| Abstract |
Lessons learned from reservoir modelling using a Discrete Fracture Network approach in the framework of EGS activity, at Soultz site or from benchmarking exercises in earlier programs, have gradually helped us to better understand a number of coupled processes. This resulted in a general improvement of the numerical tool FRACAS. The present contribution illustrates a new step forward, which is applied to simulate long term fluid circulation with thermal couplings in a multiple-cell reservoir. The new hypothesis formulated here is that the fracture network density in the stochastic approach can be regionalized in space according to the sets of micro-seismicity events recorded during the stimulation phases of the various parts of the exploited reservoir. Size distribution for the fractures follows a power law. Special care is given to the numerical treatment of seismic data sets ranging in different catalogues (e.g. only a part of the seismic events recorded in 1993, 1995, 1996, 2000, 2003 2004, 2006 and 2010 were made available for this study). Main inflow/outflow zones at the four deep wells (GPK1, GPK2, GPK3, GPK4) are considered in an explicit way. Hydraulic parameters for the long term exploitation are extracted from past studies, in the upper 3km reservoir and at 4.5 km depth and partly re-calibrated from the last 2010 circulation phase, with a GPK2 production of about 18 l/s. Extrapolation of hydraulic and thermal behaviours are made for the next 5 years, with an enlarged production rate (+25%), testing the impact of a partial re-injection of cold fluid in GPK1, as experienced in 2011. Model results do confirm in qualitative and quantitative ways, the possibility on running a balanced circulation with limited re-injection pressures. Fluids re-injected in upper parts of the reservoir through GPK1 well have a negative impact on temperature at production as they are in favor of outlet zones at intermediate depths in GPK2. |