Moving to an Autonomous Artificial Lift Concept through Automatic ESP Restart

Abstract

Electrical submersible pump (ESP) trips and unplanned shutdowns can be a major operational challenge for many oil fields. In most of these ESP trips, the ESP can be returned back to production after conducting proper troubleshooting at surface and without any downhole intervention. The process of manually restarting tripped ESPs can be a complex and costly operation, especially in an offshore environment. Alternatively, automatic ESP restart can offer great advantages by reducing the ESP downtime. Many of the variable speed drives (VSDs) available in the market offer an auto restart feature that allows the ESP to be restarted automatically without human intervention. This paper presents the concept and the application of this technique. The activation of ESP auto restart requires considerable technical review of the different trip causes and the proper restart methodology for each. Auto restart of each trip type has to be programed differently to prevent possible harm to the ESP. Specific engineering measures and procedures shall be put in place to ensure personnel and equipment safety. In this paper, some statistical tools for ESP trips and restarts are presented to measure the success of auto restart, its effectiveness, and its limitations. The obtained results from the ESP auto-restart technique show it to be both practical and beneficial; it can significantly reduce the time to put the ESP back in operation resulting in production advancement. In addition, continuous data collection and assessment of auto-restart events play an important factor in ensuring that auto-restart settings are properly applied and adjusted for each type of variable speed drive installed in the field. Finally, the paper provides several recommendations with suggested ways to improve the functionality of this feature. The technique introduced in this paper can bring artificially lifted fields closer to an autonomous and intelligent concept of operations. The presented model can serve as a good benchmarking tool for future implementation of artificial lift automation.