Conversation
pcarruscag
left a comment
There was a problem hiding this comment.
Thanks for that @WallyMaier.
The sign change is consistent also with the book by J. Blazek.
Taking away the 0.5 from the V squared also makes sense given how the speed of sound is being calculated, but I think it is needed for the Jacobian part, lines 756 and 763.
Also I noticed that in this book the inviscid flux is defined as the average of the left and right fluxes, while in SU2 it is currently the flux of the averaged quantities (like in central schemes) can you maybe have a look?
|
@pcarruscag I updated the jacobian. Regarding the flux, I am a bit confused as well. The paper referenced above says flux implemented is a higher-order scheme that employs the SLIP scheme's limiter. That being said, I do not have access to Blazek's book at the moment. |
|
I followed up on two of the references the book cites: |
pcarruscag
left a comment
There was a problem hiding this comment.
LGTM
@WallyMaier, I can take care of making this consistent with the literature as part of #691
Proposed Changes
This addressed problems with flux computation in the CUSP scheme, mentioned in issue #465.
The implementation was checked in reference to "Comparison of Several Dissipation Algorithms for Central Difference Schemes" Swanson.