The effect of discontinuities on strength of rock samples

Transcription

1 The effect of discontinuities on strength of rock samples T Szwedzicki 1 and W Shamu 2 ABSTRACT The mechanical properties of rock samples of identical lithological composition may vary significantly due to existence of material discontinuities. Laboratory tests were carried out to determine the effect of material discontinuities on rock sample mechanical properties. The investigations consisted of detecting discontinuities, observing the failure mechanisms during uniaxial compressive strength tests, and analysing of the factors affecting the strength. A non-destructive testing method has been adopted to assist in the detection of discontinuities. Laboratory testing proved that fractures leading to failures initiated on the detected discontinuities. The location, orientation, size, density and extent of the discontinuities contribute to different modes of failure and these affect the mechanical parameters of rock samples. Four modes of failure were distinguished for hard and brittle cylindrical rock samples. These modes were referred to as simple shear failure, multiple shear failure, multiple fracturing and vertical splitting. To evaluate the effect of discontinuities on mechanical parameters, the ratio of uniaxial compressive to tensile (Brazilian) strength was calculated and analysed for the different modes of failure. It was found that the value of the ratio depended on the mode of sample failure. The lowest value of the ratio was observed for shear mode of failure and the highest for vertical splitting mode of failure. INTRODUCTION Mechanical properties of rock mass and rock samples are affected by material discontinuities such as planes of weakness, mineralogical variations, bedding planes, cracks, flaws, joints, etc. On a macroscopic scale the discontinuities are characterised by distinctive joints and the tensile strength of rock mass is often considered to be zero. At the other end of the domain scale, there are microscopic imperfections like micro-defects, intergranular cracks or micro-flaws that can not be observed without the aid of a microscope. This paper describes a method of detection of discontinuities and discusses the effect of these discontinuities on the mechanical properties of rock at a mesoscopic domain scale, ie the scale of rock samples used in a laboratory. At such a scale, because of the discontinuities, the values of mechanical properties may be different from the mechanical properties of the intact rock. Although the mesoscopic discontinuities are sometimes visible, many of them remain undetected during laboratory sample preparation and testing. Mesoscopic rock samples that include discontinuities can be considered as a form of transition between the rock mass and the intact rock material. Current laboratory testing practices do not take into account the effect of mesoscopic discontinuities on the mechanical properties of the rock samples. When samples of identical lithological composition are tested, the existence of mesoscopic discontinuities results in variation in the values of mechanical properties, especially on the uniaxial compressive strength. Large variations in values of uniaxial compressive strength obtained by conventional means of determining rock strength are not uncommon. The measures used to improve the accuracy of results include: testing a statistically significant number of samples, improving precision of sample preparation, and testing a large number of samples and disregarding values that are anomalously high or low. A large number of laboratory tests produce results that do not necessarily reflect the values of mechanical properties of intact 1. Member AusIMM, Department of Mines and Energy, Darwin, Northern Territory. 2. Henry Walker Contracting Pty Ltd, Western Australia. rock material. In present practice for rock engineering design, the results of testing are arbitrarily scaled down to account for scale effects (eg Bandis, 1990). The fact that discontinuities affect mechanical properties of rock is widely acknowledged (eg Hoek and Brown, 1980; Jumikis, 1983), and various types of rock material discontinuities have been classified (eg Vutukuri, Lama and Saluja, 1974). The existence of microcracks and their effect on sample failure and on modes of failure were investigated by numerous authors (eg Horii and Nemat-Nasser, 1985; Peng and Johnston, 1970). Although various modes of failure for cylindrical samples have been observed and classified by numerous investigators (eg Reinhart, 1966; Paul and Gangal, 1966), these are not considered when interpreting the results of uniaxial compressive strength tests. However, the relation between the mesoscopic discontinuities, the mode of failure and the strength parameters has not been investigated. The objective of this investigation is to establish accurate and more reliable methods of testing of rock sample mechanical parameters by: identifying the material discontinuities that are present in rock samples which cannot be detected by current inspection techniques; analysing the effect of mesoscopic discontinuities on nucleation and propagation of failure planes through the sample; and classifying the observed modes of failure and then finding the relation between these modes, rock material discontinuities within samples and the mechanical parameters of rock samples. Understanding the role of mesoscopic discontinuities in the fracturing process will enable better interpretation of the results of strength tests. NON-DESTRUCTIVE TESTING METHOD Mesoscopic rock discontinuities are usually not visible to the naked eye. To make the discontinuities visible, a testing technique was adapted from non-destructive testing engineering. The method, similar to that used by Gardener and Pincas (1968), involves the use of a dye that, under ultraviolet light, contrasts and shows up discontinuities on the rock surfaces. The non-destructive testing technique as used for rock testing requires the following: Cleaning the sample surface: after the cylindrical rock specimen is prepared for uniaxial compressive strength testing according to the ISRM Suggested Method, (1978), the specimen surface is cleaned to remove surface grit, dust or mud that may hinder the penetration process. Application of penetrant dye: the sample is sprayed with a penetrant dye that fluoresces under ultraviolet light. The dye penetrates into surface rock material discontinuities (Figure 1a). Removal of excess penetrant dye: all the penetrant on the sample surface is removed using a chemical solvent (Figure 1b). Development: a developer is then applied to bring the dye penetrant from the rock discontinuities to the surface of the specimen and enhance the fluorescent properties of the dye to enable inspection. Inspection: the final step involves the visual inspection of the sample under ultraviolet light and then interpreting the results (Figure 1c). The AusIMM Proceedings No

2 T SZWEDZICKI and W SHAMU A B C FIG 1 - Non-destructive testing for material defects (after Betz, 1963) - a) application of penetrant dye; b) removal of excess penetrant dye; c) development and inspection. The non-destructive testing technique can be used to inspect samples prior to and after mechanical testing. The use of the fluorescent penetrant dye to detect mesoscopic rock material discontinuities proved successful and makes it possible to clearly define discontinuities that would normally have gone undetected. The technique allows for visualisation of closed cracks, bedding planes, and bonded planes of weakness which have surface traces. Obviously, discontinuities not open to the surface, where the dye could not infiltrate, can go undetected. Observations under ultraviolet light of rock fragments after mechanical testing, allows determination of whether samples failed along mesoscopic discontinuities or along new, stress induced failure planes. Discontinuities highlighted by the penetrant dye were observed to contribute significantly to the mode of the sample failure. MODES OF ROCK SAMPLE FAILURE To assess the effect of mesoscopic discontinuities on mechanical properties of rock samples, uniaxial compressive (UCS) tests and Brazilian tensile strength (BTS) tests were carried out in accordance with the Suggested Methods, International Society for Rock Mechanics, (ISRM Suggested Methods, 1978, 1979). Varieties of rock types from mines in the Eastern Goldfields, Western Australia, were tested. Rock types consisted of quartzites, ultramafics, amphibolites, serpentinites, basalts and massive sulphides. The laboratory testing was conducted on core samples of diameter ranging from 32 to 64 mm on an INSTRON closed-loop, servo-controlled, electro-hydraulic testing machine. After samples were mechanically tested to failure all rock fragments were analysed and the mode of failure was described. For hard and brittle cylindrical rocks samples, four distinct modes of failure were identified: simple shear failure, multiple shear failure, multiple fracturing and vertical splitting, (Figure 2). Simple shear is described as a failure along one or more planes (parallel to each other) situated at an oblique angle to the direction of maximum compression. When fracturing takes place along two or more planes situated obliquely to the direction of compression, but not being parallel to each other, the mode is called multiple shear. Multiple fracturing is said to take place when rock samples disintegrate, often by dynamic processes, along many planes in random directions. The vertical splitting mode denotes a sample failure along planes parallel to the direction of compression. In this mode the planes of failure propagate the length of the sample. The effect of mesoscopic discontinuities on the mode of failure and propagation of the failure planes was assessed by inspection of the rock samples and an analysis of the treated samples under ultraviolet light before and after destructive testing. From a study of the discontinuities revealed prior to destructive testing, and the rock fragments produced after failure, three types of fracture propagation were distinguished: FIG 2 - Modes of sample failure. 2 No The AusIMM Proceedings

3 THE EFFECT OF DISCONTINUITIES ON STRENGTH OF ROCK SAMPLES FIG 3 - Typical mechanisms of sample failures - a) fracture initiation from an imperfection; b) fracturing through intact rock material from an imperfection; c) fracturing along a detected plane of weakness. fracture initiation at imperfection and a subsequent propagation through intact rock (Figure 3a); fracture initiation on a number of discontinuities and propagation throughout intact rock material (Figure 3b); and fracture nucleation and propagation along detected planes of weakness (Figure 3c). It has been observed that location of mesoscopic discontinuities, their orientation, size and density contributed to the mode of failure. The location of imperfections plays a crucial role in the way the fractures initiate and propagate. In cases when the discontinuities were located at the end sections of a sample, failure took place in a single or a multiple shear mode. Figure 4 shows a dolerite sample that failed in shear mode along planes of weakness situated near the end of the sample. The discontinuities located in the centre of the cylinder tend towards nucleated multiple fracturing. The orientation of mesoscopic discontinuities greatly affected the manner in which the specimen failed. When discontinuities were oriented obliquely to the load direction, the sample failed in shear mode. Where rock material discontinuities were oriented parallel or subparallel to the load axis, failure tended to occur in a vertical splitting mode, as shown on a quartzite sample in Figure 5. Discontinuities oriented perpendicular to the load axis appeared to have less effect on the mode of failure. The other factor that had an effect on the mode of failure was the relative size and extent of discontinuities. It was observed that if the detected discontinuities were longer than per cent of the sample length the fractures tended to propagate along them forming planes of failure. Where discontinuities were shorter, the failure often initiated at these discontinuities but propagated in a random direction, Figure 6. The deeper the discontinuities, as shown by the amount of penetrant dye drawn out by the developer, the more likely they were to influence the orientation of the failure plane. FIG 4 - Failure along obliquely oriented planes of weakness - inspected under normal light before testing (top); inspected under ultraviolet light before testing (middle); inspected under ultraviolet light after testing (bottom). The AusIMM Proceedings No

4 T SZWEDZICKI and W SHAMU FIG 5 - Vertical splitting of a sample along a detected plane of weakness - inspected under normal light before testing (top); inspected under ultraviolet light before testing (middle); inspected under ultraviolet light after testing (bottom). Failure often initiated in areas of large numbers of mesoscopic discontinuities, especially at the intersections of different discontinuities. Figure 7 shows an example of fracture initiation on an intersection of detected planes of weakness in a dolerite sample.. THE EFFECT OF DISCONTINUITIES ON SAMPLE STRENGTH The uniaxial compressive strength of rock samples is a function of the mechanical properties of the intact rock and of the mechanical properties of mesoscopic discontinuities and hence is related to the mode of failure. To quantify the effect of the discontinuities on the failure mode of samples of different lithologies a dimensionless parameter was introduced. This parameter was defined as the compressive strength normalised by the mechanical property that is least affected by the mesoscopic discontinuities. Studies by Szwedzicki and Donald (1996) indicated that for heterogeneous and strong to very strong rock from Western Australia, the coefficient of variation for the uniaxial compressive strength was from 42 to 50 per cent and for the same number of tensile strength tests the coefficient of variation was from per cent. For this reason, for the purpose of this study, the uniaxial compressive strength was normalised by the tensile strength of the rock determined by the Brazilian method. During the Brazilian tensile strength test, failure is expected to propagate across a diametrical plane parallel to loading and the probability that material discontinuities are aligned exactly with the plane of failure is small. In a few cases in which samples failed off the preferred planes, the results were discarded and FIG 6 - Multiple fracturing along detected planes of weakness - inspected under normal light before testing (top); inspected under ultraviolet light before testing (middle); inspected under ultraviolet light after testing (bottom). tests were repeated on new samples. Under such conditions the tensile (Brazilian) strength of the sample can be regarded as representative of the tensile (Brazilian) strength of the intact rock, ie the strength value was not affected by the existence of mesoscopic discontinuities. For practical purposes, it is often assumed that the ratio between compressive and tensile strength for rock samples is 10:1 (Stacey and Page, 1986). However, Griffith crack theory predicts that the ratio of uniaxial compressive strength at crack extension to the uniaxial tensile strength will always be 8:1. The study revealed that the ratio depends on the mode of failure and can vary from 1 to more than 30. Low values of the ratio of compressive to tensile strength indicate that the failure of the sample in compression was influenced by discontinuities and the compressive strength predominantly represented the properties of the mesoscopic discontinuities. High values of the ratio indicate that under compression the sample fails across the intact rock and that the compressive strength represents the properties of the intact material. Of the total number of 42 samples tested, 28.6 per cent failed in simple shear failure, 35.7 per cent failed in multiple shear failure, 16.6 per cent failed in multiple fracturing and 19.1 per cent failed in vertical splitting mode. Figure 8 shows the relation between the mode of failure and the dimensionless strength ratio. Simple and multiple shear failures occurred by shearing or sliding. This type of failure occurred along extensive discontinuities that were oriented at an oblique angle to the loading direction. Where distinct mesoscopic discontinuities, such as bedding planes, or bonded joints, were detected, it was 4 No The AusIMM Proceedings

5 THE EFFECT OF DISCONTINUITIES ON STRENGTH OF ROCK SAMPLES noted that a shear failure propagated along these features. The compressive strength values for single shear mode tended to be the lowest for each rock type. The compressive to tensile (Brazilian) strength ratios ranged from 1:1 to 16:1, with an average of 6:1. It is interesting that some samples tested in compression failed in shear along planes of discontinuities at a very low load despite high values of tensile (Brazilian) strength. In multiple shearing the compressive to tensile strength ratios were generally higher than those for simple shear failure and ranged from 5:1 to 20:1, with an average of 10:1. The low values of the ratio, less than 10:1, indicated that the compressive strength obtained when samples fail in single or multiple mode may not be representative of the intact rock strength. The strength values in this mode can be used to estimate the strength along mesoscopic discontinuities within the samples. In hard and brittle samples where discontinuities were randomly oriented, the predominant mode of failure was multiple fracturing. Under peak load the samples disintegrated dynamically into small fragments. The compressive to tensile strength ratio ranged from 7:1 to 21:1, with an average of 15:1. The vertical splitting failure mode was noticed in samples where mesoscopic discontinuities were not detected or where they were detected but were located parallel or subparallel to the load axis. The developed fracture planes were aligned with the direction of maximum compression (Fairhurst and Cook, 1966). The compressive to tensile strength ratios varied from 11:1 to 31:1, with an average of 20:1, and the strength value could be taken as representative of intact rock. Non-destructive testing of rock samples prior to destruction can be used to select samples that are free from mesoscopic discontinuities. The existence of mesoscopic discontinuities is responsible for scale effects, ie strength reduction with increase of specimen size. The larger the specimen the higher the probability that the mesoscopic discontinuities will effect the strength. This technique could be used for assessing rock damage, at a mesoscopic scale, due to blasting and other mining activities. FIG 7 - An example of fracture initiation on an intersection of detected planes of weakness - inspected under normal light before testing (top); inspected under ultraviolet light before testing (middle); inspected under ultraviolet light after testing (bottom). SUMMARY AND CONCLUSIONS Rock strength parameters often show wide variations within the sample population. Such variations make it difficult to select strength parameters for engineering design. The variations are FIG 8 - Predominate modes of failure vs compressive/tensile (Brazilian) strength ratio The AusIMM Proceedings No

6 T SZWEDZICKI and W SHAMU often attributed to the existence of mesoscopic discontinuities such as planes of weakness or bonded cracks. To assess the effect of these discontinuities on strength parameters, a method of detection was devised and subsequently samples were tested for mechanical properties and their modes of failure recorded. To detect the discontinuities, a non-destructive testing method was adopted for testing of core rock samples. Rock samples were chemically treated and inspected for discontinuities before destructive testing. A dye that was visible under ultraviolet light was applied to penetrate the rock and mesoscopic discontinuities (planes of weakness, or bonded fractures). For hard and brittle rocks, four modes of failure were identified: simple shear failure, multiple shear failure, multiple fracturing and vertical splitting. The testing showed that not all modes of failure produce strength values that can be regarded as the peak strength values for the intact rock material; some values represent the strength of mesoscopic discontinuities. The effect of the discontinuities was observed to be dependent on location, size, orientation and density of discontinuities. It was found that the detected discontinuities influence the mode of failure. Shear failure was most common where discontinuities were oriented at an acute angle to the load direction and when discontinuities were revealed at the top or bottom part of the sample. When the discontinuities were located at the midheight part of the rock cylinder, the multiple fracturing mode was predominant. Vertical splitting took place where the discontinuities were not detected or were parallel to the load axis. The average value of the ratio of uniaxial compressive to tensile strength (determined by the Brazilian test) was lowest for single shear failure (6:1), slightly higher for multiple shear (10:1), higher for multiple fracturing (15:1) and the highest for vertical splitting (20:1). The value of the compressive strength obtained in shear mode represented properties along discontinuities, whereas if obtained in vertical splitting mode represented strength of the intact rock material. The compressive to tensile (Brazilian) strength ratio and the mode of failure gave an indication to the origin of the fracture propagation and must be analysed before the results can be used with a high degree of confidence in geotechnical analysis. The non-destructive method of sample inspection, when used in conjunction with an analysis of the mode of failure and the rock strength can assist in explaining the wide scatter of results obtained from the same sample population. Therefore it is recommended that this methodology be used to ensure that representative testing results of rock mechanical parameters are taken into account for rock engineering design. ACKNOWLEDGEMENT This research was carried out at the Department of Mining Engineering and Mine Surveying, Western Australian School of Mines, Kalgoorlie, Western Australia. The authors wish to thank MrDHWalker, Geomechanics Laboratory Manager, Western Australian School of Mines, for assistance in conducting the laboratory testing. REFERENCES Bandis, S C, Scale effects in the strength and deformability of rocks and joints, in Proceedings Conference on Scale Effects in Rock Masses, pp 59-76, (Balkema). Betz, C E, Principles of Penetrants, (Magnaflux Corporation: Chicago). Fairhurst, C and Cook, NGW, The phenomenon of rock splitting parallel to the direction of maximum compression in the neighbourhood of a surface, in Proceedings 1st Congress ISRM, Vol 1, pp (ISRM: Lisbon). Gardener, R D and Pincus, H J, Fluorescent Dye Penetrants Applied to Rock Fractures, International Journal Rock Mechanics and Mineral Science, 5: Hoek, E and Brown, E T, Underground Excavations in Rock, (Institution of Mining and Metallurgy: London). Horii, H and Nemat-Naser, S, Compression-induced microcracks growth in brittle solids: axial splitting and shear failure, Journal of Geophysical Research, 90(B4): ISRM Suggested Method, Suggested method for determining compressive strength of rock material, International Journal of Rock Mechanics. A Min Sci Geomech, Abst, 15(3):1978. ISRM Suggested Method, Suggested Method for Determining the Uniaxial Compressive Strength and Deformability of Rock Material, International Journal Rock Mechanics, A Min Sci Geomech Abst, 12(2):1979. Jumikis, A R, Rock Mechanics, (Trans Tech Publications:). Paul, B and Gangal, M, Initial and subsequent fracture curves for biaxial compression of brittle materials, in Proceedings 8th US Symposium on Rock Mechanics, pp Peng, S and Johnston,A M, Crack propagation and faulting in cylindrical specimens of Chelmsford Granite, International Journal of Rock Mechanics Mineral Science, 9: Reinhart, J S, Fracture of rocks, Internatioal Journal of Fracture Mechanics, Vol 2. Stacey, T R and Page, C H, Practical Handbook for Underground Rock Mechanics (Trans Tech Publications). Szwedzicki, T and Donald, D Assessment of mechanical parameters of rock using an indentation test, in Proceedings Annual Conference (The Australasian Institute of Mining and Metallurgy: Melbourne). Vutukuri, V S, Lama, R D and Saluja, S S, Handbook on Mechanical Properties of Rock, Vol 1 (Trans Tech Publications:). 6 No The AusIMM Proceedings