- Author
- Namba, Y. | Yasuno, K. | Omori, T.
- Title
- Basic Study on Forecast of Ordinary Building Fire.
- Coporate
- Kinki Univ., Hiroshima, Japan
- Journal
- Bulletin of Japanese Association of Fire Science and Engineering, Vol. 42, No. 1, 1-12, 1994
- Keywords
- building fires
- Identifiers
- Japan
- Abstract
- [ABSTRACT IN ENGLISH] The quantitative ratio of building fire is relatively high in Japan and especially the number of deaths in fire of house shows a increasing tendency lately. On the other hand, there are some studies on the forecast of seismic fire for urban conflagration risk but there are very few cases of the forecast of ordinary fire. And a lot of fire and deaths break out every year. Then it is important to study on forecast of ordinary building fire. This is the basic study for fire fighting planning and we are trying to analyze statistically the better forecast model by using data to be obtained easily. The number of fire is assumed to be a function of one or more independent variables. So the method of this study is first to classify the independent variablesinto small groups according to urban characteristics by using Factor Analysis, secondly to find the best model by Multi-Regression Analysis. Judging of the model adaptation is mainly used AIC, i.e., Akaike's Information Criteria (Akaike, 1977), which is the better evaluation method in the recent statistics. Four factors were drawed for urban characteristic data by Factory Analysis. But one of them is not connected with the cause of fire. Then the data were diveded into three groups. It was suggested that first is activity group, and that second is life one, and that third is industrial one. Industrial factor were considered such as districts were classified into three groups, i.e., commercial district, manufacturing one and residential one, according to land use ratio. Population and household were selected from life group. Quantity of consumed electric power, quantity of consumed gas and quantity of consumed water were selected from activity one. By considering these three factors, six models were made and examined. Then it was showed that [model3/population, quantity of consumed water] and [model 6/household, quantity of consumed water] are adapted by Multi-Regression Analysis and AIC from six models. Because population and household selected from life group and quantity of consumed water selected from activity one are able to obtain easily, it is seemed that these models can be analyzed and used in the other cities and towns.