Forecasting Energy Consumption and Emissions: A Data-Driven Machine Learning Approach

Abstract This paper presents a solution designed to forecast electricity consumption and carbon emissions at ADNOC Offshore sites by leveraging historical and planned production data, weather conditions, and other operational factors. The primary objective is to predict future energy demands, estimate emissions, and flag deviations for various production scenarios. Additionally, the solution generates comprehensive energy profiles for each offshore field, highlighting key energy consumers and emitters. This ensures compliance with environmental regulations, enhances operational efficiency, and aligns with ADNOC Offshore's future objectives for ESG reporting compliance. The solution transitions from an Excel-based forecasting approach to a more advanced method incorporating machine learning and time series features. By utilizing historical and planned production data, it predicts future energy demands and estimates emissions. The solution pipeline involves collecting operational data from the oil fields, aggregating it monthly, and preprocessing it based on the chosen forecasting algorithm. For short-term forecasting, a multivariate regression model, considering historical data, weather conditions, and operational factors, gave the best results. The solution includes predictive features like linear and multivariate regression analysis, time series forecasting, energy and emission conversions, and automated reporting. It also flags deviations in energy performance and establishes energy profiles and baselines. Furthermore, the solution is scalable to include all oil and gas fields. This innovative solution improves forecasting efficiency by over 95%, enhances data accuracy and quality by eliminating human errors during manual data processing, and enables the identification of major energy-consuming assets. This allows for strategic planning to ensure compliance with future goals and objectives. The novel approach is expected to optimize energy management, reduce operational costs, and enhance asset utilization.