Davis and English 



as that of the prototype, but the model is weakened by cuts to reduce effectively 

 its elastic modulus. Very complicated model construction can result from using 

 this technique, however, and in the experiments described here it was decided to 

 ignore the condition for the equivalence of stresses and deflections induced by 

 inertia forces. This neglect is probably not very important, since the tests were 

 conducted in uniform flow in which the hydrodynamic loads were nominally 

 steady. Further, the direct centrifugal stresses are usually small when com- 

 pared with the bending stresses due to the hydrodynamic loading, and the cen- 

 trifugal bending stresses depend mainly on the rake of the blades which in the 

 case of the T95 design was only five degrees aft. Blade skew and the unsym- 

 metrical distribution of material about a radial generator line will also influ- 

 ence the centrifugal stresses, but since there is no skew in T95 the effects due 

 to these features will also be small. 



Two hydroelastic model propellers were made from epoxy resin loaded 

 with fibre glass. These 10-inch diameter screws were initially formed at De 

 Havilland, Canada and then cut to the T95 design at NPL. This particular mate- 

 rial was chosen mainly because of its elastic modulus (1.8 x 10^ Ibf/in^), as 

 this together with the screw scale and tunnel conditions enabled model tests to 

 be conducted at values of n^D^ (l - a^)/E applicable to the prototype screw. 



The first hydroelastic model was run in a similar manner to the bronze 

 screw, but at particular values of rotational speed to obtain the desired values 

 of the parameter n^D^ (i - a^)/E. If it is assumed that the deflection of the 

 bronze screw was negligible, a reasonable assumption according to Ref. 17, then 

 the difference between the results from the hydroelastic screw and the bronze 

 screw are due to the deflection of the former, apart from small differences that 

 arise from experimental scatter in the results. The results obtained from the 

 two screws are compared in Figs. 28, 29, and 30, where it is clear that both the 

 thrust and torque of the hydroelastic screw have increased above the values for 

 the bronze screw. Also, since the torque increase is greater than that of the 

 thrust, the efficiency of the hydroelastic model is less than that of the bronze 

 screw, and further this reduction in efficiency is fairly significant in this case. 

 As an example, in a typical high-speed operating condition the efficiency of the 

 screw with deflected blades is about 95 percent of the undeflected value. Simi- 

 lar results to these were also found in the takeoff conditions. It may be noted 

 however, that because the T95 blades are thinner than those used in the Bras 

 d'Or application the T95 results exaggerate the effects of deflection on these 

 screws. 



The second hydroelastic model of the T95 design was used for measuring 

 blade surface strains under fully cavitating conditions and then deducing stress 

 levels. For this purpose small foil strain gauges were attached to the wetted 

 surface of each blade at a number of positions. The leads for energising the 

 gauges and conducting the output signals passed down the blades and through the 

 hollow shaft to slip-rings outside the tunnel where they were transferred to the 

 external instrumentation. The gauges were attached to the wetted surface in 

 order to ensure temperature stability which might have been difficult with 

 gauges attached to the backs of the blades and within the cavities. Temperature- 

 compensating gauges were attached to a piece of epoxy resin and situated in the 

 reservoir of the working section. The gauges and leads on the blades were 

 coated with a thin covering of epoxy resin for insulation purposes. This coating 



996 



