Case Study: Largest Oil and Gas company reduces unplanned maintenance shutdowns with Mu Sigma predictive analytics

Reducing Unplanned Maintenance Shutdown Using Predictive Analytics

Mu Sigma worked with the largest oil and gas company to address a major operational challenge in the Acid Gas Removal Unit (AGRU): predicting CO2 breakthrough events that could trigger unplanned maintenance shutdowns and production downtime. The customer wanted to automate gas flow monitoring and provide production engineers with at least 10 minutes of advance warning so they could take preventive action and reduce lean amine usage.

Mu Sigma built a neural-net-based predictive analytics solution using resampled sensor data, synthetic event-class data, and probability optimization to reduce false alarms. The model was later adapted for deployment in the plant’s Advanced Process Control system, and a monitoring UI helped engineers identify key event drivers. Mu Sigma’s solution delivered real-time alerts at least 10 minutes in advance, achieved an 86% true positive rate, and helped the company save more than $30MM per year in deferral costs.