It should be noted that tensile tests and per centages of elongation of the metal are of little use in obtaining the transverse modulus of elasticity. This modulus may vary in apparently identical shafts all the way from 11,500,000 to 12,500,000 so that in order to obtain accurate results the modulus for each shaft must be separately determined.
The Bevis-Gibson flashlight torsion meter will serve to illustrate the general method of determining the power exerted by a propeller shaft. Two blank discs are mounted on the shaft at a convenient distance apart. Each disc is pierced near its periphery by a small radial slot, and these two slots are in the same radial plane when no power is being transmitted and there is no twist on the shaft. Behind one disc is fixed a bright electric lamp masked, but having a slot cut in the mask directly opposite the slot in the disc. At every revolution of the shaft therefore a flash of light is projected along the shaft toward the other disc. Behind this disc is fitted the "torque finder," an in strument fitted with an eyepiece and capable of slight circumferential adjustment. The end of the eyepiece next its disc is masked except for a slot similar and opposite to the slot in the disc. When the four slots are set in line, a flash of light is seen at the eyepiece every revolution, and if the shaft revolves quickly enough the light will appear to be continuous. This effect is apparent at anything over 100 revolutions per minute. At lower speeds the flash is seen to be intermittent, but this in nowise affects the accuracy and reliability of the result. At each end of the shaft section, therefore, there exists what is virtually an in stantaneous shutter fixed directly to the shaft, and the only connecting link between the discs is the beam of light which flashes once in each revolution clear through the two shutters. Sup pose the shaft to be transmitting power. One disc lags behind the other by a definite amount, and although three of the slots are still in line the one in the lagging disc is not in line and therefore effectually breaks the flash and no light is seen at the eyepiece. To pick up the light again the eyepiece of the torque finder must be moved by an amount equal to the cir cumferential displacement of the lagging This is accomplished by manipulating the micrometer of the torque finder on which is a scale and vernier graduated in degrees. While the scale is fixed its vernier moves with the eyepiece, and the graduations are so marked that by the aid of a simple microscope con veniently hinged, differences of 1/100 of a degree can be readily discerned. For shafts of ordinary size the scale is set at 14.325 inches radius from the centre of the shaft, so that the degrees are apart. Since an ordinary shaft twists 1 degree in 10 feet at full power it is quite possible to get the shaft horse power to within 1 per cent of full power; but as it is frequently possible to fit the discs 40 or 50 feet apart even this accuracy may be improved upon, and the power determined to within 54 of 1 per cent of full power.
One type of the torsion dynamometer de scribed by Amster is illustrated in Fig. 9. In this form it is intended to be coupled directly by its flanges between the driving and driven shafts. It is also made, however, to receive and transmit its power by belt, in which case the dynamometer is provided with a belt pulley at each end. The two flanges are coupled to gether by a central shaft G capable of trans mitting all the power without being stressed beyond its elastic limit. The flange at the end H carries a disc M, to the outer edge of which an engraved celluloid scale U is at tached. From the flange at the end F a tube A extends toward the other end and is carried by a ballbearing within the flange at H. This tube carries two discs N and 0, in each of which slits T and P are cut. A lamp is hung be hind the scale and a mirror may be arranged, as shown, to facilitate the reading.
It will be understood that when the shaft twists under the influence of the power trans mitted the scale U moves relatively to the line through the slits PT. With each revolution of the shaft, therefore, a momentary glimpse of the scale division is seen by the eye at Q as the slits PT cross the line of sight. If the speed be sufficiently great and the torque con cast in halves so that it can be bolted together around the shaft; the head section, or so-called stump, is provided with a standing arm and is securely bolted to the shaft and turns with it. The tube section carries a corresponding stand ing arm which is free to rotate relative to the fixed arm within the limits of the angle of twist, the other end of the tube being fixed to the shaft. When the shaft is twisted the sleeve re mains untwisted, so that the twist produces a relative but small movement between the arms of the stump and sleeve. This movement is multiplied by means of gearing at the end of the arms, and converted into longitudinal motion of a light aluminum traveler adjusted to run upon the outside of the sleeve. This is accom plished by a flexible wire rope which passes around a pulley connected to the shaft and secured at both ends to a drum on the multi plying gear; the traveler is also securely fastened to the wire. All the wheels and pulleys are mounted on ball bearings, and are practically frictionless, so that when the shaft is revolving the ordinary vibration insures accuracy of position. The flange of the traveler is thus made to move along the tube by the twist im stant, a continuous impression is received of a given scale division. This reading, in conjunc tion with the speed, gives data from which the power being transmitted is calculated. This type of dynamometer is usually made for speeds of about 4,500 revolutions per minute; it has, however, been used in cases where the speed was as high as 8,500 revolutions.