search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
674 stars 626 forks source link

Numerical instability in particle energy transport #1825

Closed gforney closed 9 years ago

gforney commented 9 years ago 
SVN 15061

USCG_HAI_1000_kW_Forced_Fike.out:Numerical instability in particle energy transport,
TMP_G
USCG_HAI_250_kW_Forced_Navy.out:Numerical instability in particle energy transport,
TMP_G
USCG_HAI_250_kW_Natural_Fike.out:Numerical instability in particle energy transport,
TMP_G
USCG_HAI_500_kW_Natural_Fike.out:Numerical instability in particle energy transport,
TMP_G

There is no guidance in the User's Guide for this kind of instability.

Original issue reported on code.google.com by mcgratta on 2013-03-11 19:59:02

gforney commented 9 years ago 
Can you provide any guidance now?  I just got the same error running a model.  Don't
think there are too many particles.  Involves a sprinkler.  I can upload if necessary.
 Running SVN 15774.

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-10 13:38:49

gforney commented 9 years ago 
Yes, please upload the case.

Original issue reported on code.google.com by mcgratta on 2013-05-10 13:44:51

gforney commented 9 years ago 
Attached.

And here is the relevant part of the *.out file:

       Time Step   16500   May  9, 2013  23:35:15
       Pressure Iterations:      2
       Maximum Velocity Error:  0.16E-01 on Mesh   1 at (  15 152   3)
       ----------------------------------------------
       CPU/step:     7.661 s, Total CPU:     32.31 hr
       Time step:  0.00459 s, Total time:   323.43 s
       Max CFL number:  0.56E+00 at ( 43,131, 51)
       Max divergence:  0.39E+01 at ( 43,131, 51)
       Min divergence: -0.26E+02 at ( 51,144, 42)
       No. of Lagrangian Particles:         22153
       Total Heat Release Rate:            116.605 kW
       Radiation Loss to Boundaries:       -37.640 kW

       Time Step   16600   May  9, 2013  23:48:12
       Pressure Iterations:      3
       Maximum Velocity Error:  0.16E-01 on Mesh   1 at (  48  99   3)
       ----------------------------------------------
       CPU/step:     7.634 s, Total CPU:     32.52 hr
       Time step:  0.00459 s, Total time:   323.89 s
       Max CFL number:  0.56E+00 at ( 40,136, 50)
       Max divergence:  0.21E+01 at ( 52,116,  4)
       Min divergence: -0.23E+02 at ( 46, 72,  5)
       No. of Lagrangian Particles:         24371
       Total Heat Release Rate:            119.169 kW
       Radiation Loss to Boundaries:       -38.906 kW

Numerical instability in particle energy transport, TMP_G

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-10 13:46:59

gforney commented 9 years ago 
I'm going to run this with the latest compile on our linux cluster.

Original issue reported on code.google.com by mcgratta on 2013-05-10 20:28:53

gforney commented 9 years ago 
Ok.  Let me know if there is anything you want me to do on my end.  Thanks.  Have a
great weekend.

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-10 20:32:19

gforney commented 9 years ago 
My case failed at 313 s with the instability message about particle transport. We will
fix this before a release of FDS 6.

Original issue reported on code.google.com by mcgratta on 2013-05-12 16:54:31

gforney commented 9 years ago 
Obviously mine got a little further (to 323s) before the error, but same difference
I suppose.  Thanks for looking into this, and whenever you think you have it fixed,
even if before a release, let me know what changed and the new SVN number and I'll
recompile and give it a go.  Thanks again!

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-12 17:43:53

gforney commented 9 years ago 
Mine died at 316.38.  The gas temperature was below freezing in the cell with the droplet
and the droplet was hot. The diameter of the droplet was ~150 microns.  My guess is
whatever is causing the very cold temperatures is ultimately to blame.

Original issue reported on code.google.com by drjfloyd on 2013-05-13 12:27:31

gforney commented 9 years ago 
Well, I guess we'll figure out the difference in time of death later.  As for why it
died, how long was the sprinkler flowing before it crashed?  This is the run I discussed
with Jason F. a little on the discussion board.  Fire 1 activates sprinkler, sprinkler
flows for 10 seconds.  Fire 1 is removed.  New obstruction is placed on top with Fire
2.  Fire 2 burns.  Has anything to do with the formation of the new obstruction in
areas that might have particles?  Just a thought.

I am on business travel starting in about an hour until later in the week, but I'll
look at my output a little closer and see if I can diagnose anything from that.  Let
me know if you make any more progress.  Many thanks guys!

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-13 12:45:46

gforney commented 9 years ago 
I think the change I just committed will fix this problem.  The case has now run out
to 338 seconds, well beyond where it was failing for me.  The issue was we were applying
the min/max gas temperature clip in the wrong spot of the the evaporation loop.

Original issue reported on code.google.com by drjfloyd on 2013-05-21 13:15:28

gforney commented 9 years ago 
I think the change I just committed will fix this problem.  The case has now run out
to 338 seconds, well beyond where it was failing for me.  The issue was we were applying
the min/max gas temperature clip in the wrong spot of the the evaporation loop.

Original issue reported on code.google.com by drjfloyd on 2013-05-21 13:15:30

gforney commented 9 years ago 
Thanks Jason.  I'll give it a go and I'll be in touch after I verify.  Thanks again.

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-21 13:31:05

gforney commented 9 years ago 
Verified with SVN 15882.  Thanks Jason!

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-23 12:55:17

gforney commented 9 years ago 
Thanks for the update.

Original issue reported on code.google.com by drjfloyd on 2013-05-23 12:56:18

gforney commented 9 years ago 
Note, had a power outage over the weekend which killed the model well beyond the ~300
s where the sprinkler activated and we previously got this error.  In other words,
before the power outage, all was working fine.  But I did get the part error again
immediately when I restarted the model.  Probably not a top priority to fix as the
restart function is fragile as you note in the user's guide, but just something to
think about.

Original issue reported on code.google.com by Stephen.Olenick on 2013-05-28 14:23:31