Mineralogical Characterization of Dolomitic Aggregate Concrete: The Camarasa Dam (Catalonia, Spain)

: The Camarasa Dam was built in 1920 using dolomitic aggregate and Portland cement with two di ﬀ erent compositions: type A (dolomite and Portland cement) and type B (dolomite and sand-cement). The sand cement was a ﬁnely powdered mixture of dolomite particles and clinker of Portland cement. The mineralogy of concrete was studied by optical microscopy, scanning electron microscopy, and x-ray powder di ﬀ raction. Reaction of dedolomitization occurred in the two types of concrete of the Camarasa Dam, as demonstrated by the occurrence of calcite, brucite, and / or absence of portlandite. In the type A concrete, calcite, brucite, and a serpentine-group mineral precipitated as a rim around the dolomite grains and in the paste. The rims, a product of the dedolomitization reaction, protected the surface of dolomite from the dissolution process. In type B concrete, in addition to dolomite and calcite, quartz and K-feldspar were present. Brucite occurred in lower amounts than in the type A concrete as ﬁbrous crystals randomly distributed in the sand-cement paste. Although brucite content was higher in the type A concrete, type B showed more signs of loss of durability. This can be attributed to the further development of the alkali-silica reaction in this concrete type.


Introduction
Concrete made of dolomite aggregate, has been the source of problems such as expansion and cracking [1][2][3][4][5][6][7]. Traditionally, these deterioration problems were associated with the alkali-carbonate reaction (ACR) [8,9], which is produced when dolomitic rocks with clay minerals are used as aggregates or as a result of the reaction of dolomite with cryptocrystalline quartz [10]. Several dams, built with carbonate aggregates, exhibit signs of deterioration such as the Bimont Dam in France, constructed with dolomitic limestones [11], and the Chickamagua Powerhouse Dam in Tennessee [12].
Later, in order to reduce the costs, the Portland cement was substituted by the sand-cement mixture, and 183.839 m 3 of type B concrete were used to finish building the dam [27]. The total composition of this concrete consisted of 5.2 wt% of Portland cement and 94.8 wt% of dolomite. The sand-cement used in this dam was a homogeneous mixture of 55 wt% of clinker of Portland cement and 45 wt% of dolomite pounded together up to less than 0.13 mm in size. The composition of the two concrete types is presented in Table 1.
The concrete was studied in ten drill cores of 55 mm in diameter, obtained in 1998 by the electric power company, owner of the dam.

Analytical Methods
Thin sections were made from the cores with non-aqueous polishing fluids and epoxy resin adhesives to prevent the dissolution of mineral phases. Several thin sections were stained with Alizarin Red S+ Potassium to evaluate the composition of carbonate minerals.
Thin sections were studied by electronic microscopy with x-ray energy dispersive spectroscopy (SEM + EDS); images and qualitative analyses were produced using a Leika Stereoscan 360, an ESEM Quanta 200 FEI, XTE 325/D8395 and a field emission scanning electron microscope JEOL JSM-7001F at the Serveis Científics i Tecnològics de la Universitat de Barcelona (CCiTUB).  [26] and the location of the cores (black lines) and samples (red)).
Later, in order to reduce the costs, the Portland cement was substituted by the sand-cement mixture, and 183.839 m 3 of type B concrete were used to finish building the dam [27]. The total composition of this concrete consisted of 5.2 wt% of Portland cement and 94.8 wt% of dolomite. The sand-cement used in this dam was a homogeneous mixture of 55 wt% of clinker of Portland cement and 45 wt% of dolomite pounded together up to less than 0.13 mm in size. The composition of the two concrete types is presented in Table 1.
The concrete was studied in ten drill cores of 55 mm in diameter, obtained in 1998 by the electric power company, owner of the dam. The x-ray powder diffraction (XRPD, Siemens AG, Munich, Germany) was used to determine the mineral phases present in the concrete. Measurements were performed in a Bragg-Brentano θ2θ Siemens D-500 diffractometer (radius = 215.5 mm) with Cu Kα radiation, selected by means of a secondary graphite monochromator. The divergence slit was 1° and the receiving slit of 0.15°. The step size was 0.05° 2θ and the measuring time was 10 seconds per step. For the calculation of the mineral phase contents, the quantitative Rietveld analysis method with the FULLPROF program [28] was used. The analyses were made at the CCiTUB.

Raw Dolostones
The rocks used as aggregates were mainly constituted by dolomite up to 89 wt% and calcite up to 11 wt% ( Figure 2). The occurrence of calcite is the result of the dedolomitization process of the raw dolostones before being exploited. Brucite was been observed in any case.
Under the microscope, the dolomite crystals exhibited a turbid appearance due to the content of abundant internal inclusions forming dark and clear bands, or a zebra texture ( Figure 3). Calcite constituted the interstitial cement or was located in the rims around the dolomite crystals ( Figure 3b) or in grains of dolomite in the natural process of dolomitization forming path textures.

Analytical Methods
Thin sections were made from the cores with non-aqueous polishing fluids and epoxy resin adhesives to prevent the dissolution of mineral phases. Several thin sections were stained with Alizarin Red S+ Potassium to evaluate the composition of carbonate minerals.
Thin sections were studied by electronic microscopy with x-ray energy dispersive spectroscopy (SEM + EDS); images and qualitative analyses were produced using a Leika Stereoscan 360, an ESEM Quanta 200 FEI, XTE 325/D8395 and a field emission scanning electron microscope JEOL JSM-7001F at the Serveis Científics i Tecnològics de la Universitat de Barcelona (CCiTUB).
The x-ray powder diffraction (XRPD, Siemens AG, Munich, Germany) was used to determine the mineral phases present in the concrete. Measurements were performed in a Bragg-Brentano θ2θ Siemens D-500 diffractometer (radius = 215.5 mm) with Cu Kα radiation, selected by means of a secondary graphite monochromator. The divergence slit was 1 • and the receiving slit of 0.15 • . The step size was 0.05 • 2θ and the measuring time was 10 seconds per step. For the calculation of the mineral phase contents, the quantitative Rietveld analysis method with the FULLPROF program [28] was used. The analyses were made at the CCiTUB.

Raw Dolostones
The rocks used as aggregates were mainly constituted by dolomite up to 89 wt% and calcite up to 11 wt% ( Figure 2). The occurrence of calcite is the result of the dedolomitization process of the raw dolostones before being exploited. Brucite was been observed in any case.
Under the microscope, the dolomite crystals exhibited a turbid appearance due to the content of abundant internal inclusions forming dark and clear bands, or a zebra texture ( Figure 3). Calcite constituted the interstitial cement or was located in the rims around the dolomite crystals ( Figure 3b) or in grains of dolomite in the natural process of dolomitization forming path textures. The x-ray powder diffraction (XRPD, Siemens AG, Munich, Germany) was used to determine the mineral phases present in the concrete. Measurements were performed in a Bragg-Brentano θ2θ Siemens D-500 diffractometer (radius = 215.5 mm) with Cu Kα radiation, selected by means of a secondary graphite monochromator. The divergence slit was 1° and the receiving slit of 0.15°. The step size was 0.05° 2θ and the measuring time was 10 seconds per step. For the calculation of the mineral phase contents, the quantitative Rietveld analysis method with the FULLPROF program [28] was used. The analyses were made at the CCiTUB.

Raw Dolostones
The rocks used as aggregates were mainly constituted by dolomite up to 89 wt% and calcite up to 11 wt% ( Figure 2). The occurrence of calcite is the result of the dedolomitization process of the raw dolostones before being exploited. Brucite was been observed in any case.
Under the microscope, the dolomite crystals exhibited a turbid appearance due to the content of abundant internal inclusions forming dark and clear bands, or a zebra texture ( Figure 3). Calcite constituted the interstitial cement or was located in the rims around the dolomite crystals ( Figure 3b) or in grains of dolomite in the natural process of dolomitization forming path textures.

Concrete Petrography and Mineralogy
Type A and type B concrete can be distinguished by the color of the cement paste: type A is light grey, whereas type B is dark brown (Figure 4). The visual inspection showed significant differences between the type A and type B concrete. Type A looked more compact and was more resistant to abrasion, whereas the type B concrete was more disagreeable. It was expected, then, that the behavior of the dedolomitization reaction would be considerably different in the two concrete types.

Concrete Petrography and Mineralogy
Type A and type B concrete can be distinguished by the color of the cement paste: type A is light grey, whereas type B is dark brown (Figure 4). The visual inspection showed significant differences between the type A and type B concrete. Type A looked more compact and was more resistant to abrasion, whereas the type B concrete was more disagreeable. It was expected, then, that the behavior of the dedolomitization reaction would be considerably different in the two concrete types.
Aggregate particles from the Camarasa Dam exhibited an equant and angular shape, and did not look weathered. The most abundant aggregates were grey-brown particles constituted of pure Under the microscope, the type A and type B concrete showed a random distribution of the different particle size fractions of dolomitic particles ( Figure 5a). Unconnected porosity was randomly distributed in the paste. The thin sections stained with red alizarin show red rims in the contact between the cement paste and the particles of the dolomitic aggregate ( Figure 5b). Dolomite and calcite are the major components in the paste of both concrete types. Other crystalline phases detected were different in them (Table 2). In order to evaluate the reactions Aggregate particles from the Camarasa Dam exhibited an equant and angular shape, and did not look weathered. The most abundant aggregates were grey-brown particles constituted of pure dolomite grains.
Under the microscope, the type A and type B concrete showed a random distribution of the different particle size fractions of dolomitic particles ( Figure 5a). Unconnected porosity was randomly distributed in the paste. The thin sections stained with red alizarin show red rims in the contact between the cement paste and the particles of the dolomitic aggregate ( Figure 5b).

Concrete Petrography and Mineralogy
Type A and type B concrete can be distinguished by the color of the cement paste: type A is light grey, whereas type B is dark brown (Figure 4). The visual inspection showed significant differences between the type A and type B concrete. Type A looked more compact and was more resistant to abrasion, whereas the type B concrete was more disagreeable. It was expected, then, that the behavior of the dedolomitization reaction would be considerably different in the two concrete types.
Aggregate particles from the Camarasa Dam exhibited an equant and angular shape, and did not look weathered. The most abundant aggregates were grey-brown particles constituted of pure dolomite grains. Under the microscope, the type A and type B concrete showed a random distribution of the different particle size fractions of dolomitic particles (Figure 5a). Unconnected porosity was randomly distributed in the paste. The thin sections stained with red alizarin show red rims in the contact between the cement paste and the particles of the dolomitic aggregate ( Figure 5b). Dolomite and calcite are the major components in the paste of both concrete types. Other crystalline phases detected were different in them (Table 2). In order to evaluate the reactions Dolomite and calcite are the major components in the paste of both concrete types. Other crystalline phases detected were different in them (Table 2). In order to evaluate the reactions produced in the Camarasa Dam aggregates, we describe the differences found in the chemical and mineralogical characteristics of the two concrete types.

Type A Concrete
XRPD and SEM show that in the type A cement paste, in addition to dolomite and calcite, a significant amount of brucite and portlandite was also present ( Figure 6). produced in the Camarasa Dam aggregates, we describe the differences found in the chemical and mineralogical characteristics of the two concrete types.  EDS analyses of dolomite crystals reveal approximately 25 µm wide compositional rims (Figure 7). The chemical composition of mineral phases from these rims in the contact with the cement paste is Mg, Al, Si-rich and indicates that their mineralogy is constituted by brucite and a member of the serpentine group minerals (Figure 7).

Type B Concrete
The XRPD of the paste in the type B concrete indicates that the crystalline phases were mainly dolomite, calcite, brucite, quartz, and K-feldspar. In this type of concrete, portlandite was not detected and the contents of brucite were significantly lower than in the type A concrete ( Table 2). The presence of quartz and K-feldspar was probably due to the addition of sand from the channel river and from Cretaceous sandstones of the area during the manufacturing process of the sandcement. The SEM observations confirm the presence of brucite crystals up to 10 µm in size, with platelets or fibrous habit (Figure 8).

Type B Concrete
The XRPD of the paste in the type B concrete indicates that the crystalline phases were mainly dolomite, calcite, brucite, quartz, and K-feldspar. In this type of concrete, portlandite was not detected and the contents of brucite were significantly lower than in the type A concrete ( Table 2). The presence of quartz and K-feldspar was probably due to the addition of sand from the channel river and from Cretaceous sandstones of the area during the manufacturing process of the sand-cement. The SEM observations confirm the presence of brucite crystals up to 10 µm in size, with platelets or fibrous habit ( Figure 8). In addition, the SEM observations also showed the occurrence of ettringite and belite. Some illite elongated crystals were detected by SEM-EDS, and nests of belite, "bunch of grapes", with dark hydration rims dispersed in the interstitial component or paste could occasionally be observed (Figure 9a), indicating that the hydration of cement was not complete. Unconnected porosity was randomly distributed in the paste. Most cavities exhibited recrystallization rims or were totally or partially filled of calcite, ettringite, or portlandite (Figure 9b).

Discussion
The absence of brucite in the dolomitic rocks and its occurrence in the two types of concrete indicates that the dedolomitization reaction took place in the concrete from the Camarasa Dam. In the type A concrete, the reaction occurred on the surface of the dolomite aggregates, whereas in the type B concrete, the reaction mainly took place in the sand-cement paste and in the dolomite grain borders, although Blanco et al. [29] attributed a geological origin to the occurrence of brucite in the Camarasa Dam concrete.
In the reaction of dedolomitization, two processes are involved: the dissolution of dolomite and portlandite and the precipitation of new phases of brucite and calcite. The reaction of dedolomitization proceeds until the consumption of portlandite [15]. The dissolution rate of dolomite depends on the mineral surface, which, in turn, depends on the particle sizes, and controls the kinetics of the reaction [14,20]. The most reactive dolomite aggregate was the finest fraction, <0.13 mm in the concrete of the In addition, the SEM observations also showed the occurrence of ettringite and belite. Some illite elongated crystals were detected by SEM-EDS, and nests of belite, "bunch of grapes", with dark hydration rims dispersed in the interstitial component or paste could occasionally be observed (Figure 9a), indicating that the hydration of cement was not complete. Unconnected porosity was randomly distributed in the paste. Most cavities exhibited recrystallization rims or were totally or partially filled of calcite, ettringite, or portlandite (Figure 9b). In addition, the SEM observations also showed the occurrence of ettringite and belite. Some illite elongated crystals were detected by SEM-EDS, and nests of belite, "bunch of grapes", with dark hydration rims dispersed in the interstitial component or paste could occasionally be observed (Figure 9a), indicating that the hydration of cement was not complete. Unconnected porosity was randomly distributed in the paste. Most cavities exhibited recrystallization rims or were totally or partially filled of calcite, ettringite, or portlandite (Figure 9b).

Discussion
The absence of brucite in the dolomitic rocks and its occurrence in the two types of concrete indicates that the dedolomitization reaction took place in the concrete from the Camarasa Dam. In the type A concrete, the reaction occurred on the surface of the dolomite aggregates, whereas in the type B concrete, the reaction mainly took place in the sand-cement paste and in the dolomite grain borders, although Blanco et al. [29] attributed a geological origin to the occurrence of brucite in the Camarasa Dam concrete.
In the reaction of dedolomitization, two processes are involved: the dissolution of dolomite and portlandite and the precipitation of new phases of brucite and calcite. The reaction of dedolomitization proceeds until the consumption of portlandite [15]. The dissolution rate of dolomite depends on the mineral surface, which, in turn, depends on the particle sizes, and controls the kinetics of the reaction [14,20]. The most reactive dolomite aggregate was the finest fraction, <0.13 mm in the concrete of the

Discussion
The absence of brucite in the dolomitic rocks and its occurrence in the two types of concrete indicates that the dedolomitization reaction took place in the concrete from the Camarasa Dam. In the type A concrete, the reaction occurred on the surface of the dolomite aggregates, whereas in the type B concrete, the reaction mainly took place in the sand-cement paste and in the dolomite grain borders, although Blanco et al. [29] attributed a geological origin to the occurrence of brucite in the Camarasa Dam concrete.
In the reaction of dedolomitization, two processes are involved: the dissolution of dolomite and portlandite and the precipitation of new phases of brucite and calcite. The reaction of dedolomitization Minerals 2020, 10, 117 9 of 13 proceeds until the consumption of portlandite [15]. The dissolution rate of dolomite depends on the mineral surface, which, in turn, depends on the particle sizes, and controls the kinetics of the reaction [14,20]. The most reactive dolomite aggregate was the finest fraction, <0.13 mm in the concrete of the Camarasa Dam. The two types of concrete exhibited differences in the content of this fraction of dolomite aggregates. In the type B concrete, all of the dolomite particles of this fraction were constituents of the sand-cement and were dispersed in the Portland cement.
In the type A concrete, the occurrence of portlandite in the paste indicates that this reaction can still take place. The Mg 2+ removed from dolomite was accomplished with the OH − from portlandite, and brucite precipitates as a rim around the partially dissolved dolomite grains ( Figure 10). Camarasa Dam. The two types of concrete exhibited differences in the content of this fraction of dolomite aggregates. In the type B concrete, all of the dolomite particles of this fraction were constituents of the sand-cement and were dispersed in the Portland cement.
In the type A concrete, the occurrence of portlandite in the paste indicates that this reaction can still take place. The Mg 2+ removed from dolomite was accomplished with the OH − from portlandite, and brucite precipitates as a rim around the partially dissolved dolomite grains ( Figure 10). In addition to the precipitation of brucite, the dissolution of the Si, Al-rich phases of the cement paste removes the aluminum and silicon that substitute the Mg 2+ from brucite. In this case, in order to balance the electrical charges, a coupled substitution of OH − by CO3 2− and a Mg rich rim is formed [30,31]. This rim constitutes a barrier that makes the dissolution of the dolomite aggregates difficult, as a result, the reaction of dedolomitization slows down [14,32,33].
In the reaction of dedolomitization, calcite is formed from the CO3 2− and one Ca 2+ provided by the dissolution of dolomite, and another Ca 2+ provided by portlandite located in the cement paste. Then, most calcite is formed, not in direct contact with the dolomite aggregate, but in the contact point between the brucite rim and cement paste. A similar distribution of the products of the reaction of dedolomitization was obtained in experiments performed with a dolomite crystal immersed in an alkaline portlandite-rich solution [14]. In that case, brucite precipitated attached to the dolomite surface while most calcite precipitated in the solution.
In the type B concrete, the reaction of dedolomitization took place early, due to the high reactive surface of dolomite particles in the sand-cement paste. In this concrete, portlandite is dissolved and the OH − anions are quickly trapped by the Mg 2+ provided by dolomite in the sand-cement paste. Thus, brucite precipitates as fibrous crystals distributed in the sand-cement paste. The reaction consumes all available portlandite, and then, the coarser particles of dolomite aggregate will not be affected by the dedolomitization reaction. This process was also suggested for the dedolomitization reaction in batch experiments of dolomite dispersion in alkaline media [14].
Although the dedolomitization reaction occurred in both types of concretes, the visual inspection showed slight differences between types A and B, where A looked more compact than type B. The matrix of type B appeared more disaggregated and more fragile. However, the dedolomitization reaction was more extended in the conventional, or type A aggregate, as In addition to the precipitation of brucite, the dissolution of the Si, Al-rich phases of the cement paste removes the aluminum and silicon that substitute the Mg 2+ from brucite. In this case, in order to balance the electrical charges, a coupled substitution of OH − by CO 3 2− and a Mg rich rim is formed [30,31]. This rim constitutes a barrier that makes the dissolution of the dolomite aggregates difficult, as a result, the reaction of dedolomitization slows down [14,32,33].
In the reaction of dedolomitization, calcite is formed from the CO 3 2− and one Ca 2+ provided by the dissolution of dolomite, and another Ca 2+ provided by portlandite located in the cement paste. Then, most calcite is formed, not in direct contact with the dolomite aggregate, but in the contact point between the brucite rim and cement paste. A similar distribution of the products of the reaction of dedolomitization was obtained in experiments performed with a dolomite crystal immersed in an alkaline portlandite-rich solution [14]. In that case, brucite precipitated attached to the dolomite surface while most calcite precipitated in the solution.
In the type B concrete, the reaction of dedolomitization took place early, due to the high reactive surface of dolomite particles in the sand-cement paste. In this concrete, portlandite is dissolved and the OH − anions are quickly trapped by the Mg 2+ provided by dolomite in the sand-cement paste. Thus, brucite precipitates as fibrous crystals distributed in the sand-cement paste. The reaction consumes all available portlandite, and then, the coarser particles of dolomite aggregate will not be affected by the dedolomitization reaction. This process was also suggested for the dedolomitization reaction in batch experiments of dolomite dispersion in alkaline media [14].
Although the dedolomitization reaction occurred in both types of concretes, the visual inspection showed slight differences between types A and B, where A looked more compact than type B. The matrix of type B appeared more disaggregated and more fragile. However, the dedolomitization reaction was more extended in the conventional, or type A aggregate, as demonstrated by the higher proportion of calcite and brucite, which in turn decreased the binding properties and alkalinity of the matrix. The compression resistance of the cores from Camarasa confirmed these observations. Cabrera Vélez [34] reported mean values of 279.3 kg/cm 2 for the concrete of Portland cement and from 148.5 to 220.1 kg/cm 2 for concrete made with sand-cement. The parent density of both types of concrete [29] also showed that the alteration in the area that constituted of sand-cement was higher. Brucite has been considered as one of the main factors responsible for the loss of concrete durability [35]. However, in other cases, the highest expansion and loss of durability was attributed to the alkali-silica reaction (ASR) produced by the reactions due to the presence of quartz in the concrete matrix, which can accompany the dolomitic aggregates [36,37].
It can be considered that the concrete of the dam contains an excess of dolomite, whereas the amount of Portland cement, on the other hand, is significantly lower and therefore a limited amount of portlandite and silica gel will be produced. The pore solution in contact with these phases is subsatured in dolomite and CHS, which are thermodynamically unstable in this medium and will be dissolved. The dissolution of these phases produces two basic processes: (1) a dedolomitization reaction, which produces secondary calcite and brucite, and (2) between dolomite and the cement paste, a calcium silicate gel (CHS) gel is present, which reacts with brucite to produce a Mg-silicate gel that crystallizes in a serpentine group mineral. The formation of this mineral phase has also been reported in other concretes of dolomite aggregates [7,36,38].
The kinetics of the dedolomitization reaction is controlled by the dissolution of the dolomite and the amount of portlandite available in the system. The reaction occurs very quickly for the smallest particles. For this reason, the mineralogical evolution of the system is qualitatively and quantitatively different in each of the concretes used for the construction of the dam.
The use of a concrete made with sand-cement contributes to the system with a particle size fraction of dolomite with an enormous reactive surface that confers to the reaction of dedolomitization a very fast kinetics. This reaction is completed in the sand-cement matrix by replacing the sand-cement paste in the surface, formed by dolomite and portlandite, with a mixture constituted of calcite and brucite. The agglomerating properties of brucite are not good enough and therefore the properties of the mechanical strength of the concrete are slightly diminished, as confirmed by the resistance behavior of the concrete made with sand-cement. Once the dolomite of the paste is consumed, a small fraction of portlandite remains undissolved and is susceptible to react with the larger grain size fractions of particles in the aggregate.
In the case of Portland concrete, the proportion of dolomitic aggregates with a reactive surface equivalent to that of the sand-cement dolomite particles was significantly lower. It was observed that the dedolomitization was produced by forming a reaction rim of less than 25 µm in thickness around the dolomite particles, consisting mainly of brucite, calcite, and a gel phase, and that a significant proportion of portlandite remained in the matrix of the concrete. The formation of a reaction rim around the aggregate hinders the dissolution of the dolomite, therefore the reaction occurs very slowly.
The gel phase is unstable in the system and evolves in the presence of Mg, forming hydrotalcite and serpentine group mineral, which are more stable in the system. Reaction (2) describes how the gel reacts with brucite, which is the product of the dedolomitization reaction, to form serpentine group mineral. Studies carried out on concretes attacked by magnesium sulfates show that serpentine group minerals have a low binder capacity with an appearance similar to that of the final stages of concrete deterioration [30]. In the case of the Camarasa concrete, the magnesium source comes from the dissolution of dolomite and the presence of magnesium sulfate is not necessary.
Concrete constituted initially of dolomite and gel of the Portland cement in the appropriate ratios will evolve until reaching the equilibrium state with the formation of calcite and serpentine group mineral. The concrete of the Camarasa Dam has a large excess of dolomite that will not react. Given the reduced agglomerating properties of the serpentine group mineral, its difficulty of formation is rather beneficial, since it allows the gel life of the Portland to be extended. On the other hand, the formation of reaction rims on the surface of the dolomite is also an important factor in delaying the reactions that will produce the degradation of the concrete [39].

Conclusions
The concrete of the Camarasa Dam is made of aggregates mainly composed of dolomite. According to the dose, two type of concrete are distinguished: type A (dolomite and Portland cement) and type B (dolomite and sand-cement). The ACR occurs in both concrete types of the Camarasa Dam, forming calcite and brucite from the dissolution of dolomite in an alkaline media. In type A concrete, the dedolomitization reaction takes place on the surface of the aggregate particles and stops early, preserving the portlandite of the cement paste. In contrast, in the type B concrete, the reaction occurred in the sand-cement paste and all portlandite available was consumed very early by the dedolomitization reaction.
Although the brucite content was higher in type A concrete, type B showed more signs of loss of durability. This can be attributed to the greater development of the ASR in this concrete type.
In order to prevent the ACR, it is proposed that fractions smaller than 1 mm of the dolomite aggregate in the concrete are avoided completely.
Author Contributions: P.A. and E.T. wrote the paper; E.G. did the analytical work. All authors contributed to data interpretation and discussion. All authors have read and agreed to the published version of the manuscript.
Funding: This research was sponsored by the CICYT Spanish research project MAT2002-3345 and contract RED99-57. We thank ENDESA for providing the samples. The research was supported by the SGR-198 and SGR-707 of the Generalitat de Catalunya.