Emission Reduction in Oil & Gas Subsurface Characterization Workflow with AI/ML Enabler

Abstract

According to (McKinsey & Company, 2020), drilling and extraction operations are responsible for 10% of approximately 4 billion tons ofCO2 emitted yearly by Oil and Gas sector. To lower carbon emissions, companies used different strategies including electrifying equipment, changing power sources, rebalancing portfolios, and expanding carbon-capture-utilization-storage (CCUS). Technology evolution with digital transformation strategy is essentialfor reinventing and optimizing existing workflow, reducing lengthy processes and driving efficiencyfor sustainable operations. Details subsurface studies take up-to 6-12 months, including seismic & static analysis, reserve estimation and simulation to support drilling and extraction operations. Manual and repetitive processes, aging infrastructure with limited computing-engine are factors for long computation hours. To address subsurface complexity, hundred-thousand scenarios are simulated that lead to tremendous power consumption. Excluding additional simulation hours, each workstation uses 24k kWh/monthfor regular 40 hours/month and produces 6.1kg CO2. Machine Learning (ML) become crucial in digital transformation, not only saving time but supporting wiser decision-making. An 80%-time-reduction with ML Seismic and Static modeling deployed in a reservoir study. Significant time reduction from days-to-hours-to-minutes with cloud-computing deployed to simulate hundreds-thousands of scenarios. These time savings help to reduce CO2-emissions resulting in a more sustainable subsurface workflow to support the 2050 goal.