Subsurface Flow And Carbon Emission Optimization Using Graph Network Surrogate Models With Carbon Accounting

Abstract A framework that integrates an AI surrogate with a carbon accounting simulator to optimize well locations and control strategies in subsurface flow operations is presented in this paper. The primary objective is to maximize Net Present Value (NPV) by balancing operational efficiency with carbon emission costs within a carbon tax framework. Our methodology involves training an AI surrogate to accurately predict pressure and saturation dynamics in a 2D oil-water reservoir. These predictions are subsequently used to calculate NPV while accounting for carbon emissions as estimated by the carbon accounting simulator. By incorporating a carbon tax into the economic model, we assess the influence of carbon emissions on profitability and identify optimal well placements and control settings that enhance NPV. The optimization process leverages the surrogate model to swiftly evaluate numerous scenarios within a synthetic reservoir model. Results demonstrate that the integrated AI surrogate–carbon accounting simulator framework significantly enhances techno-economic performance. The surrogate model reliably forecasts reservoir behavior, facilitating rapid optimization across various scenarios. Notably, incorporating carbon tax considerations alters optimal strategies compared to traditional cost-focused approaches, achieving up to a 50% reduction in greenhouse gas emissions while maintaining satisfactory production efficiency. These findings underscore the critical role of carbon tax in operational decision-making. This framework offers a computationally efficient tool for industry practitioners to optimize subsurface operations, align with sustainability targets, and comply with regulatory requirements.