[yambanshee]

Good day,


I'm hoping someone could shed light onto a mistake I might have made somewhere:


I'm simulating a 1/8th section of an axial flow fan, but due to issues I have simplified the blade down to being a flat plate. The blade & hub is meant to be spinning whilst the rest of the domain is stationary. Due to the similarity, I have tried simplifying my setup down to match the SRFSimpleFoam tutorial as closely as possible.


The axis of rotation is about the y-axis, and the inlet velocity is meant to be purely axial (therefor in the negative y-axis). As such, my Urel at the inlet is uniform (0 -10 0) in the non relative reference frame, where as my innerWall (hub + blade/plate) is (0 0 0) in the rotating reference frame. I am using a cyclicAMI boundary on the two periodic zones, and my outlet is a zeroGradient U and absolute 0 pressure condition.


The setup follows very closely to that of the tutorial case, yet after the first iteration, the flow right after the inlet starts to swirl with the relative frame of reference. This is shown by the attached image of the streamlines for U (not Urel). Similarly, when the Urel streamlines are plotted, they are shown swirling at the inlet, and then being completely axial by the first cell.


I have modified the tutorial case to allow for a entrance region prior to the mixer and did not observe this same behavior. Instead it acts as would be expected.



I don't understand what could be causing this behavior, and any help would be greatly appreciated!



I have also attached my case folder for anyone curious. To save space, I have not included the mesh, so blockMesh needs to be run first.

[yambanshee]

It seems to be something to do with my cyclic boundary conditions. Attached I have a image of a different attempt. I'm using plain old simpleFoam with a inlet flow that has a component in both the axial as well as 'tangential' direction. I have also removed the cylindrical nature of the problem and made it pure cartesian.
It seems that the cyclicAMI boundary is acting more as a slip wall by the inlet. Any idea why, and how to maybe fix it?

[yambanshee]

I feel at this point it's worth making a new post due to how much the problem statement has changed from the title, but I'm going to try anyways, because I'm desperate. I've simplified the case as much as I can think of. It's a single cylindrical block domain with a swirling inlet velocity and cyclicAMI boundaries on the sides. As always, the cyclic boundaries seem to work as walls, and I can't understand why. Attached is the full case and a image of the streamlines.

[yambanshee]

Feeling very silly; but the issue seemed to be a lack of iterations. It seems the cyclic field takes many iterations before being fully cyclic. I suppose it makes sense, as it's reading both as 0 in all velocity directions, and it takes many iterations to fix that