International Journal on Engineering Performance-Based Fire Codes, Number l, p.21-23, 2010 21 REVIEW OF FIRE DETECTION PROBLEM N.K. Fong Department of Building Services Engineering, The Hong Kong Polytechnic University, Hong Kong, China ABSTRACT In order to protect life and property, a good fire and smoke detection system is necessary. However, with existing automatic fire and smoke detection system design, the major concern is false problem. In the present review, factors leading to false alarm problems are identified and remedial methods are suggested. Based from current literature survey, a number of research results for reducing false alarms such as multi-sensor technique, the use of neural networks are identified. This formed the basis for developing new fire and smoke detection techniques and would be useful for future fire protection. 1. INTRODUCTION Fire and smoke spread within the building can be affected by various factors such as the geometry, dimension, layout and usage of the building. In order to provide fire protection in the building, it is very important to detect fire at its early stage. The most common fire and smoke detection methods include the use of point type detectors (i.e. ionization smoke detectors, photoelectric detectors, heat detectors), line type detectors etc. These detection methods based on the use of fire signatures such smoke, heat. However, these detection methods have some significant drawbacks including the false alarm problem and delay in smoke detection. False alarm will dilute and discredit the valid fire intelligence. If the detection system is connected to other fire services installations such as fire shutters and smoke control systems, this may also lead to business interruption. These false alarms will also lead to the waste of fire brigade resources and reduction of the effectiveness of the system in the case of real fire. The worst of all is cry wolf syndrome. Occupants will tend to ignore the fire alarm until the fire has becoming highly destructive and life threatening. 2. LITERATURE SURVEY ON FALSE ALARM STUDIES Various countries have conducted studies attempted to identify the causes of false alarm problem and suggested remedial solutions. In U.K., a BRE information paper was published to address to this problem. Based on the BRE information paper (IP13/92) June 92, the false calls to real fires in U.K. reported in 1970 is 11:1 and the false calls to real fires in U.K. reported in 1980 is 20:1. In year 2002, British Standard 5839-1 was published to provide guideline for the design of fire detection and fire detection system in buildings. Based on BS 5839-1:2002, the no. of false alarms attended by the fire services in U.K. was over 250,000. Besides U.K., U.S. also conducted similar study. In a 1980 survey of health care facilities in the U.S., Bukowski et al. reported nuisance alarm ratios for smoke detectors at 14:1. Similar records can be found in H.K. Statistical records of fire alarms have been collected from FSD 1975 to 1995. The number of fire alarms rose from 6238 in 1975 to 31014 in 1995. The frequency of unwanted fire alarms rose from 1471 in 1975 to 18277 in 1995 and the no. of fire calls in 1997 is 35543. The number of fire calls increased to ~44000 in year 2000. The number of unwanted fire alarms per day increased from 4.03 in 1975 to 50.07 in 1995. The number of unwanted calls for 2001 to 2003 is listed as follows:  44564 fire calls were received in 2001 and ~29412 were unwanted alarm calls.  41204 fire calls were received in 2002 and ~27565 were unwanted alarm calls.  37774 fire calls were received in 2003 and ~24439 were unwanted alarm calls. In order to tackle the false problem, an exercise was conducted by the British Fire Protection Systems Association and the Suffork Fire Services in 1988. They classified the fire signals into genuine fire signal, unwanted fire signal, accidental fire signal, malicious fire signal, malfunction fire signal, unidentified fire signal Based on the records of the Hertfordshire Fire and Rescue Services, FRS analysed 4000 false alarm report. The nature of false alarm was highlighted as follows:  Human factor – 22 % caused by human 24% of these calls - actuation by manual call points 13% - accidental operation