Probabilistic Fire Simulation

Fire simulation is routinely used in performance-based fire design of buildings and other civil engineering constructions. Modern simulation tools enable to examine the adequacy of the design with respect to a wide range of fire safety objectives including safety to life, directly exposed or neighbouring property and infrastructure as well as the environment. At present, simulations are performed in a deterministic way, i.e., using fixed parameter values selected by the user. However, fire is by its nature a stochastic phenomenon, which introduces a great deal of variability to the development of a fire incident. Further uncertainties arise from deficiencies in our knowledge and models. Using a probabilistic approach to fire simulation it is possible to incorporate into the fire engineering analysis the variability and uncertainties inherently pertaining to any fire incident. Probabilistic fire simulation calls for use of the Monte Carlo technique, i.e., executing a vast number of simulations with parameter values drawn from the distributions characterising their scatter. This requires a lot of computing power, which has only recently become available for the fire engineers as a result of the enormous improvement in the performance of the personal computers. Also the development in fire safety sciences and in its engineering applications contribute to the fact that nowadays it is possible to analyse the fire incident on a probabilistic basis. The paper presents a probabilistic approach to fire simulation which is based on the use of a Probabilistic Fire Simulator (PFS), which is an EXCEL-based interface enabling to run deterministic simulation models in a Monte Carlo mode. The approach is described through application examples concerning fire safety in a cable tunnel in a nuclear power plant, in an industrial and an office building as well as concerning retrofitting a wooden facade in a concrete block of flats.