Extinguishing system query....

[David Rooney]

We are going to install an addressable coincidence detection system in a computer room - .

Automatic shutdown of the kit is not required and we have suggested they need manual Emergency Power Off button etc in line with 6266.

The question is... is it permissible to install the EPO buttons on the detection loop via I/Os, give them a different priority and use cause and effects to shut the kit down ??

[Big_Fella]

What kit are we referring to that they require shutdown?

[Galeon]

Personally , I wouldnt go there , you may well be digging a big hole for yourself , as you are now entering the world of a system akin to Santana etc.
But you may wish to appease your client , and if you come up with this solution and they are happy , I cant see a problem .
As you well know with downloads etc to the panel , you are leaving yourself exposed if it went t*ts up and shut the power off , because of this type of method.

[David Rooney]

What kit are we referring to that they require shutdown?

everything in the computer room........... complete shutdown.....

[David Rooney]

Personally , I wouldnt go there , you may well be digging a big hole for yourself , as you are now entering the world of a system akin to Santana etc.
But you may wish to appease your client , and if you come up with this solution and they are happy , I cant see a problem .
As you well know with downloads etc to the panel , you are leaving yourself exposed if it went t*ts up and shut the power off , because of this type of method.

What's Santana?? ... haven't heard of that one.

We've done similar in some MOD computer rooms, but used networked syncros - one for the fire detection and one for the EPO's and all the shutdown outputs to PDUs and UPS etc etc...

downloads have never been a problem but did once have to reboot systems with around 30k users on line.... that was a bicycle clip moment.......!!!

[Big_Fella]

I suppose it's no different really than providing an interface to shut all the power down.  I expect it's critical data stored on the computers and they are worried of such loss or corruption of data in such event, although a genuine fire could potentially cause damage to one computer before the system went off anyway.

I guess they are just being over cautious, hoping that someone may be in the room should the system activate and that person would determine if the risk is significant enough to shut down the power or weather it can be dealt with prior to having to do this.

Although this does seem like a strange situation as if such data is critical I would have thought they would have a live replication of it on other computers in either different rooms or a different building.

We have several credit card processing centres, with huge machines in rooms processing transactions/banking details/customers accounts etc from very well known banks, and they don't shut the computers down.  They are reliant on the suppression system putting it out before anything gets too major, or if something significant happens is only restricted to a very small number of computers.

[Allen Higginson]

It seems to be by local agreement - I have one site that has the main admin in the middle with three large production buildings around them.There are 8 interlock bypass keys on contactors that need permitted and installed before we start the maintenence or it starts shutting things down througout.I have others that only shut down the aircon etc.
Can't see why you can't interface between c & e programming though.

[Goodsparks]

Personally, I wouldn't as the FA system introduces another thing into the system to go wrong ! I've Just finished a small data centre with a Water mist system and ended up building a small panel to to the EPO. (Ill try and work out posting a picture) Each of the shunt-trips supplying the UPS supply, Output and A/C supplys have a auto / manual / off switch on the panel to allow them to be taken out of circuit for works on the FA system or to provide a delay between the UPS input and outputs being shut-down in the event of automatic operation in a fire condition.

A fire alarm engineer c0cking up, sounding the alarm and evacuating an office is bad, dumping a couple of grands worth of FM 200 is worse, but shutting down a data centre with thousands of users on line, and credit card transactions taking place is going to be a really bad day.

Paul

[patrickhamblin]

One major problem, as I see it, is that if power is not shut off to both air handling and the computer installation and then the gas is dumped, there will be reignition if the fire is of electrical origin, unless the power is subsequently isolated. Sometimes this is done manually, if there are suitable technicians at the site, or even from a remote location. There certainly is a reluctance by some users for this power isolation and the provision of facilities inside the room is sometimes considered unaccepable, but the provision of switching from outside the room, controlled by technicans, has been regarded as more acceptable. To leave the equipment powered up after initial extinguishment cannot be good practice.

[kurnal]

I have seen this approach  go seriously wrong once before when a co-incidence of issues came together. As other posters have indicated perhaps the most vulnerable time is during engineering work or reconfiguration of the fire alarm system. Other faults should also be considered though, such as the total loss of the system configuration and the potential for loss or corruption of any non volatile memory when some systems revert to a safety fall back of operating every interface, which in this case could be undesirable.

On the other hand the aircon and cooling  will be inevitably shut down on release of the extinguishing media and the computer systems cannot run for long without melt down through overheating.

On the larger installations there will probably be a backup system elsewhere running on hot standby so if this is in operation and maintained to a high standard a soft shut down of the other may be reasonably easy to recover.

So whilst there is no technical reason why you should not cut the power on a fire alarm originated cause and effect,  questions over overheating time, availability of hot standby, communicaion between the IT team and Facilities management to ensure before the alarm is worked on such interfaces are physically disabled by keyswitch or similar.

It may be safer if the site is staffed by an IT team 24/7 to leave it with a manual control-  add something like an additional blue break glass box to cut the power working only in co-incidence with either the double knock of two detectors or gas having dumped which should give an alert to the IT centre.

In the case you have in mind is the extinguishing system self contained with its own dedicated detectors providing the double knock or is the gas system controlled by detectors on the  main fire alarm and detection system? (I am a firm believer in keeping them self contained)

In the case I quoted on the day the fire alarm engineer was fairly routinely uploading a configuration update on the fire alarm system on site A, the hot standby machine on site B had a known fault. IT at site B were not aware of the fire alarm work going on at site A.  (and perhaps would not have forseen any problems as they did not understand the complexity of the fire alarm systems)

During the upload at site A the  upload failed for some reason and the alarm system reverted to safety sector 1- full operation of very device and interface . Gas was dumped, power shut off and damage caused to the hard drives which took a long time to recover.