Simple configuration - numerical instability

[Victor Shestopal]

Dear colleagues,

 

Attached is a very simple file. During its execution the time difference between the steps becomes shorter and shorter, and the execution breaks in "numerical instability" at step 4726 which is 178.73 seconds. The attempt to use a statement BAROCLINIC=.FALSE. does not bring any improvement.

 

Regards

 

Victor Shestopal
FIRE MODELLING & COMPUTING
Wahroonga, NSW 2076

 

 

[Tony Bova]

Hello-
Only because I've been working with them recently, I thought about the
Lagrangian particles in your simulation. At 60 s, there were almost
half a million in your simulation domain. When I removed the particle
lines from the input file, the simulation ran much faster (because it
wasn't tracking half a million or more every time step), and ran to
200 s (my TWFIN) so far without instability. You may want to change
the insertion times of the particles to every 60s, or perhaps even
longer, for such a long simulated time. That doesn't guarantee an
instability won't occur later, but it may help.
Cheers,
~tony


On Jan 12, 11:41 pm, "Victor Shestopal" <firec...@optusnet.com.au>
wrote:

>  MYLOGO.bmp
> 3KViewDownload
>
>  H2.fds
> 6KViewDownload

[rmcdermo]

Tony brings up a good point.  You can also control the number of particles with AGE.

I ran the input file (with and without BAROCLINIC) with SVN 7031 (FDS 5.5.3) and have not seen any problems out to 200 s either.  Please run with the latest code and make sure you are still seeing problems.  If so, then submit an Issue.  Strangely, I get a seg fault with OSX 64, but OSX 64 debug, OSX 32, and linux 32 all work fine.  If we need to keep working this case, let's do it on the Issue Tracker.

On a side note to anyone who is listening...  I am on a campaign to discourage the toggling of BAROCLINIC as a stability fix.  I know in the past this has been standard practice, but it is always a red herring.  If the case is unstable with BARO, then there is something else wrong.  I understand we don't always have time to figure out what that something else is, but perhaps we will look harder if BARO isn't a crutch.  Thanks.

[Kevin]

I get the same instability as you at around 180 s. I suggest the
following:

1. Use grid cells that are close to cubes.
2. Add a ZONE line to account for the fact that the compartment is
sealed until the windows break.
3. Do not use OPEN as the SURF_DEFAULT. Use WALL instead and remove
the WALL VENT lines.
4. Watch the background pressure. It was rising to about 2.5 atm by
the time the windows started to break. This would be an obvious cause
of numerical instability because in the model, a very rapid discharge
of air/smoke can cause instabilities. You may try building the
enclosure completely within the computational domain. Putting the
window at the boundary of the computational domain may not produce
accurate results.
5. You may want to add an OPEN boundary somewhere else to remove the
very large pressure rise. Maybe a LEAK would suffice.

[Pete Muir]

I have just run the sim in the background on my laptop (Dell Inspiron 9400,
Intel Core 2 2GHz, 1GB RAM) and managed to pass the second window breakage
without instability.  Detatched flame region seen as expected local to windows
when they are removed.

Perhaps the instability is processor specific?

Pete

--
You received this message because you are subscribed to the Google Groups "FDS
and Smokeview Discussions" group.
To post to this group, send email to fds...@googlegroups.com.
To unsubscribe from this group, send email to
fds-smv+u...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/fds-smv?hl=en.

[Chris]

If I would like to examine a phenomena like window breakage I would
not put the compartment exactly at the mesh boundary. And your two
meshes are not appropriate at all.

Chris

> For more options, visit this group athttp://groups.google.com/group/fds-smv?hl=en.- Zitierten Text ausblenden -
>
> - Zitierten Text anzeigen -

[Kevin]

Numerical instabilities can be somewhat random. For the reasons I
stated above, the calculation was not very robust. Given that we are
simulating turbulent phenomena, no two calculations (like no two
snowflakes) are exactly alike. One random excusion of a velocity
component in one grid cell may trigger the instability. This might
happen on one machine and not another. Most analyses with FDS involve
multiple simulations. If numerical instabilities occur often, I try to
improve the basic scenario formulation to minimize the likelihood. In
other words, you might dodge the bullet with one calculation, but
eventually you will get hit.

> For more options, visit this group athttp://groups.google.com/group/fds-smv?hl=en.- Hide quoted text -
>
> - Show quoted text -

[Thaligari sandeep]

Hello sir ,This is Thaligari Sandeep Kumar, I am very thankful for your earlier suggestions, I am now working on atriums (An atrium is “a large volume space created by floor openings connecting two or more stories and is closed at the top.”). Now here I want make the input file for this atrium (which is in cylindrical and cone shape at the top ) to study the smoke movement by using FDS, So kindly suggest me in what way I can proceed to  make use of FDS.

[mohamed assal]

Hello Sandeep

i think that you should use pyrosim or AutoCAD-fds.for Atrium
cylindrical shape.

i suggest you sandeep that next time try to post your own topic
because your topic has nothing similar to victor's topic Simple
configuration - numerical instability.

Cheers Mohamed.

[Thaligari sandeep]

Hello sir, thank you very much for your suggestion sir, this is my own topic only, kindly tell me how can i proceed with Auto-CAD-fds bcoz it is available for me.

[mohamed assal]

hello sandeep

this is pavel page Autocad to fds plugin

http://acad2fds.codeplex.com/

good chance

On 17 jan, 16:15, Thaligari sandeep <sandyo...@gmail.com> wrote:
> Hello sir, thank you very much for your suggestion sir, this is my own topic
> only, kindly tell me how can i proceed with Auto-CAD-fds bcoz it is
> available for me.
>
> On Mon, Jan 17, 2011 at 1:37 PM, mohamed assal <
>

> mohamedassalengineer...@gmail.com> wrote:
> > Hello Sandeep
>
> > i think that you should use pyrosim or AutoCAD-fds.for Atrium
> > cylindrical shape.
>
> > i suggest you sandeep that next time try to post your own topic
> > because your topic has nothing similar to victor's topic  Simple
> > configuration - numerical instability.
>
> > Cheers Mohamed.
>
> > On 17 jan, 07:02, Thaligari sandeep <sandyo...@gmail.com> wrote:
> > > Hello sir ,This is Thaligari Sandeep Kumar, I am very thankful for your
> > > earlier suggestions, I am now working on atriums (An atrium is “a large
> > > volume space created by floor openings connecting two or more stories and
> > is
> > > closed at the top.”). Now here I want make the input file for this atrium
> > > (which is in cylindrical and cone shape at the top ) to study the smoke
> > > movement by using FDS, So kindly suggest me in what way I can proceed to
> >  make
> > > use of FDS.
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "FDS and Smokeview Discussions" group.
> > To post to this group, send email to fds...@googlegroups.com.
> > To unsubscribe from this group, send email to

> > fds-smv+u...@googlegroups.com<fds-smv%2Bunsu...@googlegroups.com>
> > .

[Thaligari sandeep]

hello sir,

thank you for giving me the fds plug in but when  I am installing it along with auto-cad 2011 version ,it is saying that you have no supported auto cad version installed and ultimately the installation is cancelled. so plz tell me which version of auto cad  I should go to install your plug in ..

To unsubscribe from this group, send email to fds-smv+u...@googlegroups.com.