[Yassin]

Respected colleagues;

I want to mesh pipe with twisted plate in ICEM for generating O-Grid mesh.
I tried to do it. I could not able to sperate block for solid geometry (twisted plate) from the fluid block.
does anyone know to solve this problem?

[Stel]

Could you attach an image showing the geometry that you want to mesh?

[Yassin]

My geometry is a straight pipeline having twisted tape as seen in attached.
I generated block in ICEM and I assign middle one to a solid part.

[Stel]

Right, now I see what might be going on.

Well, first of all I should say that this is not an easy O-Grid mesh. Besides the usual O-Grid split that we use for regular pipes, you should also perform two longitudinal split blocks to isolate the twisted tape. This is to create a thin block close to the center which should be deleted (since you don't want volume mesh inside the tape solid body). A easier way to visualize why this is needed is by imaging that you have, instead of a twisted tape, a flat one stretching from the inlet to the outlet of the pipe, such that the pipe is divided into two volumes (in fact this is a good example which you could try first in order to practice the block splitting before trying the twisted one).

Now the hardest part: since the tape is twisted, you will have to perform a face to surface association between the faces of the O-Grid facing inwards and the faces of the tape point to the flow (and possibly some edge to curve associations to project the edges of the blocks to the curves that form the intersection between the tape and the pipe wall). You could try that without twisting the blocks themselves, but I'm not pretty sure that this is going to work, so maybe further vertex associations / move vertex commands should be needed.

Yeah, kinda tricky, but it should work.

Good luck.

[bluebase]

Originally Posted by Stel

You could try that without twisting the blocks themselves, but I'm not pretty sure that this is going to work, so maybe further vertex associations / move vertex commands should be needed.

I believe, you need to twist the blocking downstream, since this would also make the mesh follow the flow direction.

Anyhow, i agree with the rest of Stel's description.

[Yassin]

Respected colleagues ;

Thanks a lot for your valuable comments,
Initially, I start doing mesh for a straight plate instead of the twisted one for practicing.
I split the block and I generate the o-grid. However, I got some bad quality for some elements as seen in figure 2.jpg.

I have the following inquiry:
1. After generating o-grid, is there any way to generate o-grid without hatch area as seen in the attached (named as 3.jpg)
or is there any way to make hatch block in figure (4.jpg) similar to block shown in figure 1.jpg
2. how to do twist block?
regards

[Stel]

Not really sure if I understood what you want, but one point: while generating the O-Grid, you have to indicate which faces the O-Grid structure should be normal to, otherwise ICEM will generate trapezoidal blocks in every direction. Is this the problem that you are having right now?

[bluebase]

Originally Posted by Yassin

Respected colleagues ;

Thanks a lot for your valuable comments,
Initially, I start doing mesh for a straight plate instead of the twisted one for practicing.
I split the block and I generate the o-grid. However, I got some bad quality for some elements as seen in figure 2.jpg.

I have the following inquiry:
1. After generating o-grid, is there any way to generate o-grid without hatch area as seen in the attached (named as 3.jpg)
or is there any way to make hatch block in figure (4.jpg) similar to block shown in figure 1.jpg
2. how to do twist block?
regards

Hi Yassin,

there is a lot to unpack here.

You approach, to do a simpler subset of your original problem is always a very good idea! That way, you can figure out systematic or methodical errors much easier.
You identified correctly, that a flat plate is not much different from a twisted plate. The required blocking structure is essentially the same.
Let's keep working on the flat plate for now.


in image 1 (1.jpg) you correctly identified, that the two topologies (the inlet region, and the plate region) do not match. Actually as they are, they cannot be matched in a sensible way.
I guess, that image 3 was a previous step. image 3 is much closer to your solution than image 1.
You need to acquire the skill to understand why the two ogrids around the plate cannot fit into the blocking structure of your inlet and outlet.

In the blocking few, can you see those thin strips in the color of blocks? Those are actual blocks, but degenerated by you.

These blocks are still used to create elements, but at "zero" thickness all the elements will also have zero thickness, or some weird shape.

Blocks should also not intersect itself. Inspect the marked locations here:

4.jpg

Furthermore, you seem to have deleted more blocks then necessary. That might be the reason why you have bad elements along the edge of the plate.


I suggest the following blocking approach:
baffle_low.png
The essential structure at every cross section of the blocking should lock the same in your case. That's why i concentrated on that section only. just add splits along the axis where necessary.

1. block initialization. Associate to inlet and outlet curves and points.
2. create the ogrid. select the inlet and outlet face, and the only block to generate the ogrid in axial direction.
3. Split the ogrid layer in circumferential direction.
4. Add two radial splits to accommodate the plate
5. shift vertices to make blocks "more rectangular" (all angles of blocks should be as close to 90 degree as possible, but not more than 135, or less than 45)
6. make splits in axial direction, and associate the plate geometry
7. optionally, if you need a resolved boundary layer at the plate, you can introduce further splits to finetune the boundary layer.

At the end, remove only those blocks which are inside the plate. I guess it is best to not delete them permanently.



Best,
Sebastian


Edit: I forgot to put a background in the sketch. Now some information is masked in the black background of the forum gallery view... maybe click on the image in the gallery view again to see it separately

[Stel]

Originally Posted by Yassin

Respected colleagues ;

2. how to do twist block?
regards

What I usually do is a kind of a manual process, I create a point where I want to move a vertex to, and then I use the Associate Vertex option (click on the vertex you want to move and then the point where you want to move it). It will be a painful process to do that in your case (that would require some splits in the axial direction to twist the block little by little in the axial direction) but it should work.