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 The information provided may be out of date.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.

Extremely thin geometries

Please login with a confirmed email address before reporting spam

I created a cylinder with radius 1e-6 m and thickness 9e-10 m in 3D to make a continuum simulation of graphene. I have initial radial stress, an uniform body load along the thickness of the cylinder and use fully clamped boundary conditions on the radial edge. When I try to mesh it the process never finishes, and earlier when I had a mesh on a similar geometry the simulation didn't converge.

Question: What do I need to consider while trying to simulate extremely thin geometries?

7 Replies Last Post 25.03.2014, 00:14 GMT-4
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 10.09.2012, 07:57 GMT-4
Hi

it depends on your geoemtry overall dimensions, if averything is small 1e-9 nm, apart settign the absolute tolerance to < 1e-9 I do not see what you need to tweak.

The difficulty is when you compare a 1mm^3 obhect and you add a small part about 1nm^3 inside, then the ratio of size > 1:10'000 or slightly more qill give you some issues, I never have tw object in direct contact with size ration above 1:10'000, I then add some local internal boundaries around the small parts to respect my size ratio. Then i manually mesh these small parts first and increase the mesh per layer until I com eto the "mocro parts.

This mostly runs nicely and smoothely, but it oes not work in 1click automatic ;)

--
Good luck
Ivar
Hi it depends on your geoemtry overall dimensions, if averything is small 1e-9 nm, apart settign the absolute tolerance to < 1e-9 I do not see what you need to tweak. The difficulty is when you compare a 1mm^3 obhect and you add a small part about 1nm^3 inside, then the ratio of size > 1:10'000 or slightly more qill give you some issues, I never have tw object in direct contact with size ration above 1:10'000, I then add some local internal boundaries around the small parts to respect my size ratio. Then i manually mesh these small parts first and increase the mesh per layer until I com eto the "mocro parts. This mostly runs nicely and smoothely, but it oes not work in 1click automatic ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11.09.2012, 04:50 GMT-4
Thanks,

When I use "User controlled" mesh and scale the z-axis with 1e4 the meshing goes fine. But I get extremely dense meshing in the z axis whish is the opposite of what I want. But when I set x and y to 1e4 I get

"Mesh warning (ftet1): Failure due to incomplete boundary mesh."

Best
Martin
Thanks, When I use "User controlled" mesh and scale the z-axis with 1e4 the meshing goes fine. But I get extremely dense meshing in the z axis whish is the opposite of what I want. But when I set x and y to 1e4 I get "Mesh warning (ftet1): Failure due to incomplete boundary mesh." Best Martin

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 11.09.2012, 14:54 GMT-4
Hi

depending on what you are doing, then a regular mesh (square or "high" mesh "quality") might not be the optimum, sometimes it is imporat to have several mesh elements in the "thin" thickness direction butthe mesh elements might have a large extent in the perpendicular direction (very model and dependent variable gradient field dependent, be aware) so your mes does not necessarily need to be "regular", with ths elements, I mostly mesh the surface, and thn sweep it through the thickness with a few elements depending on my expected results. Or I mesh the short edges, before the full domain

Meshing is somewhat an "art" but with some sensible scientific touch, from the preknowledge of expected results

--
Good luck
Ivar
Hi depending on what you are doing, then a regular mesh (square or "high" mesh "quality") might not be the optimum, sometimes it is imporat to have several mesh elements in the "thin" thickness direction butthe mesh elements might have a large extent in the perpendicular direction (very model and dependent variable gradient field dependent, be aware) so your mes does not necessarily need to be "regular", with ths elements, I mostly mesh the surface, and thn sweep it through the thickness with a few elements depending on my expected results. Or I mesh the short edges, before the full domain Meshing is somewhat an "art" but with some sensible scientific touch, from the preknowledge of expected results -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 12.09.2012, 03:43 GMT-4
Hi,

I just want to mesh a cylinder with a high 1e-10m and radius 1e-6 m . Then I want to fix the radial boundary and apply a body force along the z-axis.

I don't understand how I can mesh first the radial border and then sweep this inward in the radial direction? I can't do this with "Build all", right? I think it should be just fine with very few elements in z and some powers more in the radial direction?

Best
Martin
Hi, I just want to mesh a cylinder with a high 1e-10m and radius 1e-6 m . Then I want to fix the radial boundary and apply a body force along the z-axis. I don't understand how I can mesh first the radial border and then sweep this inward in the radial direction? I can't do this with "Build all", right? I think it should be just fine with very few elements in z and some powers more in the radial direction? Best Martin

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 19.09.2012, 01:58 GMT-4
Hi

with such differences in shape, you should add a few intermediate boundaries to allow the mesher to mesh region by region (domain ba domain) with at most 1:1000 to 1:10'000 size ratio

--
Good luck
Ivar
Hi with such differences in shape, you should add a few intermediate boundaries to allow the mesher to mesh region by region (domain ba domain) with at most 1:1000 to 1:10'000 size ratio -- Good luck Ivar

Ruud Borger COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 19.09.2012, 11:20 GMT-4
Hi Martin,

As Ivar says, a Swept mesh would be best suited for these type of geometries. You can set it up as follows:
-Mesh the upper surface area, with a free triangular or free quad. If desired, right click and add a Size to specify the size.
-Next, add a Swept mesh. By right clicking this, you can determine how many layers you want through the thickness.

Attached is an example.

Best regards,

Ruud Börger
COMSOL
Hi Martin, As Ivar says, a Swept mesh would be best suited for these type of geometries. You can set it up as follows: -Mesh the upper surface area, with a free triangular or free quad. If desired, right click and add a Size to specify the size. -Next, add a Swept mesh. By right clicking this, you can determine how many layers you want through the thickness. Attached is an example. Best regards, Ruud Börger COMSOL


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25.03.2014, 00:14 GMT-4
hi
I am struck in meshing, i have 3 layers, the last two are same. i mesh the surface of first layer and then swept it down. I select second domain but it is not sweping through it. please let me know if there any example of swept meshing for multi layers.
hi I am struck in meshing, i have 3 layers, the last two are same. i mesh the surface of first layer and then swept it down. I select second domain but it is not sweping through it. please let me know if there any example of swept meshing for multi layers.

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.