Ranjith, P.G., Siew Foong, P., Hefny, A.M., Zhao, J.

Strength and mechanics of fractured rocks under triaxial loadings. ISRM 2003 Technology Roadmap for Rock Mechanics, South African Institute of Mining and Metallurgy, 2003.

Strength and mechanics of fractured rocks under triaxial loadings

P.G. RANJITH F.P. SIEW A.M. HEFNY J. ZHAO
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore

Fracture initiation and propagation, being one of the most intensive subjects in rock mechanics; contribute significantly to the deformation of jointed rock masses. An experimental study on fractured granitic rock was carried out to determine the peak strength of specimen under triaxial loading conditions. The commonly observed failure mode for rock specimens in this study was found to be the shearing of the plane of weakness. The minimum peak strength of fractured rock is observed when the joint orientation is approximately 70o to the horizontal axis. Based on experimental test data, an empirical expression was developed to accommodate the effect of joint orientation and joint trace length in estimating the peak strength of fractured rocks. The predicted values of peak strength using the proposed equation well agree with the experimental results carried out on singly fractured specimens under triaxial loading conditions. The threshold stress values of crack initiation and propagation depend on joint geometrical parameters, their degree of interconnectivity, as well as surrounding stresses on the fracture plane.

Introduction
In Singapore, due to the scarcity of land, underground space has been utilized for the development of storage facilities and deep sewerage system in rocks. Thus, a comprehensive understanding in the mechanism and the behaviour of rock structure is required as the ultimate objective is to control rock displacement into and around the underground excavations. In addition, reliable estimates of the shear strength and deformation characteristics of a rock mass are required for analysis of slopes and foundation. This research program presents the effects of stresses on the deformation characteristics of a single rock fracture subjected to triaxial loading states. Furthermore, the study investigates the effects of joint orientation and joint length on the stress-strain behaviour of jointed rocks as well as the threshold stress values of crack initiation and propagation of fractured specimens.

weakness, with respect to the principal stress direction (Donath 1972, McLamore Gray, 1967). Figure 1 shows measured variations in axial stress with the angle of inclination () of the major principal stress to the plane of weakness. Brady and Brown (1994) introduced an instructive analysis of a case in which the rock contained a well-defined, parallel plane of weakness whose normal was inclined at an angle (Figure 2) to the major principal stress direction. Each plane of weakness has a limiting shear strength (s) defined by Coulombs criterion s = c w + n tan w [1] where, w = friction angle of the plane of weakness cw = shear strength parameter of the plane of weakness n= normal stress The stress transformation equations give the normal (n) and shear () stresses on the weakness plane as: 1 1 n = ( 1 + 3 ) + (1- 3 ) cos 2 [2] 2 2 1 = (1 3 ) sin 2 [3] 2 1, 3 =major and minor principal stresses Substituting for n into Equation [1], putting s=, and rearranging, Equation [4] gives the criterion for slip on the plane of weakness. 2(c w + 3 tan w ) [4] (1- 3 )s = ( 1- tan w cot ) sin 2

Effects of joint orientation on the compressive strength of rocks

The overall strength and permeability of rock a mass and the stability of engineering structures are greatly influenced by joint orientations (Ranjith, 2000). Higher the interconnectivity of fractures which in turn lower the shear strength, the greater will be the risk of failure of a rock mass. Therefore, it is fundamentally important to study the influence of joint orientations on stress-strain characteristics of rocks under different loading conditions. The peak strength developed by transversely isotropic rocks in triaxial compression vary with the orientation of the plane of anisotropy, foliation plane or plane of

927

orientation angle as shown in Figure 3. The planar type of anisotropy (Figure 3a, Figure 3b) is the result of cleavage whereas the linear type of anisotropy or the bedding plane type of anisotropy (Figure 3c) is due to the weakness of rock along the bedding planes.

Fracture initiation
According to Eberhardt et al. (1997), the point where majority of fractures began to initiate is defined as the crack initiation threshold. With increasing loads, further cracking is observed to initiate intragranularly within the stronger quartz grains for the case of granitic rocks. This point is identified as the secondary cracking threshold. Brace et al. (1966:3948) and Latjai and Latjai (1974) define the point where the axial stress versus lateral strain curve departs from linearity as the initiation of the microcracking process which is referred to as the crack initiation stress threshold. This point represents the stress at which a significant number of critically orientated cracks initiate and propagate in the direction of major principal stress, (1). Noting the difficulty in using lateral strain data, especially in damaged samples, several researchers including Martin and Chandler (1994) suggested using the calculated crack volumetric strain to identify crack initiation. In this respect, crack initiation can be defined as the stress level at which dilation begins in the crack volume.

Figure 1: Variation of peak strength with the angle of inclination of the major principal stress to the plane of weakness for the confining pressure (Donath 1972, McLamore Gray, 1967).

Figure 2: Variation of peak strength at constant confining pressure with the angle, . The principal stress difference required to produce slip tends to infinity as 90 and as w. Between these values of , slip on the plane of weakness is possible. By differentiation, it is found that the minimum strength occurs when
+ w [5] 4 2 The variation of peak strength with the angle predicted by this theory is illustrated in Figure 2.

Figure 3: Classification of anisotropy (McLamore and Gray, 1967).

Test program
The rock specimens of cylindrical shape with approximately 47.5 mm in diameter and 97 mm in height were used in the testing program. Two types of pink Bukit Timah granitic specimens found in Singapore were used for the experimental study: (a) intact and (b) a single fracture. Six numbers of intact specimens and twenty-seven numbers of fractured specimens with a single fracture and different joint orientations (i.e. =6090 from the horizontal axis) were experimentally tested using a high pressure triaxial testing apparatus. Prior to testing, all visible surface fractures in samples were mapped relative to some co-ordinates axes. The confining pressures applied to the fractured specimens during the testing were 5, 20, 30 and 40MPa, for all the joint orientations. A schematic diagram of the triaxial testing facility used for this study is shown in Figure 4.

In view of the deficiency of the original Hoek-Brown criterion when applied to jointed rock mass, Hoek et al. (1992) modified the criterion to account for the effect of fractures on the strength of rocks. The modified criterion conforms to the strength prediction given by the original criterion, for different stress conditions, and predicts a tensile strength of zero for a rock mass. The modified criterion is expressed in the following form:
m 1 = 3 + c b 3 [6] c where, mb and a are the constants for fractured rock.
a

McLamore and Gray (1967) suggested a genetic classification of anisotropy based on the shape of the anisotropy curve between compressive strength and joint 928 ISRM 2003 TECHNOLOGY ROADMAP FOR ROCK MECHANICS

Table 1: Peak strength of singly fractured granitic rocks: experimental and theoretical values.

Failure criterion

Confining pressure, 3 (MPa)

Theoretical values Trace length, x Joint Failure Peak strength (m) orientation load Pf o 1 (MPa) ( ) (kN) 60 0.080 0.058 0.100 0.096 0.079 0.066 0.115 0.095 0.080 0.075 0.100 0.100 0.097 0.077 0.098 0.100 0.080 0.058 0.100 0.096 0.079 0.066 0.115 0.095 0.080 0.075 0.100 0.100 0.097 0.077 0.098 0.100 299.92 531.45 265.02 469.62 223.93 396.80 130.53 231.30 148.76 146.81 221.14 1827.09 222.96 220.20 325.14 2592.38 272.41 269.12 394.45 3102.38 321.86 318.03 463.76 3612.38

Experimental values Peak strength 1 (MPa) 145.83 116.79 186.49

Failure load Pf (kN) 259.50 206.95 330.46 410.41 422.33 262.86 507.38 556.58 465.20 367.50 542.94 697.10 623.37 578.29 842.09 873.20 259.50 206.95 330.46 410.41 422.33 262.86 507.38 556.58 465.20 367.50 542.94 697.10 623.37 578.29 842.09 873.20

263.61 260.15 391.86 3237.61 395.08 390.20 576.15 4593.71 482.71 476.88 698.96 5497.43 570.33 563.55 821.78 6401.15

70 80 90 60 70 80 90 60 70 80 90 60 70 80 90 60 70 80 90 60 70 80 90 60 70 80 90 60 70 80 90

235.56 238.34 148.34 286.33 311.47 262.53 207.40 306.40 400.10 351.79 327.73 475.22 507.63 145.83 116.79 186.49 235.56 238.34 148.34 286.33 311.47 262.53 207.40 306.40 400.10 351.79 327.73 475.22 507.63

Modified MohrCoulomb Equation [4]

20

30

40

Modified HoekBrown Equation [6]

20

30

40

In Table 1, the modified Mohr-Coulomb criterion is not applicable for = 90 because the function cot is not defined at = 90. Therefore, theoretical values estimated by modified Mohr Coulomb theory at = 90 should not be compared with the experimental values.

929

Table 2: Peak strength of granitic rocks with a single fracture: theoretical (Equation [11]) and experimental values.

Failure criterion

Theoretical values Trace Confining Joint Failure pressure, 3 orientation length, x Peak strength load P f o (m) 1 (MPa) (MPa) ( ) (kN) 60 5 70 80 90 60 20 70 80 90 60 30 70 80 90 60 40 70 80 0.080 0.058 0.100 0.096 0.079 0.066 0.115 0.095 0.080 0.075 0.100 0.100 0.097 0.077 0.098 0.100 120.99 76.35 152.30 139.78 204.99 153.03 335.25
*

214.39 135.29 269.88 247.69 363.24 271.17 594.06 416.37 437.95 373.70 541.84 534.42 652.49 442.04 591.83 603.12

Experimental values Failure Degree of Peak strength load Pf accuracy 1 (MPa) (%) (kN) 83.0 145.83 259.50 116.79 186.49 235.56 238.34 148.34 286.33 311.47 262.53 207.40 306.40 400.10 351.79 327.73 475.22 507.63 206.95 330.46 410.41 422.33 262.86 507.38 556.58 465.20 367.50 542.94 697.10 623.37 578.29 842.09 873.20 65.4 81.7 59.3 86.0 103.2 117.1* 75.4 94.1 101.7 99.8 75.4 104.7 76.1 70.3 67.0

(Equation [11]) developed by authors

234.97 247.15 210.89 305.78 301.59 368.22 249.46 333.99 340.36

90 *overestimated theoretical values.

predicted range. This is because in Equation [4], the function cot in the denominator is converged to zero as the angles approaches 90 and to infinity for the case of =0. The modified Mohr-Coulomb would therefore significantly overestimate the predicted peak strength of fractured rocks under such circumstances (i.e., =90 and 0). The theoretical critical joint orientation for granitic rock with a single fracture at minimum peak strength is calculated as 65 (Line AB-Figure 5) with w of 40.7 (Equation [5]). Experimental results show that the critical joint orientation, which gives minimum peak strength, is found to be approximately 70 (Line CD-Figure 5). There is a small discrepancy of 5 between the experimental and theoretical value, and the reason for this is explained below. The shape of the curves as obtained in Figure 5 is now compared to the genetic classification of anisotropy as suggested by McLamore and Gray (1967:65). It can be observed that the curves produced by the modified MohrCoulomb criterion exhibit the cleavage or planar type of anisotropy, which is further categorized as the U-type (Figure 3a). However, for the curves obtained experimentally, they are generally categorized under the bedding plane type, which is of the shoulder-type (Figure 3c). A smooth shoulder formation on either end of the Ushape suggests a sliding mode failure of the specimen in the U-shape region and splitting in the shoulder region. Thus, the occurrence of the small discrepancy in critical joint orientation between the experimental and theoretical

Figure 4: A schematic diagram of triaxial testing facility.

Test results and discussions

Effects of joint orientations on the compressive strength of rock According to Equation [4], modified Mohr-Coulomb theory incorporates the effect of joint orientations but not the joint trace length. Nonetheless, neither one of the parameters are incorporated into Equation [6], which depicted the modified Hoek-Brown criterion. Theoretical peak strength values calculated using modified MohrCoulomb criterion and modified Hoek-Brown criterion are compared with the experimental values obtained from triaxial testing on fractured granitic rocks, as given in Table 1. The theoretical values of peak strength and failure load as predicted by the modified Mohr-Coulomb are relatively consistent for joint orientations of 60 and 70. However, the experimental values for 90 are well out the 930

ISRM 2003 TECHNOLOGY ROADMAP FOR ROCK MECHANICS

result is due to the variation of the planar type that influences the critical joint orientation.

incorporated into Equation [6] as a function of sin. The resulting equation is:

1 = 3 +

71.48( 3 )0.35 sin

[7]

A back-analysis is used to obtain a relationship between the variable k in Equation [8] and the trace length of fracture, x. 71.48k ( 3 ) 0.35 [8] 1 = 3 + sin For a given values of 3= 5MPa, 1= 145.83MPa and = 60, k is estimated as 0.97. The variable (k) has been plotted against the trace length of fractures in order to develop a relationship between these two parameters (Figure 6). Figure 5: Effects of joint orientations on peak strength of rocks at different confining pressures. The modified Hoek-Brown criterion predicted a set of peak strengths for the given confining pressures regardless of joint orientations for any given granitic rock with a single fracture (Table 1). Experimental values of peak strength and failure load for joint orientation of 60 is relatively consistent with theoretical values obtained from the modified Hoek-Brown criterion but not with 70, 80 and 90. Moreover, the experimental values for the joint orientation of 90 is observed to be consistent with the theoretical values for intact granitic rocks obtained using Hoek-Brown criterion. This is only applicable when the joint oriented parallel to the line of the axial load as well as the joint trace length distributes along the specimen from top to the bottom of the specimen. Thus, it can be concluded that singly fractured granitic rocks having a joint orientation of 90 shows similar strength characteristics as observed in intact rocks under same boundary conditions. This is because with the fracture starting from the top to the bottom of the specimen in a straight line, it is difficult to mobilize opening or closing of the fracture or shearing the fracture along its plane of weakness. The inconsistency of the theoretical and experimental peak strengths for joint orientation of 70 is probably due to the fact that 70 has been the experimental critical joint orientation, which gives minimum peak strength as discussed in Figure 5. Thus, the rock specimens with critical joint orientation would experience lower peak strength as compared to the theoretical values because it has a weak plane due to the critical joint orientation. Therefore, it can be concluded that the peak strength at the joint orientation of 70 for any given confining pressure is the lowest (Figure 5). Neither the original Hoek-Brown criterion nor the modified Hoek-Brown criterion incorporates both joint orientations and the joint trace length of rock fractures into the relevant equations. The parameters in Equation [6] for granitic rock with a single fracture are determined as, a = 0.35, c = 173.3MPa and mb = 13.8. With reference to Figure 5, it is observed that the effect of joint orientation in relation to the peak strength takes a form of sinusoidal curve. Thus, the joint orientation of fracture can be

Figure 6: Variable k versus joint trace length. The exponential function is chosen to represent the variable (k) and trace length (x) due to the fact that the function is able to produce the highest correlation with respect to the relationship between k and x, as compared to other functions such as polynomial (Figure 6). Taking an average of the four exponential functions obtained with an optimum factor of safety of 1.3, the following relationship between k and x is developed as follows: k = 0.1839e18.377 x [9] Equation [9] is modified to account for the effect of different joint orientations and trace length of fractures as given below:

1 = 3 +

71.48( 3 )0.35 (0.1839e18.377x) sin

[10]

It can be seen from Table 2 that the modified Equation [10] which includes the effect of joint orientations and joint trace length, is able to estimate peak strength at various joint orientations. In contrast, the modified HoekBrown criterion is independent of x and (Equation [6]). Using Equation [10], theoretical peak strength can now be predicted and be compared with the experimental peak strengths. The results are as depicted in Table 3. The degree of accuracy in predicting the peak strength of granitic rock with a single fracture using Equation [10], 931

can reach an accuracy of as high as approximately 100.0%, with the lowest precision of 59.3%. However, it must be emphasized that Equation [10] is highly dependant on the parameter of trace length provided in any test conducted. The trace length measured in the tests shall be of fractures that distribute along the diameter of the specimen. As observed from Table 2, the accuracy in predicting strength of rocks with joint orientation of 90 is relatively low which is within the range of 59.3% to 75.4%. As discussed in the previous section, a granitic rock with joint orientation of 90 is inclined to the strength characteristics of an intact rock. The complete equation for the modified Hoek-Brown criterion with inclusion of the effect of joint orientation and joint trace length is given below:

Figure 7: Stress-strain curves for a granitic rock specimen with a single fracture (=90) at confining pressure of 5MPa.

1 = 3 +

c (mb 3 / c )0.35
sin

(0.1839e18.377 x )

[11]

Crack closure The pre-existing cracks and voids inside a rock mass close up upon the initiation of stresses. In triaxial compression test, the pre-existing cracks are forced to close in both vertical and horizontal axes. The stress threshold for the crack closure can be observed from the stress-strain curve whereby the curve depicts a slight non-linearity at the beginning of the curve. This phenomenon can be observed from Figure 7 as shown below. For a given specimen subjected to a confining pressure of 5MPa, crack closure is observed at deviator stress of 12MPa. Also, the on-set of crack closure can be determine with a higher degree of accuracy from plot of crack volumetric strain versus axial strain as suggested by Martin and Chandler (1994:644) as shown in Figure 8. The region of crack closure for the specimen shown in Figure 8 is in the range of 0 to 0.05% axial strain. During the formation of crack closure, no energy is emitted because the energy is consumed for the closing up of preexisting cracks. The crack closure stops at axial stress 15MPa and at 0.05% axial strain as determined by Figure 7. Furthermore, the closure of microcracks is represented by the decrease in crack volumetric strain at the beginning of the curve (Figure 8).

Figure 8: Volumetric strain curve for a granitic rock specimen with a single fracture (=90) at confining pressure of 5MPa.

Crack initiation Crack initiation starts when the deformation of rock becomes irreversible in which depicts the beginning of the mechanism of stable crack growth. Noting the difficulty in identifying crack initiation with the point of departure from the stress-strain curve, Martin and Chandler (1994:644) suggested in using the calculated crack volumetric strain curve to identify crack initiation. However, this method is highly dependant on the Youngs modulus and Poisson ratio taken within the region of elastic deformation. For a specimen shown in Figure 7 and Figure 8, the Youngs modulus (E) and Poissons ratio (v) are 50.5GPa and 0.2, respectively. In this respect, crack initiation can be defined as the stress level at which dilation begins in the crack volumetric plot. From Figure 7 and Figure 8, it can be seen that crack initiation started at 0.29% axial strain which corresponds to a crack initiation axial stress of 160MPa from the stress-strain curve. Crack damage The threshold stress of crack damage can be determined from the volumetric strain curve at the point of reversal of the curve. For the specimen used in Figure 8, crack damage threshold stress is determined at 0.36% axial strain which corresponds to axial stress of 185MPa. It can be deduced that the volumetric strain reversal occurred when the relative increase in lateral strain rate surpassed the axial strain rate and emerged as the dominant component in the volumetric calculation. The point of reversal is highly distinguishable in the plot of crack volumetric strain versus axial strain curve as in Figure 8.

932

ISRM 2003 TECHNOLOGY ROADMAP FOR ROCK MECHANICS

Conclusions
A series of laboratory tests on fractured rocks was carried out to determine the effects of joint orientation and joint trace length on the peak strength of rocks as well as to estimate the threshold stress values for the crack closure, crack initiation and crack damage. Findings of the study show that critical joint orientation which yields the minimum peak strength, for Bukit Timah granitic rocks with a single fracture is found to be 70. Granitic rock specimens with joint orientations of 80 and 90 exhibit similar strength characteristics as in intact rock thus their peak strength deviates significantly from the theoretical values predicted by modified Mohr-Coulomb and modified Hoek-Brown. The joint orientation and trace length are incorporated into the modified Hoek-Brown criterion as given below:
(0.1839e18.377 x ) sin The predicted values of peak strength using the proposed equation well agree with the experimental test results carried out on singly fractured specimens under triaxial loading conditions.

References
Brace W.F., Paulding, B.W., Jr, and Scholz, C (1966): Dilatancy in the fracture of crystalline rocks. Journal of Geophysical Research, Vol. 71 (16), pp. 3939-3953. Brady B.H.T. and Brown, E.T. (1985): Rock Mechanics for Underground Mining. 1st ed, Chapman & Hall, London. Donath F.A. (1972): Effects of cohesion and granularity on deformation behaivour of anisotropic rock, in studies in Mineralogy and Precambrian Geology (eds. B. R. Doe and D. K.Smith), Geol. Soc. Am. Memoir, 135, pp. 95-128 Eberhardt E., Stead, D., Stimpson, B. and Read, R.S. (1998). Identifying crack initiation and propagation thresholds in brittle rock. Can. Geotech. J., Vol. 35, pp. 222-233. Hoek E., Wood, D. and Shah, S. (1992): A Modified Hoek-Brown Failure Criterion for Jointed Rock Masses. ISRM Symposium, Eurock 1992, British Geotechnical Society, London, No. 37, pp. 209-214. Lajtai E.Z. and Lajtai, V.N. (1974): The evolution of brittle facture in rocks. Journal of the Geological Society, London, Vol. 130(1), pp. 1-18. Martin C.D. and Chandler, N. A. (1994): The progressive fracture of Lac du Bonnet Granite. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr, Vol. 31(6), pp. 643-659. McLamore R. and Gray, K.E. (1967): The mechanical behaviour of anisotropic sedimentary rocks. J. Engng for Industry, Trans. A. Soc. Mech.Engrs Ser. B, Vol. 89, pp. 62-73. Ranjith P.G. (2000): Analytical and experimental modeling of coupled water and air flow through rock joints. Ph.D Thesis, University of Wollongong, Australia.

1 = 3 +

c (mb 3 / c )0.35

Acknowledgements
The authors would like to thank the Geotechnical Laboratory technical staffs, Nanyang Technological University for their assistance in the laboratory testing. Sincere gratitude goes out to a number of colleagues for their invaluable contributions to this paper.

933

934

ISRM 2003 TECHNOLOGY ROADMAP FOR ROCK MECHANICS