Davis and English 



The tests on the simple blade shapes at De Havilland were followed by a 

 similar test on a more realistic blade in which the pitch of the sections was in- 

 corporated. At about the same time, some further tests were being conducted 

 at NPL on hydroelastically sealed model propellers in the water tunnel. Ideally 

 it would have been desirable for the same blade geometry to have been tested in 

 both establishments, but this was not possible due to the tight time schedule. As 

 a consequence, while the two sets of test results can be compared, too close a 

 correlation of the results cannot be expected since the blade shapes and thick- 

 nesses were different. The two sets of tests are described separately there- 

 fore, commencing with the static tests conducted at De Havilland. 



The blade used for the static tests had sections with circular-arc wetted 

 face cambers, as shown in Fig. 23, and an appreciable amount of leading edge 

 sweepback or skew. The test setup was similar to the one used earlier, apart 

 from the extra complication introduced because of the pitch of the blade, and is 

 shown in Fig. 24. The test blade was made from aluminum to limit the external 

 load necessary to produce easily measurable surface strains and deflections. 

 For the purposes of simulating and distributing the external hydrodynamic load 

 this load was split into 28 discrete elements and applied through rubber pads 

 bonded to the back of the blade and loaded in tension. Deflections and surface 

 strains were measured on the blade face. In the absence of any measured blade 

 pressure distributions which could be used as a basis for distributing the exter- 

 nal load, reliance had to be placed on an estimated distribution. From these 

 estimates the radial distribution of lift coefficient was almost inversely propor- 

 tional to radius, while the chordwise loading was taken as similar to that pro- 

 duced by a fully cavitating flat plate hydrofoil at incidence where the pressure 

 is concentrated towards the leading edge and the centre of pressure occurs at 

 the 25 percent chord position. In this respect, fully cavitating propeller theory 

 is not considered sufficiently refined to justify attempts at producing a more 

 realistic load distribution. It is probable, however, that the particular load dis- 

 tribution used produced larger stresses in the critical regions of the leading 

 edge and blade root than would be experienced in an actual propeller, since a 

 substantial part of the total load was situated towards the tip and leading edge 

 regions. 



The blade strains were measured with strain gauge rosettes at sixteen po- 

 sitions, thus enabling principal stresses and directions to be derived. The 

 maximum principal stresses deduced from this work have been scaled to corre- 

 spond to a 44-inch diameter propeller developing 40,000 Ibf thrust. These 

 stresses are shown in Fig. 25, where it will be seen that they always fall well 

 below a value of 40,000 Ibf/in^ which is the stipulated maximum steady stress 

 level allowable for Inconel 718 when fatigue is a factor to be considered. It will 

 be observed from these stresses that despite the incorporation of a considerable 

 amount of sweepback the stress level in the leading edge vicinity at the 70 per- 

 cent radius is higher than anywhere else in the blade. The directions of these 

 principal stresses are given in Fig. 26 and show how significant the chordwise 

 loading can become, since at the 90 percent radius the maximum principal 

 stress direction approaches the chordwise direction. 



From the above discussion it might be expected that the simple Engineer's 

 Beam Theory, (commonly used for estimating stresses in propellers, could not 

 predict reliable stresses throughout the blade. This has been found to be the 



990 



