Experimental Study on Nonlinear Flow in Granite Tensile and Shear Fractures

Abstract

Water flow experiments in single tensile and shear granite fractures were conducted to investigate the nonlinear flow behavior. Fracture geometry parameters based on single fracture wall (Rp, Z2, Z3, Z4, JRC) and two fracture walls (aperture distribution, cluster coefficient) were calculated and compared. The effect of these fracture geometry characteristics on nonlinear flow behavior was explored. Fracture transmissivity, critical Reynolds number, pressure gradient, and normalized transmissivity were analyzed in detail. The result shows that shear fracture geometry has the characteristics of higher mean aperture with higher standard deviation, more clustering contact, and rougher fracture surface. The shear to tensile fracture transmissivity ratio increases from 2.3 to around 370 with the increase of confining pressure from 2 to 60 MPa and the stress-dependence coefficient of shear fractures is close to 1/22 of tensile fractures. The critical pressure gradient increases as confining pressure, indicating that it requires a higher pressure gradient to drive the linear flow to nonlinear flow, namely, when the injection pressure gradient keeps unchanged, high confining pressure causes the nonlinear flow to turn to linear flow. The critical gradient of the shear fracture is 0.004–0.527 MPa/m on average, which is much lower than that of tensile fractures (0.011–26.862 MPa/m), meaning that a lower pressure gradient can drive the linear flow to nonlinear flow in a shear fracture. The average Forchheimer coefficient β for tensile fractures is 0.133, which is about half of the shear fracture (0.253). Regression analysis indicates that a linear relationship exists between β and the fracture geometry parameters (mean aperture η and cluster coefficient ς) and the correlation coefficient is about 0.925 and 0.920, respectively. Further investigation confirms the “weak inertial effect” with the (T0/T − 1)/Re data points falling in a straight line before the “strong inertial effect,” where the data points fall in an approximately horizontal line.