Provides a system for patient-specific cardiovascular modeling and simulation.
-
yan qing
- Posts: 32
- Joined: Tue Dec 06, 2016 4:56 am
Post
by yan qing » Thu Jul 01, 2021 2:16 am
davep wrote: ↑Fri Mar 12, 2021 3:19 pm
Hi Jimmy,
I had a look at the simulation and reproduced the crash.
Your mesh has elements that are almost flat (volume 1e-8) that are causing numerical problems. i think these flat elements are created when doing boundary layer meshing. Try creating a mesh without a boundary layer and see if that runs OK. If you really want to include a boundary layer then maybe use fewer layers or adjust layer spacing.
SV does not yet have any functionality to check mesh quality. ParaView has a
Mesh Quality filter that you could use to identify bad elements. The image I attached was create using a program I wrote using VTK (see
https://github.com/ktbolt/vtk/tree/master/mesh-quality).
Cheers,
Dave
Hi Dave,
I have the same question as Jimmy, and I use the program you wrote using VTK, then my mesh also has elements that are almost flat
(volume 1e-6).
![4.png](./download/file.php?id=5110&sid=b7641f7cec5f1c84af19a7e0da7e78e1)
- 4.png (13.78 KiB) Viewed 177 times
How can I find the elements and what should I do to fix this problem?
Cheers,
Yan
-
David Parker
- Posts: 1677
- Joined: Tue Aug 23, 2005 2:43 pm
Post
by David Parker » Thu Jul 01, 2021 10:55 am
Hi Yan,
Bad elements are typically produced from a model that has features that can't be properly meshed. How are you creating your model? Make sure all vessel junctions are well-defined, want to have the ends of branching vessels completely inside the parent vessel.
If you can upload your model (the .vtp file in your project's Models directory) I can have look.
Cheers,
Dave
-
yan qing
- Posts: 32
- Joined: Tue Dec 06, 2016 4:56 am
Post
by yan qing » Thu Jul 01, 2021 6:57 pm
Hi Dave,
I import the model from the OSMSC 0186_0001.
The model is small than its actually size, so I have scaled it. Here is the data after scaled.
https://github.com/YQ161916/mydata
Thanks,
Yan
-
David Parker
- Posts: 1677
- Joined: Tue Aug 23, 2005 2:43 pm
Post
by David Parker » Fri Jul 02, 2021 10:54 am
Hi Yan,
Element volumes around 1e-6 or 1e-7 are not ideal but will probably not cause the solver to fail. The Coronary Normal clinical case has minimum element volume of 1e-7 and it runs fine.
If your simulation is failing then there might be something wrong with the BCs. If you scale your model then you need to make sure the units you use in the simulation are consistent with the length scale of your model.
Cheers,
Dave
-
yan qing
- Posts: 32
- Joined: Tue Dec 06, 2016 4:56 am
Post
by yan qing » Sat Jul 03, 2021 7:51 pm
Hi Dave,
I will check the BCS and the units of the simulation.
Thank you very much.
Yan