Keywords

1 Introduction

The alkali–silica reaction (ASR) is a deleterious reaction affecting the long-term durability of concrete structures. ASR proceeds in the pore solution of concrete when alkali hydroxides and reactive forms of silica are present. It can be produced in any cementitious system provided that the three reaction components are available: alkali, reactive silica, and water. Reactive silica, such as strained quartz or amorphous opal, is introduced by aggregates. These silica forms dissolve in the alkaline pore solution and subsequently precipitate to form an alkali–silica gel, the ASR gel. The ASR gel, once precipitated in the concrete, generally within an aggregate, exerts mechanical stress from within the concrete, causing the development of cracking. To reduce the potential for deleterious ASR, a number of mitigation strategies are deployed in Australia, including the use of nonreactive aggregate materials, a limit on the maximum available alkali, and the incorporation of supplementary cementitious materials (SCMs). In Australia, the concrete alkali limit is 2.8 kg/m3 Na2Oeq [1] and by specifying a cement alkali content limit of 0.6% Na2Oeq [2].

Two standardized test methods have been adopted for use in Australia to determine an aggregate’s propensity to cause ASR: AS1141.60.1, the AMBT, and AS1141.60.2, the concrete prism test (CPT). These tests are carried out under accelerated conditions designed to provide aggressive reaction environments to promote ASR gel development [2, 3]. An alkali threshold is the lowest alkali level in concrete where deleterious expansion is found [3]. Threshold testing has largely been the topic of investigation in CPT-based studies, for which RILEM has a recommended testing method published as protocol AAR-3.2 [3]. Although threshold focus has been on CPT-based studies, the AMBT could prove valuable as a tool for identifying threshold behavior. The benefit of assessing aggregate reactivity using the AMBT procedure is that it enables a rapid evaluation of an aggregate’s reactivity within a 21-day timeframe (Table 1) and these benefits could also apply to threshold investigation should the method show viability. ASR gels in AMBT specimens have been observed to have compositions that vary compared with gel produced in concretes, with AMBT-induced gels having a higher sodium content compared with similar binder concretes, likely due to the external 1 M sodium hydroxide bath [4]. Regardless of gel compositional discrepancies, the AMBT remains valuable to industry for aggregate reaction screening due to its rapid timeframe and this warrants additional investigation of the test and the properties of the ASR gel it produces. The AMBT is also used to assess aggregate reactivity in the presence of SCM-incorporated binders to identify the level of SCM required to mitigate ASR for a particular aggregate. SCMs mitigate ASR through a variety of mechanisms that reduce the available alkali of the pore solution [5, 6].

Table 1 Reactivity classification for the AMBT as defined by AS1141.60.1

The focus of this study was the alkali concentration of the bath solution in modified AMBT experiments and we report expansion tests for reactive aggregates as a function of the bath solution concentration. In addition, expansion of mortars incorporating SCM-blended cements is reported for cements containing fly ash (FA) and slag (S).

2 Methods

The primary aggregate of focus was reactive river sand classified as reactive by AS1141.60.1. Reactive river sand contains 10.7% moderately strained quartz, 2% heavily strained quartz and 1.3% fine microcrystalline quartz within fragments of indurated meta-greywacke/siltstone and acid volcanic rock. Three binder material combinations were used to prepare the prisms: GP cement, FA and ground granulated blast furnace slag (S). The cement used was a Portland-type GP cement that met the specified requirements of AS3972, with an alkali content of 0.47% Na2Oe determined by X-ray fluorescence analysis. For the FA and S incorporated mixes, cement replacement percentages of 25% and 65%, respectively, were chosen because they are representative of the recommended SCM substitution rates for mitigating ASR by AS HB79 [1]. To mix the mortar, the procedure outlined in AS1141.60.1 was followed. Gauge studs were placed within the mold prior to mixing and calibrated to have a gauge length of 250 ± 1 mm. Fine aggregate was prepared in its natural unaltered grading by oven drying at 110 °C before cooling for mixing. Potable tap water was used for mixing. The mortar prisms were cured in three gang molds for 24 h before demolding and were then immersed in tap water at room temperature prior to heating to 80 °C for 24 h for equilibration prior to zero day length measurement and subsequent immersion in respective alkali baths equilibrated at 80 °C.

To assess the effect of different external alkali environments and to observe potential threshold behavior, four alkali concentrations were used as immersion baths for the mortar bars: 0.4, 0.7 and 1 NaOH and a bath of saturated Ca(OH)2 solution, which was used as the control bath with no external source of alkali. Distilled water was used to prepare the immersion solutions. The baths were kept at 80 °C throughout the duration of the test. Mortar bars were vertically oriented within the bath, supported by a stainless steel grid so that no contact with the gauge pins occurred.

To determine the comparative length change of the specimens, all comparative expansion measurements were conducted on a steel frame comparator equipped with a Mitutoyo digital micrometer. All expansion measurements are in reference to a 295-mm invar reference bar that was placed with identical positioning for each measurement and checked on a regular basis between mortar bar measurements. The mortar bars’ comparative length measurements were recorded at day 0 (immediately after removal from 80 °C water bath), then at 1, 3, 7, 10, 14, 21 and 28 days following immersion in the alkali baths. To measure relative expansion, mortar bars were removed from the alkali baths, placed in the comparator for recording to a precision of 1 micron. These comparative length measurements were carried out within 10 s of removal from the bath and were measured in the same orientation within the comparator at each age.

3 Results and Discussion

An overview of the GP cement mortar bars (no SCM) is shown in Figs. 1 and 2 displays the expansion curve plots for each binder composition over time while immersed in the respective alkali immersion baths listed. The concentrations of the solutions in each bath were measured by titration and the pH calculated is listed in Table 2, which also lists the expansion for each GP cement mortar bar (no SCM). Expansion is a strong function of the alkali content of the solution concentration. Little or no expansion was observed over the timeframe for mortar bars exposed to the saturated Ca(OH)2 solution. Expansion for the alkali solutions increased with increasing bath concentration over the time frame of the experiment. An induction period was apparent for 0.4 M NaOH where expansion was negligible up to 14 days followed by a notable increase in expansion. Expansion appeared to be increasing at 28 days. Further measurements will yield a limit for the expansion and discriminate the concentration effects on ASR. Threshold definitions have yet to be applied to AMBT-based studies. For standard reactivity assessment of the AMBT, if expansion is equal to, or exceeds 0.1% at 10 days (0.15% for natural sands such as the river sand used in the present study) or 0.3% at 21 days in bars exposed to 1 M NaOH, then the aggregate is classified as reactive (Table 1). For the GP mix immersed in 1 M and 0.7 M NaOH we observed the mortar bars exceed this expansion limit, indicating a classification of reactive. For the mortar bars exposed to 0.4 M NaOH, the expansion was within the limit, with an expansion of 0.230% observed at 21 days. Although the expansion was below the expansion limit designated for classification as reactive, the test was nonstandard and some expansion due to the reactivity of the aggregate was observed. A delay in significant expansion was observed after 14 days, suggesting an induction period prior to expansion (a delay in the onset of expansion) for the 0.4 M NaOH bath, rather than an alkali threshold, because expansion tends to be >0.3% at 28 days. For the mortar bars exposed to 0.7 M and 1 M NaOH, it appeared that, within the resolution of the measurements taken, the induction period was relatively similar, with the onset to significant expansion occurring between 3 and 7 days. When the expansion rates were compared (Fig. 1), it could be seen that increasing the NaOH concentration increased the expansion rate. It remains to be seen with this aggregate whether threshold behavior is seen in maximum expansion with respect to the change in alkali bath concentration.

Fig. 1
A multi-line graph of expression versus day. 1 M N a O H is high at (30, 0.9). 0.7 M N a O H is (30, 0.7). 0.4 M N a O H is (30, 0.4). Saturated C a O H is (30, 0.001). The values are approximate.

Percentage expansion of natural sand mortar bars immersed in 0.4 M, 0.7 M and 1.0 M NaOH solutions and saturated Ca(OH)2 solution over 28 days

Fig. 2
Four graphs of expansion versus day. The titles of the graphs are 0.5 M N a O H, 0.7 M N a O H, 1 M N a O H, and C a O H 2. It depicts G P, G P plus F A, and G P plus S. A. G P is high at (28, 0.4). B. G P is high at (28, 0.7). C. G P is high at (28, 1.0). D. All done in a horizontal line at (0, 0). The values are approximate.

Expansion of mortars submerged in a 0.4 M NaOH, b 0.7 M NaOH, c 1.0 M NaOH and d saturated Ca(OH)2 solutions at 80 °C for 28 days

Table 2 Percentage expansion of mortar bars at age 28 days for 100% GP cement binder and calculated pH for each alkali immersion bath

No significant expansion was observed for the reactive aggregate mortar bars prepared with blended cements containing FA or S (Fig. 2). These SCMs appeared to sufficiently mitigate ASR within the timeframe of the experiment. Measurements to extended ages will identify whether this is the result of complete mitigation or if the expansion-free region is an induction period where the SCM acts as an inhibitor, delaying the onset of expansion once consumed in the pozzolanic reaction.

In summary, expansion increased with increasing concentration of NaOH solution, demonstrating that reactivity increases with increasing availability of alkali. Additionally, the rate of expansion (slope of the expansion curves) was observed to increase with increasing alkali concentration in the bath solution. An induction period was observed for bars exposed to 0.4 M NaOH, and although this is not traditional threshold behavior, it does indicate that there is a mechanism affecting the time for expansion to occur at lower alkali testing environments. As data collection was limited to 28 days, the maximum possible expansion has not been determined. Further measurements at greater ages will be carried out to identify any long-term influence of alkali solution concentration on expansion to attempt to identify the origins of the expansion behavior observed, with the aim of differentiating between threshold behavior and the degree of reaction controlled as the reactivity or concentration of the immersion solution. The incorporation of SCMs in the mortar mix significantly reduced expansion, which indicated mitigation, but further measurement is required to ensure that SCMs are not mitigating the reaction only in the short term and delaying the onset of expansion.

4 Conclusions

As the alkali solution concentration increases, mortar bar expansion increases, which indicates a relationship between alkali hydroxide concentration and ASR reaction severity. Threshold behavior, as defined by RILEM, was not observed for this aggregate at lower alkali levels; however, an induction period before the onset of significant expansion was observed at the lower alkali bath concentration of 0.4 M. The expansion data presented here is limited to the timeframe of the AMBT testing criteria, so further extending AMBT experiment timeframes may clarify the relationship between alkali concentration and extent of the ASR reaction with time, including the maximum observed expansion. The investigation demonstrated that both FA and S mitigate ASR-induced expansion in aggressive accelerated reaction environments during testing at incremental concentrations up to 1 M NaOH.