Case Study: E&P Company improves offshore production and prevents pump failure with SparkCognition SparkPredict AI

Improve Offshore Production With AI

An upstream E&P Company wanted to improve the stability and availability of a critical multi-phase pump on an unmanned offshore platform to meet production targets and support aggressive growth goals. To better use its existing data platform investment, the company partnered with SparkCognition and deployed SparkPredict AI software for predictive maintenance and anomaly detection.

SparkCognition used more than 300 pre-processed data values from Cognite Data Fusion to build a normal-behavior model for the pump and later expanded the solution to key subsystems including the seal, seawater lift, and lube oil systems. The models detected 75% of known historical failures with two to 12 days of advance warning, helped prevent a pump failure that could have caused days of downtime, and is expected to deliver more than $5M in value from a single alert while avoiding production losses worth up to $500K per day.