Proceedings, 17th New Zealand Geothermal Workshop
November 8-10, p. 193-198, Auckland, New Zealand, 1995
Automatic history matching of geothermal field performance
Stefan Finsterle and Karsten Pruess
Lawrence Berkeley National Laboratory, Earth Sciences Division
University of California, Berkeley, CA 94720
We have developed inverse modeling capabilities for the multiphase multicomponent numerical simulator
TOUGH2 to facilitate automatic history matching and parameter estimation based on data obtained during
exploitation of geothermal fields. The iTOUGH2 code allows one to estimate TOUGH2 input parameters
based on any type of observation for which a corresponding TOUGH2 output can be calculated. Furthermore,
a detailed residual and error analysis is performed, and the uncertainty of model predictions can be evaluated.
This paper focuses on the solution of the inverse problem, i.e. the determination of model-related parameters
by automatically calibrating a conceptual model of the geothermal system against data obtained during field
operation. We first describe the modeling approach used to simulate fluid and heat flow in fractured-porous
media. The inverse problem is then formulated, followed by a brief discussion of the optimization algorithm.
A sample problem is given to demonstrate the application of the method to geothermal reservoir data.