Stochastic Permeability Estimation for the Soultz-sous-Forêts EGS Reservoir

Abstract

Information on the distribution of permeability at depth is of primary concern in geothermal reservoir engineering. Based on a tracer circulation test performed at the European Enhanced Geothermal System (EGS) test site at Soultz-sous-Forêts, France, three different inversion methods are applied and studied with respect to their potential for estimating the permeability distribution: A fully physical, gradient-based Bayesian inversion, a massive Monte Carlo (MC) approach with geostatistical analysis, and an Ensemble-Kalman-Filter (EnKF) assimilation. A limited, high permeability zone is a common feature of all models. It is the main flow conduit for most of the tracer flow and assumed to be associated with a primary fault zone bypassing two of the boreholes. A common finding of all approaches is that permeability in most of the reservoir is on the order of 10-15 m2; between GPK2 and GPK3 there is a good connection with a mean permeability on the order of 10-13 m2, while a barrier exists to GPK4. The Bayesian Inversion returns rock properties yielding the best fit to the observed tracer concentration at GPK3, but shows that permeability and porosity cannot be retrieved independently. The MC approach identifies various alternative flow paths corresponding to tracer concentrations almost fitting the observations at GPK3 equally well. Although the permeability field obtained by the EnKF has a poor spatial resolution, which is a consequence of only three monitoring wells, it is the only method fitting simultaneously the observations at both GPK2 and GPK4.