search

cu-numpde / 2021-community-clmurphey

2021-community-clmurphey created by GitHub Classroom
1 stars 0 forks source link

proposed contribution #2

Open clmurphey opened 2 years ago

clmurphey commented 2 years ago

I completely misread the README.md and thought we were just supposed to put an issue in the community to which we are contributing. My apologies for the delay.

Comment on the SU2 Discussion board

I intend to contribute to SU2's VandV repository. Current examples are both out of date (last updated two years ago) and focus mostly on spatial convergence. My plan is to extend these examples to include a temporal convergence study using SU2's various built-in time integration methods. I may potentially extend this contribution to look at other time-integration methods and/or use PETSc to handle the time integration.

With Jordan Perr-Sauer, I may add an additional example related to wind farms. The idea is to include a benchmark problem executed with NALU with the same problem executed with SU2. Ideally, we will include a convergence study with Jordan focusing on spatial convergence and my contribution emphasizing temporal convergence methods.

I am still awaiting feedback on my contribution proposal.

clmurphey commented 2 years ago

First portion of my contribution here. This portion is merely updating the configuration files and python wrappers within SU2's VandV repository.

clmurphey commented 2 years ago

The second part of my contribution is currently under this branch. I need to make a few updates before submitting a pull request to SU2.

clmurphey commented 2 years ago

I changed directions a bit but I am much happier with the result.

I created a repository here with two examples including time convergence studies. Ideally, I would include more examples but I have been profoundly limited by the outputs from SU2. Some example problems just won't output the parameters I would like to look for convergence. I had some strange issues with boundary conditions, which may be due to my own ignorance or may be due to the modular nature of SU2 (some numerical methods and boundary conditions just don't work with certain "solvers" yet.)

I posted a link to this repository (with a brief explanation) on my original discussion post here.