CFAST Software Limits

[tim]

Hi

I'm a researcher at a UK university who's providing a bit of
assistance to an MSc student modelling smoke propagation in a ship
using CFAST. The student's been able to pick up CFAST very quickly and
then successfully link it to a ship specific design tool. However,
she's recently hit some the limitations of CFAST. I through a post
here might prompt useful discussion...

The problem arose due to CFAST's scenario and software limits (as
described on page 56 of the user's guide). The student had intended to
to produce a model of a ship containing a large number of compartments
and vents (about 70-100 of each). After experiencing problems running
CFAST she discovered the limits on model size. I had a quick poke
about the CFAST source code and couldn't see any obvious reason for
the limits (however my Fortran is very rusty).

Given the large quantity of design information available from CAD
systems I was curious if you had considered revisiting the limits on
model size? Is it more beneficial to spend cheap CPU time running a
model with extra compartments as opposed to expensive user time
reducing the model size?

No particularly urgency need for a response, just though the feedback
may be informative and prompt a useful debate.

Many Thanks

Tim

[Barbro Maria Storm]

2009/8/26 tim <t.p.mc...@gmail.com>:

>
> Hi
>
> I'm a researcher at a UK university who's providing a bit of
> assistance to an MSc student modelling smoke propagation in a ship
> using CFAST. The student's been able to pick up CFAST very quickly and
> then successfully link it to a ship specific design tool. However,
> she's recently hit some the limitations of CFAST. I through a post
> here might prompt useful discussion...
>
> The problem arose due to CFAST's scenario and software limits (as
> described on page 56 of the user's guide). The student had intended to
> to produce a model of a ship containing a large number of compartments
> and vents (about 70-100 of each).

It's not good for this level of compartments. How about FDS?

--
Barbro Storm

[tim]

On Aug 26, 8:38 pm, Barbro Maria Storm <bar...@gmail.com> wrote:
> 2009/8/26 tim <t.p.mcdon...@gmail.com>:

A previous student used FDS as part of a very successfully MSc
project. However, CFAST has huge advantage in terms of run time.

For a tool that a designer could quickly run to examine possible
options while developing and modifying a design a two compartment
model like CFAST appears to be a winner. :-)

Tim

[Jonna]

How about splitting the simulated area in different models/
simulations?
Or do you require to simulate the whole ship in "one model"?

> Tim- Hide quoted text -
>
> - Show quoted text -

[CFAST Development]

At the moment, CFAST is dimensioned to 30 compartments and 100
horizontal flow vents. This is set in the include file cparams.fi.
While there is nothing magic about the number of compartments (it
could easily be resized and recompiled for 100 compartments, for
example), there is little if any validation for a simulation with such
a large number of compartments.

This lack of validation is true for most any model, be it CFAST, FDS,
or any other model. Still, that doesn't imply that it can't be used
either. This is an area where the use falls back on the standard
engineering judgement for the applicability of its use and validity
(for example, we have tested 100 compartments in series and the
results are as would be expected, but without experimental data of
this magnitude, we rely on our judgement of whether the answers are
reasonable or not).

If there is a desire for a larger number of compartments, we can
certainly make a change to the code base with caveats to the lack of
evaluation for larger test scenarios.