同時多発火災を想定した大規模延焼シミュレーションに基づく延焼危険性指標の提案と消防隊行動支援への活用

抄録

 It is difficult for firefighters to extinguish all fires in a large earthquake because multiple fires break out simultaneously in different places. For reducing the number of burnt-down buildings in densely built-up wooden residential areas, it is important to preferentially extinguish fires at burning buildings that are likely to spread widely. The damage of fire-spread depends on the fireproof performance of each building and the spatial characteristics of each town. However, the methods for quantifying fire-spread risk of a building have not been sufficiently discussed.
 In this paper, first, we construct a detailed simulation model that describes the spread of fire after a large earthquake. This model can take into account the fire-spread direction according to the difference of fire-prevention performance among buildings. Using this model, we propose two fire-spread indices (‘Fire-spread potential’ and ‘Burnt-down potential’) for each building based on the results of fire-spread simulation. Fire-spread potential is defined as the number of buildings destroyed by fire-spread from a building. This index can be used to clarify buildings with high-risk of fire-spread. Burnt-down potential is defined as the probability that a building is burnt down by fire-breakout from itself or by fire-spread from other buildings, and useful for discussing the safety of important places such as evacuation areas. Based on the Fire-spread potential under the assumption of Tokyo Inland Earthquake, there might be some regions with high-risk for destroying more than 300 buildings in average by a fire. Also, about a half of buildings in Tokyo Metropolitan area have the burnt-down probability 10 times higher than the fire-breakout probability of each building.
 Next, we attempt to evaluate the effects of extinguishing fires at buildings preferentially based on the Fire-spread potential under the condition when multiple fires break out simultaneously. In case that firefighters extinguish fires in order of the time reported by residents, in average more than 500 buildings might be burnt down in 61.0% of Tokyo Metropolitan area. By contrast, assuming that firefighters went to fires in descending priority order of the Fire-spread potential, the percentage of such areas could be reduced to 8.5%. These results suggest that the Fire-spread potential can be effective to identify high-risk buildings and decrease the number of burnt-down buildings.
 In the chaos immediately after an earthquake, information on the location of fires reported by residents may be sometimes imprecise. Such kind of information is likely to prevent firefighters from extinguishing fires at buildings with high fire-spread risk. Based on the simulation assuming that imprecise information on location of fires is randomly reported within 65 meters from the correct location, the number of buildings destroyed by fires increases to four times more than in the case correct information is reported.
 In order to reduce the influence of imprecise information on the location of fires, we propose that firefighters decide the priority of destinations not based on the Fire-spread potential of a reported building but on the average of the potential of a certain area around the building. In case the distance of a gap between the actual location of a fire and the reported location is 15 meters or more, the total number of burnt-down buildings can reduce by using the average Fire-spread potential of buildings within 75 meters of each reported building. However, in case the distance is less than 15 meters, it is more effective to use the Fire-spread potential of the reported building. These results suggest that it is important for firefighters to change the strategy for extinguishing fires based on the accuracy of the reported location.