Fires after earthquakes sometimes develop into conflagrations resulting in widespread losses of life and property. A geographic information system (GIS) model linked to property and valuation data is shown to be an appropriate tool for estimating urban fire losses. One approach uses a static buffering technique to define potential burnout zones that are sampled randomly to give estimates of losses. The other uses a dynamic cellular automaton technique for determining both the rate and extent of fire-spread in response to a wide range of factors including wind, radiation, sparking, branding, building separations and building claddings. The dynamic approach uses a set of 'rules' based on fire physics modified by historical data. The model runs in real time for single ignitions. The static method is used to estimate losses assuming a 12m separation will prevent fire spread All buildings are assumed combustible (upper bound case). The dynamic model assuming fire can not spread to buildings with non-combustible claddings and areas of vegetation are not flammable (lower bound case).

SUMMARYGenerally, fire resistant structures are expected to survive a fire in an adjoining compartment. Some structures, such as floors, may be designed to provide time for the occupants to escape from compartments other than the one where the fire occurs. In the fire compartment smoke development governs the time available for egress. A common misconception is that the fire resistance rating (FRR), the time an assembly will survive in a test furnace, is the time available to escape. In small compartments such as those in residential accommodation their FRR is significantly longer than the times the assemblies would survive in real compartment fires. Some fire engineering designs for retrofitted accommodation buildings use FRR times for light timber frame walls and floors as the available egress time, which is unconservative.The method of time equivalence can provide a prediction of the FRR required to survive a compartment burnout. The ratio of the total burning time of the fire to the time equivalent for the compartment and fuel load can be used to provide an estimate of the time taken for an assembly of given fire resistance to fail by multiplying the FRR by the ratio. Although this method is shown to be non-conservative when a computer model of light timber frame wall assemblies is run using both realistic time-temperature curves and the ISO-834 standard fire test time-temperature curve, it is more conservative than assuming that an assembly will last as long in a compartment fire as predicted by the FRR.

No abstract