[sTeF88]

Hi everybody
I hope that someone could help me!!
I am trying to simulate the transient flow throught a rotating splinkler, but I have some issues that I can't resolve by myself.
I have attached my .cvg model. I am interested to the interaction between fluid and walls, so I have choose the FSI physical model. In the model you can see two interface regions that I have introduced because of I thinked (probably incorrectly) that in this way the mesh between two region in motion will automatically refit.
After some iteration, the simulation crashes. I am pretty sure that the problem is the deformation of the mesh during the wall rotation.
I'd like to ask you:
1- Is the model wrong? Is there a different way of modeling the phenomenon?
2 - In the case the model was right, may I resolve the issue by introducing some seals? In this case, where I have to introduce the seals (I have observed the Wankel-SI example case, but the situation is different!)
I really trust in some help. Thanks in advance
Stef

[ksrivast]

Hello Stefano,

To begin with and clarify, it is important to note that CONVERGE adopts a cell-cutting approach on a stationary grid to construct your mesh. The mesh itself does not move or deform when boundaries move. The grid is just cut differently, after every time-step, to create a mesh during run-time.

Currently, you have your FSI boundaries directly attached to the stationary wall boundaries. When boundaries move, each vertex associated with the boundary will move. Thus, the FSI motion will stretch the triangles of any adjacent non-moving boundaries it is connected and shares vertices with. This behaviour is unphysical, and will lead to triangle intersections (which we do not allow within CONVERGE). To tackle this issue, we recommend using sealing or reloft boundaries.

We would be happy to guide you in order to construct a working case setup. Please contact support@convergecfd.com . Please use your official email for all correspondence with Convergent Science. Please mention the issues you were facing, attach the current case setup and add my name, and the cfd-online thread, as reference.

Sincerely,

[sTeF88]

Dear Kislaya Srivastava
First, thanks for your reply. I have already sent an email to CONVERGE support, but sincerly I have not added your name because I sent it before your reply on cfd-online forum.
In the email I have attached a model that is slightly different from the one I have attached here. But the issue is the same, i.e. "when boundaries move, each vertex associated with the boundary will move. Thus, the FSI motion will stretch the triangles of any adjacent non-moving boundaries it is connected and shares vertices with"
I have tried to correct the issue by using sealing, but I was not able to resolve it.
Awaiting a response from official support, I would like to ask you what do you mean by "reloft boundaries"?. I suppose the problem is that the vertex of the rotating boundary and the stationary ones are not properly aligned.
Could you give me an indication of how to approach the model?
Thanks in advance
Best Regards
Stefano

[ksrivast]

Hello Stefano,

Our support team will have a look at your case setup and guide you towards creating a working simulation suiting your needs.

If a WALL boundary connects two other WALL boundaries that are undergoing relative motion (e.g. , a connection between a rotating shaft and an annular housing),the connecting boundary's triangulation may become degenerate (as explained in my previous post). The RELOFT boundary will direct CONVERGE to calculate the quality of the triangulation at each time-step and re-triangulate the boundary if necessary. If any triangle on the RELOFT surface becomes too malformed or skewed, CONVERGE will re-triangulate the surface. This will allow for direct connection of moving and non-moving boundaries for certain boundary motions.

For more details on how to properly employ RELOFT boundaries, please have a look at our manual (v2.4/v3.0).

Sincerely,

[sTeF88]

Dear Kislaya Srivastava
I have tried to assign a RELOFT wall BC, by modifying the boundary.in file but it doesn't seem to work (the analysis starts but no reloft occurs).
Is there some flag to activate or a hidden file to insert (my version is 2.4.20)?
Thanks in advance
Stefano

[ksrivast]

Hello Stefano,

There have been some bug fixes relating to RELOFT boundaries, that are documented in our release notes. Please check to see if you're using an updated version of CONVERGE. If RELOFT boundaries are set and are being read in correctly, CONVERGE will mention it in the logfile.

If you still face issues, please include this in your query to our Support team.

Sincerely,