Discussion Forum

Dear all,

I have a problem with the mesh I am using, and I wonder if someone could help me out here. I'll try explaing it using the attached files, I apologize for the length of this post.

The file "Screenshot_20131122.png" shows the general structure of my "Mesh 2".
It also shows my geometry, which consists of an array of 3 x 3 blocks and - on the back of each block - electrodes with a size in x-y slightly smaller than the x-y areas of the blocks.

For the mesh, I want to define the following:
- A "free triangular mesh" for the front and back surfaces
- A "Swept" mesh for the blocks.
Also, for the central block (marked green in the Screenshot picture), I want a "Extremely fine" mesh both for surfaces as the block. For the neighbour blocks I want an "Extra fine" mesh.

Hence, both for "Free Triangular 1" as for "Swept 1", I have two Sizes: "Size 1" ("Extra fine") and "Size 2" ("Extremely fine").

There is also a general "Size" node just below "Mesh 2", which I have set to "Extremely Fine", though I am not sure what it serves for... Does this general size node overwrite some of the other sizes set in the sub-nodes? (It seems like that actually).

In "Free Triangular->Size 1", I select the front and back surfaces of all blocks except the center block. Hence, these are 24 surfaces in total, (8 in front and 16 in the back: both the back surfaces of the blocks and the front surfaces of the electrodes).

In "Free Triangular->Size 2", I select the surfaces for the center block. Hence, these are 3 surfaces: the front surface, the block back surface, and the electrode front surface (see CenterPixel_Surfaces.jpg).

In the Free Triangular general node, I select only the back surfaces of all the blocks (i.e. 18 surfaces in total).

For "Swept 1->Size 1", I select all neighbouring blocks (i.e. 8 in total).
For "Swept 1->Size 2", I select the center block (which has label number "8").

Fine. This all works. (I.e, I can build the mesh and then run the simulation on the mesh).
However, I have the feeling this is not completely correct.

-------//

For instance, to me it seems wrong to have only the back surfaces of the blocks in the general "Free Triangular" node selection. So, I tried putting all surfaces (front and back) into this selection.
(I.e., all surfaces from "Free Triangular-> Size 1" plus all surfaces from "Free Triangular-> Size 2".)

Then I get the error message (which appears under the Swept node:)
- "Source and destination faces must be specified"

------//

So, I guess I had to specify the source and destination faces inside the general "Swept 1" node. (Which were empty up till now).

So, for "Swept 1 -> Source Faces", I put all the front surfaces (9 in total) and for "Swept 1 -> Destination Faces", I put all the back surfaces (18 in total). Now I get the following error:
- "Each destination face must uniquely correspond to one or more source faces."

------//

When I do it the other way around: for "Swept 1 -> Source Faces", all the back surfaces (18) and for "Swept 1 -> Destination Faces", all the front surfaces (9 in total), I get the following error:
- "Failed to create swept mesh for domain."
-- "Mesh on source and destination face do not match."

------//

My conclusion is that the number of source faces and destination faces should be equal. Is that right?

So I tried that: for "Swept 1 -> Source Faces", all the front surfaces (9) and for "Swept 1 -> Destination Faces", only the "electrodes" back surfaces (9).
Now I get the following error:
- "Unsupported topology of domain." (with one corner block highlighted)
- "Unsupported topology of linking face." (with a back surface of the corner block highlighted)

I can understand this, since part of the "swept block" does not have a surface defined on the back end. But then, I don't see how to correctly define the mesh, unless I use the original set-up.
However, I have the feeling that I am doing something wrong there too, even though COMSOL does not complain.

I do apologize for this rather long (and maybe confusing) message, but I don't see how better to explain it.
Any help would be very much appreciated.

Sincerely,

Machiel