Viscous forces (interDyMFoam) differ between OF 2.0 and later versions
I am simulating boats (Fr=0.3-0.5, Re ~ e6) which are free to heave and pitch using interDyMFoam of OF-2.0.1 wit k-w-SST. Results (resistance and squat) agreed very well with experiments and the viscous force component in longitudinal direction correlated well with the friction line of Grigson. yPlus has been around 180. So far so good.
Unfortunately, if I am using a later version of OF, such as 2.1 or 2.2 viscous stress is roughly 10% less compared to the 2.0 runs, pressure force remains the same. This is weird, as I am using exactly the same case, with exactly the same coeffs for the SST model, just OF version are different.
However, using 2D flat plate simulations (wallFunctions w/ simpleFoam or interFoam) at different y+ (10-200) the result seem to correlate with the too low viscous force, no matter which OF version (2.0, 2.1 or 2.2) is used. Even if the same normal resolution which is used on the boat is utilised on the flat 2D plate, the viscous force does not correlate with established friction lines (such as from Grigson).
It seems that the "nice results" of OF 2.0 are not correct, so I am wondering which changes in OF are leading to the discrepancies? SST model has slightly changed though, but only that a roughness term has been added to later versions, which have not been activated (b1=1, F3=false).
I tried different version of OF on different computers and the differences in viscous force when using interDyMFoam can definitely be blamed on the version of OF.
Q1: Is OF really underestimating the viscous force?
Q2: How comes OF gives different results in viscous force for interDyMFoam if using OF 2.0 or a later version?
Q3: Should the viscous force (from force.dat) at flat plate not agree with traditional friction lines?
Please, let me know if you have any ideas or advice how to deal with this issue. Any feedback is much appreciated. Thanks a lot!
Ok, I had a closer look into my simulations. Attached youll find two pics. One showing total, pressure and viscous force for a run under OF2.0 which has been continued w/ OF2.2 at 4.5 seconds. The drop in shear stress is clearly visible.
The other shows wall shear stress on the hull of the SAME velocity field (at 4.5 sec) using the wallShearStress tool of OF2.0 and OF2.2. The upper one has been calculated using OF2.0 and shows higher stresses. Intergrating wallShearStress in paraview of the two different wallShearStress fields leads exactly to that ~12% difference in shear.
This leads to the conclusion that the computation of wall shear stress has changed which leads to significant differences in the result of viscous force. Maybe I am not the first who discovered it, but I am happy to share this finding :-)
Since I am rather a CFD user than a C++ coder, I could not find the difference in the code and cannot establish which formulation is the correct one. The older one definitely gives better results.
Please, feel free to comment or advise on this post. Any help to receive a closer insight is appreciated.
It looks like this may be the solution to this problem:
The only thing I am wondering about: the bug was closed mid March 2011 and OF 2.0.0 released mid June 2011 and 2.0.1 im August same year. Shouldnt be the correct wallShearStress in OF 2.0+? Or is this a different issue?
The difference in viscous force occurs at different solvers (e.g. inter(DyM)Foam, simpleFoam) but on curved bodies only. For flat plates there is no different between the OF version (2.0-2.2).
Please, correct if I am wrong.
|All times are GMT -4. The time now is 20:35.|