[EmiS]

Hello,

what kind of boundary condition should I select for outlet of a tubular geometry (3D bent pipe) when the inlet BC is given as parabolic velocity profile that oscilates in time (i.e. velocity vector changes orientation and during a part of each period, the fluid is coming out of the inlet)?

I tried to apply zero pressure at the outlet, 'outflow' and 'outlet vent'. All of them give me similar velocity field (which is not physical): there is something like a parabolic profile at the inlet, but the rest of the flow field is more or less random noise with velocity values close to 0.

This is for incompressible NS, second order upwind in space and second order time discretization. I implemented the inlet BC through a UDF. The same test case in steady settings (parabolic velocity profile at inlet and zero pressure at the outlet) works fine for me. Thank you for any hints regarding what could possibly be wrong with the test case.

Emi

[EmiS]

Dear Fluent users,

I'll try to clarify my point above.
The geometry considered is a 3D bent pipe with radius 1m (all length units are in m) - I know it's big, but it is just a test in which I'm trying to impose an oscillating parabolic velocity profile of max velocity Umax = 10 m/s.
The fluid flowing inside this pipe is water (rho = 1000 kg/m^3 and dyn. visc. = 8e-4 kg/m.s ) - Reynolds is 25000000.
I first tried to run the steady case - parabolic velocity profile at the inlet (given by a UDF) and pressure outlet set to 0 (there is also a non-slip wall and a symmetry plane). The solver used is SIMPLE with steady and laminar conditions (laminar is not the best in regards with the Re, but I'll change it later).
As observed in the screenshots below, the flow establishes slowly through the pipe over the iterations.

When I try to run under unsteady conditions (my parabolic profile is multiplied by sin(1.0*t) in the UDF expression) I need to lower the time step to 0.01s - which takes a certain amount of computational time to perform one period.
Is there a way to "speed-up" the simulation by changing the simulation settings? Should I change the solver specs for example?

The other concern is in regards to what is occurring at the outlet. When sin(1.0*t) is comprised between Pi and 2Pi, I would expect the outlet to "behave" like an inlet - but no clear velocity inflow seems to start from the outlet. The outlet BC is set to "pressure-outlet" condition (with 0 gauge pressure) and I also tried to impose a "target mass flow rate" via a UDF as suggested in another CFD-online discussion. The mass flow rate is given by another UDF function, which is set to rho*0.5*Umax*sin(1.0t) Unfortunately the issue at the outlet does not change. Would anyone have an idea where could be the problem in this simulation? Has anyone experience with a similar setup?

Thank you very much in advance for your advices.

EmiS