Effect of boundary conditions and surface damage on the shear behavior of rock joints: Tests and analytical predictions

Four series of cyclic direct-shear experiments were conducted on several replicas of three natural fractures and a tensile fracture of welded tuff from Yucca Mountain. The objective of the tests was to examine the effect of cyclic shear loading on joint shear behavior under different boundary conditions. The shear tests were performed under either different levels of constant normal load ranging between 2.7 and 113.9 kN or constant normal stiffness ranging between 25.9 and 328.1 kN/cm. Each test in the two categories consisted of five cycles of forward and reverse shear. Normal compression tests were also performed before and after each shear experiment to measure changes in joint normal deformability. In order to quantify fracture surface damage during shear, fracture-surface fractal dimensions were obtained from profile measurements before and after shear.; Based on the results of the shear tests conducted on several joint replicas under different levels of constant normal load, the shear behavior of joint replicas under constant normal stiffness was predicted by using the graphical method proposed by Saeb (1989) and Amadei and Saeb (1990). The predictions were compared to the results of actual shear tests conducted for the same range of constant normal stiffness. In general, a good agreement was found between predicted and observed shear behavior.; The results of the constant normal load shear experiments were analyzed using several constitutive models proposed in the rock mechanics literature for joint shear strength, dilatancy, and joint surface damage. Some of the existing models appear to have limitations. New constitutive models are proposed and are included in a mathematical analysis tool that can be used to predict joint behavior under various boundary conditions.; Separate from the direct shear experiments, rotary shear tests were also conducted to re-evaluate Barton's empirical failure criterion. Eleven rotary shear tests were performed on replicas of three hollow cylinders of natural fractures with Joint Roughness Coefficient (JRC) values of 7.7, 9.4 and 12.0. The replicas were made from gypsum. By varying the water-to-gypsum cement ratios from 30 to 45%, fracture replicas with different values of Joint Wall Compressive Strength (JCS) could be created. The rotary shear experiments were performed under constant normal (nominal) stresses ranging between 0.2 and 1.6 MPa. The rotary shear test results were compared with predictions using Barton's empirical peak shear strength equation. Observations during the experiments indicated that only certain parts of the fracture profiles influence fracture shear strength and dilatancy. Under relatively low applied normal stresses, the JCS does not seem to have a significant effect. As an alternative, a new procedure for predicting the rotary shear behavior of the hollow cylinder fractures was developed. The approach is based on basic fracture properties such as fracture surface profile data and the compressive strength, modulus of elasticity, and Poisson's ratio of the fracture walls. Comparison between predictions and actual shear test results shows that the alternative procedure is a reliable method.