The Efficacy of Duct Smoke Detection
Share |

The Efficacy of Duct Smoke Detection

Fire Detection Institute | Fire Protection Engineering

Does duct smoke detection really work? Is it worth the cost? Is the potential for false and nuisance alarms worth the added protection? The universal answer to all fire protection questions is: It depends. This is the first part of a two-part article that summarizes the findings of research recently conducted under the auspices of the Fire Detection Institute. This article summarizes the purpose of the research and findings in the areas of smoke-driving forces, smoke dilution in ductwork, and the effects of smoke aging on detection in ductwork. Part 2, which will appear in the Spring 2006 issue, will summarize findings regarding the effects of HVAC filters on smoke detection in ducts, smoke stratification in ducts, and the efficacy of duct detectors that use sampling tubes.

Building, fire, and life safety codes often require the installation of duct smoke detectors in heating, ventilating, and air conditioning (HVAC) ducts for the detection of fires in the HVAC system. Typically, these codes refer to NFPA 90A, Standard for the Installation of Air-Conditioning and Ventilating Systems,1to determine if and where they are required. NFPA 90A in turn refers to the National Fire Alarm Code2 for specific design, installation, and maintenance criteria for these detectors. There is no known research or data analysis that supports the prescriptive criteria in the National Fire Codes. Historically, these criteria reflect the engineering judgment of experienced practitioners in the field.


In an effort to support the committees and system designers, the Fire Detection Institute undertook a research program to study duct smoke detection and to provide technical data to the committees and system designers. The Fire Detection Institute Duct Detection Research Initiative (FDI/DDRI) employed the research assets of the Department of Fire Protection Engineering of the University of Maryland (UMDFPE) and the National Research Council Canada (NRC). The goals were to either support, change, or eliminate duct smoke detection requirements from the codes and standards, and to study the many myths about duct smoke detection. That research program has been completed, and reports have been issued. This article is a short summary of that work.


The issues identified for research related to both the use of duct smoke detectors for detection of fires within the HVAC system and as part of a larger smoke management system. Six issues were identified for investigation:

  1. Comparative Driving Forces (fire vs. fans)
  2. Dilution Effects
  3. Effects of Smoke Aging and Condensation within the HVAC System on the Performance of Duct Smoke Detectors
  4. Effects of HVAC Filters on the Performance of Duct Smoke Detectors
  5. Stratified Flow in HVAC Ducts (effect on detector or sampling tube location)
  6. Efficacy of Duct Detectors using Sampling Tubes

There are two principle uses of duct smoke detection that had to be considered in the research plan. The first was detection of smoke resulting from a fire in the building served by the HVAC system. That is, a fire external to the HVAC system. The second principle use of duct smoke detection is detection of smoke originating in the HVAC system, such as a filter fire, or smoke originating outside the building in close proximity to a fresh-air inlet.


The research included surveys of existing buildings and HVAC systems; fire, smoke, and air movement modeling; and full-scale fire testing. Modeling done by the UMDFPE team utilized a hypothetical 10-story building. The analytical methods they used and reported on are valid for smaller and larger buildings. The NRC team ran full-scale tests in a real 10-story building.


Comparative Driving Forces
The test hypothesis was that it was unnecessary to shut down the HVAC system because the driving forces due to the fire were large compared to those of the mechanical ventilation. There was a concern that the pressurization produced by the fire constituted a driving force on the air in the HVAC system ductwork that was large compared to that produced by the HVAC system fan(s). If this were the case, the fire would overpower the HVAC system, rendering the interactive control of the HVAC fans superfluous. This suggested that considerable expense was being incurred for the installation and maintenance of duct smoke detectors that did not provide a commensurate benefit.


In addition to the movement of smoke caused by buoyancy, the research also addressed air movement due to stack effect and wind.


The results indicate that the pressures produced by the HVAC system were generally larger than those produced by the fire. Since the HVAC system produced higher pressures, it also produced higher flows and was more effective in distributing smoke to the nonfire floors than the passive buoyancy, stack, and wind effects.


Consequently, permitting the HVAC system to continue to run normally during a fire will generally result in higher smoke concentrations on nonfire floors and in nonfire compartments than if the mechanical ventilation system is shut down, allowing smoke to flow subject only to the pressures resulting from the passive buoyancy, stack, and wind effects. This suggests that, unless an active smoke management strategy is in place, the HVAC systems should be shut down upon detection of a fire. The extent of this advantage depends upon the specific characteristics of the building in question, and it can be quantified by using the computational methods outlined in the UMDFPE report.


The analysis of the comparative driving forces demonstrated that the requirements for duct smoke detectors in NFPA 90A do have a technical basis. Consequently, these requirements should remain as part of the minimum compliance prescriptive standard. Where the design fire is substantially greater than the nominal 100 kW to 500 kW fire, or where the construction is substantially atypical, a performance-based analysis can be performed using the UMDFPE report as an outline guide.


Dilution Effects
One concern is whether the smoke concentration in the HVAC ducts would be too low to produce reliable detection in response to the fires they are generally expected to detect. The second concern was whether the dilution of the smoke by fresh air in the ductwork is so severe that duct smoke detection is only effective in responding to fires that would be characterized as a "major involvement."3


The UMDFPE team conducted laboratory tests using a medium-scale cone calorimeter and duct system, and developed analytical relationships. The NRC team conducted full-scale tests in the 10-story test building. The data from the NRC experiments support the relations developed by UMDFPE. As expected, the dilution of the smoke was proportional to the ratio of the air flows from the fire room (floor) and the nonfire rooms (floors). The NRC tests also demonstrated that, even with substantial dilution ratios, the commercially available duct smoke detectors were able to detect fires on the order of 100 kW to 200 kW. Thus, while dilution is important, it does not obviate the ability of commercially available duct-type smoke detectors to perform as intended.


Smoke Aging Effects
As smoke moves away from a fire source, it cools and changes characteristics. Condensation, agglomeration, and coagulation result in changes to smoke particle size, shape, and color. Soot deposition on surrounding surfaces reduces the amount of smoke in the air stream. These changes affect the response of commercial smoke detectors.


The UMDFPE team studied smoke aging using the duct on the mediumscale cone calorimeter and a laser sheet apparatus to measure smoke particle size and number. That work showed that the number of small particles at the duct inlet, compared to 3 m downstream, decreased by a factor of 10 while the number of large particles increased by a factor of two. This suggests that the aging of smoke occurs more rapidly than might have been expected. For all but two test scenarios,the NRC tests in the full-scale building showed very little difference in the response of analog smoke detectors that were in the ductwork and separated by 10 stories. The tests that varied allowed the smoke to recirculate through the ductwork for an extended period. After about 12 minutes, there was some change in detector response. Those tests would not be representative of the real use of duct smoke detection.


Thus, it appears that the aging of smoke in ductwork occurs predominantly in the first few meters. Therefore, duct smoke detectors that are located some distance from the source of smoke are not adversely affected by smoke aging. If the duct smoke detectors are of the photoelectric type, the response may be improved since the detector response is proportional to the particle diameter raised to the 2nd to 6th power, but only linearly proportional to the number of particles.4 However, this expected increase in detector signal may be offset by soot deposition on the duct walls. This may explain the lack of difference in detector signal in the NRC results.



  1. NFPA 90A, Standard for the Installation of Air-Conditioning and Ventilating Systems, National Fire Protection Association, Quincy, MA, 2002.
  2. NFPA 72, National Fire Alarm Code, National Fire Protection Association, Quincy, MA, 2002.
  3. Mnizewski, K., and Waterman, T.E., "The Effectiveness of Duct-Installed Smoke Detectors in Two Different Ventilation System Configurations," IITRI Report for the GAS, February 1979.
  4. Schifiliti, R.P., and Pucci, W.E., Fire Detection Modeling: State of the Art, the Fire Detection Institute, Bloomfield, CT, 1996.

Editor's Note About This Article

This is a continuing series of articles that is supported by the National Electrical Manufacturer's Association (NEMA), Signaling Protection and Communications Section, and is intended to provide fire alarm industry-related information to members of the fire protection engineering profession.


The Fire Detection Institute is a nonprofit corporation dedicated to improving the fire detection and alarm standards through testing programs and research. The institute has sponsored research that led to the development of heat and smoke detection modeling and research on the effects of ceiling shape and geometry, and the effects of HVAC on the response of room detectors. The Duct Smoke Detection research project was sponsored by contributions and grants from foundations, manufacturers, and interested parties. Copies of a more complete summary are available through the Automatic Fire Alarm Association at The full reports for both teams and a full set of all data collected are available from the Society of Fire Protection Engineers through their online store at For information about the Fire Detection Institute, contact the FDI Secretary at
© SFPE® | All Rights Reserved
Privacy Policy