Stochastic quantitative fire risk assessment on old buildings in Hong Kong

Abstract

Recent years, several serious fires accident in Hong Kong occurred in old buildings. It caused fatalities and raised the public awareness on the fire safety problems in the old buildings. However, since the old buildings were built over half century, it is impossible to upgrade the fire safety designs of the old buildings to the current prescriptive requirement. The performance-based fire engineering study provides an alternative approach to tackle this problem. Fire risk assessment is a probabilistic approach in the performance-based fire engineering design. It is particularly suitable to be applied to evaluate the overall performance of a fire safety design. In this paper, a stochastic quantitative fire risk assessment method is introduced. Due to the uncertainties embedded in the basic events of the fault tree or event tree, instead of the single probabilities, probability distributions are provided to describe the uncertainties of the basic events. Monte Carlo approach was adopted to carry out the fire risk analysis to obtain the probability distribution of the overall fire risk. It is much more meaningful than the traditional approach since uncertainties embedded in the probabilities of the basic events are also included in the computation. That is, all information gathered are included in the final result. Three proposed fire safety designs in an old building were evaluated by this assessment method. The results were compared and evaluated with the benchmark design that strictly complied with the prescriptive requirements. The results show that some proposed design can achieve the same or even lower fire risk level than the one complied with code.