[shainath]

Hi,
I am implementing new reconstruction method to get second order results. I have just replaced the LS method of reconstruction with the different one. If I run "viscous flow over cylinder" problem (Re=40) then Cp plot over the cylinder wall shows some oscillations but for LS given by SU2 show no oscillations in a Cp plot.
Am I missing something in this re-implementation, like avoiding reconstruction at the boundary? In general no reconstruction is done at the boundary cells/points. Please give some suggestions. Thank you.

[hlk]

Originally Posted by shainath

Hi,
I am implementing new reconstruction method to get second order results. I have just replaced the LS method of reconstruction with the different one. If I run "viscous flow over cylinder" problem (Re=40) then Cp plot over the cylinder wall shows some oscillations but for LS given by SU2 show no oscillations in a Cp plot.
Am I missing something in this re-implementation, like avoiding reconstruction at the boundary? In general no reconstruction is done at the boundary cells/points. Please give some suggestions. Thank you.

Thank you for your question.
One thing you can look into is the use of limiters. These are already used in SU2, so by looking through the code you will be able to find some examples.
I believe for higher order reconstruction you will need more boundary points than currently used by SU2.
If you would like to share your reconstruction method, once you have it working to your satisfaction you can share it on github by forking the SU2 develop branch, adding your modifications, and making a pull request for merging into the develop branch.