Testing optical flame detectors beyond international standards

Fire protection is one of the crucial safety aspects of many hazardous industries, including oil & gas. A spill of flammable liquid or flammable gas leak, if ignited, may quickly spread with undesirable consequences for personnel, plant and reputation. 

To help mitigate this risk fast and reliable fire detection is required. The requirement for short response times in the large open spaces, typical of these industries, dictates the use of optical flame detectors.

Flame detector models vary significantly in the way they work. The electromagnetic wavelengths measured, the optics and sensors used, the signal and data processing, and the decision algorithms, may be very different from one model to another. These differences result in functional differences between models. Different flame detector models, utilizing different detection technologies, or even the same detection technologies with different designs, may react very differently to the same fire and environmental conditions. This is why a thorough functional assessment of a flame detector model is crucial before it is being used in hazardous industries.

This article details one such assessment according to test procedures designed by a global energy and petrochemical company and conducted by an independent laboratory in the UK.

General objectives

Optical flame detectors use a variety of technologies, based on measurement of the optical radiation emitted from flames in the ultraviolet (UV), visible and infrared (IR) parts of the electromagnetic spectrum. Since in industrial environments radiation in these parts of the spectrum may be abundant, one of the challenges in the design and production of a reliable flame detector is to correctly identify whether the measured radiation originates from an unwanted flame or from a harmless source. Another challenge is the reliable operation of the detectors in harsh weather conditions, and when contaminated by raindrops, sea water or any other material relevant to the specific industry. The detectors should also be able to function reliably in condensed industrial environments, where the line of sight to the fire may be somewhat obscured by equipment, piping etc.  

The main objective of these tests was to evaluate the detectors in a few key areas, these were:

Flame sensitivity

• On-axis flame sensitivity
• Field of view 

Rejection of false positives

• Alarm susceptibility to sunlight 
• Alarm susceptibility to blackbody radiation and other artificial sources
• Response to hot exhaust fumes

Environmental considerations

• Environmental performance (temperature, humidity and water spray)
• Speed of response for a range of fire types (materials) and sizes
• Mechanical integrity, vibration, and alignment performance as appropriate

Real-world scenarios

• Sensitivity to larger fires which are more representative of major hazards
• Performance with optical contamination – water film, salt deposit, glycol, soot
• Detection of partially obscured fires

Detectors under test

The FlameSpec flame detector family is a range of explosion-proof optical flame detectors offering global approvals for use in hazardous areas, fire detection performance and SIL. The units can be interfaced with conventional fire panels or safety PLCs using a range of industrial outputs. All units are delivered with HART®, heated optics and adjustable sensitivity. The units log alarm events and capture sensor data for post-event analysis, the units are also available with different camera options, the choice being matched to expected fire type. The specific variants under test here were:

Device 01 FlameSpec IR3 – this device monitors IR radiation in three wavelength bands, allowing it to identify the spectral signature of hot carbon dioxide (CO₂) in flames.

Device 02 FlameSpec IR3-HD – this device uses the same detection method as the IR3 and has an integral colour video camera with internal recording of fire events.

Device 03 FlameSpec IR3 CO2L – this unit uses a modified algorithm of the first two devices so that it will not alarm to intense sources of exhausted hot CO₂ such as from helicopter engines, tanker loading trucks or other intense sources of exhausted hot CO₂.

Testing method

Testing of all flame detectors was conducted on the manufacturer’s default sensitivity setting, in the case for this article that is described as ‘medium’ sensitivity. Here we summarise the tests, in general, and add details on specific tests of interest.

Test procedure, in general

Flame sensitivity

The starting point for detector assessment was to assess the product’s performance against manufacturer statements. Today, FM 3260 and EN-54 part 10 are international standards for the performance of optical flame detectors. These documents define a standard fire test as a 12in x 12in (0.3m x 0.3m) n-heptane pan fire. FM, further define the Field of View (FOV), horizontal or vertical, as being the angle where the detection distance is at least 50% of the on-axis detection distance. Historically gasoline (petrol) had been used as the primary test fuel, and whilst this can provide equivalent response data, there are variations between gasoline types, so n-heptane was chosen as a pure compound that can be sourced globally. During the whole test programme two fire conditions were used, these were:

A) Established fire – A fire was lit but hidden from the detector using a solid metallic screen. Once the fire had evolved and become stable, which typically took 30 seconds, the screen was withdrawn, and the response measured. This was the main fire type to be used during testing.

B) Fire from ignition – As the name suggests, the fuel was lit without the use of an obscuring screen, and the device response observed.

Two fire sizes were used, these being, 12in x 12in (0.3m x 0.3m) pan and 1m diameter pan (larger fire).

Rejection of false positives

Historically, common sources of false alarm were presented to a detector without a fire being present, the detector was then moved nearer to the source until the detector responded. In later years, these tests were supplemented by determining if the detector response was desensitised by the presence of false alarm stimuli. Typical forms of unwanted alarm source included sunlight and hot blackbody radiators.

Environmental considerations

The detectors under test can be deployed almost anywhere in the world, and it is important therefore to understand what affects different temperatures – hot, cold, or thermal shock – have on the devices. Hot and cold climates are easy to visualize, the thermal shock test may relate to hot climates, but units located near the equator are subject to frequent heavy rainfall over short periods of time, whilst operating at reasonably high temperatures, particularly due to internal heating from electronic components. In addition to temperature effects, an assessment was needed to determine what impacts different forms of optical contamination had on detector response.  

Real-world scenarios

While flame detectors are regularly tested in open areas to international standards, e.g. as part of FM 3260 or EN54-10 certifications, these fires are relatively small compared to the fires more associated with major hazards.

The real-world scenarios tests therefore have two objectives: firstly, to determine fire-detection performance with a larger fire, e.g. 1m diameter fire – this test also includes how congestion, for example, from process equipment and piping, may impact the detector’s clear line of sight to a fire and hence detection performance and; secondly, to ensure the detector is still capable of responding to a flame that may be close to saturating the detector’s input signals.

The obscuration plates used were designed to block the flame signal by 25%, 50%, 75% and lastly 90%.

Starting at the maximum obtainable distance without obscuration grid the detectors were exposed to an established larger fire. The grids were then added in order of increasing blockage until the detector no longer responded; at this point the detectors were moved 10m nearer to the fire and the grid that had failed was tested again. This process was repeated until the detection response to the 90% obscuration filter was less than 10 seconds.

Other elements that present real-world challenges include optical contamination, which can come from a variety of sources. In these tests, water, salt water, glycol, light-oil and soot were evaluated.

Test results

Flame sensitivity 

The detection range to a standard fire is stated as being 30m (98ft) using medium sensitivity; within 3 seconds, and this is as approved by Factory Mutual (FM).

Rejection of false positives – susceptibility to Sunlight

Three conditions were tested, these were:

1. un-modulated sunlight – reflective disc is placed no further than 5m from the detector and positioned such that the sunlight is focused onto the detector.
2. randomly modulated sunlight in then focused onto the detector by rotating the disc.
3. regularly modulated sunlight created by rotating the chopper wheel immediately in front of the detector to produce modulations between 1 and 15Hz.

For all tests no false alarms were registered.

Rejection of false positives – detection performance in sunlight

All detectors alarmed to the fire, without the detection distance being compromised, in the presence of sunlight whether it was modulated or not.

Rejection of false positives – susceptibility to blackbody radiation

Testing was conducted using a 2.8kW heater. The heater was placed 3m away from the detector and its radiation modulated using two chopper wheels. No alarms were recorded by the detectors.

Detection performance in the presence of blackbody radiation

This tested used the same approach as the previous one, with a standard fire, from ignition, being introduced at 30m, and the modulated heater 3m from the detector.

All detectors alarmed to the fire, without the detection distance being compromised, in the presence of the modulated or unmodulated heater.

Real world scenarios – optical contamination

This test was conducted to determine if the dirty optics window fault (BIT test) is sufficiently sensitive to alarm before the detector was unable to respond to a fire. The fire under investigation, was the standard established n-heptane fire, at an on-axis distance of 30m.

The contaminants were applied to the detector window up to five times, with the materials tested including glycol, engine oil and soot. All detectors responded in a reasonable amount of time to the contaminants under investigation.

Real world scenarios – larger fire tests

The objectives of this test are twofold:

1. To verify detector sensitivity performance for a major hazard fire (larger than the specification ‘standard fire’), including impact of congestion partially obscuring the fire.
2. To ensure the detector can detect fires that may be close to saturating the input signals of the sensors.

Starting with sensitivity performance, the maximum distance available to the detectors was 130m, the detectors responded.

The detectors also responded at 40m within approximately 2 seconds when using the 90% obscuration filter.  

Obscuration plate	Distance	Device #01	Device #02	Device #03
25%	110m	2s	6s	2s
50%	80m	6s	2s	6s
75%	70m	5s	9s	5s
90%	40m	2s	2s	2s

Real world scenarios – larger fire tests – saturation tests

The saturation tests were conducted 2.5m from the first. In the first test the fire was allowed to stabilise before the detectors were exposed to the fire.  

Saturation test	Distance	Device #01	Device #02	Device #03
Established	2.5m	5s (4831)	4s (4764)	6s (4350)

In the second test the detectors were challenged to see the flames from ignition.

Saturation test	Distance	Device #01	Device #02	Device #03
Ignition	2.5m	1s (4836)	1s (4769)	6s (4354)

Summary

This article has detailed a test programme designed by a global energy and petrochemical company and conducted by an independent laboratory in the UK. Testing started in June 2023 and concluded in December of the same year.

Three triple-IR detectors were evaluated in the areas of flame sensitivity, rejection of false positives, ability to work in environmental conditions and lastly considering real-world scenarios of major hazards.

For more information please go to: www.fg-detection.com