When 30 seconds is not enough for fire protection.

18 November 2020 by Chris Downing, FIA Extinguishing Council

BS 6266 explains that “the extensive use of fire protection systems in electronic installations arises not from a high probability of fire, nor from a significant hazard to life, but from the consequences of fire loss.”  For this reason, many rightfully implement an automatic fire extinguishing system and without such a large number of IT-dependent businesses would have closed.

However, a focus on keeping the (pre-discharge) time delay short to reduce fire damage and agent decomposition may prevent recognising when a longer time delay is essential; or a different approach is required to achieve our common primary objective: safeguarding businesses availability and property.

We challenged ourselves to see how modern technology, and a review of processes, can help solve this contemporary conundrum.


BS 7273-1 clause 7.3.1 tells us “a time delay facility .. to allow personnel to evacuate … will depend upon the potential speed of fire spread and the means of escape from the protected space, it should be as short as possible and should not exceed the 30s unless a longer period is agreed with the stakeholders”, and carries a cautionary note that ‘protracted delays will allow for the further development of the fire and …’.  Readers will naturally infer this is a good reason to never exceed 30 seconds.  However, there are many instances one should or at least needs to consider it, but getting stakeholder (understanding leading to) agreement can be a challenge, and this Guidance Note strives to facilitate and foster that dialogue.


BS EN 12094 specifies the timing capability of extinguishing control panels and local codes and authorities have their own expectations.

This typically originates from historic experiences – good and bad.  However, we must also appreciate how applications have changed and evolved.

A hold button and/or pre-discharge delay is thus to allow:

  1. Escape of persons within the area
  2. Cause & effect (e.g. dampers, doors, pressure relief, etc)
  3. Intervention to avoid an unwanted extinguishing system release

Past focus – and the wording in the 2006 Standard which prevails today - was essentially on the first point, perhaps with the inference of the second, although nowadays the third is more the reason look for a longer delay.

Computer rooms and data centres have grown from small cupboards/closets through large room and halls to a scale which makes football pitches look tiny.  Escape times may thus need to be longer, however, such data facilities will have access control, minimal staffing and employ the ‘onion skin’ of security layering which, whilst serving the purpose of making penetration problematic, also impedes emergency response.

Consider This

Let’s make a hypothetical.  For instance, a large DC has the catastrophic failure of a piece of equipment which spontaneously creates a cloud of effluent the detection system regards as smoke / fire.  This could be an arcing, yet self-extinguishing, PDU or something like a failed refrigerant pipe in an air conditioning unit. Either way a large quantity of ‘effluent' is likely to trigger first stage alarm and give coincidence to another detector – triggering second stage alarm - almost instantaneously.

The less than 30-second countdown thus commences.

First and foremost – does the business’ fire strategy/policy allow staff to go to investigate the fire?  The likelihood is it may on first stage alarm (the first detector operating) but on second stage alarm (coincidence detection) it probably will not.  The likely instruction there is that the extinguishing release is imminent so all personnel should evacuate.

  • If the policy does allow them to ‘go and look’ / investigate, can your business scramble appropriate personnel in time to properly respond?
  • Can they bypass or get through all security measures, mantraps, biometrics, etc in the allotted (<30s) time? 

I suspect not.

Once outside the protected space, if they see an event unfolding, they can manually trigger the system; most panels or system have a means of manual release at the point of entrance or on/near the panel.  This does raise the question that if a responder sees flames through the door’s inspection pane (window), would they feel authorised – and ‘have the bottle’ – to operate the system themselves, but that’s perhaps a subject for a different discussion.

So, in this case, we have a false alarm which would initiate an unwanted operation (discharge) of the extinguishing system, unless someone can intervene.

Here, sensibly, they can intervene.  If the fire is known to be in a protected space, they can logically go towards that.  If the point of entrance allows them a means to assess if it is safe – e.g. glazed inspection pane in the door or remote CCTV - then they can enter.

We thus need a methodology set with enough time for them to do this – taking suitable assessment of risk and their own safety.

Once through the door, personnel can immediately assess the likelihood of a false or unwanted alarm.  Sight and smell at point of entrance make humans exceptionally good fire detectors.

As our unwanted event requires someone to enter the protected space and operate a hold off button (as BS 7273-1 clause 7.1.4 which says “the device should be located near to the exit from the protected space”), a ‘buddy’ system provides the safest methodology – with one at the button and the second presumably on “seek and search”.

If it transpires there is a need for deploying the extinguishant, releasing the hold button commences the countdown timer again.

The paradox here is you then want a short-timer, as you now know you want the releases as soon as possible; early deployment will prevent (during an incipient event) or limit the damage of flaming fire.

Perhaps the answer is an additional timer between the first detector and confirmation of the coincidence (second) detector complemented by as shorter pre-discharge delay timer?

We thus cannot prescribe: this requires a case-by-case consideration, but the technology is there for one to give thought to the above and specify for that site’s specific needs.


Businesses need to challenge themselves and honestly evaluate the means and time required to attend an unfolding event. 

  • Who can they count on to make the assessments and respond accordingly? 
  • How long is this likely to take?

This might also raise the question on whether it is worthwhile also putting an additional hold off button(s) outside the protected space, i.e. before entry?  Even along with the first-responder’s approach and/or in a Security Room / Network Operations Centre

Possibly to consider a change in access control, such as allowing specific personnel to approach a known fire location without impediments: for instance, as disabling their 2-factor authentication, to speed up their arrival.


Coincidence is to provide verification of an unfolding event, and also to avoid unwanted releases.  Normally there is a naturally occurring time delay between incoming (1st and 2nd) alarms.

Where fire growth would be expected to be slow – yet there is potential for an [near] instantaneous coincidence operation – there is good reason to make simultaneous coincidence introduce further delay prior to triggering the second stage of the extinguishing system.

Todays’ basic BS EN 12094 extinguishing control panels may not readily be able to execute this logic, but programmable addressable house systems can use the “staff alarm” function to enact this via I/O units to the extinguishing control.  Alternatively, a longer (i.e. probably greater than 30s) pre-discharge delay may be required. 

To conclude

The probability of a fire – as BS6266 implies – is perhaps remote.  The likelihood of an unwanted alarm can be greater.  However, the consequences of the system operating are only acceptable when there is a fire.

I cannot extol the need for clean gaseous extinguishing and active automatic fire protection within data centres and critical plant.  Non-Disclosure Agreements prevent us from telling you how many a crucial business operation has been saved by such systems, but we can – at relatively low cost – make these systems so much better at holistic business protection.

I firmly believe there are instances where the is good reason to challenge ourselves about the timing of operating extinguishing systems.  In those cases where slow fire growth is expected and there are impediments to a fast attendance – protraction of some kind is effectively essential.  I hope the above provides reasoning to help you explore this with your stakeholders.