Study On Risk Assessment Method By Coupling Fire Dynamics And Statistical Theory

Fire may induce considerable casualties and monetary loss, because of high occupant density, complex function and inconvenient evacuation in large public and high-rise building. With development of economy and architecture technology, these super high, super large and novel buildings will increase in recent years. The current prescriptive code is not suitable to the fire safety design of these new-style, complex and special buildings. The recommended method is based on performance-based fire protection design. However, fire risk assessment is the foundation of performance-based fire protection design, and appropriate performance-based fire protection design can not be well implemented without reasonable fire risk assessment. On one hand, until now, some fire risk assessment methods such as risk classification based on stochastic rules have become mature. However, these methods are difficult to analyze fire risk quantitatively. On the other hand, because of the complex of fire process, the uncertainty of fire being considered as a disaster and the incompletion of current information and knowledge, risk assessment can not be implemented only based on deterministic rule. Therefore, how to couple fire dynamics with statistical theory to make quantitative fire risk assessment is a problem. Based on summarizing the current fire risk assessment methods, building fire monetary loss assessment and expected fire risk to life safety assessment are presented in this study. The contents of this paper are as follows:A new building fire direct monetary loss assessment method is presented. According to various fire dynamics characteristics in different fire development process and fire protection systems, fire from ignition to spread in the whole fire compartment is divided into five phases. Firstly, fire spread probability assessment method is presented based on fire dynamics and fire protection system. Secondly, critical time of every fire phase can be calculated according to fire dynamics of every phase. Then, fire spread area can be calculated. When monetary factor is introduced, fire monetary loss can be obtained. This building fire direct monetary loss assessment method not only predicts direct fire monetary loss quantitatively, but also provides reference on fire insurance rate determination.The effect of pre-movement time on evacuation is studied with consideration of stochastic rules of pre-movement time. Based on evacuation software Gridflow, the effect of pre-movement time characterized by normal distribution and explicit value on evacuation time is studied under one floor and multi-layer building scenarios. The simulation results shows: (1) occupant evacuation time is the function of pre-movement time and occupant density. When pre-movement time follows normal distribution, for small occupant pre-movement time, evacuation time is a function of pre-movement time and occupant density. For large occupant pre-movement time, evacuation time is dominated by pre-movement time little irrespective of occupant density. When pre-movement time is characterized by explicit value, no matter how large pre-movement time, occupant evacuation time will dominated by occupant density. (2) For large mean pre-movement time, low occupant density, or wide exit, as long as little congestion and queue comes into being; the flow of people through the exit is not blocked. When pre-movement time follows normal distribution, total evacuation time also follows normal distribution shifting to right by a constant.An expected fire risk to life assessment method is presented. This method makes expect risk to life (ERL) as object. There are two key parameters: occurrence probability of fire scenario, number of injuries and deaths of every fire scenario. With consideration of some stochastic factors of fire spread and smoke movement (fire growth rate, probability of fire protection system), probable fire scenario can be constructed based on event tree and fire simulation software. Then, occurrence probability of every fire scenario at different time can be obtained according to discrete markov chain. For occupant evacuation time calculation, fire detection time is characterized by probability distribution with consideration of uncertainty of fire growth rate. Occupant pre-movement time is characterized by normal distribution. By comparison onset time to untenable conditions with evacuation time, number of injuries and deaths of every fire scenario can be calculated.