1. INTRODUCTION
The static and dynamic response of a Jointed rock mass is controlled in large part by the behavior of discontinuities consisting of joints, fractures and planes of weakness. A considerable amount of interest has been given to the development of numerical models employing finite element and discrete element idealizations for such media. The majority of these models, however, have employed an ideal, non-dilatant Coulomb type friction idealization and while they have been extremely useful for many purposes, it is well known that natural rock Joints are rough and the dilitancy they exhibit is an important phenomenon which probably should not be neglected. Rock joints are inherently rough and due to the natural processes of cracking and perhaps mineral deposition involved in their creation, contact frequently takes place over a relatively large portion of the available contact surface. This roughness and close initial mating gives rise to the phenomenon of dilatancy: the tendancy of two contacting bodies to separate during relative tangential motion due to the sliding of asperity surfaces of one body on those of the other. When this increase in contact surface volume is constrained or partially constrained, as is the case in any compliant rock mass or laboratory testing apparatus, the phenomenon of dilatancy manifests itself through increasing normal compressive stresses which in turn, can substantially increase a Joint's resistance to additional slip. Thus, dilatancy can serve as an important stabilizing effect. However, the asperity surfaces which are responsible for the dilatancy have finite strength and if sufficiently stressed, will degrade and affect a change in the subsequent joint behavior. Quantitative description of dilatancy is complex and a number of models have been developed. Patton (1966) proposed a model consisting of an interface with asperity teeth oriented at an angle with respect to the mean plane of the Joint. For "low" compressive stresses, the model behavior is characterized by dilatancy and the overriding of asperities. For "high" stresses, the model behavior is characterized by shearing through asperities. A similar model was proposed by Jaeger (1971) which featured a smooth transition from asperity overriding to asperity shearing behavior and Ladanyi and Archambault (1970) developed s model with an asperity strength. Barton (1973) proposed an empirical model based on a Joint roughness coefficient. Ghaboussi, Wilson and Isenberg (1973) and Roberds and Einstein (1978) developed models by considering elastic-plastic deformations of a joint and Goodman (1976) developed a model by idealizing an interface to consist of nonlinear springs. The Patton, Jaeger and Ladanyi and Archambault models explain the phenomonology of Joint behavior in that they describe the shape of the failure envelope. These models however, do not account for pre- sliding behavior and explicit relationships between increments of stress and deformation, which are necessary for a constitutive law to be Implementable in a general numerical procedure, are not obtainable. The Barton model is a useful empirical tool for shear strength criteria, however, an explicit relationship between stress and deformation is also unobtainable. The Ghaboussi, Wilson and Isenberg model and the Roberds and Einstein model are significant advances in Joint constitutive models in that they describe the underlying, or microstructural behavior of rock Joints including presliding elastic behavior and postsliding plastic behavior.
