Final analysis was conducted after the experimental agenda was 

 repeated for each of the three flexures representing the six components 

 of blade load. Corrections for interactions between the various load 

 components were performed for a representative condition in calm water 

 without hull pitching, as outlined by Boswell et al. (1976a, 1976b, 

 1978) . The resulting load corrections were applied to all other condi- 

 tions in the experimental agenda. The total loading components were 

 corrected for the centrifugal and gravitational loads to obtain the 

 hydrodynamic loads. Corrections were also made to the mean loads to 

 account for the influence of the d3mamometer boat. 



E. Accuracy 



The accuracy of the experiment was generally similar to that de- 

 scribed by Boswell et al. (1976a, 1976b, 1978). During the experiments, 

 the on-line analysis averaged data over many revolutions and computed 

 standard deviations of speed V, rotation speed n, forces, and moments, 

 assuming a normal distribution in the variation of these quantities. 

 From this, a variation in the measured quantities was calculated with a 

 95 percent confidence level. Model speed V, and rotation speed n varied 

 by +0.5 percent from calculated mean values. For the condition in calm 

 water with no hull pitching, the force and moment signals at each angular 

 position measured, varied by +2 to +10 percent of the calculated average 

 value. Figure 5 shows the measured variation in the raw F^ signal. 

 Note that the variation in force at each angular position was greatest 

 when the blade was nearest the model hull. The variation of the loading 

 components during the pitching and wave conditions was +10 to +20 per- 

 cent of the mean values at each angular position. These variations were 

 greater than the still water condition because each run was evaluated 

 over a certain tolerance range in pitch or wave height. It should be 

 noted that the variations from the mean represent the band in which 95 

 percent of the measured data lie. The accuracy of the mean values cal- 

 culated will be higher than the variations calculated. 



Besides the fluctuation in signals occurring in a given run, the 

 overall accuracy of the data can be represented by the repeatability 

 between different runs. An effort was made to set experimental condi- 

 tions identically on repeat runs; however, the propeller rotational 

 speed and model velocity were set by hand, so some variation was unavoid- 

 able. The variation in the measured experimental conditions and the 

 blade loading data for repeat runs is similar to that documented by 

 Boswell et al. (1978) and Jessup et al. (1977) showing that the varia- 

 tions in the mean forces and moments were +4 percent over all the runs. 



As discussed in the section on data acquisition and analysis, for 

 operation with periodic pitching either with or without waves, the data 

 were sorted and analyzed based on instantaneous position in the pitch 

 cycle, and for operation in waves without hull pitching, the data were 

 sorted and analyzed based on instantaneous position of the propeller in 

 the wave cycle. For periodic pitching runs, selection of a propeller 

 revolution at a specified pitch angle ^ in the pitch cycle necessitated 

 a tolerance of 0.05 degree to ij;; however, the average value of i^ for 



