Enhancing Energy Efficiency in Oilfield Operations: A Novel Approach Using Artificial Neural Network-Based Predictive Modeling for Electrical Submersible Pump Power Consumption

Abstract This study investigates the application of Artificial Neural Networks (ANN) to predict power consumption in Electric Submersible Pump (ESP) based on their design parameters, with a focus on the B field, which has higher power consumption per barrel of oil compared to other fields in the same region. The objective is to evaluate the ANN model's ability to predict power usage and recommend optimal ESP designs that minimize energy consumption while maintaining production targets. Data from historical ESP installations in the B field was used to train the model, which achieved a high level of accuracy, with an R² score above 0.97 across data splits. The study also demonstrates how the ANN model can be applied to existing ESPs to assess their energy efficiency. Results suggest that total power consumption could be reduced by up to 23% while still maintaining the production level. This study highlights the potential of ANN to optimize ESP performance, offering a scalable solution for managing ESP utilization and reducing both energy usage and carbon emissions. The findings contribute to more sustainable and efficient oil field operations, marking a significant advancement in the use of predictive modelling for ESP optimization.