[TheBeefBaron]

Hi all, sorry if this is a dumb question (I'm more on the CAD side than the CFD side usually) but I'm trying to run an analysis on a propeller. Besides the specifics of the problem, it's very similar to the classic OpenFOAM or Simflow propeller tutorials, where all elements are part of a rotating AMI cell zone, but the propeller boundary layer is kept static. I am having to run this with compressibility, so I'm using the rho SIMPLE solver. The problem is that in the results, I'm seeing non-zero normal velocities at the propeller boundary, and the integrated mass flow over all boundaries is also not 0, even though this boundary is set to a "non-slip" velocity condition.

I noticed there are other potential boundary conditions in Simflow, such as rotating wall velocity. Using those settings on the propeller (matching the AMI rotation) reduces the mass flow "leakage" around the prop to almost 0, but the net integrated velocity is still non-zero. Unfortunately it also looks visually the same if I intentionally mess up the rotating wall velocity magntitude, which isn't very comforting. Trying out the "Fixed-Velocity" boundary condition on the prop and setting it to 0 results in a similar result to the "non-slip" condition.

I'm curious if this could be some sort of Paraview interpolation/mapping/rounding issue? The net mass input to the input boundary is ~240 kg/s, and the leakage is ~.24 kg/s. Unfortunately due to the problem I'm trying to solve, that's a small enough amount to make a noticable difference in my results, so I need to track down the source of the problem somehow. I tested my calculator method at the interfaces and got the expected results, so I'm fairly sure I'm doing that right.

I ended up going back to the Simflow incompressible propeller tutorial and sure enough, it has a non-zero mass flow as well. It similarly shows non-zero normal velocity at the propeller interface, and a non-zero integrated normal velocity accross all boundaries. Again it's small relative to the net input (~.05 leakage relative to ~ .5 input), but 10X the % error of my problem, so definitely big enough to make me uncomfortable with the results. The attached image shows the normal velocity on the Simflow tutorial propeller boundary.

What could I be doing wrong here? Or maybe the tutorial is wrong but in a way no one generally cares about?

https://i.imgur.com/p1jIpqe.png

[TheBeefBaron]

I have an update here after investigating a bit more. If I ignore the propeller boundaries, I get a big reduction in the mass flow leakage, and surprisingly close (<.1%) results with small changes to the boundar conditions assumptions. I think that the cell-zone rotation might be causing an artificial normal velocity at that boundary interface, causing the problem. I still get some amount of non-zero integrated mass-flow, but maybe this is at the other boundries close to the rotating propeller interface? Not sure what else to try to make this problem go away completely.