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.

Generalized alpha or BDF?

Please login with a confirmed email address before reporting spam

Hello,

What is the difference between "Generalized alpha" and "BDF" time dependent solvers?
Is there any specification that each one is better for a certain type of simulation?
i.e. For a thermal-electrical-mechanical coupled transient analysis which one is better?

Best regards

4 Replies Last Post 2013年10月3日 GMT-4 13:49

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2013年10月3日 GMT-4 06:09
I'm not a big expert with numerical solvers, but roughly speaking:

1) BDF is a multistep method, this means it keeps and uses informations from previous steps to calculate the next one.
It's very good choice for stiff equations but has some problems with damping.

2) Generalized Alpha (no idea of the algorithm behind it, sorry) is usually less stable than BDF as it uses less damping but it's more accurate.

Usually you expect Generalized Alpha to perform better in structural mechanics (it was developed for that) and I would say in acoustics (you don't want to have big dampings there). No idea for other kind of physics.

Do you expect sharp transient in your solutions? In that case try generalized alpha. If you have convergence issues use BDF.

EDIT: just found this, it can help you
www.comsol.com/support/knowledgebase/1062/
I'm not a big expert with numerical solvers, but roughly speaking: 1) BDF is a multistep method, this means it keeps and uses informations from previous steps to calculate the next one. It's very good choice for stiff equations but has some problems with damping. 2) Generalized Alpha (no idea of the algorithm behind it, sorry) is usually less stable than BDF as it uses less damping but it's more accurate. Usually you expect Generalized Alpha to perform better in structural mechanics (it was developed for that) and I would say in acoustics (you don't want to have big dampings there). No idea for other kind of physics. Do you expect sharp transient in your solutions? In that case try generalized alpha. If you have convergence issues use BDF. EDIT: just found this, it can help you http://www.comsol.com/support/knowledgebase/1062/

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2013年10月3日 GMT-4 09:17
Hello Stefano,

Thank you for the information.

In my simulations, which are thermal-electrical-mechanical coupled, it would be some sharp transient for electrical current. On the other hand, when I consıder contact between parts, then convergence would be a problem.
So, maybe in case of contact it would be better to go with BDF?
Hello Stefano, Thank you for the information. In my simulations, which are thermal-electrical-mechanical coupled, it would be some sharp transient for electrical current. On the other hand, when I consıder contact between parts, then convergence would be a problem. So, maybe in case of contact it would be better to go with BDF?

Eric Favre COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2013年10月3日 GMT-4 12:40
Hello!

www.comsol.com/blogs/analyzing-electrical-thermal-conductance-contact-switch/
The corresponding model is included with 4.3b where this kind of multiphysics contact has been implemented and is ready to use.

Regarding BDF vs alpha-gen, in principle you should be able to use one or the other and get comparable results. The solution in case of any visible difference is frequently to adjust the tolerances. As usual, in a multiphysics complicated problem, always try to solve for simple problems before attacking difficult ones!

good luck,
Eric
Hello! http://www.comsol.com/blogs/analyzing-electrical-thermal-conductance-contact-switch/ The corresponding model is included with 4.3b where this kind of multiphysics contact has been implemented and is ready to use. Regarding BDF vs alpha-gen, in principle you should be able to use one or the other and get comparable results. The solution in case of any visible difference is frequently to adjust the tolerances. As usual, in a multiphysics complicated problem, always try to solve for simple problems before attacking difficult ones! good luck, Eric

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2013年10月3日 GMT-4 13:49
Ok.
Thank you for your information.

Best regards
Ok. Thank you for your information. Best regards

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.