We had an interesting discussion about the drag coefficients and drag forces; you can find the details here:
But I’d also like to briefly collect my findings maybe it can be useful for some:
The model of the spehere was a single-surface-model so snappyHexMesh was unable to create any boundary layers (as @pfernandez) highlighted. It caused the abnormal drag forces. So the model had to be split!
With the split mesh I got the following results:
Case 1: Re=10^2 (laminar flow) -> v=0.00155m/s
Calculated drag value: 1.14e-6N
Simulation drag value: 1.46e-6N
Percentage error*: 28%
*Please note that by disabling turbulence the error is reduced to 17%! Finer mesh would also lead more precise values.
Case 2: Re=10^6 (turbulent flow) -> v=155m/s
Calculated drag value: 17.05N
Simulation drag value: 41.9N
Percentage error**: 146% -> Actually it is around 5%!
**Please note that the drag coefficient for the drag force calculation was defined for smooth sphere at the actual Reynolds number. Since the mesh is not smooth the assumption was not valid. According to the plot below one can see that the drag coeff. for rough sphere is around 0.35. With this value the drag force is 39.8N hence the error is around 5%.
Concluding I can say that the simulation is a good approximation of the real thing and it could be improved more with finer mesh.
Maybe later I’ll make some convergence runs or if you are interested feel free to make a copy and give it a shot.