[piattu]

Hello Community,

I am simulating a rim-driven, ducted marine propeller in OpenFOAM. In order to save some computational costs, I was planning to simulate only 1/4 of the domain (in case of a 4-bladed propeller).
I am using OF v3.0+ as it offers a new boundary condition called cyclicPeriodicAMI.
The problem I am facing is, that the forces on the propeller and nozzle (which I am monitoring to observe convergence) oscillate a lot, instead of reaching steady state values.
I will attach two force-plots showing the results of simulating 1/4 of the domain and also those of a simulation of the whole domain. The plots show 1400 steps which equals 10 rotations of the propeller. One can easily see, that the pattern of the oscillations repeats every 360deg and what is even more confusing to me: the mean value of a 360deg rotation equals 0.25* the value computed by the simulation incorporating the whole domain. - could there be simply a bug in the calculation of the forces?
The case is rather big, but I will append any files you might think relevant to the described problem. I will start by uploading the boundary, fvSchemes and fvSolutions...

Hope someone has got any idea, as I am struggling with this for quite some time already.

Thanks for your help, Heinrich

[piattu]

Hi all,
I solved the above problem and figured I quickly answer myself here, just in case somebody else might be interested...
First thing I did, was to split one of my cyclicAMI patches into 2 parts. One of the parts was a flat surface and I could therefore (after adjusting the mesh) use the cyclic boundary condition instead of cyclicAMI.
This cyclic patch I then choose as periodicPatch for the cyclicPeriodicAMI boundaries.
Finally I specified "transform rotational;" as well as the rotation axis and centre for the cyclicPeriodicAMI.

It seems to me, OF just calculated the wrong rotational transformation in the cyclicPeriodicAMI routine if the rotation axis and centre are not specified . I am not sure, weather the step with the additional cyclic boundary was necessary. However, as the tutorial (annularCombustorTurbine) that comes with this new BC uses cyclic instead of cyclicAMI boundaries as periodic patches I figured, this might be worth a try. As this also saves me some CPU time, (where there is no AMI, no interpolation needs to be done) I did not bother to change it back...

best wishes, Heinrich