[ghost82]

Hi CFD forum!
what do you think of boundary condition to be applied on the attached figure?Simmetry or periodic?and why?
Imagine to split in half the attached geometry; the impeller moves up-down and down-up.

Daniele

[flotus1]

The choice of the appropriate boundary condition depends on two things

• tangential velocity components
• turbulence modeling approach

If the tangential velocity is zero, a symmetry boundary condition can be applied.
Otherwise, you need a periodic boundary condition to allow a tangential velocity.

Note that this only applies if you are using a RANS turbulence approach. For LES-like simulations the full geometry has to be simulated.

[ghost82]

Originally Posted by flotus1

The choice of the appropriate boundary condition depends on two things

• tangential velocity components
• turbulence modeling approach

If the tangential velocity is zero, a symmetry boundary condition can be applied.
Otherwise, you need a periodic boundary condition to allow a tangential velocity.

Note that this only applies if you are using a RANS turbulence approach. For LES-like simulations the full geometry has to be simulated.

Then, if I don't know if tangential velocity is zero I can apply periodic boundary: if it will be zero then periodic boundary will give same results as simmetry?

Yes, I'm using RANS.

Thank you

[oj.bulmer]

Try to understand the inherent definition of the two BCs.

1) Periodic: This is used when there are non zero velocity components perpendicular to the plane. In cylindrical case, it will be tangential components in theta direction. It is valid only when vectors leaving on one plane are entering in the same relative angle to the plane at the same position from the other plane. This is possible in cylindrical geometry only when there is swirl with no gradient in theta direction. Otherwise the assumption of periodic BC is not valid.

2) Symmetry: This is used when there are no velocity components perpendicular to the plane, such that the section you are simulating is same to the adjacent sections.

So by that logic, if the tangential velocity is zero and if you use periodic BC, obviously the vectors leaving in one plane and entering the other are zero, which should give you similar results as symmetry.

Please understand that if the tangential velocities are non zero and if they may have a gradient in theta direction, you need to model the full geometry.

OJ