| Abstract |
At Soultz-sous-Forêts in France, a European Enhanced Geothermal project drilled and hydraulically stimulated several wells in faulted and fractured granite during the 1990’s and 2000’s. During each stimulation, thousands of cubic meters of water were injected at high pressure into open wellbore, triggering slip on pre-existing fractures, inducing seismicity, and permanently enhancing the productivity of the wells. Understanding how storativity and permeability is distributed and created at Soultz is critical to design of future hydraulic stimulations and prediction of reservoir performance. This study investigated the permeability distribution at Soultz using pressure transient analysis in conjunction with qualitative observations from other sources of data. Pressure transient analysis is an inverse problem, which means that multiple mathematical models may be capable of matching the same data set. In this work, we proposed three candidate conceptual models that were based on well log observations. A few parameters in each model were varied until a good fit to the data was achieved. The objective was to evaluate whether the parameters used to match the data were physically plausible. A model that requires unrealistic physical parameters to match data could be considered unrealistic. At Soultz, several lines of data indicate that flow from each well is localized in a small number of fracture zones with a width of several meters. Given that interpretation, we investigated three conceptual models that were considered candidates to explain the distribution of permeability in the fracture zones. The three conceptual models were (A) flow in a crack, (B) flow in a crack embedded in an intermediate permeability and porosity damage zone, and (C) flow in a damage zone with uniform permeability and porosity. Model A, flow only in a fracture without a damage zone, did not appear to be consistent with the observations. Model A could be matched to the data, but the match required an unrealistically high hydraulic diffusivity and negative skin. If the fracture zones at Soultz were like Model A and contained only a single major flowing fracture, they could not provide the storativity needed to contain the volume of fluid injected during the Soultz stimulations. The two models involving damage zones, Models B and C, were more successful at plausibly matching the data. Of the two, the crack and damaged zone model, Model B, was more problematic, because it required an extremely negative skin. The negative skin might be explained by fracture opening, but stress calculations suggested that it is unlikely that the fracture opening could have occurred. Another problem was that the transient displayed changes in the pressure derivative that would difficult, but perhaps not impossible, to explain in light of the observations. The homogenous permeability damage zone model, Model C, matched the data well with plausible values for permeability and skin. It required a moderately negative skin, but the negative skin could be explained by slanted well penetration of the fracture zone. |