建築物火災財物損失影響因素與解釋模式Affecting Factors and Explanatory Model of Property Losses in Building Fires

無庸置疑的，建築物消防安全防護的目標除了要確保人命安全外，也要能保護財產。本文最主要的目的，是嘗試從建築物火災案例中，找出影響財物損失的重要因素及其影響權重，並據此研擬適當之對策，以減少火災時的財物損失。
本研究針對2000 年1月至2002 年2月間台灣地區曾經發生火災的建築物，以台北市、台北縣、桃園縣、新竹縣、台中縣、台南縣(市)及高雄市為主要調查地區，藉由災戶的調查及官方的火災紀錄資料，以火災發生及影響時序為架構，測量可能影響火災財物損失的因素。蒐集的資料有住戶特性、建築物空間特性、消防安全設備設施、初期反應、火勢發展與延燒情況、消防搶救及火災損失等資料。運用因素分析、相關分析及迴歸分析等統計方法，概念化及歸納出影響火災損失的重要因素，經檢驗後達顯著影響的因素有建築物結構、建築物延燒及燒毀程度、火災猛烈度、水源維護等，研究中並分析這些因素對國內建築物火災財物損失的相對影響權重。
關建字：延燒及燒毀程度、火災猛烈度、水源維護、財物損失、迴歸模型

Undoubtedly, the main goal of building fire protection is not only to ensure life safety but also to protect property. The main purpose of this study attempts to examine the important affecting factors, which contribute property loss in building fires from real fire cases, to determine relative weights of these effects and then to provide suitable strategy in order to mitigate property losses.This study focuses on the measurement of possible affecting factors on fire property losses by investigating both from real cases and official reports of building fires during January 2000 to February 2002 in Taiwan including Taipei city, Taipei county, Taoyuan county, Hsinchu county, Taichung county, Tainan county, Tainan city and Kaohsiung city. Based on the time sequence of fire occurrence and effect possible influencing factors will be measured. Contexts of investigations include occupant characteristics, building occupancy characteristics, fire protection systems, early reaction, fire development and spread, fire brigade interventions and the resulting property losses. Using statistical methods such as factor analysis, correlation analysis and regression model important factors are conceptualized and conducted. After testing significant influencing factors such as building structure, degree of fire spread and damage, fire severity, maintenance of water resource in fire fighting are examined. Relative effects among these factors for property losses are also analyzed. 
Key words: Degree of fire spread and damage, Fire severity, Fire water maintenance, Property loss, Regression model