effectiveness of sealers in counteracting alkali–silica reaction in plain and air-entrained...

Effectiveness of sealers in counteracting alkali–silica reaction in plain and air-entrained laboratoryconcretes exposed to wetting and drying, freezingand thawing, and salt water

Marc-André Bérubé, Dominique Chouinard, Michel Pigeon, Jean Frenette,Luc Boisvert, and Michel Rivest

Abstract: Low- and high-alkali, plain and air-entrained large concrete cylinders, 255 mm in diameter by 310 mm inlength, were made with a highly alkali–silica reactive limestone. After curing, a number of cylinders were sealed withsilane, oligosiloxane, polysiloxane, linseed oil, or epoxy, with others subjected to 179 freezing and thawing cycles inhumid air (one cycle per day). All cylinders were then subjected to 14-day exposure cycles, including in the most se-vere case periods of humid storage in air, drying, wetting in salt water, and freezing and thawing cycles. All low-alkalispecimens did not either expand or develop surface cracking, even those with a deficient air void system and exposedto freezing and thawing cycles. All unsealed high-alkali cylinders subjected early to a series of freezing and thawingcycles did not significantly expand during these cycles, but presented high expansion afterwards. Wetting and dryingsignificantly reduced alkali–silica reaction (ASR) expansion compared with constant humid storage; however, it pro-moted map-cracking. Regardless of the air content, freezing and thawing increased greatly the concrete expansion inthe presence of ASR, even after ASR was almost complete; freezing and thawing also greatly promoted surface crack-ing. On the other hand, all cylinders early sealed with silane, oligosilixane, or polysiloxane did not either significantlyexpand or show map-cracking, whatever the exposure conditions and the air content; these cylinders progressively lostmass with time. On the other hand, the epoxy resin was not effective. The linseed oil prevented map-cracking whilesignificantly reducing expansion, however not sufficiently. After one or 1.5 years, some expanding cylinders weresealed with silane, oligosiloxane, or polysiloxane; they started to loose mass and contracted immediately after beingsealed, whatever the exposure conditions. The results obtained thus indicate that a good sealer may greatly improve theaesthetic appearance (e.g., map-cracking) and stop expansion of ASR-affected concrete elements of 255 mm or less inthickness, made with a water-to-cement ratio in the range of 0.50, and exposed to wetting and drying, freezing andthawing, and salt water.

300Key words: air entrained, alkali–silica reaction, concrete, cracking, expansion, freezing and thawing, sealer, silane,siloxane, wetting and drying.

Résumé : De gros cylindres de béton de 255 mm de diamètre par 310 mm de longueur, de faible et haute teneurs enalcalis, ont été préparés avec un calcaire fortement affecté de réactivité alcalis-silice (RAS), avec et sans agent entraî-neur d’air. Après mûrissement, un certain nombre de cylindres ont été scellés avec du silane, de l’oligosiloxane, du po-lysiloxane, de l’huile de lin ou de la résine époxyde, pendant que d’autres cylindres ont été soumis à 179 cycles de gelet de dégel dans l’air humide (un cycle par jour). Tous les cylindres ont ensuite été soumis à des cycles d’expositionde 14 jours incluant dans le plus sévère des cas un séjour à l’air humide, une période de séchage, un trempage dansl’eau salée et des cycles de gel et de dégel. Toutes les éprouvettes de faible teneur en alcalis n’ont pas présentéd’expansion ni de fissuration superficielle, même celles pourvues d’un réseau d’air déficient et exposées à des cyclesde gel et de dégel. Tous les cylindres à forte teneur en alcalis non-scellés et exposés tôt à une série de cycles de gel etde dégel n’ont pas présenté d’expansion significative durant ces cycles, mais ont pris beaucoup d’expansion par la suite.Les cylindres exposés à des périodes de trempage et de séchage ont présenté significativement moins d’expansion associée

Can. J. Civ. Eng. 29: 289–300 (2002) DOI: 10.1139/L02-011 © 2002 NRC Canada

289

Received 29 May 2001. Revised manuscript accepted 6 February 2002. Published on the NRC Research Press Web site athttp://cjce.nrc.ca on 9 April 2002.

M.-A. Bérubé1 and J. Frenette. Département de géologie et de génie géologique, Université Laval, Sainte-Foy, QC G1K 7P4,Canada.D. Chouinard. Laboratoire de Béton, 3420, boulevard St-Joseph est, Montréal, QC H1X 1W6, Canada.M. Pigeon and L. Boisvert. Département de génie civil, Université Laval, Sainte-Foy, QC G1K 7P4, Canada.M. Rivest. Direction Ingénierie, Hydro Québec, 855, rue Ste-Catherine est, 11e étage, Montréal, QC H2L 4P5, Canada.

Written discussion of this article is welcomed and will be received by the Editor until 31 August 2002.

1Corresponding author (e-mail: [email protected]).

I:\cjce\cjce_29\cjce-02\L02-011.vpFriday, April 05, 2002 9:09:56 AM

Color profile: Generic CMYK printer profileComposite Default screen

à la RAS que les cylindres conservés constamment à l’air humide; cependant, ils ont développé une plus importante fis-suration superficielle. Quel que soit le contenu en air, les cycles de gel et de dégel ont grandement contribué à aug-menter l’expansion du béton affecté de RAS, même après que cette réaction soit presque complétée; les cycles de gelet de dégel ont aussi favorisé la fissuration superficielle du béton en cause. Par ailleurs, tous les cylindres scellés tôtavec du silane, de l’oligosiloxane ou du polysiloxane n’ont pas présenté d’expansion significative ni de fissuration su-perficielle, et ce quels que soient les conditions d’exposition et le contenu en air; ces cylindres ont progressivementperdu du poids avec le temps. Pour sa part, la résine époxyde ne s’est pas avérée efficace pendant que l’huile de lin aéliminé la fissuration superficielle et considérablement réduit l’expansion due à la RAS, mais pas suffisamment. Aprèsun an ou un an et demi, certains cylindres expansifs ont été scellés avec du silane, de l’oligosiloxane ou du polysi-loxane; ils ont alors immédiatement commencé à perdre du poids et à se contracter, et ce quelles que soient les condi-tions d’exposition. Les résultats obtenus indiquent donc qu’un bon scellant peut grandement améliorer l’aspectesthétique (e.g. fissuration polygonale) et arrêter l’expansion due à la RAS dans le cas d’éléments de béton d’épaisseurde 255 mm ou moins, fabriqués avec un rapport eau/ciment de l’ordre de 0,50 et exposés à des cycles de mouillage etde séchage, à des cycles de gel et de dégel et à de l’eau salée.

Mots clés : air occlus, réaction alcalis–granulats, béton, fissuration, gonflement, gel et dégel, scellant, silane, siloxane,mouillage et séchage.

[Traduit par la Rédaction] Bérubé et al. 300

Introduction

Alkali–silica reaction (ASR) results in internal micro-cracking and expansion of affected concrete members, dif-ferential movements, deformations, concrete spalling,extrusion of sealers along joints, and surface cracking pat-terns. However, the development of some of the above exter-nal problems can be greatly affected by the exposureconditions such as wetting and drying, freezing and thawing,and exposure to sea water or deicing salt.

Wetting and dryingWetting and drying cycles are thought to play an impor-

tant role in the development of the surface cracking patterns,such that the magnitude of this defect may not correlate withinternal deterioration (i.e., microcracking and expansion)(Nishibayashi et al. 1989). In fact, wetting and drying leadsto conditions that are relatively less conducive to alkali–sil-ica reactions (less moisture in dry weather, leaching of alka-lis by rain), in the first few centimetres of concrete than inthe middle of the concrete mass. The surface concrete thuscracks from tension under the expansive thrust of the under-lying concrete. This explains why (i) cracks observed on thesurface of ASR-affected concrete members rarely penetratemore than a few centimetres inside the concrete; (ii) im-mersed or underground parts of ASR-affected concretemembers usually deteriorate less at their surface (highermoisture) than air-exposed sections, even though signs ofASR are observed in the entire concrete mass (Bérubé et al.1989); and (iii) concrete surfaces exposed to the south (moresun and more drying) usually present more map-crackingthan the sections exposed to the north (Ludwig 1989).

Freezing and thawingFreezing and thawing cycles accelerate the deterioration

of concrete once cracking has been started by ASR. Con-versely, ASR or related deterioration can be accelerated oncethe concrete has been cracked by freezing, because moisturecan more easily penetrate into the concrete, and because theconcrete is weaker and less able to withstand the expansiveforces generated by ASR. This has been experienced by the

authors at the Québec City airport, in concrete slabs from alanding runway made with a reactive limestone. Internalsigns of ASR (microcracks in aggregate particles, siliceousgel, dark reaction rims) were observed in the overall con-crete from these slabs; however, map-cracking was only ob-served at the surface of sections in which air voids wereinsufficient (as a result of excessive local vibration after theconcrete has been placed) and in which concrete had shownmicrocracking as a result of freezing.

Sea water and deicing saltSea water and deicing salt can provide the near-surface

concrete with additional alkalis. According to some authors,this might initiate alkali reactions or at least speed them up.However, in most cases, sodium chloride does not penetrateto significant depth inside the concrete and therefore cannotinitiate or accelerate ASR in the core of relatively thick con-crete members. On the other hand, map-cracking is oftenmore important on concrete surfaces of ASR-affected com-ponents that are exposed to salt solution. As discussed else-where (Duchesne and Bérubé 1996; Bérubé et al. 2000), thisis rather explained by lower reaction and expansion in thenear-surface concrete; as a result of pH decrease in the poresolution, the corresponding concrete develops tensioncracks. This also suggests, in turn, that low-alkali and rela-tively thick concrete members are sufficiently protectedagainst ASR, even when exposed to salt solutions (Bérubé etal. 2000). However, such solutions can produce concretescaling and can move along preexisting cracks andmicrocracks generated by other mechanisms, so causingspalling, corrosion of reinforcement, and rusting.

Effectiveness of sealers against alkali–silica reactionHowever, the application of a good sealer on concrete

may reduce the influence of all exposure conditions, thus re-ducing the surface deterioration of concrete. Even more, assuggested in many papers presented at the last four Interna-tional Conferences on AAR, held in Japan, U.K., Australia,and Canada, in 1989, 1992, 1996, and 2000, respectively, theapplication of a good sealer on the surface of relatively thin

© 2002 NRC Canada

290 Can. J. Civ. Eng. Vol. 29, 2002



concrete components may reduce ASR and consequent ex-pansion. In most studies, the most interesting sealers corre-sponded to silane- and siloxane-based products.

ObjectivesThe principal objective of this study was to obtain more

information about: (i) the individual and combined effects offreezing and thawing, wetting and drying, and sodium chlo-ride exposure on expansion and surface deterioration ofplain and air-entrained concretes affected by ASR; (ii) themechanisms by which good sealers can reduce ASR expan-sion and (or) surface cracking; and (iii) the effectiveness anddurability of such sealers under severe exposure conditionssuch as those prevailing in North America.

Materials and methods

Materials

AggregatesThe coarse aggregate used for making all concrete speci-

mens corresponded to a highly alkali–silica reactive sili-ceous limestone from the Spratt Quarry (Ottawa, Ontario).This aggregate is used as a reactive control aggregate inCanada; it was supplied by the Department or Transportationof Ontario. The fine aggregate used was a non-reactive natu-ral granitic sand from the Québec City area (Canada).

CementsTwo CSA Type 10 cements were used in the study. A low-

alkali cement which contained 0.54% Na2Oe was used formaking the low-alkali concretes, while a medium-alkali ce-ment with 0.73% Na2Oe was used for making the high-alkaliconcretes.

Chemical admixtures and mixture waterReagent grade NaOH was added to the mixture water for

making the high-alkali concretes, such as to increase the al-kali content of the medium-alkali cement up to 1.25%Na2Oe. An air-entraining admixture was used for making theair-entrained concretes. The mixture water used was domes-tic tap water delivered by the municipality of Sainte-Foy(Canada). Reagent grade NaCl was used for the permanentor periodic immersion of a number of concrete specimens in6% (or 1 M) salt solution.

SealersFive different types of sealers were tested in this study:

one silane (S1), one oligosiloxane (S2), one polysiloxane(S3), one epoxy resin (S4), and linseed oil (S5). The firstthree products were selected based on their good perfor-mance when tested at the Laboratoire des Chaussées ofTransport Québec2 in accordance with the procedure pro-posed in the NCHRP Report No. 244 by Pfeifer and Scali(1981). In the tests made, concrete specimens were fabri-cated, 75 by 75 by 100 mm in size, cured for 7 days in a fogroom, allowed to dry in the laboratory during 21 days,sealed using the dosage recommended by the manufacturer(usually 2–4 m2/L), immersed for 21 days in a 15% NaClsolution, then allowed to dry in the laboratory for 21 days,

after which they were dried for three more days in adessicator. The mass of the specimens was recorded every3 days over the immersion and drying periods. The two per-formance criteria proposed in the NCHRP report (Pfeiferand Scali 1981) relate to the minimum absorption reductionduring the immersion period with respect to the control (un-sealed) specimens (>75%), and the amount of evaporatedwater during the drying period with respect to the amountabsorbed during immersion (>95%). All three silane- andsiloxane-based sealers used in the present study satisfiedboth criteria; this was not the case for the two other sealerstested, epoxy resin and linseed oil, which were retained forcomparison.

Concrete specimens

CompositionLow-alkali (1.9 kg/m3 Na2Oe) and high-alkali (4.4 kg/m3

Na2Oe) concretes were made with the above two aggregates,with or without using an air-entraining admixture. A total offour different concretes were made, the composition ofwhich is given in Table 1: (i) plain/low-alkali concrete;(ii) plain/high-alkali concrete; (iii) air-entrained/low-alkaliconcrete; and (iv) air-entrained/high-alkali concrete.

Specimen sizeThe concrete specimens were moulded in plastic pails to

give cylinders with a slightly conical shape, an average di-ameter of about 255 mm (at mid-length), a length of about310 mm, and a mass of about 38 kg. Two diametrically op-posite series of gauge lengths were located on the sides ofeach cylinder for longitudinal measurements (Fig. 1a).

Curing, sealing, and testing conditions

Initial curingAfter 7 days of humid curing in a fog room at 23°C, all

cylinders were allowed to dry for 21 days at room condi-tions, at about 35% relative humidity (RH) and 23°C.

Early treatmentsAfter curing, i.e., at 28 days, a number of cylinders were

sealed with one of the five sealers listed above (S1 to S5) inorder to evaluate their effectiveness in preventing excessiveexpansion due to ASR (Table 1). Other cylinders were sub-jected to a series of 179 consecutive freezing and thawingcycles in humid air (one cycle per day, from +25 to –18°C,with the cylinders placed above water in sealed plastic pails)(Table 1). A few specimens were also subjected to bothtreatments, with the series of freezing and thawing cyclesbeing applied prior or after the application of sealer (Ta-ble 1).

Exposure cyclesUp to one year or more, each cylinder was then subjected

to one of the 14-day exposure cycles (C1 to C5) described inTable 2. The most severe cycle, C5, included (i) 7 days ofhumid storage in air at >95% RH and 38°C, with the speci-mens placed above water in sealed plastic pails; (ii) 4 daysof drying in air at 38°C and about 30% RH; (iii) 30 min of

© 2002 NRC Canada

Bérubé et al. 291

2 D. Vézina. Personal communication.



wetting by immersion in 3% NaCl solution at 38°C, and(iv) 3 days of freezing and thawing cycling in humid air (onecycle per day, as above). The specimens subjected to cycleC1 were continuously exposed to humid air at 38°C and>95% RH, with the specimens placed above water in sealedplastic pails. Those subjected to cycle C2 were continuouslyimmersed in a 6% (i.e., 1 M) NaCl solution at 38°C.

Late treatmentsAfter one (plain concrete) or 1.5 (air-entrained concrete)

years, a number of cylinders that had not been sealed at28 days were sealed with the silane (S1), the oligosiloxane(S2), or the polysiloxane (S3) in order to determine their ef-fectiveness on concrete that was already severely affected byASR (Table 1). Other unsealed and early sealed cylinderswere subjected to 105 or 119 consecutive freezing and thaw-ing cycles in humid air (one cycle per day, as above) in orderto determine the effect of freezing and thawing on ASR-affected concrete (unsealed specimens), and to evaluate thefrost-susceptibility of an early application of sealer (Table 1).A few specimens were also subjected to both treatments, with

the series of freezing and thawing cycles being applied afterthe application of sealer (Table 1). All specimens were thenreturned to their respective exposure cycle (C1 to C5) forthe rest of the time.

Specimen identificationEach cylinder tested is identified hereafter by an alphanu-

meric number, which indicates (Tables 3 and 4) (i) if an air-entraining admixture has been used (A for air-entrained con-crete; P for plain concrete); (ii) the concrete alkali content(H for high; L for low); (iii) the main exposure cycle (C1 toC5); and (iv) all other early or late treatments to which thespecimens have been subjected, when it applies. These treat-ments include sealing (S1 to S5) and (or) series of 105 to179 consecutive freezing and thawing cycles (F). Moreover,the rank of each symbol in the specimen number indicatesthe sequence of the treatments. For instance, PLS1C4F cor-responds to a plain (P) and low-alkali (L) concrete cylinderwhich was sealed early with the silane (S1), then subjectedto the exposure cycle C4 up to one year or more, prior to be-ing exposed to a series of (105) consecutive freezing and

© 2002 NRC Canada

292 Can. J. Civ. Eng. Vol. 29, 2002

Low-alkali concretes High-alkali concretes

Parameter Plain Air-entrained Plain Air-entrained

Coarse aggregate Highly reactive siliceous limestone; 5–20 mm; 1114 kg/m3 (all concretes)Fine aggregate Non-reactive natural granitic sand; 741 kg/m3 (all concretes)Coarse:fine aggregate 1:1 (all concretes)Cement CSA type 10; 350 kg/m3 (all concretes)Mixture water Domestic tap water; 175 L/m3 (all concretes)Water/cement 0.50 (all concretes)Cement alkali 0.54% Na2Oe 0.54% Na2Oe 0.73% Na2Oe 0.73% Na2Oe

NaOH addition None None Cement increased to 1.25% Na2Oe

Concrete alkali 1.9 kg/m3 Na2Oe 1.9 kg/m3 Na2Oe 4.4 kg/m3 Na2O 4.4 kg/m3 Na2OWater-reducer None 0.49 L/m3 None 0.49 L/m3

Air-entrainer None 45.5 mL/m3 None 45.5 mL/m3

Slump 115–125 mm (all concretes)Air contenta 1.5% 5.8% (avg.) 2.4% (avg.) 6.4% (avg.)L bar (µm) 822 µm 290 µm (avg.) 692 µm (avg.) 283 µm (avg.)Specimen size Cylinders cast in slightly conical plastic pails; �38 kg in weight; �250 mm in diameter

(at mid-length) by �310 mm in lengthNo. of cylinders 6 10 20 34Exposure cyclesb C4 C2, C4, C5 C3, C4 C1, C3, C4, C5Sealers tested S1 S1 S1 S1 to S5Unsealed controls C4 C2 (2),c C4 (2), C5 (2) C3, C4 C1, C3 (3), C4 (3), C5 (3)Early sealing Sx S1C4 S1C4 (2), S1C5 (2) S1C3, S1C4 S1C3 to S5C3, S1C4 to

S5C4, S1C5 to S5C5Early freezing F — — FC3 (2) —Early sealing S1 then freezing F — — S1FC3 (2) —Early freezing F then sealing S1 — — FS1C3 (2) —Early sealing S1 and late freezing F S1C4F — S1C3F, S1C4F —Early freezing F and late sealing S1 — — FC3S1 (2) —Late sealing Sx C4S1 — C3S1, C4S1 C3S1 to C3S3, C4S1 to

C4S3, C5S1 to C5S3Late freezing F C4F — C3F, C4F —Late sealing S1 then freezing F C4S1F — C3S1F, C4S1F —

aMeasured on hardened concrete.bSee Table 2.cParenthetical values are number of specimens.

Table 1. Concrete proportionings and testing conditions.



thawing cycles (F), and returned to the exposure cycle C4for the rest of the testing period. The history of each cylin-der is indicated in Tables 3 and 4 for plain and air-entrainedconcretes, respectively.

Sealer application and dosageEach sealer was applied in two applications with about 10

minutes of drying after the first one, for a total applicationrate of 3–4 m2/L, which falls within the range recommendedby most manufacturers of sealers (i.e., 2–4 m2/L).

Periodic measurements

Mass and lengthThe mass (immediately after the humid storage, drying

and wetting periods, when applied) and the length (immedi-ately after the humid storage period) of all cylinders weremeasured periodically up to 2 years or more (Fig. 1a). Thespecimens were also examined periodically for surfacecracking.

Internal humidityAfter one year, the internal relative humidity was also

measured in two high-alkali specimens submitted to cycle

C3, unsealed and silane-sealed, respectively, immediatelyafter a period of humid storage. For each specimen, the mea-surements were performed along two small 20 mm diameterdrillholes parallel to the axis of the cylinders, at about 2 and13 cm, respectively, from the lateral surface using a com-mercial probe (Novasima MS1-E by Defensor) (Fig. 1b).

Results and discussion

Influence of alkali content on expansion due to alkali–silica reaction

Figures 2a (ALC2, ALC5) and 3 (PLC4, PLC4F) showthat all low-alkali cylinders did not either expand or developsurface cracking, whatever the exposure cycle, even the plainconcrete PLC4F exposed to more than 200 freezing andthawing cycles (i.e., 114 cycles within the 14-day exposurecycle C4 plus 105 consecutive cycles starting at one year),and the air-entrained cylinder ALC2 that was continuouslyimmersed in 1 M NaCl solution. On the other hand, all un-sealed high-alkali cylinders (all specimens numbers with an“H” in Figs. 2a and 3) presented high expansion in the longterm after being exposed to moisture (i.e., to cycle C1, C3,C4, or C5).

Influence of freezing and thawing cycles on concretenot yet affected by alkali–silica reaction

As measured on samples from each concrete batch, theair-void spacing factor (L bar) averaged from 692 µm (high-alkali) to 822 µm (low-alkali) for the plain concretes, andfrom 283 µm (high-alkali) to 290 µm (low-alkali) for the air-entrained concretes (Table 1). The unsealed high-alkali plainconcrete subjected after 28 days to 179 consecutive freezingand thawing cycles did not significantly expand during thistreatment (Fig. 3: PHFC3 up to 35 weeks), despite a highlydeficient air-void system. However, this specimen started toexpand after being subjected to humid conditions (i.e., to cy-cle C3) at a rate quite similar to that initially for the cylinderPHC3, which was subjected to cycle C3 since the beginning.It thus appears that the early application of freezing andthawing cycles did not significantly modify the further rateof ASR expansion.

Influence of exposure conditions on concrete currentlyaffected by alkali–silica reaction

Influence of wetting and drying cycles in the presence ofalkali–silica reaction (cycles C3 to C5)

A 4-day drying period for every period of 2 weeks re-duced the ASR expansion in the long term by about 40%compared with continuous exposure to humidity (Fig. 2a:AHC1 vs. AHC3), and the mass of specimens by about350 g (≈1%) over a period of 2 years (Fig. 2b). In fact, thecylinders exposed to cycle C3 are under humid conditionsduring 71% of time (10 days per 14-day cycle), compared to100% of time for cylinders exposed to C1. This reducedASR expansion, as expected. However, wetting and dryingpromoted surface deterioration (map-cracking) by creatingin the near-surface concrete less favourable conditions forASR (alkali leaching during wetting, lower humidity duringdrying), which result in tension stresses.

© 2002 NRC Canada

Bérubé et al. 293

Fig. 1. (a) Length-change measurement of a concrete cylinderand (b) relative humidity measurement along small drillholes ina concrete cylinder.



Influence of freezing and thawing cycles in the presenceof alkali–silica reaction (cycles C4 and C5)

The addition of periodic freezing and thawing cycles towetting and drying (with wetting by immersion in tap water)significantly increased the expansion in the long term(Fig. 2a: AHC3 vs. AHC4; Fig. 3: PHC3 vs. PHC4), despitethe fact that the corresponding cylinders were stored for ashorter period of time at >95% RH (7 days per 14-day cyclefor C4 vs. 10 days per cycle for C3). Their mass is alsogreater than that of specimens exposed to wetting and dryingonly (cycle C3) (Fig. 2b). In fact, the cylinders exposed tocycle C4 expanded more and are more microcraked thanthose subjected to cycle C3, then they absorbed more mois-

ture. At the end of the testing period, they are still signifi-cantly expanding while those subjected to cycle C3 are lev-elling off. This remark particularly applies to plain concretes(Fig. 3: PHC4 vs. PHC3). Freezing and thawing during cy-cles C4 or C5 also greatly promoted map-cracking (Fig. 4a).Immersion in salt water (cycle C5) rather than tap water (cy-cle C4) before the periodic freezing and thawing cycles in-creased expansion and enhanced map-cracking as well(Fig. 2a: AHC5 vs. AHC4). This is attributed to the deleteri-ous effect of sodium chloride during freezing rather than toan additional alkali supply, since the low-alkali concrete cyl-inder ALC2, which is always immersed in 1M NaCl, neverexpanded significantly (Fig. 2a), for the reasons suggested

© 2002 NRC Canada

294 Can. J. Civ. Eng. Vol. 29, 2002

Wetting at 38°C through complete immersion in

CycleHumid air curing at38°C and > 95% RHa

Drying at 38°Cand 30% RH Tap water Salt waterb

Freezing andthawing cyclesc

C1 14 days — — — —C2 — — — 14 days (6%) —C3 10 days 4 days 30 min — —C4 7 days 4 days 30 min — 3 daysC5 7 days 4 days — 30 min (3%) 3 days

aCylinders stored above water in sealed plastic pails.bImmersion in 3% or 6% NaCl solution.cOne cycle per day; 16-h freezing/8-h thawing; +23 to –18°C.

Table 2. Description of the 14-day exposure cycles to which the concrete cylinders were subjected.

Treatment at 28 days Exposure cycle Treatment at 1 year Final exposure Cylinder notation

High-alkali concrete (H)— C3 — C3 PHC3— F (105) PHC3F— S1 PHC3S1— S1 then F (119) PHC3S1FF (179) — PHFC3 (2)F (179) S1 PHFC3S1 (2)S1 — PHS1C3S1 F (105) PHS1C3FS1 then F (179) — PHS1FC3 (2)F (179) then S1 — PHFS1C3 (2)— C4 — C4 PHC4— F (105) PHC4F— S1 PHC4S1— S1 then F (119) PHC4S1FS1 — PHS1C4S1 F (105) PHS1C4FLow-alkali concrete (L)— C4 — C4 PLC4— F (105) PLC4F— S1 PLC4S1— S1 then F (119) PLC4S1FS1 — PLS1C4S1 F (105) PLS1C4F

Note: P, plain concrete; H = 4.4 kg/m3 Na2Oe; L = 1.9 kg/m3 Na2Oe; S1, sealed with silane S1; F (nnn), series of nnn consecutivefreezing and thawing cycles (one cycle per day; +23°C to –18°C), in air at >95% RH (specimens above water in sealed plasticpails); (2) denotes two cylinders (rather than usually one) from two different concrete batches and exposed to exactly the sametreatments.

Table 3. Plain concrete cylinders tested and exposure conditions.



© 2002 NRC Canada

Bérubé et al. 295

by Duchesne and Bérubé (1996) and Bérubé et al. (2000)and discussed above.

Influence of freezing and thawing cycles after alkali–silica reaction took place

The application of 105 consecutive freezing and thawingcycles after significant ASR expansion had taken place alsoinduced high additional expansion (Fig. 3: PHC3F, PHC4F),even when ASR expansion was almost complete (PHC3F).

Effectiveness of sealers on new concrete not yet affectedby alkali–silica reaction

Influence of alkali content and type of sealerAll low-alkali cylinders that were sealed early (with silane

S1) presented shrinkage all over the testing period (Figs. 5aand 6a: PLS1C4, PLS1C4F). The Canadian Standards Asso-ciation suggests a one-year, 0.04% expansion limit criterion

for concrete exposed to conditions identical to cycle C1(CSA 2000). All high-alkali cylinders sealed early with sil-ane S1, which proved to be the best product among alltested, satisfied this criterion regardless of the air contentand the exposure conditions (Figs. 5a and 6b: PHS1C3,PHS1C3F; Fig. 6c: PHS1C4, PHS1C4F; Fig. 6d: PHFS1C3,PHS1FC3). The cylinders sealed with oligosiloxane S2 orpolysiloxane S3 marginally exceeded the above limit butonly when exposed to the cycle C5 (which corresponds toFig. 5a); the oligosiloxane was slightly better than thepolysiloxane. Moreover, all cylinders treated early with sil-ane, oligosiloxane, or polysiloxane did not show map-cracking (Fig. 4b). The epoxy sealer S4 was not effective inreducing expansion (Fig. 5a) and map-cracking as well, andgreatly changed the appearance of concrete (darker colorand waxy lustre). The linseed oil prevented map-crackingwhile significantly reducing expansion; however, the expan-sion was not under the 1-year 0.04% usual limit for accep-

Treatment at 28 days Exposure cycle Treatment at 1.5 years Final exposure Cylinder notation

High-alkali concrete (H)— C1 — C1 AHC1— C3 — C3 AHC3 (3)S1 — AHS1C3S2 — AHS2C3S3 — AHS3C3S4 — AHS4C3S5 — AHS5C3— S1 AHC3S1— S2 AHC3S2— S3 AHC3S3— C4 — C4 AHC4 (3)S1 — AHS1C4S2 — AHS2C4S3 — AHS3C4S4 — AHS4C4S5 — AHS5C4— S1 AHC4S1— S2 AHC4S2— S3 AHC4S3— C5 — C5 AHC5 (3)S1 — AHS1C5S2 — AHS2C5S3 — AHS3C5S4 — AHS4C5S5 — AHS5C5— S1 AHC5S1— S2 AHC5S2— S3 AHC5S3Low-alkali concrete (L)— C2 — C2 ALC2 (2)— C4 — C4 ALC4 (2)S1 — ALS1C4 (2)— C5 — C5 ALC5 (2)S1 — ALS1C5 (2)

Note: A, air-entrained concrete; H = 4.4 kg/m3 Na2Oe; L = 1.9 kg/m3 Na2Oe; S1, silane; S2, oligosiloxane; S3, polysiloxane; S4,epoxy resin; S5, linseed oil; (2) or (3) denotes two or three cylinders (rather than usually one) from the same concrete batch andexposed to exactly the same treatments.

Table 4. Air-entrained concrete cylinders tested and exposure conditions.



© 2002 NRC Canada

296 Can. J. Civ. Eng. Vol. 29, 2002

tance (Fig. 5a); it also gave a bad appearance to concrete(nonuniform color and lustre).

Mass variationsFor each exposure cycle, a clear correlation is observed

between concrete expansion and mass variation (except forcylinders sealed with the epoxy resin) (Fig. 5a vs. Fig. 5b).This is not surprising considering that expansion is directlyrelated to microcracking, which in turn allows moisture up-take. In fact, the mass of the concrete cylinders sealed withsilane or siloxane progressively decreased by more than400 g (or 1%) over a period of 2 years, whatever the expo-sure cycle, compared with a mass increase of more than200 g for unsealed cylinders (Fig. 5b). These sealed speci-mens are still loosing mass after 2 years. At the same time,the cylinders sealed with linseed oil presented lower masslosses, and their mass started to slightly increase after about1.5 years. On the other hand, the mass of the specimenstreated with the epoxy resin has progressively increasedsince the beginning of the tests.

Evaporation during drying and absorption during wettingand humid storage

In the case of all cylinders not sealed at 28 days and ex-

posed to cycles C3 to C5, the loss in mass (moisture) duringeach 4-day drying period, and the gain in mass during eachfollowing 30-min period of immersion in tap or salt water,tend to diminish slightly with time (“unsealed” in Fig. 5c),despite increasing expansions (“unsealed” in Fig. 5a); this islikely due to additional hydration (lower internal permeabil-ity) and (or) surface carbonation of concrete (lower surfacepermeability). At the same time, all cylinders early sealedwith silane or siloxane showed relatively constant mass lossduring drying, and gain during wetting all over the testingperiod (“silane 0 w” in Fig. 5c). The effectiveness of allthree good sealers tested is related to the fact that they sig-nificantly reduced the moisture uptake in cylinders exposedto humid conditions, i.e., during the periods of wetting andhumid storage in air, while not reducing as much the mois-ture loss through evaporation during the drying periods(Fig. 5d); the moisture balance is thus negative during eachexposure cycle C3 to C5. For instance, one can see inFig. 5c that the silane-sealed cylinder exposed to cycle C5looses at equilibrium about 80 g of moisture during each 4-day drying period, while it absorbs about 20 g during eachfollowing 30-min period of immersion in salt water, for anegative balance of about –60 g per wetting and drying cy-cle; on the other hand, the unsealed specimen looses at equi-librium about 160 g of moisture during drying, while itabsorbs about 80 g during immersion in salt water, for agreater negative balance of about –80 g per wetting and dry-ing cycle. However, as shown in Fig. 5d, the cumulative lossin mass observed for the cylinders sealed with silane andsiloxane is explained by the fact that these cylinders ab-sorbed much less moisture during the rest of the exposurecycle (i.e., during the 7- or 10-day periods of humid storage)than they lost during drying and wetting, with the reverse forthe unsealed and epoxy-sealed cylinders. The cylinderssealed with linseed oil also present a negative mass balancewithin each 14-day cycle, which is, however, much less im-portant than for the cylinders sealed with silane or siloxane.

Internal humidityMeasurements made immediately after a period of humid

storage inside two 1-year-old cylinders exposed to cycle C4

Fig. 2. (a) Expansion and (b) cumulative mass variation of un-sealed air-entrained concrete cylinders subjected to exposure cy-cles C1 to C5.

Fig. 3. Expansion of unsealed plain concrete cylinders subjectedto exposure cycles C3 and C4.



© 2002 NRC Canada

Bérubé et al. 297

Fig. 4. (a) Unsealed and (b) silane-sealed air-entrained concrete cylinders subjected to exposure cycle C4, after 1.5 years.

Fig. 5. (a) Expansion, (b) cumulative mass variation, (c) mass variation during wetting and drying, and (d) mass variation at any timeof unsealed and sealed air-entrained concrete cylinders subjected to exposure cycle C5.



indicated that the relative humidity is significantly higher inthe unsealed and expansive cylinder (95% RH in center and96% near surface) than in the silane-sealed and non expan-sive specimen (86% RH in center and 81% near surface),particularly near the surface. This suggests in turn that “in-ternal” humidity conditions over 80–85% are necessary forASR expansion. On the other hand, another study (Pedneault1996) indicated that “external” humidity as low as 65% RHis sufficient to obtain expansion over 0.04% per year for thesame type of concrete, and that the more reactive the aggre-gate, the lower the critical external humidity level requiredfor significant expansion to occur. It is clear that there is aworld of difference between internal and external (or ambi-ent) humidity.

Influence of an early or late series of freezing andthawing cycles

Figure 6d shows that plain concrete cylinders subjectedearly to (179) consecutive freezing and thawing cycles priorto (PHFS1C3) or immediately after (PHS1FC3) being sealedwith silane S1 did not significantly expand during these cy-cles and afterwards when subjected to conditions promotingASR, i.e., to cycle C3. Figures 6b and 6c also show that

plain concrete cylinders sealed early with silane S1 did notexpand even after being subjected later to 105 consecutivefreezing and thawing cycles (PHS1C3F, PHS1C4F). The ef-fectiveness of the silane tested in preventing expansion dueto ASR in new concretes then appears not significantly af-fected by freezing and thawing. Recall that the cylinderPHS1C4F also suffered 117 other periodic freezing andthawing cycles when exposed to cycle C4 (three cycles eachtwo weeks).

Effectiveness of sealers on concrete already affected byalkali–silica reaction

The severely deteriorated cylinders that were sealed withsilane or siloxane after one or 1.5 years started to loose mass(i.e., moisture) (Fig. 5b) and contracted immediately afterbeing sealed with silane or siloxane (Figs. 5a and 6b:PHC3S1, PHC3S1F; Fig. 6c: PHC4S1, PHC4S1F; Fig. 6d:PHFC3S1), even in the presence of wetting and drying, andfreezing and thawing cycles (i.e., exposure cycles C4 andC5). Also, the surface cracking became less and less appar-ent owing to the fact that the cracks progressively dried.Moreover, the positive effect of the above sealers was not re-duced by the application, immediately after sealing, of a se-

© 2002 NRC Canada

298 Can. J. Civ. Eng. Vol. 29, 2002

Fig. 6. Expansion of sealed and unsealed plain concrete cylinders: (a) low-alkali specimens subjected early to exposure cycle C4;(b) high-alkali specimens subjected early to exposure cycle C3; (c) high-alkali specimens subjected early to exposure cycle C4; and(d) high-alkali specimens subjected early to 179 freezing and thawing cycles, then to exposure cycle C3.



ries of (119) freezing and thawing cycles (Fig. 6b:PHC3S1F; Fig. 6c: PHC4S1F). Also, the mass loss duringdrying, and the mass increase during wetting of these cylin-ders, which were already highly microcracked, quite rapidlyreached values that approached those for early sealed anduncracked cylinders (“Silane 1.5 yr” in Fig. 5c).

Conclusions

Influence of exposure conditions on alkali–silicareaction• In the absence of ASR (low-alkali concretes), wetting and

drying cycles and freezing and thawing cycles in humidair did not cause damage to air-entrained and plain con-cretes cylinders, 255 diameter by 310 mm in size, madewith a highly alkali–silica reactive siliceous limestone anda low-alkali cement.

• Low-alkali concrete cylinders continuously immersed in1M NaCl never expanded significantly; this suggests thatlow-alkali concretes are sufficiently protected againstASR, even when exposed to salt solutions.

• Wetting and drying significantly reduced ASR expansionof the high-alkali concrete cylinders; however, wettingand drying promoted map-cracking on their surface.

• Freezing and thawing in humid air greatly increased ex-pansion and surface cracking of the high-alkali concretecylinders currently or previously suffering of ASR, evenwhen the reactions were almost complete; however, beforeASR expansion was initiated, freezing and thawing in hu-mid air did not cause either concrete expansion or surfacecracking, even in the presence of a deficient air-void sys-tem.

• This study confirms that the exposure conditions greatlyinfluence the development of cracking on the surface ofconcrete members affected by ASR and, at least for mem-bers of less than 255 mm in thickness (= diameter of thecylinders tested), their expansion. In fact, wetting and dry-ing and freezing and thawing greatly promoted surfacecracking despite the fact that expansion due to ASR (only)is reduced when concrete is allowed to dry (lower humid-ity) or to freeze (lower temperature).

Influence of sealers on alkali–silica reaction• When applied early on (high-alkali) concrete cylinders

susceptible to ASR, the silane, the oligosiloxane, and thepolysiloxane tested reduced ASR expansion to an accept-able level, even for specimens subjected to wetting anddrying, freezing and thawing, and salt water. Moreover, thecorresponding cylinders did not show surface cracking. Theepoxy sealer tested was not effective at all. Linseed oil pre-vented map-cracking while reducing expansion; however,the expansion was not under the 1-year 0.04% usual limitfor acceptance.

• The results obtained in the laboratory thus suggest that agood sealer can reduce expansion to an acceptable level innew concrete members susceptible to ASR and exposed tosevere conditions, and the related surface cracking aswell, at least for non-massive concrete members up toabout 255 mm in thickness.

• The results obtained also indicate that a good sealer canbe applied with success to concretes that are severely af-

fected by ASR, can improve their aesthetic appearance,can stop concrete expansion and even produce contrac-tion, at least for non-massive concrete members. Theabove three good sealers were applied in the field onhighway median barriers showing different degrees ofASR deterioration. The results thus obtained are discussedelsewhere (Bérubé et al. 2002) and confirm most conclu-sions from the present laboratory study.

• Whatever the exposure conditions (all cylinders) and theinitial degree of ASR deterioration at the time of sealing(sealed cylinders), all expansion results correlated withmass and humidity variations within the concrete speci-mens tested. The effectiveness of all three good sealerstested is related to the fact that immediately after sealingand afterwards, the corresponding specimens absorbedless moisture when exposed to humid conditions (i.e., dur-ing the periods of immersion in water and storage in hu-mid air) than they lost through evaporation when exposedto dry conditions; in the meantime, the moisture contentprogressively increased with time in all unsealed and ep-oxy-sealed concrete specimens, all ASR-affected.All plain and air-entrained concrete specimens tested in

this study were made with a cement content of 350 kg/m3

and a water–cement ratio of 0.50; this is close to the compo-sition of numerous existing concrete members that are cur-rently affected by ASR. Moreover, the exposure cycles towhich the concrete specimens were subjected were quitewell-representative of the severe conditions that prevail inmany northern countries. Also, the thickness of many exist-ing concrete members is similar to or lower than 255 mm,which corresponds to the diameter of the concrete cylinderstested. The above conclusions should thus apply to variousfield concrete members, even with different permeability orof different dimensions. However, one can expect that thethinner the concrete member affected by ASR, the higher itspermeability, and the more deficient its air void system, thegreater should be the effectiveness of a good sealer in reduc-ing ASR expansion and related surface deterioration. Forsure, it is unlikely that a good sealer can reduce ASR expan-sion of massive concrete members; however, it should re-duce the development of cracking on the surface of suchmembers, by reducing the deleterious effects of all exposureconditions such as wetting and drying cycles, freezing andthawing cycles, sea water and deicing salt.

Acknowledgements

This study was supported by Hydro Québec, the Fondspour la formation de chercheurs et l’aide à la recherche duQuébec (FCAR) and the Natural Sciences and EngineeringResearch Council of Canada (NSERC). Special thanks toTransport Québec for their very kind collaboration, and par-ticularly to Daniel Vézina.

References

Bérubé, M.A., Fournier, B., and Frenette, J. 1989. Détérioration defondations de pylones d’ancrage de lignes de transportd’électricité par des réactions alcalis–granulats, performancemécanique et réparation du béton. Canadian Journal of Civil En-gineering, 16: 945–959.

© 2002 NRC Canada

Bérubé et al. 299



© 2002 NRC Canada

300 Can. J. Civ. Eng. Vol. 29, 2002

Bérubé, M.A., Dorion, J.F., and Vézina, D. 2000. Laboratory andfield investigation of the influence of sodium chloride on ASR.Proceedings of the 11th International Conference on AAR, Qué-bec, Que., pp. 149–158.

Bérubé, M.A., Chouinard, D., Pigeon, M., Frenette, J., Rivest, M.,and Vézina, D. 2002. Effectiveness of sealers in counteractingASR in highway median barriers exposed to wetting and drying,freezing and thawing, and deicing salt. Canadian Journal ofCivil Engineering, 29: this issue.

CSA. 2000. Standard CSA A23.1: concrete materials and methodsof concrete construction, and Standard CSA A23.2: methods oftest for concrete. Canadian Standards Association, Rexdale, Ont.

Duchesne, J., and Bérubé, M.A. 1996. Effect of deicing salt andsea water on ASR: new considerations based on experimentaldata. Proceedings of the 10th International Conference on Al-kali–Aggregate Reaction in Concrete, Melbourne, Australia,pp. 830–837.

Ludwig, U. 1989. Effects of environmental conditions on alkali–aggregate reaction and preventive measures. Proceedings of the

8th International Conference on Alkali–Aggregate Reaction inConcrete, Society of Materials Science, Kyoto, Japan, pp. 583–596.

Nishibayashi, S., Yamura, K., and Sakata, K. 1989. Evaluation ofcracking of concrete due to alkali–aggregate reaction. Proceed-ings of the 8th International Conference on Alkali–AggregateReaction in Concrete, Society of Materials Science, Kyoto, Ja-pan, pp. 759–764.

Pedneault, A. 1996. Développement de procédures d’essai etd’analyse pour l’évaluation du potentiel résiduel de réaction,d’expansion et de détérioration du béton affecté de réactivitéalcalis-granulats. M.Sc. thesis, Laval University, Québec, Que.

Pfeifer, D.W., and Scali, M.J. 1981. Concrete sealers for protectionof bridges structures. National Cooperative Highway ResearchProgram (NCHRP), Transportation Research Board, WashingtonD.C., Report No. 244.



effectiveness of sealers in counteracting alkali–silica reaction in plain and air-entrained...

Documents