| Title | Application of Inversion Modeling in Geothermal and Hydrothermal Reservoirs |
|---|---|
| Authors | Alireza Hassanzadegan, Mauro Cacace, Judith Sippel, Magdalena Scheck-Wenderoth |
| Year | 2014 |
| Conference | European Geothermal Workshop |
| Keywords | Geothermal reservoirs, Inversion modeling, PEST, FEFLOW |
| Abstract | In geothermal reservoir studies we often have to draw conclusions and make decisions using uncertain or incomplete data sets. Reservoir studies includes modeling of all relevant geological structures, populating the model with physical and hydraulic properties and examining their evolution due to changes in pressure, temperature and applied stress. The Berlin study area is placed in the North German Basin (NGB) and its present topographic relief is the result of Pleistocene glaciations. The groundwater in the sedimentary units below Berlin is characterized by freshwater in loose sediments at shallow depth and a salty brackish to saline groundwater within the deeper sediments. Between these two different groundwater compartments a natural hydrogeological boundary is provided by the presence of an impervious clay-enriched layer, separated by a Rupel clay layer. We are using an inversion analysis approach to estimate parameters relevant for coupled heat and transport processes and to quantify the uncertainty associated while using available local data within the regional city context. The result of this study would provide a geologically consistent model useful for the assessments of the physical hydro-thermal and mechanical process occurring in the subsurface. Such a model would therefore serve as a tool to attempt a detailed study of the potential of geothermal energy application in the sedimentary units beneath Berlin. To accomplish this task we couple a commercial finite element hydro-geological code (FEFLOW) to a parameter estimation package (PEST) and we use them to characterize the uncertainty and to estimate hydraulic parameters of interest. PEST not only can be used to calibrate the model but also to analyze the spectrum of the possible solution and consequently uncertainty range. A special algorithm (Gauss-Levenberg-Marquardt algorithm, GLMA) is used to alter the model parameters such that to improve its fit to observed data, iteratively. |