A statistical analysis of occurrence and association between structural fire hazards in heritage housing

Abstract

Reported in this paper is a novel application of statistical analysis of structural fire hazards that were found in heritage housing stock in a metropolitan area. The structural fire hazards in terms of non-compliances to the building regulations are digitised and then subjected to statistical analysis to obtain estimates of probabilities of occurrence under various conditions. The concepts of Hamming distance, Jaccard distance, virtual distance and pairwise Phi correlation coefficients are employed in the analysis to estimate the association between the fire hazards. Estimates of the probability distribution over the number of joint occurrence of hazards and pairwise joint probabilities are also obtained. In addition the 3-tuple and 4-tuple joint probabilities are analysed. Finally, logistic regression models are established to correlate each fire hazard with the others. The results show that not only the probability of occurrence of structural fire hazards is high, but probability of multiple occurrence is also significant. There are indications that some structural fire hazards are correlated. The findings of this study may assist certifying authorities, building surveyors, fire safety engineers and fire services in identifying fire hazards in heritage buildings and developing alleviating and effective strategies or solutions to protect life safety of building occupants as well as the cultural heritage values of the relevant building stock.