Experimental design and response surface method in geothermal energy: A comprehensive study in probabilistic resource assessment

Abstract

The geothermal resource assessment is a critical aspect of field development. Therefore, the numerical reservoir simulation was used to calculate geothermal power potential. However, model input in numerical simulation has a high uncertainty with a low level of confidence, especially in the early exploration stage. Looking into the different experimental design (ED) and response surface method (RSM) techniques is very relevant and essential to the geothermal resource assessment using numerical simulation. The work aims to assess all types of ED and RSM, which were the most effective and efficient regression equation by comparing the Root Mean Square Error (RMSE) and R-square value (R2) for estimated energy potential in numerical reservoir simulation. This approach using five types of ED and RSM, especially, full factorial design 2 level (FFD 2 level), full factorial design 3 level (FFD 3 level), Plackett-Burman design (PBD), Box-Behnken design (BBD), and central composite design (CCD) with 179 numerical models of the Atadei geothermal field. The regression equation based on energy potential calculations for all types of ED and RSM then integrated into Monte Carlo simulations to generate the probabilistic geothermal resource capacity. The comparison results show that BBD has the lowest RMSE value of energy potential (0.12 MW) and the highest R2 value (99.72 %). However, when compared with the probabilistic heat stored method, the results of ED and RSM are in different clusters.