Life-cycle prediction and optimization of sequestration performance in CO2 mixture huff-n-puff development for tight hydrocarbon reservoirs

Abstract

The surge in CO₂ levels in the atmosphere is responsible for the greenhouse effect. Injecting substantial quantities of CO₂ into underground sequestration has emerged as a prominent topic in recent years. Unconventional reservoirs, owing to their complex geological structures, offer secure locations for CO₂ sequestration and enhance the efficiency of hydrocarbon extraction from these intricate subsurface formations. Tight hydrocarbon (such as tight oil and gas) is one of the most representative unconventional resources and has extraordinary development potential. Given its complex pore structure and extremely low permeability, CO₂ huff-n-puff is one of the effective tertiary methods for sequestering CO₂ underground while also enhancing overall cumulative hydrocarbon recovery. As commonly-used gas solvents for increasing the production of subsurface hydrocarbons, CO₂, CH₄ and N₂ show their excellent capabilities when used individually. Their mixture can effectively re-energize reservoirs and securely store large amounts of CO₂ underground, often yielding better results than single gas huff-n-puff. However, comprehensively accounting for the synergistic effects of different gas mixture composition and huff-n-puff operations on CO₂ sequestration and hydrocarbon recovery remains a significant challenge. In this study, a promising AI-based hybrid workflow that incorporates various CO₂ sequestration mechanisms is proposed for life-cycle prediction and multi-objective co-optimization of sequestration performance during the CO₂ mixture huff-n-puff process. A field-scale reservoir numerical simulation model was established to account for the CO₂ sequestration mechanisms involved in the CO₂ mixture huff-n-puff process. Based on the complex, high-precision simulation model, the workflow integrates Temporal Fusion Transformers (TFT) with non-dominated sorting genetic algorithm III (NSGA-III) to achieve efficient proxy-based optimization. This improves the prediction accuracy of CO₂ sequestration volume, oil recovery and NPV while reducing the multi-objective optimization cost. Different optimization schemes are proposed from the perspectives of sequestration scale, productivity, and economic benefits. Compared with the CO₂ sequestration volume, oil recovery, and NPV of baseline, the optimized scheme increased by 15.06 %, 14.52 %, and 3.57 % respectively. This study aims to reduce sequestration costs while maintaining efficient energy extraction and conversion by developing an innovative and extensible workflow for evaluating CO₂ sequestration performance, providing operational guidelines for long-term CO₂ mixture huff-n-puff development.