Pool fire models

The model is a straightforward extension of a pool-fire model developed by Steward (1964), and is, of course, a drastic simplification of reality. Figure 5.4 illustrates the model, consisting of a two-dimensional, turbulent-flame front propagating at a given, constant velocity S into a stagnant mixture of depth d. The flame base of width W is dependent on the combustion process in the buoyant plume above the flame base. This fire plume is fed by an unbumt mixture that flows in with velocity Mq. The model assumes that the combustion process is fully convection-controlled, and therefore, fully determined by entrainment of air into the buoyant fire plume. 

EFFECTSGIS—This is a conglomerate package for modeling chemical releases. These releases are linked to appropriate physical phenomena models, such as pool fire models to predict the consequences on humans. The program is equipped with an internal mini-CIS system to enable calculation results, such as heat radiation, to be overlayed onto maps. 

Pool fire modeling is well developed. Detailed reviews and suggested formulas are provided in Bagster (1986), Considine (1984), Crocker and Napier (1986), Institute of Petroleum (1987), Mudan (1984), Mudan and Croce (1988), and TNO (1979). ... 

Pool fire models are composed of several component submodels as shown in Figure 3.32. A selection of these are briefly reviewed here ... 

Risk analyses can include or ignore tilt and drag effects. Flame tilt is more important flame drag is an advanced topic, and many pool fire models do not include this effect. A vertical (untilted) pool fire is often asstomed, as this radiates heat equally in all directions. If a particularly vulnerable structure is located nearby and flame tilt cotald affect it, the CPQRA should consider tilt effects (both toward and away from the vulnerable object) and combine these with appropriate frequencies allowing for the direction of tilt. 

The largest potential error in pool fire modeling is introduced by the estimate for siufacc emitted flux. Where predictive formulas arc used (especially Steflm-Boltzmann types) simple checks on ratios of radiated energy to overall combustion energy should be carried out. Pool size estimates arc important, and the potential for dikes or other containment to be overtopped by fluid momentum effects or by foaming should be considered. 

The physical models described in Chapter 2 generate a variety of incident outcomes that arc caused by release of hazardous material or energy. Dispersion models (Section 2.3) estimate concentrations and/or doses of dispersed vapor vapor cloud explosions (VCE) (Section 3.1), physical c q)losion models (Section 3.3), fireball models (Section 3.4), and confined explosion models (Section 3.5) estimate shock wave overpressures and fragment velocities. Pool fire models (Section 3.6), jet fire models (Section 3.7), BLEVE models (Section 3.4) and flash fire models (Section 3.2) predict radiant flux. These models rely on the general principle that severity of outcome is a function of distance from the source of release. 

To estimate the impact of the pool fire on the nearby personnel, start with the point source model. Assuming a head height target, use a target point at a... 

KAMELEON FIRE E-3D— This model is a program specifically designed to deal with hydrocarbon fires in the form of both liquid pool fires and gas jet fires. 

KAMELEON was validated in the SINTEE enclosed pool fire experiments. Availability of this model is restricted. 

PHAST (Process Hazard Analysis Software Tool)—This is a conglomerate package for gas dispersion and fire modeling. PHAST is capable of calculating the formation of a cloud or pool to final dispersion calculating concentrations, fire radiation, toxicity, and explosion overpressure endpoints. 

The difference between RANS and LES is depicted in Figure 20.1, which shows the temperature fields of a pool fire flame. While the RANS result shows smooth variations and looks like a laminar flame, the LES result clearly illustrates the large-scale eddies. Both results are the correct solutions of the corresponding equations. However, the time accuracy of LES is also essential for the quantitative accuracy of the buoyancy-driven flows. As Rehm and Baum have shown [10], the dynamic motions or eddies are responsible for most of the air entrainment into the fire plumes. Because these motions cannot be captured by RANS, LES is usually better suited for fire-driven flow. LES typically requires a finer spatial resolution than RANS. Examples of RANS-based fire CFD models are JASMINE, KAMELEON [11], SMARTFIRE [12], SOFIE [13], ISIS [14], and ISIS-3D [15]. Examples of LES models are the FDS [4,5] and SMAFS [16], developed at Lund University. Fire simulations using LES have also been performed by Cheung et al. [17] and Gao et al. [18],... 

Jensen, K.A., Ripoll, J.-F., Wray, A.A., Joseph, D., and El Hafi, M. On various modeling approaches to radiative heat transfer in pool fires. Combustion and Flame, 2007. 148(4), 263-279. 

Dembele, S., Zhang, J., and Wen, J.X. Assessments of spectral narrow band and weighted-sum-of-gray-gases models for computational fluid dynamics simulations of pool fires. Numerical Heat Transfer Part B, 2005.48(3), 257-276. 

Hostikka, S., McGrattan, K.B., and Hamins, A. Numerical modeling of pool fires using LES and finite volume method for radiation. Fire Safety Science—Proceedings of the Seventh International Symposium, Worcester, MA, 2003, pp. 383-394. 

ABSTRACT The determination of loads from accidental fires with realistic accuracy in the oil gas industry offshore and petrochemical industry onshore is important for the prediction of exposure of persoimel, equipment and structures to the fires. Standards, Codes of Practice and other similar publications refer to thermal loading from jet fires from 100 to 400kW/m and from 50 to 250kW/m for pool fires. The application of inappropriate fire loads may lead to incorrect predictions of fatalities, explosion of pressure vessels and collapse of structures. Further uncertainties are associated with heat transfer from the flame to pressure equipment and strucmres, and their behaviour when affected by accidental fires. The Paper presents results of a review of fire models from various Standards and Codes of Practice, and data obtained from full scale tests. A parametric study of the various methods used in the industry is presented. A simulation-based reliability assessment (SBRA) method has been applied to quantify potential accuracy range and its consequences to fire effects on structures. 

The review of fire models focuses on the hydrocarbon jet fires, but some pool fire data are also mentioned. 

FRY, C.J., The use of CFD for modelling pool fires . Packaging and Transportation of Radioactive Materials, PATRAM 92 (Proc. Symp. Yokohama City, 1992), Science Technology Agency, Tokyo (1992). 

Two approaches are available for estimating the surface emitted power the point source and solid plume radiation models. The point source is based on the total combustion energy release rate while the solid plume radiation model uses measured thermal fluxes from pool fires of various materials (compiled in TNO, 1979). Both these methods include smoke absorption of radiated energy (that process converts radiation into convection). Typical measured surface emitted fluxes from pool fires arc given by Raj (1977), Mudan (1984), and Considine (1984). LPG and LNG fires radiate up to 250 kW/m (79,000 Btu/hr-ft ). Upper values for other hydrocarbon pool fires lie in the range 110-170 kW/m (35,000-54,000 Btu/hr- ), but smoke obscuration often reduces this to 20-60 kW/m ( 6300-19,000 Btu/hr-ft ). 

The result from the solid plume radiation model is smaller than the point source model. This is most likely due to consideration of the radiation obscuration by the flame soot, a feature not treated direedy by the point source model. The differences between the two models might be greater at closer distance to the pool fire. 

Sodium spray fires The ignition temperature for Na spray fires is lower than for pool fires and can be as low as 120°C depending on the Na droplet size. The sodium oxide aerosol production rates are much higher than for a pool fire (a factor of 5 or even more). Again the various phenomena involved have to modelled in the codes. 

---