USING A LARGE RESERVOIR MODEL IN THE PROBABILISTIC ASSESSMENT OF FIELD MANAGEMENT STRATEGIES

Abstract

For the probabilistic assessment of exploitation strategies of a geothermal reservoir, alternative numerical simulation models are constructed (most likely, pessimistic and optimistic), in order to represent the uncertainty range of key reservoir properties affecting field performance. While each of these models is calibrated to the initial state and field historical behavior, they significantly diverge in the prediction of future reservoir performance. The entire set of models is used to evaluate some exploitation strategies. However, some uncertain parameters, although not directly related to reservoir properties, can also affect to some extent overall field behavior. Therefore a complete coupling of the reservoir simulation models with a probabilistic evaluation would lead to an unreasonably high number of model runs, making it computationally impractical. To overcome this problem, polynomial approximation functions have been developed for the predicted reservoir performance, based on specially designed model runs and on the recognized relationship between well deliverability and cumulative steam production. The polynomial expressions are incorporated into a complex economic model that allows the combined evaluation of the complete range of uncertainties potentially affecting a given project, without the need of specific, time consuming reservoir modeling runs.