Fire Fragility Curves for Steel Buildings in a Community Context: A Methodology

This paper proposes a novel methodology for developing fire fragility functions for an entire steel building - meaning that the function is not specific to a location within the building. The aim is to characterize the probabilistic vulnerability of steel buildings to fire in the context of community resilience assessment. In developing the fragility functions, uncertainties in the fire model, the heat transfer model and the thermo-mechanical response are considered. In addition several fire scenarios at different locations in the building are studied. Monte Carlo Simulations and Latin Hypercube Sampling are used to generate the probability distributions of demand placed on the members and structural capacity relative to selected damage thresholds. By assessing demand and capacity in the temperature domain, the thermal and the structural problems can be treated separately to improve the efficiency of the probabilistic analysis. After the probability distributions are obtained for demand and capacity, the fragility functions can be obtained by convolution of the distributions. Finally, event tree analysis is used to combine the functions associated with fire scenarios in different building locations. The developed fire fragility functions yield the probability of exceedance of predefined damage states as a function of the fire load in the building. The methodology is illustrated on an example consisting in a prototype nine-story steel building based on the SAC project.