BS5839 - Functional testing

[Colin Newman]

Reading 45.4 of the BS has caused a liitle confusion.

45.4b) states "....Thereafter, every detector should be functionally tested.  The tests used need prove only that the detectors are connected to the system, are operational and are capable of responding to the phenomena they are designed to detect."
45.4c) states "Every heat detector should be functionally tested by means of a suitable heat source, .......", but
45.4d) states "Point smoke detectors should be functionally tested by a method that confirms that smoke can enter the detector chamber and produce a fire alarm signal...."

Does this mean that if the control panel is capable of showing the fluctuations in ambient conditions i.e. temperature fluctuations for heat sensors and airborne particulate fluctuations for smoke detectors that 45.4b) is satisfied?

When testing heat detectors to conform to 45.4c) is it unnecessary to stimulate the heat detector until it provides an alarm response if a response to the applied heat can be verified at the control panel?

Do the recommendations of 45.4d) override 45.4b)?  Why does there appear to be a difference between the required responses when testing of heat and smoke detecors?
Whilst I'm at it has anyone had a functional test fail for an analogue, addressable detector when the panel has not previously registered a fault condition?

[David Rooney]

45.4b) states "....Thereafter, every detector should be functionally tested.  The tests used need prove only that the detectors are connected to the system, are operational and are capable of responding to the phenomena they are designed to detect."
45.4c) states "Every heat detector should be functionally tested by means of a suitable heat source, .......", but
45.4d) states "Point smoke detectors should be functionally tested by a method that confirms that smoke can enter the detector chamber and produce a fire alarm signal...."

Does this mean that if the control panel is capable of showing the fluctuations in ambient conditions i.e. temperature fluctuations for heat sensors and airborne particulate fluctuations for smoke detectors that 45.4b) is satisfied?

No.... detectors need to be "functionally" tested to prove a particulate entering the sensing chamber (or a rise in temperature is applied to heat detectors) will cause a response (usually sounding the alarms but not necessarily so ..... depending on the agreed strategy and cause and effects)


When testing heat detectors to conform to 45.4c) is it unnecessary to stimulate the heat detector until it provides an alarm response if a response to the applied heat can be verified at the control panel?

Yes, it is necessary. See above and 45.4b "every detector should be functionally tested........."

Do the recommendations of 45.4d) override 45.4b)?  Why does there appear to be a difference between the required responses when testing of heat and smoke detecors?
Whilst I'm at it has anyone had a functional test fail for an analogue, addressable detector when the panel has not previously registered a fault condition?

Don't understand this last point, can't see the conflict. Every detector needs to be functionally tested by application of the phenomenon it is designed to detect......

[Colin Newman]

....detectors need to be "functionally" tested to prove a particulate entering the sensing chamber (or a rise in temperature is applied to heat detectors) will cause a response ....

Since analogue detectors merely relay a value corresponding to the level of their monitored stimuli to the control panel, surely being able to see a real time fluctuation in the analogue value demonstrates a response. If the purppose is to test that the fire detection is capable of a response to "fire-like" stimuli then surely it's been achieved. 

If the purpose is to demonstrate that the detector is capable of transmitting an analogue value to the panel which corresponds to an alarm condition, why doesn't 45.4c) state that the test should provide an alarm response?

The conflict I refer to is that 45.4d) requires an alarm response, but 45.4c) doesn't.

[David Rooney]

But I think the point being missed is that the idea of these tests is to actually test the sensing element of the detector - whatever it may be - so that an engineer can't just walk around with a magnet or similar, or artificially ramp up the analogue values by playing with the control equipment - possibly via a modem -  and say he has met the criteria.

To me it's a common sense approach you need to test each detector using the phenomena they are designed to detect, and you want to check there is a response to this phenomena - the normal response being a programmed output controlled by the CIE, possibly sounders but equally an out put that should get tested annually under the cause and effect testing.

Yes you can watch analogue values rise and go beyond the standard fire threshold eg 55 on an Apollo protocol system but this value would vary on a Hochiki system that calibrates fire thresholds daily.

Regardless of needing to know what fire thresholds had been set for each detector on the system this still wouldn't prove the CIE would "respond" to the phenomena recognised by the detector.

"If the purpose is to demonstrate that the detector is capable of transmitting an analogue value to the panel which corresponds to an alarm condition, why doesn't 45.4c) state that the test should provide an alarm response?"

Because it's covered in 45.4b "........... every detector should be functionally tested........."

[Colin Newman]

I apologise if this sounds like playing with symantecs, but surely we need to be clear if we are testing the detector or the system.

If we are testing the detetctor the surely being able to see the analogue value vary with a varying stimulus is sufficient to demonstrate that sensing element is sensing and the detector is functioning, hence the detector has been subjected to a functional test.  45.4b) states that 'The [functional] tests used need prove only that the detectors are connected to the system, are operational and are capable of responding to the phenomena they are designed to detect' it doesn't state that that response should be synonymous with an alarm condition.

If we are testing the control panel's response to the magnitude of the output from the detector then it doesn't reconcile with 45.4c) since unlike 45.5d) it doesn't require an alarm response.

[kurnal]

Our learned friend CS Todd has covered this point in his guide to BS5839-1 where on page 299 he reinforces the point that the test recommended for each detector is a functional test and that it would not be sufficient to rely purely on the measurement of analogue values at the CIE equipment of an addressaable system (though the code recommends these be checked annually).

He does not amplify why this is the case however so I guess we will have to await the next expensive volume or persuade him to educate us further on the forum.

We are all aware that technology moves on apace and very often it might be the case that we carry on following the codes just because thats the way its always been done and may overlook the benefits the new technology can bring.

[Wiz]

.......The conflict I refer to is that 45.4d) requires an alarm response, but 45.4c) doesn't.

Colin, I see what you mean but although 45.4 d) mentions; '....produce a fire alarm signal', I would say that the term in 45.4 c) 'Every heat detector should be functionally tested...' means that the functional test ensures a fire condition is produced. Surely it wouldn't make sense not to?

[Colin Newman]

I agree that it would appear to make sense that an end to end functional test is required which includes activation of each automatic detector, transmission of detector condition to the control panel, determination of fire condition, activation of the cause/effect matrix, transmission of signals to alarm devices, control interfaces, alarm receiving centres etc.  However, that is not what is described in 45.4.

The emphasis appears to be on testing the ability of the detector to provide a response when subjected to a fire-like stimulus as described in 45.5b) "The tests used need prove only that the detectors are connected to the system, are operational and are capable of responding to the phenomena they are designed to detect."

Whilst intitiating an alarm condition is one way to indicate that a detecor is able to respond to a fire-like stimulus, so is verifying fluctuations in analogue values carresponding to fluctuations in ambient conditions.

For example, the analogue value of an Appollo xp95 heat detector has an approximately linear correlation to ambient temperatue, therefore, by monitoring the analogue value over a relevant time period, fluctuations are likely to be witnessed which correspond to building usage.  This would identify if the detector is capable of responding to heat stimulus.  Is it then necessary to demonstrate that the heat detector is capable of responding to more heat stimulus?  As I've questioned before are we testing the system or the detector?

[Wiz]

Colin, I feel the recommendations of BS5839-1 are trying to ensure that any testing is as true a test of the whole system as possible.

If, for example, we relied wholly on the fact that an addressable heat detector just appeared to have a fluctuating analogue value, this wouldn't necessarily confirm that it was operating correctly. If, for example, we didn't go to physically check the detector to ensure that access to the sensing thermistor wasn't blocked by debris etc., then maybe we might find that the analogue response wouldn't be a true reflection of the temperature surrounding the detector e.g. could it take a temperature of, say, 100C to actually cause the detector to give the analogue value of 55 (which would initiate an alarm) because of the debris blocking access of warm air to the thermistor?

I'm sure BS is just trying to ensure a testing regime that covers a true reflection of what creates an alarm condition in a real fire condition. And that relying totally on the apparent information of the analogue value might not be enough of a test.

With reference to your point regarding the Apollo XP95 heat detector, it is only the XP95 Temperature Detector part no. 55000-400 (Standard) that produces an analogue value that numerically relates to the sensed temperature in degrees C, e.g. a sensed temperature of 40C will produce an anlogue count of 40 (approx). An analogue count of 55 is the alarm condition, so the detector 'trips' the alarm at 55C. I believe that Apollo designed this as a 'feature'.
The High Temperature version part no.55000-401 does not have the 'numerical' relationship, and does not produce an analogue value of 55 (fire condition) until 90 degrees C

 

[David Rooney]

The High Temperature version part no.55000-401 does not have the 'numerical' relationship, and does not produce an analogue value of 55 (fire condition) until 90 degrees C
 

And further to that as mentioned earlier, some systems such as Hochiki ESP Protocol calibrate detectors each 24 hours and set varying analogue fire thresholds depending on the ambient conditions measured at the time making it quite difficult to check that each device has reached or passed the point at which the CIE would react.

Why is it such an issue ?

I don't understand what is to be gained by viewing rising and falling analogue levels over witnessing a zone lamp being illuminated ?

Or are you trying to get an amendment written to clarify what is already written in the guide that will cost us all another £200 for the privilege ??

[Colin Newman]

Wiz, I fully appreciate the need for a physical check and wouldn't consider standing at a control panel and looking at numbers on the screen to represent a substitute for functional testing.  For a start there's no way of telling whether a cover had been left on a detecteor without physically going to check.

The point I'm making is that fluctuating analogue values are an indicator that the sensing element, A/D converter, Comms and signal path are operational.

David, I'm not trying to be clever (I gave up on that years ago 'cos I was no good at it!), nor do I want another amendment to the BS.

The reason that this is an issue, is that a number of hospital installations are carrying out the functional tests as described, and due to the size of the installations, testing often goes on for days. In many cases, to prevent days of alarms sounding, significant portions of the system are puit into "test" mode such that the alarms don't sound and interfaces don't actuate.  This can be manually overridden, but still there is an increased level of risk during the testing process.  Coupled with this I have trawled the test records for a number of hospital sites for a period of some years and have yet to identify a recorded instance of a detector failing a functional test without there having been a related fault indicated at the control panel. 

Whilst cost is always an issue in the NHS, my main concern is that the increased levels of risk exhibited during functional testing does not appear to be justified if the test always passes i.e. the test doesn't identify a fault that isn't already indicated.

[David Rooney]

Well I've certainly come across analogue detectors that appear normal on the control panel but haven't responded to functional testing - not many but a few - and certainly I wouldn't  certify a system as operational and compliant with BS if I hadn't carried out a functional test of the detectors simply because a statistic showed they never went wrong without indicating a fault on the CIE.

[Wiz]

Colin, I can see why you are interested in this issue.

I can confirm that I would agree that an analogue addressable detector that exhibits changes in its analogue value when subjected to the environmental condition that is being monitored is likely to be working absolutely fully and correctly. I don't have any memorable experience of this not being so in the 25 odd years I have used analogue addressable systems.

Since BS5839-1 are just recommendations I can't see why you couldn't come up with some other testing regime that would suit your requirements if you could get the relevant parties to agree it as a variation.

I can especially see that if you were able to monitor the analogue values of heat detectors over a period of time during temperature changes then there could be an argument that they were probably working o.k. without having to take them all the way up to a fire condition.

But i'm pretty sure that, with the critical nature of fire alarm systems, many people would insist on adhering to the BS recommendations at the very least.

[Colin Newman]

David, if you've got any details of analogue detection failures you could let me have I'd be really grateful.

Wiz, I fully appreciate the critical nature of the fire detection system, especially in hospitals, which is exactly the point that I'm making, if it's partially disabled to facilitate a test which we have a 99.9% confidence that it will pass, why endure the partial loss of utility?

Since we are required to demonstrate that our risk is managed ALARP, it strikes me that this area of petential risk should be considered.  In addition, the NHS as a public sector organisation has a duty to achieve value for money so there's an incentive to mitigate the cost of functional testing if it doesn't appear to provide any real benefit.

I'm not suggesting we tear up the BS and unilateraly decide to dispense with faunctional testing, but we should maximise on our investment in the technology and where functional testing is proven not to add the same or greater benefit than the cost (both risk and financial), then we should reduce our testing activity, maybe by extending the test interval rather than dispensing with it.

Open to being shown the folly of my thinking, just not had anyone present me with evidence that functional testing identifies a significant number of failures that aren't already indicated by the automatic system monitoring.

[Wiz]

Colin, I'm somewhat surprised that no panel manufacturer, to my knowledge, has developed an addressable device monitoring software package. If it was available, it might provide sites like your own with a useful tool that might allow you to change your system testing regime to more suit the conditions it was experiencing.

I envisage a software package running on a PC connected to the system and where every device's analogue value was monitored against general environmental conditions. E.g. changes in temperature and air quality etc. This might even be  possible by comparing one device against other addressable devices in known similar environments. The idea would be that the software would flag up devices in which the analogue values didn't respond in the manner expected.