Optimizing Injection-Production Allocation in Geothermal Systems Using Advanced Look-Ahead Reservoir Models

Abstract

Abstract This paper introduces a novel approach to optimize the allocation of injection and production wells in geothermal systems, based on the Look-Ahead Models (LAMs) framework that was originally designed for hydrocarbon reservoir optimization using numerical simulation. The goal is to improve energy extraction efficiency and to extend the sustainable operation of geothermal reservoirs while minimizing computational cost. The approach consists of using reduced order (proxy) models to gather information about future performance and feed it into the high-fidelity model (HFM). This way, only one detailed simulation is required to globally maximize enthalpy extraction and thermal efficiency over time. LAMs are advanced predictive tools that integrate future predictions of well performance and reservoir behavior into a fine scale full-field reservoir models using subrogate models. Adapting these models to geothermal systems, requires accounting for the specific heat and mass transfer dynamics, including factors like thermal breakthrough and reinjection effects. This adaptation involves modifying existing LAM frameworks to accurately represent the complex interactions in geothermal reservoirs particularly in terms of Temperature response at production wells. Simulation results using this adapted LAM approach on a theoretical geothermal reservoir exhibit marked improvements in both energy extraction efficiency and reservoir longevity, delaying thermal breakthrough by 5 years. The method shows potential in reducing operational costs and enhancing the predictability of reservoir behavior, which is critical for long-term sustainability. Using LAMs enables quasi-global optimization while keeping computational cost to a minimum. Because the LAM models are used as guides, the algorithm itself is extremely robust and can be integrated into our existing in-house reservoir simulator with minimum disruption.