Risk assessment: What is it worth? Shall we just do away with it, or can it do a better job?. (November 2017)

• Record Type: Journal Article
• Authors: Pasman, Hans J.; Rogers, William J.; Mannan, M. Sam
• Is Part Of: Safety science. Volume 99: Part B (2017)
• Page Start: 140
• Page End: 155
• Publication Date: 2017-11
• Subjects: Process hazards -- Consequence analysis -- Hazard identification -- Risk management; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2017.01.011
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Evidence is summarized on the uncertainties in QRA results. Ways how to characterize the quality of QRAs are reviewed. Hazard identification methods by system approach are described. Consequence and failure determination shortcomings and improvements are highlighted. New tools, risk metrics and an optimal risk acceptance policy are described. Abstract: Much has been written about the value of risk assessment, in particular the quantified version, QRA, here considered in the context of industrial processing oil, gas, and chemicals. It is plagued by mistrust due in part because estimated uncertainties are generally not included. QRA is done for predictive purpose, to design risk reducing measures, and to base safety decisions on, but many major accidents occur according to scenarios not foreseen in the analysis. Yet, QRA forces us to think. Predicting what can happen under specified conditions is the most elementary step in better safety. So, we must improve the methods and in particular to model sensitive uncertainties and establish confidence limits about our predictions. The present paper will review in short the state of affairs, and it will summarize recent contributions shedding light of what is currently wrong with QRA. The paper will further present for the various aspects and steps to be done in a QRA directions, in which progress is being made and where promising possibilities may appear. The latter are according to the opinion and the scope of the authors, so hopefully there is much more! Anyhow, due to complexities of socio-technical systems a QRA will require more effort, so to keep it practical large scale application of computerization will be a must. And, to improve trust in QRA we must acknowledge variability and quantify the uncertainties as we see them; in addition, there will be hidden uncertainties, which gives reason to think in resilience terms too.