reflection boundary condition?
 

Hi ,

I wanted to program a simple Euler equation solution for and airfoil using MacCormack scheme, for a supersonic case using finite difference method. I try to apply reflection at the wall (symmetry) without any results, using ghost points. I wanted to ask if it should be applied before the predictor or before the corrector step or both. And if yes, how. I simply could not figure out why the normal component of the velocity equals to zero when applying for 1 iteration.

I would appreciate any suggestion.

Re: reflection boundary condition?
 

Since the intermediate step flow variable Q* has no apropriate physcial BC, but it must be consistent with original physical quantity. So I suggest in both steps of Macormack scheme, use reflection BC.

Re: reflection boundary condition?
 

Yep, your non-reflecting boudnary conditions have to follow the same loops as your 'internal' solution. If you would do it time accurate (say 2nd order) you would have to do the same time integration for your boundary conditions. From personel experience I can say that this really helps convergence. No more bouncing pressure wave going up and down the domain. To bad not more commercial codes have this feature. Did anybody do this in an implicit code?

Re: reflection boundary condition?
 

Oeps, forgot: If you have a fully supersonic case (in/outlet)then it shouldn't help you much as pressure waves do not travel upstream from your outlet and to your inlet. But do correct me if I'm wrong here.

Re: reflection boundary condition?
 

Guys,

Thank You very much for all advices. However my confusion is the following. I am looking at the predictor and I do not understand why should the velocity component normal to the wall go to 0.0 if I set appropriate values at ghost points for a certain boundary.

Or should IC be set for the particular boundary appropriately in order to get reflection BC working?

Thanks.

Re: reflection boundary condition?
 

What woudl be the difference and significance of doing this in an implicit code rather than an explicit code?

Thank you.