CFD Online URL
[Sponsors]
Home > Forums > CFX

desperate Fatal overflow in linear solver - transient

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   January 4, 2010, 08:13
Default desperate Fatal overflow in linear solver - transient
  #1
Senior Member
 
Join Date: Jul 2009
Posts: 207
Rep Power: 8
kingjewel1 is on a distinguished road
Hi there,

I'm a bit desperate here, tried everything i can think of

I'm running a simulation of a ventilated room (with k-e) under transient conditions and the solver has a fatal error after the first iteration. Steady state runs nicely even under high resolution scheme. Any Ideas? please

+----------------------+------+---------+---------+------------------+
| K-TurbKE | 2.00 | 4.1E-02 | 8.9E-01 | 6.3 2.2E-02 OK|
| E-Diss.K | 2.26 | 5.4E-02 | 1.9E+00 | 7.1 3.8E-04 OK|
+----------------------+------+---------+---------+------------------+
----------------------------------------------------------------------
COEFFICIENT LOOP ITERATION = 2 CPU SECONDS = 1.244E+02
----------------------------------------------------------------------
| Equation | Rate | RMS Res | Max Res | Linear Solution |
+----------------------+------+---------+---------+------------------+
| U-Mom | 0.00 | 1.3E-13 | 1.8E-12 | 3.2E+10 * |
| V-Mom | 0.00 | 1.3E-13 | 2.9E-12 | 3.1E+10 * |
| W-Mom | 0.00 | 4.1E-13 | 3.0E-12 | 9.6E+09 F |
| P-Mass | 0.00 | 0.0E+00 | 2.9E-24 | 9.7 9.1E+10 * |
+----------------------+------+---------+---------+------------------+
| Contaminant | 3.85 | 7.4E-03 | 2.2E+00 | 6.0 3.2E-02 OK|
+----------------------+------+---------+---------+------------------+

+--------------------------------------------------------------------+
| ERROR #004100018 has occurred in subroutine FINMES. |
| Message: |
| Fatal overflow in linear solver. |
+--------------------------------------------------------------------+

+--------------------------------------------------------------------+
| An error has occurred in cfx5solve: |
| |
| The ANSYS CFX solver has terminated without writing a results |
| file. Command on host phipc (PHI-PC) exited with return code 0. |
+--------------------------------------------------------------------+

End of solution stage.

+--------------------------------------------------------------------+
| The following transient and backup files written by the ANSYS CFX |
| solver have been saved in the directory C:\Users\Phi\Documents\New |
| folder\Single\Standrews8_transient_013: |
| |
| 777_full.trn |
+--------------------------------------------------------------------+
Attached Images
File Type: png Standrews_8.png (28.3 KB, 33 views)
kingjewel1 is offline   Reply With Quote

Old   January 4, 2010, 08:30
Default
  #2
New Member
 
Join Date: May 2009
Posts: 17
Rep Power: 7
shogologo is on a distinguished road
Which Timestep did you chose?
shogologo is offline   Reply With Quote

Old   January 4, 2010, 08:32
Default
  #3
Senior Member
 
Join Date: Jul 2009
Posts: 207
Rep Power: 8
kingjewel1 is on a distinguished road
Quote:
Originally Posted by shogologo View Post
Which Timestep did you chose?
Timestep is 10s for total of 30mins. Also gives the same problem with adaptive timestepping and regardless of initial value file.
I'm worried about the simulation always giving RMS Courrant Number=0.00 and then jumping suddenly 999 after first iteration then crashing.

Last edited by kingjewel1; January 4, 2010 at 08:52.
kingjewel1 is offline   Reply With Quote

Old   January 4, 2010, 09:26
Default
  #4
New Member
 
Join Date: May 2009
Posts: 17
Rep Power: 7
shogologo is on a distinguished road
I managed a similar type of simulation.
My suggestion is to reduce timestep so to get Courant between 2 and 10 (or at least 30).

So I would try reduce to 1e-2 / 1e-3 sec...
shogologo is offline   Reply With Quote

Old   January 4, 2010, 12:45
Default
  #5
Senior Member
 
Join Date: Jul 2009
Posts: 207
Rep Power: 8
kingjewel1 is on a distinguished road
Quote:
Originally Posted by shogologo View Post
I managed a similar type of simulation.
My suggestion is to reduce timestep so to get Courant between 2 and 10 (or at least 30).

So I would try reduce to 1e-2 / 1e-3 sec...
Thank you. Its running with a Courant number at 21 with a 0.1 second timestep. Why does it require such a small timestep for this type of simulation? I don't need this much precise detail, as the code will take, I calculate, 60 days to run 30 mins of simulation.

What was your simulation?
kingjewel1 is offline   Reply With Quote

Old   January 5, 2010, 05:51
Default
  #6
New Member
 
Join Date: May 2009
Posts: 17
Rep Power: 7
shogologo is on a distinguished road
You can try to speed up the timestep "on the fly" to understand which is the characteristic timescale.

It should depend mainly by grid size and Reynolds number.
shogologo is offline   Reply With Quote

Old   January 5, 2010, 08:17
Default
  #7
Senior Member
 
Attesz's Avatar
 
Attesz
Join Date: Mar 2009
Posts: 352
Rep Power: 7
Attesz is an unknown quantity at this point
Hi,

why are you doing the simulation so long (i'm just interested)? Doing a steady simulation first, and after that a transient for 5min for example, will not be better? (just an idea, i don't know your goals).

Regards,
Attesz

Last edited by Attesz; January 5, 2010 at 12:20.
Attesz is offline   Reply With Quote

Old   January 5, 2010, 12:17
Default
  #8
Senior Member
 
Join Date: Jul 2009
Posts: 207
Rep Power: 8
kingjewel1 is on a distinguished road
I'm looking at the movement of particles/pathogens in a hospital room. The idea is to investigate different scenarios including, coughing, sneezing, bed making etc. In the end I'd also want to look at the way displacement ventilation suspends certain particles too. What do you think?

How do you change the timestep on the fly without going back into pre?
kingjewel1 is offline   Reply With Quote

Old   January 5, 2010, 12:31
Default
  #9
Senior Member
 
Attesz's Avatar
 
Attesz
Join Date: Mar 2009
Posts: 352
Rep Power: 7
Attesz is an unknown quantity at this point
I'm not experienced in transient simulation, but perhaps the value of timesteps can be adjusted by clicking on the "Edit run in progress..." button (I've never tryed it). In my centrifugal compressor simulation I've used adaptive timestep, to reach always the desired Courrant number (it was very important at me), and the timestep will be set "automatically" by the solver.

However, your simulation is very interesting!

Good luck,
Attesz

Attached Images
File Type: jpg Vágólap01.jpg (36.6 KB, 169 views)
Attesz is offline   Reply With Quote

Old   January 5, 2010, 14:53
Default
  #10
Senior Member
 
Join Date: Jul 2009
Posts: 207
Rep Power: 8
kingjewel1 is on a distinguished road
Thanks! I've just tried it and it works great.
kingjewel1 is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Fatal Overflow when using RNG k-e Pascal CFX 2 February 5, 2008 17:41
free C code for large sparse matrix linear solver ztdep Main CFD Forum 7 May 24, 2007 15:14
Linear Iterative Solver + Elliptic PDE cfd101 Main CFD Forum 0 November 14, 2005 19:59
linear solver overflow peggy CFX 1 February 8, 2001 02:39
solver for linear system with large sparse matrix Yangang Bao Main CFD Forum 1 October 25, 1999 05:22


All times are GMT -4. The time now is 07:37.