numerical instability in with multiple meshes

[fire-tom]

Hello everybody,

I was running the following test_case
http://code.google.com/p/fds-smv/source/browse/trunk/FDS/trunk/Test_cases/Atmospheric_Effects/lee_waves.fds

It contains the following mesh:

&MESH IJK=200,1,50, XB=0.0,10000.0,-1.0,1.0,0.0,2500.0 /

-->everything seems to work (6000s, -55Pa<pressure<95Pa).


When I try to define two meshes

&MESH IJK=100,1,50, XB=0.0,5000.0,-1.0,1.0,0.0,2500.0 /
&MESH IJK=100,1,50, XB=5000.0,10000.0,-1.0,1.0,0.0,2500.0 /

first the system behaves well. After about 3000s, the system begins to
oszillate very quickly (-3500Pa<pressure<3500Pa). It doesn't matter
wether MPI is used or not. The border of the meshes becomes visible
because the oszillation is not synchronous in the two meshes.

My first fds-goal is to reproduce nice fluid dynamics patterns like
karman vortices and to use all cores of my notebook for this.

Can anybody tell me, what must be considered, when meshes a splitted?
I tried using a finer mesh, but no way.
Is the 2-dimensional nature of the system the reason for the instable
behaviour?

Thanks for any help,
greetings,

Thomas

[SimonFrey]

Dear Thomas,

In this case, you have to add:

&MISC PRESSURE_CORRECTION=.TRUE./

Like this, the pressure field in the two meshes is linked and
information on pressure perturbations get transferred between the two
meshes.

Best Regards
Simon

On Aug 31, 10:19 pm, fire-tom <thomas.rimm...@gmail.com> wrote:
> Hello everybody,
>

> I was running the following test_casehttp://code.google.com/p/fds-smv/source/browse/trunk/FDS/trunk/Test_c...

[Alexander Ramlow]

Thomas,

Correct me if I'm wrong, but it looks like your mesh is only one grid cell thick!?! I would increase the number of cells in the y-direction descibed by JBAR or the J component in the MESH line. depending on which FDS version you are going to use. Looks like FDS5 here, but just in case you or someone might try FDS4 as well.

I would use at least 10 cells in either direction of the space of the mesh.

Cheers

Alex

[Kevin]

One cell thick in the y direction is how you do a 2-D calculation.

And, yes, PRESSURE_CORRECTION=.TRUE. is needed in a case like this.

> >> Thomas- Hide quoted text -
>
> - Show quoted text -

[fire-tom]

Thanks for the answers!

Of course I added the PRESSURE_CORRECTION immediately. What happened
was the following:
When using the command
fds5 leewaves.fds
I get nice leewaves, no signs of two meshes, no global oszillation.

When using the command
mpiexe -n 2 fds5_mpi leewaves.fds
the first grids has a more or less stable pressure (5 * 10^4 Pa)
the second grid looses pressure (5 * 10^4 Pa --> -2*10^6 Pa)
and the calulation aborts after about 600 sec.
See output and setup below.

Greetings,
Thomas

SETUP:
http://code.google.com/p/fds-smv/source/browse/trunk/FDS/trunk/Test_cases/Atmospheric_Effects/lee_waves.fds
with the following changes
...

&MESH IJK=100,1,50, XB=0.0,5000.0,-1.0,1.0,0.0,2500.0 /
&MESH IJK=100,1,50, XB=5000.0,10000.0,-1.0,1.0,0.0,2500.0 /

&TIME T_END=6000.0 /

&MISC
PRESSURE_CORRECTION=.TRUE.,U0=5.0,RADIATION=.FALSE.,LAPSE_RATE=0.005 /
...


OUTPUT:

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2500.0
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2500.00691
0.00000
-0.00038
0.00000
0.00000
0.00000
0.00000

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.7
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2486.6
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2486.61293
0.00000
9.71381
0.00000
0.00000
0.00000
0.00000

Time Step: 1, Simulation Time: 1.58 s
-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.7
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2497.6
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2497.58737
0.00000
2.69770
0.00000
0.00000
0.00000
0.00000

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.4
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2496.8
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2496.82806
0.00000
3.39625
0.00000
0.00000
0.00000
0.00000

Time Step: 2, Simulation Time: 3.16 s
-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -11.0
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2497.4
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2497.36688
0.00000
-10.96469
0.00000
0.00000
0.00000
0.00000

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -10.5
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0-2497.2
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
-2497.23825
0.00000
-10.54504
0.00000
0.00000
0.00000
0.00000

Time Step: 3, Simulation Time: 4.74 s
-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -9.6
...
...
...
-2602.40679
0.00000
3.91254
0.00000
0.00000
0.00000
0.00000

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0*******
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0*******
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
************
0.00000
************
0.00000
0.00000
0.00000
0.00000

-100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0*******
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0*******
100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-300.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
IERROR= 103

0.00000
************
0.00000
************
0.00000
0.00000
0.00000
0.00000

Time Step: 422, Simulation Time: 665.66 s

STOP: Numerical Instability

STOP: Numerical Instability

[Kevin]

PRESSURE_CORRECTION is still something that we are working on. It
appears that you've found a case where it is not working. I suspect it
has to do with the atmospheric stratification. Could you start an
Issue Tracker for this. Post the exact same input file that has given
you trouble, and note your FDS version numbers. Then we can work out
the bug step by step. Thanks.

On Sep 1, 5:13 pm, fire-tom <thomas.rimm...@gmail.com> wrote:
> Thanks for the answers!
>
> Of course I added the PRESSURE_CORRECTION immediately. What happened
> was the following:
> When using the command
> fds5 leewaves.fds
> I get nice leewaves, no signs of two meshes, no global oszillation.
>
> When using the command
> mpiexe -n 2 fds5_mpi leewaves.fds
> the first grids has a more or less stable pressure (5 * 10^4 Pa)
> the second grid looses pressure (5 * 10^4 Pa --> -2*10^6 Pa)
> and the calulation aborts after about 600 sec.
> See output and setup  below.
>
> Greetings,
> Thomas
>

> SETUP:http://code.google.com/p/fds-smv/source/browse/trunk/FDS/trunk/Test_c...

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

[fire-tom]

Hello again,

the "loss of pressure"-problem seems to be resolved after an update to
version 5.2. Problematic version was 5.1. The first problem (global
oscillation) has been resolved with adding PRESSURE_CORRECTION.

Thanks a lot,

Thomas

> > > > >> I was running the following...
>
> Erfahren Sie mehr »