This presents a significant sustainability dilemma for the oil and gas sector, affecting its environmental performance within the ESG framework. Despite numerous companies in the industry striving for long-term net-zero objectives, accurately forecasting future energy demand remains a complex problem.
Approach: To tackle this challenge, we have deployed machine learning-based solution at Saudi Aramco. This innovative approach aims to precisely forecast the energy consumption and greenhouse gas emissions linked to the injected and disposed water at Gas & Oil Separation Plants (GOSP). Leveraging cutting-edge machine learning algorithms, the solution performs data analysis and improves existing correlations to predict water-cut levels for targeted oil production, as well as the energy demand for specific operations. Moreover, the solution incorporates individual facility sites as well as overall network optimization across multiple fields.
By accurately predicting water-cut levels and energy requirements, this solution facilitates a shift in operational strategies, promoting operational efficiency while meeting environmental, social, and governance (ESG) standards. The presented work outlines the entire process, including the fundamental components of the solution and a case study utilizing actual operational data from a typical GOSP.
Results:
In practice, the teams managing GOSP operations rely on long-term forecasts of water-cut provided by reservoir experts to plan for water injection and separation needs. This involves estimating the energy required for tasks like water pumping and crude separation. However, historical data reveals that the actual water-cut often differs significantly from these forecasts. Therefore, it's more practical to predict energy demand based on past changes in water-cut. This approach helps optimize equipment operation, leading to more efficient energy use and lower carbon emissions. Implementing this solution has led to a notable reduction in energy consumption during hydrocarbon production processes. Initial assessments suggest a potential decrease of up to 4% in energy consumption and associated GHG emissions annually for the selected GOSPs.
Novelity: In this paper, we introduce a groundbreaking approach that harnesses the power of advanced AI models to optimize resources in oil and gas operations. By employing cutting-edge machine learning algorithms, we pave the way for sustainable hydrocarbon production while adhering to Environmental, Social, and Governance (ESG) standards. The methodology and case study presented here offer a blueprint that can be replicated across various facilities and extensive pipeline networks.