The Utilization of Machine Learning and Deep Learning to Predict Real-Time ESP Well Status

Abstract

Abstract One of the most common, efficient and reliable artificial lift methods used for lifting high volumes of fluids from wellbores is the Electric Submersible Pump (ESP). Thus, monitoring the live status of this ESP is essential to determine whether a well is on stream or off. In this research, a new methodology that utilize advanced Machine Learning (ML) to recognize ESP performance status in real time and provide it to engineers instantaneously. In this research, the proposed methodology was to develop an intelligent system that consciously monitor ESP sensors and provide a reliable predicted well status to engineers. Firstly, the proposed system actively fetches real-time data from ESP sensors. Secondly, an advanced pool of ML algorithms was created, and trained on historical well status data. Finally, the real-time acquired data are fed to the advanced ML model to automatically identify ESP well status to engineer and notify them in advanced. After finalizing the advanced ML system, it was evaluated on its performance to accurately predict the real-time status of ESP well and provided system users with targeted results. In addition, performance of the system on extraction, mapping and mitigation attributes were evaluated using ROC-AUC performance matrix. Validating and testing the ML model disclosed a promising outcome scoring accuracy exceeds 98 % which indicated high reliability of the model to accurately predict well status instantaneously. The developed ML system enabled engineers in the office to actively monitor well performance status fast and securely. Moreover, by implementing such a system, significant impact on our operation were achieved leading to cost and time savings. The developed ESP well model enhances the way artificial lift engineers visualize and analyze wells status in real time. It's worth noting that the proposed system lead to fast and substantial improvements in achieving a desired result field wise. Also, the system provides a detailed description and analysis of the well status to engineers. The developed ESP performance recognition system has achieved significant fiancial and time saving especially in offshore location where a reliable ESP well and field status are critical in decision making saving millions of dollars as well as thousands of engineering hours.