Simulating Gas Jet Fire with FDS

[doug.w....@exxonmobil.com]

I’m trying to simulate a horizontal jet fire of a light gas in order to see
the location and size of the convection zone around the tip of the flame.
The fire is essentially in the open, although there is a grated steel deck
5 m above the release point. Can a highly porous object (steel grating) be
simulated? I’ll settle for an open-area simulation. I want to predict
heat fluxes on the deck at 5-m elevation, and at that elevation convection
dominates near where the end of the flame turns upward and crosses that
deck. Is this kind of simulation feasible with FDS, and am I approaching
it at all correctly?

The gas contains ~85 mol% methane and has a molecular weight of ~19.7
kg/kgmole. The release is ~23.5 kg/s through a hole with an area
equivalent to a diameter of 100 mm. I think I have to simulate the
opening as square. The gas velocity is high (sonic), and I understand
that a large difference between the gas velocity and the surrounding air
velocity (5 m/s) can be problematic. Do I need a finer mesh near the
jet zone? Should I take advantage of symmetry about the y-axis, perhaps
by specifying only one mesh cell in the y direction? It’s really a 2-D
problem, x vs z (elevation). The characteristic fire diameter D*
defined in the User Guide is ~16 m, but it’s not obvious how to use that
in selecting mesh sizes, since a jet fire is so long and thin. An
empirical correlation predicts a curvilinear flame length of ~42 m.

I also have some cases with lower flow rates, where the flame is shorter
and turns up at the end but doesn’t touch the deck at 5-m elevation
(based on a simple model).

I’m sending the input file and the error message I received, which says
"Program Exception - access violation". FDS appears to pick up some but
not all of the specified input parameters, and it creates some output
files.

Thanks very much for any help you can provide.

(See attached file: horiz_jet_fire.fds) (See attached file:
FDS_screen_capture.doc)

Doug Hissong
ExxonMobil Upstream Research Company
Phone (713) 431-4517 Fax (713) 431-6387

[drjfloyd]

Have you read the FDS documentation:

Theory Manual 2.1.8 Limitations

Low Speed Flow Assumption - The use of FDS is limited to low-speed
flow with an emphasis on smoke
and heat transport from fires. This assumption rules out using the
model for any scenario involving
flow speeds approaching the speed of sound, such as explosions, choke
flow at nozzles, and detonations.

Your inlet boundary condition is 3000 kg/m^2/s of a 19 g/mol gas at
ambient temperature and pressure. This is Mach 10.

>  horiz_jet_fire.fds
> 4KViewDownload
>
>  FDS_screen_capture.doc
> 167KViewDownload

[drjfloyd]

Also it should be noted that you have specified MASS_FLUX(1) which
means FDS is going to be looking for SPECies(1), but you have not
defined any additional species in this input file as both the SPEC
line and the READ line have been commented out.

On Jun 15, 11:04 am, doug.w.hiss...@exxonmobil.com wrote:

>  horiz_jet_fire.fds
> 4KViewDownload
>
>  FDS_screen_capture.doc
> 167KViewDownload

[Guillermo Rein]

The low Mach number limitation for this problem does not affect the
results if the physical domain is split into two zones. One zone where
(Mach number) Ma>0.3 and another where Ma<0.3. A sonic jet expands
rapidly and its velocity drops to Ma<0.3 in a short distance (in the
order of 0.5 m) from the discharge nozzle. It happens that for most
cases of interest, the Ma>0.3 zone falls within the lift-off zone,
where the flame has not yet established. Thus, a CFD model of the
flame in the second zone could be attempted with FDS.

The research group at CERTEC (http://certec.upc.es/eng/main.html) is
working on this problem (I am helping them a bit). They have performed
extensive experiments with propane flames up to 10 m long (presented
at the recent Sixth Mediterranean Combustion Symposium in Corsica) and
are now simulating these using FDS. It is a very challenging task but
results are promising. If you are interested in their progress, please
contact them, or contact me and I will direct you to them.

Cheers
Guillermo Rein

[rmcdermo]

I think that both Jason and Guillermo have good points: Jason's that
this problem needs careful consideration and Guillermo's that the
problem is possible to solve with FDS given that the flow slows
relatively quickly.

In a previous life, I designed low-NOx burners with sonic nozzles.
Though I never got to try it, my thought was that the near field of
the jet could be simulated using a compressible RANS code and that
this could supply the bc for a low-Mach LES code (e.g. FDS) by writing
out the velocity, species, and temperature profiles at a plane
downstream of the jet.

Another issue that needs to be addressed is what combustion model will
be used. With sonic nozzles, as Guillermo was saying, the jet lifts
off and stabilizes where the flame speed equals the speed of the fuel/
air mixture, which at that point is premixed to a large degree. FDS
is not designed for turbulent premixed or partially premixed flames.
So, attacking this problem would be a research effort (as I suppose
CERTEC is investigating), but I certainly think it is possible.
However, it may be the case that the premixed region is also
relatively short-lived in the context of calculating heat fluxes at a
distance 5 m from the nozzle.

Cheers,
Randy

[MGM]

I'm part of the CERTEC group. We are working on vertical propane jet
fires. Guillermo has helped us to find an alternative way to be able
to use FDS (thank you very much!)... It is true that FDS could be
useful for simulating jet fires, but we continue working on that...
so, maybe we can talk about that if you are interested, I think it
would be useful for you and us.

Cheers

Mercedes G.

[rmcdermo]

Mercedes,

I am definitely interested in your approach, especially any
modifications to the turbulent combustion model.

Best,
Randy

[Guillermo Rein]

Randy,

This study relates to risk in the process industries and is an
engineering study, as oppose to an accurate and detailed simulation of
all the mechanisms in the 10-m long jet fire (a fundamental work like
that would be rather unrealistic in less than three years time, I
think). This first attempt will not include modifications to the
combustion model of FDS, but maybe the results could be used to guide
future changes to the code. Instead the current objective is know
whether FDS can reproduce approximately the most important risk-
related characteristics, like flame length, maximum temperature and
radiation emission power.

G.