Influence of microstructure on fracture propagation in rock

Abstract

Influence of Microstructure on Fracture Propagation in Rock This paper describes the results of research to correlate the fracture resistance with microstructural features of Salem limestone and Berea sandstone. Tests were conducted on wedge-loaded double-cantilever-beam specimens containing machined slots as crack starters. The fracture resistance of Salem limestone measured in terms ofR, the energy dissipated per unit area of projected surface, was found to increase in the initial stage of crack extension but finally reached a constant value which was strongly dependent on crack orientation with respect to the bedding plane. For this limestoneR ranges from about 50 joules/m2 to 230 joules/m2 (1 joule/m2 = 104 erg/cm2). The sandstone which is a more compliant rock exhibited similar fracture behavior while theR values ranged from 465 joules/m2 to 1580 joules/m2. In addition, tests in liquid nitrogen which were aimed at eliminating plastic deformation in the rocks during fracture showed little difference inR for the limestone but a substantial reduction inR for the sandstone which, in the latter case, may be caused by differential thermal expansion between the quartz grains and the calcite cement. Acoustic emissions were detected in both rocks at very early stages in the tests indicating the occurrence of microcracking near the initial slot tip at low loads. The mode of fracture and the fracture path in both materials were identified by fractography over a broad range of magnifications. The evidence gained from this work strongly points to the existence of an extensive array of microcracks produced in a region surrounding the main crack tip and which advances with it. The energy dissipated in fracturing of rock is associated with the creation of the large amount of surface area contained in this microcrack array. This picture provides a self-consistent explanation for the puzzling discrepancy between the measured tensile strength of rocks and the strengths predicted from measuredR values.