Residue Scaling in moose

[Max]

Hi all,

I am simulating a viscoplasticity solid mechanics problem coupling with heat conduction equation, I just found out my residual norm for 'chi' (temperature) and the displacement field are not the same order of magnitude.

I have read about the Scaling option to give better convergence, I am wondering can someone point me to a couple of exmaples ? 

Thanks,

Max

3 Nonlinear |R| = 8.272558e-01

      0 Linear |R| = 8.272558e-01

      1 Linear |R| = 3.721311e-01

      2 Linear |R| = 1.345872e-01

      3 Linear |R| = 4.131587e-02

      4 Linear |R| = 1.609853e-02

      5 Linear |R| = 4.858614e-03

      6 Linear |R| = 1.629061e-03

      7 Linear |R| = 5.060169e-04

      8 Linear |R| = 1.631621e-04

      9 Linear |R| = 5.342711e-05

     10 Linear |R| = 1.637924e-05

     11 Linear |R| = 5.142239e-06

    |residual|_2 of individual variables:

                     chi:    4.65369e-10

                     disp_x: 0.000314444

                     disp_y: 0.000577023

 4 Nonlinear |R| = 6.571377e-04

 Solve Converged!

[Andrew....@csiro.au]

from memory you just add

scaling = 1000

to your ‘chi’ Variable (assuming your ‘chi’ residual is 1000 smaller than your displacement residuals)

a

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/73dd1929-38c8-4fc0-bd41-5286b7a58817%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Derek Gaston]

Do the solve look ok though?  This might not actually be a problem.

Derek

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/D3478752.2666A%25Andrew.Wilkins%40csiro.au.

[Max]

The solution right , I am just wondering if the scaling would help the results to converge faster ? 

Max

[Andrew....@csiro.au]

i believe it would in certain cases.  just give it a try and see.  In my experience if the residuals are different by a factor of 100 it makes no difference whatsoever.

 

a

 

 

Ph: +61 7 3327 4497.  Fax: +61 7 3327 4666
Queensland Centre for Advanced Technologies
PO Box 883, Kenmore, Qld, 4069 
 

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/de1f7416-1594-45a5-baab-095f58671e04%40googlegroups.com.

[Benjamin Spencer]

For coupled thermal/mechanics problems, we've found that the difference in the scaling isn't really an issue.  Usually the thermal solution converges more quickly than the mechanics problem, so it's pretty normal to see a far lower final residual in the thermal problem, so that the difference in the residuals in the converged solution is far larger than the difference in the initial residuals.

- Ben

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/c9013f70d1024f83afc4e5a0b19f2006%40exch3-cdc.nexus.csiro.au.