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 a thin flat plate

Please login with a confirmed email address before reporting spam

I'm working on a 3D conduction solution of a spot heater on a thin flat plate with dimensions 61cm x 30.5 cm x 0.635 cm. This aspect ratio is challenging to mesh in a way that does not produce errors. I need advice on solving this meshing issue. I have attached my current solution.

Mike

5 Replies Last Post 2009年11月5日 GMT-5 10:54
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 2009年11月4日 GMT-5 01:46
Hi

First of all you have forgotten the attachment !
Second for me such a plate could be meshed straightforward with a surface mapped mesh and then a sweep with 2-3 elements in the thickness, no ?
This is the theory, I agree, because depending on the size of your central "hot spot" it could make life slightly tougher.
With mappd mesh I usually make also square geometries, it eases life, even by making some volumes not stricly required, such as if your "hot spot" is a surface elements on the top of the plate, I would make a full volume of the area of the "hot-spot" and going all the way through the plate (the plate is then split in 2 items/volumes/subdomains). You can then mesh them independently which is far easier with mapped meshing.

The fact that you make a few extra internal boundaries mostly does not make the model heavier, at least not in a significant way.

You can also try with the shell elements of the structural model for the plate, but I would run a few test comparing the volume calculations and the shell results, on a simpler model, just to check.

Finally you can always build up the full shell model by equations, but will probably take some more time, but you lear a lot of COMSOL internals like that

Good luck
Ivar

Hi First of all you have forgotten the attachment ! Second for me such a plate could be meshed straightforward with a surface mapped mesh and then a sweep with 2-3 elements in the thickness, no ? This is the theory, I agree, because depending on the size of your central "hot spot" it could make life slightly tougher. With mappd mesh I usually make also square geometries, it eases life, even by making some volumes not stricly required, such as if your "hot spot" is a surface elements on the top of the plate, I would make a full volume of the area of the "hot-spot" and going all the way through the plate (the plate is then split in 2 items/volumes/subdomains). You can then mesh them independently which is far easier with mapped meshing. The fact that you make a few extra internal boundaries mostly does not make the model heavier, at least not in a significant way. You can also try with the shell elements of the structural model for the plate, but I would run a few test comparing the volume calculations and the shell results, on a simpler model, just to check. Finally you can always build up the full shell model by equations, but will probably take some more time, but you lear a lot of COMSOL internals like that Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2009年11月4日 GMT-5 15:28
My mistake on the attachment. My spot is quite small (0.3175 cm diameter). I'm working through creating a mapped mesh. Would you recommend creating a cylinder for the spot, a cylinder of something like a 10 cm radius and then splitting the plate into 8 blocks? Or would it be better to forgo the bigger cylinder and just split the plate into 8 blocks with the hot-spot cylinder in the middle?

Mike
My mistake on the attachment. My spot is quite small (0.3175 cm diameter). I'm working through creating a mapped mesh. Would you recommend creating a cylinder for the spot, a cylinder of something like a 10 cm radius and then splitting the plate into 8 blocks? Or would it be better to forgo the bigger cylinder and just split the plate into 8 blocks with the hot-spot cylinder in the middle? Mike


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 2009年11月5日 GMT-5 02:58
Hi

I ment not necessarily a cylinder, if you use mapped mesh (quads) use rather rectangles or cubes.

if yo have a large ratio of size try using the "russian doll" principle, add a few cubes/rectangles around the small spot with indreasing sizes.

The principle is that the FEM mesh should be adapted to the gradients of your (heat) flux, with a small hot spot on a large plate you need a coarse mesh "far" away but very dense locally. If the ratio is too big COMSOLwill not mesh, then by adding an intermediate boundary (or several) and starting with the smallest one then normally you can get around. You can also increase the default mesh sizes as you get further apart

Good luck
Ivar

PS: sorry I have no time to open your model just now
Hi I ment not necessarily a cylinder, if you use mapped mesh (quads) use rather rectangles or cubes. if yo have a large ratio of size try using the "russian doll" principle, add a few cubes/rectangles around the small spot with indreasing sizes. The principle is that the FEM mesh should be adapted to the gradients of your (heat) flux, with a small hot spot on a large plate you need a coarse mesh "far" away but very dense locally. If the ratio is too big COMSOLwill not mesh, then by adding an intermediate boundary (or several) and starting with the smallest one then normally you can get around. You can also increase the default mesh sizes as you get further apart Good luck Ivar PS: sorry I have no time to open your model just now

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2009年11月5日 GMT-5 10:14
We're on the same path. Maybe a larger cylinder makes the transition smoother? I've attached a new model with mapped mesh. This model is fairly resource intensive, so I'm going to investigate reducing the number of elements. As you eluded to before, the small spot size creates some difficulties for the mapped mesh. I'd appreciate any feedback you may have about the mesh.

Mike
We're on the same path. Maybe a larger cylinder makes the transition smoother? I've attached a new model with mapped mesh. This model is fairly resource intensive, so I'm going to investigate reducing the number of elements. As you eluded to before, the small spot size creates some difficulties for the mapped mesh. I'd appreciate any feedback you may have about the mesh. Mike


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 2009年11月5日 GMT-5 10:54
Hi

indeed it gets heavy, I would suggest to try (but this depends on how much more you want to add on the model) to use 1/4 with symmetry BC's, then you could also simulate the plate thickness as an equation w.r.t. a surface element.

Finally, if you look at the temperature gradients, I beleive you are overdoing it, I would use far larger meshes, quicker when you get away from the hot-spot

good luck
Ivar
Hi indeed it gets heavy, I would suggest to try (but this depends on how much more you want to add on the model) to use 1/4 with symmetry BC's, then you could also simulate the plate thickness as an equation w.r.t. a surface element. Finally, if you look at the temperature gradients, I beleive you are overdoing it, I would use far larger meshes, quicker when you get away from the hot-spot good luck Ivar

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.