Healing_cad_model

Dear Coreform community,
I hope everyone is doing well. I have a curved model as shown below. It consists of 4 volumes. The model has been generated on Solidworks by means of bending operation on the initially straight version of this model. However, due to bending, the volumes are having gaps/overlaps due to which meshing is difficult.

I can redo the CAD model by other methods, however, I wish to ask if it is possible to mend this model in Cubit itself & thereon obtain a uniform HEX mesh ?

I would highly appreciate if it were possible to do so! This image won’t be so helpful therefore I’m also sharing the STEP file for simplicity.

curved_version.STEP (1.5 MB)

With profound regards :relaxed:
Aakash

I use these command to check the cad quality but I believe autoheal can also fix some cad issues.

cubit.cmd("validate vol all")
cubit.cmd("autoheal analyze vol all")

There is also a heal option on the GUI when importing that might help

Hey,
Yes, I had tried those things. Unfortunately, something else needs to be done to get it sorted.
Nevertheless, thanks for the suggestion.

Hi Aakash,

First locate the area of the overlap.

  1. Click on the Power Tools tab and select volume all
  2. Use an automatically calculated small size based on the model dimensions
  3. Analyze the model for problems
  4. Right click on the Vol1-Vol3 overlap and and select Draw Overlaps.

There are several ways to try to correct this overlap. I used Tweak Replace with Surface (you have to scroll the combo box to find this option on the panel).
image

tweak surface 18 replace with surface 1

I then used a tolerant imprint to get the imprint and merge correct. See Geometry/Volume/Imprint and Merge/Tolerant Imprint.

set merge tolerance 5e-3
imprint tolerant volume all merge

After meshing the model and and using Mesh/Node/Quality, I find no coincident node pairs.

Hi Mr. Merkley,
I will follow the steps to achieve the same. It is great to know that Coreform has powerful tools to could overcome such issues :relaxed:

Thank you with warm regards,
Aakash

Hi Mr. Karl,
Can I ask how did you obtain such an amazing radial mesh especially for that green volume ? Could you please share the commands if you still have ?

With regards,
Aakash

I you look closely, you will notice that the mesh is not perfectly radial. There are two layers of mesh in the middle and one layer on the outer edge.

I just let Cubit mesh this with an automatic size and didn’t specify anything.

If you want to enforce a radial mesh on these surfaces set the meshing scheme on the endcaps to map.

Hi Mr. Karl,
Well, in the image below you may find the best attempt I could do to obtain a radial mesh for the four volumes using the HOLE scheme. Except for the pink volume (encircled) all other volumes came out perfectly fine. Its a triple layered stenotic artery model. Due to large deformations the non radial mesh gets quite distorted during the analysis.

Could there be a better strategy to achieve what I desire ? You may please find the STEP file in the attachment.

Stenotic_Artery.step (299.0 KB)

With regards,
Aakash

Hi Mr. Karl,
Is it possible for you to upload this healed CAD model over here, in case you still have it. I wish to compare it with mine.

Thank you very much.

Best regards,
Aakash

I will upload tomorrow. As far as getting the radial elements, I would do a webcut through the radial direction in the middle of the yellow volume. That additional geometry will create a boundary to help guide the mesh.

Karl

Perfect, thank you!

Aakash

If you still have you curved_version_STEP (or download it from this thread), try these commands.

reset
import step "C:/Users/karl/Downloads/curved_version.STEP" heal
tweak surface 18 replace with surface 1
webcut volume all with plane normal to curve 25 fraction .5 from start
set merge tolerance 5e-3
imprint tolerant volume all merge
mesh Volume 3 7
volume all except 3 7 scheme submap
mesh volume all

Thank you so much for the instructions. I will compare & give the feedback!

Regards,
Aakash

Hi Mr. Karl,
Thank you very much for the description. This methodology works. However, at the same time I believe that a nicely created CAD model takes us a step closer to its successful discretization.

Nevertheless, the power tools of Coreform to address such situations are remarkable!

Thanks with regards,
Aakash