Whilst personnel are required to be in and around hazardous locations the demand for effective and reliable audible and visual alarm notification will remain a primary concern for any life safety focussed system. However, field devices such as alarm horn sounders or warning beacons can unfortunately be specified using criteria that may not accurately reflect the necessary real-world level of performance. Out of sight and out of mind until that singular moment when they are needed, the risks are obvious. So, what are these criteria and why do they cause confusion.
Audible warning signals
The decibel unit is a precise measure of performance for audible warning signals but its relationship to power, distance and frequency can often mislead. One issue is an appreciation that an increase or decrease of 3dB represents a halving or doubling of the sound pressure level (SPL). However, in practice, a change of about 10dB is required before the sound subjectively appears to have reached these levels. The smallest change we can hear is about 3dB. As a dB is by definition 10x the log of the ratio of the powers: in units of dB = 10 x log(P1/P0), an output power reduction of -3 dB means that the ratio of output to input power is 10-3/10= 0.5 – as stated above the reduction of 3dB is a 50% drop in power. The main difficulty in relating to a dB scale is visualising that large differences in power are represented by relatively small increases in the dB level. For example, going from a level of 100 dB to 110dB equates to an increase in the SPL from ten billion times the threshold of hearing to one hundred billion times the threshold of hearing; an increase of +10dB does not convey the magnitude of the change.
The perceived sound level of an alarm tone is determined by several factors, including that the human ear is not equally sensitive to all frequencies, preferring a range of 2kHz and 5kHz. This difference in sensitivity to different frequencies is more pronounced at lower SPL’s than at higher SPL’s. For example, a continuous alarm tone at 50Hz would need to be 15dB higher than a 1kHz alarm tone at a level of 70dB in order to give the same subjective ‘loudness’. Additionally, an alarm tone with multiple frequencies and/or a temporal pattern will be more distinct from any ambient noise than a continuous tone. For this reason, ISO8201 temporal pattern alarm tones have been adopted by the International Marine Organization and more recently stipulated in NFPA 72 2016 which in turn correlates with ANSI S3.41, American National Standard Audible Emergency Evacuation Signal. The temporal pattern consists of an ‘On’ phase lasting 0.5 second, an ‘Off’ phase lasting 0.5 second for three successive ‘On’ periods followed by an ‘Off’ phase lasting 1.5 seconds.
To normalise the frequency dependence associated with human hearing a weighting is applied. The ‘A’ weighting network weights a signal in a manner which approximates to an inverted equal loudness contour at low SPL’s, the ‘B’ network corresponds to a contour at medium SPL’s and the ‘C’ network to an equal loudness contour at high SPL’s. The ‘A’ weighting network is the most widely used. SPL quoted in dB(A) at a distance of 1m can be converted to an estimate of effective distance as the SPL reduces by 6dB as the distance doubles. The effective distance of a 100dB(A) @ 1 metre alarm in an ambient of 65dB(A) is the distance at which the sounder output level reduces to 70 dB(A). So, a 100dB(A) @ 1m sounder has an effective area, assuming a uniform sound distribution radius of 32m at 70dB(A). However, the output from a 121dB(A) @ 1 metre sounder does not attenuate to 70dB(A) until approximately 300 metres from the source, giving the unit ten times the effective distance and, even more importantly, 100 times the coverage area (see fig 1).
By comparison to audible signals the measures of performance for visual signals are numerous and even more prone to misinterpretation. The Xenon strobe warning beacon has been the standard for visual warning signalling in hazardous locations, but the next generation of LED warning beacons are now increasingly becoming the preferred choice due to their low current draw, low in-rush and longer life – this has further complicated end users ability to accurately compare the effectiveness of devices.
The required brightness of the Xenon strobe has traditionally been specified in Joules; based upon on incorrect assumption that all strobes are equivalent in output. The number of Joules is a measure of energy input rather than a reliable indication of light output. A legacy visual warning beacon may be correctly referred to as a 21 Joule beacon yet only have an effective intensity of 355 candela whereas a contemporary 21 Joule beacon has an effective intensity of over 1,250 candela.
Unfortunately, the candela unit of measure is not a panacea when comparing beacon performance or designing warning systems due to the various use of peak, calculated and effective candela values. To ensure accurate data a spectrometer should be used for measuring the average effective luminous intensity of an entire beacon lens – which is then translated into an effective candela figure (cd). In the case of a flashing beacon such as a Xenon strobe the pulse duration is measured between the 10% of peak amplitude for the leading and trailing edges of the pulse. Light levels are collected during the pulse period, these are calculated using the Blondel-Rey formula. The measured effective candela (cd) is the intensity that would appear to an observer if / when the light was burning steadily. The measured peak candela is the maximum intensity measured generated by a flashing device during its light pulse.
The effective luminous intensity (Ieff), expressed in candela (cd), is calculated for each pulse measured using the following Blondel-Rey formula:
I(t) is the instantaneous value in candela (cd).
a = visual time constant. 0.2 (night-time) or 0.1 (daytime).
t2 – t1 is the light pulse duration as measured between the 10 % of peak amplitude for the leading and trailing edges of the pulse.
Therefore, when evaluating the output data of a visual signalling device, it is worth considering how the data has been established. Rules of thumb and calculations based on the energy of the flash tube within a Xenon strobe beacon as discussed above, have customarily been used to give an indication of effectiveness. However, when comparing light outputs derived by calculations based on energy alone, to measured outputs with a spectrometer or similar, the output is often overstated.
The ‘calculated’ light output data is based upon the assumptions that the effective candela can be found by allocating 50 cd for every 1 Joule of energy and for the peak candela by assuming that 1 Joule of energy supplied to a flash tube assimilates to 100,000 cd of peak candela. Compliance with legacy specifications requiring light outputs of over 2,000,000 cd has necessitated the continued use of ‘calculated’ values yet, as with audible signals this focus on the ‘headline number’ has detracted from the all-important measure of real world effectiveness. As an example, a legacy 5 Joule Xenon strobe beacon with a measured effective candela of 29 cd would have an effective warning distance of approximately 8m whereas the use of its calculated effective candela figure (5 Joule x 50cd = 250cd) would infer an effective warning distance of approximately 25m.
The relationship between effective candela and the range of the visual signal can be separated into the distance where the beacon can will provide adequate warning and that where the beacon can be considered as offering an effective warning. The following formula can be used to convert effective candela into effective warning distance, in other words, to alert rather than inform:
Ieff(av) = Effective Candela
d = Distance (m)
The formula below may be used to convert effective candela into adequate warning distance or range, based on normal visibility in daytime conditions.
Ieff(av) = Effective Candela
d = Distance (feet)
Lb = Foot-Lamberts background illuminance (normal day time conditions, Lb = 2919 ft-L)
From the above two formulas the table below gives an indication of both warning distance and range of a visual signal given an effective candela measurement.
The future
Life safety system designs decisions are increasingly focussed on the direct correlation between measured performance and effectiveness of audible and visual signals. The rising demand for compliance to test standards such as UL464, UL1638 and UL1971 referenced by NFPA 72, as well as EN54-3 and EN54-23 from the CPR directive, have highlighted the need to control the efficacy of notification devices. Fortunately for designers the next generation of signals are solving the contradictory requirements of greater sound and light output for less current over an increasing temperature range. High output LED’s are beginning to offer true low power replacements for the traditional Xenon strobe whilst features such as automatic synchronisation are reducing installation and system costs. However, diligence is still very much required when considering performance data.
For more information, go to www.e2s.com
About the Author
Brett Isard is the Managing Director of E2S Warning Signals. Involved with notification devices and life safety solutions for over 30 years – initially with Clifford & Snell prior to the launch of E2S Warning Signals in 1992.
