In this study, a probabilistic model of spotting fires by firebrands was developed. When a large earthquake hits a city, it is conceivable that multiple urban fires may simultaneously break out and spread involving extensive urban area. In order to implement effective fire safety measures in the regional disaster prevention plan for reducing the damage by post-earthquake fires, it is essential to develop a simulation model which can predict fire spread behavior involving broad area of a city reasonably. Among the mechanisms of urban fire spread, fire spotting is the most difficult to develop a prediction model in spite of its significance in fire spread. In this paper, the fire spread in the Sakata City Fire in 1976 was simulated by using Monte Carlo method for validating the proposed model of spotting fires, where the proposed model was incorporated into an existing physics-based model for urban fire spread. The obtained results showed that estimated macroscopic fire spread behavior such as number of spotting fires roughly corresponded with the survey report of that time.
In the standard verification method for fire resistance of Building Standards Law of Japan, the temperature of fully developed fire is calculated by a simple analytical formula considering fire load density, constituent materials of the room and opening geometry. The calculation results are expressed by two parameters, fire temperature rise coefficient and fire duration. Even though the accuracy of fire temperature calculation is important for the fire resistance design of elements of construction, the accuracy of the calculation formula is not quantified yet. In this study, the accuracy of the calculation formula was analyzed by comparing with existing full-scale experiments. The errors in fire temperature rise coefficient and fire duration were quantified. As a result, it has been shown that the calculation formula tends to underestimate the fire temperature rise coefficient, while the fire duration is overestimated. In terms of equivalent fire duration, the formula seems to give slightly optimistic results in average but with considerable scatter.
The full scale tests were carried out to grasp behavior of water film supplied by water discharge system under heating conditions. The specimen was a steel plate of 2438 mm height, 1219 mm width and 6 mm thickness. The intermediate furnace at Tokyo University of Science was used as radiation panel. The radiation panel was placed parallel to the specimen at some distance. The experimental conditions have changed the heat flux rate from radiation panel to specimen covered with water film, the water supply rate gushed from water discharge head and so on. The water film was almost uniform from the top of the specimen to the bottom by using this water discharge system. As the results of experiments, the behavior of water film is clearly involved to the water supply rate and the relation could be marshaled by using Reynolds number based on a simple viscous layer model. The water film temperature became higher so that the water supply rate decreases. Based on the experiment results the empirical formula has been developed to calculate the temperature-rise of water under any heating conditions.