148]. Three main tween Ca and Li are unknown and could be signi"cant.

lithium compounds emerged as candidates for controlling ASR: LiOH, Work by Feng and co-workers [145] focused on Vycor glass disks LiCO3 and LiNO3. From the research completed in the late 1980s through and glass disks sandwiching a cement paste and exposed to solutions early 2000, several mechanisms were postulated by which lithium of NaOH, NaOH plus calcium, and NaOH plus calcium and lithium.

compounds may control ASR: Their "ndings showed that in the presence of lithium and calcium, a layer of Li-Si crystals was formed on the surface of the glass disk that

• Lithium may reduce the dissolution rate of silica, helped to protect the silica from further dissolution. Furthermore, the

• Lithium may impede formation of alkali-silica gel, presence of calcium formed a dense rigid alkali-silica gel composed of

• Lithium may reduce repulsive forces between the colloidal gel parti-Li with low-Ca concentrations. These authors postulated that this gel cles of ASR, thus reducing swelling, and would have a low capacity to absorb moisture and thus a low swelling

• Lithium may incorporate into the alkali-silica gel, altering its proper-capacity.

ties and expansion.

Later work by Tremblay and co-workers [155] did not support many previously held theories about the mechanisms by which LiNO3 may Kawamura and Fawa [149] and Mo [150] both noted changes in the reduce ASR. Mechanistic work relying on pore and immersion solution chemical composition of the alkali-silica gel in mortars as dosages of analysis, XRD, SIMS and petrographic examination provided results LiOH and Li2CO3 increased, leading to gels with a lower CaO/SiO2 ratio. suggesting the following classes of mechanisms may be unlikely as the Silica appeared to react with lithium preferentially over other alkali cause of ASR reduction by Li: pH reduction, increased solubility limit species. Mo and co-workers [151] hypothesized that the smaller ionic of silica to prevent gelation, formation of a protective coating, non-radius and higher charge density of lithium (compared to Na or K) was expansive crystalline product, and non- to less-expansive amorphous responsible for the latter phenomenon, but none of these studies could gel. From their study, the mechanism most supported was that the explain why the reaction product with lithium was less expansive. Later presence of lithium provided increased chemical stability and reduced work by Leeman and co-workers [163] and Bulteel and co-workers dissolution of reactive silica due to reasons other than pH reduction or

[152] supports the hypotheses that lithium may reduce the solubility formation of a protective coating [155]. Clearly, this debate has not and dissolution rate of silica. been settled and more research into the impact of lithium salts on differ-From laboratory studies and "eld experience, LiNO3 emerged as the ent aggregate mineralogies is needed to de"nitively make conclusions preferred lithium compound for controlling ASR, as many other lithium on the mechanisms by which lithium controls alkali-silica reaction in salts increase the pH of pore solution by reacting with Ca(OH)2; for concrete. Furthermore, the long-term ef"cacy of lithium nitrate example: needs to be veri"ed through correlation of laboratory studies and the growing body of information available from "eld exposure. This is Li2 CO3 CaOH2 2Li 2OH CaCO3 : 8 speci"cally important in light of shrinking resources and availability for low-calcium low-alkali "y ash to control ASR.

LiNO3 does not signi"cantly raise the pore solution pH due to the relatively high solubility of the calcium analog Ca(NO3)2 [141]. 4.6. Research needs for alkalis Initially, it was proposed that the dosage rate of lithium to control ASR be based on the molar concentration of sodium and potassium sup- Future research should address the apparent disconnect between plied by the portland cement. A molar ratio [Li]/[Na + K] of 0.74 or accelerated laboratory tests, often conducted at temperatures of 38 to greater emerged from work by McCoy and Caldwell [135] as the dosage 80 °C and near-100% relative humidity, and variable real-world condi-rate needed to control ASR, and this dosage rate was used until the mid tions experienced by a concrete structure over its service life. This to late 2000s. At this time, research by Tremblay and co-workers [144] may involve working with the growing database of "eld exposure showed that one dosage rate of lithium was not equally effective across sites to identify how "uctuating environmental conditions can impact all classes of aggregates. They investigated 12 aggregates of different the availability of alkalis in "eld concrete.

mineralogies and showed that 50% of the reactive aggregates responded The current understanding of the threat posed by alkali release by well to the standard dosage (0.74 molar ratio). For three of the aggre- aggregates is primarily based on accelerated solution-leaching studies gates tested, a higher dosage of LiNO3 ranging from 0.75 to 1.04 was in the laboratory, as well as several anecdotal cases involving structures required to control ASR. For the remaining three aggregates, a 1.11 in the "eld. Systematic research is needed to establish the expected molar ratio was not suf"cient to limit ASR below the 0.04% expansion alkali release from aggregates of a range of mineralogies in concrete

138 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 with a range of alkali loadings. This is particularly important for long-life and Thaulow [161] showed, using electron probe microanalysis of con-hydraulic structures where the combination of alkali release, alkali mi- crete, that gel formed within or close to aggregates was low in calcium gration, and a near-constant supply of moisture in the structure could and high in alkali. But as the distance from the aggregate increased, result in an alkali concentration suf"cient to trigger and then maintain the calcium content of the gel increased, presumably due to close expansive ASR. This research would be facilitated by the many outdoor association with the calcium-rich cement paste. Thomas [162] com-exposure sites that have been constructed over the past couple of de- pared the composition of ASR gels found in 7-year-old laboratory cades; specimens on these sites are well-documented with respect to concrete and a 55-year-old concrete dam with the composition of their materials, initial alkali loadings, and expansive behavior. Re- calcium-silicate hydrate (C-S-H) in the concrete from the dam. The searchers should couple testing of exposure site specimens with labora- data expressed as alkali/silica and calcium/silica atomic ratios are pre-tory tests such as the forthcoming RILEM AAR-8 method. sented in Fig. 5. The composition covers a wide range, but it appears The widespread use of SCMs in mitigating ASR demands an im- that there is a reasonable relationship between the alkali and calcium proved understanding of alkali release and binding by these materials. contents (i.e., as the calcium content increases, the alkali content Fly ashes, with their widely variable chemistry (sometimes with very decreases). This supports the concept of a cation exchange with the high alkali contents), are perhaps in greatest need of attention. In calcium replacing the alkali in the initially alkali-rich reaction products.

North America, the future of reliable sources of "y ash are uncertain as This exchange occurs as a portion of the gel migrates away from the utilities shift to fuels other than coal, and other SCMs are being looked aggregate particle and comes into contact with the calcium-rich cement to for ASR mitigation. It is likely that the same questions regarding alkali paste. The data in the "gure indicate that the process continues slowly availability will abound in new mitigation materials. Research on alkali as the concrete ages and that the composition of the reaction product release from SCMs should be conducted in parallel with research on may ultimately approach something similar to C-S-H. This "nal alkali release from aggregates. product appears to retain very little alkali, indicating that almost all Another signi"cant research need is to examine the differences of the alkali that originally participated in the reaction process may be between Na and K with respect to ASR because most accelerated tests eventually recycled to participate in further reaction.

use NaOH, rather than KOH, to boost alkali loadings. However, most Thomas [162] suggested that alkali recycling might partially explain cements contain a higher proportion of K2SO4 versus Na2SO4. Investiga- why in many large concrete dams, the expansion continues for decades tions should focus on whether the composition of alkali boosting and does not appear to reach a maximum value. This is not the case for solutions in ASR studies should be matched to the chemistry of the laboratory tests (e.g., ASTM C1293 [61] where specimens are stored cement. Studies of expressed pore solution are particularly valuable, over water), in which the expansion-time relationship is typically rep-but are not without "aws. Some researchers suggest that it may be resented by an S-shaped curve, reaching a maximum after a few years more dif"cult to express Na than K, and have observed a change in the or less. In large structures, alkali recycling may continue to fuel ASR ratio of these species as increasing pressures are applied [156-158]. (until all the reactive silica or portlandite are consumed), whereas in Even the most advanced apparatus can only express about 40% of the laboratory specimens, leaching of the alkalis from the specimens even-available pore solution at a pressure of 1000 MPa [159]. The composi- tually reduces the alkali concentration to a level below that necessary to tion of the remaining 60% remains uncertain. Rather than pursuing sustain ASR.

higher pressure pore solution expression devices, the authors encour-age researchers to agree upon a standardized pressure and overall pro- 5.2. Effect of calcium on the properties of ASR gel and its expansion cedure, enabling them to better compare results obtained in different laboratories. A number of experimental studies have concluded that signi"cant In each of the above areas, it will be critical to identify where alkalis expansion only occurs when an adequate supply of calcium is available in concrete reside. This is likely to require a combination of bulk analysis in the form of portlandite. In systems with abundant alkali hydroxides techniques (e.g., water- and acid-soluble alkali testing, and pore solu- and reactive silica, but no portlandite, silica dissolves and remains in tion analysis) and micro-analytical techniques (e.g., SEM-EDS, EMPA, solution [44,163]. Although the precise role that calcium plays in gel and micro-XRF). expansion remains unclear, a series of mechanisms have been proposed, as listed below [164]:

5. Calcium

• High calcium concentrations in the pore solution prevent the diffusion Two major roles played by calcium in the alkali-silica reaction are: of silica away from reacting aggregate particles [18,165].

• If calcium is not available, reactive silica dissolves in alkali hydroxide

• Exchange with alkalis in the reaction products (gel) alkali solution without causing damage [50,164].

recycling, and

• The formation of calcium-rich gels is necessary to cause expansion,

• Its impact on the composition and properties of the reaction products (gel), and its expansion.

7-Year-Old Concrete: Gel in aggregate 0.3 7-Year-Old Concrete: Gel in paste 55-Year-Old Concrete Gel 5.1. Alkali recycling C-S-H 0.2 (Na+K)/Si Hansen [160] "rst proposed that calcium may replace alkalis (Na and K) in the reaction products (i.e., ASR gel), thereby regenerating alkalis for further reaction. Note that the pore solution pH can therefore 0.1 increase by dissolution of portlandite to compensate for Ca uptake, according to Eqs. (5a) and (5b). Hansen's hypothesis was based on the 0.0 observation of depleted portlandite in the cement paste immediately 0.0 0.4 0.8 1.2 1.6 2.0 in contact with ASR gel in cracks. He observed a darkening of the Ca/Si paste adjacent to gel-"lled cracks or around the periphery of reacting aggregates and attributed this to the reduction in the quantity of Fig. 5. Composition of alkali-silica gels in concrete.

portlandite, which he con"rmed by petrographic examination. Knudsen From Ref. [162].

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 139 (a) (b)

Flint relict A2 B1 B2 A2 B2 B1 Flint B2 relict A1 Fig. 6. BSE images of polished samples from (a) portland cement concrete showing high-alkali gel (A2) within "int grain and high-calcium gel (A1) around periphery of grain, and (b) "y ash (40%) concrete showing high-alkali, low-calcium gel (B1) within dissolved (relict) "int grain and gel (B2) penetrating and saturating cement paste surrounding "int relicts without causing damage. Width of view approximately 2 mm for both images.

From Ref. [51].

either directly or indirectly through the formation of a semi- signs of damage to concrete (Fig. 6b). They attributed this to the very permeable membrane around reactive aggregate particles [51,164, low viscosity of the reaction product in the absence of calcium. By con-166]. Fig. 6 shows the formation of a high-calcium, alkali-silica reac- trast, other workers [169,170] showed that the addition of CaO can in-tion product in concrete containing reactive "int aggregate. duce expansion in concretes containing "y ash or silica fume.

• Calcium increases the viscosity and yield strength of gel, resulting in

• Urhan [171], summarizing observations from a number of studies, larger stress magnitudes generated by the gel expansion [51,53]. proposed that there is a progressive passage from low viscosity alka-li-silica gel to C-S-H and that, this is accompanied by a change in the physical and mechanical properties of the gel. Fig. 7 illustrates The important role of available calcium (in the form of portlandite) the changes in composition, physical and mechanical properties, and is supported by the following reported observations: structure of the reaction product. Gel with low calcium content may have a high swelling capacity, but the viscosity is very low. On the

• Expansion of mortars or concretes containing reactive aggregate can other hand, gel high in calcium is more rigid, and does not swell.

be prevented by removal of the Ca(OH)2 prior to immersion of the Although not explicitly stated by Urhan [171], at some point the sample in alkaline solutions. Chatterji [18] removed the Ca(OH)2 composition of the gel is such that the gel has suf"cient swelling ca-from concrete by leaching, and Thomas [164] carbonated the pacity and viscosity to cause damage to the surrounding cement Ca(OH)2 by exposing mortar bars to an enriched-CO2 environment. paste. However, earlier work of Struble and Diamond [172] did not re-

• Pozzolans are effective in controlling the expansion of concretes (and port a noticeable reduction in the free swelling and swelling-pressure mortars) immersed in alkaline salt (including alkali hydroxide) solu- characteristics of calcium bearing silica gels in comparison with tions [51,137,167,168]; and this has been attributed to the consump- sodium silica gels.

tion of Ca(OH)2 by the pozzolanic reaction, which reduces the availability of calcium for ASR. Bleszynski and Thomas [51] showed considerable evidence of reaction in concrete containing 40% "y ash and a reactive "int sand after 4 years immersion in 1 NaOH at 80 °C, 6. Role of aluminum but no expansion. ASR gel was observed to be migrating into and sat-urating the cement paste surrounding reactive "int particles, with no Research on the role of Al on ASR has been limited; but there is evi-dence that the presence of soluble Al may help in mitigating ASR. Al-rich SCMs (e.g., "y ash, metakaolin) were found to be more effective against Na+, K+ concentrations decrease Changes in ASR than SCMs that were only rich in silica (e.g., silica fume) [173-175].

chemical Ca2+ concentration increases composition Schumacher and co-workers [176,177] and later Warner [178] showed evidence that "y ash with Al2O3 N 20 wt.% were better able to control viscosity increases ASR than comparative "y ashes with lower alumina but similar calcium Changes in volume changes decrease physical and oxide contents. Warner et al. [179] demonstrated that binary and terna-water absorption decreases mechanical ry blends of SCMs rich in Al2O3 had several effects on pore solution properties chemistry compared to mixtures with lower Al2O3: (1) the pore solu-mechanical strength increases tion pH was lower; (2) aluminum was found in increased amounts in Structural pore solution and this remained stable and/or increased up to 90 days low viscosity gel high viscosity gel crystallization C-S-H of age; (3) potassium amounts were reduced in pore solution; and changes (4) ternary blends were most effective at depleting CH. They also dem-Fig. 7. Changes in the nature and properties of the alkali-silica reaction product [171]. onstrated that "y ashes with Al2O3 b 20% could be augmented with

140 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 alumina powder to improve ASR mitigation ef"ciency. Their work the response to changing alkali concentrations of the gel phase mainly further showed that ternary blends of "y ash with either metakaolin in the undersaturated region and, if possible, determine the role of Al.

or ultra-"ne "y ash (both rich in Al2O3) showed improved ASR mitiga- In a real concrete undergoing ASR, the concrete pore solution alkalinity tion compared to just "y ash on its own. Work by Shafaatian [91] is a dynamic system changing as the reaction progresses and potentially shows signi"cant reduction of ASR expansion when Al(OH)3 was even "uctuating due to alkali recycling with calcium [162,164]. What admixed in ASTM C1260 mortars. remains unclear is how stable the system is. Binding alkalis in the Possible explanations for the bene"cial effects of Al on ASR can be short-term has been demonstrated in the laboratory. However, a summarized as: reduction in binding capacity as concrete ages (e.g., pore solution pH de-creases due to ongoing ASR or carbonation depleting Ca(OH)2) could be

• Reducing silica dissolution rate from aggregates, by Al adsorption on detrimental to such mitigation measures.

the surface of reactive silica and/or formation of a diffuse zeolite barrier [180-182];

• Reducing the pore solution pH by formation of C-A-S-H, which 7. Effect of exposure conditions improves alkali binding capacity over C-S-H [183];

• Reaction with and consumption of portlandite to form calcium alumi- Although ASR expansion is signi"cantly dependent on the availabil-nate phases; ity of moisture [2,195,196], research on interactions of water and ASR

• Reducing swelling of ASR gel; gel (and how gel properties are affected by moisture content, temper-

• Densifying pore structure, and reducing permeability of concrete. ature, and pressure) has been very limited. Most available conclusions are based on expansion of concrete stored at particular relative humid-ity (RH) as opposed to direct testing of the gel. These results are Among the above possible mechanisms, limited past research has dif"cult to interpret and reproduce since they are also greatly affected been conducted on the "rst and second mechanisms. Chappex and by the elastic and fracture properties of concrete. It is believed that an Scrivener [182,184] demonstrated that presence of aluminum in internal RH N 80% is required to support expansive ASR [197]; but pore solution can limit the dissolution of silica from aggregates. Two specimens stored at drier environments may still undergo silica potential mechanisms have been suggested to explain how Al may dissolution and gelation reactions, although swelling will only occur impact the dissolution of silica in alkaline environments. (1) Al can after the specimens are moved back to a humid environment [1,198].

react with the silica and form an aluminosilicate layer on the silica Larive et al. [199] argued oppositely that water only increases concrete surface, which could protect the substrate silica against further attack expansion when it is available during formation of the ASR gel; any of hydroxyl ions [185]. (2) Presence of Al in the solution, passivates subsequent water additions will not cause extra swelling [200]. Poyet or poisons the surface of silica glass through surface adsorption of et al. [201] showed that the RH threshold to support expansions de-the Al(OH) 4 ions. The negatively charged aluminosilicate surface sites pends on temperature, and concrete at 59% RH and 60 °C can undergo thus formed, repel OH ions and thus depress their attack on the silica ASR. This is understandable as the moisture content of concrete, and structure [180,186]. thus the moisture available for swelling of ASR gel, is dependent not Aluminum is known to interact with C-S-H to form the aluminum- only on the internal RH but also on the temperature. This is further modi"ed calcium silicate hydrate C-A-S-H. Komarneni and co-workers an indication that accelerated laboratory tests that do not mimic and

[187-189] demonstrated through NMR studies on tobermorite that Al signi"cantly depart from the real life cyclic temperature and moisture was mainly found in C-S-H in tetrahedral coordination Al[IV] occur- conditions may be poor indicators of real concrete behavior and ring in both Q2 and Q3 chain sites that linked across interlayers. durability.

They suggested that this linkage played an important role in cation ex- Using the analogy with C-S-H, it may be presumed that ASR gel has change properties. Other researchers [190,191] also demonstrated a certain moisture content and expansion capacity at 100% RH, and will that Al can be incorporated in C-S-H as Al[IV] and Al[VI]. Lognot and undergo partly irreversible shrinkage and swelling with cycling RH co-workers [192] proposed that Na+ played an important role in pro- [202,203]. The magnitudes of shrinkage and swelling are also likely de-viding charge balance when Al substituted for Si in C-S-H. Recently, pendent on gel composition and temperature. Such information on ASR Sun et al. [193] prepared Al-substituted C-S-H by precipitation and gels is largely unavailable. By exposing dried synthetic gels to water, showed signi"cant increases in basal spacings when Al was incorpo- Struble and Diamond [172] measured the swelling capacity of a number rated into C-S-H. They concluded that alumina was present in the in- of ASR gels and reported signi"cantly different results (e.g., 0.5% to terlayer and that the substitution of Si by Al provides a () site that 81.6% free swelling) even for gels with fairly similar compositions. It is must be charge balanced by a net (+) charge (e.g., Na+, K+), resulting important, however, to note that pre-drying could disturb the structure in alkali binding. and properties of gels. Also, in exposure to ambient CO2 (e.g., while dry-The work of Hong and Glasser [183] showed that C-A-S-H may ing), ASR gel is gradually carbonated, which affects its properties [204].

be better able to bind alkalis and reduce the pH of pore solution. No data are available on the rheological properties (e.g., yield strength, They demonstrated that for synthetic C-A-S-H with low Ca/Si viscosity) of ASR gel as a function of its composition, moisture content, ratio (0.85), the alkali binding capacity was dramatically improved and temperature.

for both Na+ and K+ in comparison to a C-S-H not incorporating Research is needed to determine how much moisture is absorbed by alumina. Chappex and Scrivener [194], however, suggested that in- ASR gel at a given temperature and relative humidity, and how much corporation of alumina may not increase the alkali binding capacity swelling this causes. Does the swelling stop under con"ning pressure; of C-S-H in real cementitious materials (compared to synthesized in other words, what level of stress is generated when the gel is re-C-A-S-H previously studied). They suggested that reduction of sil- strained from expanding? Can this stress be relieved by hydraulic "ow ica dissolution as a result of adsorption of aluminum on the silica of the gel through cement paste's pore network (which depends signif-surface is more likely to explain the bene"ts of aluminum in reduc- icantly on the yield strength and viscosity of the gel, as well as perme-ing ASR. ability of the matrix)? And how are all these parameters affected by A central theme from these studies emerged; a better understanding the gel composition, moisture content, and temperature? Answering in real cement pastes and concretes is essential for further study and these questions are important for correctly predicting the ASR perfor-determination of the ef"cacy of alumina modi"ed C-A-S-H. In fact, mance of concrete in a given location/exposure environment and to Hong and Glasser [183] highlighted this shortcoming, stating that to better assess how certain species (e.g., Li, Ca, Al) affect the properties understand the behavior of real concretes, it is important to consider of ASR gel.

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 141

8. ASR mitigation charged rebar) shows higher penetration depth but comes at the cost of OH release at the cathode (negatively charged rebar) due to electro-A classic conclusion that is drawn from the above discussions is that chemical reduction of water. Currently, the most viable mitigation ASR requires four main prerequisites: (1) a source of reactive silica method for existing structures is limiting concrete's access to external (commonly contained in aggregates), (2) alkalis or more correctly, moisture by improved drainage, application of penetrating sealers, OH ions in concrete pore solution to attack silica, (3) a source of solu- and use of cladding [8]. Application of restraint and con"nement (e.g.,

ble Ca (e.g., portlandite) to react with dissolved silica and form deleteri- FRP wrapping of columns, enclosing affected concrete with new ous gel, and (4) access to moisture to allow gel expansion [1,2]. As a prestressed concrete) has shown some success. Other methods for result, mitigation methods are often targeted at eliminating one or treating the symptoms (and not the cause) of ASR include crack "lling more of such prerequisites. (to reduce the risk of subsequent damage by freeze-thaw or corrosion),

Mitigation methods can be divided into methods that are applicable and slot cutting (e.g., in dams) to relieve ASR induced compressive for new concrete and those that can be used for existing structures. For stresses.

new concrete, the following mitigation options could be available [25, 62]: 9. Modeling state of the art and future prospects

• Use of non-reactive aggregates (e.g., those producing less than 0.04% Reliable modeling of ASR could be an extremely useful tool for expansion in ASTM C1293 test [61] or less than 0.1% in ASTM C1260 prediction of the remaining life of affected structures, for optimum test [60]): Due to limited availability of truly non-reactive aggregates scheduling of repairs, and for design of these repairs. The available ex-in many locations, this option is not often practical. perimental methods are especially inaccurate with respect to predicting

• Limiting the alkali content of concrete: For example, a maximum alka- the future performance and progression of ASR in structures in service, li content of Na2Oe = 1.8 kg/m3 of concrete (to be achieved using low and reliable models could have great utility in supplementing laborato-alkali cement) is speci"ed by AASHTO-PP65 [62] for structures with ry experiments and "eld monitoring.

moderate ASR risk. A major challenge with this approach is that it ASR modeling can be characterized in terms of the method (numer-only considers the alkali contribution of portland cement and does ical or analytical), the scope (structural or material-level), and the not account for alkalis in aggregates, SCMs, and other admixtures. As physical assumptions underlying the model. Analytical models are discussed in Section 4 above, these non-cement alkalis could be signif- fast, but provide relatively crude estimations. Numerical and "nite icant and lead to deleterious ASR. However, it is unclear what fraction element models (e.g., Fig. 8) can be computationally expensive, but of the total alkalis in these non-cement sources is leachable and avail- can provide a more detailed understanding of the processes being sim-able for ASR. For structures with high ASR risk (e.g., infrastructure ulated. Mechanical or chemo-mechanical models can be formulated at components utilizing moderately to highly reactive aggregates), limit- the material or at the microstructure level, with only the former being ing concrete's alkali content should not be the sole mitigation strategy. practical for engineering applications and the latter being preferable

• Use of supplementary cementitious materials (SCM): These primarily for scienti"c investigation. Predictive approaches for kinetics require reduce the alkalinity (pH) of pore solution by OH consumption and modeling of diffusion and chemistry. As these are not well understood, alkali binding, consume portlandite, and reduce permeability and it is expected that future works will considerably improve these aspects.

mass transport of concrete, all due to the pozzolanic reaction [205, In the current state, physics-based models only exist at the microscale 206]. Using a suf"cient quantity of SCM is the most common method level for mechanical aspects, whereas all other models used in practice of ASR mitigation for new structures. SCMs with low CaO and low al- are largely empirical.

kali contents have been shown to be the most effective [205]. Howev- The analytical model from Bazant and Steffens [79] assume that the er, using SCMs is not without challenges: (a) high replacement levels kinetics of ASR is driven by the diffusion of the alkalis to the aggregates, of portland cement with SCM can negatively impact other concrete and that the reaction occurs at the aggregate surface. Homogenization properties such as setting, strength development and freeze-thaw/ techniques are then employed to derive the macroscopic swelling [79, scaling durability, especially in cold construction season; (b) the avail- 209]. This model does not however produce reasonable predictions of ability of high quality SCMs (e.g., low alkali class F "y ash) in some lo- the free expansion of the laboratory samples in most cases. Building cations has become uncertain as power plants shift to fuels other than on the work of Larive [210] and the assumption that the reaction occurs coal, and new environmental regulations have led to producing ashes that no longer meet concrete speci"cations (ASTM C618 [207]). This, in turn, puts more pressure on other SCM sources (e.g., blast furnace slag). There is a clear need for development and evaluation of non-conventional, yet economically viable SCM sources for ASR mitigation.

• Use of lithium admixtures: Laboratory and "eld studies, as summa-rized in Section 4.5 above, have documented the effectiveness of admixed lithium compounds (e.g., LiNO3) in mitigating ASR in new concrete; although, the mechanism of mitigation is not clearly under-stood. The main challenges with these admixtures are their relatively high cost and limited availability of lithium. Li is a scarce and strategic element, making up less than 0.002% (20 ppm) of the earth's crust.

New ASR inhibiting chemical admixtures that are cheaper and com-posed of more abundant ingredients should be sought.

In comparison with new structures, options for ASR mitigation in existing and already affected structures are much more limited. Lithium treatment (e.g., topical application, vacuum impregnation, electro-chemical impregnation) has had limited success due to the low penetra-tion depth of Li (10 to 20 mm) [208]. Electrochemical impregnation Fig. 8. ASR models can be formulated from different microscopic foundations, integrate (where surface applied Li+ is electrostatically attracted to negatively diffusion, and treat the microstructure explicitly or through homogenization.

142 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 at the aggregate surface, Multon and colleagues [211] have proposed data, especially in conditions more similar to concrete's interior; that the shape of the expansion curve versus time could be explained e.g., accounting for aggregate composition and heterogeneity, local partly by the need to "ll the interfacial transition zone pores before pore solution composition at silica surface, and effects of temperature, paste cracking can occur. The model they derived has been used to pressure/restraint, and humidity on the reaction parameters. This reproduce a number of experimental observations, notably linking the knowledge could be very powerful in developing more ef"cient ASR aggregate gradation and the effect of applied stress on the measured inhibiting admixtures, and ability to predict and optimize service life expansion [72,212]. This model has been applied to real structures by through computer models and laboratory and "eld experiments.

Grimal and co-workers [213,214]. However, more than 20 largely

• Aggregate reactivity depends not only on the type of silica mineral it empirical parameters are used in this model, and the microscopic foun- contains, but also on the size/distribution of these minerals within ag-dations do not match experimental observations (e.g., this model gregate structure. Reactive aggregate size and content have docu-assumes surface reaction of aggregates, which is not always the case) mented effects on ASR expansion, and often an intermediate particle

[215,216]. Microscopic approaches to modeling have tried to minimize size (i.e., the pessimum size) and intermediate content (i.e., the the number of free parameters by deriving the behavior of ASR from pessimum content) result in the highest expansion. Also, it is impor-

"rst principles. tant to note that contrary to a common belief, ASR does not always If the progress of the reaction is tracked as a function of the amount occur at the aggregate-cement paste interface, but may only occur of gel produced, the diffusion and chemical aspects become decoupled in the interior of some aggregates.

from their mechanical consequences. Ichikawa and Miura [217] have

• Internal alkali release from non-portland cement sources in concrete developed a chemical and thermodynamic model of the reaction, can be a signi"cant contributor to ASR in structures with long service which allowed them to predict the expansive pressure of the gel and lives and/or low cement alkali loadings. Standardized methods for its density. Using fracture-mechanical approaches, Reinhart and Mielich measuring the potential releasable alkalis from aggregates and SCMs

[218] analytically, as well as Dunant and Scrivener [219] using "nite are needed. These methods must include criteria for interpreting re-elements, showed that the shape of the expansion curve could be ex- sults in a manner relevant to "eld performance.

plained from "rst principles. The expansion is explained by the forma-

• LiNO3 has emerged as the preferred lithium compound for controlling tion of cracks in the aggregates and the paste. The cracks themselves ASR, as many other lithium salts increase the pH of pore solution.

are caused by the swelling pressure induced by ASR gel pockets. Dunant However, not a single prescribed Li dosage or [Li]/[Na + K] molar and Scrivener showed that gel pockets located through the aggregates ratio is effective against aggregates of different reactivity. Future re-were a necessary aspect of the models. The micro-mechanical approach search on the long-term ef"cacy of LiNO3, in better correlating labora-was further validated by modeling concrete under load and identifying tory and "eld data to determine an optimum Li dosage, and in better the contributions of individual aggregate classes [71,220]. This approach understanding of the mechanisms of ASR mitigation by Li is bene"cial.

to modeling ASR provided insights on the degradation process, which

• Presence of calcium is required for formation of ASR gel; otherwise the could not have been obtained using purely experimental approaches. dissolved silica could remain innocuously in the pore solution. In addi-However, it involves complex micro-mechanical modeling using a tion, Ca plays two speci"c roles with respect to ASR: (a) it can replace purpose-developed framework [221,222]. Recent developments by alkalis in ASR gel (i.e., recycle alkalis back into pore solution), and thus Giorla and colleagues [223,224] indicate that although the ultimate maintain a high pH in the pore solution, and (b) it impacts the prop-free expansion is not affected by the rate of the reaction, laboratory erties (e.g., viscosity, yield strength, swelling capacity) of ASR gel.

tests may overestimate the damage compared to the "eld where the

• There is evidence that presence of soluble Al in pore solution helps in slower rate of reaction allows creep to relax most of the damage in mitigating ASR. The exact mechanisms are unknown, but the effects the cement paste. Although microstructural simulations provide in- could be due to reducing silica dissolution, improving alkali binding sights on the validity of the models used, they cannot currently be and reducing the pH, consuming portlandite, reducing swelling of used to provide answers to engineering problems. the gel, or densifying the pore structure and reducing the permeability The future of ASR modeling will probably see a combination of micro of concrete.

and macro scales, where the macroscopic models are derived from the

• Limited qualitative information is available on the composition versus microscopic simulations. In particular, an open question in the study rheological and swelling properties of ASR gels. Research is needed of ASR is the anisotropy of the expansion under various loading to quantify the effect of gel composition, temperature, humidity, conditions, themselves frequently a product of the concrete expansion. and pressure/restraint on the gel's structure, moisture uptake, and Another axis of development is in predicting the ASR kinetics. Currently, swelling.

models assume an Arrhenius law to extrapolate from laboratory

• There is a strong need for synthesis, development, and assessment of experiments to the "eld. However, although this works well for the new ASR mitigation methods (e.g., new or unconventional SCMs, new expansion, it is likely to be wrong for the evolution of mechanical prop- chemical admixtures, and innovative treatment methods) applicable erties, particularly strength. A model integrating the research in ther- to new and, more importantly, to existing structures.

modynamics and chemistry with micro-mechanics is still forthcoming.

• Considerable progress has been made in understanding the micro-This is of special practical importance as the interaction of creep and mechanics of ASR. This has led to predictive models valid for samples ASR in structures is critical in planning for repairs. Finally, there are cur- of laboratory size. Much is still needed to integrate the chemical as-rently no published models which account for the effects of SCMs or pects of the reaction in these models. There are also a number of other ASR-inhibiting admixtures in concrete. well-established macroscopic models, used for the management of real structures. These are unfortunately largely empirical and need

10. Conclusions to be constantly amended to track the observed changes in the "eld.

In future, reactive-transport and micromechanical models should The following summarizes the most important conclusions of this serve as a base for formulating more robust macroscopic predictive paper: models.

• ASR is a result of a number of sequential reactions: dissolution of silica, formation of sol, formation of gel, and swelling of the gel. The under- Acknowledgments lying chemical reactions along with some thermodynamic informa-tion were presented in this paper. However, more research is The authors would like to thank Mr. Asghar Gholizadeh for his assis-needed to better quantify the thermodynamics and kinetics tance in reviewing the manuscript and providing insightful comments.

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 143 References [32] D.E. Clark, E.L. Yen-Bower, Corrosion of glass surfaces, Surf. Sci. 100 (1980) 53-70.

[33] V.S. Molchanov, N.E. Prikhidko, Corrosion of silicate glasses by alkaline solutions,

[1] A.B. Poole, Introduction to alkali-aggregate reaction in concrete, in: R.N. Swamy Russ. Chem. Bull. 6 (10) (1957) 1179-1184.

(Ed.), The Alkali Silica Reaction in Concrete, Van Nostrand Reinhold, New York, [34] H. Schloze, Chemical durability of glasses, J. Non-Cryst. Solids 52 (1-3) (1982)91-103.

1992. [35] S. Sjberg, Silica in aqueous environments, J. Non-Cryst. Solids 196 (1996) 51-57.

[2] D.W. Hobbs, Alkali-Silica Reaction in Concrete, Thomas Telford, London, 1988. [36] B.C. Bunker, Molecular mechanisms for corrosion of silica and silicate glasses, J.

[3] MERL Report-09-23, New Recommendations for ASR Mitigation in Reclamation Non-Cryst. Solids 179 (1994) 300-308.

Concrete Construction, U.S. Bureau of Reclamation, Denver, Colorado, 2009. [37] J.V. Walther, H.C. Helgeson, Calculation of the thermodynamic properties of

[4] F.M. Salles, J.C. Pínfari, S.C. Kuperman, C. Mizumoto, H. de Mayo Bernardes, Jaguari aqueous silica and the solubility of quartz and its polymorphs at high pressures hydropower plant evaluation, diagnosis and control of a structure affected by and temperatures, Am. J. Sci. 277 (10) (1977) 1315-1351.

alkali-silica reaction, 14th ICAAR, International Conference on Alkali Aggregate [38] ASTM C1285 02 (Reapproved 2008), Standard test Method for Determining Reactions, Austin, Texas, 2012. Chemical Durability of Nuclear, Hazardous, and Mixed Waste Glasses and Multi-

[5] G.A. Thompson, R.G. Charlwood, R.R. Steele, D.D. Curtis, Mactaquac generating phase Glass Ceramics: The Product Consistency Test (PCT), American Society for station intake and spillway remedial measures, Proc. 18th Int. Cong. on Large Testing and Materials, West Conshohocken, Pennsylvania, 2008.

Dams, Durban, South Africa, V.1 1994, pp. 342-368. [39] R.W. Douglas, T.M.M. El-shamy, Reaction of glasses with aqueous solutions, J. Am.

[6] L.J. Malvar, Guidelines for ASR prevention in DoD air"eld pavements, transporta- Ceram. Soc. 50 (1) (1967) 1-8.

tion.wes.army.mil/tsworkshop2012/submission%20by%20number/124/124% [40] P.M. Dove, D.A. Crerar, Kinetics of quartz dissolution in electrolyte solutions using a 20paper.pdf. hydrothermal mixed "ow reactor, Geochim. Cosmochim. Acta 54 (1990) 955-969.

[7] K. Torii, C. Sannoh, Y. Kubo, Y. Ohashi, Serious damages of ASR affected RC bridge [41] H. Maraghechi, Development and Assessment of Alkali-activated Recycled Glass-piers and their strengthening techniques, 12th ICAAR, International Conference based Concretes for Civil Infrastructure(Ph.D. Dissertation) The Pennsylvania on Alkali-Aggregate Reaction in Concrete, Beijing 2004, pp. 1283-1288. State University, 2014.

[8] B. Fournier, M.-A. Bérubé, K.J. Folliard, M. Thomas, Report on the diagnosis, prognosis, [42] J.P. Icenhower, P.M. Dove, The dissolution kinetics of amorphous silica into sodium and investigation of alkali-silica reaction (ASR) in transportation structures, FHWA chloride solutions: effects of temperature and ionic strength, Geochim.

Report# FHWA-HIF-09-004, Federal Highway Administration, Washington, D.C., 2010 Cosmochim. Acta 64 (2000) 4193-4203.

[9] L.W. Hixson, NextEra and NRC to continue monitoring ASR degradation at [43] R.D. Rimstidt, H.L. Barnes, The kinetics of silica-water reactions, Geochim.

Seabrook Station Nuclear Power Plant, Enformable Nuclear News2012. (April 24, Cosmochim. Acta 44 (1980) 1683-1699.

http://enformable.com/2012/04/nextera-and-nrc-to-continue-monitoring-asr- [44] F. Gaboriaud, A. Nonat, D. Chaumont, A. Craievich, Aggregation and gel formation degradation-at-seabrook-station-nuclear-power-plant/). in basic silico-calco-alkaline solutions studied: a SAXS, SANS, and ELS study, J.

[10] NRC Information Notice 2011-20: Concrete Degradation by Alkali-Silica Reaction, Phys. Chem. B 103 (1999) 5775-5781.

United States Nuclear Regulatory Commission, http://pbadupws.nrc.gov/docs/ [45] L.S. Dent Glasser, N. Kataoka, The chemistry of alkali-aggregate reaction, Cem.

ML1122/ML112241029.pdf. Concr. Res. 11 (1981) 1-9.

[11] L.A. Cooley, J.W. Brum"eld, ASR benchmarking workshop "nal report, www.fhwa. [46] F. Gaboriaud, D. Chaumont, A. Nonat, B. Hanquet, A. Craievich, Study of the dot.gov/pavement/pub_details.cfm?id=404. in"uence of alkaline ions (Li, Na and K) on the structure of the silicate entities in

[12] T. Miyagawa, K. Seto, K. Sasaki, Y. Mikata, K. Kuzume, T. Minam, Fracture of silico alkaline sol and on the formation of the silico-calco-alkaline gel, J. Sol-Gel reinforcing steels in concrete structures damaged by alkali-silica reaction: "eld Sci. Technol. 13 (1998) 353-358.

survey, mechanism and maintenance, J. Adv. Concr. Technol. 4 (3) (2006) 339-355. [47] G.W. Scherer, Structure and properties of gels, Cem. Concr. Res. 29 (1999)

[13] G.M. Ahlstrom, The United States Federal Highway Administration's alkali-silica 1149-1157.

reactivity development and deployment program, 14th ICAAR, International [48] X. Hou, R.J. Kirkpatrick, L.J. Struble, P.J.M. Monteiro, Structural investigations of Conference on Alkali Aggregate Reactions, Austin, TX, 2012. alkali silicate gels, J. Am. Ceram. Soc. 88 (4) (2005) 943-949.

[14] T.E. Stanton, Expansion of concrete through reaction between cement and [49] D.W. Hobbs, Alkali-silica reaction in concrete, in: J. Bensted, P. Barnes (Eds.),

aggregate, Proc. Am. Soc. Civ. Eng. 66 (10) (1940) 1781-1811. Structure and Performance of Cements, 2nd ed.Spon Press, London, 2002.

[15] ASTM C227 03, Standard Test Method for Potential Alkali Reactivity of Cement- [50] S. Diamond, S. Nishibayashi, M. Kawamura, ASR another look at mechanisms, Aggregate Combinations (Mortar-bar Method), American Society for Testing and Proc. 8th International Conf. on Alkali-Aggregate Reaction (ICAAR), Kyoto 1989, Materials, West Conshohocken, Pennsylvania, 2003. pp. 83-94.

[16] F.P. Glasser, Chemistry of alkali-aggregate reaction, in: R.N. Swamy (Ed.), The [51] R.F. Bleszynski, M.D.A. Thomas, Microstructural studies of alkali-silica reaction in Alkali Silica Reaction in Concrete, Van Nostrand Reinhold, New York, 1992. "y ash concrete immersed in alkaline solutions, Adv. Cem. Based Mater. 7 (2)

[17] T.C. Powers, H.H. Steinour, An interpretation of some published researches on the (1998) 66-78.

alkali-aggregate reaction; part 1 the chemical reactions and mechanisms of [52] A.M. Buckley, M. Greenblatt, The sol-gel preparation of silica gels, J. Chem. Educ. 71 expansion, J. Am. Concr. Inst. 26 (6) (1955) 497-516. (7) (1994) 599-602.

[18] S. Chatterji, The role of Ca(OH)2 in the breakdown of portland cement concrete due [53] A. Gholizadeh Vayghan, F. Rajabipour, The composition-rheology relationships in to alkali-silica reaction, Cem. Concr. Res. 9 (1979) 185-188. alkali-silica reaction gels and the impact on the gels' deleterious behavior, Cem.

[19] S. Diamond, A review of alkali-silica reaction and expansion mechanisms, 2. Concr. Res. (2015) (in preparation).

Reactive aggregates, Cem. Concr. Res. 6 (1976) 549-560. [54] M. Kunitz, Syneresis and swelling of gelatin, J. Gen. Physiol. 12 (2) (1928)

[20] ASTM C295 03, Standard Guide for Petrographic Examination of Aggregates for 289-312.

Concrete, American Society for Testing and Materials, West Conshohocken, [55] T. Ichikawa, M. Miura, Modi"ed model of alkali-silica reaction, Cem. Concr. Res. 37 Pennsylvania, 2003. (2007) 1291-1297.

[21] I. Sims, P. Nixon, RILEM recommended test method AAR-1: detection of potential [56] S. Chatterji, A.D. Jensen, N. Thaulow, P. Christensen, T. Denmark, Studies of alkali-alkali-reactivity of aggregates petrographic method, Mater. Struct. 36 (2003) silica reaction. Part 3. Mechanism by which NaCl and Ca (OH)2 affect the reaction, 480-496. Cem. Concr. Res. 16 (1986) 246-254.

[22] I. Sims, P. Nixon, RILEM recommended test method AAR-0: detection of alkali [57] M. Prezzi, P.J.M. Monteiro, G. Sposito, The alkali-silica reaction, part I: use of double reactivity potential in concrete outline guide to the use of RILEM methods in layer theory to explain the behavior of reaction-product gels, ACI Mater. J. 94 (1) assessments of aggregates for potential alkali-reactivity, Mater. Struct. 36 (2003) (1997) 10-16.

472-479. [58] F.A. Rodrigues, P.J. Monteiro, G. Sposito, The alkali-silica reaction: the surface

[23] M. Thomas, B. Fournier, K. Folliard, J. Ideker, M. Shehata, Test methods for evaluat- charge density of silica and its effect on expansive pressure, Cem. Concr. Res. 29 ing preventative measures for controlling expansion due to alkali-silica reaction in (4) (1999) 527-530.

concrete, Cem. Concr. Res. 36 (2006) 1842-1856. [59] E. Garcia-Diaz, J. Riche, D. Bulteel, C. Vernet, Mechanism of damage for the alkali-

[24] J. Lindgrd, . Andiç-Çakr, I. Fernandes, T.F. Rønning, M. Thomas, Alkali-silica re- silica reaction, Cem. Concr. Res. 36 (2) (2006) 395-400.

actions (ASR): literature review on parameters in"uencing laboratory performance [60] ASTM C1260 14, Standard Test Method for Potential Alkali Reactivity of testing, Cem. Concr. Res. 42 (2012) 223-243. Aggregates (Mortar-bar Method), American Society for Testing and Materials,

[25] M.D.A. Thomas, B. Fournier, K.J. Folliard, Report on determining the reactivity of West Conshohocken, Pennsylvania, 2014.

concrete aggregates and selecting appropriate measures for preventing deleterious [61] ASTM C1293 08b, Standard Test Method for Determination of Length Change of expansion in new concrete construction, Federal Highway Administration report Concrete Due to Alkali-Silica Reaction, American Society for Testing and Materials, FHWA-HIF-09-001, National Research Council, Washington, D.C., 2008 West Conshohocken, Pennsylvania, 2008.

[26] M.D.A. Thomas, B. Fournier, K.J. Folliard, Protocol for selecting ASR-affected [62] AASHTO PP 65-11, Determining the Reactivity of Concrete Aggregates and structures for lithium treatment, Federal Highway Administration Publications Selecting Appropriate Measures for Preventing Deleterious Expansion in New FHWA-HRT-04-113 (2004) and Techbrief FHWA-HRT-06-071, 2006. Concrete Construction, American Association of State Highway Transportation

[27] S.L. Brantley, J.D. Kubicki, A.F. White, Kinetics of Water-Rock Interaction, Springer, Of"cials, 2013.

New York, 2008. [63] CAN/CSA-A23.2-27A-09, Standard Practice to Identify Potential for Alkali-reactivity

[28] A.K. Varshneya, Fundamentals of Inorganic Glasses, 2nd ed. Academic Press Inc, of Aggregates and Measures to Avoid Deleterious Expansion in Concrete, Canadian San Diego, CA, 1994. Standards Association, 2009.

[29] J.E. Shelby, Introduction to glass science and technology, R. Soc. Chem. 2nd Ed. [64] F. Rajabipour, H. Maraghechi, G. Fischer, Investigating the alkali silica reaction of (2005) 72-110. recycled glass aggregates in concrete materials, ASCE J. Mater. Civ. Eng. 22

[30] R.K. Iler, The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface (2010) 1201-1208.

Properties, and Biochemistry, Wiley, New York, 1979. [65] H. Maraghechi, S. Shafaatian, G. Fischer, F. Rajabipour, The role of residual cracks on

[31] L.L. Hench, D.E. Clark, Physical chemistry of glass surfaces, J. Non-Cryst. Solids 28 alkali silica reactivity of recycled glass aggregates, Cem. Concr. Compos. 34 (2012)

(1978)83-105. 41-47.

144 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146

[66] T.R. West, Geology Applied to Engineering, Prentice Hall, Englewood Cliffs, New aggregate, 5th ICAAR, International Conference on Alkali Aggregate Reactions, Jersey, 1995. Cape Town, South Africa, 1981.

[67] P. Mehta, P.J.M. Monteiro, Concrete: Microstructure, Properties, and Materials, 3rd [101] Y. Wang, G. Yu, M. Deng, M. Tang, D. Lu, The use of thermodynamic analysis in ed. McGraw-Hill, New York, 2005. assessing alkali contribution by alkaline minerals in concrete, Cem. Concr. Compos.

[68] M.A.T.M. Broekmans, The Alkali-Silica Reaction: Mineralogical and Geochemical 30 (2008) 353-359.

Aspects of Some Dutch Concretes and Norwegian Mylonites(Ph.D. Dissertation) [102] O. Batic, P. Maiza, J. Sota, Alkali-silica reaction in basaltic rocks NBRI method, Cem.

Utrecht University, 2002. Concr. Res. 24 (1994) 1317-1326.

[69] H.E. Vivian, The effect on mortar expansion of particle size of the reactive [103] D. Constantiner, S. Diamond, Alkali release from feldspars into pore solution, Cem.

component in aggregate, Aust. J. Appl. Sci. 2 (1951) 488-492. Concr. Res. 33 (2003) 549-554.

[70] S. Poyet, A. Sellier, B. Capra, G. Foray, J.M. Torrenti, H. Cognon, E. Bourdarot, [104] E.R. Giannini, Evaluation of Concrete Structures Affected by Alkali-Silica Reaction Chemical modelling of alkali silica reaction: in"uence of the reactive aggregate and Delayed Ettringite FormationPh.D. Dissertation UT Austin, 2012.

size distribution, Mater. Struct. 40 (2) (2007) 229-239. [105] J.H. Ideker, K.J. Folliard, M.C.G. Juenger, M.D.A. Thomas, ASR in Texas, USA laborato-

[71] C.F. Dunant, K.L. Scrivener, Effects of aggregate size on alkali-silica reaction ry and "eld experience, in: M. Tang, M. Deng (Eds.),Proceedings of the 12th Inter-induced expansion, Cem. Concr. Res. 42 (6) (2012) 745-751. national Conference on Alkali-Aggregate Reaction in Concrete, Beijing, China 2004,

[72] S. Multon, N. Leklou, L. Petit, Coupled effects of aggregate size and alkali content on pp. 1062-1070 (October).

ASR expansion, Cem. Concr. Res. 38 (3) (2008) 350-359. [106] E.R. Giannini, A.F. Bentivegna, K.J. Folliard, Strain gradients in concrete affected by

[73] W. Jin, C. Meyer, S. Baxter, Glascrete concrete with glass aggregate, ACI Mater. J. alkali-silica reaction: a laboratory simulation, Adv. Civ. Eng. Mater. 3 (1) (2014) 97 (2) (2000) 208-213. 388-403.

[74] W.J. French, Avoiding concrete aggregate problems, Improving Civil Engineering [107] J. Chen, A.R. Jayapalan, J.-Y. Kim, K.E. Kurtis, L.J. Jacobs, Rapid evaluation of alkali-StructuresOld and New 1995, pp. 65-95. silica reactivity of aggregates using a nonlinear resonance spectroscopy technique,

[75] M.C.G. Juenger, C.P. Ostertag, Alkali-silica reactivity of large silica fume-derived Cem. Concr. Res. 40 (2010) 914-923.

particles, Cem. Concr. Res. 34 (2004) 1389-1402. [108] I. Sims, P. Nixon, B. Godart, Eliminating alkali-aggregate reaction from long-service

[76] A.J. Mass, J.H. Ideker, M.C.G. Juenger, Alkali silica reactivity of agglomerated silica structures, 14th ICAAR, International Conference on Alkali Aggregate Reactions, fume, Cem. Concr. Res. 37 (2007) 166-174. Austin, Texas, 2012.

[77] S. Diamond, N. Thaulow, A study of expansion due to alkali-silica reaction as [109] H. Chen, J.A. Soles, V.M. Malholtra, Investigations of supplementary cementing ma-conditioned by the grain size of the reactive aggregate, Cem. Concr. Res. 4 terials for reducing alkali-aggregate reactions, Cem. Concr. Compos. 15 (1993)

(1974) 591-607. 75-85.

[78] Y. Shao, T. Lefort, S. Moras, D. Rodriguez, Studies on concrete containing ground [110] S. Diamond, Effects of two Danish "y ashes on alkali contents of pore solutions of waste glass, Cem. Concr. Res. 30 (2000)91-100. cement-"y ash pastes, Cem. Concr. Res. 11 (3) (1981) 383-394.

[79] Z.P. Bazant, A. Steffens, Mathematical model for kinetics of alkali-silica reaction in [111] J.H. Ideker, T. Drimalas, A.F. Bentivegna, K.J. Folliard, B. Fournier, M.D.A. Thomas, concrete, Cem. Concr. Res. 30 (3) (1999) 419-428. R.D. Hooton, C.A. Rogers, The importance of outdoor exposure site testing, 14th

[80] D.W. Hobbs, W.A. Gutteridge, Particle size of aggregate and its in"uence upon the ICAAR, International Conference on Alkali Aggregate Reactions, Texas, Austin, 2012.

expansion caused by the alkali-silica reaction, Mag. Concr. Res. 31 (1979) [112] ASTM, C114-09, Standard Test Method for Chemical Analysis of Hydraulic Cement, 235-242. American Society for Testing and Materials, West Conshohocken, Pennsylvania,

[81] I. Moundoungou, D. Bulteel, E. Garcia-Diaz, V. Thiéry, P. Dégrugilliers, J.G. 2009.

Hammerschlag, Reduction of ASR expansion in concretes based on reactive chert [113] ASTM C311/C311M 13, Standard Test Method for Sampling and Testing Fly Ash aggregates: effect of alkali neutralisation capacity, Constr. Build. Mater. 54 or Natural Pozzolans for Use in Portland-cement Concrete, American Society for (2014) 147-162. Testing and Materials, West Conshohocken, Pennsylvania, 2013.

[82] E. Brouard, Potentially reactive aggregates with a pessimum effect: pessimum ef- [114] M.-A. Bérubé, J. Duschene, Relationships between portlandite depletion, available fect mechanisms, review of PRP quali"cation tests and conditions of use of these alkalis, and expansion of concrete made with mineral admixtures, 9th Internation-aggregates, Proceedings of the 14th International Conference on Alkali Aggregate al Conference on Alkali-Aggregate Reaction 1992, pp. 287-293.

Reactions (ICAAR), Austin, Texas, 2012. [115] M. Shehata, M.D.A. Thomas, Alkali release characteristics of blended cements, Cem.

[83] E. Garcia-Diaz, D. Bulteel, Y. Monnin, P. Degrugilliers, P. Fasseu, ASR pessimum be- Concr. Res. 36 (2006) 1166-1175.

haviour of siliceous limestone aggregates, Cem. Concr. Res. 40 (4) (2010) 546-549. [116] M. Shehata, M.D.A. Thomas, The effect of "y ash composition on the expansion of

[84] T. Drimalas, J.H. Ideker, G. Arrietta, K.J. Folliard, B. Fournier, M.D.A. Thomas, concrete due to alkali-silica reaction, Cem. Concr. Res. 30 (2000) 1063-1072.

Evaluating combinations of aggregates in the accelerated mortar bar test, Proc. [117] P.R. Rangaraju, J. Olek, Potential for acceleration of ASR in presence of pavement 14th Int. Conf. on Alkali-Aggregate Reactions in Concrete, Austin, Texas, 2012. deicing chemicals, IPRF (Innovative Pavement Research Foundation) Research, Re-

[85] R.P. Martin, C. Bazin, F. Toutlemonde, Alkali-aggregate reaction and delayed port. No. 01-G-002-03-92007.

ettringite formation: common features and differences, Proc. 14th Int. Conf. on [118] American Concrete Pavement Association (ACPA), Alkali-silica reaction: old issue, Alkali-Aggregate Reactions in Concrete, Austin, Texas, 2012. new cause? R&T Update, ACPA2005.

[86] F. Rajabipour, H. Maraghechi, S. Shafaatian, ASR and its mitigation in mortars con- [119] R. Joel, Accelerated alkali-silica reactivity in portland cement concrete pavements taining recycled soda-lime glass aggregates, Proceedings of the 14th International exposed to runway deicing chemicals, Engineering Brief No. 70, Federal Aviation Conference on Alkali Aggregate Reactions (ICAAR), Austin, Texas, 2012 (10 pp.). Administration, 2005.

[87] X. Hou, L.J. Struble, R.J. Kirkpatrick, Formation of ASR gel and the role of C-S-H and [120] S. Colwell, Local airport problems a bellwether? Colorado Springs Gazette2006. ().

portlandite, Cem. Concr. Res. 34 (2004) 1683-1696. [121] L. Kilzer and A. Gathright, Finding concrete problems at Denver International

[88] D.S. Lane, Alkali-Silica Reactivity in Virginia, Virginia Transportation Research Airport: DIA runways meant to last decades, but repairs cast shadow on defunct Council, Charlottesville, Virginia, USA, 1994. company, Rocky Mountain News, Jul. 15, 2006.

[89] J.J. Kollek, S.P. Varma, C. Zaris, Measurement of OH concentrations of pore "uids [122] C. Giebson, K. Seyfarth, J. Stark, In"uence of acetate and formate-based deicers on and expansion due to alkali-silica reaction in composite cement mortars, Proc. ASR in air"eld concrete pavements, Cem. Concr. Res. 40 (2010) 537-545.

8th Int. Cong. on the Chemistry of Cement, Rio de Janeiro, V.3 1986, pp. 183-189. [123] P.R. Rangaraju, J. Desai, Effectiveness of "y ash and slag in mitigating alkali-silica

[90] S. Diamond, Effects of microsilica (silica fume) on pore-solution chemistry of reaction induced by deicing chemicals, ASCE J. Mater. Civ. Eng. 21 (2009) 19-31.

cement pastes, Communications of the American Ceramic Society 1983, pp. 82-84. [124] S. Hayman, M.D.A. Thomas, T. Drimalas, K. Folliard, Deterioration of concretes

[91] S. Shafaatian, Innovative Methods to Mitigate Alkali-Silica Reaction in Concrete exposed to potassium acetate due to alkali-silica reaction, 14th International Materials Containing Recycled Glass Aggregates(Ph.D. Dissertation) Pennsylvania Conference on Alkali-Aggregate Reaction, Austin, USA, 2012.

State University, University Park, Pennsylvania, USA, 2012. [125] M.D.A. Thomas, S.A. Hayman, T. Drimalas, K.J. Folliard, Deterioration of cement

[92] R. Goguel, Alkali release by volcanic aggregates in concrete, Cem. Concr. Res. 25 paste exposed to potassium acetate solution, 13th International Conference of (1995) 841-852. Cement Chemistry. Madrid, Spain, 2011.

[93] E. Poulsen, T.S. Hansen, H.E. Sorensen, Release of alkalis from feldspar in concrete [126] J.G.M. Wood, Comparison of "eld performance with laboratory testing: how safe and mortar, ACI SP:192-49, American Concrete Institute 2000, pp. 807-824. and economic are current AAR speci"cations, in: M.-A. Bérurbé, B. Fournier, B.

[94] M.A. Bérubé, J. Duchesne, J.F. Dorion, M. Rivest, Laboratory assessment of alkali Durand (Eds.),Proceedings of the 11th International Conference on Alkali-contribution by aggregates to concrete and application to concrete structures Aggregate Reaction in Concrete. Quebéc City 2000, pp. 543-552.

affected by alkali-silica reactivity, Cem. Concr. Res. 32 (2002) 1215-1227. [127] M.-A. Bérubé, J. Frenette, Testing concrete for AAR in NaOH and NaCl solutions at

[95] D. Lu, X. Zhou, Z. Xu, X. Lan, M. Tang, B. Fournier, Evaluation of laboratory test 38 °C and 80 °C, Cem. Concr. Compos. 16 (1994) 189-198.

method for determining the potential alkali contribution from aggregate and the [128] C. Sannoh, K. Torii, Suppressing effect of various mineral admixtures on combined ASR safety of the Three-Gorges dam concrete, Cem. Concr. Res. 36 (2006) deterioration caused by ASR and chloride attack, in: M.A.T.M. Broekmans, B.J.

1157-1165. Wigum (Eds.),Proceedings of the 13th International Conference on Alkali-

[96] F. Locati, S. Mar"l, E. Baldo, Effect of ductile deformation of quartz-bearing rocks on Aggregate Reaction in Concrete, Trondheim, Norway, C. Sannoh, K. Torii, 2008.

the alkali-silica reaction, Eng. Geol. 116 (2010) 117-128. [129] P. Demars, P. Gilles, E. Dondonne, G. Lefebvre, A. Darimont, G. Lorenzi, G. Henriet, A.

[97] C.W. Parmelee, A.J. Monack, The solubility of feldspars in water, J. Am. Ceram. Soc. Marion, The degradation of the bridge deck slabs in Belgium mainly involves alka-13 (6) (1930) 386-392. li-aggregate reactions, in: M.A.T.M. Broekmans, B.J. Wigum (Eds.), Proceedings of

[98] J.H. Ideker, B.L. East, K.J. Folliard, B. Fournier, M.D.A. Thomas, The current state of the 13th International Conference on Alkali-Aggregate Reaction in Concrete, the accelerated concrete prism test, Cem. Concr. Res. 40 (2010) 550-555. Trondheim, Norway, 2008.

[99] P.E. Grattan-Bellew, Laboratory evaluation of alkali-silica reaction in concrete from [130] A. Shayan, A. Xu, G. Chirgwin, H. Morris, Effects of seawater on AAR expansion of saunders generating station, ACI Mater. J. 95 (1995) 126-134. concrete, Cem. Concr. Res. 40 (2010) 563-568.

[100] G.E. Blight, J.R. McIver, W.K. Schutte, R. Rimmer, The effects of alkali-aggregate [131] M.-A. Bérubé, J.-F. Dorion, Laboratory and "eld investigations of the in"uence of reaction on reinforced concrete structures made with witwatersrand quartzite sodium chloride on alkali-silica reactivity, in: M.A. Bérurbé, B. Fournier, B.

F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146 145 Durand (Eds.),Proceedings of the 11th International Conference on Alkali- [162] M.D.A. Thomas, The role of calcium hydroxide in alkali recycling in concrete, in: J.

Aggregate Reaction in Concrete. Québec City 2000, pp. 149-158. Skalny, J. Gebauer, I. Odler (Eds.), Materials Science of Concrete Special Volume on

[132] A. Shayan, Expansion of AAR-affected concrete under aggressive marine Calcium Hydroxide in Concrete, American Ceramic Society, Westerville, OH 2001, conditions: a look at possible effects of complex interactions, in: B. Fournier pp. 269-280.

(Ed.), Marc-André Bérubé Symposium on Alkali-Aggregate Reactivity in Concrete, [163] A. Leemann, G. Le Saout, F. Winnefeld, D. Rentsch, B. Lothenbach, Alkali-silica Montréal 2006, pp. 369-389. reaction: the in"uence of calcium on silica dissolution and the formation of

[133] A.K. Suravanshi, J.D. Scantlebury, S.B. Lyon, Mechanism of Friedel's salt formation in reaction products, J. Am. Ceram. Soc. 94 (4) (2011) 1243-1249.

cements rich in tri-calcium aluminate, Cem. Concr. Res. 26 (1996) 717-727. [164] M.D.A. Thomas, The role of calcium in alkali-silica reaction, in: M. Cohen, S.

[134] Q. Yuan, C. Shi, G. De Shutter, K. Audenaert, D. Deng, Chloride binding of cement- Mindess, J.P. Skalny (Eds.), Materials Science of Concrete The Sidney Dia-based materials subjected to external chloride environment a review, Constr. mond Symposium, American Ceramic Society, Westerville, OH 1998, Build. Mater. 23 (2009) 1-13. pp. 325-331.

[135] W.J. McCoy, A.G. Caldwell, New approach to inhibiting alkali-aggregate expansion, [165] S. Chatterji, N.F. Clausson-Kass, Prevention of alkali-silica expansion by using slag-ACI J. Proc. 47 (5) (1951) 693-706. portland cement, Cem. Concr. Res. 14 (1984) 816-818.

[136] Y. Sakaguchi, M. Takakura, A. Kitagawa, H. Takahiro, T. Fuminori, The inhibiting [166] M.D.A. Thomas, P.J. Nixon, K. Pettifer, The effect of PFA on alkali-silica reaction, in:

effect of lithium compounds on alkali-silica reaction, Proceeding of the 8th V.M. Malhotra (Ed.), , 2nd CANMET/ACI Conference on the Durability of Concrete, International Conference on Alkali-Aggregate Reaction, Kyoto 1989, pp. 229-234. ACI SP-126American Concrete Institute, Detroit 1991, pp. 919-940.

[137] S. Chatterji, N. Thaulow, A.D. Jensen, Studies of alkali-silica reaction. Part 4. Effect of [167] M. Kawamura, O.A. Kayyali, M.N. Haque, Effects of a "y ash on pore solution different alkali salt solutions on expansion, Cem. Concr. Res. 17 (1987) 777-783 composition in calcium and sodium chloride-bearing mortars, Cem. Concr. Res.

(repeated as 122). 18 (1988) 763-773.

[138] D.C. Stark, Lithium salt admixturesan alternative method to prevent excessive [168] M.M. Alasali, V.M. Malhotra, Role of concrete incorporating high volumes of "y ash alkali-silica reactivity, 9th ICAAR, International Conference on Alkali Aggregate in controlling expansion due to alkali aggregate reactions, ACI Mater. J. 88 (1991)

Reactions, London, 1992. 159-163 (Discussion in Vol. 89, (1992), 110-111).

[139] J.S. Lumley, ASR suppression by lithium compounds, Cem. Concr. Res. 27 (2) [169] M. Tang, Y.F. Ye, M.Q. Yuan, S.H. Zheng, The preventive effect of mineral admix-(1997) 235-244. tures on alkali-silica reaction and its mechanism, Cem. Concr. Res. 13 (1983)

[140] V.S. Ramachandran, Alkali-aggregate expansion inhibiting admixtures, Cem. 171-176.

Concr. Compos. 20 (2) (1998) 149-161. [170] H. Wang, J.E. Gillott, Mechanism of alkali-silica reaction and the signi"cance of

[141] S. Diamond, Unique response of LiNO3 as an alkali silica reaction-preventive calcium hydroxide, Cem. Concr. Res. 21 (1991) 647-654.

admixture, Cem. Concr. Res. 29 (8) (1999) 1271-1275. [171] S. Urhan, Alkali silica and pozzolanic reactions in concrete. Part 1: interpretation of

[142] M.D.A. Thomas, R. Hooper, D. Stokes, Use of lithium-containing compounds to published results and an hypothesis concerning the mechanism, Cem. Concr. Res.

control expansion in concrete due to alkali-silica reaction, Proceedings of the 17 (1987) 141-152.

11th International Conference on Alkali-Aggregate Reaction (ICAAR), Quebec [172] L.J. Struble, S. Diamond, Swelling properties of synthetic alkali silica gels, J. Am.

City 2000, pp. 11-16. Ceram. Soc. 64 (11) (1981) 652-655.

[143] C.L. Collins, J.H. Ideker, G.S. Willis, K.E. Kurtis, Examination of the effects of [173] T. Ramlochan, M.D.A. Thomas, R.D. Hooton, The effect of pozzolans and slag on the LiOH, LiCl, and LiNO3 on alkali-silica reaction, Cem. Concr. Res. 34 (8) (2004) expansion of mortars cured at elevated temperature; part II: microstructural and 1403-1415. microchemical investigations, Cem. Concr. Res. 34 (2004) 1341-1356.

[144] C. Tremblay, M.A. Bérubé, B. Fournier, M.D.A. Thomas, Performance of lithium- [174] J. Duchesne, M.-A. Berube, Long-term effectiveness of supplementary cementing based products against ASR: effect of aggregate type and reactivity, and reaction materials against alkali-silica reaction, Cem. Concr. Res. 31 (2001) 1057-1063.

mechanisms, Proc. 7th CANMET/ACI Int. Conf. on Recent Advances in Concrete [175] W. Aquino, D.A. Lange, J. Olek, The in"uence of metakaolin and silica fume on the Technology 2004 (Suppl. Papers), pp. 247-267. chemistry of alkali-silica reaction products, Cem. Concr. Compos. 23 (2001)

[145] X. Feng, M.D.A. Thomas, T.W. Bremner, K.J. Folliard, B. Fournier, New observations 485-493.

on the mechanism of lithium nitrate against alkali silica reaction (ASR), Cem. [176] K. Schwing (m. Schumacher), Use of Fly Ash in the Mitigation of Alkali-Silica Concr. Res. 40 (1) (2010)94-101. Reaction in ConcreteMaster's Thesis Oregon State University, 2011.

[146] X. Feng, M.D.A. Thomas, T.W. Bremner, B.J. Balcom, K.J. Folliard, Studies on lithium [177] K. Schumacher, J.H. Ideker, New considerations in predicting mitigation of alkali-salts to mitigate ASR-induced expansion in new concrete: a critical review, Cem. silica reaction based on "y ash chemistry, ASCE J. Mater. Civ. Eng. 27 (4) (2015).

Concr. Res. 35 (9) (2005) 1789-1796. [178] S. Warner, The Role of Alumina in the Mitigation of Alkali-Silica ReactionMaster's

[147] K.J. Folliard, M.D.A. Thomas, B. Fournier, K. Kurtis, J.H. Ideker, Interim recommenda- Thesis Oregon State University, 2012.

tions for the use of lithium to mitigate or prevent alkali-silica reaction (ASR), FHWA [179] S. Warner, J.H. Ideker, K. Schumacher, Alkali-silica reactivity and the role of Report No. FHWA-RD-06-073, Federal Highway Administration, McLean, Virginia, alumina, Proceedings of the 14th International Conference on Alkali-Aggregate 2006. Reactivity in Concrete (ICAAR), Austin, Texas, 2012 (10 pp.).

[148] X. Feng, M.D.A. Thomas, T.W. Bremner, K.J. Folliard, B. Fournier, Summary of [180] R.K. Iler, Effect of adsorbed alumina on the solubility of amorphous silica in water, J.

research on the effect of LiNO3 on alkali-silica reaction in new concrete, Cem. Colloid Interface Sci. 43 (2) (1973) 399-408.

Concr. Res. 40 (4) (2010) 636-642. [181] K.-J. Huenger, The contribution of quartz and the role of aluminum for understand-

[149] M. Kawamura, H. Fuwa, Effects of lithium salts on ASR gel composition and expan- ing the AAR with greywacke, Cem. Concr. Res. 37 (2007) 1193-1205.

sion of mortars, Cem. Concr. Res. 33 (6) (2003) 913-919. [182] T. Chappex, K.L. Scrivener, The effect of aluminum in solution on the dissolution of

[150] X. Mo, Laboratory study of LiOH in inhibiting alkali-silica reaction at 20 °C: a con- amorphous silica and its relation to cementitious systems, J. Am. Ceram. Soc. 96 tribution, Cem. Concr. Res. 35 (3) (2005) 499-504. (2013) 592-597.

[151] X. Mo, C. Yu, Z. Xu, Long-term effectiveness and mechanism of LiOH in inhibiting [183] S.Y. Hong, F.P. Glasser, Alkali sorption by C-S-H and C-A-S-H gels: part II, role of alkali-silica reaction, Cem. Concr. Res. 33 (1) (2003) 115-119. alumina, Cem. Concr. Res. 32 (7) (2002) 1101-1111.

[152] D. Bulteel, E. Garcia-Diaz, P. Dégrugilliers, In"uence of lithium hydroxide on alkali- [184] T. Chappex, K.L. Scrivener, Alkali "xation of C-S-H in blended cement pastes and silica reaction, Cem. Concr. Res. 40 (4) (2010) 526-530. its relation to alkali silica reaction, Cem. Concr. Res. 42 (2012) 1049-1054.

[153] C. Tremblay, M.A. Bérubé, B. Fournier, M.D.A. Thomas, K.J. Folliard, Effectiveness of [185] J.C. Sang, R.F. Jakubic, A. Barkatt, E.E. Saad, The interaction of solutes with sili-lithium-based products in concrete made with Canadian natural aggregates sus- cate glass and its effect on dissolution rates, J. Non-Cryst. Solids 167 (1994) ceptible to alkali-silica reactivity, ACI Mater. J. 104 (2) (2007) 195-205. 158-171.

[154] T. Drimalas, J.H. Ideker, A.F. Bentivegna, K.J. Folliard, B. Fournier, M.D.A. Thomas, [186] B.R. Bickmore, K.L. Nagy, A.K. Gray, A.R. Brinkerhoff, The effect of Al(OH)4 on the The long-term monitoring of large-scale concrete specimens containing lithium dissolution rate of quartz, Geochim. Cosmochim. Acta 70 (2006) 290-305.

salts to mitigate alkali-silica reaction, ACI Spec. Publ. SP289-18 (2012) 1-17. [187] S. Komarneni, R. Roy, D. Roy, C. Fyfe, G. Kennedy, A. Bothner-By, J. Dadok, A.

[155] C. Tremblay, M.A. Bérubé, B. Fournier, M.D. Thomas, K.J. Folliard, Experimental Chesnick, 27Al and 29Si magic angle spinning nuclear magnetic resonance spectros-investigation of the mechanisms by which LiNO3 is effective against ASR, Cem. copy of Al-substituted tobermorites, J. Mater. Sci. 20 (1985) 4209-4214.

Concr. Res. 40 (4) (2010) 583-597. [188] S. Komarneni, D.M. Roy, D.M., C.A. Fyfe, G.J. Kennedy, Naturally occurring 1.4 nm

[156] J. Duschene, M.-A. Bérubé, Evaluation of the validity of the pore solution expression tobermorite and synthetic jennite: characterization by 27Al and 29Si MASNMR method from hardened cement pastes and mortars, Cem. Concr. Res. 24 (3) (1994) spectroscopy and cation exchange properties, Cem. Concr. Res. 17 (1987) 891-895.

456-462. [189] S. Komarneni, M. Tsuji, Selective cation exchange in substituted tobermorite, J. Am.

[157] D. Constantiner, S. Diamond, Pore solution analysis: are there pressure effects? in: Ceram. Soc. 72 (1989) 1668-1674.

K.L. Scrivener, J.F. Young (Eds.), Mechanisms of Chemical Degradation of Cement- [190] H. Stade, D. Müller, On the coordination of Al in ill-crystallized C-S-H phases based Systems, E & FN Spon, London 1997, pp. 22-29. formed by hydration of tricalcium silicate and by precipitation reactions at

[158] M.-A. Bérubé, C. Tremblay, Chemistry of pore solution expressed under high ambient temperature, Cem. Concr. Res. 17 (1987) 553-561.

pressure: In"uence of various parameters and comparison with the hot-water [191] R.J. Kirkpatrick, X.D. Cong, An introduction to 27Al and 29Si NMR spectroscopy extraction method, 12th ICAAR, International Conference on Alkali Aggregate of cements and concretes, in: P. Colombet, A.R. Grimmer (Eds.), Application of Reactions, Beijing 2004, pp. 833-842. NMR Spectroscopy to Cement Science, Gordon and Breach, Amsterdam, The

[159] M. Cyr, P. Rivard, F. Labrecque, A. Daidié, High-pressure device for "uid extraction Netherlands 1994, pp. 55-77.

from porous materials: application to cement-based materials, J. Am. Ceram. Soc. [192] I. Lognot, I. Klur, A. Nonat, NMR and infrared spectroscopies of C-S-H and previous 91 (8) (2008) 2653-2658. Al substituted C-S-H synthesized in alkaline solution, in: P. Colombet, A.R.

[160] W.C. Hansen, Studies relating to the mechanism by which the alkali-silica reaction Grimmer, H. Zanni, P. Sozzani (Eds.), Nuclear Magnetic Resonance Spectroscopy proceeds in concrete, J. Am. Concr. Inst. 15 (1944) 213-227. of Cement Based Materials, Springer-Verlag, Berlin, Germany, 1998.

[161] T. Knudsen, N. Thaulow, Quantitative microanalyses of alkali-silica gel in concrete, [193] G.K. Sun, F.J. Young, R.J. Kirkpatrick, The role of Al in C-S-H: NMR, XRD, and Cem. Concr. Res. 5 (1979) 443-454. compositional results for precipitated samples, Cem. Concr. Res. 36 (2006) 18-29.

146 F. Rajabipour et al. / Cement and Concrete Research 76 (2015) 130-146

[194] T. Chappex, K.L. Scrivener, The in"uence of aluminum on the dissolution of [210] C. Larive, Apports Combinés de L'expérimentation Et de La Modélisation La amorphous silica and its relation to alkali silica reaction, Cem. Concr. Res. 42 Compréhension de L'alcali-Réaction Et de Ses Effets MécaniquesPhD Dissertation (2012) 1645-1649. Ecole Nationale des Ponts et Chaussés, Paris, 1998.

[195] H. Olafsson, The effect of relative humidity and temperature on alkali expansion of [211] S. Multon, A. Sellier, M. Cyr, Chemo-mechanical modeling for prediction of alkali mortar bars, 7th ICAAR, International Conference on Alkali Aggregate Reactions, silica reaction (ASR) expansion, Cem. Concr. Res. 39 (6) (2009) 490-500.

Ottawa 1986, pp. 461-465. [212] S. Multon, F. Toutlemonde, Effect of applied stresses on alkali-silica reaction-

[196] T. Kurihara, K. Katawaki, Effects of moisture control and inhibition on alkali silica induced expansions, Cem. Concr. Res. 36 (5) (2006) 912-920.

reaction, 8th ICAAR, International Conference on Alkali Aggregate Reactions, [213] E. Grimal, A. Sellier, Y. Le Pape, E. Bourdarot, Creep, shrinkage, and anisotropic Kyoto 1989, pp. 629-634. damage in alkali-aggregate reaction swelling mechanism; part 1: a constitutive

[197] D. Stark, The moisture condition of "eld concrete exhibiting alkali-silica reactivity, model, ACI Mater. J. 105 (3) (2008) 227-235.

ACI Spec. Publ., SP126-52 1991, pp. 973-988. [214] E. Grimal, A. Sellier, S. Multon, Y.L. Pape, E. Bourdarot, Concrete modeling for

[198] S. Multon, F. Toutlemonde, Effect of moisture conditions and transfers on alkali expertise of structures affected by alkali aggregate reaction, Cem. Concr. Res. 40 silica reaction damaged structures, Cem. Concr. Res. 40 (2010) 924-934. (4) (2010) 502-507.

[199] C. Larive, A. Laplaud, O. Coussy, The role of water in alkali-silica reaction, 11th [215] J.M. Ponce, O.R. Batic, Different manifestations of the alkali-silica reaction in ICAAR, International Conference on Alkali Aggregate Reactions, Quebec, 2000. concrete according to the reaction kinetics of the reactive aggregate, Cem. Concr.

[200] D.G.R. Bonnell, Studies in gels, IV: the swelling of silica gel, Trans. Faraday Soc. 28 Res. 36 (6) (2006) 1148-1156.

(1) (1932) 1217-1220. [216] M. Ben Haha, E. Gallucci, A. Guidoum, K.L. Scrivener, Relation of expansion due to

[201] S. Poyet, A. Sellier, B. Capra, G. Thevenin-Foray, J.-M. Torrenti, H. Tournier-Cognon, alkali silica reaction to the degree of reaction measured by sem image analysis, E. Bourdarot, In"uence of water on alkali-silica reaction: experimental study and Cem. Concr. Res. 37 (8) (2007) 1206-1214.

numerical simulations, ASCE J. Mater. 18 (4) (2006) 588-596. [217] T. Ichikawa, M. Miura, Modi"ed model of alkali-silica reaction, Cem. Concr. Res. 37

[202] R.A. Helmuth, D.H. Turk, The reversible and irreversible drying shrinkage of (9) (2007) 1291-1297.

hardened Portland cement and tricalcium silicate pastes, PCA Bull., 215, Portland [218] H.W. Reinhardt, O. Mielich, A fracture mechanics approach to the crack formation Cement Association, Skokie, Illinois, 1970. in alkali-sensitive grains, Cem. Concr. Res. 41 (3) (2010) 255-262.

[203] W. Hansen, Drying shrinkage mechanisms in portland cement paste, J. Am. Ceram. [219] C.F. Dunant, K.L. Scrivener, Micro-mechanical modeling of alkali-silica reaction-Soc. 70 (5) (1987) 323-328. induced degradation using the AMIE framework, Cem. Concr. Res. 40 (4) (2010)

[204] B. Fournier, M.-A. Bérubé, Alkali-aggregate reaction in concrete: a review of basic 517-525.

concepts and engineering implications, Can. J. Civ. Eng. 27 (2000) 167-191. [220] C.F. Dunant, K.L. Scrivener, Effects of uniaxial stress on alkali-silica reaction

[205] M.D.A. Thomas, The effect of supplementary cementing materials on alkali-silica induced expansion of concrete, Cem. Concr. Res. 42 (3) (2012) 567-576.

reaction: a review, Cem. Concr. Res. 41 (2011) 209-216. [221] C. Dunant, P.N. Vinh, M. Belgasmia, S. Bordas, A. Guidoum, Architecture tradeoffs of

[206] S. Shafaatian, A. Akhavan, H. Maraghechi, F. Rajabipour, How does "y ash mitigate integrating a mesh generator to partition of unity enriched object-oriented "nite alkali-silica reaction (ASR) in accelerated mortar bar test (ASTM C1567)? Cem. element software, Revue Européenne de Mécanique Numérique 16 (2007)

Concr. Compos. 37 (2013) 143-153. 237-258 (March).

[207] ASTM C618 12a, Standard Speci"cations for Coal Fly Ash and Raw or Calcined [222] C.F. Dunant, P. Kerfriden, S. Bordas, K.L. Scrivener, T. Rabczuck, An algorithm to Natural Pozzolan for Use in Concrete, American Society for Testing and Materials, compute damage from load in composites, Front. Arch. Civ. Eng. China 5 (2)

West Conshohocken, Pennsylvania, 2012. (2011) 180-193.

[208] M.D.A. Thomas, B. Fournier, K.J. Folliard, J.H. Ideker, Y. Resendez, M.D.A. Thomas, B. [223] A.B. Giorla, Modelling of Alkali-Silica Reaction under Multi-axial LoadPhD Fournier, K.J. Folliard, J.H. Ideker, Y. Resendez, The use of lithium to prevent or mit- Dissertation EPFL, 2013.

igate alkali-silica reaction in concrete pavements and structures, FHWA Report# [224] A.B. Giorla, K.L. Scrivener, C.F. Dunant, In"uence of visco-elasticity on the stress FHWA-HRT-06-133, Federal Highway Administration, Washington, D.C., 2007 development induced by alkali-silica reaction, Cem. Concr. Res. 70 (2015) 1-8.

[209] G.M. Idorn, A discussion of the paper mathematical model for kinetics of alkali-silica reaction in concrete, in: Z.P. Bazant (Ed.),Cem. Concr. Res., 31(7) 2001, pp. 1109-1110.