Study of the alkali-silica reaction rate of Spanish aggregates. Proposal of a classification based in accelerated mortar bars tests and petrographic parameters

1. INTRODUCTION

Concrete is considered a durable material. However, environmental exposure, loads and the passage of time can cause material deterioration. According to the ACI Durability Guide (11. ACI 201.2R-08. (2008) Guide to durable concrete, reported by ACI committee 201, Ed. American Concrete Institute.
), one of the main causes of deterioration of concrete is the alkali-aggregate reaction. Until the 1990s it was considered that in Spain these aggregates were not potentially reactive and that there was no deterioration due to this expansive process, except for some dams with granitic aggregate (22. Soriano, J. (1987) Reactions d’interaction entre certains granulats et la phase interstitielle du beton, In Pore Structure and Materials Properties. 25-32. Chapman & Hall Ed., London, (1987).
, 33. Soriano, J.; García Calleja, M.A. (1989) Áridos reactivos. Acción del hidróxido cálcico sobre áridos silicatados. In III Congreso de Geoquímica España. 9-15. España, (1989).
). However, from the 90s of the 20th century, and mainly linked to accelerated curing in precast pieces, cases of significant deterioration in different concretes began to be identified in Spain (44. Menéndez, E. (1993a) Deterioro de materiales artificiales I. Reacción álcali-árido. La humedad como patología frecuente en la edificación. Colegio oficial de aparejadores y arquitectos técnicos de Madrid, 163-169. Madrid, (1993).
, 55. Menéndez, E. (1993b) Estudio microestructural de productos de reacción álcali-árido en hormigones curados a alta temperatura. Mater. Construcc. 43 [232], https://doi.org/10.3989/mc.1993.v43.i232.664.
). The late recognition of concrete with an alkali-aggregate reaction is due to the fact that Spanish aggregates are mostly slow reacting, a concept that had not been developed until then. The concept of reaction rate is introduced later, although it must be taken into account that aggregates have a strong local component (66. Menéndez, E. (2010) Análisis del hormigón en estructuras afectadas por reacción Árido-Álcali, ataque por sulfatos y ciclos Hielo-deshielo. Ed. IECA, España, (2010).
).

The petrographic analysis of thin sections, with polarized light with crossed Nicols, allows characterizing the reactivity of the aggregates. The quartz grains are analyzed to observe those that show undulatory extinction (reactive due to their high density of dislocations) or straight extinction of light or dark color (non-reactive). Bragg (77. Bragg, D. (1993) Alkali-aggregate reactivity in Newfoundland: Field and laboratory investigation. Newfoundland Department of Mines and Energy, Geological Survey Branch, Report. 93-1, 113-126. Canada.
, 88. Bragg, D. (1995) Petrographic examination of construction aggregates of Newfoundland. Department of Natural Resources. Geological Survey. Report 95-1, 77-104. Canada.
), developed a petrographic method to define the reaction rate of aggregates as a function of the quantity of quartz particles with undulatory extinction: < 1% non-reactive (very slow), 1-10% slight ratio (tendency to non-reactive), > 10% and < 20% moderate ratio (tendency to reactive) and > 20% fast ratio (reactive). For its part, B. Wigum (99. Wigum, B.J. (1995a) Examination of microstructural features of Norwegian cataclastic rocks and their use for predicting alkali reactivity in concrete. Eng. Geol. 40 [3-4], 195-214. https://doi.org/10.1016/0013-7952(95)00044-5.
), considers that the volume of highly reactive particles required to produce the expansion is very small, although the required amount of reactive particles to produce expansion in slow reacting aggregates cannot be established. According to Lagerblad and Trägård, and Wigum (1010. Lagerblad, B.; Trägårdh, J. (1992) Slowly reacting aggregates in Sweden - Mechanism and conditions for reactivity in concrete. Proc. 9th Int. Conf. Alkali-Aggregate Reaction in Concrete, Concrete Society Publication CS-104. 2, 570-578. London (1992).
, 1111. Wigum, B.J. (1995b) Ph D reactions in concrete properties, classification and testing of norwegian cataclastic rocks. University of Trondheim, Norway (1995).
) there is no pessimum effect for slow-reacting rocks, and 100% reactive particles may be required.

Petrographic analysis has been mainly used to analyze the potential reactivity of aggregates, based on the number of reactive particles. However, in general, petrography are usually only descriptive if they are not performed by petrologists with enough experience in ASR (alkali-silica reaction). For this reason, most of the ASR regulations require the analysis of the potential expansion of specimens in mortars or concrete. In addition, petrographic analysis is the most widely used method based on the ASTM C1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
) standard (UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
) in Spain). A comparative summary between different petrographic analysis standards and expansion-based test methods is described in (1414. Blight, G.; Alexander, M. (2011) Alkali-aggregate reaction and structural damage to concrete. engineering assessment, repair and management. CRC Press. Taylor & Francis Group, London, (2011).
, 1515. Furny, J.; Kerkhoff, B. (2007) Diagnosis and control of alkali-aggregate reactions in concrete, concrete technology. PCA R&D, Illinois.
).

On the other hand, Bragg and Foster (1616. Bragg, D.; Foster, K. (1992) Relationship between petrography and results of alkali-reactivity testing, samples from Newfoundland, Canada. The 9th International Conference on Alkali-Aggregate Reaction in Concrete, 127-135. London, UK (1992).
) investigated the relationship between the petrographic examination and the results of accelerated test of mortar bars (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
). Aggregates with alkali-reactive minerals between 15% and 40% were rated as good, tending to be reactive, while those with over 40% alkali-reactive rocks were rated as highly reactive. The agreement between petrography and accelerated mortar bar testing was found to be 83%, although this correlation is questionable for slow reactive aggregates.

Jensen (1717. Jensen, V. (2012) Reclassification of alkali aggregate reaction. In Proceedings of the 14th Conference on Alkali-Aggregate Reaction in Concrete, 10, Austin.
) suggests a classification of reactive minerals and rock constituents, divided into three groups according to their reactivity (very fast, fast and slow), based mainly on the lists in table A.1.2 and table A.1.3 of the recommendation of RILEM AAR-1.1 (1818. AAR-1.1 (2016) Detection of potential alkali-reactivity Part 1: petrographic examination method. RILEM Recommendations for the prevention of damage by alkali-aggregate reactions in new concrete structures. State-of-the-art report of the RILEM Technical Committee 219-ACS. Springer, Switzerland.
). For its part, Jensen, 2012 (1717. Jensen, V. (2012) Reclassification of alkali aggregate reaction. In Proceedings of the 14th Conference on Alkali-Aggregate Reaction in Concrete, 10, Austin.
), proposes the inclusion of an additional category of highly reactive aggregates, which would include fine dolomite crystals and expansive clay minerals, while granite aggregates are included in the slow reactions group, where structural effects may appear 10 years after the completion of a construction project.

G. Neto et al. (1919. Gomes Neto, D.P.; Conceiçãoc, H.; Carvalho Lisboa, V.A.; Soares de Santanaa, R.; Silva Barreto, L. (2014) Influence of granitic aggregates from northeast brazil on the alkali-aggregate reaction. Mater. Res. 17 [1], 51-58. https://doi.org/10.1590/S1516-14392014005000045.
) have carried out several studies trying to link the potential reactivity of aggregates with the dissolution of the silica of different aggregates (2020. Bulteel, D.; Garcia-Diaz, E.; Vernet, C.; Zanni, H. (2002). Alkali-silica reaction: A method to quantify the reaction degree. Cem. Concr. Res. 32 [8], 1199-1206. https://doi.org/10.1016/S0008-8846(02)00759-7.
, 2121. Gao, X.X.; Cyr, M.; Multon, S.; Sellier, A. (2013) A comparison of methods for chemical assessment of reactive silica in concrete aggregates by selective dissolution. Cem. Concr. Comp. 37, 82-94. https://doi.org/10.1016/j.cemconcomp.2012.12.002.
). He classified them into two groups, according to the reaction kinetics. The first group consists of vitreous or amorphous minerals such as volcanic glass and opal, in which the reaction develops very quickly. In the second group, which includes deformed crystalline minerals such as quartz deformed by tectonic processes, the reactions and expansions are slow (2020. Bulteel, D.; Garcia-Diaz, E.; Vernet, C.; Zanni, H. (2002). Alkali-silica reaction: A method to quantify the reaction degree. Cem. Concr. Res. 32 [8], 1199-1206. https://doi.org/10.1016/S0008-8846(02)00759-7.
, 2121. Gao, X.X.; Cyr, M.; Multon, S.; Sellier, A. (2013) A comparison of methods for chemical assessment of reactive silica in concrete aggregates by selective dissolution. Cem. Concr. Comp. 37, 82-94. https://doi.org/10.1016/j.cemconcomp.2012.12.002.
).

Granitic aggregates are widely used in concrete structures around the world and they can show different responses in terms of alkali-aggregate reaction, although they are generally considered slow-reacting (2222. ABNT. NBR 15577-1 (2008) Aggregates-Alkali-aggregate reactivity Part 1: Guide for the evaluation of potential reactivity of aggregates and preventive measures for its use in concrete, ABNT: Rio de Janeiro, (in Portuguese).
, 2323. Ponce, J.M.; Batic, O.R. (2006) Different manifestations of the alkali-silica reaction in concrete according to the reaction kinetics of reactive aggregate. Cem. Concr. Res. 36 [6], 1148-1156. https://doi.org/10.1016/j.cemconres.2005.12.022.
).

According to Jensen, Alaejos and Lanza (1717. Jensen, V. (2012) Reclassification of alkali aggregate reaction. In Proceedings of the 14th Conference on Alkali-Aggregate Reaction in Concrete, 10, Austin.
, 2424. Alaejos, P.; Lanza, V. (2012) Influence of equivalent reactive quartz content on expansion due to alkali silica reaction. Cem. Concr. Res. 42 [1], 99-104. https://doi.org/10.1016/j.cemconres.2011.08.006.
), the deformation of quartz with the development of subgrains and the presence of microcrystalline and cryptocrystalline phases, provide important characteristics for the evaluation in petrographic analysis. For its part, Wigum (99. Wigum, B.J. (1995a) Examination of microstructural features of Norwegian cataclastic rocks and their use for predicting alkali reactivity in concrete. Eng. Geol. 40 [3-4], 195-214. https://doi.org/10.1016/0013-7952(95)00044-5.
, 1111. Wigum, B.J. (1995b) Ph D reactions in concrete properties, classification and testing of norwegian cataclastic rocks. University of Trondheim, Norway (1995).
) studied deformed granite rocks, observing that the variables having the greatest influence on expansion are the presence of quartz subgrains, the total surface of the grain boundary and the size of the grains.

Tiecher et al. (2525. Tiecher, F.; Rolim, P.H.; Hasparik, N.P.; Dal, Molin, D.C.C.; Gomes, M.E.B.; Glieze, P. (2012) Reactivity study of Brazilian aggregates through silica dissolution analysis. In: Proceedings of the 14th Conference on Alkali-Aggregate Reaction in Concrete, 10, Austin.
) studied the dissolution of three Brazilian aggregates (granite, mylonite and quartzite) with different degrees of deformation. Grains with higher deformation (quartzite and mylonites), with marked deformation bands and undulatory extinction, dissolve more easily and produce greater expansions in the mortar bar test, compared to granite, which had a higher content of recrystallized quartz subgrains. Based on these results, it is considered that there is no conclusive proof regarding the influence of the aggregate size on the reactivity of quartz and quartz subgrains with smaller dimensions. Dolar-Mantuani, Sims et al. and Grattan-Bellew (2626. Dolar-Mantuani, L.M.M. (1981) Undulatory extinction in quartz used for identifying potentially reactive rocks. In Proceedings of Conference on Alkali-Aggregate Reaction in Concrete, 252 [36], 11, Cape Town, South Africa (1981).
, 2727. Sims, I.; Hunt, B.; Miglio, B. (1992) Quantifying microscopical examination of concrete for aar and other durability aspects. Spec. Public. 131, 267-287.
, 2828. Grattan-Bellew, P.E. (1986) Is high undulatory extinction in quartz indicative of alkali-expansivity of granitic aggregates?, In Proc. 7th International Conference on Concrete Alkali-Aggregate Reactions, Canada.
) also analyzed petrographically different species of SiO₂ and they related, qualitatively, the state of the quartz lattice with its reactivity, taking into account the undulatory extinction angles of quartz.

Prendes et al., and Menéndez et al. (2929. Prendes, N.; Menéndez, E. (2007) Digital image processing and MEB (BSE) Techniques in the identification and quantification of minerals phases present in cement and concrete. MRS Online Proc. Library 1026. 404. https://doi.org/10.1557/PROC-1026-C04-04.
, 3030. Menéndez, E.; García-Rovés, R.; Prendes, N. (2015) Metodología avanzada de evaluación petrográfica de áridos para predecir su potencial reactividad frente a los álcalis del hormigón, In Proc. IV Congreso Nacional de Áridos, Madrid.
), quantitatively analyzed the external perimeter of grain and the internal perimeters of contact between subgrains with deformation, defining an index of the reactivity of the quartzs (I_QR). This is a dimensionless index with values between 0 and 1, being the value to start considering reactivity <0.39, with I_QR values closer to 0 being more reactive.

Most of the European and international regulations base the clasification of the aggregates, with respect to the alkali-aggregate reactivity, in standards and procedures of petrographic characterization and in accelerated tests of mortar bars and concrete prisms. The standards used in Spain are: UNE-EN 932-3, UNE-EN 932-3/A1, UNE 146508, UNE 146509, UNE 83967, UNE 83968 and UNE 83969 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
, 31-3531. UNE-EN 932-3 and UNE-EN 932-3/A1:2004. (2004) tests for general properties of aggregates. part 3: procedure and terminology for simplified petrographic description. UNE: Madrid, Spain.
32. UNE 146509:2018. (2018) Determination of the potential reactivity of aggregates. Concrete prisms method; UNE: Madrid, Spain.
33. UNE 83967:2016 EX. (2016) Concrete durability. Test method for the assessment of potential expansion due to alkali-aggregate reactivity of concrete mixes. Method semi-accelerated of concrete prisms. UNE: Madrid, Spain.
34. UNE 83968: 2016 EX. (2016) Concrete durability. Evaluation of the expansion of mortar bars using potential reactive and non-reactive aggregate mixes, against alkali-silica and alkali-silicate. Accelerated method of mortar bars. UNE: Madrid, Spain.
35. UNE 83969:2017 EX. (2017) Concrete durability. Evaluation of the expansion of mortar bars using potential reactive binders and aggregates against alkali-silica and alkali-silicate. Accelerated method of mortar bars. UNE: Madrid, Spain.
). On the other hand, the prevention strategies for ASR applied in France, North America and Australia are collected in the Monograph Nº 430 of the CSIC (36), in addition to the recommendations of the Spanish EHE (3737. EHE-08. (2008) Instrucción del hormigón estructural, Centro de Publicaciones Secretaría General Técnica Ministerio de Fomento, España, (2008).
) and a proposal for a comprehensive prevention strategy for AAR. This proposal is based on a specific characterization of the concrete components and the test of real concrete mixtures, using semi-accelerated expansion methods (3636. Menéndez, E. (2019) Estrategia integral de prevención de la reacción árido-álcali, monografías del IETcc, 430. Ed. CSIC, Madrid.
).

Velasco-Torres et al. (3838. Velasco-Torres, A.; Alaejos, P.; Soriano, J. (2010) Comparative study of the alkali-silica reaction (ASR) in granitic aggregates. Estud. Geológ. 66 [1], 105-114, https://doi.org/10.3989/egeol.40133.091.
) carried out a comparative study of two granitic rocks extracted from two dams affected by alkali-silica reaction, with slow and fast reactions, respectively, classifying the reactivity of the rocks according to their geological term. They concluded that the reaction can be slow or fast for any given type of rock depending on its components and/or its microstructural characteristics. They also concluded that, for the time being, there are no sufficiently reliable methods to evaluate the reactivity of slow reacting aggregates, which are the most frequent in Spain.

In fast-reacting aggregates, the attack by the concrete pore solution begins to dissolve the microcrystalline quartz zones by contact with the cement paste. This generates a large amount of gel, which accumulates in the surrounding paste in a short period. In addition, there is also a slow reaction, caused by deformed or microcracked quartz. On the other hand, in slow-reacting aggregates, the concrete pore solution slowly enters the aggregate, mainly through microcracks and, to a lesser extent, through subgrain boundaries. The expansive gel fills the fissure system, giving rise to a silica solution.

Rocker et al. (3939. Rocker, P., Mohammadi, J., Sirivivatnanon, V.; South, W. (2015) Linking New Australian alkali silica reactivity tests to world-wide performance date. Proceedings of the Biennial National Conference of the Concrete Institute of Australia in conjunction with the 69th RILEM Week, 2015, pp. 502 - 513, Ed. Concrete Institute of Australia.
), indicate that the term slow reacting aggregates has been introduced since the 90s and it is widely used throughout the literature (40-4940. Shayan, A.; Morris, H.A. (2001) Comparison of RTA T363 and ASTM C1260 accelerated mortar bar test methods for detecting reactive aggregates. Cem. Concr. Res. 31 [4], 655-663. https://doi.org/10.1016/S0008-8846(00)00491-9.
41. Stark, D.; Morgan, B.; Okamoto, P. (1993) Eliminating or Minimizing Alkali-Silica Reactivity. Strategic Highway Research Program, Washington DC, (1993).
42. Bérubé, M.; Fournier, B. (1993) Canadian experience with testing for alkali-aggregate reactivity in concrete. Cem. Concr. Compos. 15 [1-2], 27-47. https://doi.org/10.1016/0958-9465(93)90037-A.
43. Shayan, A. (2007) Field evidence for inability of ASTM C 1260 limits to detect slowly reactive Australian aggregates. Aust. J. Civ. Eng. 3 [1], 13-26. http://doi.org/10.1080/14488353.2007.11463917.
44. Lindgård, J. (2011) RILEM TC 219-ACS-P: Literature survey on performance testing, SINTEF Building and Infrastructure, Norway, (2011).
45. Castro, N.; Sorensen, B.E.; Broekmans, M.A. (2012) Quantitative assessment of alkali-reactive aggregate mineral content through XRD using polished sections as a supplementary tool to RILEM AAR-1 (petrographic method). Cem. Concr. Res. 42 [11], 1428-1437. https://doi.org/10.1016/j.cemconres.2012.08.004.
46. Nixon, P.; Lane, S. (2006) Experience from testing of the alkali reactivity of European aggregates according to several concrete prism test methods. Partner Report 3. Norway.
47. Islam, M.S. (2010) Performance of Nevada’s aggregates in alkali- aggregate reactivity of Portland cement concrete. UNLV Thes. Dissert. Profess. Papers Capsto. 243. https://doi.org/10.34917/1452669.
48. ACI221.1R-98. (1998) State-of-the-art report on alkali-aggregate reactivity, Reported by ACI Committee 221, (1998).
49. Shayan, A. (2011) Aggregate selection for durability of concrete structures. Proceedings of the ICE-Construction Materials. 64 [3], 111-121. https://doi.org/10.1680/coma.900018.
). According to Shayan (4343. Shayan, A. (2007) Field evidence for inability of ASTM C 1260 limits to detect slowly reactive Australian aggregates. Aust. J. Civ. Eng. 3 [1], 13-26. http://doi.org/10.1080/14488353.2007.11463917.
), it is important to apply a reliable alkali-silica reaction test method, to provide the expansion limit for the classification of aggregates as “non-reactive”, “slowly reactive” or “reactive”. This classification was considered necessary due to the large number of cases of AAR observed over time, due to have been great damages in significant structures in Australia, because of use of aggregates with slow reaction rate by the presence of meta-basalts or granite gneisses (4949. Shayan, A. (2011) Aggregate selection for durability of concrete structures. Proceedings of the ICE-Construction Materials. 64 [3], 111-121. https://doi.org/10.1680/coma.900018.
).

Another suggested method is RILEM AAR-4 (5050. On behalf of the membership of RILEM TC 219-ACS, Nixon P.J., Sims I. (2016) RILEM Recommended test method: AAR-4.1-Detection of potential alkali-reactivity-60 °C Test method for aggregate combinations using concrete prisms. In: Nixon P., Sims I. (eds) RILEM Recommendations for the prevention of damage by alkali-aggregate reactions in new concrete structures. RILEM State-of-the-Art Reports, vol 17. Springer, Dordrecht. https://doi.org/10.1007/978-94-017-7252-5_6.
), which allows identifying the reactivity of slow-reacting aggregates (5151. Ponce, J.M.; Batic, O.R. (2006) Different manifestations of the alkali-silica reaction in concrete according to the reaction kinetics of reactive aggregate. Cem. Concr. Res. 2006; 36 [6], 1148-1156. https://doi.org/10.1016/j.cemconres.2005.12.022.
).

The Australian standard AS 1141.60.1 (5252. AS 1141.60.1:2014. (2014) Methods for sampling and testing aggregates Potential alkali-silica reactivity - Accelerated mortar bar method, AS: Australia.
) applies new limits to detect slow reacting aggregates, replacing the classification of “uncertain reactivity” with that of “slow reaction” in the accelerated expansion test of mortar bars. The results of the tests showed that the proposed limits could distinguish between harmless aggregates and slow reaction. This standard classifies aggregates as non-reactive (< 0.10% at 21 days), slow reactive (< 0.10% at 10 days and < 0.30% at 21 days) and reactive (≥ 0.10% at 10 days and ≥ 0.30% at 21 days), using the accelerated mortar bar method. On the other hand, the UNE 146508 (13) has the following limits: at 14 days (≤ 0.10% non-reactive and ≥ 0.20% reactive) and 28 days (≤ 0.20% non-reactive and ≥ 0.20% reactive); and between 14 and 28 days is doubtful with expansion between > 0.10% and < 0.20%.

In the present work, 68 Spanish aggregates and sands, mainly siliceous, have been analyzed. These aggregates come from quarries in operation and have mainly caused problems due to alkali-silica reactivity, this damage appearing as early as one year after the concrete manufacturing and up to 50 years. This is not only due to the reaction rate of the aggregates, but also to the type of curing, the environmental conditions of exposure and the geometry and volume of the concrete structure.

A petrographic characterization has been carried out and the reactivity index of the quartz has been determined. The open porosity has also been determined and accelerated expansion tests of mortar bars, extended in time up to one year, have been carried out.

With these parameters, the reaction speed of the aggregates has been analyzed and a reaction speed classification for Spanish aggregates has been proposed, based on the accelerated mortar bar test.

2. MATERIALS AND METHODS

2.1 Materials and localization

 ⌅

As it is mentioned before, the materials tested are 68 Spanish aggregates, mainly from quarries sited in areas of siliceous soils. In addition, most of the aggregates are siliceous gravels and sands. The denomination and properties of the samples are collected in Table 1.

Table 1.  Denomination of aggregates and their properties.

Denomination	Nº of order	Nature of aggregate (*)		Preparation type		Size of aggregate
00-X-Y-z	1 to 49	Siliceous (red)	S	Cracking	T	Sand	a
	50 to 53	Granitic (blue)	G	Rolling	R	Gravel	g
	54 to 56	Dolomitic (cyan)	D	Mix	M
	57 to 67	Limestone (green)	C
	68	Basaltic (orange)	B
(*) Code of colors in Figures, except in Figure 1

These aggregates are mainly located in the siliceous part of the soil of Spain. Their location is represented in the lithological map of the Península Ibérica (Iberian Peninsula) (Figure 1).

Figure 1.  Location of aggregates in the lithological map of the Atlas Nacional de España (Spanish National Atlas) (5353. de Terán, M.; Solé-Sabarís, L. (1978) Geografía general de España, Ed. Ariel, España, (1978).
).

A Portland cement CEM I-42.5 R was used to prepare mortars, following UNE-EN 197-1 and UNE-EN 196-2 standards (5454. UNE-EN 197-1:2011. (2011) Cement - Part 1: Composition, specifications and conformity criteria for common cements. (UNE): Madrid, Spain.
, 5555. UNE-EN 196-1:2018. (2018) Methods of testing cement - Part 1: Determination of strength. (UNE): Madrid, Spain.
). The chemical composition of this cement, expressed as the most stable oxide, is shown in Table 2.

Table 2.  Chemical composition of Portland cement CEM-I 42.5R.

Component	LOI1	SiO2	Al2O3	Fe2O3	CaO	SO3	MgO	K2O	Na2O	Na2Oeq1
Mass (%)	3.18	18.81	5.15	3.18	63.70	2.69	1.50	1.02	0.19	0.86

¹ LOI: Loss of ignition; Na₂O_eq = Na₂O + 0.658 K₂O

Mineralogical characterization was performed by X-ray diffraction (XRD). Thus, the crystalline compounds present in the cement were determined by XRD. Diffraction data of the cement were recorded using a D8 Advance powder crystal X-ray diffractometer (Bruker) with 2.2 kV Cu anode ceramic X-ray tube. Crystalline compounds were identified with the DIFFRAC.EVA v4.2.1 software, which supports a reference pattern database, derived from the Crystallography Open Database (COD) for phase identification. The semi quantitative analysis of cement showed Alite (C₃S), Belite (C₂S) and Brownmillerite (C₄AF)as major crystal compounds and Gypsum (CaSO₄·2H₂O) as a minor compound. All of them are characteristic components of a Portland cement.

2.2. Test methods

 ⌅

2.2.1. Open porosity

 ⌅

The open porosity (accessible to water) and the density were determined following the standard process UNE 83980:2014 (5656. UNE 83980:2014. (2014) Concrete durability. Test methods. Determination of the water absorption, density and accessible porosity for water in concrete. (UNE): Madrid, Spain.
) and calculated according to Equation [1]:

$Open porosity,\%=\frac{m_2-m_1}{m_2-m_3}.100$  [1]

where m₁ is the weight of the sample after drying (110°C ± 5°C for 24h), m₂ is the weight of the sample after vacuum conditions and m₃ is the apparent weight of the mortar sample (hydrostatic weight, i.e. underwater weighing).

This analysis allows measuring the relative quantity of aggressive entry in the different aggregates analyzed.

2.2.2. Petrography

 ⌅

The aggregates were observed by optical microscopy using 25 µm thick sections. These sections were prepared in the Instituto Geominero of Spain. The analysis was done by means of polarized microscopy in transmission mode using an OLYMPUS BX51 microscope, with objectives 4x, 10x, 20x and 40x. The treatment of the photographs was done using the software Analysis docu (Olympus Soft Imaging Solutions GmbH).

The petrographic analysis is used mainly to identify the mineral phases and their morphology. In the case of the alkali-silica reaction it is possible to analyze the presence characteristics of reactive minerals, for instance, the undulatory extinction of the quartz and the degree of it. The quartz, depending on its crystallographic characteristics, may contribute with siloxane groups to the interstitial solution, favoring the nucleation of neo-formed phases and the process of alkali-silica reaction.

With respect to the classification of reactivity of the aggregates the criteria of the RILEM Recommendation AAR-1.1 (1818. AAR-1.1 (2016) Detection of potential alkali-reactivity Part 1: petrographic examination method. RILEM Recommendations for the prevention of damage by alkali-aggregate reactions in new concrete structures. State-of-the-art report of the RILEM Technical Committee 219-ACS. Springer, Switzerland.
) is used. The potential reactivity of the aggregate is classified in terms of alkaline reactivity due to the presence of mineralogical phases potentially reactive. The classification is based in three classes:

• Class I: Unlikely reactivity.

• Class II: Uncertain reactivity. The aggregate cannot be classified directly in class I or III.

• Class III: Reactivity very likely.

2.2.3. Quartz reactivity index

 ⌅

The quartz reactivity index is a dimensionless parameter that is defined as ratio between the quartz perimeter and the total perimeter of quartz subgrains. It is usually calculated in quartz with undulatory extinction, Equation [2]:

$I_{Qr}=\frac{P_{ext}}{\sum P_{int}}$  [2]

where, I_Qr is the Index of reactivity of quartz, P_ext is the external perimeter of the quartz grain in µm andΣ P_int is the summation of internal perimeters of the sub-grains of quartz, in µm.

The petrographic classification was done according with the standard ASTM C294 (5757. ASTM C294-19. (2019) Standard descriptive nomenclature for constituents of concrete aggregates. ASTM International, West Conshohocken, PA. https://doi.org/10.1520/C0294-19.
), with a classification more detailed than the standard UNE-EN 932-3 (3131. UNE-EN 932-3 and UNE-EN 932-3/A1:2004. (2004) tests for general properties of aggregates. part 3: procedure and terminology for simplified petrographic description. UNE: Madrid, Spain.
). The grains of quartz were analyzed in angles of 11º with crossing Nicols and the images were combined to obtain an image with the borders of sub-grains. Later, the perimeter and the length borders were measured using the program ImageJ (Wayne Rasband, National Institutes of Health, USA), this software was used to carry out the picture treatment, the extraction of information and the count. Treatment of information was done using Microsoft Excel.

A scheme of the calculation process of I_Qr is shown in Figure 2.

Figure 2.  Numerical calculation of the IQr, measured with the ImageJ (2929. Prendes, N.; Menéndez, E. (2007) Digital image processing and MEB (BSE) Techniques in the identification and quantification of minerals phases present in cement and concrete. MRS Online Proc. Library 1026. 404. https://doi.org/10.1557/PROC-1026-C04-04.
, 3030. Menéndez, E.; García-Rovés, R.; Prendes, N. (2015) Metodología avanzada de evaluación petrográfica de áridos para predecir su potencial reactividad frente a los álcalis del hormigón, In Proc. IV Congreso Nacional de Áridos, Madrid.
).

The values of the I_Qr vary between 0 and 1. The qualification of the grains of each grain of quartz could be:

• I_Qr ≥ 0,39: Non-reactive particles

• I_Qr < 0,39: Potential reactive particles. The lowest values are the most reactive.

A representative number of particles of quartz, between 10 and 15, must be analyzed in each aggregate. The average value is the result for I_Qr of an aggregate (2929. Prendes, N.; Menéndez, E. (2007) Digital image processing and MEB (BSE) Techniques in the identification and quantification of minerals phases present in cement and concrete. MRS Online Proc. Library 1026. 404. https://doi.org/10.1557/PROC-1026-C04-04.
, 3030. Menéndez, E.; García-Rovés, R.; Prendes, N. (2015) Metodología avanzada de evaluación petrográfica de áridos para predecir su potencial reactividad frente a los álcalis del hormigón, In Proc. IV Congreso Nacional de Áridos, Madrid.
).

2.2.4. Expansion by UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
) similar to ASTM 1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
)

 ⌅

The expansion is, along with the petrography, the most frequent parameter to analyse the potential reactivity of the aggregates. In addition, the accelerated mortar bar test is the most used due to the short duration and flexibility.

The alkali-aggregate test method used in this research work is detailed in the Spanish standard UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
), equivalent to ASTM C1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
). In preparing the mortar bar specimens, the coarse aggregates were washed, dried (105 °C ± 5 °C), crushed, and sieved into the five fractions (from 0.160 mm to 5 mm), as per the requirement of UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
). Then, the potential reactivity of the aggregates in the mortars was evaluated in three mortar prisms (2.5 × 2.5 × 28.5 cm³) for each aggregate. They were prepared mixing 400 g of cement (CEM-I 42.5R) and 900 g of aggregate with a water-to-cement ratio (by weight) of 0.47, and the graded aggregates to total cement ratio (by weight) of 2.25. Special moulds were used with a stainless-steel gauge stud into both ends of the longitudinal section of the prism. The effective gauge length was 254 ± 2.5 mm. The mortar test specimens were demoulded after 24 hours, and then, stored immersed in water in closed containers which were placed in an oven maintaining the temperature of 80 °C ± 1 °C for 24 h. Thereafter, they were removed from the containers for which the zero readings were recorded. Afterward, the prisms were submerged in the 1 N NaOH soak solution at 80 °C ± 1 °C in plastic containers held in an oven at 80 °C ± 1 °C for a further fourteen days. Subsequent expansion readings were made from 2 to 14 days in accordance with UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
). However, additional readings were measured over 90-days and 365-days periods instead of the conventional 14-days period prescribed by the standard. Mortar expansion was calculated according to Equation [3], and the average expansion of the three prisms for each exposure time is given as the mortar expansion result:

$Mostrar bar expansion \left(\%\right)=\frac{(L_n-L_0)}{L_c}\times 100$  [3]

where L_n is the length at the testing time, L₀ is the initial length after 24 h of water immersion at 80 °C ± 1 °C, L_c is the calibration length (L_c = 254 mm according to UNE 80113 (5858. UNE 80113:2013. (2013) Test methods of cements. Physical analysis. Determination of the autoclave expansion. (UNE): Madrid, Spain.
))

The ASR classifications of the aggregates were evaluated based on the 14-day expansion upper limit of mortar bars of 0.10% (non-reactive aggregate), prescribed by the Spanish standard UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
), similar to the ASTM C1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
). Expansions over 0.20% indicate a potentially reactive aggregate. Nevertheless, additional 28-day expansion evaluation is made when the 14-day expansion results are between 0.10% and 0.20%. If it is beyond 0.20% at 28 days, the aggregate is considered potentially reactive.

This test methodology is used as a reference to do a classification of the potential reactivity of the aggregates. However, it has the limitation of not detecting the potential reactivity of aggregates with pessimum effect, even for aggregates known to be slowly reactive (e.g. granites). The ASR classifications of the aggregates were evaluated based on the 14-day expansion upper limit of mortar bars of 0.10% (non-reactive aggregate), prescribed by the UNE 146508 standard (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
). Expansions over 0.20% indicate a potentially reactive aggregate. Nevertheless, additional 28-day expansion evaluation is made when the 14-day expansion results are between 0.10% and 0.20%. If it is beyond 0.20% at 28 days, the aggregate is considered potentially reactive.

2.2.4.1. Expansion AS 1141.60.1 (5252. AS 1141.60.1:2014. (2014) Methods for sampling and testing aggregates Potential alkali-silica reactivity - Accelerated mortar bar method, AS: Australia.
)

 ⌅

The methodology of this test is the same as the previous one, but with a different test time, in this case the duration of the test is 21 days. This standard classifies aggregates as non-reactive (<0.10% at 21 days), slowly reaction (< 0.10% at 10 days and < 0.30% at 21 days) and reactive (≥ 0.10% at 10 days and ≥ 0.30% at 21 days).

3. RESULTS AND DISCUSSION

3.1. Open porosity

 ⌅

The open porosity is determined in each of the 68 aggregates, using a representative sample of them. The results are grouped by type of aggregate. Figure 3 presents the dispersion graphic, with the mean value of porosity indicated by a solid point, for each type of aggregate.

Figure 3.  Open porosity of each type of aggregates, (basalt isn´t included, because there is a single result).

The values of open porosity can be classified as follows: low ≤ 5%, medium > 5% and 10% and elevated when a value is > 10%, according to the standard UNE 83980:2014 (5656. UNE 83980:2014. (2014) Concrete durability. Test methods. Determination of the water absorption, density and accessible porosity for water in concrete. (UNE): Madrid, Spain.
).

It should be noted the low porosity of the granites that, although there are few samples, have very little dispersion. This low porosity makes it difficult for aggressive ions to enter the aggregate grains, which justifies the low reaction rate. The basalt isn´t represented in the Figure due to the fact that there is just a single sample, its porosity being 7.9%.

For its part, dolomitic and limestone aggregates have high porosity and, consequently, if they have reactive siliceous particles inside, they will react relatively quickly. With regard to siliceous aggregates, they show a lot of dispersion, although the average value is similar to limestone and dolomites. However, their dispersion can be attributed to the large number of samples of different types and origins.

The relationship between crushed and pebble gravel and silica sand has also been analysed. In principle, crushed gravel and sand should be more porous than pebble sand. However, a clear relationship has not been found in either gravel or sand. This behaviour is attributed to the homogeneity of the crushed aggregates and the different origins of the aggregates.

3.2. Petrography and Quartz Reactivity Index

 ⌅

In the petrographic analysis, the typology of the aggregates is characterized and its potentially reactive phases are quantified. In addition, the I_Qr of the quartz particles is calculated, classifying these in terms of their reactivity.

Table 3 shows the petrographic classification of aggregates, according to ASTM C295 (5959. ASTM C295 / C295M-19. (2019) Standard guide for petrographic examination of aggregates for concrete.
) and RILEM AAR-1.1 (1818. AAR-1.1 (2016) Detection of potential alkali-reactivity Part 1: petrographic examination method. RILEM Recommendations for the prevention of damage by alkali-aggregate reactions in new concrete structures. State-of-the-art report of the RILEM Technical Committee 219-ACS. Springer, Switzerland.
). In addition, the I_Qr values and the reticular state of the quartz are collected.

Table 3.  Petrographic classification of aggregates, according ASTM C295 (5959. ASTM C295 / C295M-19. (2019) Standard guide for petrographic examination of aggregates for concrete.
) and RILEM AAR-1.1 (1818. AAR-1.1 (2016) Detection of potential alkali-reactivity Part 1: petrographic examination method. RILEM Recommendations for the prevention of damage by alkali-aggregate reactions in new concrete structures. State-of-the-art report of the RILEM Technical Committee 219-ACS. Springer, Switzerland.
). Index of reactivity of quartz (Q) and reticular state of them.

Sample	Petrographic classification ASTM C295	Petrographic classification RILEM AAR-1.1	IQr	Reticular state of the Q
01STa	Siliceous sand	Class II-S	0.55	Q with undulatory extinction, Q microcrystalline and cryptocrystalline in chert, clays and others
02STa	Siliceous sand	Class III-S	0.23	Q with undulatory extinction, Q microcrystalline and cryptocrystalline in matrix
03STa	Siliceous sand	Class II-S	0.65	Q stable monocrystalline, Q deformed and chert
04STa	Siliceous sand	Class II-S	0.78	Q with undulatory extinction in low proportion
05STa	Siliceous sand	Class II-S	0.67	Predominant Q stable monocrystalline, Q polycrystalline, Q slightly deformed
06STa	Siliceous sand	Class III-S	0.37	Predominant Q stable monocrystalline, Q polycrystalline, Q slightly deformed
07STa	Siliceous sand	Class III-S	0.31	Predominant Q monocrystalline deformed, Q microcrystalline, cherts and clays
08STa	Siliceous sand	Class II-S	0.46	Predominant Q monocrystalline, some limestone and Q microcrystalline and cherts
09STa	Siliceous sand	Class III-S	0.52	Undulatory extinction Q polycrystalline
10STg	Gravel milonite	Class III-S	0.11	Q with undulatory extinction, Q microcrystalline and cryptocrystalline
11STg	Siliceous gravell	Class III-S	0.21	Undulatory extinction, Q polycrystalline
12STg	Siliceous gravell	Class II-S	0.67	Q deformed in sedimentary rocks
13STg	Opal	Class III-S	0.001	Q without crystalline structure defined and Q cryptocrystalline filling partially holes
14STg	Siliceous gravel	Class III-S	0.36	Undulatory extinction, Q polycrystalline
15STg	Siliceous gravel	Class III-S	0.23	Q with undulatory extinction, Q microcrystalline and cryptocrystalline in matrix
16STg	Calcedony	Class III-S	0.001	Q microcrystalline and cryptocrystalline
17STg	Siliceous gravel	Class II-S	0.57	Q with undulatory extinction in quartzite, y Q microcrystalline y saccharoids
18STg	Siliceous gravel	Class III-S	0.31	Predominant Q polycrystalline microcrystalline and deformed
19STg	Siliceous gravel	Class III-S	0.30	Predominant Q monocrystalline and Q polycrystalline microcrystalline
20STg	Siliceous gravel	Class III-S	0.20	Predominant Q deformed and Q
21STg	Siliceous gravel	Class II-S	0.53	Predominant Q monocrystalline , some limestone, Q microcrystalline, cherts
22STg	Siliceous gravel	Class III-S	0.55	Q with undulatory extinction in quartzite, fragments of chalcedony high reactive
23STg	Siliceous gravel	Class III-S	0.30	Undulatory extinction Q and Q microcrystalline
24STg	Siliceous gravel	Class II-S	0.47	Undulatory extinction Q with Q microcrystalline
25STg	Siliceous gravel	Class III-S	0.39	Undulatory extinction Q with Q microcrystalline and cryptocrystalline in cherts, clays and others
26SRa	Siliceous sand	Class II-S	0.69	Q with strength extinction and some with undulatory extinction
27SRa	Siliceous sand	Class II-S	0.56	Predominant Q stable monocrystalline, and some Q deformed and chert
28SRa	Siliceous sand	Class II-S	0.61	Predominant Q stable monocrystalline, and some Q deformed and chert
29SRa	Siliceous sand	Class III-S	0.47	Q stable monocrystalline, with frequent Q deformed
30SRa	Siliceous sand	Class III-S	0.21	Undulatory extinction Q in polycrystalline aggregates
31SRa	Siliceous sand	Class II-S	0.63	Predominant Q stable monocrystalline, some Q deformed and chert
32SRa	Siliceous sand	Class II-S	0.73	Predominant Q stable monocrystalline, and Q deformed in polycrystalline particles and some chert
33SRa	Siliceous sand	Class III-S	0.65	Predominant Q stable monocrystalline, and deformed Q and chert
34SRa	Siliceous sand	Class III-S	0.41	Q with undulatory extinction, Q cryptocrystalline in cherts, clays and others
35SRa	Siliceous sand	Class III-S	0.34	Q with undulatory extinction, Q microcrystalline and saccharoids in cherts, clays and others
36SRa	Siliceous sand	Class III-S	0.38	Predominant Q monocrystalline and Q polycrystalline, with particles chert
37SRa	Siliceous sand	Class II-S	0.62	Q stable and undulatory extinction in polycrystalline Q
38SRa	Siliceous sand	Class III-S	0.49	Q stables, undulatory extinction Q in polycrystalline
39SRa	Siliceous sand	Class II-S	0.71	Predominant Q stable monocrystalline, and some Q deformed in polycrystalline
40SRa	Siliceous sand	Class II-S	0.61	Predominant Q stable monocrystalline, with some Q deformed in polycrystalline and chert
41SRg	Siliceous gravel	Class II-S	0.64	Q polycrystalline, Q stable, others polycrystalline with undulatory extinction and Q microcrystalline
42SRg	Siliceous gravel	Class III-S	0.36	Undulatory extinction Q, Q microcrystalline and cryptocrystalline in cherts, clays and others
43SRg	Siliceous gravel	Class III-S	0.34	Undulatory extinction Q, Q microcrystalline and cryptocrystalline in cherts, clays and others
44SRg	Siliceous gravel	Class III-S	0.46	Undulatory extinction Q, Q cryptocrystalline in cherts, clays and others
45SRg	Siliceous gravel	Class III-S	0.51	Q polycrystalline with stable Q stables, others polycrystalline with undulatory extinction and Q microcrystalline
46SMa	Siliceous sand	Class II-S	0.62	Q with straight extinction, and some with undulatory extinction
47SMa	Siliceous sand	Class III-S	0.39	Undulatory extinction Q, Q microcrystalline and cryptocrystalline in cherts, clays and others
48SMa	Siliceous sand	Class III-S	0.37	Undulatory extinction Q, Q microcrystalline and cryptocrystalline in cherts, clays and others
49SMg	Siliceous gravel	Class III-S	0.36	Undulatory extinction Q in polycrystalline
50GTg	Granodiorite	Class II-S	0.41	Q deformed with ramified cracks and borders with ramified cracks and with sutured edges
51GTg	Granite	Class III-S	0.37	Crystals micro granular, cracking saccharoids
52GTg	Granodiorite	Class II-S	0.52	Micro cracking and cracks intergranular
53GTg	Granodiorite	Class II-S	0.46	Micro cracking and cracks intergranular
54DTa	Calcite dolomite	Class I	0.90	Low proportion of Q with straight extinction
55DTa	Dolomitic sand	Class I	0.90	Low proportion of Q with straight extinction
56DTg	Dolomite gravel	Class I	0.90	Low proportion of Q with straight extinction
57CTa	Calcite sand	Class I	0.90	Q with straight extinction in low proportion
58CTa	Calcite sand	Class I	1.00	Q it isn´t observed
59CTg	Limestone	Class I	0.90	Q with straight extinction in low proportion
60CTg	Calcite gravel	Class I	1.00	Don´t have Q
61CTg	Calcite gravel	Class I	1.00	It isn´t observed Q deformed
62CTg	Calcite sand	Class I	1.00	It isn´t observed Q deformed
63CTg	Calcite sand	Class I	0.89	Microgranular crystals with straight extinction
64CRa	Calcite sand	Class II-S	0.79	Undulatory extinction Q, Q microcrystalline in very low proportion
65CRa	Calcite sand	Class II-S	0.64	Undulatory extinction Q in very low proportion
66CRg	Calcite gravel	Class II-S	0.54	Undulatory extinction Q in very low proportion
67CMa	Calcite sand	Class I	0.90	Q with straight extinction in low proportion
68BTg	Basalt	Class II-S (*)	0.90	Q with straight extinction in low proportion
(*) Basaltic rocks have small crystals of quartz, and this affect to the classification of RILEM AAR-1.1

Most of the gravels and siliceous sands have quartz particles with undulatory extinction and many of them show I_Qr values lower than 0.4%. This implies a high reactivity. The granites show a moderate presence of reactive phases and I_Qr values at the limit of reactivity. This corroborates the slow reaction kinetics of this type of aggregate and the large volume of concrete necessary for the alteration evidence to be significant.

Regarding the classification of RILEM AAR-1.1 recommendation, all siliceous aggregates are classified as potentially reactive (Class III) or as uncertain reactivity (Class II). However, limestone, dolomite and several granites that have a low percentage of potentially reactive phases are also classified as uncertain reactivity. These results indicate that this classification method is not very precise since it only detects one of the limestone aggregates as non-reactive. Although, in practice, the limestone aggregates have not been found to have expansion, while all the granites tested presented alkali-silica reaction in field concrete structures (mainly dams and bridges).

Figure 4 shows the classification of aggregates according to the classification of the RILEM AAR1.1 recommendation, the percentage of reactive phases analyzed by petrography and the relationship between both methods.

Figure 4.  % potentially reactive phases versus AAR1.1 classification.

It is observed that the classification of the RILEM recommendation AAR1.1 is capable of detecting potentially reactive aggregates (Class III) quite rigorously. However, many of the aggregates that have a high percentage of potentially reactive particles are classified as doubtful, although, in practice, these aggregates show reactivity in construction sites (such as granites and many of the siliceous aggregates).

Regarding the I_Qr, it can be seen in Figure 5 that most of the siliceous aggregates would be classified as potentially reactive. However, the dispersion of the results is high, so they could be classified as non-reactive, although they mostly have a slow or very slow reaction rate. For its part, granites have little dispersion, mainly due to the low number of samples and low porosity. In the I_Qr classification, the granites would be very close to the reactivity limit, so this parameter (I_Qr) is better for classifying granites than other test methods, although the reactivity range should be extended to 0.55. This would also allow the inclusion of reactive siliceous aggregates that are not classified as such with the limit of 0.39 of the I_Qr value. Figure 5 shows the average results and the dispersion of the different types of aggregates analyzed.

Figure 5.  Reactive quartz index (IQr) of the aggregates (Basalt isn´t included because there is a single result).

If the percentage of reactive phases is represented versus the I_Qr parameter (Figure 6), it is verified that, in general, there exists a very good correlation for siliceous, granitic and limestone aggregates, although each type of aggregate shows different slopes. The largest dispersions are produced for siliceous aggregates with a high percentage of potentially reactive phases, since the I_Qr will underestimate them.

Figure 6.  % potentially reactive phases versus reactivity index of aggregates.

3.3. Expansion

 ⌅

The expansion has been analyzed under the test conditions indicated in the ASTM C1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
) standard (similar to the UNE 146508 (1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
)), although the measurements have been prolonged up to 1 year, in order to analyze the long-term behavior and observe the trend changes in the expansion of aggregates over time.

Figure 7 shows the expansion up to 28 days of testing, as well as the limits to consider aggregates as reactive according to ASTM C1260 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
) and AS 1141.60.1 (5252. AS 1141.60.1:2014. (2014) Methods for sampling and testing aggregates Potential alkali-silica reactivity - Accelerated mortar bar method, AS: Australia.
). For its part, Figure 8 shows the expansion results of the aggregates up to 365 days. In the first case, it is observed that only non-reactive aggregates stabilize their expansion before 28 days, while in the 365 days test most of the siliceous and granitic aggregates and some of the limestone continue to expand, with different reaction rates. It is found that granites and limestone with silicon inclusions expand much more slowly than siliceous aggregates.

Figure 7.  Expansion at 28 days.

Figure 8.  Expansion at 365 days.

Observing in detail the expansion up to 365 days, it can be seen that, before the first 150 days of testing, there is a change in trend in the expansion curves of the reactive aggregates. Figure 9 shows the mean value and the dispersion of the number of days elapsed until the change in trend in the expansion curve by type of aggregate. If this change in trend is analyzed for each type of aggregate, it is observed that it occurs around 40 days for siliceous aggregates, around 60 days for granites and over 65 days for limestone aggregates. This is associated with the reaction rate in aggregates, the fastest being siliceous, followed by granites and limestone with quartz inclusions.

Figure 9.  Change in reaction speed of each aggregate (basalt isn´t included because there is a single result).

If we takes a limit of 0.25% expansion at 365 days, to consider the aggregates as potentially reactive, 7 siliceous aggregates, 7 limestone aggregates and all the dolomites and the basalt are qualified as non-reactive. Moreover, 42 siliceous, 4 limestone and the 4 granites aggregates are considered as potential reactive.

Definition of Expansion Rate as a Function of the Porosity, Petrographic Analysis and Quartz Reactive Index (I_Qr) and Accelerate Mortar Bar Test.

The open porosity of each aggregate has been plotted versus the reactivity index of quartz in order to analyse the possible relationship between high porosity and high reactivity, but no significant relationship is observed due to the different nature of the aggregates (Figure 10). It should be noted the high porosity of some siliceous, limestone and dolomitic sands, which is associated with the retention of water between small particles.

Figure 10.  Open porosity versus reactivity index of quartz of each aggregate.

In Figure 11, the porosity versus the expansion of each aggregate at 14 days, 28 days and 365 days is represented. In each bar, the data on the left corresponds to the expansion at 14 days; the one in the middle shows the expansion at 28 days and the one on the right the expansion at 365 days.

Figura 11.  Expansion at 14 days, 28 days and 365 days vs porosity.

Although the hydroxyl ions have to be in contact with the aggregate to start the reaction, in the case of siliceous aggregates there is no clear relationship between open porosity and reaction rate over time. However, in granite aggregates, with very low porosity, a slowdown of the reaction is observed due to this characteristic. The limestone aggregates are mostly non-expansive and those that show expansion are due to the inclusions of deformed quartz.

On the other hand, in the case of siliceous aggregates with very low porosity, the alteration of the aggregates occurs from the homogeneous surface, so they initially show low expansion (66. Menéndez, E. (2010) Análisis del hormigón en estructuras afectadas por reacción Árido-Álcali, ataque por sulfatos y ciclos Hielo-deshielo. Ed. IECA, España, (2010).
, 6060. Idorm, G.M.; Johansen, V.; Thaulow, N. (1992) Assessment of causes of cracking in concrete. Material Science in Concrete III. American Ceramic Society, New York.
); although at 365 days they show a high expansion, like the rest of siliceous aggregates.

Monograph No. 230 of the IETcc-CSIC (3636. Menéndez, E. (2019) Estrategia integral de prevención de la reacción árido-álcali, monografías del IETcc, 430. Ed. CSIC, Madrid.
) contains a proposal on the qualification of the reaction speed, defining five groups:

• Non-reactive aggregates (<0.10% at 14 days or> 0.20% at 28 days, according to the limits of ASTM C 1260 and UNE 146528 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
  ,1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
  )).

• Slow reaction rate aggregates (0.10 to 0.20% at 14 days or between 0.20% and 0.25% at 28 days).

• Moderate reaction aggregates (0.25% to 0.35% between 14 days 28 days).

• Fast reaction aggregates (0.35% to 0.45% between 14 days 28 days)

• Very fast reaction aggregates (>0.45% between 14 days 28 days).

The expansion up to 28 days is represented in the scheme with the reaction rate rating described above (Figure 12), and it is found that most of the slow-reacting aggregates cannot be detected with the usual classification of the Standard Test Method for Potential Alkali Reactivity of Aggregates (ASTM C 1260 and UNE 146528 (1212. ASTM C1260-21. (2021) Standard test method for potential alkali reactivity of aggregates (mortar-bar method). ASTM International, West Conshohocken, PA, (2021) https://doi.org/10.1520/C1260-21.
,1313. UNE 146508:2018. (2018) Test for aggregates. determination of the alkali-silica and alkali-silicate potential reactivity of aggregates. accelerated mortar bar test. UNE: Madrid, Spain.
)). This method would classify as non-reactive all granites, limestones with siliceous inclusions, and more than a quarter of siliceous aggregates, all of which have shown reactivity on site. The potential reactivity of these aggregates appears with the extension of the test time.

Figure 12.  Classification of the aggregates according to the proposal of qualification at 14 days and at 28 days.

Representing the expansion of each aggregate at 14 days and 28 days versus the amount of quartz multiplied by the reactivity index of this quartz, we can divide the aggregates by their reaction rate into three groups: non-reactive, slowly reaction and reaction fast. This classification is carried out based on the total percentage of reactive quartz (I_Qr·Q_t) and the expansion value at 14 days and 28 days. The fast reaction would have a value greater than 30% of the total percentage of reactive aggregates and a percentage of expansion at 28 days > 0.20%. The slow reaction ones between 5% and 30% and a percentage of expansion at 14 days > 0.10%, the non-reactive ones would be those with a total percentage of reactive quartz ≤ 5% and a 28-day expansion percentage < 0.20%.

In Figure 13, the total percentage of reactive quartz (I_Qr·Q_t) versus the expansion value at 14 days and 28 days is shown. Moreover, in Figure 14 the total percentage of reactive quartz (I_Qr·Q_t) versus the expansion value at 14 days, 28 days and 365 days is shown. Most of the siliceous aggregates are classified as fast reacting, while granites, some siliceous and limestone with quartz inclusions are classified as slow reacting. For its part, dolomites and most limestone are rated as non-reactive.

Figure 13.  Expansion at 14 and 28 days versus IQr of aggregates.

Figure 14.  Expansion at 14, 28 and 365 days versus IQr of aggregates.

In the case of the Spanish aggregates studied, just 1 of the siliceous aggregates can be considered as non-expansive, while 13 are slow reactive, 31 are considered as potential reactive aggregates and 4 are considered as potential reactive aggregates with pessimum effect. For its part, the 4 granites and 3 of the 11 limestone aggregates are considered as slow reactive aggregates. The limestone aggregates considered as slowly reactive have some reactive particles of quartz in their composition. Finally, neither dolomites nor the basalt can be considered as potential reactive aggregates.

CONCLUSIONS

The open porosity should have a direct relationship with the potential reactivity of the aggregates. However, as these are aggregates of a different nature, this correlation cannot be observed. However, the low open porosity, as in the case of granites and quite a few siliceous aggregates, show the difficulty of entry of the OH^- groups that causes the breaking of the siloxane bridges of the deformed quartz.

On the other hand, the expansion extended in time up to 1 year, makes it possible to verify the evolution of this expansiveness, especially for slow-reacting aggregates. However, the extension of these tests to a normative level is not considered, since it would not be feasible to wait this time in the vast majority of concrete works.

The relationship between the percentage of total reactive quartz in an aggregate and the expansion extended to 1 year, allows classifying reactive aggregates, both slow reacting and those classified with the accelerated mortar bar method as potentially reactive. The total reactive quartz of an aggregate (I_Qr·Q_t) is calculated by multiplying the reactivity index of the quartz by the total quantity of quartz in the concrete aggregate. According to the results obtained in this work, the following classification is proposed:

If one of the two conditions is not fulfilled, the total percentage of reactive aggregates (I_Qr·Q_t) prevails.