Fire Load Survey and Statistical Analysis

Abstract

Fire load surveys in many building types are presented and analysed. They are based on two datasets which are studied in two separate chapters, after a brief presentation of the measurement method. The first dataset concerned fire loads in industrial and commercial buildings in Switzerland. ETH (Eidgenossische Technische Hoschule Zurich) and VKF (Swiss Fire Authority, Berne) inspected 95 companies in about 40 industrial sectors leading to 336 measures. A statistical analysis is made and the agreement between those data and two probability laws is studied. It is shown that lognormal law provides a better agreement than Gumbel law. The second dataset concerns public buildings (such as shopping centers, hotels and hospitals) and offices in France for a total of 139 rooms. Those data were collected in the framework of “Projet National pour l’Ingenierie de la Securite Incendie” agreed by the French Reseau Genie Civil & Urbain (PNISI: National Project for Fire Safety Engineering). The same process as previously is applied: a chi-square test. Here, differences are not significant between Gumbel law and lognormal law. The explanation is related to the ratio of standard deviation over mean value which is much smaller for this dataset. It leads to equivalent distribution functions. The composition of fire load is also analyzed and reveals that wood material continues to be present in great proportion.