Just-in-time maintenance: a blind test on a drilling contractor's compressor
Sensor and maintenance history blended and auto-grouped, with adaptive learning that removes retraining, tested blind against a compressor's recorded rebundling, bearing and sensor events.
Challenge
An offshore drilling contractor wanted to cut non-productive time by maintaining equipment just in time, on predicted component failure, and to extend maintenance intervals on condition.
Solution
Arundo ingested and blended sensor and maintenance data, auto-grouped the sensors by cluster relationship, built a custom interface integrated into the operator's workflow, and used adaptive learning to avoid model retraining and still catch the unexpected. The approach was tested blind on a compressor with a known history: a rebundling, a thrust-bearing replacement and a sensor calibration.
Impact
Anomaly detection on the compressor's history as the basis for just-in-time maintenance.