Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Detector environment and detector response : a survey

Holmstedt, Göran LU ; Magnusson, Sven Erik LU and Thomas, Philip H (1987) In LUTVDG/TVBB--3039--SE
Abstract
1. The survey has mainly concentrated on the following items: the false alarm problem, the problem of the fire not being detected due to the fact that pre-fire heating and ventilation dominate flow inside the compartment, a description of detector sensitivity to fire signatures. engineering design methods for the siting of detectors.

2. The statistical as well as practical experience suggests that alarm systems in Sweden, follow international trends regarding rates of false alarms.

3. For existing systems, especially the older ones, introduction of a proper alarm organisation, regular training of the technical maintenance staff and/or desensitization may be appropriate measures to reduce the false alarm rate.

4.... (More)
1. The survey has mainly concentrated on the following items: the false alarm problem, the problem of the fire not being detected due to the fact that pre-fire heating and ventilation dominate flow inside the compartment, a description of detector sensitivity to fire signatures. engineering design methods for the siting of detectors.

2. The statistical as well as practical experience suggests that alarm systems in Sweden, follow international trends regarding rates of false alarms.

3. For existing systems, especially the older ones, introduction of a proper alarm organisation, regular training of the technical maintenance staff and/or desensitization may be appropriate measures to reduce the false alarm rate.

4. Analysis shows that desensitization of existing systems must be done with care so life hazard is still kept under control.

5. In addition to the remedies mentioned under point 3 it might be effective to revise the siting standard rules in RUS 110 in the light of the information obtained for example in the Norwegian investigation.

6. A revision of RUS 110 should also be based on scaled down physical modelling and the use of numerical modelling (field modelling) in order to investigate local flow field characteristics: the influence of beamed ceilings, ventilation in and outflow. a thermally stratified pre-fire atmosphere, etc.

7. In order to take consideration of various fire developments rates and varying ceiling heights, standard siting distance should be calculated according to the design model described in NFPA 72E Appendix C for heat detectors.

8. For smoke detectors, a more rational use of the information obtained in the EN54/9 sensitivity test standard fires should be investigated.

9. Point 3 mentioned existing systems. For new and future systems the gradual improvement in software control (alarm verification procedures. automatic compensation for regular variation in ambient conditions, automatic change of sensitivity to compensate for the gradual accumulation of dust and dirt, filtering of electronic signal spikes, checking of electric circuit integrity etc) is thought to dramatically reduce false alarm rates.

10. Efficient pattern-recognition algorithms for separating fire process environments from the aerosols produced by welding, trucks exhaust, smoking etc may be hard to produce: the

aerosol characteristics are too much alike. Algorithms based on use of data from a group of sensors ("group decision") should be more reliable but may be too complex. In theory, if sensor response to monodisperse aerosols is known and with standard test fire aerosol characteristics experimentally measured, sensor output during the standard test fires should be easily calculable and a first check obtained on the discrimatory capability of pattern recognition algorithms. No information of this kind has been found. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Book/Report
publication status
published
subject
in
LUTVDG/TVBB--3039--SE
pages
77 pages
publisher
Department of Fire Safety Engineering and Systems Safety, Lund University
report number
3039
ISSN
0282-3756
language
English
LU publication?
yes
id
5b1ca2d2-bbd1-4412-8d1d-56c2582f5a60 (old id 1267856)
date added to LUP
2016-04-01 17:10:03
date last changed
2018-11-21 20:47:11
@techreport{5b1ca2d2-bbd1-4412-8d1d-56c2582f5a60,
  abstract     = {{1. The survey has mainly concentrated on the following items: the false alarm problem, the problem of the fire not being detected due to the fact that pre-fire heating and ventilation dominate flow inside the compartment, a description of detector sensitivity to fire signatures. engineering design methods for the siting of detectors.<br/><br>
2. The statistical as well as practical experience suggests that alarm systems in Sweden, follow international trends regarding rates of false alarms.<br/><br>
3. For existing systems, especially the older ones, introduction of a proper alarm organisation, regular training of the technical maintenance staff and/or desensitization may be appropriate measures to reduce the false alarm rate.<br/><br>
4. Analysis shows that desensitization of existing systems must be done with care so life hazard is still kept under control.<br/><br>
5. In addition to the remedies mentioned under point 3 it might be effective to revise the siting standard rules in RUS 110 in the light of the information obtained for example in the Norwegian investigation.<br/><br>
6. A revision of RUS 110 should also be based on scaled down physical modelling and the use of numerical modelling (field modelling) in order to investigate local flow field characteristics: the influence of beamed ceilings, ventilation in and outflow. a thermally stratified pre-fire atmosphere, etc.<br/><br>
7. In order to take consideration of various fire developments rates and varying ceiling heights, standard siting distance should be calculated according to the design model described in NFPA 72E Appendix C for heat detectors.<br/><br>
8. For smoke detectors, a more rational use of the information obtained in the EN54/9 sensitivity test standard fires should be investigated.<br/><br>
9. Point 3 mentioned existing systems. For new and future systems the gradual improvement in software control (alarm verification procedures. automatic compensation for regular variation in ambient conditions, automatic change of sensitivity to compensate for the gradual accumulation of dust and dirt, filtering of electronic signal spikes, checking of electric circuit integrity etc) is thought to dramatically reduce false alarm rates.<br/><br>
10. Efficient pattern-recognition algorithms for separating fire process environments from the aerosols produced by welding, trucks exhaust, smoking etc may be hard to produce: the<br/><br>
aerosol characteristics are too much alike. Algorithms based on use of data from a group of sensors ("group decision") should be more reliable but may be too complex. In theory, if sensor response to monodisperse aerosols is known and with standard test fire aerosol characteristics experimentally measured, sensor output during the standard test fires should be easily calculable and a first check obtained on the discrimatory capability of pattern recognition algorithms. No information of this kind has been found.}},
  author       = {{Holmstedt, Göran and Magnusson, Sven Erik and Thomas, Philip H}},
  institution  = {{Department of Fire Safety Engineering and Systems Safety, Lund University}},
  issn         = {{0282-3756}},
  language     = {{eng}},
  number       = {{3039}},
  series       = {{LUTVDG/TVBB--3039--SE}},
  title        = {{Detector environment and detector response : a survey}},
  url          = {{https://lup.lub.lu.se/search/files/4895418/4450372.pdf}},
  year         = {{1987}},
}