Using fluid-induced seismicity to infer permeability

Abstract

For engineering applications involving fluid flow through porous rocks underground, the permeability of a rock mass is an important parameter. Imaging of the 3D permeability distribution is generally done by a history-matching approach: fluid pressure due to injection in a well with known pressure and flow rates is numerically simulated, and the rock permeability is adjusted so as to match predicted pressure values to measured observations. These pressure observations are made at producing wells. Because there will normally be few wells, the observations, while being dense in time, are spatially sparse. The idea of this paper is to augment these downhole pressure measurements by using microseismic data to infer pressure at seismic event locations.The mechanism of pressure-induced seismicity is the reduction of effective normal stress across a plane of weakness. The rock strength can be represented as a critical pressure - the pore pressure above which the rock will fail. The rock strength is highly heterogeneous, because of the existence of weaknesses such as joints, bedding planes, and clay bands, and stronger regions such as sandstone channels. So the rock strength will be random. We model the rock strength as a Weibull-distributed critical field. A microseismic event occurs where pressure exceeds this critical field, and so is effectively a point pressure measurement, with an uncertainty given by this Weibull distribution. The idea is to augment the well-pressure observations with these “virtual” pressure measurements at seismic event locations.We model pressure diffusion using a finite volume approach, and examine the inversion, for permeability, of two different types of data, (1) pressure measurements in boreholes, and (2) virtual pressure measurements at seismic event locations. We show that the two types of data provide complementary information.