Heskestad VS Maccaffrey Plume

[Firas Shawash]

HI All,
I noticed that there are considerable differences in the results when
switching between Heskestad and Maccaffrey Plumes.
CFAST6.2 user manual and technical reference guide do not refer to
any difference in the results when using any of the plumes.

Can anyone help and explain the variance of the results ?

[Gabriele Vigne]

Hi Firas,

A preliminary quick answer could be that the formulae are different
and consequently the results. If you read the technical reference (pg.
24) you can find both approaches and relatives formulae.

A more refined answer would be that there are a lot of formulae that
pretend to empirically predict the smoke entrainment into the fire
plume, not just Heskestad and McCaffrey. Just to cite a few more,
extensive work has been done by Zukoski, Thomas, Delichatsios and more
recently Harrison for spill plumes.

The real problem is to understand when a formula is more appropriate
for a given scenario and actually this is not as clear as it should
be. Recent research done by the University of Jaén in a large fire
test facility has shown a major difference when comparing the real
data obtained with a heptane pool fire and the analytical results
obtained from different approaches.

This research will be presented in the SFPE conference on Performance-
Based Codes and Fire Safety Design that will be held in Hong Kong,
June later this year (Uncertainty of Smoke Filling Calculations in a
Large Atrium through Comparison to Experimental Results, G.Vigne, C.
Gutierrez-Montes, G. Rein).

Regards,

--------------------
Gabriele Vigne
Senior Engineer
Arup Fire

[CFAST Development]

I'm not surprised they are not identical as they are different emperical correlations.  For typical fires, they do provide a bit different results.  For example, for the default test case supplied with the model, they average difference in upper layer temperature, lower layer temperature, and layer height are 0.27 C, -0.05 C, and 0.03 m, respectively.  If you have cases where they are dramatically different, we'd be interested in them to provide additional testing for our model. 

[CFAST Development]

We'd be very interested in a copy of any paper that becomes available from your presentation.

[Firas Shawash]

I noticed that Heskestad is more conservative in smoke spread, whereas
Maccaffrey is more conservative in the temperature.
Both parameters (Temperature and Smoke spread) are important for
determining the tenable conditions.
I am working on when to use both plumes.

[Firas Shawash]

I think the fact that there is no justification will affect the
acceptability of using CFAST to calculate the tenable conditions.

[Maira Santana]

There's a justifiction. It's all empirical studies, trials to reach a natural behavior for smoke. None empirical formulae will fill all requires on the same way, in CFAST or any avaiable model. CFAST give us the possibility to evaluate the plumes for each case. If you use other model, you will find conditions that are not so tenable as well. I use the same simulation in CFAST, ARGOS, OZONE in all plumes avaiable and ALL results are quite different, as expected. In each one there is a different level of accurance and all them less accurate that CFD models. Each model was create to reach different proposals with different effort, avaiable data, ...

2012/3/7 Firas Shawash <fshawa...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "CFAST" group.
To post to this group, send email to cf...@googlegroups.com.
To unsubscribe from this group, send email to cfast+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/cfast?hl=en.

--
Maira Leal Andrade Santana, Engª
(+55 11) 94 99 45 82

P.S.: Favor confirmar o recebimento deste.

[Gabriele Vigne]

What Maira says is absolutely right.

If you use OZone you will need to choose among 4 different air
entrainment correlations and if you use BRANZFIRE, just to cite other
zone model, you will need to decide which formula is more suitable
for
your scenario choosing between McCaffrey and Delichatsios.
It is part of your work to judge the results of a model and convert
them into proper (wise) design. Models need to be used carefully but
also books need to be read carefully (Enclosure Fire Dynamics, An
introduction to Fire Dynamics, etc.)

Regards,
--------------------
Gabriele Vigne
Senior Engineer
Arup Fire

On Mar 10, 11:12 am, Maira Santana <maira.sant...@gmail.com> wrote:
> There's a justifiction. It's all empirical studies, trials to reach a
> natural behavior for smoke. None empirical formulae will fill all requires
> on the same way, in CFAST or any avaiable model. CFAST give us the
> possibility to evaluate the plumes for each case. If you use other model,
> you will find conditions that are not so tenable as well. I use the same
> simulation in CFAST, ARGOS, OZONE in all plumes avaiable and ALL results
> are quite different, as expected. In each one there is a different level of
> accurance and all them less accurate that CFD models. Each model was create
> to reach different proposals with different effort, avaiable data, ...
>

> 2012/3/7 Firas Shawash <fshawash1...@gmail.com>

[Manuel]

Hello.

The results between Heskestad and McCaffrey do not only differ a bit. As I calculated a typical fire of 6 MW on an area of 20 square meters you can see a lot difference. Only as example the Upper Layer Temperatures. You find more diagrams in an excel-sheet within the zip-folder.

 

You shouldn't wonder about that because McCaffrey doesn't take the fire area in account. With the Heskestad plume, however, you have to care about the area. Since I do, I wondered about another issue. There is the possibility to define the burning area time dependent. As I did I expected to get different results with the Heskestad plume. But this isn't the truth. The results are all the same. The model only takes the constant fire area defined.

 

Can you please give me a hint how to consider a growing design fire. Only assuming a growth in HRR is not appropriate as Heskestad takes the buring area into account for calculating the entrainment. So there will ba a mistake without this function.

 

In the attached Zip-folder you find three fire objects, all with a maximum HRR of 6MW. The first defines an area of 20 m² (4,472 x 4,472). The second one a growing area till 20 m². And a third one with 4m² to test the Plume models.

All had been calculated twice in the same one-room geometry with CFAST 6.1.1, once with Heskestad and once with McCaffrey-Plume. You find all the *.in-files and resultfiles within the zip-folder.

 

Thanks a lot Manuel

[Richard Peacock]

You are correct.  At the moment, it simply takes the fire area as defined in the input file as a characteristic fire area.  This needs to be fixed in the next release of the model; it should be a simple fix.  I've created an issue in the CFAST issue tracker for this at

http://code.google.com/p/cfast/issues/detail?id=85

Rick Peacock

[jdreis-Atkins]

 

since they're all empirical, should have a look at the original experiments that served the basis of the correlations. while they likely have been extrapolated, and non-dimensionalized, i would be wary to apply, say, for example McCaffery correlation to a fire larger than about 500kW. I believe the fires he used were quite small, bench-scale even...