A Deep Learning Framework for Thermal Enhanced Oil Recovery Optimization of Hydrogen from H2S – A Maari Reservoir Study

Abstract

A particularly corrosive and poisonous by-product of a range of feedstocks, including fossil resources like coal and natural gas as well as renewable resources, is hydrogen sulfide (H2S). H2S is also a possible source of hydrogen gas, a significant green energy carrier. Our business would greatly benefit from the recovery of H2 from chemical compounds that have been classified as pollutants, such as H2S. Due to the large volumes of H2S that are readily accessible across the world and the expanding significance of hydrogen and its by-products in the global energy landscape, attempts have been undertaken in recent years to separate H2 and Sulphur from H2S using various methods. In addition to deep gas reservoirs, hydrogen sulfide may be found in a wide range of other reservoir types. Due to their low use, these gas reserves often have little economic viability. Thanks to novel strategies for converting hydrogen sulfide into hydrogen and its remaining components, it has become possible to efficiently recover hydrocarbons and its hydrogen sulfide components. This paper introduces a unique deep learning (DL) architecture for improving field recovery over time based on thermal-enhanced recovery. We investigated performance of the framework on the Maari Field in New Zealand. The ultimate goal is to optimize recovery and, within the limits of processing, reach a specific volume of H2S. The optimization results indicate the ability to increase oil and natural gas recovery while constraining H2S levels within the reservoir and converting the associated H2S into hydrogen. The deep learning architecture that has been built provides a technique for developing field strategies to improve sustainability for thermal-enhanced recovery strategies. The framework is flexible enough to incorporate additional reservoir and production parameters.