Accelerating CCUs Storage Site Screening Using AI/ML: A Data-Driven Approach to Early-Stage Subsurface Evaluation

Le résumé fourni par la source

Abstract As the energy landscape shifts toward sustainability, the role of subsurface gas storage particularly in porous geological reservoirs has become a necessity for carrying the future of energy systems such as hydrogen, CO2, and natural gas storage. However, the selection of suitable underground sites, especially depleted fields and aquifers, remains demanding and challenging. Traditional approaches rely on manual data mining, quality control, analytics, and prolonged report reviews, placing significant demands on subsurface specialists, inflating costs, and delaying project timelines. Addressing these limitations, this paper introduces an AI and machine learning framework for streamlining early-stage screening of gas storage sites. The methodology was validated across more than ten gas fields, with an AI/ML-assisted comparative analysis leading to focused evaluations specifically, Chance of Success (CoS) calculations on the most promising candidates. The approach demonstrates its value through assessments of sites spanning a wide risk spectrum, with cases such as structurally robust anticlines facing minimal aquifer risk and, conversely, sites prone to aquifer interactions and complex containment challenges. AI-driven models in this workflow deliver enhanced subsurface risk assessment by detecting faults, seal weaknesses, fluid anomalies, and aquifer signatures using large historical datasets. This enables the elimination of zones with unmitigable risks and drive site selection with greater confidence. The integration of structural integrity, reservoir continuity, aquifer proximity, and analog geodata further strengthens the models ability to calculate CoS scores with embedded uncertainty, streamlining the ranking process for a diverse set of geological storage candidates. Preliminary project experience demonstrates that this approach can reduce the timeline of the initial site screening phase by 20-30%. This helps geoscience and reservoir engineering teams to focus their efforts and resources on locations with the highest potential, while complementing established simulation workflows, rather than replacing them. The proposed AI framework thus offers a rapid, scalable, and repeatable screening process, automating risk checks and blending advanced analytics with domain expertise. Ultimately, this study advances in subsurface site selection by accelerating business decisions and supporting the global shift to cleaner energy through smarter, faster geological screening and de-risking, directly fueling progress toward net-zero targets.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.