Model with L turn

c0384

Hi

I am modelling a rotating blade by using k-epsilon model with SFCD discretization scheme. There is a L turn flow path before the fluid enter the rotating blade. CFD results show that fluid loss all its energy before entering the rotating blade, and this is identified as the reason of why the torque on blade is a lot smaller than experimet value. L turn seems to give a big error to my simulation. I tried to change the discretization scheme to MARS and I get a little of improvement for my torque but still small if compared to experiment value.I suspect this unbelievable results might be due to improper use of CFD model. Can any advise me on what is the appropriate CFD model for my simulation, or any ways I can improve my CFD results?

Thanks in advance.