On the use of conservatism in risk assessments. (February 2016)

• Record Type: Journal Article
• Authors: Aven, Terje
• Is Part Of: Reliability engineering & system safety. Volume 146(2016:Feb.)
• Page Start: 33
• Page End: 38
• Publication Date: 2016-02
• Subjects: Conservatism -- Risk assessments -- Knowledge; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2015.10.011
• Languages: English
• ISSNs: 0951-8320

Abstract

Abstract: It is common to use conservatism in risk assessments, replacing uncertain quantities with values that lead to a higher level of risk. It is argued that the approach represents a practical method for dealing with uncertainties and lack of knowledge in risk assessment. If the computed probabilities meet the pre-defined criteria with the conservative quantities, there is strong support for the "real risk" to meet these criteria. In this paper we look more closely into this practice, the main aims being to clarify what it actually means and what the implications are, as well as providing some recommendations. The paper concludes that conservatism should be avoided in risk assessments – "best judgements" should be the ruling thinking, to allow for meaningful comparisons of options. By incorporating sensitivity analyses and strength of knowledge judgements for the background knowledge on which the assigned probabilities are based, the robustness of the conclusions can be more adequately assessed. Highlights: The paper clarifies the meaning of conservatism in risk assessments. It is shown that a conservative policy has severe shortcomings. It is concluded that conservatism should be avoided in risk assessments. "Best judgements" should be the ruling thinking. Sensitivity analyses and strength of knowledge judgements are required.