Stop tripping the plant
Flame detectors are used widely across oil and gas installations because they respond quickly to a fire, indoors or outdoors. The detector output may initiate alarms, shutdowns, suppression release and emergency response. The challenge for the facility operator begins when the detector alarms for the wrong reason
A false alarm is not just an irritation. On an FPSO or other petrochemical facility, one unwanted trip can carry a significant cost. Lost production is the obvious one, but it is rarely the only one. There is also a less visible cost: confidence. If operators see repeated alarms from the same detector, or in the same process area, and each time there is no fire, the system starts to lose authority. People may not consciously ignore it, but responses can become slower. A safety system must be trusted, and false alarms steadily erode trust.
UNDERSTANDING THE ‘WHY’
Modern optical flame detectors usually look for a combination of spectral and temporal features. A triple infrared detector, for example, does not just measure “heat”. It compares energy in the hot carbon dioxide emission band with adjacent guard bands and looks for flame-like flicker and signal relationships. A UV/IR detector uses two different sensing principles. Both technologies can be very robust when correctly applied, but neither should be treated as universal.
The installation environment matters. Welding, hot exhausts, flare reflections, modulated sunlight and reflections from metallic surfaces can all create problems. Some are obvious during design; others only appear once the plant is operating. This is why site experience and realistic false alarm testing matter. A detector that performs well against standard fire tests may still be challenged by a particular installation if the false alarm sources in that area were not considered.
Flare reflection is a good example. A flare is a real flame, but it is not the hazardous fire that the detector is there to detect. On an FPSO, reflected flare radiation can move across structures, pipework, handrails or wet deck surfaces. To a conventional optical detector, this may produce changing infrared energy which looks more fire-like than a simple static hot object. Desensitising the detector may reduce the trips, but that is not a very elegant answer if it also compromises genuine fire detection.
A HYBRID APPROACH
This is where a newer hybrid approach is useful. By combining triple IR flame detection with video analytics, the detector has more information on which to base its alarm decision. This approach is known as Video Enhanced Flame Detection, or VEFD. The triple infrared section provides the fast spectral response to a fire, while the integral near-infrared camera allows the video image to be analysed for flame shape, size and apparent movement.
The result is a more robust detection decision because spectral information and image behaviour are assessed together in a combined device. The key point is not that video replaces infrared flame detection. It does not. Instead, it adds a second layer of discrimination, which is particularly valuable where reflected flare radiation can otherwise look like a fire to a conventional flame detector.
CONSIDERING FIELD OF VIEW
The field of view is another area that deserves attention, because a detector can only make decisions based on what it can see. F&G mapping tells us what the detector should cover, but it does not prove the coverage provided by the installed detector. A practical field-of-view validation record, captured at commissioning and linked to the detector tag, is therefore valuable. It gives maintenance teams something real to compare against later, rather than relying on memory or drawings, and helps prevent the plant being tripped by a detector that is simply looking in the wrong place.
ADDITIONAL CONSIDERATIONS
Voting and cause-and-effect logic also have a role, but this must be engineered carefully. Too much voting can delay action or reduce availability, but sensible logic can reduce unnecessary executive actions without ignoring the first warning.
Finally, event data is becoming increasingly important. If a detector records video, sensor data and decision parameters before and after an alarm, the investigation changes completely. Instead of guessing whether the cause was sunlight, hot work, flare reflection or contamination, engineers can review what happened. If a setting change or re-aiming decision is made, it can be validated against the original event.
False alarms do happen, but they should not be solved by simply reducing detector sensitivity. The best results come from selecting the right technology, validating what each detector can see, applying sensible logic and using recorded data to remove guesswork. That maintains system confidence whilst reducing the chance of tripping the plant for the wrong reason.
Dr Eliot Sizeland is vice president of business development at Fire & Gas Detection Technologies.
www.fg-detection.com
