Assessment of highway pavement concrete suffering from alkali-silica reaction: case study

3.1. Visual inspections and macroscopic analysis

The drilled cores were subjected to a thorough visual inspection, photographed (Figure 2) and measured. Macroscopic analysis was performed on halves of 6 cores cut vertically. The remaining core halves were left as control and reference. The homogeneity of the concrete, general composition, type of fine and coarse aggregate as well as entrapped air and cracks were taken into consideration. After macroscopic analysis, the specimens were cut into 30±2 mm sections, polished on SiC powders (180, 320, 600, 1000) and dried at 50°C for 24 h. The Damage Rating Index (DRI) method was used to quantify the internal deterioration of the concrete pavement (1818. Villeneuve, V.; Fournier, B.; Duschene, J. (2012) Determination of the damage in concrete affected by ASR - The damage rating index (DRI). Proc. 14^th ICAARAustin, USA
). A petrographic examination was performed using a stereomicroscope at 30x mag. in order to identify and count defects associated with ASR. The assessment was made on the basis of summing up individual factors with different weights adopted in accordance with (1818. Villeneuve, V.; Fournier, B.; Duschene, J. (2012) Determination of the damage in concrete affected by ASR - The damage rating index (DRI). Proc. 14^th ICAARAustin, USA
). The higher the DRI value, the higher the ASR-induced concrete degradation rate. The analysis was performed on 6 specimens (10x10 cm).

3.2. Mechanical properties

In order to determine the compressive strength and the bulk density of the highway pavement concrete, sections of small cores (Ø ≈ 100 mm and 200 mm long) were cut out of each drill core. As in the case of the E-modulus tests, part of these sections were exposed to moist and warm storage (40ºC and 100% RH). After cutting and grinding and prior to testing, these specimens were stored under laboratory conditions (23ºC and 50% RH until thermo-hygro equilibrium), weighed and measured to determine the bulk density.

To determine the modulus of elasticity, sections 170‒200 mm long were cut out of the small cores. Then the end faces of the sections were ground plane-parallel. Afterwards the E-modulus tests were carried out.

The specimens were initially loaded up to the σ_o upper stress of almost 30 MPa. This corresponds to about one third of the average compressive strength of the specimens of 85.4 MPa. Afterwards, the specimens were unloaded to σ_u of about 0.5 MPa. Then two more loading and unloading cycles were applied. Finally, the Young’s modulus was determined as the secant modulus from the third rising branch of the respective stress‒strain lines.

The E-modulus tests were carried out on specimens stored in laboratory conditions (23ºC and 50% RH) and specimens that were exposed beforehand to a moist-warm storage (40ºC and 100% RH) to initiate ASR.

3.3. Microscopic analysis

For petrographic analysis, a polarising microscope (Olympus BX51) equipped with a camera and an automatic moving table (Prior ES11BX/B) were used. Thin sections of about 20±2 µm were used for plane-light (PPL) and cross-polarised light (XPL) observations at magnifications from 40 to 400x. The specimens cut from the concrete cores were lapped and then polished (45x30 mm) to identify the alkali-silica gel. A layer of carbon was sprayed onto the samples and a strip of conductive tape was attached to provide conductivity. A scanning electron microscope (SEM) equipped with energy dispersion X-ray analysis (EDX) was used to study the microstructure of the concrete. A Zeiss sigma VP microscope in backscatter mode at an accelerating voltage of 20 kV was used.

4.1. Visual inspections and macroscopic analysis

The thickness of the structural concrete of the roadway slabs was from 285 to 247 mm. The drill cores from the structural concrete did not differ from each other with regard to the concrete composition and structural design. The maximum grain size was approx. 32 mm. The aggregate consisted mainly of crushed, fine-crystalline, siliceous aggregates and sand, as is typical for northern Germany. The concrete contained neither siliceous chalk nor opaline sandstone. The concrete was well compacted. It contained relatively few coarse pores and air-voids due to the air-entraining process. However, in all the investigated core specimens cracks typical of ASR were found (Figure 2). Some of the cracks run through the aggregates, others along the contact zones. As a rule, these cracks run vertically from the upper face of the component to a depth of 50~100 mm and then incline into the direction parallel to the surface. On the fracture surfaces some fractures are visible, running through the coarse aggregates. The edges of the aggregates or the fracture surfaces running through them show reaction edges typical of ASR. The fine parts of the matrix have been torn out by the rolling over of tyres, which was probably caused by high traffic volumes, while the somewhat deeper embedded, coarser grains of sand have remained fixed in the matrix. There were no signs of structural loosening or sanding due to freeze‒thaw cycles on any of the surfaces. It is known from the literature (88. Mielich, O. (2019) Alkali-silica reaction (ASR) on German motorways: an overview. Otto-Graf-J. 18, 197‒208.
, 1919. Bérubé, M-A.; Chouinard, D.; Pigeon, M.; Frenette, J.; Boisvert, L.; Rivest, M. (2002) Effectiveness of sealers in counteracting alkali-silica reaction in plain and air-entrained laboratory concretes exposed to wetting and drying, freezing and thawing, and salt water. Can. J. Civ. Eng. 29 [2], 289‒300. https://doi.org/10.1139/l02-011.
) that, in the case of permanent access of alkali from de-icing agents and the presence of heavy traffic, which is of particular importance for the stresses of the road structure, the alkali-silica reaction may occur faster. When there is a simultaneous interaction of cyclic freezing and thawing, the concrete deterioration may be more severe. Bérubé et al. (1919. Bérubé, M-A.; Chouinard, D.; Pigeon, M.; Frenette, J.; Boisvert, L.; Rivest, M. (2002) Effectiveness of sealers in counteracting alkali-silica reaction in plain and air-entrained laboratory concretes exposed to wetting and drying, freezing and thawing, and salt water. Can. J. Civ. Eng. 29 [2], 289‒300. https://doi.org/10.1139/l02-011.
) revealed that, regardless of the air content, freezing and thawing greatly increased the concrete expansion in the presence of ASR, even after the ASR was almost complete; freezing and thawing also greatly promoted surface cracking.

Summarised results of the calculation of the DRI are presented in Figure 3. The average value of the DRI from six specimens was 384±32, which suggests that the concrete pavement was damaged due to ASR. Sanchez et al. (2020. Sanchez, L.F.M.; Fournier, B.; Jolin, M.; Mitchell, D.; Bastien, J. (2017) Overall assessment of Alkali-Aggregate Reaction (AAR) in concretes presenting different strengths and incorporating a wide range of reactive aggregate types and natures. Cem. Concr. Res. 93, 17‒31. https://doi.org/10.1016/j.cemconres.2016.12.001.
) and Fournier et al. (2121. Fournier, B.; Fecteau, P.L.; Villeneuve, V.; Tremblay, S.; Sanchez, L. (2015) Description of petrographic features of damage in concrete used in the determination of the DRI. Québec: Département de géologie et de génie géologique, Université Laval.
) suggested a classification of ASR damage exposed to a moist-warm storage for the DRI using the weighting factors proposed by Villeneuve et al. (1818. Villeneuve, V.; Fournier, B.; Duschene, J. (2012) Determination of the damage in concrete affected by ASR - The damage rating index (DRI). Proc. 14^th ICAARAustin, USA
), according to which the tested pavement concrete was moderately damaged as a result of ASR (3 on a five-point scale).

4.2. Mechanical properties

The compressive strength for cores cured in moist conditions (40°C, 100% RH) was 87.9±8.1 MPa, and for cores cured in laboratory conditions (23°C, 50% RH) - 82.9±11.2 MPa. The average value was 85.4 MPa. Since the strength of the differently pre-treated test specimens differed only slightly, in the following evaluation they were therefore considered as belonging to one and the same population. Taking into account the slimness and diameter of the specimens, these correspond to the compressive strengths of cubes with an edge length of 150 mm.

The determined gross densities were between 2.34 and 2.49 kg/dm³, with an average value of 2.39 kg/dm³. The modulus of elasticity was determined as the increase in the secant modulus from the third stress (including the pre-loading cycle) between the σ_u (upper) and σ_o (lower) stress levels used in the testing cycle, as summarised in the following Table 1.

The modulus of elasticity ranges from 17.1-28.8 GPa, with an average value of 23.1 GPa. The modulus of elasticity itself correlates significantly with the extent of the crack formation, which was described in Table 1 as a few, strong and severe damage due to ASR. Only the modulus of 28.8 and 27.6 GPa determined with two specimens: no. 2 and 8, come close to the values to be expected for a concrete with the strength of approx. 85.6 MPa presented here. For the other values, the modulus lies between 17.1 and 25.3 GPa. Considering the existing compressive strength, these values appear to be clearly too low. Probably the size and cracks network did not significantly affect the strength results, but the changes which suggest internal damage are visible in the values of the modulus of elasticity.

The average compressive strength (85.4 MPa) and the determined gross densities were 2.39 kg/dm³, which was within the range to be expected for concrete of the present quality (2222. Rostàsy, F.S. (1983) Baustoffe. Verlag W. Kohlhammer, Stuttgart - Berlin - Köln - Mainz. pp. 243.
). The calculated parameters according to DIN EN 13791 allow one to classify the investigated structural concrete into the compressive strength class C60/75 according to EN 206-1. Accordingly, it was a high-strength concrete. The average value of the modulus of elasticity was 23.1 GPa, which was below the value according to DIN 1045 (= 40 GPa) (2020. Sanchez, L.F.M.; Fournier, B.; Jolin, M.; Mitchell, D.; Bastien, J. (2017) Overall assessment of Alkali-Aggregate Reaction (AAR) in concretes presenting different strengths and incorporating a wide range of reactive aggregate types and natures. Cem. Concr. Res. 93, 17‒31. https://doi.org/10.1016/j.cemconres.2016.12.001.
). The above results are in line with (88. Mielich, O. (2019) Alkali-silica reaction (ASR) on German motorways: an overview. Otto-Graf-J. 18, 197‒208.
, 2323. Reinhardt, H-W; Mielich, O. (2012) Mechanical properties of concretes with slowly reacting alkali sensitive aggregates. Proc. 14^th ICAAR, Austin, USA.
), where the substantial reduction in mechanical properties, particularly the tensile strength and the static modulus of elasticity, as a result of ASR was shown. However, no significant compressive strength reductions due to damage caused by the ASR occurred during the moist warm storage. Instead, the concrete achieved a slightly higher strength during the moist warm storage. These increases in strength are generally attributable to the post-hydration process that takes place at these temperatures or to the bonding of crack surfaces with freshly formed ASR gel (2424. Bödeker, W. (2003) Alkalireaktion im Bauwerksbeton - Ein Erfahrungsbericht. Deutscher Ausschuß für Stahlbeton, Heft 539, Beuth Verlag GmbH, Berlin-Wien-Zürich, 1. Auflage.
).

4.3. Microscopic analysis

In the concrete cores during petrographic analysis, three main genetic rock types were found: sedimentary, metamorphic rock and mineral clasts (Figure 4). Four aggregate size fractions can be distinguished: 16‒32 mm: sandstone and silt, 4‒16 mm: sandstone, weakly metamorphosed silt shale (phyllite), limestone, quartz, amphibolite, 0.063‒4 mm: quartz, feldspar, limestone, chert, and below 0.063 mm: limestone, quartz, feldspar, amphibolite.

As coarse aggregate, sandstone (Figure 4a), a massive silt with aleuritic microstructure (Figure 4b) and weakly metamorphosed silt shale were found. The sandstone was characterised by well sorted grains with low sphericity, and subrounded. Plagioclase was present as well as carbonate crystals. As a binder, remains of primary clay matrix, secondary siliceous cement and carbonate cement were identified. In the silt, a primary clay matrix was abundant and the secondary matrix was cement. The silt shale was composed of quartz, carbonate, plagioclase, feldspar, muscovite and opaque mineral. As fine aggregate, grains of quartz and limestone were found as well as chert particles (Figure 4c). In most cases, the chert was formed by chalcedony and quartz, but it also contained carbonate.

The pavement concrete was air-entrained to ensure frost resistance. The air-void system was visible on the thin sections, but almost all the air-voids were partially or fully filled by ettringite (Figure 5). However, this should not significantly affect the concrete degradation caused by ASR.

In all the analysed specimens the characteristic ASR map of microcracking was visible, both on whole cores and on micro scale specimens. The microcracks were observed in the fine and coarse aggregate and cement matrix (Figures 6‒8). Most of the cracks were propagating from aggregate grains through the cement matrix (Figure 6). They were empty or lined with ASR gel. The cracks located in the coarse and fine aggregate contained Si-Ca-K-Na gel (Figures 6 and 7). The microscopic analysis did not reveal any de-icer ions in the CSH. There was no presence of either chloride ions, ASR gel or Friedel’s salt in the cement matrix.

The presence of ASR products in the cracks in the coarse and fine aggregate grains clearly indicates the relationship of these cracks and the cement matrix with the expansive alkali-silica reaction. It is assumed that the concrete pavement degradation was directly due to the presence of chert (above all chalcedony), siltstone (the potentially reactive element was the siliceous cement) and silt shale. The siliceous cement of siltstones has usually not been connected to a risk of ASR, but a microcrystalline fine-grained quartz in siltstones is potentially ASR susceptible and also generally deleterious in the aggregate for concrete, especially when associated with clay minerals (2525. Vola, G.; Berra, M.; Rondena, E. (2011) Petrographic quantitative analysis of ASR susceptible Italian aggregates for concrete. Proc. 13^th Euroseminar on Microscopy Applied to Building Materials (EMABM 2011).
). The quartz of the cement was very fine-grained and siltstones strongly dominated in the aggregate by volume. Sand grains were mainly monomineral. However, they also contained reactive forms of quartz, which was found using thin section analysis and confirmed by SEM analysis. The sand particles were partially or completely converted into ASR gel, which clearly confirms their ASR potential (2626. Jóźwiak-Niedźwiedzka, D.; Antolik, A.; Dziedzic, K.; Gméling, K.; Bogusz, K. (2021) Laboratory investigations on fine aggregates used for concrete pavements due to the risk of ASR. Road Mater. Pavem. Design. 22 [12], 2883‒2895, https://doi.org/10.1080/14680629.2020.1796767.
). The ASR ceases when the reactive aggregates, the hydroxyl, or the alkalis’ sodium and potassium ions are exhausted (2727. Fernandes, I.; Noronha, F.; Teles, M. (2004) Microscopic analysis of alkali-aggregate reaction products in a 50-year-old concrete. Mater. Charact. 53 [2-4], 295‒306, https://doi.org/10.1016/j.matchar.2004.08.005.
). And although the alkali-silica gel has a much lower Na content compared to K, which may suggest no effect of the NaCl de-icer (or the use of cement with a high K₂O content), the reactive minerals in the aggregate in combination with cement alkali and moisture caused significant damage to the concrete. No presence of Friedel’s salt was found, but its formation, which was assumed earlier to increase the sodium concentration in the concrete pore solution, turned out not to be required for ASR and does not result in an increase of the pH (88. Mielich, O. (2019) Alkali-silica reaction (ASR) on German motorways: an overview. Otto-Graf-J. 18, 197‒208.
). Bérubé et al. (2828. Bérubé, M.A.; Dorion, J.F.; Duchesne, J.; Fournier, B.; Vézina, D. (2003) Laboratory and field investigations of the influence of sodium chloride on alkali-silica reactivity. Cem. Concr. Res. 33, 77-84. https://doi.org/10.1016/S0008-8846(02)00926-2.
) showed that the high sodium concentration in the near-surface layer of the concrete might not cause severe ASR owing to a decreased OH^- concentration in this area. Heising et al. (2929. Heisig, A.; Urbonas, L.; Beddoe, R.E.; Heinz, D. (2016) Ingress of NaCl in concrete with alkali reactive aggregate: effect on silicon solubility. Mater. Struct. 49, 4291-4303. https://doi.org/10.1617/s11527-015-0788-y.
) stated that the chloride binding on hydrated cement phases and the corresponding release of OH^- ions does not significantly promote ASR.

Previous research on degraded pavement concrete (44. Frýbort, A.; Všianský, D.; Štulířová, J.; Stryk, J.; Gregerová, M. (2018) Variations in the composition and relations between alkali-silica gels and calcium silicate hydrates in highway concrete. Mater. Charact. 137, 91-108.
, 13-1513. Marfil, S.A.; Maiza, P.J. (2001) Deteriorated pavements due to the alkali-silica reaction: A petrographic study of three cases in Argentina. Cem. Concr. Res. 31, 1017-1021. https://doi.org/10.1016/S0008-8846(01)00508-7.
14. Hong, S.H.; Han, S.H.; Yun, K.K. (2007) A case study of concrete pavement deterioration by alkali-silica reaction in Korea. Int. J. Concr. Struct. Mater. 1 [1], 75‒81.
15. Glinicki, M.A.; Jóźwiak-Niedźwiedzka, D.; Antolik, A.; Dziedzic, K.; Dąbrowski, M.; Bogusz, K. (2022) Diagnosis of ASR damage in highway pavement after 15 years of service in wet-freeze climate region. Case Stud. Constr. Mater. 17, e01226. https://doi.org/10.1016/j.cscm.2022.e01226.
) presented similar observations regarding the ASR as a cause of its destruction; however, they mentioned other reactive aggregates. Rocks such as black shales, green schists and metabasalts used as concrete aggregate revealed the ASR potential (44. Frýbort, A.; Všianský, D.; Štulířová, J.; Stryk, J.; Gregerová, M. (2018) Variations in the composition and relations between alkali-silica gels and calcium silicate hydrates in highway concrete. Mater. Charact. 137, 91-108.
). Also a strained and microcrystalline quartz as well as volcanic glass were recognised as reactive components (1313. Marfil, S.A.; Maiza, P.J. (2001) Deteriorated pavements due to the alkali-silica reaction: A petrographic study of three cases in Argentina. Cem. Concr. Res. 31, 1017-1021. https://doi.org/10.1016/S0008-8846(01)00508-7.
), as well as black shale and sea-dredged natural sand (1414. Hong, S.H.; Han, S.H.; Yun, K.K. (2007) A case study of concrete pavement deterioration by alkali-silica reaction in Korea. Int. J. Concr. Struct. Mater. 1 [1], 75‒81.
) and a quartzite aggregate (1515. Glinicki, M.A.; Jóźwiak-Niedźwiedzka, D.; Antolik, A.; Dziedzic, K.; Dąbrowski, M.; Bogusz, K. (2022) Diagnosis of ASR damage in highway pavement after 15 years of service in wet-freeze climate region. Case Stud. Constr. Mater. 17, e01226. https://doi.org/10.1016/j.cscm.2022.e01226.
).

Alkali-silica reaction has been identified in more than 50 countries worldwide (22. Pyy, H.; Holt, E.; Ferreira, M. (2011) An initial survey on the occurrence of alkali aggregate reaction in Finland, Customer Report VTT-CR-00554-12, pp. 27.
), but it should be noted that the types of aggregates that caused ASR are very different at these sites. For example, in northern Europe, Danish aggregate is young sedimentary rock and Icelandic aggregate is young volcanic rock, while in Sweden and Norway, old crystalline rock is the main source of aggregate (22. Pyy, H.; Holt, E.; Ferreira, M. (2011) An initial survey on the occurrence of alkali aggregate reaction in Finland, Customer Report VTT-CR-00554-12, pp. 27.
).

However, one should also pay attention to the research methods. Seyfarth et al. (1616. Seyfarth, K.; Giebson, C.; Ludwig, H. (2022) ASR related service life estimation for concrete pavements. Mater. Construcc. 72 [346], e287. https://doi.org/10.3989/mc.2022.15921.
) stated that at highway sections built after 2005, a sufficiently low ASR potential was found. This has been explained by the effectiveness of the new regulations for preventing ASR in highway pavement concrete in Germany. The research results are consistent with the above statement.

The results for cores taken from the damaged concrete pavement, improved test methods and the development of national methods of preventing the alkali-silica reaction in concrete will contribute to the avoidance of reactive aggregate in the construction of durable concrete pavements.