[ndrich]

Hi all,

I'm using Ansys Fluent to model oscillatory flow within a cylindrical pipe. I'm driving the oscillatory flow in the pipe using mass flow inlet and pressure outlet boundary conditions. The profile for both the mass flow inlet BC and pressure outlet BC are represented as sine waves of a given amplitude and phase.

The simulation starts up without issue (converge to <1E-6 for all metrics in between 20 and 50 iterations) and runs until the pressure wave is near its minimum value. When the pressure wave approaches its minimum value threshold, omega residual suddenly stops decreasing - no oscillations at all, just steadily approaches an asymptote and does not improve further. After about 30 time steps worth of this behavior, the sim diverges.

Here are my observations from troubleshooting.

1. It doesn't seem to matter where I start the simulation from, it always diverges near the minimum value of the pressure wave. I've tried running the sim with different boundary conditions (pressure inlet, pressure outlet) and switching the inlet back to mass flow, I've tried starting the sim from different points along the sine wave, and I've tried initializing the model to different ICs.
2. When walking the model up to the point where the strange omega residual behavior begins, I see a spike in turbulent kinetic energy (TKE) concentrated at the inlet. This TKE is not distributed evenly across the entire inlet, either - it's present in the core flow, and zero near the wall. This zero TKE region extends about 3 or 4 times further into the core flow than the thickness of the boundary layer, so I don't believe it to be a boundary layer thing.
3. When I let the model continue stepping, this region's TKE grows in magnitude, pressure decreases locally, and velocity magnitude increases in the exiting-from-domain direction. It's almost as though something is causing this region to 'suck' mass flow from the interior domain faster and faster to satisfy the mass flow boundary condition.

Other possibly relevant info: my mesh is pretty uniform, IE fully quad and highly orthogonal. I'm not using any inflation layers and instead am using biasing to achieve greater mesh density near the wall. I'm running the RSM-kw turbulence model.

Does anyone have any insight into what may be going on here? Happy to provide any additional info needed.

Thanks in advance!

NR