Evaluating uncertainty in steel-composite structure under fire : application of the ME-MDRM

Probabilistic applications of structural fire engineering will be central to the realisation of complex buildings, where reliance on precedent / experience is insufficient, and an adequate level of safety must be explicitly demonstrated. The practical application of probabilistic methods in structural fire safety has, to date, been limited to simplistic structural systems due to a lack of an efficient, unbiased calculation methodology. Herein, two examples are presented whereby the Maximum Entropy Multiplicative Dimensional Reduction Method, ME-MDRM for short, is applied to estimate the probability density function of performance metrics output from finite element analyses. Given a limited number of model realisations (order of 101), it is shown that the MEMDRM can give valuable insight into the distribution of failure time, and maximum supported load under ISO fire conditions for a composite column, and composite slab panel, respectively. The former is benchmarked against Monte Carlo Simulations, with excellent agreement. Finally, limitations in the stochastic data and in the application of the ME-MDRM itself are discussed, with future research needs identified.