Javid Ashrafifar - Graduate student of Earthquake engineering, Sharif University of Technology, Tehran
Arman Delaviz - Graduate student of Earthquake engineering, Sharif University of Technology, Tehran

In this paper, Bayesian regression models of fire following earthquake (FFE) ignition are developed that gives a relationship between the intensity of the earthquake and the number of fires that are caused by gas and electrical network damages. At the first step, the data of FFE recent events are collected, and then the Bayesian regression technique is employed to develop a new type of fire following ignition models. Two separate models are developed, the first one for fires that are caused by damage to the gas network, and the second one for fires that are caused by damage to electrical systems. Diagnostics of these models evaluated and final models presented. These models can predict the number of FFE events by determining the source of FFE (such as the damage of gas network), and this is the main difference of this study compared to the recent studies. Finally, for a case study, the developed models are used to predict the mean number and standard deviation of fires that are caused by damage of gas network from a scenario earthquake at the North Tabriz fault.

fire, ignition, earthquake, FFE, Gas network damage, Bayesian model