Connect inflation layers around a corner between two split bodies

Hi, 

Thank you very much for your question. 

Unfortunately, it is a bit hard for me to understand the block topology in this case. In other words, I am not understanding where each block ends. Is there any chance you can post some additional images to understand this; or perhaps a very simple Fidelity Pointwise file showing this particular topology? 

In addition, it seems that you are using a regular extrusion to create the boundary layers. Have you tried using the T-Rex algorithm for this? 

Thanks much!

Hi,

of course. I am using a CAD model and create the blocks from its topology.

Here you can see the white part. This should be the block with inflation layers defined as a porous volume.

The other part above is a different block and both inflation layers should be connected.

The connected inlation layers should be around the edges like the red lines you can see here:

I haven't tried T-Rex for this yet, I only used the normal extrusion. Do you think this would work better?

Thanks for your help!

Hi, 

Thank you very much for your quick reply. 

It is good to hear that you are familiar with the recommended workflow.

You will likely generate the smoothest (and faster) grid with the use of ASM to create the surface grid and AVM to create the two blocks. Inside AVM, you will be able to set up the T-Rex algorithm for each block with the corresponding boundary conditions.

Thank you very much for your time.

Hi,

I tried T-Rex several times and got the following problem now:

As you can see, the inflation layers alsways end in the edges between the blocks. (The porous body is between the red lines, the other 2 bodies are fluid)

Is there a specific setting to connect it around the corner and also between blocks in general?

Thank you!

Hi, 

Thank you very much for the additional image. Now I better understand the topology you are working with. 

The T-Rex algorithm on the lower red interface domain stops because this domain hasn't been set to a Match T-Rex boundary condition. In addition, you need the Push Attributes turned on. 

As for the top red interface domain, T-Rex won't be able to make the 90 degree turn with the topology as-is. Even if you set the top red domain to a Match T-Rex BC, both BCs, the Wall and  the Match will be coplanar preventing the result you are looking for. 

One possible option in this case would be to change the topology slightly and move the red block down just enough to allow the T-Rex algorithm to make the desired turn (image below).

Please let me know if this would be a reasonable approach for you., 

Best regards.

Hi,

Yes I will try that.

But in this case, extruding in normal direction should also work, right?

Because it connects smoother than T-Rex cells.

Thank you!

Hi, 

You are very welcome. 

In my experience, T-Rex is usually what will allow for a nice transition in corner cases like this one. As far as I can tell, and considering I don't have the entire geometry, my last image looks pretty similar to your original image depicting your end goal. Based on this, T-Rex with the slight topology change would be my recommendation. 

That being said, you are welcome to give the normal extrusion a try but I suspect you may need to define additional topology in that corner in order to make the turn you are looking for. 

I hope this helps. 

Best regards. 

Hi, 

You are very welcome. 

In my experience, T-Rex is usually what will allow for a nice transition in corner cases like this one. As far as I can tell, and considering I don't have the entire geometry, my last image looks pretty similar to your original image depicting your end goal. Based on this, T-Rex with the slight topology change would be my recommendation. 

That being said, you are welcome to give the normal extrusion a try but I suspect you may need to define additional topology in that corner in order to make the turn you are looking for. 

I hope this helps. 

Best regards. 

Hi,

 the normal extrusion worked pretty well around the corner:

After working with this, I got two more questions.

1. How is it possible to define conformal surface meshes between the two large areas to create a periodic boundary (next picture, 2 red arrows show the 2 domains of the boundary layer at the bottom and the surface mesh above)?

I only found out that I can go by rotating. But this would destroy the block with my boundary layer because the extrude also creates a separate block.

Is there a way to simply select the surfaces and set matching surface meshes as a boundary condition?

2. Do you know how to use hex dominant meshing for volume meshes? I only found the surface mesh option with hexes and tets. I am pretty sure there is also an option to use it in volume meshes somewhere.

Thanks for your help, I really appreciate it!

Hi, 

Thank you very much for the follow up question(s). 

• Perhaps I am not understanding correctly from the images, but it seems that you obtained with the extrusion something similar to what I showed with T-Rex. Based on the images, it seems that T-Rex should provide more flexibility in this case.
• The advantage of using T-Rex in this case is that you can use the Create, Periodic, Extrusion functionality to create the periodic domains, then assemble your block, and then let the solver complete the entire grid (aniso and iso) while maintaining the periodicity (i.e., both domains are the same). If the geometry you are showing is just a "slice" of the entire geometry, you will need that periodicity. In other words, perhaps I am misunderstanding, but it seems that you are doing more work than needed using the normal extrusion in this case.  
• As for your second question: 
  • You can create quad-dominant domains (i.e., surface grids containing triangles and quads)
  • You can create a block (i.e., volume grid) containing prisms and hexes in the anisotropic region (i.e., boundary layer) and tets and pyramids in the isotropic region. 
  • You can also create a block containing prisms and hexes in the anisotropic region and hexes, pyramids and tets in the isotropic region. This are called Voxel meshes. 

If at all possible, please review some of the videos presented in the User Manual (under Grid, Solve) and let us know what would be your end goal. 

Best regards.

Hi,

Thanks for your advice.

I have now tried using your advices regarding T-Rex. It might be better for the periodic boundary.

I still do not get the connection between the blocks properly.

I defined all the walls for the inflation layers and created a Match-boundary for the domains separating my blocks.

Then the following appears:

Push attributes is turned on and the boundary conditions are set like this:

Do you have any suggestions which setting is missing in my case?

Thanks a lot!

Hi, 

Thank you very much for the update on this. 

Unfortunately, it is hard to understand your T-Rex set up and to answer your questions by looking at two images. At this point, I would recommend you log a support ticket in the Cadence Online Support (COS) portal if possible. 

If you are a student, you may not have access to log cases. In this case, I would recommend you work with the person in charge of the license (usually a professor) who should be able to log a case so we can take a good look at the grid for a more accurate resolution. 

Thanks much for your time. 

Best regards.