Intact and fractured rock samples were studied in the laboratory in order to understand more fully the mechanism of closure of fractures subjected to high confining stresses and the resultant effect on sample permeability. Confining stresses applied to the samples ranged from 3.5 to 20.5 MPa, and the closure of fractures was observed by monitoring changes in the hydraulic conductivity of the specimens. Test results suggest that some resealing may occur due to crushing and realignment of mineral grains along a fracture surface. The closure of fractures is dependent upon the strength of the rock mass, the physical nature of the fracture, and the fluid pressure present in the fracture. Flow rates through fractures obeyed the cubic law although the induced permeability resulting from the fractures was very sensitive to changes in aperture, and was affected by the matrix permeability of the rock mass and the roughness of the fractures.
The general approach to analysis of fluid flow through fractures has been to equate fluid flow through fractures with viscous, incompressible flow between smooth parallel plates (see Snow, 1965). By considering the flow to be laminar, and the plates to be horizontal, the flow velocity and hydraulic gradient relationship becomes: (mathematical equation)(available in full paper)
3 LABORATORY PROCEDURES
The Hassler cell apparatus was used for testing the hydraulic conductivity of the cylindrical specimens. Specimen dimensions were kept as uniform as possible to reduce the size effects that have been observed by other researchers including Gale and Raven (1980). Thirty-four samples were prepared from intact NX sized core specimens which ranged in diameter from 500 mm to 510 mm. The length of the specimens varied between 950 mm and 1150 mm. Four different lithologies were tested: Berea Sandstone, Hartshorne Sandstone, basalt from the Umatilla member in the Columbia Plateau Series, and gneiss from the Orofino Metamorphic Series near Dworshak Dam, Idaho. All samples were tested using straight flow along a single horizontal fracture that ran along the longitudinal axis. Fractures were induced in the specimens by a modified Brazilian test procedure, in which triangular platens were used to ensure even loading and splitting. All fracture surfaces were mapped using a LVDT and contour plots were generated to help characterize fracture roughness, and nonplanarity. Samples were _first saturated and initially tested for their hydraulic response under loading, in order to determine as accurately as possible their matrix permeability. The samples were then fractured and the surfaces of the induced fractures were characterized. The samples were resaturated and then reloaded into the Hassler cell permeameter to test their fracture permeability. Precautions were taken in order to ensure that the fractures were as horizontal as possible before final emplacement of platens and sealing of the pressure chamber. Water was employed as the confining medium, with nitrogen gas used to apply both the confining and infection pressures. Flow rates through the samples were measured as a function of both confining and head pressures, during both loading and unloading cycles.