PROBABILISTIC RESOURCE ASSESSMENT USING THE NGATAMARIKI NUMERICAL MODEL THROUGH EXPERIMENTAL DESIGN AND RESPONSE SURFACE METHODS (ED AND RSM)

Abstract

Reservoir simulations are becoming prevalent in geothermal resource assessments. Natural-state reservoir simulation models based on robust conceptual models are used to simulate a number of development scenarios in geothermal prospects to evaluate the long-term reservoir performance. The Experimental Design and Response Surface Methods (ED and RSM) workflow was applied to the natural-state reservoir model of the Ngatamariki geothermal field in order to assess the effects of uncertain parameters and scenarios to the simulated field generation capacity. The workflow provides a systematic way of building multiple versions of the TOUGH2 model through a designed pattern of parameter combinations. The results from multiple model runs are used to evaluate the uncertain parameters, weigh their significance, and create a response surface model. The response surface (polynomial) lends itself well to Monte Carlo methods of generating the probabilistic range of generation capacity. From the current practice of providing deterministic results from reservoir simulation scenario runs, a full probabilistic assessment was demonstrated through the ED and RSM workflow. In addition, the uncertainty analysis performed provides a basis for managing development risks and uncertainties.