Railroad ballast of granites and basic rock in tropical regions: rela onships between petrography, physical-mechanical proper es and alterability

Recebido: 4 de maio de 2020 Aceito para publicação: 20 de agosto de 2020 Publicado: 30 de julho de 2021 Editor de área: Francisco Thiago Aragão ABSTRACT The objective of this study was to evaluate, comparatively, the evolution of this degradation of three rock types, two granitoids and a microgabbro which, eventually, could be used as ballast, due to the proximity of the quarries to the design route of a railroad to be built between São Paulo and around Campinas, State of São Paulo, southeastern Brazil. They were tested for alterability in sodium sulfate and ethylene glycol solutions. Based on detailed petrographic analysis, Micro-Deval and Slake Durability Index tests were performed with and without conjugation to the referred alterability tests. The results obtained are punctual, valid and positive for the samples of the rock masses carefully selected and described, considering the different characteristics of the granitoids belonging to the Morungaba and Cantareira Complexes. In the case of basic rock, the microgabbro, should not be used as ballast in the studied tropical climate area, since the exposed fragments that make up the ballast are intensely affected by chemical weathering (notably reactions accelerated by water and temperature), which culminate in the decomposition of primary minerals and, consequently, in the faster physical-mechanical degradation of this rock. In practice, more frequent periodic maintenance and higher costs.


INTRODUCTION
Railroad is an essential and strategic means of transport that plays a key role in the economy of many countries. In recent years, Brazil has carried out research to implement new ballast railroads and/or readjust the existing ones, in general for the !low of agricultural products between producing regions and ports. The studies also focus on projects to be construct that involve passenger transport, interconnecting, for example, some of the main cities in the southeastern region through medium-and high-speed trains (Cagnon et al., 2015;Revista Ferroviária, 2019).
Ballast railroads are the most used all over the world for several reasons, including savings in implementation, ef!icient rainwater drainage and easy maintenance. Regarding the maintenance cost, a remarkable technological advance is occurring in the manufacture of rails and sleepers, and it is also necessary to encourage studies on the knowledge of the technological characteristics of the most adequate rocks for use as aggregate in the ballast. Some of the most recent studies include those on petrographic aspects, mechanical properties to provide engineering design parameters and alterability (Apaydin and Yilmaz, 2020;Esfahani et al., 2019;Paiva et al., 2018;Van Blerk et al., 2017;Sadegui et al., 2016), especially considering the weathering conditions imposed on this rocky materials in a tropical climate, dominant over a large area of the Brazilian territory. In other words, it is important to stress that the rock aggregates used to make up the ballast of railroads are subjected to constant exposure to the weathering, being under stronger in!luence of the environment in terms of decomposition of minerals by chemical reactions. That is, studies on alterability become as important as those on geomechanical characterization of rocks aiming at this type of use.
In this approach, the present study is part of a multidisciplinary line of research, involving the expertises of Engineering Geology and Transportation Engineering.
In addition to the physical-mechanical quali!ication, the objective was to evaluate aggregates comparatively, based on non-routine and poorly disseminated tests in studies of characterization in Brazil, the evolution of the degradation of three rocky materials from quarries located in the city of São Paulo and in the surroundings of the city of Campinas (southeastern Brazil). These materials were chosen for several reasons, among them: they have been used for decades as aggregates for concrete and paving, but still little analysed for railroad ballast, and for logistical reasons, due to the proximity to the layout of the route planned for the Inter Cities Train "TIC" (Revista Ferroviária, 2019).
To this end, based on detailed petrographic analysis (mineralogy, texture, degree of alteration, microcracking, among others), physical-mechanical and alterability tests were carried out, combined or not with Micro-Deval abrasion and Slake Durability Index tests, on samples collected in sites of the massifs that presented better quality, which were adopted as parameters to quantify the mechanical degradation of the studied rocks. The proposed characterization, from the technological point of view, represents very well the mechanical demands (compression, abrasive wear, crushing and toughness) and as well as the weathering effects imposed on the ballast of a railroad in use in a tropical climate.

MATERIAL AND SAMPLING
Among the lithologies studied by Remedio (2017) to compose railroad ballast, Morungaba Complex (MOR), Cantareira Granite (CAN) and Basic Intrusives of the Serra Geral Formation (GAB) were selected, due to their technological qualities and the proximity to the possible route TRANSPORTES | ISSN: 2237-1346 designed for the Inter Cities Train "TIC" that will connect the capital of the State of São Paulo, Brazil, to the city of Americana. These rocks are exploited in several active quarries near existing railroads and have been in operation for decades, driving the growth of the metropolitan regions of Campinas and São Paulo (Figure 1).

Figure 1.
Location of the area. The three geological units that make up the present study are highlighted in bold in the legend A fundamental aspect to be highlighted (but usually neglected in Brazilian engineering practice) is that, in the case of railroad ballast, the analysis of the !ield works should precede the sampling steps and, obviously, the technological tests. There is no homogeneous rock massif, but higher or lower geological variability. In other words, every study should begin with a detailed geological description of the quarry, especially the heterogeneities of the rock mass regarding the exploitable rocks among those existing.
Based on these premises, before sampling, geological observations and petrographic descriptions were made on the fronts of the quarries, where rocks are extracted for engineering applications to de!ine the portions of the massif with better quality. Rocks for ballast are extracted from these portions, where the aggregate is required for its mechanical quality and, especially, for the greater resistance to weathering. The other portions of the massif are used for the extraction of rocks for other applications, such as concrete and paving, which use aggregates with binders, in which other properties gain relevance, for example, the alkali-aggregate reactivity and the adhesion to the asphalt.
Subsequently, new observations were made, with regard to the integrity found in the diverse dimensions of the disassembled rock, along with the collection of blocks and irregular fragments of such material ( Figure 2). Coarse aggregates were produced in laboratory, for the exact origin of the material, different from what occurs when the material is crushing in the quarry, which may contain rocks from different domains of the rock mass.  The methodological !lowchart ( Figure 3) is show below: (I) Geological !ield recognition to de!ine the quarry sector to be studied; (II) Collection of samples in front of rock dismantling (blocks and irregular fragments). After crushing, the material was sieved, homogenized and quartered, with the separation of the aliquots for the mechanical and accelerated alteration tests; petrographic analysis and determination of physical indexes; (III) Determination of the values of the mechanical and alterability testing; Micro-Deval abrasive wear (MD and MDR) and Slake Durability Index (Id) served as a reference to assess the mechanical degradation of rocks: MD associated with soundness tests by use of sodium sulfate and Id to soundness tests by immersion in ethyleneglycol; (IV) Integrated assessment of data obtained in technological characterization tests.

TEST METHODS
Petrographic analyses and physical-mechanical testing procedures (physical indices, uniaxial compression strength -σdry and σwet, Los Angeles abrasion -LAA, crushing strength -SC, Treton toughness index -T, Micro-Deval abrasion -MD and Slake Durability Index -SDI) and alterability followed the guidelines of the Brazilian Association of Technical Standards (ABNT), American Standard Testing Machines (ASTM) and National Department of Transportation Infrastructure (DNIT). Table 1 presents the synthesis of these tests and their respective technical standards, type/amount of sample and execution procedures.
The uniaxial compressive strength tests were carried out on specimens dried in an oven at 110º C and also on saturated specimens, to assess the possible hydraulic weakening of each rock in relation to the efforts applied on them. The hydraulic weakening coef!icient R (Kowalski, 1970 according to Mesquita, 2002 andMattos et al., 2013), relates the strength of a dry and saturated material and allows to evaluate the effect of water on the strength of rocky materials, and is calculated by Equation 1. R = (σwet / σdry) x 100 (1) where σwet and σdry are the tensile strength of the material in saturated and dry conditions, respectively.
To obtain information about the weathering of rocks as fragments (ballast), two types of alterability tests were conducted: with sodium sulfate and with ethyleneglycol, all with oven drying. During these cycles, all samples were qualitative and quantitatively evaluated by measuring mass losses, by visual examinations and photographic records of any damages (!issures, cracks, !laking, oxidation and disintegration) in the specimen. Additionally, the MD and SDI tests, which are not routine in studies on the characterization of aggregates in Brazil, were adopted as auxiliary parameters to quantify the disaggregation (Figure 4), because they are tests performed with saturated samples, in which the altered material is discarded from the sample mass.
Based on the guidelines of the D6928 standard (ASTM, 2017), the MD test was used to quantify the abrasion resistance and durability of rocky aggregates. This test is widely used worldwide, but not in Brazil. Each test used a sampled with about 1,500g of fragments with diameters passing through a 19.50 mm sieve, and retained in a 9.50 mm sieve. The "Micro-Deval Residual" (MDR) index, proposed by Benediktsson (2015), was also calculated by the ratio of the sum of the masses retained in the sieve of the lowest fraction of the chosen grade of the test to the initial sample mass. TRANSPORTES | ISSN: 2237-1346  In the case of the SDI test, although originally used in soft and low-strength rocks, with an expansive nature and with high disintegration when wetted (Franklin and Chandra, 1972), there are several examples of application of this test in research involving the alterability of igneous rocks and their behavior in geological massifs used by engineering (Momeni et al., 2017;Gupta and Ahmed, 2007;Arel and Tugrul, 2001). For comparative purposes between the rocks, adjustments to the guidelines of the D4644 standard (ASTM, 2016) were established, using approximately 500 g of fragments with diameters between 19.5 mm and 9.5 mm (same grade of the MD test) before and after the accelerated alteration tests. In the present study, !ive complete cycles were run for each rock. Table 2 lists the classi!ication of rock durability proposed by Gamble (1971), taking into account the Id values obtained in the tests, and each sample retained in the drum should be visually categorized in comparison to the standards established in the D4644 standard (ASTM, 2016), as follows: Type I: retained samples remain practically unchanged; Type II: retained samples consist of large and small fragments and Type III: retained samples are exclusively small fragments.

RESULTS AND DISCUSSION
The results of the tests performed are presented following the order of the methodology used in the study.

Petrographic analysis
The information regarding the degrees of microcracking and rock alteration are discussed below, with the most important petrographic features analyzed at the microscopic level presented in Table 3 and Figure 5. The studied materials refer to intrusive rocks with homogeneous aspects, massif and isotropic structures with equigranular to unequigranular phaneritic textures (discretely porphyritic). Considering the composition, there are two granitoids: the !irst, called Morungaba Granite (MOR), a !ine-grained and pink-colored syenogranite; the second, Cantareira Granite (CAN), a coarse-grained, porphyritic and poikilitic monzogranite, light gray in color. The other rock is a medium to !ine grained microgabbro (GAB), dark gray in color.
TRANSPORTES | ISSN: 2237-1346 The microcracking degree of the three materials is scarce to low, essentially intragrain, relatively more intense in the ma!ic minerals (olivine and clinopyroexenes) of the microgabro. As for the alteration of this rock, it is more intense close to the concentrations/aggregates of ma!ic minerals. The predominant secondary minerals occurring in a powdery form, represented by chlorite, biotite, !ibrous amphiboles and cloropheite, which would represent a !ine greenish aggregate and possibly composed of chlorite, !ibrous amphiboles, greenish biotite and clay minerals, such as celadonite and nontronite (especially on clinopyroxenes and olivines). As shown in the Figure 5, olivines are the most intensely microfractured minerals, with open microcrack systems and !illed with a thin mass of secondary minerals, such as opaques, chlorite, amphiboles and some mineral clay.

Physical-mechanical indices
Information from the petrographic analysis allows to set some correlations with the values of the physical indices. A brief comparison between the studied materials (Table 4 and Figure 6) indicates the GAB microgabbro as the rock with relatively more expressive values of apparent porosity (ηa = 1.73%) and water absorption (αa = 0.59%).
In the !irst case, as shown in Figure 7, it exceeds the maximum value of 0.8% recommended by the NBR 5564 standard (ABNT, 2012). These higher values of ηa and αa seem to be related to the coarser particle size of the rock and to a greater communicability of its pore network. They depend on the conditions of mineralogical alteration (especially of plagioclases) and the microcracking pattern with predominance of intragrain !issures with some transgrain cracks. In relation to the apparent speci!ic mass (ρa), the three rocks meet the requirements of standard NBR 5564 (ABNT, 2012), which recommends a minimum value of 2,500 kg/m³. As expected, the microgabbro was the rock with the highest ρa (2,940 kg/m³), with a content of 37% heavy minerals (augite, opaques and olivine).   With respect to uniaxial compressive strength, both under dry condition and saturated in water, all the rocks presented values that exceed the 100 MPa required by the NBR 5564 standard (ABNT, 2012), as one of the requirements for use in railway ballast. They had very high (above 170 MPa) to extremely high (above 260 MPa) strength in the case of MOR syenogranite, a solid, homogeneous, compact and !ine-grained rock. Values of σsat decreased for the three studied rocks, with R indices that varied from 95.2% to 93.0%, indicating, respectively, a decline in mechanical strength of the order of 4.8% to 7.0%. The reduction for GAB microgabbro was slightly more pronounced and is related, in large part, to its higher values of porosity, water absorption, microcracking conditions and mineral alteration which, together, are responsible for resulting in a relative weakening of the rock. The average results of all mechanical properties are in Figure 7. MOR and CAN granites exceeded the requirement of strength to Treton impact of 25% recommended by the NBR 5564 standard (ABNT, 2012). However, considering the Los Angeles abrasion and strength to crush tests, the most representative for being performed with larger volume samples, the three samples studied met the regulatory requirements, including the more conservative limits of the EN13450 standard (DIN, 2013), which set forth a maximum LAA abrasion loss of 24% for passenger trains, an index better related to the present study.
These results are valid for the conditions and sampling sites in the quarries visited and should not be used as a criterion for the acceptance or rejection of these rocks for use as aggregate in railroad ballast, since all materials are explored and commercialized for other purposes, notably for use in concrete with Portland cement and in paving. In addition, the more extensive the study area, the more important the geotechnical quali!ication of the rock mass. In practice, as stressed by Remedio et al., (2018), it is essential to carry out studies at the extraction sites to optimize the dismantling and crushing operations and, thus, to meet the strict selection criteria for use as ballast in a large engineering work, as may be the case of the Inter-Cities Train (TIC). Anyway, it is reiterated that the scope of this research was to compare different types of rock aggregates, discussing their suitability and limitations aiming at their use as railroad ballast, but not specifying them and also avoiding any type of commercial connotation as to the quarries studied.
As for the Treton Impact, a limitation of this test refers to the "artisanal" preparation to remove the edges and avoid elongated axes, to approach the equidimensional shape and adapt to the test grade (12.5 to 9.5 mm). Small differences in the shape of the fragments that make up the standard sample can cause the impact of the test socket to be, even partially, absorbed by slightly more prominent fragments (Figure 8). That is, it may not accurately represent the fracture toughness of the tested material. A schematic illustration of this test is given below. In this sense, some international references highlight the need to use the grade effectively used in ballast to specify rocks for this purpose. According to Lim (2004) and Bach (2013), the rupture indices of the fragments are very poorly related to those of the conventional crushing test (that is, with small particles), but very well with tests carried out in proportionally larger devices using larger particles. As for impact strength, the EN 14350 standard describes the sample and test conditions: about 3 kg fragments between fractions 40 and 31.5 mm and an apparatus similar to the national Treton, but with obviously larger dimension. This equipment does not exist in Brazil.
MD and SDI tests are not conventionally applied in aggregate studies, but were used as a reference for mechanical degradation in the alterability tests, thus, their results are discussed together with the alterability tests, below.

Alterability tests
The accelerated alteration tests were carried out considering the feasibility of the experiments and also to check their suitability in testing laboratory conditions that seek to re!lect the conditions of the rocky materials studied.

Soundness of aggregates by use of sodium sulfate
Although the differences in texture between the two granitoids (MOR and CAN) are pronounced, they were not suf!icient to obtain signi!icant results in mass losses. The low degree of alteration of these rocks also contributed to the low laboratory degradation observed, with average mass losses of 0.53% and 0.47%, respectively.
In relation to microgabbro (GAB), with a slightly greater loss of mass (0.67%), a greater effect of the crystallization of sodium sulphate inside the rock was expected during the soundness tests, as it has porosity (1.73%) above speci!ication of NBR 5564 (ABNT, 2012), since it has !ine to medium grains and subophitic texture. Considering the composition, the mineral most intensely altered by chloritization is olivine, which is also the most intensely microfractured mineral, !illed with secondary minerals that correspond to 8% total rock, including opaques, chlorite, !ibrous amphiboles and clays (celadonite and nontronite). This would result in greater cracking of the materials subjected to cycling, which did not occur, possibly due to the discontinuous and intragrain microcraking pattern, as shown in petrographic analyses.
When the rock samples were subjected to soundness cycles and combined with the MD test, the results of the degradation surprisingly were lower compared to the fresh rocks (Table 5), except for CAN granitoid, which showed an increase of only 1% of its value. The low degradation of the rocks due to accelerated alteration with Na2SO4 is because the tensile stress cycles, exerted by the crystallization and dilation of the salts in the rock voids (pores and !issures), were not enough to cause signi!icant fragmentation during the Micro-Deval assay.
As already mentioned for crushing and impact strength, there is also an international speci-!ication for Micro-Deval abrasion. The standard EN 1097-1 (DIN, 2011) presents in Annex A, for tests of railroad ballasts, the 50 mm sieves for passing particles and 31.50 mm for retained particles. It uses 10 kg aggregates, and a cylinder height 2.6 times greater than the conventional drum, without water. In order to better represent the tropical climate conditions, comparative tests with this apparatus and the Los Angeles abrasion test ("G" grade), both with saturated samples, would be advisable. TRANSPORTES | ISSN: 2237-1346

Soundness by immersion in ethyleneglycol
As previously described, the petrographic analysis of the microgabbro indicated the occurrence of expansive clays (celadonite and nontronite) in the microcracks among the secondary minerals that, in the presence of water, produce internal stresses in the rock leading to its degradation. This motivated the performance of soundness tests with immersion in ethyleneglicol, an organic compound that easily penetrates the microcracks of the rock and by expansive forces lead to its fragmentation. With particle size fractions between 19.00 and 9.52mm, comparative tests with ethylene glycol indicated the following mass losses: 0.39% (MOR), 0.69% (CAN) and an expressive value of 41% (GAB), with important evidence of damage (!issures, cracking, !laking and fragmentation) in the samples of this last rock. Considering the amount and the low quality (several fragments) of sample available from material degraded by ethylene glycol (Figure 9B-9C), SDI tests were performed to quantify the mechanical degradation of material saturated in this substance, comparing to samples of fresh rocks as a reference.

Relationships between petrography and alterability tests
In general, the information and evidence presented by the three types of alterability tests are corroborated by the following mineralogical characteristics of the three rocks studied, in part already discussed by Santos et al., (2019): 1) The MOR sienogranite has continuous fractures in two different directions !illed by !ine mass of secondary minerals (muscovite, carbonates, clay minerals and others). The microcracking of the quartz crystals which correspond to 31% of the constituent minerals are rare and closed. These characteristics result in the lower MD coef!icient (4.28%) of the rock when compared to the other two studied. A smaller amount of fragmented material is also observed when the MDR index (88.67%) is applied. This coef!icient represents, therefore, greater resistance of the rock producing more particles retained on the 9.50 mm sieve (Fig. 3). Petrographic aspects also lead to the lowest mass losses observed in the lower values of mass losses veri!ied in soundness tests by immersion in ethyleneglycol.
2) For the CAN monzogranite, the average of the MD (7.3%) and MDR (80.7%) coef!icients are intermediate in relation to the other two non-cycled MOR and GAB rocks. The effect of abrasive wear can be explained by its phaneritic discreetly porphyritic texture, characterized by microcline mega crysts. Microscopically, quartz crystals correspond to 29% of the total rock and have a high degree of microcracking, !illed with secondary minerals (epidote and carbonates).
The petrographic study also showed a recrystallization indicated by the strong undulating extinction of quartz crystals, because they were subjected to conditions of ductile-rupture tectonic stresses.
3) The GAB microgabbro consists of 27% augite and 8% secondary minerals (iron oxides and hydroxides and celadonite and nontronite clay minerals). These minerals have lower abrasive resistance and consequently re!lect in their lower strength, especially when compared to granite rocks where the percentage of quartz crystals exceeds 20%, resulting in a higher average of the coef!icients MD (10.3%) and MDR (79.1%) of the rocks studied. Similar trends veri!ied in the ethyleneglycol oven tests, where the presence, even subordinate, of expansive minerals lead to very intense degradation and fragmentation of the microgabbro samples. These simulations represent, in a more accelerated way, the chemical weathering processes to which the exposed fragments that make up the ballast will be subjected, in a region of tropical climate. The most expressive alterability implies the faster physical-mechanical degradation of this rock. In practice, more frequent, periodic and higher-cost maintenance.

CONCLUSIONS
The three rocks have physical-mechanical characteristics that are highly suitable for use in railroad ballasts, however, as in the studied region they would be exposed to conditions of altitude tropical climate, the microgabbro should not be used, due to the high susceptibility to alterations of its minerals under these climatic conditions.
In the case of the two granitoids studied, for correct application, the necessary tests should be carried out, considering the punctual character of the results, which are from the best quality portions of the rock mass observed in the present study. The technological information already obtained is quite detailed and can serve as subsidies to government agencies responsible for making decisions regarding the use of the materials available in the studied region, for use in ballasted railroads.
As for the tests, the following aspects stand out: • The Micro-Deval test is widely used, especially in Europe, USA and Canada. It provides important information on the behavior of rock aggregates for engineering works, particularly when used in railway ballast and possibly should be part of the NBR 5564 (ABNT, 2012) under review. • It is recommended to continue studies with larger equipment that allow tests on aggregates in the dimensions that make up the railroad ballast. This because with the particle size distribution used in the present study, it was already possible to obtain important results, on wet abrasive wear of the rocks that best represents the !ield conditions, especially when they are used in tropical regions.
• The perspective of using more robust equipment also in crushing and impact tests is also important, aiming at establishing more effective parameters for Brazilian rocks to be speci!ied for conventional railroad ballast or for passenger transportation.
• The values obtained in the SDI tests proved to be good indicators of the more intense degradation of the microgabbro when subjected to the soundness test by immersion in ethylene glycol, which should be mandatory for the characterization of any basic rock, in terms of the potential for use as aggregate without binders and.