AI Driven Carbon Storage Well Connectivity - A Kemper CO2 Storage Analysis

Abstract

Abstract Carbon dioxide (CO2) from industrial process emissions is collected before it is discharged into the atmosphere and then stored in subterranean geologic formations in order to support sustainability goals. Carbon capture storage and utilization represents an important part of reducing carbon dioxide emissions into the atmosphere and maximize value generation. Carbon capture, utilization and storage is a useful instrument for combating CO2 reduction in the environment since it enables business to function while producing less greenhouse gases. But storage needs to be useful, economical, and safe. Both onshore and offshore regions may have storage formations, and each kind of geologic formation has advantages and disadvantages of its own. The placing of CO2 underground for safe long-term storage and encompasses the evaluation of a variety of factors to ensure the gas is effectively trapped. In general, a range of factors are to be taken into consideration in order to determine the favorable formations, such as saline formations, oil and natural gas reservoirs, impenetrable coal seams, organic-rich shales, and basalt formations. The Kemper CO2 storage Project, which is being built in Kemper County, Mississippi, is crucial to understand and acquire experience in how to effectively sequester CO2 and gaining expertise with managing carbon storage projects. It was first built by Mississippi Power, a division of Southern Company, and is alternatively referred to as Plant Ratcliffe or the Kemper County Energy Plant. The project’s objective was to incorporate CO2 sequestration to lower the volume of carbon emissions it generated. Kemper County was chosen as the plant’s location in order to use the region’s undeveloped brown coal potential and offer regional variety, which would assist to balance the state’s energy demand and produce energy. The paper describes a novel directed well connectivity method that makes use of similarity learning to determine well connection for carbon storage. When comparing the connectivity between the injectors and producers, similarity learning can help with connectivity determination for CCUS through categorization and grouping as well as by incorporating different to non-Euclidean metrics in order to assess similarity. In order to foresee prospective CO2 sequestration issues or take corrective action, similarity learning may also be used to detect abnormalities. The association between rising CO2 injection volumes and subsequently observed CO2 quantities in the characterization wells was confirmed by the results, which showed a high correlation between the injector and producer wells. For CCUS well location and sequestration optimization, the approach could serve as a valuable toolkit.