Evaluation of rock mechanical behaviors under ... - Springer Link

Rock Mech Rock Engng (2009) 42: 73–93 DOI 10.1007/s00603-008-0170-2 Printed in The Netherlands

Evaluation of rock mechanical behaviors under uniaxial compression with reference to assessed weathering grades By

A. Basu, T. B. Celestino, A. A. Bortolucci Departamento de Geotecnia, EESC, Universidade de S~ao Paulo, S~ao Carlos, Brazil Received April 18 2007; Accepted January 24 2008; Published online May 5 2008 # Springer-Verlag 2008

Summary A weathering classification for granitic rock materials from southeastern Brazil was framed based on core characteristics. The classification was substantiated by a detailed petrographic study. Indirect assessment of weathering grades by density, ultrasonic and Schmidt hammer index tests was performed. Rebound values due to Schmidt hammer multiple impacts at one representative point were more efficient in predicting weathering grades than averaged single impact rebound values, P-wave velocities and densities. Uniaxial compression tests revealed that a large range of uniaxial compressive strength (214– 153 MPa) exists in Grade I category where weathering does not seem to have played any role. It was concluded that variability in occurrences of quartz intragranular cracks and in biotite percentage, distribution and orientation might have played a key role in accelerating or decelerating the failure processes of the Grade I specimens. Deterioration of uniaxial compressive strength and elastic modulus and increase in Poisson’s ratio with increasing weathering intensity could be attributed to alteration of minerals, disruption of rock skeleton and microcrack augmentation. A crude relation between failure modes and weathering grades also emerged. Keywords: Weathering classification, granite, index tests, uniaxial compressive strength, Young’s modulus, Poisson’s ratio, failure modes

1. Introduction The ongoing process of weathering in nature produces progressive but intricate changes in the chemical, mineralogical and physical, and thus microstructural properties of rock materials especially at shallow depths where most engineering works are confined. As southeastern Brazil experiences a subtropical climate, chemical weathCorrespondence: Prof. T. B. Celestino, Universidade de S~ao Paulo, S~ao Carlos, Brazil e-mail: [email protected]

74

A. Basu et al.

ering is the dominant weathering process in this region. Being the most developed and industrialized area of Brazil, this region experiences frequent engineering activities (Zuquette et al., 1994) where rock weathering is a major concern. Nevertheless, a lack of studies focusing mainly on gradational changes of rock materials from this region due to increasing weathering intensity, quantitative assessment of the degree of weathering by index tests and rock mechanical behavior with reference to weathering grades seems to exist. The present study is to explore all these issues for granitic rock materials from southeastern Brazil. The investigation consists of three major components: 1) framing a weathering classification for the granitic rock materials based on core characteristics followed by substantiation of the framed classification by a detailed petrographic study; 2) indirect assessment of weathering grades by density, ultrasonic and Schmidt hammer index tests and 3) a critical evaluation of rock mechanical behavior under uniaxial compression with reference to degree of decomposition.

2. Samples Granitic core samples of various weathering grades collected from the site of Porto Goes Hydroelectric Power Plant were used for the investigation. No soil material was included in this study. The rocks belong to the Itu Granitic Complex (IPT, 1981) and NW–SE trending faults are predominant in the proximity of the site (Fig. 1). Therefore, only intact and uniformly weathered cores (devoid of densely spaced fractures, any shear signature or mixed lithology) were selected for the investigation. Preliminary visual inspection of the cleaned core surfaces ascertained comparable mineralogy and texture of the samples. All samples, brown=pinkish brown in color, were medium grained granites and displayed equigranular phaneritic texture.

3. Weathering classification ‘‘The descriptions and classifications of weathered rocks for engineering purposes has been a subject of debate since engineering geologists first produced standards and codes’’ (ANON, 1995). Consequently, a number of classification schemes for both weathered rock mass and material have evolved in last five=six decades. The most widely used weathering classification systems for rock materials in the world (e.g. ANON, 1995; BS 5930, 1999) by and large resemble the 6-fold classification scheme developed by Moye (1955) in which Grades I–IV represent rocks whereas higher grades stand for soils. The grades in this 6-fold scheme are based on chemical rather than mechanical weathering (Hencher and Martin, 1982). As mentioned before, in southeastern Brazil, chemical weathering supported by rock structural discontinuities is the most dominant weathering process. In the borehole logs of the collected cores, degree of alteration is indicated according to the proposed classification by Vaz (1996). However, this classification is mainly about rock masses and not for materials. Moreover, in this classification, there are only three divisions of weathered states of rocks whereas it is well-acknowledged that although the common 6-fold scheme is adequate for general descriptions, even subdivision of

Evaluation of rock mechanical behaviors under uniaxial compression

75

Fig. 1. Site location and geology (IPT, 1981) in and around the site

the grades may be justified if a more detailed description is required; for example while relating laboratory test results with degree of decomposition. In the present study, a set of five recognition factors (discoloration and=or staining, grain luster, grain boundaries, relative strength and disintegration) was identified and

76

A. Basu et al.

Table 1. Characteristics and assigned weathering grades (WG) of the investigated granitic materials Granite cores

Characteristics

WG

No discoloration Grains have vitreous luster Equigranular texture with intact grain boundaries

I

Slight to moderate staining Grains have vitreous to sub-vitreous luster Intact grain boundaries

II

Grains have sub-vitreous to dull luster White clay minerals are common Intact grain boundaries Not easily broken by a geological hammer

II–III

Moderately decomposed Abundant soft white clay minerals can be scratched by nail Intact grain boundaries Can be broken easily by a geological hammer

III

Highly decomposed (powdery feldspars) with loose grain boundaries Large pieces can be broken by hand Does not readily slake in water

IV

used to describe gradational changes of the granitic materials caused by weathering. The idea was to capture detailed decompositional variations from the rocks. Table 1 presents the weathering classification for the granitic materials which is by and large conformable to the common 6-fold material classification scheme. However, an intermediate class ‘Grade II–III’ is assigned in the Table 1 with its distinct differentiable decompositional characteristics compared to Grades II and III. It should be noted that ‘‘fresh’’ rock is straight out of the oven – only really found at great depth. In the present study, therefore, Grade I rocks are the cores for which there is no perceptible signature of weathering (within shallow depths where most engineering works are confined). 3.1 Petrography The weathering grades in Table 1 have an indirect connotation to gradational decomposition of individual minerals and induced microcracking. Although petrographic


77

Fig. 2. Photomicrographs according to weathering grades

study is useful to gain a direct control on these issues, it has not received significant attention in regular engineering environment because of time-consuming nature of such study (Martin, 1986). A detailed petrographic study of thin sections (orthogonal to core axis) was undertaken to ascertain the assigned weathering grades.

78

A. Basu et al.

Thin section study validates the equigranular texture with intact grain boundaries (contrast: prominent, sharpness: sharp) and the granitic composition (by volume) of Grade I rocks composed mainly of K-feldspar, quartz, plagioclase and biotite (Fig. 2a). The term ‘equigranular’ is used here essentially to indicate a texture devoid of any recognizable phenocrysts and groundmass. The rocks are ‘medium-grained’ with an average grain size of 2–6 mm (Hutchison, 1974). Cloudy K-feldspars (subhedral, 2– 14 mm) do not show any sign of alteration. Microperthitic textures are common. Quartz grains (anhedral, 0.2–8 mm) are diagnosed readily by their typical first order gray interference color or by distinctive yellowish=whitish appearance and absence of cleavage, twinning and alteration. Plagioclases (euhedral to subhedral, 0.1–6 mm) commonly display lamellar twinning and are unaltered in general; however, occasional sericitization points to pre-weathering=deuteric alteration. Biotite grains (euhedral to subhedral, 0.3–3.5 mm) have their unique strong pleochroism in shades of brown and green with no sign of alteration (Fig. 2b). Opaque minerals are present as accessory primary phases. Intragranular microcracks in quartz grains are occasionally noticeable (Fig. 2c). Except cleavage cracks and a few tiny cracks across cleavage planes, biotite hardly exhibits other cracks. However, microcracks occur infrequently around biotite grains especially when the neighboring minerals are quartz. With the onset of weathering (Grade II), slight staining is apparent. Although the minerals do not show any significant increase in degree of alteration (Fig. 2d), slight sericitization of plagioclases (Fig. 2d) and chloritization of biotites (Fig. 2e) are common. Quartz grains display more intragranular microcracks than the Grade I rocks and sometimes, transgranular cracks across quartz and K-feldspar grains are noticeable (Fig. 2f). Intragranular cracks in K-feldspars are also observed. With the advancement of weathering (Grades II and III), plagioclases become slightly to moderately sericitized (Fig. 2g). Slight to moderate chloritization of biotites is also common (Fig. 2h). Grain boundaries, although stained, are intact in general. Transgranular microcracks at this stage are more common than in Grade II rocks (Fig. 2i) and are sometimes filled (with Fe-oxides or with unrecognized materials). Intragranular cracks in plagioclases are rare. Microcracks often occur around biotites. As weathering intensifies (Grade III), rocks become notably stained and the grain boundaries are intact to semi-intact (contrast: distinct to faint, sharpness: clear to diffuse) (Fig. 2j). Plagioclases are highly to completely sericitized (Fig. 2j). Biotites are moderately to highly chloritized (Fig. 2k). Slightly to moderately decomposed K-feldspars at times show more intragranular cracks than the lower grade rocks (Fig. 2l). With further advancement of weathering (Grade IV), rocks are considerably stained with open grain boundaries (Fig. 2m). Completely altered plagioclases and biotites are common (Fig. 2m and n, respectively). Moderately to highly decomposed K-feldspars are noticeable (Fig. 2m). Microcrack density drops significantly compared to lower weathering grades as considerably augmented clayey domains at this stage of weathering mask the presence of microcracks under microscope (Fig. 2o). Thin section appearances in terms of staining, alteration of individual minerals, nature of grain boundaries and microcrack patterns depict a clear gradational weathering effect on the investigated rocks. Thus, the petrographic study substantiates the framed weathering classification.


79

4. Prediction of weathering grades by index tests Descriptive 6-fold intact rock (or rock material) weathering classification scheme for engineering purposes is formulated to address the need for a common but simple basis of communication with underlying messages mainly on the possible ranges of mechanical properties. Because the six-fold weathering classification scheme is based on subjective criteria, identifying and assigning weathering grades objectively and quantitatively by index tests have obvious advantages (Dearman and Irfan, 1978; Irfan and Dearman, 1978; Hencher and Martin, 1982; Martin, 1986; Martin and Hencher, 1986 etc.). It should be noted that indices themselves can be useful measures for some classifications which may directly shed light on correlatable rock mechanical properties without referring to any ‘‘weathering classification’’. But, as the weathering classification has been used for over decades and pervaded all aspects of geological engineering in weathered profiles, it seems unavoidable to provide a quantitative link between indices and weathering grades. Moreover, while considering rock masses, one should bear in mind that zones of different weathering grades may be very difficult to distinguish (de Mello, 1972), as they merge into one another and their depths of occurrence and thicknesses vary erratically. In these contexts, the index with minimum overlapping of ranges in relation to weathering grades of a followed weathering classification for a particular rock type with a certain geological history should be considered as the best index in discriminating different weathered states of that rock. A large number of studies have been devoted to achieve this goal and the Schmidt hammer has drawn considerable attention in recent years in this issue. Preceded by a background on weathering grade assessment by the Schmidt hammer, the present study evaluates indirect assessment of weathering grades of the investigated rocks by Schmidt hammer test along with density and ultrasonic tests.

4.1 Background on weathering grade assessment by the Schmidt hammer The Schmidt hammer consists of a spring loaded piston. When the hammer is pressed orthogonally against a surface, the piston is automatically released onto the plunger and the rebound height of the piston is considered to be the index of surface hardness. Partial consumption of the impact energy is caused by the interaction between the plunger and the surface and by the mechanical friction in the instrument. As the Schmidt hammer has got a plunger diameter of 15 mm, energy absorptions by polymineralic rock surfaces depend on the number, proportional areas and bonding of the constituent mineral grains the plunger hits considering other factors (such as surface smoothness, moisture content and mass of the materials) are invariable. Therefore, even from a single rock type, rebound value ranges of adjacent weathering grades are very likely to overlap as weathering produces intricate changes in mineralogy and microstructures and characteristic rebound value of a rock material is determined averaging several random single impact rebound values from its surface. A brief literature review is presented here to express the nature of relation between rebound value ranges and degree of decomposition. The weathering classifications (in material or mass scale) used in the following references comply by and large with the

80

A. Basu et al.

conventional 6-fold classification scheme. Hencher and Martin (1982) advocated that Schmidt hammer could be used as an index tool over the full range of weathering. They provided non-overlapping in situ N hammer value ranges (>45, 25–45, 0–25, no rebound) for differentiating weathered states (from Grades II to V) of igneous rocks of Hong Kong. Ebuk (1991) referring this work, however, indicated that 12% of 133 samples otherwise graded as highly decomposed (Grade IV) showed no rebound by N hammer. Irfan and Dearman (1978), based on the investigation of SW England granite, advised that the Schmidt hammer should be used only for relatively strong materials giving rebound value > 40. Karpuz and Pasamehmetoglu (1997) proposed distinct L hammer rebound value ranges (54–61, 39–54, 28–39, 18–28 and 2:1) were air-dried to constant mass. Dry density (dry), the most often used physical index, was calculated from the measured volume and mass for all specimens. The specimen numbers within each assigned weathering grade are as per the descending order of dry (Table 2). For ultrasonic and Schmidt hammer investigation, a tailor-made test setup schematically represented in Fig. 3 was used. Direct pulse transmission technique was employed for ultrasonic testing with the coaxial arrangement of the specimen and the transducers at a constant coupling pressure. P-wave velocities (VP) were determined (Table 2) from the measured travel times through the specimens. Five random test points (any two points separated by more than one plunger diameter) on each specimen surface were selected to apply each of L- and N-type Schmidt hammers (impact energies: 0.735 and 2.207 Nm, respectively). Considered five single impacts by each hammer were those which did not cause any visual damage of the


81

Table 2. Dry densities, ultrasonic velocities and rebound values Sp. No.

WG

dry (gm=cm3 )

VP R L (m=sec) mean

RN mean

Range dry

VP

RL

RN

S1 S2 S3 S4 S5

I I I I I

2.67 2.66 2.66 2.65 2.65

5201 5514 5218 5362 5423

53.12 55.38 51.60 52.55 53.49

63.04 64.74 63.60 62.09 62.28

2.67–2.65 5514–5201

55.38–51.60 64.74–62.09

S6 S7 S8 S9 S10

II II II II II

2.65 2.64 2.64 2.63 2.63

4726 4952 4859 4698 4733

51.98 50.85 48.77 46.69 47.06

59.07 60.01 57.75 56.61 57.56

2.65–2.63 4952–4698

51.98–46.69 60.01–56.61

S11 S12 S13 S14 S15

II–III II–III II–III II–III II–III

2.64 2.63 2.62 2.59 2.55

4741 4684 4430 3987 4481

46.50 48.01 43.79 41.76 42.33

53.26 57.75 53.02 47.71 52.07

2.64–2.55 4741–3987

48.01–41.76 57.75–47.71

S16 S17 S18 S19

III III III III

2.58 2.54 2.49 2.46

3868 4000 3642 3747

42.71 37.89 40.62 36.24

48.09 45.11 46.01 43.16

2.58–2.46 4000–3642

42.71–36.24 48.09–43.16

S20

IV

2.11

1938

20.64

24.21

–2.11–

–20.64–

–1938–

–24.21–

RL and RN are the averages from five random single impact rebound values from individual samples. WG Weathering grade; dry dry density; VP ultrasonic P-wave velocity; RL and RN rebound values by L and N type hammers, respectively

Fig. 3. Schematic representation of the laboratory setup for ultrasonic and Schmidt hammer tests. Note that during ultrasonic testing, thin insulating foam paper was placed beneath the rock core

rock surface. Rebound values were normalized in horizontal direction according to Basu and Aydin (2004) for nullifying gravity effect. For each specimen, the normalized rebound values (R) by individual hammer were averaged (Table 2). For each of dry and VP , ranges in adjacent grades overlapped (Table 2) which could partially be attributed to unavoidable experimental errors. Ranges of R in neighboring grades also overlapped where RN produced less overlaps than RL (Table 2).

82

A. Basu et al.

Fig. 4. Rebound nature of Schmidt hammer multiple impacts (at one representative point) with reference to degree of weathering. RL2 –RL1=RN2 –RN1 is the difference between the 2nd and the 1st impact rebound values

For examining the rebound nature of the Schmidt hammers due to multiple impacts at one point in response to weathered states of the investigated rocks, one representative test point on each specimen giving a rebound value 2 of the mean value of individual hammers was chosen. Test point comprising single mineral grain resulting in desirable R was not considered. Four consecutive impacts including the initial impact were performed at the chosen point. Trends of R in each sequence were plotted and were related to weathering grades (Fig. 4). In general, the steepest increase in R (or maximum compaction) was caused by the second impacts whereas minor increase in R resulted from following impacts (Fig. 4). Minor decreases in R at the third and=or the fourth impacts were also noted (Fig. 4) which actually were the results of minor chipping or fragmentation just next to the center of the compaction zone. Grade IV specimens, however, could not sustain the entire impact sequence (Fig. 4). It is notable that although there are some overlaps of the trend lines among neighboring weathering grades in terms of absolute R, the differences of R between the second and the first impacts are quite typical of the weathering grades (Fig. 4). This implies that nature of compaction at that point is representative of the behavior of constituent minerals and microstructures or in other words, weathered state of the material. N hammer performed better than L hammer in discriminating grades which


83

indicates that stress wave induced by higher impact energy influences larger volume in terms of compaction than that induced by lower impact energy. In order to obtain an additional control on the nature of compaction during the impact sequence, VP was measured after each blow. It was anticipated that VP would reflect the compaction pattern. Surprisingly, VP did not show any increase with the progress in impact sequence, rather VP dropped significantly in few test sequences. This observation could be explained by the fact that volume of the compaction zone due to Schmidt hammer multiple impacts is not very significant compared to the volume of the specimen particularly in a polymineralic medium grained rock where the impact energy dissipates quickly along the grain boundaries. The compaction zone, existing only in the close proximity of the specimen surface, does not significantly modify the propagation time of the ultrasonic wave with a small beam spread angle. The occasional falls in VP , on the other hand, are caused by probable cracks (induced by multiple impacts) that propagate deep inside the specimen causing delay in wave propagation. In order to verify such explanation, an additional investigation was carried out on an artificial plaster specimen (dry ¼ 2.08 gm=cm3, close to that of Grade IV material) with the same volume as the rock specimen. Although the plaster specimen had lower density than the Grade IV specimen, VP was higher (2142 m=sec) in the plaster than in the rock specimen which implies more scattering of ultrasonic waves at grain boundaries and microcracks in Grade IV rock material than in the plaster (assumed to be virtually homogeneous material). Only N hammer was used in this part of the investigation. After two consecutive impacts at one point of plaster surface, minor increase in the velocity was observed whereas after two consecutive impacts at one point of the Grade IV specimen, velocity dropped. Figure 5 shows the effect of multiple impacts by N hammer in both plaster and Grade IV rock. In case of a homogeneous material (plaster), propagating hemispherical stress waves induced by multiple impacts cause a volume increase of the compaction zone whereas volume of the compaction zone due to multiple impacts is very limited in a heterogeneous

Fig. 5. Effect of N hammer multiple impacts on plaster and highly weathered granite

84

A. Basu et al.

material (weathered granite) (Fig. 5). Stress concentrations occur around rock grains directly in contact with the plunger tip resulting in irregular and local grain crushing and extensional cracking (Fig. 5).

5. Rock mechanical behavior under uniaxial compression Strength and deformational properties of intact rock under uniaxial compression are basic in geotechnical investigations. Following is a critical evaluation of mechanical behavior of the granites under uniaxial compression in relation to weathering grades. Use of the same specimen for both Schmidt hammer and unaixial compression tests can be misleading as hammering may induce microcracks in the specimen and lower its compressive strength (Aydin and Basu, 2005). Therefore, a separate set of 20 representative specimens (diameter 75 mm; length-diameter ratio 52:1) of different weathering grades was prepared for the uniaxial compression test and the cut faces were ground to achieve the accuracy in the parallelism and smoothness of the specimen ends and their perpendicularity to the axis (ASTM D4543, 2001). Due to limited volumes of Grade IV cores and significant sample loss while machining, no Grade IV specimen could be obtained. All specimens were air-dried to constant mass. A computer-controlled, servo-hydraulic machine (MTS 815 Rock Mechanics Test System#) was used for the compression test (Fig. 6). Axial and circumferential extensometers were attached to the specimen for deformation measurements (Fig. 6).

Fig. 6. Uniaxial compression test setup


85

The overall experimental procedure complied with ASTM D2938 (2001) and ASTM D3148 (2001). For each specimen, the experiment was carried out in two steps. The first step involved loading the specimen (with the attached extensometers) uniaxially up to around 55% of the estimated peak load to measure the deformation. This load level was chosen as at around 50% of the ultimate strength of granite specimens, load induced cracks start to generate (Lau and Chandler, 2004). The second step was to load the specimen (without extensometer) uniaxially until the specimen failed.

5.1 Uniaxial compressive strength Uniaxial compressive strengths (UCS) for the tested specimens are summarized in Table 3. Specimen numbers are according to the descending order of UCS values within each assigned weathering grade (Table 3). A large range of UCS values (214–153 MPa) exists in Grade I category (Table 3). Such wide ranges of UCS values for Grade I granitic rocks are reported by several other researchers elsewhere (e.g. 150–300 MPa by Dearman and Irfan, 1978; 175– 275 MPa by Irfan and Powell, 1985; 141.24 37.33 MPa by Gamon, 1985; 196.45– 116.30 MPa by Basu, 2006). Gamon (1985) attributed the low strengths of granites from earlier studies to probable incorrect test procedures. Although the dimensional and shape tolerances for uniaxial specimens specified in the standards (e.g. ISRM, 1979; ASTM D4543, 2001) are often too demanding to be satisfied, specimens with

Table 3. Mechanical properties under uniaxial compression Sp. No.

WG

UCS (MPa)

E (GPa)

E=UCS

U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U11 U12 U13

I I I I I I I I I I II II II

214 204 200 199 188 186 185 179 164 153 161 147 134

70.17 67.92 65.72 65.62 67.47 64.86 68.79 61.49 66.92 65.00 55.23 61.18 58.82

327.15 332.83 327.85 330.31 358.41 347.98 371.04 343.17 407.23 425.39 343.32 415.77 437.84

0.22 0.22 0.23 0.24 0.20 0.18 0.25 0.21 0.23 0.20 0.27 0.29 0.24

U14 U15 U16 U17

II–III II–III II–III II–III

137 125 109 107

54.51 47.19 50.35 51.71

396.87 376.98 462.48 485.13

0.27 0.30 0.26 0.27

U18 U19 U20

III III III

88 85 73

41.08 52.00 42.70

464.71 613.86 585.49

0.29 0.32 0.28

Range UCS

E

214–153

70.17–61.49

0.18–0.25

161–134

61.18–55.23

0.24–0.29

137–107

54.51–47.19

0.26–0.30

88–73

52.00–41.08

0.28–0.32

WG Weathering Grade; UCS uniaxial compressive strength; E Young’s modulus; Poisson’s ratio

86

A. Basu et al.

minor deviations from the tolerance specifications can be tested authentically in uniaxial compression machines with an adjustable upper platen. Therefore, such a wide range of UCS values cannot result from a sophisticated testing procedure as followed in this study. Referring to Lumb’s (1983) study (UCS: 80–140 MPa for Grade I granites), Gamon (1985) indicated that the uniaxial compressive strengths of the specimens containing shear planes are not representative of the rock material and should be disregarded. However, the fact that specimens (from same Grade I granitic rock) devoid of any shear or other potentially vulnerable macrocracks leading to wide variation in strengths is common in the literature and also is true for the present investigation. Probable causes responsible for large strength variations in the rocks where weathering does not seem to have played any role are discussed below. Influence of weathering on rock strength will be presented thereafter. As the role of microcracks is of great importance in understanding the mechanical behavior of rocks (Akesson et al., 2004), this discussion focuses mainly on microcraking. In Grade I rocks, UCSs are very sensitive to existing microcracks as failure in strongly bonded materials occur by the coalescence of existing and loading-induced microcracks (Aydin and Basu, 2005). In a comprehensive review on rock microcracks, Kranz (1983) emphasized that local stresses are increased by twin lamellae, kink bands and deformation lamellae, stress concentrations at grain boundary contacts and around crystalline cavities. Local strength is reduced along cleavage planes, along grain boundaries and along any internal surface as a result of corrosion by chemically active fluids. Microcracks may result from any loading process (e.g. tectonic, gravitational or thermal loading) (Davis and Reynolds, 1984). Microcracking can also be induced by spatial and temporal changes in temperature due to differential thermal expansion between grains with different thermoelastic moduli and thermal conductivities, for example, differential thermal expansion between adjacent quartz and feldspar grains may be the main cause of microcracks in granite (Kranz, 1983; Davis and Reynolds, 1996). As NW–SE faults are predominant in the proximity of the site (Fig. 1) and the area experiences a subtropical climate, the variability in microcrack density, length, distribution and orientation in Grade I granites may be attributed mainly to tectonic loading and/or differential thermal expansion of adjacent quartz and feldspar grains. Rock failure in relation to external stress states and crack features has been a subject of principal research for several decades. However, failure behavior of different minerals contributing to the overall rock failure is not well-understood. Minerals in a rock behave differently during cracking and the failure mechanisms are dependent on both mineralogy and grain orientation (Li et al., 2003). Li (2001) studied the failure behavior of minerals in unweathered Hong Kong granite under uniaxial compressive condition to various load levels. He found that failure initiation of granite is due to the cracking of quartz grain microscopically. The higher elastic modulus and brittleness of quartz relative to other minerals in granite may be the main reason for quartz cracking first. Load-induced axial cracks also propagate in feldspar grains but at high load level (at about 75% of the peak load). Eberhardt et al. (1999), however, observed that 50% of the observed microcracks (in 3 mm average grain sized granite loaded beyond crack initiation threshold) occurred along grain boundaries and 50% occurred within feldspar grains. In the present


87

investigation, quartz intragranular cracks are more common than cracks within other minerals in Grade I rocks. This gives an indication that quartz intragranular cracks might have played a strong role in initiating failure of Grade I granites. Li (2001) indicated that biotite is the only mineral in granite that undergoes plastic deformation when loaded and local tensile stress field is created at the edge of hard minerals (particularly quartz). When subjected to uniform compressive stress, this local tensile stress field leads to formation of microcracks around biotite at low loading level. Therefore, biotite can speed up the failure process by inducing microcracks. Influence of biotite on microcrack initiation and propagation was also indicated by Nishiyama et al. (2002). The petrographic study of the present investigation revealed that although biotite grains hardly exhibit cracks except a few cleavage cracks, microcracks occasionally occur around biotite grains especially when the neighboring minerals are quartz even in the artificially unstressed Grade I rocks. This observation supports that biotite grains have a great capacity to adjust local stress and minor variation in biotite percentage, distribution and orientation could speed up or slow down the rock failure process. It should be noted that ultimate failure of Grade I granites could be influenced by various factors. For example, minor changes in mineralogy, grain size, shape, orientation and distribution, microcrack density, orientation, length and distribution together may contribute in an extremely complex manner to overall rock failure under a single loading condition (e.g. uniaxial compression). It is hard to predict the distribution of potential zones of stress concentration inside the specimen and how they would result in a definitive failure. Because of such complex fracture mechanics, the overall rock failure process is not yet fully understood. However, from the present petrographic observations and above discussion, it is probably apt to say that variability in occurrences of quartz intragranular cracks and in biotite percentage, distribution and orientation might have played a key role in accelerating or decelerating the failure processes of the investigated Grade I specimens significantly. Table 3 shows a decreasing trend of UCS values as weathering intensifies (with minor overlaps of UCS ranges in adjacent weathering grades) that can be attributed to the evolution of rock microfabric with the onset and advancement of weathering. The sequence of microfabric evolution of granitic rocks with increasing weathering is summarized as initial water ingress along primary microcracks and open mineral cleavages; solution along grain boundaries and within feldspars; increased intensity of microfracturing by opening grain boundaries, expanding biotite and possibly destressing quartz crystals and continued weathering of feldspars resulting in a variety of microfabric features (Baynes and Dearman, 1978; Hencher and Martin, 1982; Ebuk, 1991; Irfan, 1996; Aydin and Duzgoren-Aydin, 2002). Lumb (1983) found that the greatest relative change in strength occurs from Grades II to III granitic rocks of Hong Kong. The study of UK granites by Dearman and Irfan (1978) showed that the minimum loss of strength from Grades I to III granites could be as high as 33%. Gupta and Rao (2000) reported a strength loss of 23% from Grades I to II granites and of 48% from Grades II to III granites from India. Basu (2006) found a strength loss of 59% from Grades I to III Hong Kong granites. In the present study, the strength loss from Grades II to III rocks is around 44%.

88

A. Basu et al.

5.2 Deformational behavior The axial and diametral strains were determined following the axial and circumferential extensometer strain calculation methods. The stress-strain curves for all specimens obtained by processing the raw data in Microsoft Excel# are presented collectively in Fig. 7. Although the stress-strain curves mainly represent the elastic deformational stage for each specimen, the subtended angle between the axial and diametral stressstrain curves becomes larger as weathering intensifies (Fig. 7). For Grade I rocks, the stress-strain curves form a nearly unique deformational behavioral cluster whereas the clusters representing higher weathering grades become increasingly overlapping particularly in adjacent weathering grades (Fig. 7). Average slope of the more-or-less straight line portion of each stress-strain curve was calculated to obtain Young’s modulus (E) and Poisson’s ratio () following ASTM D3148 (2001) for every specimen (Table 3). In case of three specimens (Sp. Nos. U6, U9 and U17), the maximum applied load with the extensometers was well below 50% of the actual peak load. However, it is believed that this did not induce serious error in calculating elastic constants as these specimens already achieved their linear elastic deformational stage partially. Loss of elasticity (E) with increasing weathering intensity is conspicuous from the present study (Table 3) as found by other researchers elsewhere (e.g. Hamrol, 1961; Lumb, 1983; Gupta and Rao, 2000; Basu, 2006).

Fig. 7. Stress-strain curves under uniaxial compression (note: the maximum applied load does not represent the peak load)


89

Fig. 8. Correlation between uniaxial compressive strength and Young’s modulus

Fig. 9. Distribution of data points on E vs. UCS log–log plot with respect to the range of modulus ratio from 200:1 to 500:1

90

A. Basu et al.

E plotted against UCS shows a good linear correlation (Fig. 8) which conforms to the findings of other researchers. Deere and Miller (1966) indicated that for most rocks, the ‘modulus ratio’ (E=UCS) lies in the range from 200 to 500. In the present study, however, this ratio shows a slight deviation from the said range (Table 3 and Fig. 9). The increasing trend of this ratio with the progress of weathering complies with the findings of Kulhawy (1975) for granitic rocks whereas Lumb (1983) reported a decreasing trend for granitic rocks. It is apparent from the Table 3 that increases with increasing weathering intensity and the values range from 0.18 to 0.32 for the investigated rocks (from Grade I to Grade III). For the same segment of the weathering spectrum, Gupta and Rao (2000) reported a smaller range of for Indian granites whereas a larger range was found by Basu (2006) for Hong Kong granites. The common finding of the previous studies on granitic materials and the present investigation is that alteration of minerals to clay particles, rock skeleton disruption and microcrack augmentation cause a reduction in E and an increase in .

Fig. 10. Failure modes: a) axial splitting (in Grade I Sp.), b) single shear plane failure (in Grade II Sp.), c) single shear plane failure (in Grade III Sp.), d) failure along two shear planes (in Grade II–III Sp.), e) axial splitting (in Grade III Sp.), f) single shear plane failure (in Grade I Sp.)


91

5.3 Failure modes The stress-strain curves for brittle rock material under uniaxial compression examined by many researchers could be divided into four phases namely crack closure, linear elasticity, stable crack growth and unstable crack growth. Consequently, the rock fails with fractures developed from the coalescence of several microcracks. As failure modes of rocks could provide useful information in an engineering environment, the failed specimens were examined. The most common failure types were failure by axial splitting (Fig. 10a) and along single shear plane (Fig. 10b and c). Failure along two shear planes was observed only for one specimen (Fig. 10d). Axial splitting took place because of the development of one or more major cracks in the direction of the applied load. A closer inspection of the shear planes revealed coalescence of smaller axial cracks. Occasionally, a combination of the above failure modes seemed to be present in a few failed specimens. When the failure modes were examined with respect to weathering grades, a crude relation emerged. Most Grade I specimens failed in axial splitting mode whereas weathered ones usually underwent shear failure. This observation is in line with the findings by Lumb (1983) and Gupta and Rao (2000). However, in the present investigation, exceptions were also noted (Fig. 10e and f).

6. Conclusions Rocks from Itu Granitic Complex in southeastern Brazil were investigated. A weathering classification for rock materials nearly conformable to the common 6-fold classification scheme was framed. Grade II–III was recognized as an intermediate weathered state based on its characteristic decompositional differences with Grades II and III. The framed classification was substantiated by a detailed petrographic study. Indirect assessments of weathering grades based on dry density, P-wave velocity and Schmidt hammer single impacts were evaluated. Ranges of each of these indices produced significant overlaps in higher neighboring grades. Changes in rebound values due to repetitive hammering at one representative test point proved to be more efficient in predicting weathering grades than conventional averaged single impact rebound values, P-wave velocities and dry densities. The N hammer performed better than L hammer in discriminating weathering grades. Ultrasonic velocity was not sensitive enough to capture compaction patterns due to multiple hammer impacts. Uniaxial compression test results revealed that a large range of uniaxial compressive strength (214–153 MPa) exists in Grade I category where weathering does not seem to have played any role. It is concluded that although a number of variables could be responsible for such huge range of uniaxial compressive strength, variability in occurrences of quartz intragranular cracks and in biotite percentage, distribution and orientation might have played a key role in accelerating or decelerating the failure processes of the investigated Grade I specimens significantly. Deterioration of uniaxial compressive strength and elastic modulus and an increasing trend of Poisson’s ratio with increasing weathering intensity were evident from the present study. Overall, degradation pattern of mechanical properties of the investigated granites is in accord with the findings of the previous studies on granitic materials from other parts of the world. Such depreciation is due to the evolution of rock

92

A. Basu et al.

microfabric (alteration of minerals to soft clay particles, disruption of rock skeleton and microcrack augmentation) with increasing weathering intensity. The modulus ratio depicted an overall increasing trend with advancement of weathering. In terms of relation between failure modes under uniaxial compression and weathering grades, a crude pattern emerged. Most Grade I specimens failed in axial splitting mode whereas majority of weathered specimens underwent shear failure. Acknowledgements The authors gratefully acknowledge Prof. Giovanni Barla and the anonymous reviewer for their valuable comments that provided opportunity to improve the layout and the clarity of the article. guas The authors are thankful to Mr. Paulo V.C.B. Braun of EMAE – Empresa Metropolitana de A e Energia S.A. for kindly providing the cores. The technical support by Mr. Benedito Osvaldo de Souza and Mr. Decio Aparecido Lourenc° o is also gratefully acknowledged. The study was funded by Fundac° ~ao de Amparo a Pesquisa do Estado de S~ao Paulo (FAPESP).

References Akesson U, Hansson J, Stigh J (2004) Characterization of microcracks in the Bohus granite, western Sweden, caused by uniaxial cyclic loading. Engng Geol 72: 131–142 ANON (1995) The description and classification of weathered rocks for engineering purposes: Geological Society Engineering Group Working Party Report. Q J Eng Geol 28: 207–242 ASTM (American Society for Testing and Materials) (2001) Standard test method for unconfined compressive strength of intact rock core specimens, Designation D2938 ASTM (American Society for Testing and Materials) (2001) Standard test method for elastic moduli of intact rock core specimens in uniaxial compression, Designation D3148 ASTM (American Society for Testing and Materials) (2001) Standard practice for preparing rock core specimens and determining dimensional and shape tolerances, Designation D4543 ASTM (American Society for Testing and Materials) (2001) Standard method for determination of the point load strength index of rock, Designation D 5731 Aydin A, Basu A (2005) The Schmidt hammer in rock material characterization. Engng Geol 81: 1–14 Aydin A, Duzgoren-Aydin NS (2002) Indices for scaling and predicting weathering-induced changes in rock properties. Env Eng Geosc VIII: 121–135 Basu A (2006) Mechanical characterization of granitic rocks of Hong Kong by improved index testing procedures with reference to weathering induced microstructural changes. PhD Thesis, The University of Hong Kong, Hong Kong Basu A, Aydin A (2004) A method for normalization of Schmidt hammer rebound values. Int J Rock Mech Min Sci 41: 1211–1214 Baynes FJ, Dearman WR (1978) The relationship between the microfabric and the engineering properties of weathered granite. Bull Int Assoc Engng Geol 18: 191–197 BS (British Standard) (1999) Code of practice for site investigations, No. 5930 Davis GH, Reynolds SJ (1996) Structural geology of rocks and regions, John Wiley, New York Dearman WR, Irfan TY (1978) Assessment of the degree of weathering in granite using petrographic and physical index tests. Proc. Int. Symp. on Deterioration and Protection of Stone Monuments, Paris, pp 1–35 Deere DU, Miller RP (1966) Engineering classification and index properties for intact rock. Tech. Rep. No. AFWL-TR-65-116, Air Force Weapons Lab., New Mexico Eberhardt E, Stimpson B, Stead D (1999) Effects of grain size on the initiation and propagation thresholds of stress-induced brittle fractures. Rock Mech Rock Engng 32(2): 81–99 Ebuk EJ (1991) The influence of fabric on the shear strength characteristics of weathered granites. PhD Thesis, The University of Leeds, United Kingdom


93

Gamon TI (1985) The influence of weathering on the engineering properties of the Hong Kong granite. PhD Thesis, The University of Newcastle upon Tyne, United Kingdom Gupta AS, Rao KS (2000) Weathering effects on the strength and deformational behavior of crystalline rocks under uniaxial compression. Engng Geol 56: 257–274 Hamrol A (1961) A quantitative classification of weathering and weatherability of rocks. Proc. 5th Int. Conf. Soil Mech. Found Engng., Vol. 7, pp 771–774 Hencher SR, Martin RP (1982) The description and classification of weathered rocks in Hong Kong for engineering purposes. Proc. 7th SE Asian Geotech. Conf., Hong Kong, pp 125–142 Hutchison CS (1974) Laboratory handbook of petrographic techniques, John Wiley, New York IPT (Instituto de Pesquisas Tecnologicas) (1981) Mapa Geol ogico do Estado de S~ao Paulo Irfan TY (1996) Mineralogy, fabric properties and classification of weathered granites in Hong Kong. Q J Eng Geol 29: 5–35 Irfan TY, Dearman WR (1978) Engineering classification and index properties of a weathered granite. Bull Assoc Engng Geol 17: 79–90 Irfan TY, Powell GE (1985) Engineering geological investigations for pile foundation on a deeply weathered granitic rock in Hong Kong. Bull Assoc Engng Geol 32: 67–80 ISRM (International Society for Rock Mechanics) (1979) Suggested methods for determining the uniaxial compressive strength and deformability of rock materials. Int J Rock Mech Min Sci Geomech Abstr 16: 135–140 Karpuz C, Pasamehmetoglu AG (1997) Field characterization of weathered Ankara andesites. Eng Geol 46: 1–17 Kranz RL (1983) Microcracks in rocks: a review. Tectonophysics 100: 449–480 Kulhawy FH (1975) Stress deformation properties of rock and rock discontinuities. Eng Geol 9: 327–350 Lau JSO, Chandler NA (2004) Innovative laboratory testing. Int J Rock Mech Min Sci 41: 1427–1445 Li L (2001) Microscopic study and numerical simulation of the failure process of granite. PhD Thesis, The University of Hong Kong, Hong Kong Li L, Lee PKK, Tsui Y, Tham LG, Tang CA (2003) Failure process of granite. Int J Geomech 3: 84–98 Lumb P (1983) Engineering properties of fresh and decomposed igneous rocks from Hong Kong. Eng Geol 19: 81–94 Martin RP (1986) Use of index tests for engineering assessment of weathered rocks. Proc. 5th Int. IAEG Cong., Buenos Aires, Vol. 1, pp 433–450 Martin RP, Hencher SR (1986) Principles for description and classification of weathered rocks for engineering purposes. Site Investigation Practice: Assessing BS5930, Geological Society, London, Eng. Geol. Spec. Pub., Vol. 2, pp 299–308 de Mello VFB (1972) Thoughts on soil engineering applicable to residual soils. Proc. 3rd Southeast Asian Conf. Soil Eng., Hong Kong, pp 5–34 Moye DG (1955) Engineering geology for the Snowy Mountain scheme. J Inst Eng 27: 287–298 Nishiyama T, Chen Y, Kusuda H, Ito T, Kaneko K, Kita H, Sato T (2002) The examination of fracturing process subjected to triaxial compression test in Inada granite. Engng Geol 66: 257–269 Vaz LF (1996) Classificac° ao genetica de solos e horizontes de alterac° ao de rocha em regi~ oes tropicais. Rev Bras Geotec 19: 117–136 Zuquette LV, Pejon OJ, Sinelli O (1994) Engineering geological zoning of S~ao Paulo State (Brazil) – Scale 1:500,000. Proc. 7th Int. IAEG Cong., pp 1187–1195