S U M M A RYWe propose an approach for estimating the permeability tensor using seismic emission induced by borehole hydraulic tests or by a £uid injection of an arbitrary nature. This approach provides a single estimation of the permeability tensor for the complete heterogeneous rock volume where the seismic emission was recorded. The approach is an extension of the method proposed by Shapiro et al. (1997) for the isotropic case. It is based on the hypothesis that the triggering front of the hydraulic-induced microseismicity propagates like a low-frequency second-type compressional Biot wave (corresponding to the process of pore-pressure relaxation) in an e¡ective homogeneous anisotropic poroelastic £uid-saturated medium. The permeability tensor of this e¡ective medium is the permeability tensor of the heterogeneous rock volume upscaled to the characteristic size of the seismically active region. We demonstrate the method using the microseismic data collected during the Hot Dry Rock Soultz-sous-Foreª ts experiment (Dyer et al. 1994). These data show that the corresponding rock volume is characterized by a signi¢cant permeability anisotropy caused by oriented crack systems. The maximal principal component of the permeability tensor has a subvertical orientation. It is about seven times larger than the minimal subhorizontal principal component.

Hydraulic fracturing is a common procedure to increase the permeability of a reservoir. It consists in injecting high‐pressure fluid into pilot boreholes. These hydraulic tests induce locally seismic emission (microseismicity) from which large‐scale permeability estimates can be derived assuming a diffusion‐like process of the pore pressure into the surrounding stimulated rocks. Such a procedure is applied on six data sets collected in the vicinity of two geothermal sites at Soultz (France) and Ogachi (Japan). The results show that the method is adequate to estimate large‐scale permeability tensors at different depths in the reservoir. Such an approach provides permeability of the medium before fracturing compatible with in situ measurements. Using a line source formulation of the diffusion equation rather than a classical point source approach, improvements are proposed for accounting in situation where the injection is performed on a well section. This technique applied to successive fluid‐injection tests indicates an increase in permeability by an order of magnitude. The underestimates observed in some cases are attributed to the difference of scale at which the permeability is estimated (some 1 km3 corresponding to the seismic active volume of rock compared to a few meters around the well for the pumping or pressure oscillation tests). One advantage of the proposed method is that it provides permeability tensor estimates at the reservoir scale.

Summary We propose an approach for estimating the permeability tensor using seismic emission induced by borehole hydraulic tests or by a fluid injection of an arbitrary nature. This approach provides a single estimation of the permeability tensor for the complete heterogeneous rock volume where the seismic emission was recorded. The approach is an extension of the method proposed by Shapiro et al. (1997) for the isotropic case. It is based on the hypothesis that the triggering front of the hydraulic‐induced microseismicity propagates like a low‐frequency second‐type compressional Biot wave (corresponding to the process of pore‐pressure relaxation) in an effective homogeneous anisotropic poroelastic fluid‐saturated medium. The permeability tensor of this effective medium is the permeability tensor of the heterogeneous rock volume upscaled to the characteristic size of the seismically active region. We demonstrate the method using the microseismic data collected during the Hot Dry Rock Soultz‐sous‐Forêts experiment (Dyer et al. 1994). These data show that the corresponding rock volume is characterized by a significant permeability anisotropy caused by oriented crack systems. The maximal principal component of the permeability tensor has a subvertical orientation. It is about seven times larger than the minimal subhorizontal principal component.

No abstract