Deep learning-enhanced global sensitivity analysis for uncertainty quantification in THMC coupled scCO2-EGS

Quan Gan, Hongli Song, Derek Elsworth, Sida Jia, Junjun Chen, Funing Ma, Qian Li, Yaling Yang, Xiaoping Wang and Zhenxue Dai

Energy, 2025, vol. 335, issue C

Abstract: Understanding thermo-hydro-mechanical-chemical (THMC) coupling is essential for optimizing subsurface resource extraction. However, existing models struggle with computational efficiency and uncertainty quantification due to strong nonlinearities and intricate multi-physics interdependencies, hindering the identification of key parameters and the optimal design of complex processes involving fluid flow, heat transfer, mechanical effects, and reactive transport. This study proposes an integrated framework that employs deep learning-based surrogate modeling to accelerate global sensitivity analysis (GSA), enabling the identification of optimal control portfolios through quantitative sensitivity indices to decouple uncertainties across parameters, physical fields, and spatial domains. The framework is applied to a THMC coupled model of supercritical CO2-enhanced geothermal system (scCO2-EGS), simulating formation fluid distribution, petrophysical property evolution, and mineral reactions. To reduce computational costs, ResNet-18 serves as surrogate models for efficient prediction. By applying GSA to quantify the influence of parameters on the output at each model grid, this framework supports sensitivity evaluation across THMC fields and spaces. Findings indicate that wellbore pressure and injection rate control CO2 plume behavior, while porosity-permeability evolution is mainly influenced by wellbore pressure and fracture spacing. Hydraulic-mechanical fields dominate CO2 plume and porosity-permeability evolution, while mineral reactions exhibit strong additional coupling with the chemical field. In addition, spatial sensitivity patterns provide prompt, optimal and available observational data for monitoring. These results prove its efficiency and interpretability for evaluating uncertainties across parameters, physical fields, and spatial domains, offering broad applications in multi-physics coupled systems under uncertainty, including geothermal energy, CO2 storage, and nuclear waste management.

Keywords: THMC coupling model; Global sensitivity analysis; scCO2-EGS; Deep learning; Porosity-permeability evolution (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225037284
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:335:y:2025:i:c:s0360544225037284

DOI: 10.1016/j.energy.2025.138086

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().