[Jan Smedseng]

Hallo.

Actually, I'm designing a 3D CFD code. Now, I got problems on my inlet and outlet boundary condition. The problem is, that so many things have to work correctly so you cannot really debug.

What do I have to specify on a inlet and outlet boundary condition for incompressible flow using velocity specified at the inlet and pressure at the outlet.

Until now, I have specified:

Inlet - Momentum:
Massflow for convective transport

Outlet - Momentum:
Nothing

Inlet Pressure:
Uncorrected Massflow

Outlet Pressure:
Uncorrected Massflow
aP_p += (Density * (Df.Af) * Af)/(af.PN)

aP_P is the pivot matrix coefficient
Df is the following 3x3 matrix: Df = [Df_u, 0, 0; 0, Df_v, 0; 0, 0, Df_w]
Df_u is V/aP_u;
Af is the area normal vector
PN is the vector from the current cell center to the cell center of the neighbour cell.

The massflow at the outlet is scaled so it is equal to the inlet massflow.
OutletMassFlow = -InletMassFlow

I know, that I'll have some diffusive terms in the momentum equation, but first it also have to work without too.

Does anybody has knowledge about this boundary conditions or a really good paper/reading?

Thanks in advance,
Jan

[Jan Smedseng]

Hallo.

I just uploaded a video, showing the problem, I described.

Maybe somebody has an idea, what could be wrong. The velocity at the outlet changes frequently.

https://youtu.be/9RMRI3KNMrU

[FMDenaro]

The video does not help me so much in understanding what happens...

As you are testing (and debugging) a CFD code, I suggest starting with a straight pipe with the same inlet and outlet BCs in steady laminar condition to check your solution. Use Re=1. The pressure gradient along the axis must result constant and a quadratic streamwise velocity profile must be computed after a sufficient lenght. So, ensure a sufficient length of the pipe.

[Jan Smedseng]

Thank you, I'll try that.

Do I have to specify the velocity or rather the massflow in the boundary conditions of the momentum equation for the convective term in the incompressible case?

[FMDenaro]

Originally Posted by Jan Smedseng

Thank you, I'll try that.

Do I have to specify the velocity or rather the massflow in the boundary conditions of the momentum equation for the convective term in the incompressible case?

try to specify the plug velocity profile as inlet condition, it must develop a boundary layer until to reach the quadratic velocity profile

[Jan Smedseng]

I already did that. But what about the convective flux at the outlet condition? I have three different books and two different answers at this question, thats my problem.

[FMDenaro]

Originally Posted by Jan Smedseng

I already did that. But what about the convective flux at the outlet condition? I have three different books and two different answers at this question, thats my problem.

for laminar fully developed conditions, at the outlet you can specify vanishing normal derivatives for the velocity

[Jan Smedseng]

That means, nothing to define at the outlet condition.

I uploaded a video of a simple, rectangular cannel:
https://youtu.be/gTIvibW0Nf8

The channel is quite short, but my ansys licence just ran out, so I have to renew it to generate new meshes.

But this example shows, that there has to be something wrong with the inlet condition. The high velocity at the symmetry planes is not a display bug as I thought.
In the end of the video, I'm showing the pressure. Looks strange. But the poisson-equation it self works. I tested it without any sources with some dirichlet and neumann boundary conditions and it did, what is should do. The gradient calculation is also correct. I use both (poisson equation and gradient calculation) for the calculation of the wall distance and that works correctly.

Thank you, that you take your time for that.

[FMDenaro]

for incompressible flows, either velocity or pressure define a mathematically well posed problem.
Fixing the velocity inlet and a free stream condition along with Neumann condition for the pressure equation must work...

check if you effectively satisfy the divergence-free constrain

[FMDenaro]

PS: please provide simple curves of the variables, your videos do not help so much