Indirect Estimations of Frictional Coefficients of Fractures in Sandstones for Analysis of Injection Induced Microseismicity

Abstract:

The frictional coefficient of fractures, a fundamental parameter needed to analyze a variety of geomechanical problems for microseismicity, is normally determined from laboratory shear tests. However, recovered rock cores are rarely available because of difficulties and high cost in getting undisturbed core samples. In that case, the frictional coefficient should be either assumed or estimated indirectly. We investigate the frictional property of fractures of various sandstones in laboratory tests and attempt to correlate that with other properties measureable relatively readily even without cores. We use various sandstones obtained from different depths of a 1 km deep borehole drilled for coal bed methane development in a Paleozoic sedimentary basin, South Korea. The sandstones have various physical properties (e.g. P-wave velocity (V_P) of 2253-5038 m/s) and chemical compositions in terms of clay content (5-31%). We conduct direct shear tests in an artificial saw-cut fracture in the sandstones and determined frictional coefficients in a range of 0.36-0.57. The frictional coefficients have an inverse-linear correlation with clay contents measured from XRD analysis. These results are also quite consistent with those from previous clay gouge experiments (Takahashi et al., 2007; Tembe et al., 2010; Kohli & Zoback, 2013). They also have a linear correlation with V_P. Our study demonstrates that frictional coefficients can be estimated empirically from such properties. To check feasibility of such an approach, we apply the obtained empirical relation to the borehole where cores were recovered. The clay contents in sandstone formations are estimated from the borehole gamma ray log calibrated using the XRD clay content data. Clay content estimated from gamma ray varies significantly with depth in a range of 0-45%. This range of clay content corresponds to frictional coefficients of 0.25-0.58. Comparison between estimated and measured frictional coefficients shows a reasonable (but quite scattered within ±20%) agreement. Estimation of frictional coefficients is further constrained using a dual variable regression analysis of both clay content and V_P of the rocks (within ±10%). We believe that calibration of frictional coefficient-properties relation is necessary to apply this approach to other regions.