High-Fidelity Numerical Investigation of CO2-Plume Geothermal Systems Using Near-Well Local Grid Refinement

Abstract

CO2-Plume Geothermal (CPG) systems circulate supercritical CO2 through naturally permeable formations, combining geothermal power generation with geologic CO2 storage. Their performance is governed by strongly nonlinear pressure, temperature, and phase-composition gradients in the near-well region, yet meaningful evaluation requires repeated simulation across a broad parameter space, which uniform fine-scale discretisation renders computationally prohibitive. This thesis develops, verifies, and applies a near-well local grid refinement (LGR) framework for thermal-compositional CPG simulation in the open-source simulator open-DARTS. Refined patches are embedded in a structured parent grid through a face-overlay identification of coarse--fine connections, and consistency across the non-conforming interfaces is enforced by a flow-based transmissibility correction derived from deterministic steady-state solves on local fine-scale support grids, applied to both the hydraulic and thermal coupling. Verification against uniformly refined references shows that the benefit of the method is concentrated where it is designed to act: two-phase responses dominated by near-well mobility and CO2 breakthrough are reproduced more accurately than on the coarse grid, at approximately 11% runtime overhead compared with a factor of roughly 23 for uniform refinement. The implementation further reproduces the principal hydraulic and circulation behaviour of an independent CMG-GEM reference simulation. The verified framework is then applied to a 50-year sensitivity study of a CO2--water aquifer CPG system covering horizontal permeability, permeability anisotropy, spatial heterogeneity, and reservoir depth. The results show that net power output is controlled not by the CO2-rich production rate alone, but by its joint evolution with produced-fluid composition and producer wellhead state: stronger inter-well connectivity generally raises the production rate at the expense of CO2 purity and retention, and the best-connected heterogeneous realization yields the poorest net electricity. Under the fixed permeability assumption and the adopted direct-CO2 conversion model, reservoir depth is the only parameter that improves energy and storage performance simultaneously, and the minimum permeability required for positive net power decreases with depth, indicating that CPG candidates should be screened on a coupled permeability--depth envelope rather than on either property alone.