convergence problem Log Out | Topics | Search
Moderators | Register | Edit Profile

FlexPDE User's Forum » User Postings » convergence problem « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Nestor Rueda Vallejo (ruedavallejo)
Member
Username: ruedavallejo

Post Number: 4
Registered: 10-2009
Posted on Sunday, October 25, 2009 - 11:38 pm:   

Hi Group
I’m trying to model a coupled magnetic-thermal problem. I already solve both problems independently and they run ok, but when I put these two equations together the model stops suddenly without any correct result.
Which is wrong with the code?
Best Regards.
Top of pagePrevious messageNext messageBottom of page Link to this message

Marek Nelson (mgnelson)
Moderator
Username: mgnelson

Post Number: 146
Registered: 07-2007
Posted on Monday, October 26, 2009 - 07:07 pm:   

This script ran to completion in three seconds. It did not "stop suddenly". How long did the independent scripts run? About three seconds?

It appears that the equations in this script are not coupled. You need to do something to couple them, e.g. add a term in each equation that depends in some way on the other variable.
Top of pagePrevious messageNext messageBottom of page Link to this message

Nestor Rueda Vallejo (ruedavallejo)
Member
Username: ruedavallejo

Post Number: 5
Registered: 10-2009
Posted on Monday, October 26, 2009 - 09:53 pm:   

Hi Marek, thanks for your answer.
I guess I ask the wrong question. The true problem is the physical meaning of the results. Why I’m having such a bad plots.
On the other hand, I realize that there is no need to couple these two equations because the heating is generated by the joule effect, not by eddy currents, so the only variable which I need is the current density through the conductors not the magnetic potential.
One last question: how can I delete the previously attached file?
Top of pagePrevious messageNext messageBottom of page Link to this message

Marek Nelson (mgnelson)
Moderator
Username: mgnelson

Post Number: 148
Registered: 07-2007
Posted on Tuesday, October 27, 2009 - 02:54 pm:   

What is it about the plots that you do not like?

The mesh is crude because you are running against the node limit of the student version. This makes a couple of the contour plots rough. Overall, the plots look pretty good to me, but I do not know what you are expecting.

I can remove the previously posted file if you like. Your login does not have access to do it.
Top of pagePrevious messageNext messageBottom of page Link to this message

Nestor Rueda Vallejo (ruedavallejo)
Member
Username: ruedavallejo

Post Number: 6
Registered: 10-2009
Posted on Wednesday, October 28, 2009 - 12:29 am:   

Hi Marek, these are the plots which I dislike it!
vector_plot
And these is the plot which I expect, it was solved independently:
vector_plot2
And yes please delete the attached file of my initial post.
Top of pagePrevious messageNext messageBottom of page Link to this message

Marek Nelson (mgnelson)
Moderator
Username: mgnelson

Post Number: 150
Registered: 07-2007
Posted on Wednesday, October 28, 2009 - 03:29 pm:   

If you use
SELECT OVERSHOOT=1e-4
then this script runs correctly.

OVERSHOOT controls the accuracy in the conjugate gradient solver. The default value is 1e-3. So 1e-4 will make the solver try harder to follow the solution.

Previously, I ran your script using the FlexPDE6 student version and it solves fine without this modification, but the 5.10s Student version appears to have a problem with scaling the two equations and needs a tighter control on the conjugate gradient solver.

PS - I removed your script as you requested. In the future if you do not wish to post the script publically, you can email it to support.
Top of pagePrevious messageNext messageBottom of page Link to this message

Nestor Rueda Vallejo (ruedavallejo)
Member
Username: ruedavallejo

Post Number: 7
Registered: 10-2009
Posted on Thursday, October 29, 2009 - 02:28 am:   

Thanks Marek you said a Keyword: scaling! next step is to make a simmetry around x axis to make better use of meshing capabilities.
Regards

Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action:

Topics | Last Day | Last Week | Tree View | Search | Help/Instructions | Program Credits Administration