search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
625 stars 610 forks source link

PROBLEM WITH MULTI-MESH #2214

Closed gforney closed 9 years ago

gforney commented 9 years ago 
Hello everybody,

I have a problem with my compute, In fact, when i run my compute using multi-mesh with
multi-processors, i have bad values. But when i run with one Mesh using one Processor,
i have of good values.

I want use a multi-mesh in order to have a small time of computation

Could you help me to resolve my problem

Thans for your answers

Original issue reported on code.google.com by bradyaxel@yahoo.fr on 2014-09-24 13:36:52

gforney commented 9 years ago 
I will take a look at it. What version of FDS are you using?

Original issue reported on code.google.com by mcgratta on 2014-09-24 13:41:16

gforney commented 9 years ago 
Could you be more specific. What do you mean when you say "bad values"?

Original issue reported on code.google.com by mcgratta on 2014-09-24 13:48:17

gforney commented 9 years ago 
I took a quick look at these since they are using HVAC.  If I run both for a short bit,
there are different velocities coming out of the HVAC VENTs which are split over multiple
meshes. With the multimesh case showing about twice the velocity of the single mesh
case.  I think I am not aggregating the boundary conditions correctly for the case
where an HVAC VENT spans more than one mesh which is resulting in not having the same
wall velocity in HVAC_BC.

Original issue reported on code.google.com by drjfloyd on 2014-09-24 14:07:12

gforney commented 9 years ago 
OK, I'll turn it over to you.

To the original poster of the issue, do not split vents between meshes. This is probably
a good idea for the fire too.

Original issue reported on code.google.com by mcgratta on 2014-09-24 14:10:40

gforney commented 9 years ago 
(No text was entered with this change)

Original issue reported on code.google.com by mcgratta on 2014-09-24 14:10:52

gforney commented 9 years ago 
I am using FDS6, and for the bad values, i want say that, during the first minutes,
with one mesh,i have my desired boundary conditions. But when, i use multi-mesh, i
have other values.

I would know, how we have to put multi-mesh to have the same values when we are in
one mesh

thanks for your help

Original issue reported on code.google.com by bradyaxel@yahoo.fr on 2014-09-24 14:54:29

gforney commented 9 years ago 
Read the comments above. Apparently this is a bug. We suggested a work-around until
a new version is released.

Original issue reported on code.google.com by mcgratta on 2014-09-24 14:56:40

gforney commented 9 years ago 
Unlike leakage which is simpler to handle over multiple meshes, actual HVAC nodes will
need a lot more information sharing over meshes.  I started to look in to this but
it is quickly becoming a rabbit hole.  For now I'll add a note to the guide to avoid
splitting HVAC vents over multiple meshes.  I'll put this on hold for a bit.

Original issue reported on code.google.com by drjfloyd on 2014-09-24 20:08:30

gforney commented 9 years ago 
In addition to manual update have added error trapping until the boundary condition
routines are reworked.

Original issue reported on code.google.com by drjfloyd on 2014-09-24 20:27:36

gforney commented 9 years ago 
Hello,

Concerning the MULTI-MESH you can see an other example which doesn't work when we divide
one MESH into several MESH.

I have attached the file

My case is similar to this example, i have a HVAC system and Leaks.

I want know, why i use one MESH, my compute have the good boundary conditions, but
with several MESH like in the example, i have the bad values

Thanks for your help

Original issue reported on code.google.com by bradyaxel@yahoo.fr on 2014-09-25 08:54:33

gforney commented 9 years ago 
comment 11 was also posted in the forum.  See response there:
https://groups.google.com/forum/?fromgroups=&nomobile=true#!topic/fds-smv/47M-MI2i7A4

Please keep comment threads in one location.  It becomes an extra burden on the developers
if the same thread is active in multiple locations.

Original issue reported on code.google.com by drjfloyd on 2014-09-25 12:39:01

gforney commented 9 years ago 
pushed up changes yesterday that made it through firebot except for door_crack.  Going
to call this fixed but will investigate door_crack.

Original issue reported on code.google.com by drjfloyd on 2014-09-26 11:30:11

gforney commented 9 years ago 
Think I have this fixed completely and will push up once we get through the current
release cycle.

Original issue reported on code.google.com by drjfloyd on 2014-09-30 19:21:42