[Bourke]

Hey,

I've looked through the forums trying to find a similar issue to this one but none of them quite helped me solve my issue so I thought it best to make a new post.

I'm modelling a 2D flow around a cylinder which contains an internal cavity and orifice which is connected to the external flow.
Within the cavity, there is a membrane which I'm trying to model as a velocity inlet.

I'm using ICEM to create the geometry and the mesh from there. Once the mesh is created, I try and directly load it into fluent or import the created mesh file.

I define my parts and boundary conditions in ICEM, but they change somewhat when I'm loading the ICEM mesh in fluent. This is not the major problem because I can identify which boundary condition should be which when in fluent, but for some reason, my membrane part (which I want to define as a velocity inlet) becomes merged into another part of the geometry.
I make sure to associate all edges and points when I'm creating the geometry and I've tried following the online ANSYS tutorials to make sure everything is alright, but I can't figure out why the membrane section is becoming merged into the other geometry.

If there is any advice you can lend me for this problem, it would be very appreciated.
Thanks a lot!

P.S. I'm attaching some images which will hopefully clarify what I'm describing.

[bluebase]

Hi Bourke,

i don't have a solution to your problem, just a few suggestions to isolate the cause.
Reassociate alle curves, edges,... to the corresponding parts. Especially make sure that boundary curves are split into seperate parts according to your designed boundary conditions. (The velocity inlet curve should be an individual curve and not continously attached to the wall curve).
Then reconvert the structured mesh to unstructured after you changed the associations. And, export it to fluent again.

Will this solve your problem?

Apart from the original problem, i have a small suggestion to your blocking inside the cavity. It needs an additional o-grid to improve mesh quality in the "corners" of the left and right block next to the central block of the cavity. See the attachement.
Also, on top of the cavity, the mesh jumps from a very small size to a big size. This will lead to lower accuracy and potential convergence issues in this area. Try to keep the change of volume/area between neighbouring elements low.

with regards,
Sebastian

cavitymesh.jpg

[Bourke]

Hi Bluebase,

Thanks for your reply,

Yeah, I've made sure that the boundary curves are separate but for some reason during the process, they become merged to the wall curve.
Its just a frustrating issue when you have to keep re-associating and then setting up the mesh again so I was just wondering if there was something blatantly obvious that I was doing wrong.

In terms of the mesh, thanks for your advice.
The mesh I posted a picture of is not final, it was just a sample in order to demonstrate the geometry.
But yes, I think I will add another o-grid to help the mesh quality. Also yes, I intend to smooth the transition from the mesh atop the synthetic jet to the surrounding mesh.

Thanks again for your consideration!

[Upyoung]

Originally Posted by bluebase

Hi Bourke,

i don't have a solution to your problem, just a few suggestions to isolate the cause.
Reassociate alle curves, edges,... to the corresponding parts. Especially make sure that boundary curves are split into seperate parts according to your designed boundary conditions. (The velocity inlet curve should be an individual curve and not continously attached to the wall curve).
Then reconvert the structured mesh to unstructured after you changed the associations. And, export it to fluent again.

Will this solve your problem?

Apart from the original problem, i have a small suggestion to your blocking inside the cavity. It needs an additional o-grid to improve mesh quality in the "corners" of the left and right block next to the central block of the cavity. See the attachement.
Also, on top of the cavity, the mesh jumps from a very small size to a big size. This will lead to lower accuracy and potential convergence issues in this area. Try to keep the change of volume/area between neighbouring elements low.

with regards,
Sebastian

Attachment 43966

dear,

I just came across the same situation, my experience is delete all curves at once ICEM opened, then draw then by connecting two points other than do a topology repairing.


hope this will help.

[Bourke]

So I've had to just re-associate curves again when this issue is encountered, it seems there's no way around it and I'm not sure why it happens.

Otherwise I've tried improving the mesh. I've encountered another issue however (of course ) where when I'm trying to set edge parameters, despite not having "copy edge parameters" on, the parameters from 1 edge copies to another.
Really frustrating cos I want the mesh to be uniform all the way around the cylinder up until above the synthetic jet orifice.

Its clearer in the image attached what I mean.
When I set edge parameters for edge 1, they auto copy to edge 2 and I end up with a really dense mesh in the bottom patch, whereas the side edges of the cylinder stay the same if I change them.

Any advice?

[bluebase]

Hi Bourke,

this issue might rise up when you delete blocks. When you delete a block without the permanent option, then the connectivity between blocks is maintained. You can check this in the part VORFN.
To solve you problem, activate the VORFN part and delete a block permanently which is attatched to your south-facing block. Now, all shadow connectivity is removed. So, your east and west blocks will probably be detatched, too.

With regards,
Sebastian

[Bourke]

Originally Posted by bluebase

Hi Bourke,

this issue might rise up when you delete blocks. When you delete a block without the permanent option, then the connectivity between blocks is maintained. You can check this in the part VORFN.
To solve you problem, activate the VORFN part and delete a block permanently which is attatched to your south-facing block. Now, all shadow connectivity is removed. So, your east and west blocks will probably be detatched, too.

With regards,
Sebastian

Sebastian,

This indeed did work. Thanks for your help!