Proposal of a New Parameter for the Weathering Characterization of Carbonate Flysch-Like Rock Masses: The Potential Degradation Index (PDI)

The Alicante Flysch sequence (Fig. 1) is composed of pelagic sediments, predominated by sequences of grey marls and thin white marly limestones (hemipelagites) that constitute the rythmite predominated by marls. This sequence may overlap calcarenitic turbiditic episodes. However, the sedimentological complexity of the Flysch formation is even greater because some superposed composite gravitational processes such as mélanges and debrites are also present (Cano and Tomás 2013a).

In this study, five slopes were selected and fully characterized (Fig. 1). 117 intact rock samples were extracted from all of the strata present in the selected slopes and were described in detail in the field. They were geologically classified as: (a) thick bedding calcarenites [Grainstone of turbiditic facies of channel (Ta-b)]; (b) thick bedding calcarenites [grainstone of turbiditic facies of channel (Ta-b) or sheet flood facies (Tb, Tb-c)]; (c) Thin bedding calcarenites [turbiditic thin beds of fan fringe facies (Tb-c-d)]; (d) poorly cemented thin bedding calcarenites [turbiditic thin beds of fan fringe facies (Tb-c-d)]; (e) slightly marly limestones; (f) marly limestones; (g) silty calcareous marls; (h) silty marls; (i) calcareous marls–marls; (j) sheet silty marls; (k) soft marls; (l) sheet marls; (m) soft calcareous mélanges; and (n) calcareous debrites.

At regional level, the area is fractured into several blocks linked to several main dextran fault systems, such as the Cadiz-Alicante fault (N70E), which generated N20E oriented folding, the Vinalopo fault system (N155E) and the Socovos fault system (N120E) that generated N70E oriented folding (Guerrera et al. 2006).

At local scale, six sets of discontinuities were observed in the slopes in the study, five of which are of tectonic origin (J1–J5). The sixth set corresponds to bedding. These five tectonic joints present a quasi-perpendicular disposition to the bedding, generating prismatic blocks with a variety of sizes and shapes for the different lithologies (Fig. 2a).

The two first sets (J1, J2) exhibit a very large persistence (>20 m), and the isolated blocks present a parallelepiped shape. Their average spacings are 28 and 18 cm, respectively. J3 and J4 present a much lower persistence (1.4 and 2.2 m, respectively) and average spacings of 47 and 25 cm, respectively. J5 presents short to very short persistence (1.1 m) and 60 cm of average spacing.

The sixth set of discontinuities corresponds to bedding, which shows a very large persistence. As the studied slopes are heterogeneous, the bedding has been specifically described for each of the lithologies outcropping in the study area. Typically, the thick bedding calcarenites units are 25–120 cm thick and the bedding spacing is 7–30 cm. They are thick bedded blocky, consisting of tabular blocks formed by two to five other intersecting discontinuity sets. The size and shape of these blocks is different to other, less competent lithologies. This is because some discontinuities are well cemented, behaving as if there were no such discontinuity. The slightly marly limestones units are 12–60 cm thick, the bedding spacing is 10–30 cm and they are prismatic blocky. The calcareous mélange units are 50–200 cm thick and exhibit a chaotic structure with a high erratic discontinuity surfaces density that results in heterometric rock blocks (centimetric to decimetric) with different morphologies. The thin bedding calcarenites (C) units are 5–20 cm thick, the bedding spacing is 3–10 cm and they are very blocky with prismatic shape. The thin bedding calcarenites (L) units are 5–30 cm thick and the bedding spacing is 3–10 cm. They are thin bedded blocky consisting of parallelepipedic blocks. The only observed unit of calcareous debrites in the studied slopes is 600 cm thick and exhibits a chaotic structure constituted by blocks and a calcareous matrix with a high erratic discontinuity surfaces density that generates heterometric rock blocks (centimetric to decametric block size) with different morphologies. The only unit of soft calcareous mélange encountered is 20 cm thick and has a chaotic structure. Thin bedding silty calcarenites units are 2–50 cm thick, the bedding spacing is 2–10 cm. They are thin bedded blocky consisting of parallelepipedic blocks. The marly limestones units are 10–80 cm thick and the bedding spacing is 8–30 cm. They are very blocky with prismatic shape, in the same manner as the other lithologies listed below. The silty calcareous marls units are 15–50 cm thick and the bedding spacing is 8–15 cm. The silty marls units are 20–70 cm thick and the bedding spacing is 10–14 cm. The calcareous marls–marls units are 15–160 cm thick and the bedding spacing is 9–30 cm. The sheet silty marls units are 30–160 cm thick and the bedding spacing is 10–23 cm. Soft marls are 25–40 cm thick and the bedding spacing is 10–15 cm.

Regarding openings, it should be noted that in the studied rock masses the presence of infilling in the openings of discontinuities, which are extension joints (not veins), highly depends on the lithology of the unit. As such, only in the following lithologies, which showed high contents of carbonate, are discontinuities filled by free crystals of calcite: thick bedding calcarenites, slightly marly limestones, calcareous mélange, thin bedding calcarenites (C), calcareous debrites, marly limestones, silty calcareous marls and soft calcareous mélange. In these carbonatic lithologies, the opening of joint J1 varies between 0.5 and 10 mm and is completely or partially filled by calcite. The width of J2 ranges from very tight (<0.1 mm) to open (1 mm) and the filling is also made of calcite that can partially fill the discontinuity space or even not be present, leading to a partially open to open discontinuity. Opening ranges are from 0.5 to 5 mm for J3 and from 0.1 to 3 mm for J4. In both cases, the joint is filled or partially filled by calcite. J5 exhibits 1 to 3 mm calcite infilling width (Fig. 2b). Otherwise, in the marly lithologies, the discontinuities also present an aperture varying between <0.1 mm to 10 mm, but without calcite infill. Finally, the joints that correspond to bedding are mainly very tight or they are cemented in the more calcareous lithologies. Additionally, a patina of iron oxide and manganese oxide species in dendritic form has been commonly observed in all sets of discontinuities, including bedding.

Both intact rock and discontinuities exhibited a weathered state, from the face of the slope to certain depth. The degree of alteration changed according to the lithology and the depth, according to Cano and Tomás (2015).

Generally, both, the tectonic discontinuities and the bedding are planar, mainly rough or sometimes slightly rough in the more carbonatic lithologies and smooth or slightly rough in the marly lithologies. Occasionally, the tectonic discontinuities of the units of thick bedding calcarenites, mainly the thickest ones, show evidence of karstification, which generates more undulated or stepped, and very rough joints, and also greater apertures (20 mm). Occasionally, the bedding surface presents load casts at the bottom of calcarenite units of metric thickness at the contact with marly lithologies, generating very rough joints with large undulation. Sometimes, the bedding presents slickensided surfaces.

Water flow was not observed, although after downpours the discontinuities and the intact rock of marly lithologies were wet.

The effect of the excavation method on the disturbance of a rock mass is a well-known phenomenon (Romana 1993; Hoek et al. 2002). However, a large number of factors can influence the degree of disturbance in the rock mass surrounding an excavation, and it may never be possible to quantify these factors precisely (Hoek et al. 2002). Four of the five studied slopes were excavated using mechanical methods and the fifth slope is a natural hill. Due to the heterogeneous nature of these slopes, when the cuts are excavated by means of mechanical methods, the low competence lithologies (e.g., soft marls, silty marls, etc.) are sheared through the rock matrix. However, in the high competence sets (e.g., thick bedding calcarenites) the blocks are broken off through their discontinuities, dislocating them towards the slope face. This is the reason why in the slope face of these sets, the aperture of the discontinuities is higher than that observed in the inner part of the slope, in which the blocks have not been perturbed by the mechanical action of the excavation (Fig. 2c, d). Another cause of the greater discontinuity opening in the more competent blocks from the slope face is the dislocation of rock blocks due to removing underlying unit effects of erosion and/or differential degradation.

The main aim of this paper is to characterize the slaking behaviour of the different carbonate lithologies outcropping in the study area, to aid the prediction of their weathering behaviour following the excavation of a slope or cutting. To this end, the different lithologies were identified and described in the field and their mineralogical characteristics obtained. 5-cycle slake durability tests were performed on intact rock samples. Fragment size distribution curves were also obtained and their morphology analysed. A modified parameter (D _RP) based on the original “disintegration ratio” (D _R) proposed by Erguler and Shakoor (2009), has also been proposed. From D _RP, a novel parameter named potential disintegration index (PDI) based on the change in the D _RP ratio between slake cycles has been defined. The combined use of the PDI, together with the analysis of the shape of the particle size distribution curves, and the behaviour of retained fragments throughout the five cycles of the slake durability test (changes in size and shape) has allowed a new classification of the slake behaviour of these lithologies to be proposed. Additionally, the proposed slake behaviour classes were compared with the weathering patterns and weathering profiles observed in the same lithologies in the field (Cano and Tomás 2015) (Fig. 3). The methodology used in this study is described in the following paragraphs.

In this study, the different lithologies were described in the field using a simplified geological classification of rocks based on their genetic classification, structure, composition and grain size (Geological Society of London 1977). Additionally, a mineralogical characterization of the samples by X-ray diffraction was performed. Because some of the samples were of marly composition, they were characterized in two different stages. Firstly, X-ray diffractograms of all the samples were obtained. Secondly, X-ray diffractograms of the oriented aggregate of samples with high phyllosilicate content were obtained, to identify them according to Robert and Tessier’s (1974) methodology. Finally, for some representative samples, the carbonate contents obtained from the interpretation of the X-ray diffractograms were compared with those obtained using the Bernard calcimeter test (ASTM 2007a), to validate these results.

Data were collected and interpreted using the XPowder software package, (Martin 2004) whose qualitative search-matching procedure was based on the ICDD-PDF2 database.

The Slake Durability Test (SDT) is one of the simplest tests in rock mechanics, and is the most widely used test worldwide for characterizing the environmental weathering resistance of rock. Although originally the slake durability test was developed for testing the weathering potential of shales, mudstones, siltstones, and other clay-bearing rocks (Franklin and Chandra 1972), the slake durability index is also typically used for testing weak rocks such as mudstones, marls, ignimbrites, conglomerates, and poorly cemented sandstones (Sabatakakis et al. 1993; Santi 1998; Czerewko and Crips 2001; Erguler and Ulusay 2009; Miščcević and Vlastelica 2011). Consequently, although in the Flysch formation there are some very competent, hard turbiditic rocks that show very high durability indices, to classify the Flysch lithologies using a uniform weathering potential criteria, the slake durability test was used for testing all of the samples.

The durability of weak rocks is usually assessed using the second-cycle slake durability index (Id₂). Nevertheless, some authors (Gamble 1971; Taylor 1988; Moon and Beattie 1995; Ulusay et al. 1995; Bell et al. 1997; Gökçeoğlu et al. 2000; Erguler and Shakoor 2009; Miščcević and Vlastelica 2011) have suggested that index values taken after three or more cycles of slaking and drying may be useful when evaluating higher durability rocks, such as those in this study.

As part of a previous study, it was observed that within the study area some intact rock samples showed high Id1 and Id2 indices [after Franklin and Chandra (1972) and Gamble (1971)], in contrast with the weathering behaviour observed in situ. As such, it was concluded that the observed weathering of the rocks was much higher than that predicted by the SDT indices (Cano and Tomás 2015).

This appeared to be related to the fact that, despite the high Id₂ values obtained, the sample retained in the drum was extremely fragmented and visually appeared highly degraded. However, the fragments were larger than 2 mm, leading to the high Id₂ values obtained (see Fig. 4).

The American Society for Testing and Materials requires that as part of the slake durability test, in addition to the Id₂ index, the fragments retained in the drum should be qualitatively categorized as type I material (primarily large fragments), type II material (mixture of large and small fragments), or type III material (primarily small fragments) (American Society for testing and Materials (ASTM) 2004). However, Erguler and Shakoor (2009) demonstrated these categories are insufficiently detailed to give a refined classification. As a consequence, it is obvious that the Id₂ index does not adequately reproduce the real degradation properties of the Flysch lithologies studied, providing optimistic values.

It should be noted that rock specimens found on superficial parts of slopes usually show signs of weathering or even severe degradation. As a consequence, the Flysch rock samples tested correspond to intact rocks that were obtained from the inner part of the slope. Subsequently, the intact rock samples were transported to the laboratory in plastic bags, and maintained at a constant temperature. The time between storage and testing was always less than 1 week. While the tests were performed, the laboratory temperature was also kept constant (24 ± 2 °C) to conserve humidity and temperature conditions.

The tests were performed according to the ASTM (2004), with five test cycles performed. Five cycles were chosen because of: (a) the need to compare hard and soft lithologies using the same parameter; (b) the existence of some durable rocks which are unaffected by a low number of cycles; (c) the need to study the rocks’ long-term weathering behaviour; and (d) the need to avoid an excessively long test period.

It is unrealistic to classify carbonate Flysch lithologies according to their in situ weathering behaviour after long-term exposure to real conditions based solely on the indices obtained from the slake durability test. As such, many authors have combined (in various different ways) the results of the slake durability test with the analysis of particle size distribution curves (Erguler and Shakoor 2009; Erguler and Ulusay 2009; Gautam and Shakoor 2013, 2015), with the aim of improving the characterization of various lithologies according to their weathering behaviour.

The procedure adopted as part of this study was as follows. Firstly, ten 40–60 g pieces of intact rock were taken from the study area, providing a total sample mass of 450–550 g. The samples were dried for 24 h in an oven at 105 °C, sieved, and immediately placed in the slake durability test apparatus. After the first test cycle, the sample retained in the drum was dried, sieved and weighed. This procedure was repeated for a further four cycles, giving a total number of five test cycles. The sieving procedure adopted for determining fragment size distribution curves was the same as that used for soils (ASTM 2007b), using standard sieves whose aperture sizes were 40, 31.5, 25, 20, 12.5, 10, 6.3, 5 and 2 mm. The results were plotted in semi-logarithmic scale, to show the fragment size distribution of samples before and after each test cycle. The curves were plotted on the same graph, to easily observe changes in the samples after each slaking cycle. The sieve apertures used were shown on the x-axis in semi-logarithmic scale, and the percent passing (by weight) on the y-axis. It is important to note that the samples were sieved with extreme care, to avoid further fragmentation of the particles retained in the drum, which could have been mistakenly attributed to the effects of the previous slaking cycle.

Erguler and Shakoor (2009) proposed the disintegration ratio (D _R) parameter as the sole indicator of the effects of each slaking cycle on the fragment size distribution curves. It is defined as:where A _C is the area under any size distribution curve and A _T is the total area encompassing the whole range of fragment size distributions. In this study, a similar parameter is proposed. However, as the graphs used show percent passing (by weight), as opposed to percent retained (by weight), when the parameter is equal to 1 (A _C = A _T) this represents the maximum degree of degradation possible. To avoid confusion, the proposed new parameter has been named D _RP (Fig. 5).

Despite the fact that D _RP can be a good indicator of the degradation potential of the weakest rocks, it has been observed that the size and shape of the fragments that are retained in the drum between cycles changes from one cycle to the next, throughout the five slaking cycles. This observation agrees well with in situ observations, where the form and visual condition of lithologies change over time. For this reason, it is proposed that the change in the calculated D _RP value between slaking cycles is evaluated. To this end, a logarithmic curve was fitted to the D _RP values obtained for each sample and the R ² and typical error values were calculated (Fig. 6). From this curve, the number of cycles required for a sample to reach 50 % of the maximum possible degradation (D _RP = 1) could be estimated. This number of cycles is denominated N ₅₀. In the example shown in Fig. 6, from the fitting equation:

The Nr value for D _RP = 1 is Nr = N ₅₀ = 290 cycles. However, owing to the fact that the slaking resistance of the rocks in the study varied greatly, the range of N ₅₀ values was very large, varying from N ₅₀ = 2 in rocks that were very susceptible to degradation, to N ₅₀ = 8.10¹⁹ in rocks which were not. As such, a new parameter was defined to aid the classification of samples—the Potential Degradation Index (PDI). This is calculated as:

The values calculated for the samples in this study vary from 0.8 to 46, with a value of PDI = 5.7 calculated for the example in Fig. 6.

As such, using only the Potential Degradation Index, the potential long-term degradation of a given sample from a carbonate lithology may be assessed. However, with the aim of refining the classification limits, to better distinguish between the numerous lithotypes present in the study area, a qualitative study was performed on the fragments retained in the drum during the different cycles of the slake durability test, and a classification was proposed. Three factors are proposed to standardize this classification: roundness, number of fragments, and fragment size (Table 2; Fig. 7).

Using these standard factors, 11 slaking behaviour patterns were defined based on the changes observed in the fragments. A distinction is made between three different textures present in the lithotypes: compact, laminated and poorly cemented (Table 3).

A qualitative analysis of the changes in morphology in the six fragment size distribution curves obtained for each sample after five slake durability test cycles was performed. The curves showing a similar morphology were grouped.

Using all of the aforementioned parameters (behaviour in terms of changes in the retained fragments, changes in the shape of the fragment size distribution curves, and principally the potential degradation index parameter), all of the samples taken as part of the study were classified, and limits between the different classes were established.

However, the procedure required for this classification was somewhat laborious. To improve ease of use, the changes in the morphology of the fragment size distribution curves throughout the five slaking cycles were analysed quantitatively, and a parameter which is simpler to obtain (but correlates well with the potential degradation index) was proposed. For the 117 samples in the study, parameters habitually used in sieve analysis in soil mechanics or sedimentology were calculated, such as the coefficient of uniformity (C _U), curvature coefficient (C _C), diameter for which 50 % of the sample passes (D ₅₀), percentage of sample passing a certain sieve size, etc. as well as combinations of these parameters which described the shape of the fragment size distribution curves. Using the IBM SPSS Statistics program, possible correlations between these parameters and the Potential Degradation Index were investigated, with the R ² and typical error values calculated for each curvilinear approximation. The parameters that showed the best fit were the D ₅₀ range (i.e. the variation between the initial D ₅₀ and the value after the fifth slaking cycle) and the sample percentage passing through the 12.5 mm sieve (P_12.5). These parameters have the additional advantage of quantifying the changes in shape of the fragment size distribution curves (Fig. 8). Of these two parameters, the best correlation was given by P _12.5. The parameter also has the advantage of allowing the same number of classification groups to be established.

Once all of the samples were classified, the behaviour of the different lithotypes from which the samples were extracted was studied and compared with the weathering patterns and weathering profiles defined by Cano and Tomás (2015).