The resistance of rock to tensile fracture may be measured by its fracture energy G•, which is found to range from 40 to 200 J/m e in tests on nine types of sedimentary and crystalline rock. Differences in microstructure among the rocks tested are the principal cause of differences in the steady state value of G•, in the distance that a crack must advance before steady state fracturing is attained, and in the amplitude of the fluctuation of G• that accompanies crack advance. When nearly continuous surfaces of weakness are present, as in the Salem limestone, G• is low and attains steady state after only a small amount of crack advance. When a preexisting, interconnected network of microcracks is exploited by the fr9cture process, G• is large, and steady state is attained only after extended crack propagation. The sensitivity of G• to crack speed and the presence of water is low under the test conditions used in all the rocks examined. However, the magnitude of G• measured in a given type of rock is found to depend on the configuration of the test specimen and on components of stress near the crack tip that do not influence crack growth in linearly elastic materials. The conditions under which G• can be considered a material property are therefore restricted.
INTRODUCTIONPeck et al. [1985] (hereafter referred to as paper 1) used the fracture energy G• as a measure of the resistance of rock to the propagation of tensile cracks. Measurements of Gt with a standardized test method on nine different types of rock show that Gt ranges from 40 to 200 J/m '•. Here we present evidence on the relation of Gt to the structure and composition of rock. We also consider the sensitivity of G t to changes in the state of stress, chemical environment, and fracture rate. Because these are quantities likely to be quite different in the field from the conditions that can be attained in the laboratory, it is important to know how sensitive Gt is to them when test data are to be applied outside the laboratory. The test results also bear on the question of the degree to which G t can be considered a material property.
METHODS AND MATERIALSAll tests were done with wedge-loaded, double cantilever beam test specimens; the measurement procedures and specimen design used are described in paper 1. The results reported there show that even under the best of conditions and with samples of the most homogeneous rocks, G t cannot be reproduced to better than about 10%. Because of this, the dependence of G t on variables such as crack speed is more easily found with interrupted tests than by comparison of the results of tests done on different specimens. (In an interrupted test the crack advance is started under one set of conditions and continued until a steady value of G t is observed; test conditions are then changed, and as crack growth continues, the change in G t is noted.)The structure of freshly formed fracture surfaces was observed with both optical and scanning electron microscopes; care was taken to minimize the disturbance of the surface ...