Fire and Gas Detection
- Strategy objectives
- Current knowledge of detection effectiveness
- Areas of Uncertainty
- Industry practice
- Strategy development issues
There are two principal types of detector which are commonly in use in off-shore installations: heat, flame & smoke, and flammable gas instruments. The most significant for risk reduction are gas detection systems, since they give the earliest warning of hazardous situations. Infra Red (IR), line-of-sight or point type detectors which identify an accumulation of gas and acoustic leak detectors, are also used. The OSD strategy is to promote the use of a combination of sensors, thereby giving early leak detection with the acoustic detectors and identifying a gas cloud accumulation with the IR type sensors.
Objectives in terms of effective gas detection revolve around the life cycle of the system:
- To specify performance requirements based on installation specific hazard scenarios;
- To identify where deficiencies exist with regard to detector specification and effective operation, including maintenance;
- Initiate research to increase knowledge and understanding in ill-defined areas of detection systems effectiveness; and
- Promote the use of a consistent methodology in the location of detectors, to maximise detector effectiveness.
Current knowledge of detection effectiveness
Vapour state detector specification
The key requirements of the system are that it should (a) respond effectively and reliably to the hazard and (b) be tolerant of the environment and working procedures. This requires knowledge of the following:
- Real-time ventilation surveys are critical in detector performance. Re-circulation and/or dead zones need to be identified. This is rarely carried out, and is an OSD priority area of concern;
- Detector head location and coverage should be specific to the hazard conditions within the installation area covered. Head location commonly follows a grid pattern;
- Detector sensitivity to oil or mist releases is variable and there are large uncertainties with calibration systems. OSD regard the use of liquid leak detection as a future concept to be pursued;
- Many gas detectors are sensitive to low (geometric) sunlight, fog / condensation and reflections. Newer models overcome this problem;
- Coverage by point detectors is sensitive to the ventilation regime. Beam instruments give wider coverage;
- Point detectors can give a higher maintenance burden than Line of Sight Detectors;
- An analysis of 8 years of data relating to hydrocarbon releases (HSE, 1999b) indicates that across all installations and detection systems an effective detection rate of about 60% has been recorded.
Liquid or droplet mixture detection specification
Oil mists are generated by the release of flammable liquids under pressure. Oil mists are very flammable and can ignite at a lower temperature than most hydrocarbon gases. Most oil mist detectors are optical beam devices. The current evidence is that gas detectors do not seem able to detect oil mist releases. Wormald & Shell have developed a mist detector for installation in ships engine rooms.
Fire and smoke detection
Fire may be detected by heat rise or flame sourced radiation in the UV, visible and IR spectrum ensuring that all types of fire will alert the detector system. Current knowledge is that these instrument systems are generally reliable and effective. They are independent of the ventilation regimes, unlike gas detectors.
Various rules of thumb are used to determine the location and coverage of the different types of fire detector. Point heat detectors in open, naturally ventilated areas are sited at approximately a density of 1 per 25m2 and at spacing of 7m with a maximum distance from bulkheads of 3.5m. In enclosed mechanically ventilated modules, they are sited at approximately 1 per 37 m2 and 9m apart with a maximum distance from bulkheads of 4.5m.; They are not applied in areas with high ceilings above 8m (point heat detectors have poor sensitivity with height).
Flame detectors are sited such that their vision cone covers areas where fire may occur. For IR flame detectors around 15m is considered a reasonable range because of obscuration by smoke and lack of sensitivity at the periphery of their field of view. They are generally sited at the corners of an area or module. CAD tools are used to optimise their coverage at the design stage. Triple band IR systems are less prone to false alarms.
Point smoke detectors rely on the transport of combustion products (particulates and gases) to the detector by convection. The numbers of detectors can be reduced with increased ceiling height because of more uniform distribution, although the concentration will be less and the sensitivity of the detectors must be adequate. Current smoke detectors are located at not more that around 7.5m apart and are not appropriate for high ceilings (>10.5m).
Detector head spacing is governed by the size and geometry of the area (confinement and congestion), ventilation and the nature of the release. Typically the minimum spacing in congested areas is around 5m based. Problems can occur in large volumes or trapped volumes where there is local confinement that restricts the venting path. Gas/vapour tends to slump, particularly in low air movement areas, after a liquefied gas release because the local gas cloud is relatively dense and cold. Detector heads should be located in a 3D pattern with some heads at low level in modules that are liable to have gas vapour slumping.
Duct sampling. Either point or beam detectors can be installed in HVAC and other ductwork. Generally beam detectors are not an area of concern as they are more likely to provide coverage across the duct.
Acoustic detectors. Location is based on identifying the potential sources of leaks, e.g. all joining parts in high pressure gas installations. An ultrasound map of the background noise can be determined to decide the alarm level and assist with selection of the optimal location. Care must be taken to avoid acoustic reflections that may produce false alarms.
It is possible to use the detectors in a grid system as for the concentration-based detectors above. However, the location criteria are different and are not as well understood as for point catalytic/IR detectors because of their newness.
Areas of uncertainty
Guidelines for optimal combination of point, beam and acoustic detectors are dispersed in various documents and lack useable detail.
Conclusions from analysis of JIP data are not particularly useful for effective detector set-up, i.e. insufficient number of release scenarios to cover the conditions found in practice. Location of detectors for various types of flammable gas release: methane, propane and condensates are not defined.
There is sparse information on effectiveness of gas detectors to detect low rate releases (<1.0kg s-1) but which may become significant over a period.
Response of instruments calibrated for methane to higher chain hydrocarbons are not well defined and awareness in the Industry is low (methane and C5-C8 hydrocarbons)
Relation of measurement parameters of beam (%LEL.m) and acoustic (dB) to flammable hazard (e.g. explosive cloud volume) is not well understood.
Application of intelligent data processing to assist in better detection of fire/gas incidents is not used offshore, i.e. pattern recognition technology. It is an area of potential benefit.
Industry practice is based on the general guidance provided in the UKOOA guidelines (UKOOA, 1995) that is then translated into more specific rules and guidance in company codes of practice (e.g. Shell, 1995 and BP, 1997). BS EN ISO 13702 contains high level advice on detection systems.
A JIP of large scale experimental study of gas build-up from high pressure releases in naturally ventilated offshore modules was completed in 2000. This JIP study did not address detection directly, however the data is being used for evaluation of the effectiveness and optimisation of gas detection networks.
The principal factors explored in the tests were release rate, direction and location, module wall configurations, and wind speed and direction (both external and internal). The main findings were:
- The use of a grid based on 5 m spacing for point detectors was successful in detecting releases where clouds formed within the module;
- When only small clouds formed, or cloud growth was slow, detection times increased and in some cases the releases were not detected at all;
- Halving the spacing between detectors slightly reduced detection times but at the expense of a large increase in the number of detectors required. Doubling the detector spacing caused a large increase in detection times;
- The IR point detectors performed better than the catalytic detectors, both in the number of releases detected and in the detection time;
- The IR beam detectors showed good performance in the initial configuration and when comparing their performance to rows of IR point detectors; and
- Well-placed detectors can improve the performance of a detection system and this emphasises the need for knowledge of dispersion, the processes being undertaken in a module, and the equipment layout when deciding detector placement.
Strategy development issues
- Further, more specific guidance is needed on the most effective location of detectors in a system. This guidance should be based on experimental studies (e.g. JIP) of:
- Methane releases;
- Higher fraction, e.g. propane, releases; and
- Condensates (C5-C8 hydrocarbon) releases.
- The response of sensors calibrated to one type of hydrocarbon when exposed to longer or shorter chain hydrocarbons specific to the well fluids should be investigated with the aim of improving overall detection performance.
- In assessment and inspection the dutyholder should be made aware of real time ventilation data as a key element in installing an effective gas (and smoke) detector system.
- The integral use of acoustic leak detectors and IR systems needs to be studied and guidance developed from the findings.