Ramachandran, G., 1972. AN APPLICATION OF THE THEORY OF EXTREME VALUES FOR ESTIMATING THE DELAY IN THE DISCOVERY OF A FIRE. Fire Research Notes 929
The economic benefits from fire detection measures depend upon the reduction they can give in the delays in discovering fires. For evaluating these benefits data on the times of origin of fires are required. These data are not available from the offlcial reports of the fire brigades. To supplement the information given in fire reports a team from the Station has visited a series of fires in Hertfordshire and Buckinghamshire. These were fires that required the use by the fire brigades of one or more jets. The team has been able to make estimates of maximum possible delays in 14 recent fires in industrial buildings. The buildings involved in these fires were not provided with detectors. From these data it appears that, in the absence of detectors, the average delay in the discovery of all fires in industrial buildings would be of the order of 110 minutes. For the small sample of 14 fires visited the expected delays ranged from 6.1 minutes to 96.7 minutes. These results are based on an application of the statistical theory of extreme values. It has been assumed that the (parent) probability distribution of delays in discovery is of a simple exponential form. The object of this work is to demonstrate the use of extreme value theory in analysing data, such as those collected by the Station's Fire Visiting Team, for providing results useful for practical purposes and not at this stage to present specific conclusions. There may be other areas in fire research for which extreme value techniques would be useful.