Convergence Issue

[PUAE2014]

Hi, 

I changed material property such that it is updated based on the variable value at qp as solved from kernel. I am getting error saying "Solve failed and timestep already at or below dtmin, cannot continue!". I tried using customized time stepper but it dint work. Please suggest what can be done in this regard. 

Thanks!

 

[Tonks, Michael R]

The problem is the solve is not converging and it keeps cutting the time step until it hits the dtmin value. This could be happening because of a bug or because you have a really hard problem to solve. Or, it may be that your residual is starting really small so it just can't shrink it enough to meet the relative convergence tolerance. In that case, you can just set an absolute tolerance to avoid the problem.

To help us know which one it is, what are the residual values that MOOSE prints to the screen during the first attempt and solving the timestep?

Mike

--
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 http://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/07d5a037-578a-4661-afbf-db7c55b3b601%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

*************************

Michael R. Tonks

Fuel Modeling and Simulation Department

Idaho National Laboratory

208-526-6319

michae...@inl.gov

[Peterson, JW]

On Wed, Nov 12, 2014 at 2:20 PM, Tonks, Michael R <michae...@inl.gov> wrote:
> The problem is the solve is not converging and it keeps cutting the time
> step until it hits the dtmin value. This could be happening because of a bug
> or because you have a really hard problem to solve. Or, it may be that your
> residual is starting really small so it just can't shrink it enough to meet
> the relative convergence tolerance. In that case, you can just set an
> absolute tolerance to avoid the problem.
>
> To help us know which one it is, what are the residual values that MOOSE
> prints to the screen during the first attempt and solving the timestep?

Please send your input file to the list.

--
John

[PUAE2014]

Please find the attached input & log file. I have also attached my kernel & material codes.

[Slaughter, Andrew E]

Sudipta,

I took a look at your log file and there doesn't seem to be anything alarming, your problem is just not converging. You likely have issues with your physics. I suggest checking your calculations, especially in your PFDiffusion material which is quite complicated.  You may consider seeing if you can get the problem to converge with a simpler material, the work on getting your more complicated material working.

--

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 http://groups.google.com/group/moose-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/408eac33-b5eb-4e29-a529-9aeec10fb34b%40googlegroups.com.

[Cody Permann]

Sudipta, 

Why does your name display as PUAE2014? Does that stand for something?

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/CAGWYgTk9Ab1RGRovV2Hrn7xQ6ikMuEES9bGsEx_suggZy9PYMA%40mail.gmail.com.

[Tonks, Michael R]

I've found that it is always good idea to debug a new kernel with a constant mobility before I make the mobility a function of the concentration.

Once you know your kernel works, then you can make the mobility a function of the concentration and then debug the material.

Mike

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/CAK06EuWbej3ntRkQaTV_Nn7MFPC_pqMd71reMzMFoHMG-1Fffw%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

[PUAE2014]

Yes Mike, I tried with constant mobility as well. Its giving me the same convergence problem. I decoupled my equations & tried to use a simpler version in kernel. I even tried my material in one of the example CH files without changing the kernel. I ran into same problem in all the cases.

That's why I was wondering if I am doing something wrong? Do you think It has anything to do with solver options?

Cody, I changed my display name in some other group, not sure how it got changed here as well. It does not really signify anything. I will change it back to my name. Thanks for drawing attention to that.  

[Tonks, Michael R]

Your solve options look fine. I will try to look at it tomorrow. 

Mike

To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/24a736c5-1963-4454-b887-535efe74e600%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Daniel Schwen]

Sudipta,
if you update to the very latest MOOSE version (from this evening) you
can try running the Jacobian analyzer on your input file:

$MOOSE_DIR/python/jacobiandebug/analyzejacobian.py -r inputfile.i

Let me know how that works for you. The tool is still in development.
Daniel

> https://groups.google.com/d/msgid/moose-users/CA%2BEq24_bmBwGH%3Dvfjv-EfoCeiQtzg9Txf33h4drrx93fkwh7Kg%40mail.gmail.com.