Creep of rock in situ, like most rock mass behavior, will be largely governed by the behavior of the natural discontinuities -- bedding planes, faults, and joints, in particular. Several past studies have investigated various aspects of the time-dependent deformation of rock discontinuities; examples include Amadei and Curran (1980), Engelder and Scholz (1976), Johnson (1975), Kaiser and Morgenstern (1979), Solberg et al. (1978), and Wawersik (1974). These studies have provided much insight into the general phenomenon, but they have also produced some inconsistent conclusions and left many unanswered questions. This paper will attempt to shed some light on one of these questions: What is the influence of stress level on the creep of unfilled rock joints? The discussion will follow a two-part approach to the problem: (1) laboratory creep experiments on small scale jointed specimens, (2) a simple theoretical mechanism to explain, at least in qualitative terms, some of the results from the experimental program.
Uniaxial creep tests were conducted on intact and jointed samples of a specially formulated gypsum plaster "synthetic rock." Synthetic rock materials have several advantages over natural rock for long-term creep testing: (a) higher creep rates, (b) lower strength and therefore lower loading requirements, (c) simpler specimen preparation, and (d) more controlled and consistent properties. Gypsum based plasters have been used previously in studies of short-term strength and deformability of jointed samples (Patton, 1966; Einstein and Hirschfeld, 1973). A gypsum plaster mix originally developed by Nelson (1968) for a study of short-term jointed rock behavior was used for all of the creep tests in our investigation. The mix consists of three ingredients: (a) water, (b) Hydrocal B-11 gypsum plaster (a product of the U.S. Gypsum Coo), and (c) Celite (a product of the Johns-Manville Co.), a fine-grained diatomaceous earth used as an admixture to prevent bleeding of the water during mixing and curing. All creep tests were performed on 32 mm. (1.25 in.) by 32 mm. (1.25 in.) by 121 mm. (4.75 in.) tall rectangular prismatic specimens loaded in uniaxial compression. Specimen fabrication was carefully controlled to minimize variations in properties among samples. Jointed specimens were prepared by carefully sawing intact samples and then sanding the joint surfaces with 60 grit paper, producing a joint peakfriction angle, øj, of 42º . The uniaxial constant-stress compressive loading was applied by dead weights using a specially modified soil consolidation frame. Test duration in most cases was approximately 50 hours.
A series of sets of tests on intact specimens loaded to 0.2sc, 0.4sc, and 0.5sc stress levels was performed to provide information on the general nature of the creep behavior for the gypsum synthetic rock and its stress dependence. In order to minimize the influence of material variability between specimens in these and all other test series reported herein, three tests were run in each set and an average or "composite" power law creep function was fitted to the entire set of data using conventional least-squares regression analysis.
