PII:S0886-7798(98)00023-6 Effect of Intermediate Principal Stress on Strength of Anisotropic Rock Mass Bhawani Singh, R. K. Goel, V. K. Mehrotra, S. K. Garg, and M. R. Allu Abstract -- The Mohr-Coulomb criterion needs to be modified for highly anisotropic rock material and jointed rock masses. Taking % into account, a new strength criterion is suggested because both a2 and a3 would contribute to the ¢ormal stress on the existing plane of weakness. This criterion explains the enhancement of strength (%- a3) in the underground openings because a~along the tunnel axis is not relaxed significantly. Another caLtse of strength enhancement is less reduction t~n the mass modulus in tunnels due to constrained dilatancy. Empirical correlations obtained from data from block shear tests and uniaxial jacking tests have been suggested to estimate new strength parameters. A correlation for the tensile strength of the rock mass is presented. Finally, Hock and Brown theory is extended to account for az A common strength criterion for both supported underground openings and rock slopes is suggested. © 1998ElsevierScienceLtd Introduction C omputer modelling of rock structures has reached an advanced stage of development, in which analytical studies are used in the practical design of support systems for tunnels, caverns, slopes and dam foundations, etc. In mining engineering as well, computer modelling has gained wide popularity. Experience suggests that the reli- ability of computer modelling is increased if realistic consti- tutive equations and parameters are used in the analysis. Experience has led to the following strategy of refinement in the design of support systems. 1. In feasibility studies, empirical correlations may be used for estimating rock parameters. 2. For major projects, in-situ tests should be conducted at the design stage to determine the actual rock parameters. It is suggested that exLensive in-situ triaxial testing (with ~, ~2 and ~3 applied on sides of the cube of rock mass) should be conducted because cr 2is found to affect both strength and modulus of deformation of the rock mass. This is the moti- vation for the research, which points to the urgent need for in-situ triaxial tests. (Significant developments have al- ready taken place on polyaxial tests on concrete, and have been reported in the Lecture Notes of a Workshop on Behaviour of Concrete Under Multiaxial States of Stress held at New Delhi, India, in 1987. For example, sophisti- cated equipment may be adopted for in-situ triaxial tests on rock masses; and the fziction between flat jacks and sides of the rock mass may be reduced significantly by inserting sets of three thin Teflon-coated steel plates in between them). Present addresses: Bha.wani Singh, Professor, Dept. of Civil Engineering, University ofRoorkee, Roorkee - 247 667, India; R. K. Goel, Scientist, CMRI l~gional Centre, CBRI, Roorkee - 247 667, India; V. K. Mehrotra, Supdt. Engineer, Upper Ganga Canal Modernization, U.P. Irrigation, Roorkee - 247 667, India; S. K. Garg, SRF, Dept. of Civil Engineering, University of Roorkee, Roorkee - 247 667, India; M. R. Allu, Student, Dept. of Civil Engineering, University of Roorkee, Roorkee - 247 667, India. 3. At the initial construction stage, instrumentation should be carried out in the drifts, caverns and other important locations in order to obtain field data on displace- ments, both on the supported excavated surfaces and within the rock mass. Instrumentation is also essential for moni- toring construction quality. Experience has confirmed that in complex geological environments, instrumentation is the key to achieving a safe and steady tunnelling rate. These data should be utilized in the computer modelling for back analysis of parameters (Sakurai 1993). At present, non-linear back analysis is difficult and prone to mistakes and misinterpretations. However, expe- rience on many future projects should make advances possible. 4. At the construction stage, forward analysis of rock structures should be carried out using back-analyzed pa- rameters of rock masses. Repeated cycles of back analysis and forward analysis will eliminate many inherent uncer- tainties in the geological mapping and engineering behaviour of the rock mass. Stresses and strains vary considerably within the ground and with excavation phases. Failure will develop where stresses exceed the strength. The zone of failure will propa- gate with phases of excavation. Thus, the sequence of excavation would govern the shape of final failure zone and needs to be simulated. 5. The aim of computer modelling should be to design site-specific support systems, not simply to analyse the strains and stresses in the idealized geological environ- ment. In the case of a non-homogeneous and complex geological environment, which is difficult to predict, slightly conservative values of rock parameters may be assumed for the purpose of designing site-specific remedial measures and accounting for inherent uncertainties in geological and geotechnical investigations. Stresses in the shotcrete may be reduced significantly if the spray of the shotcrete is slightly delayed in non-squeezing ground conditions. The delay, however, should be within the stand-up time. This basic "design as you go~ strategy of computer modelling has been suggested by Hoek (1997). Tunnelling and Underground Space Technology, Vol. 13, No. 1, pp. 71-79, 1998 C 1998 Elsevier Science Ltd Printed in Great Britain, All rig]~ts reserved 0886-7798/98 $19.00 +0.00 Pergamon