Velocity of solution

Topics: Geometry, Solving, 4.3, 4.3a

Thread index  |  Previous thread  |  Next thread  |  Start a new discussion

RSS FeedRSS feed   |   Email notificationsTurn on email notifications   |   7 Replies   Last post: January 23, 2013 9:08pm UTC
Martin Kandl

Martin Kandl

January 23, 2013 9:04am UTC

Velocity of solution

Hi,

Since it depends what speed calculations?
I created two identical geometry.
One in 2D it is counted in minutes.
Second in 3D it is counted for 30 minutes to 1 hour.

It is possible to calculate the 3D somehow accelerate?

Thank you.

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Edgar Kaiser

Edgar Kaiser

January 23, 2013 3:10pm UTC in response to Martin Kandl

Re: Velocity of solution

There are several options:

- Reduce mesh resolution
- Take advantage of any symmetries in the model
- Buy a faster computer ;-))

Cheers
Edgar

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Martin Kandl

Martin Kandl

January 23, 2013 5:04pm UTC in response to Edgar Kaiser

Re: Velocity of solution

Thank for amswer..

BTW If both geometries are calculated on the same computer, so the quality of the computer should not play a role.

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Ivar Kjelberg

Ivar Kjelberg

January 23, 2013 6:04pm UTC in response to Martin Kandl

Re: Velocity of solution

Hi

no 3D has n^3 nodes versus n^2 for 2D and n for 1D its simple time consuming rule, now this is not the full story, but a good approximation.
Conclusion: always try to reduce the number of dimension of a model, and a model 10% correct in 2D gives better results than waiting for a 3D model to solve and anyhow not knowing if something is wrong and that one need to wait 2-3 solver times anyhow ;)

--
Good luck
Ivar

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Martin Kandl

Martin Kandl

January 23, 2013 7:08pm UTC in response to Ivar Kjelberg

Re: Velocity of solution

Thank you.

But I do not know if I'm wrong but I understood the problem that the model is slower in 3D than in 2D.
I should therefore reduce the 3D model?

And one more question what is Llimitfivar?

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Ivar Kjelberg

Ivar Kjelberg

January 23, 2013 7:37pm UTC in response to Martin Kandl

Re: Velocity of solution

Hi
Yes a model will be far slower in 3D than in 2D because there is far more to calculate.
But I cannot tell you anything about that variable, I do not know that one ;)
Probably "Support" can tell (I'm not from COMSOL ;)
Have you tried to search the help and the pdf documents ?

--
Good luck
Ivar

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Martin Kandl

Martin Kandl

January 23, 2013 7:54pm UTC in response to Ivar Kjelberg

Re: Velocity of solution

I did try searching the help, but nothing.

I have another meaningless error:

Failed to evaluate variable Jacobian.
- Variable: mod1.AzY
- Geometry: 1
- Domain: 2
It is nonsense because it appears in the 2D model.

You do not have at least some e-mail address for such support?
Because when I write to support, no reply.

Thank.

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse

Ivar Kjelberg

Ivar Kjelberg

January 23, 2013 9:08pm UTC in response to Martin Kandl

Re: Velocity of solution

Hi

its the one on the main web site, but recently you can ask for support cases via your login info (click on your name top right of the present web page or login first if your name is not there)

It would help if you explained a bit more which physics you are using, 2D, 3D, etc as there are a few thousand variables inside COMSOL and their name depends on the physics, the type of modelling etc, it's not easy for us out here to understand exactly what you are doing

The Jacobian is the matrix COMSOL uses to map the deformations and to sense where to push next step for the solver iterations, your BC must be continuous functions (or scalars) else COMSOL cannot estimate the Jacobian

--
Good luck
Ivar

Reply  |  Reply with Quote  |  Send private message  |  Report Abuse


Rules and guidelines