Lumped parameter models provide an attractive alternative to numerical modeling of geothermal reservoirs with the distinct advantage of having to deal with fewer modeling parameters. Hence when such models are used, the aim of the modeling work is then to determine the parameters that best describe the system often through history matching. Once the parameters of the model are determined, we proceed to make future predictions under given production scenarios. However, throughout the modeling work, it is very important to asses the uncertainty that arises from (i) "measurement" errors or noise in observed data, (ii) modeling errors, (iii) nonlinear relationship between model parameters and observed response and (iv) non-uniqueness of the problem. Furthermore, it is crucial that this uncertainty be reflected to the future predictions. Hence instead of dealing with a single deterministic response, one can analyze various possible outcomes of the future predictions.

In this work, we first introduce briefly two methods that can be used for predicting the uncertainty in future flow behavior predicted by lumped-parameter models; The Randomized Maximum Likelihood (RML) and the Ensemble Kalman Filter (EnKF). A synthetic application is given for comparing both methods. Then the RML is used to analyze the uncertainty regarding the future predictions of a lumped parameter model on a real field example. The field at study is the Balcova-Narlidere, a low temperature geothermal field located at the west coast of Turkey.

Press the Back button in your browser, or search again.

Copyright 2007, Stanford Geothermal Program: Readers who download papers from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the original publisher.