1. Introduction
Aggregates with distinguishable different origins are widely utilized for a variety of construction applications especially as a road stone, railway ballasts, and some concrete applications. An enormous amount of aggregate is used annually worldwide. Currently, the demand of crushed stone aggregates increases because of the increasing expansion of highway and other construction projects and the decreasing availability of global natural aggregate resources. Different types of rocks have different impacts on construction. The quality of aggregates is of considerable significance in determining their suitability for any engineering application [
1].
Igneous rocks are commonly hard and dense, which results in an excellent source of aggregate materials. However, certain extrusive rocks are too porous to be used as aggregates while some highly siliceous igneous rocks tend to chemically react with alkali when they are used as concrete aggregates. Fractures in some rocks may render them unsuitable for aggregate use. Similarly, some lava-flow rocks are considered unsuitable for aggregates. When they contain flow banding, they are strongly jointed or brecciated. Furthermore, pyroclastic volcanic materials such as ash and tuff may be unsuitable unless they have become indurated by heating or compacted and cemented during burial.
The study of the engineering properties of rock materials as well as their respective mineralogical and textural characteristics decisively determines the rock’s strength and its capability from failure [
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12]. The engineering parameters of igneous rocks are controlled by several inherent and environmental parameters while one of the most significant parameters is the alteration. The inherent parameters can be determined by their petrographical properties, which control the engineering properties of igneous rocks. Many studies have been concentrated on granitic rocks [
13,
14]. However, several authors have conducted analogous studies in acidic-intermediate volcanic rocks [
15,
16], metamorphic rocks [
17,
18], and mafic [
5,
19] and ultramafic rocks [
11,
12,
19,
20,
21]. Alteration is a critical factor as increased percentages of certain secondary minerals such as serpentine and talc affect negatively the mechanical properties of ultramafic aggregates due to their layered structure, cleavage, and platy or fibrous crystal habit [
11,
20]. Chlorite is a common secondary mineral in mafic rocks and it is known to have a critical effect on the freeze-thaw durability of the aggregates in concrete. Clay minerals are common secondary minerals in intermediate-acidic rocks such as andesites and dacites.
The prediction of one engineering parameter through the other is an important field of research. Kazi and Al-Mansour [
22] obtained strong correlations between uniaxial compressive strength, Schmidt hammer, and Los Angeles abrasion after testing volcanic and plutonic rocks. Chargill and Shakoor [
23] established a non-linear inverse relation between the compressive strength and LA abrasion after testing sedimentary and metamorphic rocks. Christensen [
24] focused on the relationship between the mechanical strength of serpentinized rocks with their physical and mechanical properties. Kahraman [
25] studying a variety of igneous, sedimentary, and metamorphic rocks has reported good correlations between the uniaxial compressive strength and the LA abrasion. He has also mentioned that, when these rocks are classified into classes of porosity, the correlation coefficients increase. Ugur et al. [
26] have also pointed out high correlations between the LA abrasion and the compressive strength, Schmidt hardness, and the point load index in a variety of aggregate rocks. Giannakopoulou et al. [
9] mentioned inverse relationships between the point load index, the compressive strength, and the LA abrasion for ultramafic aggregate rocks.
The most common statistical method used to correlate geological data and more specifically engineering properties of aggregate rocks is regression analysis [
2,
8,
9,
11,
12,
20,
27]. Some researchers have noticed the importance of other statistical methods such as factor analysis and Q-mode analysis in a wide range of geological subjects [
28,
29,
30,
31]. Factor analysis is a multivariate, well-known statistical technique, which uses uncorrelated variables called factors and explain the variance observed in the original dataset [
32,
33]. This technique has been successfully applied in hydrochemistry [
34,
35], geochemistry [
36,
37,
38,
39,
40], and less in engineering geology [
41,
42].
The goal of this study is to investigate the interrelationships between the engineering parameters of igneous rocks derived from various localities in Greece (
Figure 1) using factor analysis. In order to export when more representative results could be exported, a wide range of igneous lithotypes were collected and studied, characterized by a great variety of engineering properties, and of petrographic features, which enhances the suggestions of previous researchers by the interaction of the engineering parameters for having as much statistical significance as possible.
3. Materials and Methods
In order to investigate the petrographic characteristics and the engineering properties of the aggregate rocks, ultramafic, mafic, and intermediate-acidic aggregate blocks were collected from quarries and other outcrops from the studied areas, according to the EN 932-1 [
55] standard. The samples were subsequently prepared in order to be suitable for all the engineering tests, which were performed according to European and International standards.
The petrographic features of the studied samples were examined in thin sections using a polarizing microscope (Leica Microsystems Leitz Wetzlar, Germany), according to the EN-932-3 [
56] standard for a petrographic description of aggregates. The mineralogical composition of the studied samples was also determined with X-ray Diffraction, according to EN-932-3 [
56] using a Bruker D8 advance diffractometer with an Ni-filtered CuKα radiation. Random powder mounts were prepared by gently pressing the powder into the cavity holder. The scanning area for bulk mineralogy of specimens covered the 2θ interval 2–70° with a scanning angle step size of 0.015° and a time step of 0.1 s. The mineral phases were determined by using the DIFFRACplus EVA 12
® software (Bruker-AXS, Gmbtl, Karlsruhe, Germany) based on the ICDD Powder Diffraction File of PDF-2 2006.
The determined physical properties are the moisture content [
57], total porosity, and dry density [
58]. Geometrical properties included the flakiness index (I
F) [
59] and the elongation index (I
E) [
60]. The studied samples have been crushed in a laboratory-jagged crusher. The mechanical properties of the Los Angeles abrasion value (LA), uniaxial compressive strength (UCS), the point load index (Is
(50)), and the Schmidt hammer value (SHV) were also determined. The Los Angeles abrasion (LA) test measures the resistance of aggregates to abrasion, attrition, and grinding, which indicates that the lower LA abrasion values of rocks correspond to more resistant rocks in abrasion and attrition. This test was carried out in accordance to the ASTM C-131 [
61] standard using the “B” gradation. The uniaxial compressive strength (UCS) is one of the most significant engineering properties of rocks. The UCS test was carried out in six cylindrical rock specimens with height/diameter ratios between 2 and 3. Their diameters range from 51 to 54 mm (ASTM D-2938 [
62]) and the average values were used for each set of specimens. The point load index (Is
(50)) is used in order to obtain an indirect measure of the uniaxial compressive strength, according to the ISRM [
63] standard. The Schmidt hammer test is a non-destructive method to characterize the rock hardness and strength. The test was carried out using the L type Schmidt hammer on cylindrical specimens [
58]. The physicochemical properties, which were calculated for this study is the Soundness test (S) [
64] and the methylene blue test (MB
F) [
65]. The soundness test is used for the assessment of the ability to resist the aggregates in the excessive volume changes relative to the changes in the physical environment. The MB
F is an indirect method for determining the swelling clay minerals in the aggregate rocks. This test was determined on the aggregate fraction of 0–0.125 mm. Furthermore, loss on ignition (LOI) in all samples was determined according to the ASTM D7348-13 standard [
66].
5. Discussion
The study of the engineering parameters of rocks is of special significance since they are extensively used in many engineering projects. Relationships between mechanical properties have been reported by Ugur et al. [
26], Kahraman [
25], Petrounias et al. [
11], Kazi and Al Mansour [
22], and Al-Harthi et al. [
1] investigating various igneous, sedimentary, and metamorphic aggregate rocks. A number of researchers such as Christensen [
24] and Petrounias et al. [
11] have also stated interrelationships between physical and mechanical properties. Moreover, several scientists have suggested negative correlations between the total porosity and the dry density (ρ
d) [
68,
69,
70,
71,
72]. In addition, numerous other researchers have interpreted the behavior of mechanical parameters in relation to their petrographic characteristics. Fortes et al. [
73] stated that mineralogy combined with the textural and physical characteristics such as porosity and moisture content are modulatory factors for the mechanical behavior of aggregate rocks. Sabatakakis et al. [
74] have shown a direct influence of microstructure on the strength of various sedimentary and igneous rocks. Numerous scientists have studied the impact of primary or secondary minerals contained in a variety of lithologies on their physical, mechanical, and physicochemical properties [
6,
11,
12,
75,
76,
77,
78,
79]. In this paper, the interrelationship between the engineering parameters was identified by using factor analysis based on the petrographic characteristics of the tested igneous rocks used as aggregates.
The R-mode factor analysis suggested a three-factor model for expressing the interrelations of the investigated parameters. The three factors reflect the three different strong trends among groups of interrelated parameters. The differences of the parameters are associated with differences of the petrographic characteristics of the studied rocks.
The poles of the engineering properties of the first factor (~36% of the total variance) are inversely correlated. The positive pole contains high loadings for LA abrasion, moderate loadings for n
t and S, and weaker loadings for w and I
E, which indicates that, with the increase of n
t, w, I
E of the investigated igneous rocks, their resistance in abrasion (LA) decreases. Additionally, the decrease of the resistance in excessive volume changes (S) is related to the increase of n
t, w, I
E of the investigated igneous rocks. The positive pole overall highlights the strong interrelation between the resistances of the investigated rocks in abrasion (LA) with their total porosity (n
t). There are several researchers who have associated physical properties such as n
t with the petrographic characteristics of the aggregate materials [
11,
12,
80]. Generally, in
Table 5, high correlations have been depicted between physical and mechanical properties such as n
t and LA, w and UCS, and n
t and UCS. These high correlations are due to the wide range of the engineering property results of the studied samples depending on their variable petrographic characteristics, which is shown in
Table 2. The secondary phyllosilicate minerals (i.e., serpentine, chlorite, clay minerals) seem to be the dominant minerals influencing the engineering properties of igneous rocks used as aggregates [
11]. Serpentine is the dominant secondary mineral in the studied ultramafic rock samples, which has been observed by the microscopic study, and it seems to determine n
t as well as w and S. Rocks with a high content of serpentine such as BE.01 (
Figure 2b) was presented as more capable to incorporate water in their structure because of the ability of serpentine to form foliated masses contributing to the development of more porous areas, which result in higher values of n
t and in higher values of w in contrast to less serpentinized ultramafic rocks such as GE.34, GE.30, and BE.67 (
Figure 2a,c,d). Furthermore, the existence of the mesh texture of serpentine creates weak planes, which allow hydrous solutions (MgSO
4) to flow along them. Subsequent crystallization and expansion of the salts cause failures of the rock structures [
12], which highlights the strong interrelation of n
t with S. Chlorite. This is the dominant alteration product of the mafic rocks and significantly influences their total porosity. Rocks with high contents of chlorite such as ED.66A (
Figure 4d) are considered more capable to incorporate water in their structure due to the platy and tabular structure of chlorite contained. This acts in similar ways to serpentine and results in higher values of n
t, w, S, and, consequently, LA abrasion in contrast to rocks such as KIL.5 and KIL.3 with less to minor chlorite contained [
11]. The intermediate-acidic rocks present big differences in their engineering properties. Both albitite and granodiorite are plutonic and, therefore, quite compact rocks present mainly a granular texture responsible for their good cohesion and low porosity (
Figure 6c,d) (BE.139, BE.150), which results in low n
t values and, subsequently, contributes on their low LA abrasion values. Dacites and andesites include many altered phenocrysts of plagioclase, which commonly transform to clay minerals and particularly in swelling clay minerals. These minerals, even in low percentages, are capable of adsorbing water in their phyllosilicate structure, which results in the increase of n
t and secondarily of w [
11]. Samples BE.82B, BE.101B, and GE.22 potentially contain a low amount of swelling clay minerals presenting higher n
t values compared to the other volcanic rocks, which contributes to the decrease of their resistance in abrasion (LA).
The negative pole shows strong negative loadings for SHV and moderate loadings for UCS and Is
(50), which indicates that these three different mechanical tests reflect the mechanical strength of rocks and present positive trends among them, which can be seen from
Table 5. The relationship between SHV with UCS and Is
(50) as well as the relation between the last two parameters are positive due to the similar nature of these mechanical tests. Sabatakakis et al. [
81] and Giannakopoulou et al. [
9] have also reported similar pairs of relationships. In addition, they are in accordance with Rigopoulos et al. [
42] who examine various ophiolitic rocks. UCS is negatively related with LA abrasion since the presence of phyllosilicate minerals may create artificial surfaces of a weakness that results in the decrease of UCS and the increase of LA abrasion values simultaneously (i.e., serpentine and chlorite in ultramafic and mafic rocks, respectively) (
Table 5). The petrographic study verifies the above relation as the most altered (ED.59, ED.26B) ultramafic and mafic samples presenting lower UCS values and lower resistance in abrasion in contrast to less altered ones (GE.34, ED.93). This interrelation is in accordance with Kazi and Mansour [
22], Kahraman [
25], and Ugur et al. [
26]. Furthermore, physical properties such as w and n
t seems to influence negatively UCS [
11,
24] (
Table 5). Regarding the intermediate-acidic rocks, they present a great variety in their mechanical properties (
Table 2) due to their variable petrographic features. More specifically, this range is due to the further discrimination of this lithological group into plutonic and volcanic rocks as well as the different textures contained in these rocks. The porphyritic texture of the volcanic rocks seems to influence negatively on the mechanical behavior of the tested samples in contrast to the granular texture of the plutonic ones.
The second factor (~27% of the total data variability) is also bipolar and correlates physical, physicochemical, mechanical properties as well as LOI, which is considered to be an indirect index for the alteration degree of rocks expressed by the presence of serpentine in ultramafic rocks and by the presence of chlorite in mafic rocks [
7]. Other researchers [
7] have also cited similar ranges in LOI values to those of
Table 2. More specifically, the positive pole of this factor displays high loadings for MB
F, moderate loadings for physical w, and weak loadings for n
t, S, LOI, which indicates that mineralogical components are the key parameters influencing the MB
F (GE.4, BE.01, BE.82B, and BE.88). This happens because swelling clay minerals can adsorb more water in their structure than other minerals. Similar conclusions have been cited in trachytes by Rigopoulos et al. [
42]. The negative pole alike enhances the moderate relations between the physical and the mechanical properties such as ρ
d with UCS and Is
(50) [
2], which can be seen in
Table 5 due to the wide range of these engineering properties (
Table 2) relative to their various petrographic features.
The third factor has little effect on the engineering properties (~14% of the total variance). It presents only a positive pole correlating the studied geometrical properties with the LOI. More specifically, it shows high loadings for I
F and moderate loadings for I
E and LOI, which indicates that the flakiness and the elongation index increase in more altered rocks (BE.01B, BE.12, BE.103, and GE.26). Regarding the ultramafic rocks, these relationships attributed to the presence of serpentine, which belongs to the phyllosilicate subclass of minerals and promotes the production of flaky and elongated aggregate particles during the crushing process [
27]. In mafic rocks, the presence of chlorite is responsible for the increase of I
F and I
E since it also belongs to the phyllosilicate subclass of minerals acting similarly to serpentine.
The first two factors together account for the ~63% of the total variance and, hence, the factor scores (
Table 6) of the first factor plotted against the factor scores of the second one on a scatter diagram are shown in
Figure 8. In this diagram, we can observe the engineering properties of the collected igneous aggregate rocks, which have been grouped into the positive and negative poles of factor 1 and factor 2. In order to give the best interpretation of the above relations, the diagram was divided into two clusters (A, B) of the samples while the samples display further differentiations.
In
Figure 8, we can observe two groups of ultramafic rocks that present a clear gap between them and are detected in two clusters A and B. We also observe two groups of intermediate-acidic rocks divided into cluster A and B. The investigated mafic rock samples do not appear to have significant geological variation, which results in the two observed groups fitting within the same cluster (B). More specifically, rocks detected in cluster A are the most serpentinized tested ultramafic rocks as well as the intermediate-acidic volcanic rocks (dacites and andesites), which display higher values of n
t, w, S, LA, I
E, MB
F, and LOI than those of cluster B. Giannakopoulou et al. [
12] investigated the engineering properties of ultramafic rocks and have concluded that, with the increase of the serpentine percentage, n
t, w, LA, and S increased, respectively. Petrounias et al. [
11] found similar conclusions regarding the relation between n
t and LA. Furthermore, samples of cluster A display lower values of ρ
d and of mechanical properties such as UCS, SHV, and Is
(50) when compared with those of cluster B. For example, samples GE.30 and BE.67, which display a low percentage of serpentine (
Figure 2c,d) have lower values of n
t, w, S, LA, and I
E and higher values of UCS, SHV, and Is
(50) (
Table 2). This is in contrast to sample BE.01, which is presented as more serpentinized (
Figure 2b), and presents high values of n
t, w, S, LA, and I
E and lower values of the referred UCS, SHV, and Is
(50) (
Table 2). The plutonic intermediate-acidic rocks (granodiorite and albitite). These are detected in the B cluster presenting low values of n
t, w, S, and LA, and simultaneously high values of UCS, SHV, and Is
(50) in contrast to the volcanic intermediate-acidic rock samples (dacites and andesites), which are detected in cluster A.
Rigopoulos et al. [
42] proposed similar relationships between engineering parameters when investigating data of common lithologies. In this study, more representative rocks have been investigated such as in the number of lithological type indicating similar results to Rigopoulos et al. [
42] and, simultaneously, presenting more statistically accurate conclusions about the engineering behavior of these types of rocks.