In the UK there are no strict mandatory replacement periods, within British Standards, for smoke alarms and detectors. The aim of this research was to identify the optimum replacement periods for optical smoke alarms and detectors used in domestic and commercial environments, respectively.
Background
The timely response of smoke alarms and detectors during a fire will provide sufficient warning to occupants, allowing them time to intervene or escape. These devices are expected to deteriorate with time and may operate later than expected. Whilst manufacturers provide recommended replacement periods there are no replacement periods proposed within the British Standards BS 5839-1 (Code of practice for non-domestic premises) and BS 5839-6 (Code of practice for domestic premises). Whilst some countries do have replacement periods for optical smoke alarms and detectors, a literature review revealed that there appears to be no independent evidence to support these.
This research aimed to identify any changes in performance and propose suitable replacement periods by measuring the response of a number of smoke alarms and detectors, from the field, in domestic and commercial environments. This work focused on optical smoke devices and no ionisation devices were included in the study.
A stakeholder group comprising the Fire Industry Association, Scottish Government, Scottish Fire and Rescue Service, Detector Testers and the BRE Trust collectively funded and steered the research work, which was performed by BRE Global.
Method
The Trutest (see Figure 1), was selected to be used for this research work as it was portable, easy to assemble, produced a gradual increase in aerosol concentration and gave a measurement value at the time of response. It also proved to have a test volume large enough to accommodate all the measured smoke alarms and detectors. The growth rate used for the tests was 0.5% obs./ft per minute up to a maximum measurable aerosol concentration of 6% obs./ft.
To quantitatively identify acceptable performance limits, ten new approved domestic optical smoke alarms and ten new approved commercial optical smoke detectors, from different manufacturers, were tested using the Trutest in the BRE Global labs.
Using the results of the measured devices, acceptable performance limits were proposed by identifying the range between the least and most sensitive responses. The limits for both smoke alarms and detectors were determined as 1.4–3.8 % obs./ft and devices tested subsequently in the field were expected to respond within these limits.
When testing the commercial smoke detectors, the silicone membrane provided with the Trutest was placed over the mouth of the testing chamber which formed a seal with the head of the detector (as can be seen in Figure 1). To allow for the testing of domestic smoke alarms in-situ, the Trutest was adjusted by removing the membrane and placing sticky foam along the inside of the testing chamber which formed a seal when it was pressed up against the ceiling.
A subjective ‘cleanliness’ score was given for each device from a 5-point scale (1 = very clean, 5 = very dirty) to identify whether there was any correlation of cleanliness with smoke sensitivity response.
Some commercial detectors contain drift compensation, which is an algorithm used to automatically adjust the sensitivity of a smoke detector over its lifetime to accommodate for changes, such as the accumulation of dust, to ensure consistent alarm response. Another feature of detectors is that they can be addressable systems that provide status information about individual detectors or conventional systems that only provide status information about the zone or loop. Those detectors containing drift compensation, and whether they were addressable or conventional, were noted to allow for comparisons in performance between them to be made.
Responses of domestic smoke alarms
During the period from February 2019 to January 2020, a total of 85 domestic smoke alarms were tested; comprising 23 different alarm models from eight different manufacturers and their age range was 0–12 years.
Figure 2 shows all of the results and the blue line represents the average response result for the alarms tested of a specific age. The dotted orange line represents the line of best fit of the results for all of the alarms tested. The solid horizontal red lines on the graphs indicate the upper and lower limits previously determined. Four alarm responses (5%) were outside the lab-based limits for domestic optical smoke alarms.
Responses above the top red line would be expected to give a later response during a fire (low sensitivity) whereas those below the bottom red line correspond to high sensitivity and these may be more likely to produce false alarms.
The line of best fit, shows that for older alarms their sensitivity slightly increases (i.e. it takes less smoke for the device to operate) with age but remains far from the lower limit. If the line of best fit was linearly extended it would intercept the lower limit at 21 years.
It was recommended that all smoke alarms should be replaced no later than 12 years after their date of manufacture (which is normally shown on the device). Some recommendations were also made for immediate replacement under certain specific circumstances.
By comparing the smoke sensitivity responses with the cleanliness scores there was a clear correlation that dirtier smoke alarms responded sooner, indicating that they were getting more sensitive as they got more contaminated.
Responses of commercial smoke detectors
During the period from April to December 2019, a total of 107 commercial smoke detectors were tested; comprising 20 different detector models from nine different manufacturers and their age range was 0–30 years. The optical smoke detectors tested were from office spaces, university premises and hotels.
Figure 3 shows all of the results and the blue line represents the average response result for the detectors tested of a specific age. The dotted straight orange line represents the line of best fit of the results for all of the smoke detectors tested. The number in brackets, in the legend, indicates the number of associated detectors.
As before, responses above the top red line indicate a delayed response to fire and those below the bottom red line indicate a quicker response. Again, the line of best fit shows that for older detectors their sensitivity slightly increases with increasing age. There were 14 detectors out of the 107 detectors tested (13%) that had a response outside the lab-based limits. Some of those were quite close to the limits, but there was no correlation of devices falling outside the limits with age.
Figure 3 also shows the results with the optical smoke detectors tested (addressable, conventional and those which had drift compensation). In terms of the different types of detectors, those that were addressable without drift compensation most consistently remained within the limits. The conventional detectors, especially as they got older, had the highest failure rate.
The results demonstrate that there is a 0% failure of smoke detectors up to 10 years, so there is no justification for replacing them before then. On the basis of discontinuity in the data for detectors, without drift compensation that are 15–20 years old, those in clean environments (such as circulation spaces) should be replaced at 15 years. However, it may be appropriate to reduce this period in dirtier environments, such as utility rooms and loft spaces. The maximum time for the replacement of detectors without drift compensation should be 25 years, after which the system should be regarded as non-compliant with BS 5839-1. However, the maximum time for the replacement of detectors with drift compensation is 30 years, after which the system should be regarded as non-compliant with BS 5839-1. Prior to 30 years, it is considered that reliance can be placed on the initiation of a fault warning (when further compensation for drift is impossible) to indicate the need for detector replacement.
In terms of sensitivity response with the cleanliness of the smoke detectors, it appears that as they get dirtier, they appear to get more sensitive. This behaviour was also observed with smoke alarms.
Further analysis was performed on the smoke detectors and alarms tested to determine the response range of individual models, but these are not reported here.
Recommendations
As well as the proposed replacement periods for smoke alarms and detectors, four recommendations resulted from this work, which included the mandatory labelling of the installation date on these devices. Also, the periodic on-site measurements of smoke detector response was recommended in areas to potentially reduce the likelihood of disruption (e.g. airports, shopping centres).
During measurements on-site, dust and oil particulates build-up was observed on the thermal element of heat alarms located in kitchens. As it was suspected that this build-up could reduce the sensitivity of heat alarms a research programme investigating this was recommended and is currently in progress. Any organisations interested in collaborating with this work should contact BRE directly.
Conclusions
In total 107 commercial smoke detectors (20 models) and 85 working domestic smoke alarms (23 models) were tested. This sample size was smaller than anticipated and therefore a recommendation is made for further work which would provide greater confidence in the results. Both smoke alarms and detectors demonstrated a slight increase in sensitivity with time.
It was recommended that all smoke alarms should be replaced no later than 12 years after their date of manufacture.
The replacement period proposed for smoke detectors without drift compensation in clean environments is 15 years and with a maximum replacement period of 25 years. The proposed maximum replacement period for detectors with drift compensation in clean environments is 30 years.
The recommendations resulting from this work are expected to be included during the next revisions of BS 5839-1 and BS 5839-6.
For more information, go to https://www.bregroup.com/firesafetyresearch
