Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

3D structural model converging or badly defined?

Please login with a confirmed email address before reporting spam

I have a 3D structural model with about 1M DoF.
It's currently running on a workstation with 2 Xeon processors Quad-Core (Total 8 cores) with 24GB of RAM.

It's been running for 24 hours and I'm wondering if I should just kill the solver and try to redefine my model or is it "normally" solving it.

Here is what the log looks like for the moment:

Iter ErrEst Damping Stepsize #Res #Jac #Sol
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
1 0.00044 0.0100000 0.00045 4 1 4
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
2 0.00091 0.0039136 0.00092 5 2 6
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
3 0.00052 0.0128438 0.00053 6 3 8
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
4 0.00053 0.0131304 0.00054 7 4 10
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
5 0.0011 0.0030533 0.0011 8 5 12
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
6 0.001 0.0104015 0.0011 9 6 14
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
7 0.0015 0.0055642 0.0015 10 7 16
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
8 0.0013 0.0010043 0.0013 12 8 19
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
9 0.0041 0.0001691 0.0041 13 9 21
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
10 0.0062 0.0002497 0.0062 14 10 23
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
11 3.5e+007 0.7500000 0.08 16 11 26
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
12 1.3 0.0001000 1.3 20 12 31
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
13 0.93 0.0001000 0.93 22 13 34
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
14 1.2 0.0001000 1.2 23 14 36
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
15 1.2 0.0001000 1.2 24 15 38
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
16 1 0.0001648 1 25 16 40
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
17 1.6 0.0001000 1.6 26 17 42
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
18 4e+007 0.7500000 1.7 28 18 45
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
19 0.95 0.0100000 1 31 19 49
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
Warning: Inverted mesh element near coordinates (0.00121766, 0.0998856, 0.00217183)
Warning: Inverted mesh element near coordinates (0.00121095, 0.0993358, 0.00215987)
20 1.3e+006 0.7500000 1.6 35 20 54
Warning: Inverted mesh element near coordinates (0.00121319, 0.0995191, 0.00216386)


Looking at the error estimate (ErrEst), it seems to oscillate a lot so I'm wondering if it's worthwhile to let it run or if I should try to redo my model. There's a lot of "inverted mesh element" warning but since I had a lot of difficulty just to mesh my model without ending up with 4M DoF...I'm rather pleased with the mesh as it is.

What do you think?
Is it a "normal" run or is it going nowhere due to the error oscillating that much?
(It's running the Nonlinear solver and MUMPS)

Also, should I restrict COMSOL to just use 4-cores instead of 8 in order to get more RAM per core (since I "only" have 24GB).

Thanks.

3 Replies Last Post 2 mar 2011, 15:00 GMT-5
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 mar 2011, 12:54 GMT-5
HI

indeed when the error estimate oscilaltes and exploses to 1E6 or more I usually abort and change my time or parameter steps.
You might try a stop, look at the results and do a restart, sometimes it manages to finish (if already almost there but if it's at 10% take a closer look and try to improve initial conditions or BC or whatever so that it solves quicker on next try

--
Good luck
Ivar
HI indeed when the error estimate oscilaltes and exploses to 1E6 or more I usually abort and change my time or parameter steps. You might try a stop, look at the results and do a restart, sometimes it manages to finish (if already almost there but if it's at 10% take a closer look and try to improve initial conditions or BC or whatever so that it solves quicker on next try -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 mar 2011, 13:22 GMT-5
This might be a silly question but how do you know what's the total progress made?

It keeps going to 100% for each solver (Nonlinear and MUMPS) passing through Matrix Factorization, Solving linear system, etc.

But you're right, an error estimate of 1.3E6 is not really an error anymore, it's simply on the wrong track (I first thought that it was 1.3E-06....)

Do you think I should also restrict COMSOL to 4-cores since I have 24GB RAM? (At 8-cores I end up max with 3GB RAM per core which is not that much...).

Thanks Ivar,
I'm slowly learning....
This might be a silly question but how do you know what's the total progress made? It keeps going to 100% for each solver (Nonlinear and MUMPS) passing through Matrix Factorization, Solving linear system, etc. But you're right, an error estimate of 1.3E6 is not really an error anymore, it's simply on the wrong track (I first thought that it was 1.3E-06....) Do you think I should also restrict COMSOL to 4-cores since I have 24GB RAM? (At 8-cores I end up max with 3GB RAM per core which is not that much...). Thanks Ivar, I'm slowly learning....

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 mar 2011, 15:00 GMT-5
Hi

things take time, I should say it but I estimated that it took me more than a year before I really feltr confortable, and I still do stupid errors ;)

I would leave COMSOL sort out the np use, persnally I use only half of the max as I have multithreaded cores and I do not geain that much using all rather than 50% of them. Then I can still use the WS for other things, like a sencond COMSOL session to test out things

--
Good luck
Ivar
Hi things take time, I should say it but I estimated that it took me more than a year before I really feltr confortable, and I still do stupid errors ;) I would leave COMSOL sort out the np use, persnally I use only half of the max as I have multithreaded cores and I do not geain that much using all rather than 50% of them. Then I can still use the WS for other things, like a sencond COMSOL session to test out things -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.