[aerosayan]

I have a good enough 2D structured solver. However, there aren't much OpenSource tools that allow easy grid generation like in ICEM-CFD or Pointwise.

GMSH is amazing, however we need to write a fairly complex input file for any non-standard case.

Looking at some images of CFD domains attached, it dawned on me that it is possible to create a semi-automatic 2D structured grid generator. Of course the grid generated will be in multiple blocks. I have a rudimentary AMR system in development, that will adaptively refine the grid blocks in levels. The AMR system refinement would look like the first picture.

I want to hear some counter points before I dive into this.

Why do you think it might be difficult or even not possible?

[arjun]

Give me your email, I will send you something in the direction you are looking in.

I have also written (not completely) something something simple to generate simple 2d meshes for wildkatze.

Since it was not completely finished it was not formally released.

[aerosayan]

Originally Posted by arjun

Give me your email, I will send you something in the direction you are looking in.

I have also written (not completely) something something simple to generate simple 2d meshes for wildkatze.

Since it was not completely finished it was not formally released.

Really appreciate it. I sent you my email in private message. Thanks!!


My current idea is from the time I was making an advancing front triangulation algorithm for [redacted], but realized that handling multiple bodies is insanely difficult, because of the presence of multiple holes in the domain. Then I realized, I could partition the domain in such a way, there would be no holes, and the algorithm would work well.


The results worked out well.

[arjun]

Originally Posted by aerosayan

Really appreciate it. I sent you my email in private message. Thanks!!


My current idea is from the time I was making an advancing front triangulation algorithm for [redacted], but realized that handling multiple bodies is insanely difficult, because of the presence of multiple holes in the domain. Then I realized, I could partition the domain in such a way, there would be no holes, and the algorithm would work well.


The results worked out well.

https://www.cs.cmu.edu/~quake/triangle.html