[crazyshock]

Hi,
I am trying to validate my 2d NS code (WENO5) by simulating flow over a flat plate.
These are my Boundary conditions (i am using non dimensional variables):

Inlet:


Upper Boundary condition:


Lower Boundary Condition (Reflection):




Outlet Boundary Conditions:



I am getting erreneous results using these conditions. Can somebody please verify if these are correct?

I took these boundary conditions from MacCormak's paper (A Numerical Method for Solving the Equations of Compressible Viscous Flow)


Note: For energy expression at boundary, so it reduces to

[FMDenaro]

I don't understand what you are doing... that is a 2D RANS simulation of a developing turbulent BL??

[crazyshock]

No I am solving just the compressible Navier Stokes equations with no turbulence models. Since i am using the conservative form, the variables i am using are rho, u, v and E (total energy). Stating turbulent BL on the topic name might have been a misnomer. This is just a simple DNS solution I am going for.

[FMDenaro]

Originally Posted by crazyshock

No I am solving just the compressible Navier Stokes equations with no turbulence models. Since i am using the conservative form, the variables i am using are rho, u, v and E (total energy). Stating turbulent BL on the topic name might have been a misnomer. This is just a simple DNS solution I am going for.

What you are doing makes no sense. With a 2D NS code, you can only simulate the laminar zone of the developing BL. That means you can solve until Rex<O(10^5). But you need a refined grid such to resolve the thickness delta(x).

[crazyshock]

I see, then i am just solving for the laminar zone with a Re of 100. Please ignore the turbulent part. I just want to validate this thing and get it over with

[FMDenaro]

In solving the low-Mach flow using a compressible flow solver, you must be aware of:
1) the problem is stiff in the limit M->0
2) for subsonic flow you need to prescribe correct inflow/outflow condition according to the characteristic direction. You have to let free one condition in inflow and you have to set one condition in outflow.

Be also careful to the upper conditions, the normal component velocity is zero only asymptotically

[sbaffini]

Originally Posted by FMDenaro

In solving the low-Mach flow using a compressible flow solver, you must be aware of:
1) the problem is stiff in the limit M->0
2) for subsonic flow you need to prescribe correct inflow/outflow condition according to the characteristic direction. You have to let free one condition in inflow and you have to set one condition in outflow.

Be also careful to the upper conditions, the normal component velocity is zero only asymptotically

I can only confirm Filippo's doubts. In particular:

1) You are using fully supersonic bcs at both inlet and outlet.
2) Your domain seems to be square. Give this BL some space

For what concerns the MacCormak's method, you need to verify that it is well suited for your case (I also suspect that for M->0 that might not be the case).

However, your solution doesn't appear to be qualitatively wrong.

[crazyshock]

Thank you so much for the inputs. So the method I am going to go with is to ask my professor to provide the BCs he used in his code validatons

[selig5576]

I am by no means an expert as Dr. Denaro is, however from my *very limited* experience with compressible flows, the AUSM+ scheme is quite nice. It provides accurate numerical solutions in the presence of shocks (when compared to MacCormack). AUSM+ coupled with WENO5 should then provide a very accurate solution.