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 packed beads

Please login with a confirmed email address before reporting spam

Hello,

I would like to seek for some advice when meshing a packed beads contained in a box. Screenshots are attached.

I have 100 spherical beads (radius~2mm) packed inside a box whose wall has a finite thickness (~1mm). Each bead-bead and bead-wall boundary make contacts with a small circular face (rather than a point-contact). Packed beads make contacts with an enclosing box on 5-sides, including 4 walls and a floor. This is a microwave heating model so the maximum element size during meshing is governed by dielectric constant of the material and wavelength, which are known. Beads are assumed to be zeolite and box is assumed to be mica.

I tried refining meshes multiple times but I cannot seem to get around the following error messages

  • Face is (or has a narrow region that is) much smaller than the specified minimum element size.
  • Domain has a region that is much thinner than the specified minimum element size.

These errors seems to occur on the outer boundaries of the beads around its circular contact with other beads, and also on the inner domain of the box that surrounds the packed beads (see attached). The minimum and average quality of the mesh are 0.08 and 0.63, respectively. This is after applying 'remove details' virtual operation on the geometry. I mostly adjusted the size of the mesh to each domain to setup the mesh (attached). Attempts to use swept mesh were thus far unsuccessful. How can I improve the mesh?

I would appreciate any suggestions!



2 Replies Last Post 2021年8月15日 GMT-4 22:24
Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 3 years ago 2021年8月10日 GMT-4 20:10
Updated: 3 years ago 2021年8月10日 GMT-4 20:11
  1. Those are not error messages. They are warning messages. The key question is whether such warnings imply a risk to the validity of the model. Sometimes these types of warnings are important and you have to address them. But also, believe it or not, sometimes these warnings can be safely ignored.
  2. Your statement that "the maximum element size during meshing is governed by dielectric constant of the material and wavelength..." is a bit of an oversimplification. You also need to keep elements small compared to scale-lengths of any high-gradient field regions and regions with detailed small geometries (often the same as the high-gradient field regions), if those regions contribute meaningfully to the physics you are trying to model. Sometimes, such regions are much smaller than would be naively set by overly-simple rules like always setting the max element size < lambda/N, where N is some modest number (typically between 6 and 20). The trick here is knowing what (level of detail to) a mesh you can throw away, and knowing what to keep! (Just ask Kenny Rogers about that!).
  3. In your case, you need to do the same kind of assessment in regard to thermal modeling, in addition to the RF modeling. That is, your mesh needs to be suitable for both.
  4. You can (and should) set the mesh sizes manually. Don't let the code choose the mesh for you based on the physics. You can find mesh setting options pretty easily. Set them in volumes, on surfaces, and maybe even on edges, if you need to, thinking about the physics in all cases and trying to understand what matters and what doesn't. By the way, I also recommend starting with a simpler model than your 100-beads case. Walk first, then jog, then run!
-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
1. Those are not *error* messages. They are *warning* messages. The key question is whether such warnings imply a risk to the validity of the model. Sometimes these types of warnings are important and you have to address them. But also, believe it or not, sometimes these warnings can be safely *ignored*. 2. Your statement that "the maximum element size during meshing is governed by dielectric constant of the material and wavelength..." is a bit of an *oversimplification*. You also need to keep elements small compared to scale-lengths of any high-gradient field regions and regions with detailed small geometries (often the same as the high-gradient field regions), *if* those regions contribute meaningfully to the physics you are trying to model. Sometimes, such regions are *much smaller* than would be naively set by overly-simple rules like always setting the max element size < lambda/N, where N is some modest number (typically between 6 and 20). The trick here is knowing what (level of detail to) a mesh you can throw away, and knowing what to keep! (Just ask Kenny Rogers about that!). 3. In your case, you need to do the same kind of assessment in regard to thermal modeling, in addition to the RF modeling. That is, your mesh needs to be suitable for both. 4. You can (and should) set the mesh sizes manually. Don't let the code choose the mesh for you based on the physics. You can find mesh setting options pretty easily. Set them in volumes, on surfaces, and maybe even on edges, if you need to, thinking about the physics in all cases and trying to understand what matters and what doesn't. By the way, I also recommend starting with a simpler model than your 100-beads case. Walk first, then jog, then run!

Please login with a confirmed email address before reporting spam

Posted: 3 years ago 2021年8月15日 GMT-4 22:24
  1. Those are not error messages. They are warning messages. The key question is whether such warnings imply a risk to the validity of the model. Sometimes these types of warnings are important and you have to address them. But also, believe it or not, sometimes these warnings can be safely ignored.
  2. Your statement that "the maximum element size during meshing is governed by dielectric constant of the material and wavelength..." is a bit of an oversimplification. You also need to keep elements small compared to scale-lengths of any high-gradient field regions and regions with detailed small geometries (often the same as the high-gradient field regions), if those regions contribute meaningfully to the physics you are trying to model. Sometimes, such regions are much smaller than would be naively set by overly-simple rules like always setting the max element size < lambda/N, where N is some modest number (typically between 6 and 20). The trick here is knowing what (level of detail to) a mesh you can throw away, and knowing what to keep! (Just ask Kenny Rogers about that!).
  3. In your case, you need to do the same kind of assessment in regard to thermal modeling, in addition to the RF modeling. That is, your mesh needs to be suitable for both.
  4. You can (and should) set the mesh sizes manually. Don't let the code choose the mesh for you based on the physics. You can find mesh setting options pretty easily. Set them in volumes, on surfaces, and maybe even on edges, if you need to, thinking about the physics in all cases and trying to understand what matters and what doesn't. By the way, I also recommend starting with a simpler model than your 100-beads case. Walk first, then jog, then run!

Robert, thank you for responding with an insightful post! I tested out some mesh variations and have been getting consistent results even with those warning messages. Your explaination that they are not 'error' messages and that these messages should be used as an indicator rather than an instruction helped me build confidence with the mesh. I appreciate your kind insights!

>1. Those are not *error* messages. They are *warning* messages. The key question is whether such warnings imply a risk to the validity of the model. Sometimes these types of warnings are important and you have to address them. But also, believe it or not, sometimes these warnings can be safely *ignored*. >2. Your statement that "the maximum element size during meshing is governed by dielectric constant of the material and wavelength..." is a bit of an *oversimplification*. You also need to keep elements small compared to scale-lengths of any high-gradient field regions and regions with detailed small geometries (often the same as the high-gradient field regions), *if* those regions contribute meaningfully to the physics you are trying to model. Sometimes, such regions are *much smaller* than would be naively set by overly-simple rules like always setting the max element size < lambda/N, where N is some modest number (typically between 6 and 20). The trick here is knowing what (level of detail to) a mesh you can throw away, and knowing what to keep! (Just ask Kenny Rogers about that!). >3. In your case, you need to do the same kind of assessment in regard to thermal modeling, in addition to the RF modeling. That is, your mesh needs to be suitable for both. >4. You can (and should) set the mesh sizes manually. Don't let the code choose the mesh for you based on the physics. You can find mesh setting options pretty easily. Set them in volumes, on surfaces, and maybe even on edges, if you need to, thinking about the physics in all cases and trying to understand what matters and what doesn't. By the way, I also recommend starting with a simpler model than your 100-beads case. Walk first, then jog, then run! Robert, thank you for responding with an insightful post! I tested out some mesh variations and have been getting consistent results even with those warning messages. Your explaination that they are not 'error' messages and that these messages should be used as an indicator rather than an instruction helped me build confidence with the mesh. I appreciate your kind insights!

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.