Assessment of uncertainty in future performance predictions by lumped-parameter models for single-phase liquid geothermal systems

• We present a method for selecting appropriate lumped parameter.
• An RML like procedure is given to model uncertainties in future predictions.
• Correctly quantifying uncertainty requires correct models to be used.
• We give application of the methodology to real field data.

A stochastic simulation methodology is presented for assessing the uncertainty in future pressure and/or temperature data simulated by using history-matched lumped-parameter models for single-phase liquid water geothermal systems. The methodology consists of a two-step procedure; first selecting the appropriate lumped-parameter model that can best describe the geothermal system based on history matching observed pressure and/or temperature datasets and then using a randomized maximum likelihood (RML) like method for the assessment of uncertainty. Any uncertainties in both the model and the measured data may be incorporated into the future performance predictions for the pressure. Once the uncertainty in predicted performance is characterized and assessed, it becomes possible to make reservoir management decisions that account for an incomplete knowledge of the actual geothermal system. One synthetic application and one field application from the Balçova-Narlıdere geothermal field in İzmir, Turkey are presented to illustrate the methodology.