A pool fire is a turbulent diffusion fire burning above a horizontal pool of vaporising hydrocarbon fuel where the fuel has zero or low initial momentum. Fires in the open will be well ventilated (fuel-controlled), but fires within enclosures may become under-ventilated (ventilation-controlled). Pool fires may be static (e.g. where the pool is contained) or 'running' fires. Pool fires represent a significant element of the risk associated with major accidents on offshore installations, particularly for Northern North Sea (NNS) installations that may have large liquid hydrocarbon inventories.
The objectives of this area of fire hazard assessment are:
There are major uncertainties in the behaviour and properties of fires of condensate and higher molecular weight and multi-component materials and very large flames of all materials; behaviour of running fires and of liquids released from pressurised containment
The influence of water deluge and foam on fuel distribution and pool fire mass burning rates and the influence of pool shape on radiation and soot shielding is not well understood.
The effect of scale on fire size, geometry and radiation, particularly for very large fires and the ability to predict the overall behaviour of large hydrocarbon pool fires in offshore structures is poor.
Validation and accuracy of field model applications to offshore compartment fires is questionable.
There is no difference in the burning rate between pool fires on water or steel.
Fuel-controlled pool fires are characterised by rapid rise in temperature (up to 1300 ° C) and high heat fluxes (up to 320 kW m-2) in insulated compartments.
The burning rate (kg s-1 m-2) is not dependent on pool area at large scale and ignition criteria for external flames.
Pool spread characteristics of pool fires on steel plates are not well understood.
Phase I of the JIP:
...included a review of open hydrocarbon pool fire models. Three types of model were evaluated: semi-empirical models (e.g. WHAZAN), field models (e.g. CFD models) and integral models (falling between semi-empirical and field models). It was concluded that well-validated, semi-empirical models represented the best available models for the prediction of heat fluxes to objects outside flames, provided that such models are used within their range of validity.
Compartment fire modelling, have two types of Code: zone models and field models. It was concluded that zone models (typically used for modelling fires within buildings) encounter severe limitations in modelling large offshore compartment fires.
Phase II of the JIP included a fire model evaluation exercise. This considered three jet-fire scenarios, but no pool-fire scenarios. However it did generate high quality data that were considered suitable for future pool fire model evaluation. It has been recognised from Phase II of the JIP that a more extensive fire model evaluation exercise is warranted, involving a greater number of models and test scenarios.
POOLFIRE6 (Rew and Hulbert, 1996), sponsored by HSE has been validated against a wide range of test data and found to give good results for all fuel types except methanol. The model performs as well for smoky, obscured heavy hydrocarbon fuels as it does for clean burning fuels such as LNG. Other pool fire models widely used include that of Mudan and Croce (SFPE, 1995) and the pool fire models contained in Shell's FRED software, DNV's PHAST and NEPTUNE software packages and the 'Yellow' Book (and associated software, EFFECTS).
There is currently no reliable, general method for calculating ventilation-controlled compartment burning rates, and how they respond to changes in enclosure geometry, ventilation and fuel spill size. Without this basic knowledge of the rate of fuel consumption it is impossible to assess the internal temperature and size of external projected flame.
Currently no criteria have been established to determine when pool fires in compartments self-extinguish. Some preliminary data on this were produced by HSL (Atkinson, 2001) for fires in fully welded compartments. These tests also showed that relatively small cracks in key locations could prevent self-extinction.
The large external flames projected from ventilation-controlled fires can threaten escape routes on offshore installations and a reliable method to assess the size and shape of external flames is needed. The key variable in such an assessment is the mass flow of unburned fuel from the fire. Computational methods might provide useful support to experimental work in this area.
External flaming may not occur if the burning hydrocarbon spill is sufficiently distant from the vent and there is no piloting at the vent. In this case unburned fuel vapour is released. The potential for the generation of explosive mixtures remote from ventilation controlled offshore fires should be investigated.
Software packages commonly used for offshore QRA studies include codes such as ARAMAS, NEPTUNE and PLATO. These codes appear only to model open pool fires, which would not represent the particular features of confined or ventilation-controlled fires (e.g. external flaming).