Case Study: a major oil and gas company predicts ESP failures with Avathon

Major Oil and Gas Company - Customer Case Study

A major oil and gas company was struggling to predict costly electric submersible pump (ESP) failures. They faced an insurmountable barrier due to poor sensor data quality and quantity, caused by the pumps' remote locations, frequent power outages, and insufficient data recording. They partnered with the AI solutions provider Avathon to find a way to implement predictive maintenance despite these data challenges.

Avathon implemented two custom machine learning approaches. They first used a normal behavior model that successfully identified 5 out of 7 historical failures 13-35 days in advance. A second, more scalable clustering and classification model was also created. Together, these AI-powered solutions could have saved the company over 70 days of deferred production, demonstrating that predictive analytics is possible even with imperfect data.