(162c) Digitalization in Process Safety: Fire Risk Prediction Analytics

As PETRONAS move towards digitalization, we envision to converge HSSE and operational excellence and realized the value of our extensive process data. One of our focus is to identify emerging process safety risk utilizing process safety data. With that, we embark on a pilot analytics project with the aim to enable timely interventions to prevent process safety incidents.

The analytic model is build based on Data-driven model and SME-input model. Data-driven model learns the pattern of past incidents while the SME input model detects abnormality based on available process safety information. Weightage is assigned to process indicators and alarm excursions to obtain a risk score and provide an early warning for intervention.

Furthermore, skeptical analysis was conducted against process safety leading indicators determine if there are any early warning indicators prior to an incident.

The pilot model was run on 5-years data in one of our facility with a selected recall accuracy and prediction accuracy. This set the basis of minimum viable product. Real time monitoring is currently in place for the pilot model. The results will serve as a basis for future risk analytic projects.