$site = "publications.iafss.org"; $fullsite = "publications.iafss.org"; $basePath = "/home2/firesag5/private/data/"; ?>
Ramachandran, G., 1972. EXTREME VALUE THEORY AND FIRE LOSSES - FURTHER RESULTS. Fire Research Notes 910
In a previous paper the author illustrated the use of the theory of extreme values for analysing the largest losses due to fires in buildings. In this paper the theory is extended so as to cover the top 17 extreme losses in the textile industry. A few statistical problems concerning these extremes and their averages are discussed. Using the estimated values of the parameters of these extreme value distributions a method for assessing the total loss in smaller fires in a given year is also illustrated. This method could also be used to estimate the expected loss in a particular building of known value at risk. Problems for further research have also been suggested. Conceptually, the intensity function of the probability distribution of fire loss is 'u' shaped. But, neglecting the infant and early stages of growth of fire this function increases exponentially. In 1967, there were about 105 fires in the textile industries with individual losses in the range between Â£55 and Â£10,000. The overall average loss in these fires was about Â£2,200. In the same year and in the same industry sprinklered buildings had an average loss of Â£1,600 for a comparable loss range. Hence in non-sprinklered buildings the average loss was about Â£2,800 indicating a saving of Â£1,200 per fire due to sprinklers in the range considered. It is extremely unlikely that the total loss in all smaller fires (costing less than Â£10,000) in the textile industries in 1967 was more than Â£300,000.