Application of Supersoft Decision Theory in Fire Risk Assessment

Abstract

The application of Supersoft Decision theory (SSD) to fire safety problems, and of decision analysis in general to decisions involving a high degree of epistemic uncertainty, are discussed. SSD and two traditional decision analytic methods employed earlier within the context of fire engineering are compared, particularly regarding how uncertainties are dealt with and the robustness of decisions – robustness concerning the likelihood that the alternative adjudged to be best will change when a reasonable degree of change in assessments of either the probabilities or the utilities involved occurs. Substantial differences between the three methods in decision robustness were noted. It was found that, since traditional decision analysis involving precise probability and utility values gives no indication of robustness, it can lead to incorrect conclusions, making it unsuitable in the present context. It is argued that methods not providing the decision maker with information on decision robustness are unsuitable in situations involving a high degree of epistemic uncertainty. A procedure is suggested involving use of Supersoft Decision theory and extended decision analysis to facilitate the choice between different fire protection alternatives for the case of a specific building.