Calibration of a Geothermal Model Using PEST

Abstract

The natural state model of a geothermal reservoir numerically represents the state of the reservoir prior exploitation. Starting with an existing natural state model that gives a generally good match between simulated and measured down-well temperatures, inverse modeling is conducted to further improve the model. The inverse modeling software PEST is used to automatically calibrate the model. PEST utilizes advanced regularization and singular value decomposition techniques to ensure the numerical stability of the inversion process. Permeability values of the reservoir rock are used as parameters that PEST automatically adjusts so that the difference between the measured and the simulated temperature is reduced. Significant improvement in the match to the measured temperatures is achieved after several optimization runs.