THE STRAIN GAUGE: AN AID TO THE DEVELOPMENT OF MARINE

Transcription

1 THE STRAIN GAUGE: AN AID TO THE DEVELOPMENT OF MARINE TRANSPORT by H. B. BOYLE National Physical Laboratory, Hovercraft Unit, Hythe Strain gauges measure directly strains in models and marine components and are parts of torque and thrust transducers and of force and moment balances. Sixcomponent balances of strut and shear-plate design and special sliprings are described. Introduction In 1957, the Ship Division of the National Physical Laboratory (NPL) started to use strain gauges in a force measuring transducer. As an innovation, it was treated by many with suspicion and by some with strong disbelief. Since that time the strain gauge has ousted all other forms of electro-mechanical transducer for measurement of force, and it gradually replaces other systems used in the Division. Force transducers measuring simultaneously up to six components have been produced. Devices measuring force and moment span the ranges from 0 ito 12,000 lbf (05 N to 60 kn) and from 0 1 to 15,000 lbf ft (0 15 Nm to 20 knm). The ranges cover both model and full-scale research on a variety of vessels including conventional ships, high-speed boats, hydrofoil and hovercraft. The advantages of small size, low mass and accuracy, the ability to convert any loaded structure into a force transducer, and the convenience of direct strain measurement made possible experiments that would have been difficult, if not impossible, with other transducer systems. It would be difficult to think of any single more important aid used for research into hydrodynamics in the Division than strain gauges. The Gauges The first gauges used at the Ship Division were manufactured by NPL and were bonded with Bakelite cement, whereas current practice employs epoxybacked self-temperature-compensated foil gauges for nearly all applications. Various glues are used, hot setting adhesives being preferred wherever possible. In 1960 a brief note in an American magazine recorded the development of a strain gauge having a gauge factor of about 120. A set of four gauges were ordered and the first semi-conductor strain gauges arrived in the Division. They were used to measure propeller shaft thrust fluctuations on a large passenger liner travelling across the North Atlantic. This type of gauge is now considered as a useful and normal extension of strain gauge techniques and is normally applied to the measurement of low skin strains (not more than 10 pin/in) in stable temperature environments. Energisation, Amplifiers and Slip Rings The majority of strain gauge systems are energised by a 5 V a.c. carrier system operating in the range from 1 to 3 khz. On occasion, d.c. energisation is used for high-frequency dynamic applications or where supply and signal leads are extremely long. In a carrier system, the a.c. amplifier and phase sensitive demodulator are incorporated in one unit or strain indicator; with d.c. units, separate high-stability amplifiers are employed together with some form of d.c. calibration. Slip rings, as signal transfer devices, have been fully described (1) and only a simple very successful 'homemade' type of slip ring may be mentioned which is constructed by bonding strips or silver 1/32in thick by in wide (0.8 x 6mm) to a rubber sheet. The assembly is wrapped around the shaft and attached to it by means of suitable straps or "hose clips". The details of the ring joint are shown in Fig. 1. Silver-graphite brushes are used in the normal way. This type of ring has performed admirably on shafts from 3 to 24in (75 to 600mm) diameter. O'OIO'' JRRASS STRIP TO WHICH RING 5 SOLDERED SILVER SLIP RING SOLDER FILLED JOIN Fig. 1. Details of ring joint of a silver slip ring. GAP WIDE LEAD OUT WIRES 1

2 Strain Measurement On the model scale it is often necessary to measure strain on dynamically similar models. A model of a fully cavitating propeller made of glassfibre filled resin (Fig. 2) was gauged on its blade surfaces for a deter- reduced number of leads within the shaft. The records were very good and noise free with both a.c. and d.c. energisation. With time the insulation broke down, but regular drying out allowed the tests to be completed. A further application is the determination of the propeller shaft torque on full-scale ships. A strain gauge torsionmeter compares favourably with other types of ship torsionmeter and has the unique advantage that values of torque can be determined without prior calibration, provided the modulus of rigidity of the shaft material is known. Strain gauge torsionmeters have been used over a wide range of shaft sizes. Fig. 3 shows a shaft section complete with commercial and home-made rings. The order of Fig. 2. Glassfibre filled resin model propeller. of car itating mination of the skin strain distribution. The difficulty with this type of experiment is to attach the gauges and complete the wiring and waterproofing of the installation without either significantly increasing the blade thickness or causing significant irregularities on the surface. Extremely small epoxy-backed foil gauges were used, 1200 and 90 rosettes together with single gauges. Larger gauges were bonded with their metal foil side to the blade surface so as to increase the effective thickness of the waterproofing layer. Tinned copper wire rolled into flat strip 000l5 by 0060 in (004 x 1 5 mm) was used for lead-out wires. The strip was bonded to the surface and allowed to follow the natural contour of the blades. Any change of direction was accommodated by overlapping a second strip and soldering. All leads were led to grooves in the propeller boss and joined to conventional insulated lead-out wires passing through the hollow drive shaft. For waterproofing, two thin coats of epoxy resin were applied over the whole surface. At some soldered joints local extra coatings were needed because the resin tended to "fall away" from high spots. The system was installed in a water tunnel and slip rings were fitted at the drive end of the propeller shaft. All gauges had one terminal connected at the propeller end to a common lead fed to one slipring. The other gauge connections were fed to the remaining rings. Although this method of slipring connection is considered to be bad practice, the technique was chosen to provide a maximum number of signals with a Fig. 3. Shaft section with commercial and 'home-made' slip rings. accuracy of a strain gauge torsionmeter depends on many factors; with care accuracies within ±1 % of full scale can be achieved. Research has shown that the torsionmeter is stable over long periods and that the signal output exhibits good linear properties. Force Balances Many types have been developed, from single-force devices to complete six-component balances. A few of them are described here. Cantilever balance This type of balance has a wide range of application for ship model experiments and a selection is shown in Fig. 4. For single force measurements, e.g. with a drag balance for hovercraft experiments, a hinge or pivot is incorporated at the loading point and the gauges are situated as near as possible to the position of maximum bending moment. Where a pivot is not practicable, as when holding a submerged body in a rigid attitude, two sets of gauges are placed a reasonable distance apart along the cantilever and the sets 2

3 are so connected that their individual outputs subtract. The resultant signal is directly proportional to force and independent of the point of application as long as the line of action of the force is not between the gauge stations. are heavy. are suitable only for calm water conditions and cannot be used in experiments involving waves, measure the bearing friction of the transmission shaft to the propeller, do not lend themselves to electrical recording. To overcome these difficulties, a strain gauge dynamometer as been developed. Obviously it is difficult to achieve the same order of accuracy but by making a range of units, all easily interchangeable, a satisfactory accuracy can be obtained. Two possibilities may be considered: direct substitution of a strain gauge dynamometer for the mechanical type, or measurement of thrust and torque directly at the propeller boss. Method (b) provides a simpler and cheaper dynamometer and also eliminates the problems posed by measurement of bearing friction and of inertia forces acting upon the shaft when the model is run in waves. Typical designs are shown in Fig. 6. Fig. 4. Several types of cantilever balance. Thrust and torque dynamometer Measurement of propeller thrust and torque in a ship model forms one of the most important basic experiments performed in any ship model towing tank. In the past this measurement has been done by means of mechanical dynamometers (Fig. 5) which have been developed into extremely accurate dynamometers but show a number of inherent disadvantages. They Fig. 6. Strain gauge dynamometers measuring both thrust and torque. Fig. 5. Mechanical dynamometer for ship model towing tank. The transducer element forms part of the propeller shaft and takes the shape of a rectangular bar, the long axis being the axis of rotation. Torque is measured by foil gauges bonded at 45 to the rotational axis. The maximum strain is of the order of 65 p. in/in. Thrust is measured by semi-conductor gauges placed on opposite parallel faces, along and transverse to the rotational axis: the maximum strain is of the order of 5p.in/in. The low compressive strain and the relatively high torsional strain make it impossible to align the thrust gauges so that they do not respond to torsional strain, and due to unavoidable mismatch they also respond to a bending moment. In order to eliminate these unwanted signals, two techniques have been developed. Torque interaction is almost completely eliminated by applying a portion of 3

4 a torque signal derived from a separate torque bridge (2). The bending moment interaction is eliminated by resistively shunting that half of the thrust bridge which gives the higher output for a given bending moment. This can be achieved satisfactorily only if a parallelsided transducer element is used, since the gauges must be situated exactly on opposite sides of the shaft axis to avoid any phase shift between the output signals from the two half-bridges as the shaft rotates. Calibration results have been excellent, the calibration factor for thrust being constant within ± 05 per cent, that for torque within ± 025 per cent. The sensitivity of both bridges changes with temperature linearly over the usual range of application (15 to 25 C). the scatter about the mean line being less than ±025 per cent for both bridges. Calibrations are therefore taken over a range of temperatures. During tests the constancy of the water temperature ensures a high degree of zero stability. Development of the six-component balance The earliest form of a multi-component strain gauge balance consisted of three cantilevers and measured three components of force. Attempts were made to isolate each flexure so that each cantilever carried only one component of load. This technique is often used but, in the author's opinion, only complicates the balance design without achieving the desired effect. Results of equal merit are obtained by placing gauges on neutral axes or by relying on the cancellation of the bridge network to eliminate unwanted signals. Coupled with the techniques described already, this can produce balances of simple design. This philosophy also led to the design of the first six-component balance (Fig. 7). Three struts separate two plates, the "force plate" carrying the test vehicle Fig. 7. Six-component balance formed by three measuring struts. and the "earth plate" bolted to a support structure. All loads are transmitted from one plate to the other through the three struts which in encastré bending give two force directions and in direct axial loading the third. Algebraic sums of the independent outputs give the three corresponding moments (Fig. 8). Since nine signals are produced, electrical summing circuits *c ca.in $Ysnh. PITcH PQCE ROIL UOM(N1 THREE STRUT - SIX COMPONENT FORCE BALANCE Fig. 8. Output circuits for six-component balance formed by three struts. are provided to give the six components of loading. Conventional gauges are used in bending and semiconductor gauges for the axial loading. To increase the semi-conductor strain level, the struts are shaped as double tapers reducing at the centre where the semiconductor gauges are located and where the maximum direct strain and the minimum bending strain are found. The struts are encased in flexible oilfilled bellows giving a high degree of temperature stability for the semi-conductor gauges. The advantages of this balance are that the construction is simple, the strut design is controlled by the force system alone, the physical position of the struts controls the sensitivity to moments. The original balance was developed for a hydrofoil project and the prototype was built, gauged and calibrated in about four weeks. Since then, a number of similar balances have been built to carry forces up to 5000 lbf (25 kn) and moments up to 2000 lbf ft (2750 4

5 Nm). For a given set of forces and moments the interaction values are low, not normally exceeding 5 per cent for full load in all planes; the maximum interaction occurs on the semi-conductor gauges. In tests the measured values of interaction are higher since test conditions may not be similar to the original design condition. Interaction, within reason, does not prohibit satisfactory measurement but complicates balance calibration because (I) calibration is time consuming and expensive, (2) precise multidirectional loading is difficult to achieve for large loads where deadweight calibration is not practicable, (3) the balance must have a very high degree of stability so that frequent calibrations are not required. For the type of experiment performed in the Ship Division of NPL, a wide range of force measurement is required and many multi-component balances have to conform to certain hydrodynamic requirements, for instance to fit into a given shape. Hence many different balances are constructed and sometimes have a life limited to one set of experiments only. A balance therefore should be cheap to produce and have little or no interaction. From these considerations a new type of balance has been built, the shear plate balance (3) which measures the linear strain resulting from shear. Its advantage is that the end fixing conditions are not critical. The simple type of flexure unit shown in Fig. 9 has excellent characteristics and is insensitive in all directions other than the measuring direction. A combination of such individual units may measure up to six components of loading (Fig. 10). Fig. 10. Prototype six-component balance made up from simple flexure units. Acknowledgment The work described has been carried out at the National Physical Laboratory and the author is indebted for help to the staff of the Ship Division of NFL, in particular to members of the Equipment Group. Fig. 9. Simple flexure unit based on shear plate principle. References D. A. DREW, "SlipringsA Review", Strain, Vol. 2, No. 4, 1966, p H. B. BOYLE, "Method of reducing signal interaction of a strain gauge balance", Strain, Vol. 4, No. 2, 1968, p H. B. BOYLE, "The shear plate as a multi-component balance", Strain, Vol. 4, No. 2, 1968, p

6 Reprif2ted from "Strain" Vol. 6, No. 4, October, 1970