Hi, I tried to simulate the 24" propeller fan which used in the air-cond outdoor unit in order to obtain the fan curve. In my case, the full 3D propeller fan is model with the bell-mouth (casing) together. The upstream of the fan is set as Pressure inlet 0Pa (atmospheric). The downstream is a rectangular air chamber (cross section is about 3 x of the fan flow area). The air chamber is set as wall and the opening of the chamber is set as Pressure Outlet. The pressure outlet is set as different pressure and the corresponding mass flow rate of the fan is obtained and the fan curve (Pressure vs massflow)is plotted The rotating fan zone is set as moving reference frame. Compared with my test result, for the same flow rate my simulation result always show lower static pressure, especially for higher flow rate region. Percentage different can go up to 80%. I have tried to use different turbulence model (ke-standard, ke-RNG, kw, RSM etc) and schemes (1st order, 2nd order etc) but the differences are very small. I am wondering where is going wrong with my model and hopefully can obtain some help here. My model is meshed with tetra unstructured mesh with reasonable fine mesh. I have tried to include boundary element mesh by using ICEM, but still not give very good result. Anybody have validated propeller fan model? can I try to remodel it again to see will I can the same result, hopefully from there I know where have I make mistake.

Thanks in advance

Have you tried doing a mesh independant study? also what about sliding mesh method...that might be useful as well to compare with the MRF.

Hope that helps.

Hi zongtwi,

Thanks for your reply. I didn't actually conduct a proper mesh independent study. but I have tried a coarse mesh and fine mesh. The fine mesh is about 1/2 of the coarse mesh size. It seems that the coarse mesh, have nearer result to the test, but for fine mesh, the air flow rate simulated is even lower and further away from the test. Yes I have tried sliding mesh by using the same model. I just changed the rotating boundary to interface and run the sliding mesh. However, the results still not much improved. I have another question, in the fan simulation, should we go for enhance wall function or standard wall function will do? I understand that the enhance wall function require the mesh to be fine enough so that the y+ is around 1. Actually I have tried these two functions before with different mesh size, but still not much different between them. Will the boundary layer elements affect so much the results? Is there any guide on how many layers of boundary layers should I consider?

Thanks again