Investigation of the Use of the Ensemble Kalman Filter (EnKF) for History Matching Pressure and Temperature Data from Geothermal Reservoirs

Abstract

In this study we investigate the use of the Ensemble Kalman Filter (EnKF) method for estimating model parameters and quantifying uncertainty of future performance predictions of reservoir models for liquid dominated geothermal reservoirs. Specifically we concentrate on the performance and accuracy of the method. We couple the Ensemble Kalman Filter with lumped parameter models (tank models) for testing the method. The lumped parameter models used in the study are capable of modeling the average pressure behavior of liquid dominated geothermal reservoirs. This is accomplished by solving the mass balance simultaneously on all tanks that represent the various components of a geothermal reservoir (components such as the aquifer or the reservoir itself). The model parameters that are used in the inversion process are mainly recharge indices between tanks, storage capacities and initial pressures of the tanks.

Our main goal in this study is to have a clear understanding about the Ensemble Kalman Filter method and how it performs. We first present synthetic examples then use the method on real field data. The method seems to be very advantageous in terms of speed compared to other gradient-based history matching procedures (e.g., the Levenberg-Marquart) even though the problem we are dealing with in this study is composed of only a few model parameters.