This paper has experimentally studied the windage torque in a rotor-stator system with superimposed central inflow and rotor-mounted protrusions. A novel measurement method has been proposed, and the basic principle is to transform the torque of rotating components into static torque measured by using a static torquemeter. Compared with the previous research, the difference of the moment coefficient for the free plain disk in this paper is within 10%. The disk models used in the experiments included a plain disk and a rotor with 18 protrusions. Plain-disk results were obtained with axial clearances varying from 4.5 mm to 40.5 mm and a stator of the same diameter. Two test cases were performed: one was the case where the flow structure was dominated by the superimposed flow and the other was where rotation dominated the flow structure. For the plaindisk case, as turbulence parameter increases, the sensitivity of the torque to variations of G value also increases, leaving the moment coefficient as a function of the rotational and throughflow Reynolds number only. Comparing to the flow parameter, gap ratio and shroud-clearance ratio have weaker influence on frictional moment coefficient. The rotor with protrusions results showed that the gap ratio had negligible effect on the moment coefficient for the former case; however, the torque decreased by approximately 20% with the decrease of the gap ratio for the latter case. It was also found that, for different configurations, the deviation in the moment coefficient was attributed to variations in form drag. In addition, the moment coefficient was affected by the orientation of bolts with respect to the direction of rotation. The empirical correlations have been proposed for the windage losses of various bolt configurations, and a further discussion about minimizing the windage losses was conducted.