Maya Yermekova, Zeshan Hyder, Agustin Gambaretto, Akshay Dhavale
SLB
Sucker rod pumps (SRPs) remain the leading artificial lift (AL) method worldwide, with technological advancements tracing back to the wave equation development in the 1960s. Leveraging edge-based technologies, a new workflow has been developed to enhance existing Pump-Off Controller (POC) capabilities. This workflow integrates machine learning (ML)-driven dynamometer card classification for real-time event detection with an advanced logic system that autonomously optimizes SRP operating setpoints. Operating within an Industrial Internet of Things (IIoT) framework, it continuously analyzes high-frequency dynamometer card and pump data.
The workflow consists of two distinct control mechanisms tailored for SRPs:
Fast Loop Mitigation Controls – These controls utilize classified surface and downhole dynamometer cards in real time. Designed for rapid response, they detect and mitigate common SRP issues such as flatlining, fluid pound, gas interference, and tagging as they occur.
Production Optimization (POPT) Algorithm – This algorithm collects and evaluates operational data within a dynamically shifting time window. By synthesizing historical trends into performance indicators, it forecasts the optimal pump operating setpoints to enhance efficiency and production.
Testing results highlight the significant advantages of combining both control systems. Across tested wells, inferred production increased by an average of 15%, while runtime improved by 3%. Additionally, by maintaining optimal pump fillage, cycling was reduced by 29%, leading to more stable operating conditions.
This workflow represents a holistic approach to SRP optimization, bridging short-term issue mitigation with long-term production enhancement. By integrating real-time anomaly detection with predictive optimization, it provides a comprehensive and adaptive solution for maximizing well performance and reliability.