Optimizing production well geometry in the Utah FORGE geothermal project using machine learning and fluid flow modeling

Abstract

This study addresses the critical challenge of optimizing well placement in Enhanced Geothermal Systems (EGS), specifically within the framework of the Utah FORGE geothermal project, as part of the 2023 Society of Petroleum Engineers (SPE) Geothermal Datathon. Effective well placement is essential for enhancing geothermal production efficiency and maximizing resource utilization. We employed a discrete fracture network (DFN) modeling approach, utilizing the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm from the Scikit-Learn library to analyze microseismic event location data. Through rigorous simulations conducted in the open-source GeoDT fluid flow simulator, we identified an optimal production well configuration characterized by a spacing of 400 m, an injection rate of 0.03 m³/s, and alignment parameters that significantly improve thermal recovery. The results indicate a projected net present value (NPV) of $75 million over a 20-year operational horizon, underscoring the economic potential of optimized well placement strategies. This study offers valuable insights for the operation of the FORGE geothermal site. More importantly, it exclusively utilizes open-source tools, enhancing accessibility and adaptability for the broader geothermal community.