friction & windage losses prediction - help needed

alex

hi! i am working on a rotor stator issue. I have to study friction and windage looses of this rotating machinery. I make steady state and transient runs and i have the same issue. I have two domain (1 stationnary 1 rotating domain). I modeled the machine on CFX 10.0, the rotational axis is z.I computed torque_z()@"rotorwall". By multiplying this value by omega (rotational velocity), i find the friction and windage looses for rotating part. Experimental datas show that frictions and windage looses are twice larger than cfx predictions. IT's very strange.

note: if i compute the torque_z()@"statorwall" and multiply this value by the angular velocity of the rotor and if i make the sum with the fricions and windage looses due to the rotating part, then i find a good result. But, i believe that not a good physically approach.

Question 1: how can i take account for the effect of the static parts on the frictions and windage looses?

Question 2: why cfx underpredict the experimental data result by 50 % by just taking account for the torque of the rotating wall?

Please help me. CFX support don't have any clear answer concerning this problem.

Regards.

Alex