Luis Vargas Rojas, Tiago Meira de Borba, Amin Adolfo Claib
Sensia Global
In the realm of artificial lift systems, the monitoring and optimization of Electrical Submersible Pumps (ESPs) are paramount due to the high costs associated with unintended shutdowns and failures. Traditional methods, ranging from simple logic-based local alarms to sophisticated 24/7 surveillance centers, have shown tangible improvements in ESP operation. However, scaling such surveillance systems across large asset fleets cost-effectively remains a significant challenge.
Recent advancements focus on integrating real-time automated event detection engines with AI/ML capabilities to address this challenge. These systems process downhole, surface, and electrical sensor data to identify potentially harmful patterns without requiring human intervention. Unlike traditional multi-level threshold alarms, these engines adapt to each well, minimizing manual finetuning and effectively managing the large volume of notifications and alarms generated by evolving operational strategies and well conditions. The robust design of these engines ensures they are noise-tolerant and capable of differentiating between various sensor failures and harmful ESP conditions, thus reducing false alarms.
Field results from the Oriente Basin, Ecuador’s largest brownfield with approximately 100 producer wells, illustrate the effectiveness of these digital solutions. The integration of a smart digital system equipped with AI/ML allowed for the prediction of critical events, reducing the time of action and avoiding production losses. This led to a significant extension in ESP run life, from 247 days to 950 days. By creating a digital ecosystem that integrates surveillance tools, frequent data gathering, and advanced detection methods, operational efficiency and productivity were maximized while reducing ESP failure events.
The synergy of AI/ML, field knowledge, and real-time data facilitated intelligent actions in digital field operations. This included detecting and ranking operational events, generating focused lists of potential failure threats, and providing actionable insights. Consequently, the number of ESPs needing to be shut down annually was significantly reduced, highlighting the potential of digital transformation in optimizing ESP operations.