Dynamic assessment of safety barriers preventing escalation in offshore Oil&Gas

Offshore Oil&Gas is an industrial sector where risk assessment and management has a pivotal role. Moreover, the installation of Floating Production Storage and Off-loading units (FPSOs) in remote regions may hide the emergence of unexpected risks due to emergency and rescue difficulties caused by harsh climate. For this reason, the Norwegian Petroleum Safety Authority (PSA) is committed to ensure appropriate safety monitoring and management within the petroleum sector. This study focuses on safety barriers preventing fire escalation on representative Oil&Gas FPSO unit, such as the recently inaugurated Goliat platform in the Barents Sea. Specific safety barrier management strategy and a safety barrier status panel are currently employed for this unit. In this context, a safety barrier is a technical system aiming to prevent, control or mitigate an undesired event or accident. The purpose of the panel is to provide an overview of the status of the barrier functions and elements in the area to protect. However, an overall risk picture reflecting the real conditions of the platform is not evaluated as support to safety-related decisions. Bayesian Networks are suggested for this purpose. The approach can describe an accident scenario as a succession of events originating from an initial cause and leading to a damage. Between each event, the safety barriers are interposed to represent prevention, control or mitigation measures. Failure probabilities describe the performance of safety barriers and, ultimately, progression towards the final consequences. The probabilities used within the study are derived from relevant norms, standards and guidelines, while the values for update are obtained from the RNNP project (risikonivå i Norsk petroleumsvirksomhet: trends in risk level in Norway’s petroleum activity) by PSA. Barriers of ignition prevention, fire and gas detection and escalation prevention (including related sub-systems) are the focus of the study. Bayesian Networks proved to be, not only an intuitive technique for barrier management due to its potential on graphic representation, but also a reliable model to assess and continuously improve barrier performance.