Crack Features and Shear-Wave Splitting Associated with Fracture Extension during Hydraulic Stimulation of the Geothermal Reservoir in Soultz-sous-Forêts

Abstract

The recent tomography results obtained within the scope of the Enhanced Geothermal System (EGS) European Soultz project led us to revisit the meso-fracturing properties of Soultz test site. In this paper, we develop a novel approach coupling effective medium modeling and shear-wave splitting to characterize the evolution of crack properties throughout the hydraulic stimulation process. The stimulation experiment performed in 2000 consisted of 3 successive injection steps spanning over 6 days. An accurate 4-D tomographic image was first carried out based upon the travel-times measured for the induced seismicity [Calò M., Dorbath C., Cornet F.H., Cuenot N. (2011) Large-scale aseismic motion identified through 4-D P-wave tomography, Geophys. J. Int. 186, 1295-1314]. The current study shows how to take advantage of the resulting compressional wave (Calò et al., 2011) and shear-wave velocity models. These are given as input data to an anisotropic effective medium model and converted into crack properties. In short, the effective medium model aims to estimate the impact of cracks on velocities. It refers to a crack-free matrix and 2 families of penny-shaped cracks with orientations in agreement with the main observed geological features: North-South strike and dip of 65°East and 65°West [Genter A., Traineau H. (1996) Analysis of macroscopic fractures in granite in the HDR geothermal well EPS-1, Soultz-sous-Forêts, France, J. Vol. Geoth. Res. 72, 121-141], respectively. The resulting output data are the spatial distributions of crack features (lengths and apertures) within the 3-D geological formation. We point out that a flow rate increase results in a crack shortening in the area imaged by both compressional and shear waves, especially in the upper part of the reservoir. Conversely, the crack length, estimated during continuous injection rate phases, is higher than during the increasing injection rate phases. A possible explanation for this is that cracks remain large because the system has time to relax. We also calculate the extension and opening rates during all hydraulic stimulation sets. While the opening rate is unchanged, the extension rate varies depending on the stimulation phase. It is also shown to be higher around and above the open-hole section than below. This can indicate a potential upward path that makes fluid percolation easier within the granite formation, this path being induced by the temperature gradient. We also compare the evolution of crack extension during injection with shear-wave splitting. Split shear waves were recorded at 2 stations during hydraulic stimulation and processed in terms of splitting parameters. The fast shear-wave polarization remains constant and parallel to the maximum horizontal stress orientation while the amplitude of splitting varies with time. We observe a good agreement between travel-time differences and crack extension rates during the first 4 days of the stimulation experiment. Afterwards, these two parameters depart from each other. This study emphasizes the added value of the coupling between effective medium modeling and shear-wave splitting to monitor meso-scale cracks in reservoirs submitted to hydraulic stimulation.