[MTabone]

Hi, I have spent a very long time trying to replicate the Co-Simulation tutorial (but with my own geometry not the tutorial's imported geometry) Abaqus Co-Simulation: Mechanical Coupling, to be able to perform FSI through co-simulation between Star CCM+ and Abaqus but unfortunately I've had trouble with creating my own geometry and correctly meshing it. What happens is that I get a Negative Cell Volume Error usually after over a 1000 iterations, but changes depending on the mesh that I create. I am pretty certain the negative cell volume errors are caused by the prism layers at the base of the plate. I have successfully run the Tutorial version without issues but I have not been able to understand how to correctly re-create the geometry myself which sounds like something simple but I can't seem to understand how to get the prism layers to wrap around the plate (like in the tutorial's imported Model) rather than embed themselves into the floor below plate.

Please click on pictures to see pictures and to understand what is happening and what is my issue:

This is a view of the full model:
WholeMeshView.jpg


This is what I'm aiming for (Tutorial's Mesh):
CorrectPrismLayers.PNG


And this is what I'm getting:
MyMesh.jpg


What I'm doing to start my own simulation from scratch:

I create all the geometry in Star CCM+ by just right-clicking and creating new a new block part for the plate which I create first. I have all the coordinate geometry correctly mapped out with the Abaqus geometry (Plate) which I am very confident is all correct and which I made myself in Abaqus as well. Then I create another block part around the plate which is the fluid domain. The bottom face of the plate merges with the floor of the fluid domain. (Could I be doing something wrong creating the geomtry like this?)

I then generate the mesh selecting Polyhedral Mesh, Surface Remehser and Prism Layer Meshers. I only change the base size of the Mesh and refine the plate cell size a little more for a higher mesh concentration on the plate. This how I usually create my mesh. Then I just generate a volume mesh. (I know I could use the surface wrapper but it doesn't seem to help me as I have limited experience and the tutorials I've done haven't seemed to be applicable to this sort of thing).

The simulation can perform well but half the time I get a Negative Cell Volume Error depending on the Mesh due to mesh morphing and the other half I have fully working simulations which actually do what I want it to, but if I could fix this problem it would be a major bonus as this is a University project which is very important for me and I need to get this right.

I'm very new to Star CCM+ and my understanding of meshing is quite basic but I've done a lot of the tutorials and I understand that the Surface Wrapper could potentially help me but I have no idea what options to use in the surface wrapper and I've been playing with it but haven't been able to make any progress. Any advice to point me in the right direction would be greatly appreciated.

Thanks Again for taking the time to read this. Any advice on what I could do would be greatly appreciated as it will solve a problem I've been working on for longer then I care to think.

[Bin]

Hi MTabone,

I am interested to study the co-simulation and FSI as well. Do you mind to send me the tutorial so that I can learn from it?

my email address is: bin6331@hotmail.com

Thank a lot in advance.

As for your meshing problem, I'm not sure whether this helps or not, but maybe you can try applying 'thin mesher' on the flat plate when you generate the mesh. Thin mesher seems like very useful when meshing a thin structure.

Regards,
Bin

[CFDPosterBoy]

In the prism layer expert properties, take a look at the Concave Angle Limit (that's my first guess). My second guess is the Boundary March Angle, but that shouldn't come into play since both surfaces (the bottom and the outside of the flap) have prism layers.

In my experience, meshing and FSI simulation stability tend to be improved by using overset meshing instead of mesh morphing. You might want to look into that. Basically, the 'background' mesh is just the open volume (ignore the flap). Then, a second 'overset' mesh that wraps the surface of the flap is set on top of the background. The cells in the overset are allowed to morph, but since that mesh is only a few elements thick, you avoid a lot of the problems of large deflection mesh morphing.

If this sounds like it might work for you, try doing the tutorial on overset meshing then hit me up for some more details. I do this stuff everyday, so I've learned a few tricks.

Good luck!

[MTabone]

Originally Posted by CFDPosterBoy

In the prism layer expert properties, take a look at the Concave Angle Limit (that's my first guess). My second guess is the Boundary March Angle, but that shouldn't come into play since both surfaces (the bottom and the outside of the flap) have prism layers.

In my experience, meshing and FSI simulation stability tend to be improved by using overset meshing instead of mesh morphing. You might want to look into that. Basically, the 'background' mesh is just the open volume (ignore the flap). Then, a second 'overset' mesh that wraps the surface of the flap is set on top of the background. The cells in the overset are allowed to morph, but since that mesh is only a few elements thick, you avoid a lot of the problems of large deflection mesh morphing.

If this sounds like it might work for you, try doing the tutorial on overset meshing then hit me up for some more details. I do this stuff everyday, so I've learned a few tricks.

Good luck!

Cheers CFDPosterBoy, I'll check those two options out!


Bin => The link for the tutorial is already posted in the first sentence of my first post.

http://stevedocs.cd-adapco.com/ViewD...Den%3D.html%23

[Bin]

Hi MTabone,

Previously I do not have the access ID for Steve Portal and I'll need to ask the person-in-charge to get it for me. So I was thinking maybe I can get the file from you straight away. But I get the access ID now, so I just got it via the link you posted. Still, thanks a lot.

Bin