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 The information provided may be out of date.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.

Meshing of a thin hemispherical shell

Please login with a confirmed email address before reporting spam

Hi

I am trying to find the eigenfrequency of a hemispherical shell. This hemisphere has a diameter of 200 um and height of 140 um, and the shell thickness is 0.5 um.

The hemispherical shell is created by:
1. Using the boolean operation "Difference" to subtract a sphere of radius 99.5 um from a sphere of radius 100 um. This creates a spherical shell.
2. Using the boolean operation "Difference" to subtract the top portion of the sphere using a block drawn at z = 40 um.

I meshed with "Physics-controlled mesh" with element size "Coarser". Because of the relatively thin shell compared to the size of the hemisphere, the number of elements is pretty huge. The file size is too large to be attached here. Is it possible to use the "Swept" function to mesh the hemisphere similar to the way a large, thin rectangular plate is meshed? In meshing the thin rectangular plate, the source and destination faces are defined; the element size can be made large on the face of the plate, but the element size can be made very fine across the thickness of the shell. I would like to mesh the hemispherical shell the same way, with larger-sized elements throughout the face of the sphere, and smaller elements along the shell thickness. Does anyone have any suggestions on how that can be done? Many thanks.

Erwin

5 Replies Last Post 2012年7月25日 GMT-4 13:49
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年6月6日 GMT-4 00:53
Hi

often thin shell are better meshed as "shells" i.e. a surface with the thickness attached to the "physics" as you stated you get very many elements. Ideally you should have at least 2-5 elements in the thickness too !!

There are ways to revolve mesh with the sweep function, start from a 2D workplane (you might ness to do it in 2 *180° and not in one sweep. There were a thread or two about this anly a month or so ago, try a search

--
Good luck
Ivar
Hi often thin shell are better meshed as "shells" i.e. a surface with the thickness attached to the "physics" as you stated you get very many elements. Ideally you should have at least 2-5 elements in the thickness too !! There are ways to revolve mesh with the sweep function, start from a 2D workplane (you might ness to do it in 2 *180° and not in one sweep. There were a thread or two about this anly a month or so ago, try a search -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年7月19日 GMT-4 11:45
Hi Ivar,

Thank you for your suggestion. The shell feature does help in reducing the number of elements and speeding up the simulation. It is also more convenient to change the shell thicknesses to experiment with different values.

Apart from the eigenfrequency analysis, I am also performing time-dependent analysis to study how the displacement of the rim of the hemisphere changes with time. I fixed the bottom of the hemisphere, applied a sinusoidal load at one vertex of the hemisphere (not at the rim or bottom), and I plotted 1D graph of displacement vs time at a vertex on the rim. The rim has only 4 vertices for me to choose to study the displacement vs time. If I would like to study the displacement vs time at any point on the rim (not at the vertices), how do I do that? Is there a feature in COMSOL to extract all the data at every mesh point? Thank you once again.

Regards
Erwin
Hi Ivar, Thank you for your suggestion. The shell feature does help in reducing the number of elements and speeding up the simulation. It is also more convenient to change the shell thicknesses to experiment with different values. Apart from the eigenfrequency analysis, I am also performing time-dependent analysis to study how the displacement of the rim of the hemisphere changes with time. I fixed the bottom of the hemisphere, applied a sinusoidal load at one vertex of the hemisphere (not at the rim or bottom), and I plotted 1D graph of displacement vs time at a vertex on the rim. The rim has only 4 vertices for me to choose to study the displacement vs time. If I would like to study the displacement vs time at any point on the rim (not at the vertices), how do I do that? Is there a feature in COMSOL to extract all the data at every mesh point? Thank you once again. Regards Erwin

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年7月19日 GMT-4 14:09
Hi

you can define a cut line or cut surface by adding a node to your Data Set, and then plot along the line/surface

often I add several lines / surfaces to cut my volumes such to have several entities to pick and plot varables along these internal lines, it's handier than defining cut lines with interpolations (I find it so)

You can set up your own data lists in tables by adding nodes to the Derived Variables node, and then plot he table outputs

--
Good luck
Ivar
Hi you can define a cut line or cut surface by adding a node to your Data Set, and then plot along the line/surface often I add several lines / surfaces to cut my volumes such to have several entities to pick and plot varables along these internal lines, it's handier than defining cut lines with interpolations (I find it so) You can set up your own data lists in tables by adding nodes to the Derived Variables node, and then plot he table outputs -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年7月19日 GMT-4 14:29
Hi Ivar

Thank you. I will definitely give it a try.

Regards
Erwin
Hi Ivar Thank you. I will definitely give it a try. Regards Erwin

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年7月25日 GMT-4 13:49
Hi Ivar

I tried to add cut lines by drawing several more spheres and performing the necessary boolean operations and converting them to shells (as I am working on a thin hemispherical shell). However, I cannot "form a union" of my structures, and I cannot mesh.

Is there a way to export all the results to Matlab? I have saved my comsol simulation as .m file, and I opened the .m file in Matlab. However, can I define in Matlab, perhaps using coordinates, any arbitray point on the rim of the hemisphere the displacement-time graph I want to look at? My mesh element is very fine, and there are many mesh nodes along the rim. When I looked at the code, there is a line which looks for the displacement-time results as:

model.result('pg5').feature('ptgr1').selection.set([2]);
This means it is going to give me the displacement-time graph at vertex 2 (because of the [2]). I would like to specify a particular point by using coordinates.

Thanks again.

Regards
Erwin
Hi Ivar I tried to add cut lines by drawing several more spheres and performing the necessary boolean operations and converting them to shells (as I am working on a thin hemispherical shell). However, I cannot "form a union" of my structures, and I cannot mesh. Is there a way to export all the results to Matlab? I have saved my comsol simulation as .m file, and I opened the .m file in Matlab. However, can I define in Matlab, perhaps using coordinates, any arbitray point on the rim of the hemisphere the displacement-time graph I want to look at? My mesh element is very fine, and there are many mesh nodes along the rim. When I looked at the code, there is a line which looks for the displacement-time results as: model.result('pg5').feature('ptgr1').selection.set([2]); This means it is going to give me the displacement-time graph at vertex 2 (because of the [2]). I would like to specify a particular point by using coordinates. Thanks again. Regards Erwin

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.