(162b) Finding Unknown Anomalies before They Risk Process Safety

Four out of every five oil historical refinery accidents resulted from unexpected, unknown process and mechanical anomalies rather than known, expected failure modes. These unknown anomalies caused unexpected expenses of more than $100 million and ruined excellent company safety records. The oil refinery industry works diligently to address findings from all accidents to avoid repeat incidents. However, the next unknown anomaly which may cause a large property damage loss lurks obscured in a fog of maintenance and operations variance. Rather than wait fearfully for the next unknown anomaly to surface, AI Factory detects all known and unknowns as early as possible. This paper will explain how the latest Artificial Intelligence and Deep Learning techniques from BHGE, a GE company, help eliminate operational surprises. Because refineries generate a wealth of rich data, they need a fundamentally different approach to use information to detect anomalies which lead to 9 digit accidents. Using AI Factory, refiners will find unknown anomalies early and avoid catastrophic surprises. In addition, they can now rapidly refine data driven models and visualize unique, industrial scale operations.