Tricks in simulating Rotor37 with RANS
 

Many people have studied NASA rotor 37 to validate their codes. I have also done that for months. I have already noticed that one can hardly achieve good agreement with exp data especially for the adiabatic efficiency comparison. It's very common to result in lower efficiency than exp.
I have tested many turbulence models and even refined the grid as many as 2mil nodes, but the efficiency is still relatively low.
Does anybody have experience to solve the problem?
I think it is very wonderful to share our experiences, no matter how happy or how miserable they are, to help other people to accelerate their study progress.