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.
Shape function selection in 4.0?
Posted 9 mai 2010, 03:34 UTC−4 10 Replies
Please login with a confirmed email address before reporting spam
Thx?
Please login with a confirmed email address before reporting spam
I havent got V4 yet (Switzerland is too remote you know ;) but if I remember from the beta, and if it's the shape order you are looking for I beleive its under mesh "discretisation", sorry "discetization" in US, try an indexed search on the pdf docs:
p106 of my "old" beta doc mph_ug.pdf
Discretization: "You can change the order using the "Element order" list."
hope this helps
Ivar
Please login with a confirmed email address before reporting spam
Well, I see the place that lists the Shape Function in the
Settings Window of the Equation View (e.g., 'shlag'
as applied to the dependent variable 'u'). However the
order of shlag isn't listed and there doesn't seem to be
any way to modify the information in the table.
In the Documentation I find under "Physics Interface Common Settings"
and under that "Discretization" the following:
"Discretization
"To display this section, select Show More Options from the View menu in the Model Builder.
The Element order (or, more precisely, the order of the shape function) directly affects the number of degrees of freedom in the solution and the accuracy of the solution. Increasing the order of the elements roughly corresponds to a uniform mesh refinement. Most physics interfaces uses Lagrange elements, which can be of order 1 to 4, with 2 being the default order. You can change the order using the Element order list. "
However, I am not getting a "Show More Options" from the View menu in the Model Builder (Only the
"Show geometry labels" check box), hence there is no Element order list to make changes in.
I would note that the "Show more options" is also a checkbox that can be changed under
the Toolbar -> Options -> Preferences -> Model builder. I have that check box checked, but
it doesn't seem to have any effect.
Dave
Please login with a confirmed email address before reporting spam
I have in mind that I finally found it under one of the mesh subtabs. but its a few months ago, but I remember I spent some time before I localised it ;)
Have fun Comsoling
Ivar
Please login with a confirmed email address before reporting spam
Mystery remains unsolved, but I'm still looking. If anyone else can point me to the information it would be much appreciated. Thx.
(It was easy under 3.5a.)
Please login with a confirmed email address before reporting spam
to the right of the Model Builder tab, there is a "View" menu button. It has a small upside-down white triangle on it. There, select "Show more options". Then go to the physics node, for example, Laminar Flow. In the settings windows, you find the Discretization area and Element order. See attached screenshot dicretization.png. I highlighted with red the areas of interest.
We added the "Show more options" feature to make the interface cleaner.
Niklas Rom, COMSOL
Ivar, nope, nothing under the mesh tabs.
Mystery remains unsolved, but I'm still looking. If anyone else can point me to the information it would be much appreciated. Thx.
(It was easy under 3.5a.)
Attachments:
Please login with a confirmed email address before reporting spam
Got it! Thanks!
This selects the element order of the default assigned type. How does one select among various element types, or shape functions (shlag, shdisc, shbub, etc.)?
Thanks again for your help,
Dave
Please login with a confirmed email address before reporting spam
It is also not currently possible to modify the integration order. It is fixed at twice the order of the shape function. Support for non-Lagrange shape functions in the mathematical interfaces will be added soon.
There is a rather painful work-around you can use today. You can add a weak contribution to your PDE and then add an auxiliary dependent variable. This auxiliary dependent variable allows you to select from Lagrange or discontinuous shape functions and also change the integration order. You have to deselect the degrees of freedom from the original PDE when solving so that only the auxiliary variables are solved for. This is done in the variables node under study (see the attached model).
The shbub element is discontinued. From the release notes:
"None of the special basis functions or elements for the finite element formulation of flow problems included in version 3.5a are available in version 4.0. However, the new stabilization functionality in version 4.0 for fluid flow is identical to using the bubble elements in 3.5a.
Other special elements that were available in 3.5a will not be re-implemented in version 4. The reason for this is that the stabilized formulation in version 4.0 gives high accuracy to a relatively small computational cost compared to the special elements."
/ Dan
Attachments:
Please login with a confirmed email address before reporting spam
Many thanks for this information, and for the included model file.
Can't spend time on it just now, but will later and get back with comments.
Thanks again,
Dave
Please login with a confirmed email address before reporting spam
I just starting using comsol 4 and I'm trying to implement a simple Stokes/Navier-Stokes problem starting from the equation model and it is quite painful.
There is a rather painful work-around you can use today. You can add a weak contribution to your PDE and then add an auxiliary dependent variable. This auxiliary dependent variable allows you to select from Lagrange or discontinuous shape functions and also change the integration order. You have to deselect the degrees of freedom from the original PDE when solving so that only the auxiliary variables are solved for. This is done in the variables node under study (see the attached model).
I've opened the attached model and I tried to run it again, but it's not working on my Comsol (v.4.0.0.929).
It is also not currently possible to modify the integration order. It is fixed at twice the order of the shape function. Support for non-Lagrange shape functions in the mathematical interfaces will be added soon.
Will it be possible in the near future to be able to select, for example, P2-P1 elements for this kind of problems?
Thanks
Please login with a confirmed email address before reporting spam
I finally got around to checking out the model file you attached and see what you mean.
Will look forward to seeing support for non-Lagrange shape functions in the mathematical
interfaces in a future release. The shdisc shape function is of particular interest.
Thanks.
Dave
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.