Estimation of risk under fire design situation

Experience and available data indicate that depending on particular conditions and applied fire safety system, die probability of fire flashover and possible structural failure may be expected within a broad range. It appears that Bayesian causal networks provide a capable tool for analysing efficiency of fire safety system and for estimation of risk under fire design situation. The network consisting of twelve chance nodes, two decision and five utility nodes is used to analyse a fire safety system. Chance nodes of the network include fire start, detection, tampering, sprinklers, smoke detection, fire brigade, fire flashover, and structural collapse. The decision and utility nodes are implemented to enable estimation of risk under various decisions concerning sprinklers and structural protection. It is shown that the probabilities and estimated risks due to structural failure expressed in terms of economic consequences are particularly dependent on the conditional probability of sprinklers are acting given the fire started. It appears that obtained findings including estimates of risks provide valuable information for making decisions concerning fire application of sprinklers and structural protection.