Dynamics of Fluids in Fractured Rocks
Concepts and Recent Advances
In Honor of Paul A. Witherspoon's 80th Birthday

Lawrence Berkeley National Laboratory, Berkeley, CA, February 10-12, 1999

Evaluation of geothermal well behavior using inverse modeling

Stefan Finsterle1, Grimur Björnsson2, Karsten Pruess1, and Alfredo Battistelli3

1Lawrence Berkeley National Laboratory, Earth Sciences Division
University of California, Berkeley, CA 94720

2Orkustofnun, Grensasvegur 9, 108 Reykjavik, Iceland

3Aquater SpA, San Lorenzo in Campo (PS), Italy

Abstract. Characterization of fractured geothermal reservoirs for numerical prediction of fluid and heat flow requires determination of a large number of hydrologic, thermal, and geometric properties. For use in a computer model that is based on a simplified conceptual model, these properties must be capable of reflecting the complex multiphase flow behavior in a fracture network, including fracture-matrix interaction. We discuss the potential of inverse modeling techniques to provide model-related input parameters based on a joint inversion of field testing and actual production data from a geothermal reservoir. Using synthetically generated data, we demonstrate the need to simultaneously analyze multiple data sets in a joint inversion. The impact of parameter correlations on the estimated values and their uncertainties is also discussed. Inverse modeling techniques are then applied to data from the Krafla geothermal field, Iceland, in an attempt to estimate some critical reservoir parameters such as steam saturation after 20 years of production from that two-phase system. We conclude that inverse modeling is a powerful tool not only to provide input parameters to a numerical model, but also to improve the understanding of fractured geothermal systems. Its efficiency and the insight gained from the formalized error analysis allow an evaluation of alternative conceptual models, which remains the most crucial step in geothermal reservoir modeling.