Quantitative Risk Analysis for The Response of Steel Beams in Fires

This paper develops a probabilistic approach though event trees for the scenarios and Monte-Carlo simulation for the propagation of uncertainties (e.g. in input parameters, models) concerning the evaluation of time and consequences of the failure of structural elements in fires. The time to failure is the output of specific deterministic two zone models (e.g. Ozone) following current regulatory guidelines for possible fire scenarios and input parameters for the design of restrained and unrestrained beams. As a consequence, probabilistic distributions of times replace fixed times to failure of structural elements. These probability distributions are combined in event trees to determine the risk curve for different scenarios, design parameters and fire safety systems such as detection, sprinklers and insulation of beams. The consequence in the risk curve is represented by the magnitude of the probability to failure whereas the associated likelihood is calculated for all possible scenarios. The results allow the relative risk assessment of different design alternatives and the underpinning of sensitive parameters.