Use of machine learning techniques in reservoir and seal characterization to estimate carbon storage potential of the Alaska North Slope and Upper Miocene, onshore Louisiana, United States

Geologic carbon storage is widely viewed as a critical component of the energy transition and is now receiving significant attention in the United States near legacy hydrocarbon fields. Louisiana and Alaska are two states that contain large amounts of petroleum infrastructure such as pipelines, roads, surface facilities, etc. that could be repurposed for CO₂ storage. This dissertation seeks to make initial carbon storage estimates for these areas at a site-specific and then a regional scale. To calculate initial carbon storage, chapter II explores how self-organized maps can be combined with post-stack inversion to better define high quality sands. The research highlights a project sized target within the middle Miocene containing 16, 24, and 52 megatons (P₁₀, P₅₀, P₉₀) of estimated CO₂ storage. Using self-organized maps and inversion attributes together is an effective way for investigating an area for carbon storage targets. In chapter III, the scope is expanded to the Greater Kuparuk Field on the Alaskan North Slope, where a pre-stack inversion enhances the interpretation and allows for geobody extractions that delineate sand trends. Improved mapping from seismic inversion and well log measurements provide a new Ivishak carbon storage estimate beneath the Kuparuk oilfield of 1.2, 1.8, and 3.8 (P₁₀, P₅₀, and P₉₀) gigatons. Finally in chapter IV, Shublik seal integrity is evaluated as a possible top seal for carbon storage. Elastic rock properties, seismic attributes, and shale-gouge analysis are combined to identify the faults that pose the highest risk for leakage. Results show the Shublik largely blankets the area, and seal integrity along with thickness is consistent except for the heavily faulted eastern area.