Stanford Geothermal WorkshopFebruary 9-11, 2026

Simulating long-term performance of enhanced geothermal systems (EGS) is computationally intensive, limiting the scope of reservoir model calibration and sensitivity analysis. This work introduces a novel reduced-order model, Fast Marching Method-based Simulation (FMM-SIM) to accelerate the geothermal reservoir simulation to make it feasible for real-field applications including Utah FORGE project. The proposed FMM-SIM is a reduced-order modeling approach that transforms 3D fine-scale simulation into a multi-resolution representation using Diffusive Time of Flight (DTOF) which represents the propagation time of the ‘pressure front’ in the reservoir. Our proposed method utilizes a finite-volume Fast Marching Method to efficiently compute DTOF, which then serves as the spatial coordinate for the multi-resolution representation. To ensure accuracy, full 3D resolution is retained near the wellbore and hydraulic fractures, while the remainder of the reservoir is represented by a sequence of 1D grid. The 3D and 1D domains are connected through non-neighbor connections that account for both fluid and heat transmissibility. Whereas conventional reservoir simulations can be computationally intensive to capture complex physics including thermal, compositional and geomechanical effects, FMM-SIM preserves the essential details while reducing high fidelity simulation time by orders of magnitude. We applied the proposed method to both a synthetic model and Utah FORGE model, achieving around 15x speedup in simulation time, allowing each full simulation to finish in about one hour. Using a synthetic model designed after a commercial-scale EGS project, we examined the influence of key parameters in EGS development, including 3C (connectivity, conductivity and conformance), geomechanical effects and well control effects on thermal breakthroughs. Barton-Bandis fracture closure model represents fracture closure and dilation during fluid circulation. We also evaluated intermittent thermal extraction strategies that aim to postpone premature thermal breakthrough. Subsequently, we carried out data assimilation using a multi-objective genetic algorithm to construct a dynamic reservoir model that integrates diverse datasets such as the reference DFN model, the native-state model, DSS measurements, and one month of circulation test data from Utah FORGE project. Integrating the geomechanical module enhanced the accuracy of the history match and indicated that some degree of fracture closure occurred during the circulation period. The calibrated model was then used to test intermittent thermal extraction strategies at the Utah FORGE site, revealing the balance between energy production rate and long term cumulative energy recovery when the production well is periodically shut in. All simulation studies are grounded by accelerated reservoir simulations using FMM-SIM, which enables comprehensive parametric studies and dynamic reservoir modeling. The results demonstrate that FMM-SIM is an effective tool for optimizing geothermal energy extraction through fast and reliable reservoir simulation.

Topic: Reservoir Engineering