Data-driven methodologies for process risk assessment
I am a visiting scholar at MKOPSC and a Ph.D. candidate at Memorial University of Newfoundland. Process digitalization demands a transition of conventional risk assessment approaches to data-driven approaches to determine system state and take preventive measures to maintain a system to an acceptable risk level. A mishap will result in loss of life, property damage and environmental degradation. Lessons from previous accidents are a good source of learning. These learnings assist in tracking and avoiding future adverse events. My work aims to develop data-driven methodologies to leverage data and convert it into meaningful information. Data extraction is performed by employing natural language processing on the databases such as Chemical Safety and Hazard Identification Board (CSB). It will help identify critical parameters such as causal factors, contributing factors, failure scenarios, and consequences associated with an incident. Hence, these features will turn into incident scenarios, from root causes to consequences to perform risk analysis. This work will assist in predicting accidents caused resulted in fires and explosions in chemical processing industries and serve as an essential tool of Safety 4.0 and a mechanism to enable process safety in the process system digitalization.