Encyclopaedia Index

6. CFD applied to fire-spread phenomena

Contents


    6.1 A typical scenario
    6.2 The physical and mechanical processes considered
    6.3 How PHOENICS accounts for the processes
    6.4 Some typical findings

6.1 A typical scenario

Let it be supposed that, possibly as a result of an explosion, followed by rupture of gas and oil pipes, a fire starts, and spreads through an oil platform.

Let it be further supposed that fire-fighting devices are activated.

How well can the spread of the flame, and the success of the efforts to contain and extinguish it, be predicted by CFD techniques in general and by EXPLOITS in particular?

Will it be possible, with the aid of EXPLOITS, to improve the design of the fire-fighting measures?

These are the questions which will now be addressed.

6.2 The physical and mechanical processes considered

(a) Fuel-supply rates

Whereas the gas-dispersion process is slow, and may be treated as a steady-state one, while on the other hand the explosion and blast phenomena are extremely rapid, the fire-spread process proceeds at an intermediate rate.

Explosion and blast occupy milli-seconds, whereas fires may last for hours.

Crucial to the development of the fire are the rates of supply of fresh gas and oil to the conflagration. These are to be regarded as INPUTS to the CFD calculation.

However, the rate at which the liquid fuel vaporises, so entering the gaseous phase in which alone significant exothermic reaction can occur, is an OUTPUT of that calculation; for it depends on the flow of heat and fluids.

(b) Radiation

It is for this reason that RADIATIVE heat transfer assumes an importance in fire-spread simulation which it possesses in none of the other three hazards which EXPLOITS deals with.

Heat radiated from the flame to the liquid-fuel surface, and to the metal surfaces oveer which it flows, substantially controls the rate at which the oil vaporises,

It is essential, therefore, to include radiative transfer in the simulation procedure,

This entails that one must calculate not only the distributions of gas composition and temperature throughout and surrounding the platform, but also the SOOT concentration at each point; for that is what mainly affects the radiation.

(c) Chemical kinetics

To calculate the soot concentration, it is necessary to have quantitative data on the kinetics of the relevant chemical reactions; and these must be used in a manner which takes account of the fluctuations of gas condition resulting from turbulence.

The chemical-kinetic data are not yet very securely established.

Therefore, influenced also by the fact that fuel-supply conditions are themselves never much better than guesses, the modeller of fires may well decide NOT to use precious human and computer time on soot kinetics,

If he does take a short cut, it will usually be by guessing the emissivity of the flame; but certainly not by neglecting radiation.

(d) Water-spray and foam effects

Fires may be extinguished by the injection of foam, or of sprays of water, the effects of which therefore require to be simulated.

Both foam and sprays necssitate the activation of two-phase- simulation capabilities, in the first case without, and the second case with, relative motion between the liquid and the gaseous materials.

In both cases, the latent heat of vaporisation of the liquid has to be taken into account.

6.3 How PHOENICS accounts for the processes

PHOENICS possesses (and therefore EXPLOITS does also) all the necessary equipment for simulating the processes and phenomena just described.

Of particular importance is the new IMMERSOL technique, which makes it possible, without inordinate expense, to handle the radiative, convective and in-solid-conductive processes simultaneously.

The multi-fluid model of PHOENICS, in particular, makes it possible to account for the influence of turbulent fluctuations on smoke production, as the following gas-turbine related picture shows.

The multi-fluid model predicts a different soot distribution from that which a conventional (ie single-fluid) nodel would predict; and there is every reason to presume that the former is the more correct. Soot-concentration distributions compared; multi-fluid above; single-fluid below.

The differences are not great in this case; but sometimes they can be significant.

Two-phase effects

PHOENICS is well-supplied, as has aready been mentioned, with means of simulating two-phase phenomena, having been indeed the first general-purpose code to do so, in the early 1980's.

To illustrate this, it may suffice to show one more picture from the spray-drier sequence introduced above.

This now follows.

Vaporisation-rate distribution

6.4 Some typical findings

CHAM did not participate in the SCI's fire-spread exercise, for lack of resources.

It is therefore not in a position to present "typical findings" at the present stage.

Nevertheless, it is very desirable to establish whether or not the (as CHAM believes it to be) most powerful fire-spread simulator in existence performs well in the circumstances of the SCI trials.

Funding for such a study is now being sought.

It should however also be mentioned that PHOENICS users other than CHAM have used the code for fire-spread simulations of various kinds; and that the JASMINE code employed by the Fire Research Station is, in essence, an early version of PHOENICS.

wbs