Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

Two mathematically identical equations draw diffrent graphs!

Please login with a confirmed email address before reporting spam

I am trying to solve a dimensionless transient problem with PDE solver.

I had an issue in solving
dtheta/dt + grad cdot (- grad theta) = 0
, which should be identical to
dtheta/dt = d2theta/dx2
in 1-dimension with the PDE solver in COMSOL: these two have different
results!
I attached two mph files showing the difference.

I want any comment. I hope I made some syntax error.


2 Replies Last Post 11 lug 2012, 23:57 GMT-4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 lug 2012, 07:05 GMT-4
Hi Sun Ung Kim,

I did not go through your model files but what you are observing might be related to the fact that mathematical equations which are equivalent on the continuous level, i.e., in their analytic form, may show a different behaviour/solution if you a apply a discrete numeric scheme. In particular, the discretized equations are usually no longer equivalent.

It's been a while but I worked on very similar problems a couple of years ago. Maybe this one can give you some idea on how much a solution may vary depending on the actual formulation used for the implementation.

www.comsol.com/papers/4887/download/Weddemann.pdf
Hi Sun Ung Kim, I did not go through your model files but what you are observing might be related to the fact that mathematical equations which are equivalent on the continuous level, i.e., in their analytic form, may show a different behaviour/solution if you a apply a discrete numeric scheme. In particular, the discretized equations are usually no longer equivalent. It's been a while but I worked on very similar problems a couple of years ago. Maybe this one can give you some idea on how much a solution may vary depending on the actual formulation used for the implementation. http://www.comsol.com/papers/4887/download/Weddemann.pdf

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11 lug 2012, 23:57 GMT-4
Thank you for your reply. Probably my problem is different from what you were looking at.
My problem is solved by changing quadratic Lagrange shape function to Hermite shape functions.
Thanks anyway, and have a good day!
Best,
Sun Ung Kim





Received message from COMSOL:

Dear Sun Ung,

The difference in results is because second-order derivative terms in the f
coefficient are not accurately represented when using quadratic Lagrange
shape functions. One workaround is to change the discretization to use
Hermite shape functions instead, see attached model.

With best regards,
Thank you for your reply. Probably my problem is different from what you were looking at. My problem is solved by changing quadratic Lagrange shape function to Hermite shape functions. Thanks anyway, and have a good day! Best, Sun Ung Kim Received message from COMSOL: Dear Sun Ung, The difference in results is because second-order derivative terms in the f coefficient are not accurately represented when using quadratic Lagrange shape functions. One workaround is to change the discretization to use Hermite shape functions instead, see attached model. With 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.