Flame Detector Voting Query


Thread Starter


Hello everyone,

I've had a question from one of our offshore assets on UV flame detectors.

Inside the turbine compressor enclosure, there are two groups (circuits) of three flame detectors. The first group is in the "engine" enclosure and the second group is pointed at the drive shaft/turbine end. Both groups (6 FDs in total) are in the same fire zone.

The cause and effects and testing procedure state that an executive action will occur if 1 FD from each group (kind of like 2oo6, not sure if this is accurate name) trigger. However, in actuality the system acts as 2oo3 within each group, i.e. if 2 FDs in the engine enclosure trigger then we'll get the executive actions (alarm, co2 release, etc).

My question is, has anyone seen this 2oo6 (I'm calling it this, though it may have a different name) and is this preferable to 2oo3? I can see the logic, you want to verify that there is an actual flame in the compressor (same FZ). But 2oo3, to me, makes more sense, because you don't have to wait until the flame/fire spreads to the other group's detection range. Some reference reading material would also be very appreciated. Thank you.

Assylbek (Instrument Engineer, 1Y experience, Aberdeen, North Sea)

Yes, 2 of 6 is a legitimate logic term. M of N voting has many applications!

Nuke's use a 2 of 4 strategy. Even supplied a 3 of 75 SIS system some years ago!

Regards, Phil Corso

The voting from your description appears to be 1oo3 for each flame detector group (e.g. any single FD in a group is a vote to trip, e.g. executive action), which is then voted 2oo2 to take the final executive action (e.g. a vote from each detection group (2 ea) is required for an executive action).

This is not 2oo6 vote since flame detection by two flame detectors in a single group will not cause a trip. This voting also requires detection of a flame in two different areas of the fire zone, which means that the fire will have to expand to be present in two different detection zones (this assumes that the two groups cannot see a fire in the other's detection area). If all six of the flame detectors will see any flame in the compressor fire zone, the (1oo3) 2oo2 (1oo3) vote makes more sense. Without knowing the geometry of the two detection areas, it is hard to judge the extent of a fire which would be detected by both detection groups.

Flame detector voting is determined by three general factors. The first is how many of the flame detectors will see the flame for a given fire situation, the second is what protection is required to minimize the potential for a spurious trip, and the third is the risk of the hazard. Assuming there are two separate fire detection areas, your logic regarding that it is better to take executive action if a true flame is detected in the detection location of a group to minimize the potential damage to the compressor and to address any safety consequences is reasonable.

In general, the sooner a hazard is detected, the more likely that the consequence will be prevented or minimized by actions of the safety system. A risk assessment of the consequence of a fire for your particular installation should be done to find the risk from a safety consequence and from financial loss perspective(damage to the compressor due to a fire). An assessment of the potential financial loss due to a spurious trip should be done and factored in to find the correct balance of redundancy.

If all three FD in a group will detect a flame in their protected area (but not the other group's area), a vote of 2oo3 generally provides a good balance between safety (1oo3) and spurious trip (3oo3) requirements. If you then vote the two groups 1oo2, you should provide adequate protection for the complete fire zone.

I would check with the manufacturer for their recommendations as they have the most experience with their compressor installations.

William (Bill) L. Mostia, Jr. PE
ISA Fellow, SIS-TECH Fellow,
FS Eng. (TUV Rheinland)
SIS-TECH Solutions, LP
Thanks for the quick reply gents. I think will perform some form of risk assessment using FD coverage areas and safety/cost analysis.

Currently the system is in 2oo3 in each group, so I don't think it's very critical. But it would be good for me to learn how to review and analyse system suitability and even start thinking about optimisation.