Jet fire

A jet fire is a flame sustained by a directed release of combustible material that mixes with surrounding air as it burns. The release can arise from a pressurised leak or from an intentionally operated flare. Its momentum influences the flame's direction and shape, while buoyancy and the surrounding wind also affect its development. Beyler describes these fires as turbulent diffusion flames, meaning that fuel and air mix during combustion.

The distinction between exposure outside the flame and direct flame impingement is central to assessing the hazard. A distant surface may receive predominantly thermal radiation, whereas an engulfed target also experiences strong convective heating. The release direction, flame dimensions and target position therefore matter alongside the total heat release rate. A radiation calculation for a distant target cannot represent every consequence of a nearby jet.

Release conditions and possible fire sequences

Beyler's release event tree distinguishes immediate ignition from ignition after fuel has accumulated or dispersed. Immediate ignition of a release through a relatively small opening can produce a jet flame. The material can be gas or liquid, with a liquid release potentially involving flashing into vapour and accumulation. The flame representation must therefore match the state and behaviour of the released material.

A release can lead to different fire types and a changing sequence. Based on Beyler (2002), pp. 3-268–3-269.

A pool fire involves burning associated with an accumulated liquid surface. A directed jet and an accumulated pool can arise within the same release sequence, but their geometries differ. Delayed ignition of a dispersed cloud can also propagate back towards the release and establish subsequent steady burning. Treating the initial event and the continuing fire as separate stages preserves these differences.

An intentionally operated flare shares important flame characteristics with an accidental jet fire. However, flare composition, outlet diameter and stack height may be known design inputs. For an accidental release, those inputs can require estimation from the failed equipment and release conditions. Similar combustion physics does not mean that the source uncertainty is the same.

The EPRI/NRC methodology provides examples involving hydrogen relief valves, valve packing and flange leaks. It distinguishes jet flames from tank explosions and delayed cloud ignition. The report was prepared for fire risk assessment at US nuclear power facilities. Its scenario categories can clarify the sequence being considered, while its numerical screening assumptions remain tied to that purpose.

The distinction also matters when considering a boiling liquid expanding vapour explosion. A sustained jet exposure and a sudden vessel failure are different stages or outcomes. Beyler describes accidents in which burning released fuel heated neighbouring liquefied-gas vessels before later failures. A calculation of the original jet therefore does not automatically describe every subsequent exposure.

Momentum, buoyancy and flame stability

Fuel emerging from an opening carries momentum. Mixing with surrounding air supports combustion and changes the moving stream, while hot gases experience buoyancy. Depending on their relative influence, the flame can be predominantly momentum-controlled or buoyancy-controlled. Beyler consequently separates the corresponding flame-length relationships instead of assigning one universal length to a given release diameter.

At relatively low release velocities, the flame can remain attached near the outlet. Increasing velocity can cause its base to lift away. At still higher conditions, a stable flame may cease to be maintained and blowout can occur. An obstruction can also stabilise a flame, so the arrangement around the release belongs in the assessment.

A lifted flame does not mean that fuel has stopped being released. Its burning region is displaced from the opening. The handbook explains this position through the balance between local flow velocity and turbulent burning velocity. Changes in release conditions can move the flame base to another location where that balance is possible.

These mechanisms limit simple comparisons based only on outlet velocity. The same velocity does not uniquely determine flame length, attachment or radiation for every fuel and opening. Density, burning characteristics and the flow regime also enter the cited relationships. A prediction should retain those conditions instead of treating a laboratory observation as a universal jet-fire rule.

Wind, geometry and the location of exposure

Crosswind can bend a jet flame and change its length and width. The handbook reviews experiments with several fuels and distinguishes vertical flame height from distance along an inclined flame. These geometric quantities answer different questions. A height above the outlet cannot be used directly as the full length of a tilted flame.

The correlations described in the chapter have stated experimental ranges. One group of wind-tunnel tests used burner diameters from 6 to 22 mm, jet velocities from 13 to 200 m/s, and crosswinds from 2.6 to 8.1 m/s. These conditions identify the supporting evidence. They do not establish equal accuracy for every large accidental release.

Release direction can determine which equipment lies within the flame path. The EPRI/NRC hydrogen assessment explicitly considers orientation relative to potential targets. For a relief valve, the discharge direction may constrain the possible flame direction. For a less clearly defined leak, the assumed direction requires its own scenario basis.

Distance also requires a defined reference. In the point-source radiation model, it is measured from a representative point near the flame centre. Distance from the pipe opening is generally a different quantity. A long flame can therefore place the representative radiating region substantially closer to a target than the release point itself.

The receiving surface orientation further affects radiation. A surface facing the representative source receives a different projected exposure from the same surface turned obliquely. This geometric effect is separate from any change in flame power. Source location and surface orientation both need definition before numerical distances can be compared.

One historical hydrogen incident in the EPRI/NRC appendix illustrates the importance of localisation. Direct impingement severely damaged equipment near the release, while damage decreased farther away. A later release through an excess-flow valve directed flame towards a fence. The account identifies changing release geometry rather than a uniform exposure around the original source.

Radiation models and their limits

Flame radiation depends on the amount of energy emitted and the geometric relationship to the target. A simple model replaces the flame by a point at its centre. Beyler gives the following expression for a target outside the flame, together with the relation between radiated power and total heat release.

q̇″ = Q̇_r cos θ / (4πR²)

Q̇_r = χ_r Q̇

Symbol	Meaning	Unit
q̇″	Incident radiative heat flux in the simplified model	kW/m²
Q̇r	Total power emitted as radiation	kW
Q̇	Total heat release rate	kW
χr	Fraction of heat release emitted as radiation	dimensionless
R	Distance from the representative flame centre to the target	m
θ	Angle between the surface normal and direction towards the source	degrees, used consistently in the cosine

This expression distributes radiant power over a spherical area and includes the receiving surface's projected orientation. It does not resolve how emission varies along the flame. The chapter reports useful comparisons for flare radiation, but also identifies errors close to the source. The point approximation therefore requires separation large enough for the finite flame geometry to be adequately represented.

A solid-flame model retains a simplified emitting shape, such as a tilted cylinder. Its result depends on surface emissive power, a geometric configuration factor and atmospheric transmission. A line-source model instead sums contributions along the flame length. These approaches expose different geometric detail; none eliminates the need to justify the flame and emission inputs.

The radiative fraction is not a universal constant for a fuel. Beyler reviews effects of combustion efficiency, soot, mixing and heat transferred to entrained air. The chapter also reports changes with jet velocity and crosswind. Laboratory trends provide evidence of these influences, but their quantitative transfer to large flames requires care.

Atmospheric attenuation is another separate influence. The basic point expression shown here contains no transmission factor. Its use therefore embodies that simplification. A more detailed assessment must retain consistency between emitted power, any attenuation model and the incident quantity being reported.

The spatial distribution of emission is another source of approximation. Propane experiments reviewed by Beyler found the strongest radiation near the flame centre, decreasing towards either end. A uniformly emitting cylinder discards that variation. A line representation can retain variation along its length, provided the assigned distribution has supporting evidence.

The chapter also describes a three-dimensional model incorporating flow, turbulence, combustion and soot. Its reported comparison with field experiments used methane jets up to 254 mm in diameter and obtained heat-flux agreement within 15%. That result belongs to the identified model and experiments. It is not a general accuracy statement for every computational jet-fire model.

Worked radiation comparison

Consider an illustrative 10 MW jet with an assumed radiative fraction of 0.20. The handbook lists a value of 0.20 among the reported methane results, but also explains its dependence on operating conditions. It is selected here solely to demonstrate the equation. It is not a recommended fraction for every methane release.

Distance and orientation change the calculated far-field radiation. Based on Beyler (2002), pp. 3-298–3-299.

The assumed radiated power is 2,000 kW. For a target facing the representative source at 50 m, the equation gives approximately 0.0637 kW/m². At 100 m, the result is approximately 0.0159 kW/m². Doubling distance reduces the calculated flux to one quarter because the spherical area increases with the square of distance.

Keeping the target at 50 m but turning its normal through 60° halves the result to approximately 0.0318 kW/m². The calculation uses the same emitted power and changes only projected orientation. These three cases demonstrate the distance and angular terms independently. All intermediate calculations retain unrounded values.

The example does not establish that an actual 10 MW jet has a particular flame length. It assumes that the chosen positions are sufficiently far from the flame for the point representation. Nor does it calculate atmospheric attenuation, direct contact or the thermal response of a target. Those omissions define the scope of the reported values.

A numerical flux is also not automatically a consequence criterion. The target's material, duration of exposure and response require separate treatment. The handbook frames its radiation methods as inputs to an assessment of particular impacts. A distance derived from the equation needs that additional basis before it can be interpreted as a protection distance.

Direct impingement and structural exposure

An impinging jet heats the target through both radiation and convection. The handbook reports particularly severe exposures where large-scale jet flames engulfed cylindrical targets. In the cited tests, average heat fluxes over the impingement area were approximately 200 kW/m² for sonic natural gas and 150 kW/m² for two-phase liquefied petroleum gas. These are results for those experimental arrangements.

Direct impingement combines radiative and convective heating. Based on Beyler (2002), pp. 3-301–3-302.

The corresponding reported peaks reached approximately 300 and 250 kW/m² respectively. A peak local value differs from an average over the target area. Neither value describes every location on a vessel or every release condition. Retaining the fuel state and target geometry prevents those measurements from becoming universal design inputs.

Beyler describes roughly comparable radiative and convective contributions in some high-flux regions. Other gaseous propane tests reported a different balance. A radiation-only estimate therefore cannot generally reproduce total impingement heating. The relevant exposure includes the local gas motion and thermal environment as well as radiation.

A prescribed hydrocarbon fire curve describes a temperature–time exposure associated with oil or gas fire testing. Van Rossum distinguishes such agreed histories from temperatures developing in an actual scenario. The curve and the local jet-impingement flux consequently describe different inputs. Substituting one for the other requires a thermal assessment rather than a change of label.

Mitigation, continuing release and assessment scope

Water application can affect exposure without necessarily extinguishing the jet. Beyler reviews experiments showing reduced radiation at water-spray rates below complete suppression. Fleming likewise distinguishes cooling of exposed construction from successful extinction. A reduction in measured radiation is therefore a different result from stopping fuel flow or eliminating the flame.

A deluge system supplies open discharge devices after a separate detection system operates its valve. That system description does not specify its effectiveness against a particular jet. Fleming emphasises the empirical basis and limits of suppression calculations. The cited jet-spray observations therefore cannot supply a general deluge application rate or guaranteed performance.

For hydrogen safety, continuing release and ignition sequence remain particularly relevant. The EPRI/NRC report explains that loss of a flame while gas continues escaping can permit a flammable cloud to form. Its discussion of source isolation reflects that change in scenario. Extinction and removal of the fuel source are consequently distinct outcomes in the hazard analysis.

The final assessment needs a defined release, flame representation, target geometry and exposure duration. It also needs the applicable model limitations and supporting experimental conditions. The EPRI/NRC report leaves selection and validation of detailed computational tools outside its hydrogen scenario rules. Its screening assumptions cannot replace a validated thermal model for a different application.

See also

• Fireball

References

• Beyler, C. L. (2002). Fire hazard calculations for large, open hydrocarbon fires. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., Chapter 3-11). National Fire Protection Association.
• Electric Power Research Institute, & U.S. Nuclear Regulatory Commission. (2005). EPRI/NRC-RES fire PRA methodology for nuclear power facilities: Volume 2, Detailed methodology (EPRI 1011989; NUREG/CR-6850).
• Fleming, R. P. (2002). Automatic sprinkler system calculations. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 4-72–4-87). National Fire Protection Association.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering.