Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development and testing of an aerodynamic marine propeller test facility : an investigation into making… Davis, Karl Richard 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-0199.pdf [ 38.89MB ]
Metadata
JSON: 831-1.0080786.json
JSON-LD: 831-1.0080786-ld.json
RDF/XML (Pretty): 831-1.0080786-rdf.xml
RDF/JSON: 831-1.0080786-rdf.json
Turtle: 831-1.0080786-turtle.txt
N-Triples: 831-1.0080786-rdf-ntriples.txt
Original Record: 831-1.0080786-source.json
Full Text
831-1.0080786-fulltext.txt
Citation
831-1.0080786.ris

Full Text

D E V E L O P M E N T AND T E S T I N G O F A N A E R O D Y N A M I C M A R I N E P R O P E L L E R T E S T F A C I L I T Y : A N INVESTIGATION INTO M A K I N G BOATS F L Y by Karl Richard Davis B.E. Hons., University of Canterbury, 2000 A THESIS SUBMITTED IN P A R T I A L FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF APPLIED SCIENCE . in THE F A C U L T Y OF G R A D U A T E STUDIES DEPARTMENT OF M E C H A N I C A L ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A April 2004 © Karl Richard Davis, 2004 Library Authorization In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Karl Davis 19th April, 2004 Name of Author (please print) Date (dd/mm/yyyy) Title of Thesis: Development and testing of an aerodynamic marine propeller test facility: An investigation into making boats fly Degree: Master of Applied Science Department of Mechanical Engineering The University of British Columbia Vancouver, BC Canada Year: 2004 Abstract Current techniques for performance testing marine propellers involve the use of extremely expensive facilities such as towing tanks or cavitation tunnels. These methods are also limited as to the maximum propeller size they can test. If it were possible to perform such testing in a more commonly available wind tunnel, marine propeller testing and research could be carried out more widely and upon larger propellers. A device for testing marine propeller performance in a wind tunnel has been designed, built and commissioned. The device features a marine propeller that is 24 inches in diameter that can be spun at up to 3000rpm. Instrumentation for measuring the propeller's thrust, torque, rotational speed and the wind tunnel free stream velocity are integral features of the device. Dimensional analysis was applied to ensure the applicability and validity of the results determined in air to the hydrodynamic case. A series of tests were performed with the equipment to investigate the repeatability of the results, the effect of varying Reynolds number, and to evaluate the device as a comparative tool. The repeatability was found to vary with test conditions. For the preferred test speed of 1500rpm, a repeatability of 1.3% was achieved at peak efficiency. The Reynolds number investigation showed that as the Reynolds number increased, the Reynolds number dependence of the results was reduced. The goal, however, of finding a test speed which exhibited Reynolds number independence was not achieved. The comparative performance test involved testing two identically specified propellers. Subtle differences in the performance characteristics of the two propellers were observed. i i Table of Contents Abstract i i Table of Contents i i i List of Figures v List of Tables vi Acknowledgements vii 1 Introduction 1 1.1 Propeller Testing Theory 1 1.2 Current Propeller Testing Techniques 3 1.3 Research Goals and Motivation 5 1.4 Aerodynamic Testing 6 1.5 Previous Work 8 2 Experimental Design 12 2.1 Test Rig Design 12 2.1.1 Row Similarity 12 2.1.2 Design Considerations 14 2.1.3 Detail Design and Equipment Selection 16 2.2 Test Rig Construction 26 2.3 Test Rig Commissioning 28 2.3.1 Equipment Problems 29 2.3.2 Signal Noise and Drift 31 2.3.3 Other Commissioning Issues 36 3 Experimental Results and Discussion 37 i i i 3.1 Repeatability 37 3.2 Reynolds Number Dependence 44 3.3 Comparative Performance Test 54 3.4 Propeller Geometry Investigation 59 3.5 Error Discussion 63 4 Conclusions and Recommendations for Future Work 71 4.1 Conclusions 71 4.2 Recommendations for Future Work 72 4.2.1 Performance Improvements to Experimental Apparatus 72 4.2.2 Further Research 73 References 75 Appendix A - Engineering Drawings 76 Appendix B - Instrumentation Wiring, Isolation and Shielding Diagram 90 Appendix C - Failure Effects, Modes and Criticality Analysis 91 Appendix D - Typical Thrust and Torque Calibration Charts 95 Appendix E - Repeatability Results at 500,1000 and 2000rpm 98 Appendix F - Reynolds Number Dependence Comparison Plots between Successive Test Speeds 108 Appendix G - Calculation of parasitic thrust and torque 124 iv List of Figures Figure 1.1: Typical propeller performance chart (Source: Lewis 1998) 3 Figure 2.1: General Arrangement of Test Rig 16 Figure 2.2: Air bushing loading diagram 21 Figure 3.1: Repeatability of Efficiency Measurements at 1500rpm 39 Figure 3.2: Repeatability of Torque Coefficient Measurements at 1500rpm 40 Figure 3.3: Repeatability of Thrust Coefficient Measurements at 1500rpm 41 Figure 3.4: Blade element velocity and force diagram 46 Figure 3.5: Efficiency Reynolds number analysis 48 Figure 3.6: Torque coefficient Reynolds number analysis 49 Figure 3.7: Thrust coefficient Reynolds number analysis 50 Figure 3.8: Efficiency comparison 55 Figure 3.9: Torque coefficient comparison 56 Figure 3.10: Thrust coefficient comparison 57 Figure 3.11: Comparison of Propeller Geometries 60 Figure 3.12: Comparison of Propeller Section Profiles 61 Figure 3.13: 2400rpm and 2500rpm Efficiency Comparison 66 Figure 3.14: 2400rpm and 2500rpm Torque Coefficient Comparison 67 Figure 3.15: 2400rpm and 2500rpm Thrust Coefficient Comparison 68 v List o f Tables Table 2.1: Preliminary Experimental Test Conditions 23 Table 3.1: Performance parameter repeatability at peak efficiency 43 Table 3.2: Test Reynolds Number Ranges 45 Table 3.3: Performance parameter trends with successive speed change (roughly proportional to Re) 47 vi Acknowledgements Firstly I would like to express my gratitude to my supervisors, Dr. Sheldon Green and Dr. Carl Ollivier-Gooch, for their continued faith and patience throughout the battle. I wish to extend my thanks to the following people: Mr. Glenn Jolly for his tireless technical assistance; Isabella L i , for her able laboratory assistance, and more importantly her friendship and chocolate; the many generations of my Green College extended family, who have enriched my life in ways I never imagined possible; and lastly to the snow, the hills and the waters surrounding U B C for preserving my sanity. Funding for this research was provided by the Natural Sciences and Engineering Research Council of Canada. Greatly appreciated financial support was also provided by the R. Howard Webster Foundation Fellowship vii 1 Introduction The basic concepts underlying marine propeller testing are relatively simple, and yet the practicalities of performing such tests are far from this. The following chapter presents these concepts, as well as the practical ways in which these are implemented in real testing. The drawbacks of the current testing methods are then discussed, which leads into a discussion of the goals and motivation of this thesis. Finally, a review of pertinent previous work is provided at the conclusion of this chapter. 1.1 Propeller Testing Theory Propeller performance characteristics are defined in terms of the thrust produced by and torque supplied to the propeller. From these the propeller efficiency may be determined. For a given propeller these parameters change with the ratio of the propeller blade tip velocity to the relative velocity between the propeller and the far field flow. This ratio is termed the propeller advance ratio. These performance measures are affected by a range of factors such as hull-propeller interactions and water conditions. Consequently, a form of standardised testing known as an open water test has been developed to enable more objective comparison between propellers. The open water test consists of testing the propeller in as isolated condition as possible, with outside influences such as the boat hull, waves and wind removed. Such tests are usually carried out in a towing tank or cavitation tunnel on scaled down propeller models. As with any scale model testing, non-dimensionalising the 1 characteristic parameters is necessary to ensure the results produced are generally applicable. The pertinent dimensionless groups for propeller testing are given below: J = V / n D Advance Ratio (1) K Q = Q / p n 2 D 5 Torque Coefficient (2) K T = T / p n 2 D 4 Thrust Coefficient (3) Re = p Co.? n D V( f + (0.7 n )z) /\i Reynolds Number (4) Ca = ( P - P v ) / p V 2 Cavitation Index (5) T] = ( J / 2 7 t ) ( K T / K Q ) Efficiency (6) where: (m/s) (rev/s) (m) (m) (kg/m3) (kg/m-s) (kg-m/s2) (kg-m2/s2) (kg/m-s2) (kg/m-s2) These are the standard parameter definitions used for marine propeller testing1. The advance ratio characterises the relationship between the boat speed and the propeller rotational speed. The torque and thrust coefficients are simply non-dimensional forms of the torque supplied to, and thrust produced by the propeller. The cavitation index essentially defines the likelihood of cavitation to occur within the given flow conditions. A slightly different formulation of the Reynolds number is used. The length scale used is the propeller chord at 70% of its radius. Similarly the velocity scale is taken as the vector summation of the free stream velocity and the propeller linear velocity again at 70% of V Linear forward speed n Rotational speed (RPS) D Propeller diameter Co.7 Propeller chord at 70% radius p Fluid density M- Fluid viscosity T Propeller thrust Q Propeller torque Pv Liquid vapour pressure P Fluid static pressure 1 Most of these groups are not strictly non-dimensional and are defined in this manner essentially for historical reasons. Nonetheless, as their units are defined only in terms of revolutions (which are proportional to the non-dimensional radians) they behave in the same manner as non-dimensional groups. 2 the propeller radius. This definition of the Reynolds number is basically an average Reynolds number for the propeller. In addition to this set of defining parameters is the propeller efficiency, which is included separately as it is a combination of the advance ratio, thrust and torque coefficients. As such it is not an independent parameter. Nonetheless, it is included as it is very important for quantifying propeller performance. 10 03 2 .0.4 , . / \ 0,08 0.06 004 0.0? 0,2 0,4 0.6 0.8 J » V * / P i t 10 12 Figure 1.1: Typical propeller performance chart (Source: Lewis 1998) A typical propeller performance chart is shown in figure 1.1. The key characteristics of this plot are that the torque and thrust coefficients both gradually drop off to zero, whilst the efficiency curve shows a gradual increase with increasing advance ratio, before it drops quite sharply to zero after peak efficiency is reached. 1.2 Current Propeller Testing Techniques Three methods are typically used in the testing of marine propellers. These methods are towing tank tests, cavitation tunnel tests and sea trials. Towing tank tests involve mounting the propeller on a long shaft well clear of the propeller driving apparatus. Instrumentation on the shaft measures the torque supplied to and thrust produced by the propeller. The propeller is pulled through the towing tank by a carriage at a set velocity with a set rotational speed. This is done for a range of advance ratios to produce plots of efficiency, torque and thrust coefficients. Cavitation tunnel tests are performed similarly to towing tank tests, with the key difference being that the propeller is kept laterally stationary (whilst rotating) as the water is pumped past it at a given velocity. Sea trials are carried out in quite a different manner from the other tests. These are performed with the propeller running on a boat. The tests are carried out by running the boat at a range of test speeds with a range of drogues attached to more heavily load the propeller. Instrumentation is added to the propeller shaft to measure the thrust, torque and propeller rotational speed. Towing tank and cavitation tunnel tests afford considerably better accuracy than that of sea trials. Wind, waves and water currents can potentially influence sea trial tests significantly, and these factors are very difficult to avoid or account for. Furthermore, providing instrumentation within a marine environment to accurately measure the torque and thrust on the propeller shaft can be very challenging. The primary advantage of sea trials is that large propellers can be tested with little added complication. In contrast, the physical size and power requirements of towing tank and cavitation tunnel facilities restricts the maximum size of propeller that can be tested. At present the largest worldwide propeller test facility can test propellers up to 1.4m in diameter, with more typical sizes being 250mm in diameter. Towing tank and cavitation tunnel facilities 4 suitable for testing marine propellers are not widely available and are generally very expensive to use. At present there are roughly 15 laboratories worldwide with these facilities. 1.3 Research Goals and Motivation This piece of work has been an intermediary step towards a larger goal of our research group. The end goal of this research is to experimentally investigate, and ultimately optimise the geometry of a novel propeller modification known as the ducted tip propeller. The ducted tip modification is made by attaching a flow-through duct to the tip of each propeller blade. This device has been shown in sea trials (Hordnes and Green 1998) and cavitation tunnel testing (Straver 2002) to improve propeller performance and reduce cavitation under certain operating conditions. These improvements are achieved by redistributing the vorticity shed from the propeller blade tip. The duct transforms the usually small, rapidly spinning tip vortex, into a larger more slowly moving flow region. Whilst Hordnes and Green's sea trials showed improved overall performance, Straver's laboratory testing showed reduced performance. Differing duct geometries were used in their work, which implies sensitivity of the propeller performance to precise geometry. Consequently, investigating and ultimately optimizing the duct geometry is a major long term goal of this research. Furthermore, because of the relatively small performance changes anticipated with altered duct geometry, gaining sufficient accuracy from the experimental apparatus is considered to be a significant challenge. 5 In order to investigate and optimise the geometry of the ducted tip propeller a suitable test apparatus is required. Previous experimental work within our group has been performed at the Institute of Marine Dynamics (IMD) in St. Johns, Newfoundland. This arrangement has been found to be less than satisfactory, due to the physical distance between supervisor and students, which makes communication very difficult. In addition, there was a lack of technical control over the testing process, as it was carried out in a semi-commercial facility. Furthermore, the maximum size (250mm diameter) of propeller that could be tested at the IMD did not lend itself well to the experimental optimization that was desired. As such it was decided that an experimental facility would be developed within the University of British Columbia Department of Mechanical Engineering, and this has been the focus of this thesis. It was decided that the testing would be carried out in a wind tunnel, as an aerodynamic test rather than the more conventional hydrodynamic test. 1.4 Aerodynamic Testing Performance testing of marine propellers is generally carried out in a highly specialized and expensive facility such as a cavitation tunnel or a towing tank. If the same testing could be performed in a more commonly available wind tunnel, experimental marine propeller research might be carried out more readily and economically. A number of advantages exist to performing such testing in air instead of water. Firstly, suitably sized wind tunnels such as the University of British Columbia Department of Mechanical Engineering boundary layer wind tunnel are considerably less expensive to use and more commonly available than conventional propeller test facilities. 6 Consequently, development of a suitable apparatus for aerodynamic testing of propellers might allow experimental marine propeller research to be carried out more readily and economically. Secondly, aerodynamic testing allows many problems associated with using water as a working fluid to be avoided. Such problems include incompatibility of precision sensory equipment with water and corrosion issues due to the use of chlorine (or other water sanitizers). A particularly important problem avoided by aerodynamic testing, is the experimental error associated with water seals. These seals, located between the rotating shaft and housing affect the thrust and torque measurements incurring substantial errors. In aerodynamic testing such seals are not required. Another important advantage afforded by aerodynamic testing is the ability to test much larger propellers. Power requirements for testing marine propellers in water generally restrict the maximum propeller size (for example, at the IMD the maximum diameter is approximately 250mm). In comparison, a 24-inch diameter propeller is being tested in this work. This is a significant advantage with regard to investigation of the ducted tip propeller, as well as other experimental marine propeller research. The combined work of Hordnes and Straver indicates that the ducted tip propeller performance is sensitive to the tip's detailed geometry. Investigating and altering this geometry at the propeller scale at which Straver worked (250mm diameter) would be extremely difficult (in these tests the ducts were only a few millimetres on diameter), however, in using aerodynamic testing, with a propeller diameter of 24-inches, considerably more insight might be 7 gained. This is particularly the case when methods of repeatably attaching and detaching the ducted tips are established. With respect to general propeller experimental research, the ability to test larger propellers is similarly advantageous, as at a suitably large scale, features such as blade shape and surface finish can be more easily assessed, and accurate manufacturing and machining of propeller parts can often be performed more easily. Whilst aerodynamic testing offers a range of advantages over conventional hydrodynamic testing, certain disadvantages do exist. Due to the lower density of air than water, the loadings measured in air will be considerably smaller than in water, making them potentially more difficult to measure accurately. Disadvantages also exist as a consequence of testing larger propellers, as the propellers will be both more expensive, as well as much more difficult to physically handle. The validity of such testing is also a key concern. It is not immediately obvious that the results found in air will in fact apply in water. Because, for example, cavitation, an important flow feature in determining marine propeller performance, does not occur in air. Considerable thought and investigation has been applied to resolve these issues. 1.5 Previous Work Very little work has been done in the area of aerodynamic marine propeller testing. As part of a PhD. thesis Kotb (1984) carried out research with such a device. Kotb's work was primarily focused upon the flow field downstream of windmills and both 8 aerodynamic and marine propellers. Whilst this device was constructed, it appears that very little work was done regarding the intricacies of aerodynamic testing of marine propellers. Of the work published by Kotb, all tests using the device were performed using lightweight propellers resembling the type used in air, rather than the much heavier marine type. Given the dimensions of the device it would appear that it would be suitable for considerably smaller and/or slower turning marine propellers than the current device. Furthermore, Kotb was unable to reliably implement a torque transducer. In a collaborative project between Massachusetts Institute of Technology and the University of Michigan, a helicopter hover test stand for researching advanced rotor blade concepts has been built. The test stand features a 10 foot diameter rotor, and instrumentation capable of measuring the lift and moment generated by the rotor. Investigations into ducted tip propellers have been carried out within the University of British Columbia Department of Mechanical Engineering. Hordnes and Green (1998) carried out sea trials of a 45ft seine boat, comparing the performance of a 36 inch diameter conventional tip to a ducted tip propeller. The trials included both quantitative measurement of propulsive efficiency as well as observations of cavitation inception. The quantitative work revealed an increase in overall efficiency of up to 6% (at high advance ratios) over the conventional propeller. The cavitation observations were performed by a diver taking video recordings of the propeller whilst the boat was tied to the dock. The observations revealed a 50% reduction in the cavitation inception index for the ducted tip propeller. 9 Experimental comparison of a 250mm diameter conventional and ducted tip propeller was carried out by Straver (2002). This work was performed in both a towing tank and cavitation tunnel. The towing tank results showed a general decrease in propeller efficiency over a wide range of test conditions. These results contradict those of Hordnes and Green (1998). The cavitation tunnel testing showed improved cavitation performance of the ducted tip propeller, with the tip vortex appearing to be substantially diffused by the duct. Significant work has been carried out into an array of other marine propeller tip modifications such as the ducted tip. Platzer and Souders (1979) reviewed efforts to alleviate the effects of tip vortices on marine propellers prior to 1980. From this review it was concluded that bulbous tips, porous tips and linear mass injection tips had potential to delay tip vortex cavitation in practical applications. Bulbous propeller blade tips were investigated by Crump (1948). These were shown to improve cavitation performance with little overall performance penalty. Conversely, Johnsson and Rutgersson (1991) carried out a similar study upon bulbous tips, with their results indicating reduced cavitation performance and overall performance. Itoh et al. (1987) found that bladelets could delay cavitation inception and slightly increase propeller efficiency. This contradicted the earlier work on bladelets of 10 Goodman and Breslin (1980), which showed reduced cavitation performance and overall performance. Porous blade tips were investigated by both Mani et al (1988) and Arakeri (1985). These were shown to improve cavitation behaviour with no significant performance penalty. One of the key conclusions that can be drawn from the various propeller modification tests that have been carried out is that the results are highly inconsistent. The two most likely causes of these inconsistencies are that the exact geometry of these modifications is very important, or that it is very difficult to produce high quality marine propeller test results. It seems likely that both of these are contributing factors, and the focus of the present work is to address these issues and ultimately overcome them. 11 2 Experimental Design The present chapter describes the experimental test equipment used for this project. The design considerations that influenced the final design are presented. The key aspects of the construction process follow the design section, and the chapter closes with a description of the rather arduous task of commissioning. 2.1 Test Rig Design The design of a marine propeller aerodynamic test rig presents some unique challenges. Questions of how to ensure that the results found in air apply in water, and how to measure the relatively small loads produced whilst restraining an apparatus with substantial kinetic energy must all be resolved. The following section discusses these issues and explains how they have led to the final design of the apparatus. 2.1.1 Flow Similarity Before serious consideration could be given to developing the apparatus, the applicability of aerodynamic testing upon a marine propeller had to be investigated. This has been done using dimensional analysis. The dimensionless groups, as identified in chapter 1, are listed below. The propeller blade tip Mach number has been added to this set as for testing in air it must also be considered. J = V / n D Advance Ratio (1) KQ = Q / p n 2 D 5 Torque Coefficient (2) K T = T / p n 2 D 4 Thrust Coefficient (3) Re = p Co.? n D V( 52 Ca = ( P - P v ) / p V 2 + (0.7 7i )2) l\i Reynolds Number (4) Cavitation Index (5) Ma = n n D / a Propeller Blade Tip Mach Number ,(6) 12 To ensure flow similarity these groups must be matched between the two flow situations. The matching of advance ratio, torque and thrust coefficients is implicit, as they are the variables being measured. Matching of the Mach number is generally only considered important for Mach numbers greater than 0.3. The hydrodynamic Mach number remains well below this due to the relatively high speed of sound in water. For the aerodynamic flow situation this places a constraint upon the maximum propeller speed and diameter. Cavitation does not exist in air, so matching of the cavitation number is not possible. For the proposed future duct geometry optimization work, this should have little detrimental effect. This is because any gains in performance achieved by optimizing the duct are expected to be due to a reduction in the intensity of the propeller tip vortex. This reduction in tip vortex intensity will reduce the probability of cavitation occurring in water, and potentially further improve the propeller performance. This implies that whilst true flow similarity may not be achieved, the optimization results should remain valid in water. Reynolds number matching is not practical, as it would require the aerodynamic Mach number to exceed 0.3. Hydrodynamic Reynolds numbers are typically on the order of 10,000,000, whilst for the aerodynamic tests these will be around 1,000,000. The results of the work carried out by Straver (2002) indicate that for Reynolds numbers greater than 200,000, performance is independent of Reynolds number. As such, it is anticipated that 13 exact Reynolds number matching will not be necessary. Nevertheless, efforts shall be made to perform tests over the range of Reynolds number afforded by the equipment to verify this. 2.1.2 D e s i g n C o n s i d e r a t i o n s During the design process several key considerations were established which have been broadly grouped under the headings of accuracy of results, clear airflow and mechanical strength. The significance of each of these considerations, which have shaped the design into its final form, are discussed below. 2.1.2.1 A c c u r a c y o f R e s u l t s Ensuring that the data obtained are sufficiently accurate is the most important design consideration when developing a new experimental apparatus. Because the primary purpose of the test rig is to allow experimental optimisation to be performed, a particularly high level of accuracy is required from the apparatus. It is anticipated that the complete optimisation process will reveal improvements of the order of 5 % . This implies that the overall uncertainty of measurements taken will need to be less than about 2 % . Obtaining measurements of this accuracy is ordinarily not unreasonably challenging. However, under important operating conditions, specifically near peak propeller efficiency, the torque and thrust both become small. This makes them much more difficult to measure accurately. Hence, achieving the required level of accuracy across the full testing spectrum is a surprisingly challenging task. 14 2.1.2.2 Clear Airflow The testing is intended to mimic standard open water testing of propellers. As such, the propeller should be mounted in a manner which minimizes extraneous influences upon the airflow. In particular, the effect of all test rig components and sensing equipment on fluid flow, both upstream and downstream of the propeller, must be carefully considered. This is necessary so that the results produced indicate the isolated performance of the propeller. This also makes the results potentially suitable for comparison with equivalent hydrodynamic tests. 2.1.2.3 Mechanical Strength The physical strength of the test rig is naturally an important design consideration. Several loading conditions must be allowed for including the static deadload of the propeller, as well as fatigue loadings on rotating parts due to cyclically varying forces. Loads created by centripetal forces due to out of balance rotating equipment must also be allowed for. Of particular concern is the propeller, which due to its size, is the most likely component to produce significant out of balance forces. Furthermore, attaching ducted tips to the propeller may introduce considerable imbalance. Additionally, the vibration characteristics of the test rig assembly must be considered. This is to ensure that the propeller rotation will not excite natural frequencies in the test rig, which might lead to mechanical failure. Specifically of concern is the lateral, torsional and axial vibration of the shaft, coupling and instrumentation assembly. 15 2.1.3 Detail Design and Equipment Selection The array of commonly conflicting considerations just presented have moulded the design into that shown in Figure 2.1. The propeller is mounted on a cantilevered shaft. This shaft is radially restrained by three air bushings. The propeller thrust is channelled through the driveline onto a set of thrust bearings, which are then linked to a load cell, which provides the axial restraint for the shaft and propeller. A rotary torque sensor is located within the drive train between the motor and the propeller shaft. A n explanation of each of the key features and why they were selected follows. Thrust Load Cell \ Secondary Shaft > Thrust Bearings> Torque Sensor • Air Bushings Propeller Electric Motor Air Bushings Cantilevered Propeller Shaft F igure 2.1: Genera l A r r a n g e m e n t of Test R i g 16 2.1.3.1 Propeller A four blade, 24-inch diameter, bronze propeller with pitch to diameter ratio of 1.0 was chosen for the test rig. The propeller model selected was the Dyna-Quad manufactured by the Michigan Wheel Corporation. In selecting the propeller size, there were several important considerations. Firstly, the propeller needed to be large enough to allow ducts to be easily attached and properly aligned. Conversely, advantages existed in using a smaller propeller, which is less expensive, can be more readily physically handled, and is more easily mechanically restrained when mounted on a rapidly spinning shaft. The 24-inch size represented an amicable compromise between these considerations. Weighing 18kg, it could be reasonably physically handled and mechanically restrained, whilst being large enough to facilitate attachment of ducts. Two models of propeller, both manufactured by the Michigan Wheel Corporation were looked at. The Dyna-Quad is primarily used on large (over 40 foot) cruising or fishing boats. The alternate propeller, known as the Work-Horse, is commonly used on tugboats. As cavitation is most likely to occur under high load and low speed (ie: low advance ratio) operating conditions, which are common for tugboats, it is suspected that the Work-Horse propeller is designed to perform well under these conditions and therefore already reduce cavitation. As such, it would probably benefit less from the addition of the ducted tips. Consequently the Dyna-Quad was chosen, as it would allow the effect of the ducted tips to be more easily determined. A notable disadvantage of this decision was 17 that the results will have less practical applicability for high load propellers. However, as a trade-off to allowing greater theoretical insight, this was considered appropriate. Consideration was given to using propellers with various numbers of blades. Fewer blades would make attachment of the ducted tips simpler. It was decided that the more commonly used number of four, for a propeller of this type, was more appropriate. The pitch ratio of 1.0 was chosen, as it was considered to be a typical ratio, and also the results of the work performed by Straver on ducted tip propellers were most positive at this ratio. Some minor customizations were also made to the propeller. The standard 1.5 inch diameter tapered bore was altered to a 2.0 inch straight bore, with dual keyways located 180° apart. The bore was increased to reduce shaft stresses. The taper was replaced with dual keyways because the standard taper is orientated in the reverse direction from that required for an open water test. 2.1.3.2 Shaft The propeller shaft has been designed in a cantilever arrangement, with a 12-inch (or one half the propeller diameter) spacing between the propeller hub and the front bearing. The shaft is 3 inches in diameter and made from 4140 steel, which has been chrome plated and surface hardened. 18 With respect to airflow, the selection of the distance between the propeller hub and the first bearing is very important. In the hydrodynamic testing carried out by Straver, the propeller was located approximately three diameters from the test rig. In contrast, the rig built by Kotb used a distance of approximately one diameter between the propeller hub and supporting structure. Due to strength considerations, particularly with respect to the air bushings and lateral shaft vibration, the distance for this rig has been chosen as one half the propeller diameter. At this distance, about 12% of the disk area, not including the propeller shaft, is obstructed. Without carrying out preliminary testing upon the effects of the downstream supporting structure, it is not possible to comment on the proximity of this obstruction in an informed manner. Such testing would be of limited worth however, as the strength and vibration requirements of the test rig are essentially non-negotiable. As such, the shaft has been designed to position the propeller in as clear air as a conservative engineering design allows. The shaft diameter has been selected based upon the size of the largest commercially available air bushing. This was also considered to be the largest sensible size for the shaft with respect to its weight and cost. The shaft has been chrome plated and surface hardened so that it is compatible for use with the air bushings. 2.1.3.2.1 Shaft Vibration The vibration characteristics of the shaft assembly have been considered to ensure that the shaft and propeller rotation will not excite natural frequencies, which might lead to mechanical failure. Specifically of concern are the lateral, torsional and axial vibration modes of the shaft and its assembly. 19 Conventional conservative engineering design is to ensure that the calculated natural frequencies are at least twice the running/excitation frequency of the device. The testing speed range of the propeller is proposed to be between 2000-3000rpm. This indicates a minimum natural frequency of 6000rpm for conservative design. The importance of vibration is more than just whether the equipment will hold together mechanically. As we are attempting to take very accurate measurements, any vibration transferred to the measurement devices will contribute to signal noise and make obtaining accurate data difficult. The lateral vibration natural frequency has been determined by modeling the shaft as a cantilever with a mass equal to that of the propeller fixed to the end. For the chosen geometry the lateral vibration natural frequency is approximately 9400rpm. Clearly this is suitably high to provide some confidence that lateral vibration should not be a problem. It is worth noting that if the geometry were arranged as previously planned, with the distance between the propeller and first bearing being equal to one propeller diameter, this frequency would be approximately 3500rpm. Furthermore, the inherent assumptions in these calculations are expected to result in an overestimate of the natural frequency, indicating that such an arrangement would not be permissible. Torsional and axial vibration calculations were performed by modeling each of the parts of the drive train as discrete components, having either stiffness or mass. Their results yielded a natural frequency of 970rpm and 600rpm for the torsional and axial vibration 20 modes respectively. This was not expected, as the size of the shaft makes this system appear very stiff under torsional and axial load. However, the torque and axial load sensing equipment is necessarily comparatively soft, resulting in these low natural frequencies. These natural frequencies are well below the test speed range. However, to reach the test speeds, they must be 'run-through' during both acceleration and deceleration. Furthermore, various harmonics are expected to exist above the natural frequencies and therefore potentially within the test speed range. The exact response of the apparatus to excitation is not practical to model. As such, during the commissioning process it was intended that the vibrational behaviour of the rig would be investigated by slowly adjusting the running speeds so that any resonances could be found under controlled conditions, and thereby avoided during testing. 2.1.3.3 Air Bushings Three 3-inch diameter air bushings have been used to radially restrain the propeller shaft. These were chosen in favour of more conventional bearings due to their very low resistance torque and nil axial restraint. These features allow significantly improved measurement accuracy, as they permit the propeller thrust to be applied directly to a load cell and the propeller torque to be measured with minimal interference from the bearings. Ii h FOOB Out of balance force M Propeller mass g Gravitational constant w Shaft weight per unit length h Length between propeller and first bushing I2 Length between bushings FA Reaction force on front bushings FB Reaction force on rear bushing Figure 2.2: Air bushing loading diagram 21 The bushing size was selected for its load bearing capacity and because it was the largest that was commercially available. The load bearing capacity of the air bushings has been the primary strength design constraint. The shaft length has been determined based upon the required bearing spacing, as has the spacing between the propeller and the first bushing. For the purpose of bushing load calculations, the shaft was modeled as a beam supported by pin joints. The modeled arrangement is shown in figure 2.2. Clearly this model has some significant limitations in that the bushings (particularly the front two) will not behave purely in this manner, as they will certainly provide some moment resistance, however, this is seen as the most sensible approximation that can be reasonably calculated. The out of balance force is the most important loading. The maximum continuous out of balance force has been estimated to be 500N. This equates to the force created by 17 grams of out of balance mass located on the propeller tip, or roughly 10% of a duct's mass. This produces a radial load of 80kg on the bushings, which is their rated load. The overload scenario is considered to be the event where a ducted tip becomes detached from the propeller. The attachment of these tips is considered to be a major challenge. As such, it seems prudent to design for this possibility. In this event an out of balance force of up to 4.5kN (for a 150 gram duct rotating at 3000rpm) would be produced. This would impose a load of 450kg on the air bushings. The manufacturer's stated maximum overload for the bushings is 455kg. This condition is expected to cause permanent damage to the air bushings, but should not result in catastrophic failure. 22 Aligning three air bushings on a rotating shaft would ordinarily be a difficult task. This is due to the very fine tolerances which these devices require. To avoid this problem a unique design of air bushing has been used. These air bushings, manufactured by New Way, mount within their housings upon o-rings. These o-rings are compliant enough to allow the bushings to self-align. 2 . 1 . 3 . 4 E x p e r i m e n t a l T e s t C o n d i t i o n s Early in the design process it was identified that a sound estimate of the measured variables was required. These would be used to establish the range of measurements required, and from this the type, size and accuracy of transducers could be selected. Artv.inc. R.ilio Free Stream Propeller Velocity SpfOd (m/s) (RPM) K , Thrust (N) KQ Torque (Nm) Power (Wiilf.) 0.05} 1.5 3,000 0.48 197 0.074 19 5,873 0.10 3.0 3,000 0.46 189 0.071 18 5,665 0.15} 4.6 3,000 0.44 181 0.069 17 5,457 _ ! 6.1 3,000 0.42 172 0.066 17 5,248 0.25] 7.6 3,000 0.39 163 0.062 16 4,957 0.30J 9.1 3,000 0.37 155 0.059 15 4,665 6.35| 11 3,000 0.35 145 0.056 14 4,457 0.40| 12 3,000 0.33 135 0.054 14 4,249 0.45J 14 3,000 0.30 125 0.050 13 3,957 0.50I 15 3,000 0.28 116 0.046 12 3,666 0.55! 17 3,000 0.26 106 0.043 I _ 3,416 0.60J 18 3,000 0.23 96 0.040 , _ 3,166 0.65J 20 3,000 0.21 85 0.036 9.1 2,874 0.70J 20 2,812 0.18 66 0.033 7.2 2,127 0.75] 20 2,625 0.16 49 0.029 5.7 1,562 0.80| 20 2,461 0.13 37 0.026 • 5 1,149 0.85| 20 2,316 0.11 27 0.023 • 824 0.901 20 2,187 0.09 19 0.019 2.5 581 0.95] 20 2,072 0.06 12 0.016 i.9 412 1.00| 20 1,969 0.04 7.1 0.013 r ~ T~4 282 Table 2.1: Preliminary Experimental Test Conditions Table 2.1 shows the preliminary test conditions selected. In producing this table the thrust ( K T ) and torque (KQ) coefficients were estimated using GAWN series propeller 23 data, that has been adjusted for the selected propeller using KT and KQ multipliers sourced from Hydrocomp (2001). In selecting the test conditions, several constraints and restrictions had to be adhered to. Firstly, the rotational speed had to be kept less than 3000rpm in order to keep the propeller blade tip Mach number below 0.3. In addition, the maximum wind tunnel speed of 20m/s could not be exceeded. Once these constraints were imposed, the primary goal of selecting the remaining test condition parameters became maximizing measurement accuracy. At low advance ratio the wind tunnel speed becomes small and consequently the accuracy of measurement reduces. The advance ratio has been limited to a minimum of 0.05 because of this. At lower advance ratio the wind tunnel speed accuracy would result in poor quality data. This minimum advance ratio therefore selects the maximum thrust and torque and has been the basis of the size selection of the torque and thrust transducers. At the other end of the advance ratio scale, the thrust and torque becomes very small and therefore difficult to measure accurately. The advance ratio has been chosen to range up to 1.0 as this represents a reasonable compromise between choosing a suitably large range of advance ratios and measuring reasonably sized variables. Based upon the minimum torque and thrust measurements the accuracy of the required thrust and torque transducers was determined. 24 2.1.3.5 Instrumentation The most challenging aspects of instrumentation have been negotiated using air bushings. If conventional bearings were used the thrust and torque sensors would have needed to be located between the propeller and the first bearing to avoid parasitic loadings imposed by the bearings. As such, their accuracy would have been severely compromised by the large bending moments imposed upon the shaft at that location. To measure the torque supplied to the propeller a proprietary rotary torque sensor has been mounted between the propeller shaft and a smaller secondary shaft (also mounted in air bushings). The selected torque meter is a 23Nm unit made by Sensor Developments Incorporated (model 01324). The secondary shaft is used to measure the propeller's thrust. Mounted to this shaft is a set of thrust bearings, which are connected to a load cell which measures the thrust produced by the propeller. To minimize the axial load transferred from the secondary shaft to the electric motor the two are coupled via a servometer type bellows coupling which has minimal axial stiffness. The chosen thrust load cell was a 20kg device manufactured by Precision Transducers Ltd (model PT2000). The torque sensor also contains a speed encoder, which is used for determining the propeller rotational speed. The wind tunnel free stream velocity is measured using a differential pressure transducer connected across the contraction section of the wind tunnel. 25 2.1.3.6 Electric Motor The selected motor and drive combination is a 15 horsepower (HP) electric motor manufactured by W E G , with a maximum running speed of 3600rpm controlled by a W E G CFW-09 variable speed drive. The motor has been sized based upon the required maximum propeller speed of 3000rpm and maximum power input of approximately 6kW (8 HP). The 15 HP unit was chosen in favour of the next smallest 10 HP unit, as it provided a substantial power safety margin at only a modest extra cost. This safety margin was necessary due to the unknown power losses in bearings and motor efficiency, as well as the unknown effect of making modifications to the propeller. 2.2 Test Rig Construction By the completion of the design phase, thirteen complete engineering drawings had been produced (refer appendix A). These drawings detailed the major constitutive parts of the test rig, as well as the arrangement of the overall assembly. The drawings were supplied to the U B C Chemical Engineering workshop which fabricated all of these parts excluding the propeller shaft and nosecone. The propeller shaft and nosecone were manufactured by Westcan Engineering and Machining. Westcan also dynamically balanced both propellers, as well as the complete propeller, nuts, washer and shaft assembly. 26 Assembly of the manufactured and proprietary components was relatively trouble free, until the alignment of the three shafts (propeller shaft, thrust transducer shaft and electric motor shaft) was inspected. Alignment of the shafts had been an important consideration during the design process, as poor alignment might lead to excessive vibration. It had been expected that alignment of the shafts would have been a trivial matter, as a result of the tight tolerances specified on the shaft mounting surfaces. However, initial inspection revealed the primary (propeller shaft) and secondary (thrust transducer shaft) were out of alignment by over 0.7mm. This was due in part to the motor mounting surface not being level, as well as the manufacturer's dimensions on the air bushings being inaccurately reported. The motor mounting surface was improved by painstaking work (approximately four hours) with a hand file. Ideally this would have been done using a milling machine, however the work piece was too large to fit under any of the available machines. Alignment of the shafts was then carried out using the reverse dial indicator method. This technique involves taking measurements with a dial indicator and magnetic base. Measurements are taken from one shaft to the other, from the top, bottom and each side of the first shaft to the second, and then back from the second shaft to the first. From these eight measurements, the exact relative positions of the two shafts, in both the horizontal and vertical plane, and in the angular and lateral directions can be determined. The vertical alignment could then be adjusted using precision shims, and the horizontal plane was dealt with using a customised jig with setscrews. The propeller shaft was bolted down rigidly as the reference, and then the motor was aligned to this. Once these 27 two outer shafts were aligned the thrust transducer shaft was then aligned by clamping it in place using rigid shaft couplings. The alignment process was very challenging, as repeatably moving the motor very small distances (a few thousands of an inch) was very difficult. Furthermore, only one plane could be aligned at a time, however it was found that movement in one plane might affect the other. Similarly, the alignment might change after the motor was bolted down, however it could only be moved whilst unbolted. Eventually the shafts were aligned to within 1/1000 of an inch, which took approximately one month to achieve. This level of alignment was chosen as it was the highest that could be reasonably achieved with the available equipment, and took little more time to achieve than a lower level of alignment, once the appropriate equipment and systems were in place. 2.3 Test Rig Commissioning The commissioning of the test rig has been the most challenging and time-consuming part of this project. Extensive failure of high quality, expensive proprietary equipment has been one of the most frustrating problems encountered. The delays associated with isolating such problems, returning and having these items repaired or replaced has also been a source of frustration. Refining the instrumentation arrangement to minimize signal noise and drift has been another significant challenge which has proved difficult to overcome. 28 In general, extensive commissioning problems are expected to stem from poor detailed design. In this case however, the benefit of hindsight has revealed few design oversights that might have made the commissioning process less arduous. 2.3.1 Equipment Problems The electric motor variable frequency drive has been very problematic. Upon delivery the unit was found not to function at all. It was then replaced by the manufacturer. The controller is a very sophisticated unit that is meant to be able to maintain a chosen speed very accurately. It also has included a function which allows the maximum torque output to be regulated. Extensive problems were encountered relating to the controller being unable to maintain a constant speed. The controller is specified to be capable of maintaining a set speed to within 0.5rpm, however, cyclic variations of up to 50rpm were commonly encountered at certain operating speeds. The manufacturers technicians provided seemingly endless suggestions to remedy this problem, all of which failed. It appears that the controller was unable to cope with controlling the combination of the large rotational inertia of the propeller, coupled through a relatively torsionally soft driveline. Some of the technicians' suggestions in fact made the controller become unstable and resulted in the motor producing sufficient torque to damage the bellows coupling. Eventually the controller was switched from 'sensorless vector mode', a supposedly very sophisticated controller mechanism, to 'variable frequency mode', which is a much simpler form of control (these two modes are the user selectable controller options of the hardware and are both open loop control systems). This resolved the issues with the controller. 29 The most expensive single component purchased as part of the test rig assembly was the rotary torque sensor, and it has also been the largest source of commissioning problems encountered. Upon installing the device it initially appeared to function properly, however, after carrying out a preliminary calibration it was found that the calibration constant quoted by the manufacturer was about twelve times lower than that determined by the preliminary calibration. Quite extensive calibration documentation was supplied with the unit, and so it was assumed for quite some time that there was some problem with the cabling or data acquisition system, and not the sensor itself. Further investigation showed that with the rotary encoder (which is integral to the torque sensor) disconnected that the sensor would work properly. This was quite a puzzling problem. Eventually it was established that one of the sensors strain gauges was erroneously grounded to the sensors shaft. This meant that when used in isolation the device would function properly, but, with the rotary encoder connected, which separately grounded the unit, the sensor malfunctioned. This explained why the manufacturer was able to calibrate the device without finding the problem. At this point the unit could then be returned to the manufacturer for repair. From initially discovering the problems with the torque sensor to finally having it repaired and returned took in excess of six weeks. As earlier mentioned, the bellows coupling was damaged due to the electric motor producing excessive torque. Two of these couplings were destroyed in the process of ironing out the motor controller issues. 30 In order to reduce instrumentation signal noise a filter was added to the data acquisition system. The unit used was an Iotech DBK18 filter card. This card was found to be very temperamental. It would often stop working properly, although usually in a particularly subtle manner. When this happened it would continue to process the signals, but do so with a very slow response time. The overall affect of this was that when a load was applied to one of the instruments, the signal would qualitatively appear to behave correctly, but the quantitative data was incorrect. This meant that the filter could stop functioning during a test, and be producing meaningless data, without any obvious indication of a problem. It was found that replacing a removable chip used for setting the cutoff frequency of the filter remedied this problem. It appeared that possibly the chip had been overheating and causing this problem. A similar problem was encountered with the thrust load cell strain gauge amplifier. This device was designed and built by the Mechanical Engineering Department instrumentation technicians, and is one of many of these units in service within the department. The device was found to stop functioning commonly. In some cases this could be fixed by simply switching its power supply off and then on, although on two occasions it was necessary to replace failed components within the device. 2 . 3 . 2 S i g n a l N o i s e a n d D r i f t The close physical proximity of the large electric motor to the array of sensitive instrumentation on the test rig has led to several confounding problems relating to signal noise and drift. 31 Preliminary testing of the instrumentation showed that the torque and thrust signals became very noisy when the electric motor was switched on. To reduce this problem the DBK18 filter card was added to the instrumentation setup. A low pass filter was implemented with a cutoff frequency of 1Hz on the thrust channel, and 0.1 Hz upon the torque. The lower cutoff frequency was chosen for the torque as it exhibited more low frequency noise than the thrust. Once the random noise on the torque and thrust signals was removed, it was found that when the motor was switched on that the signals steady state values would both change considerably. Initially this was thought to be caused by a mechanical coupling between the thrust and torque signals as the shaft rotated. Further investigation showed that with the motor mechanically de-coupled from the test rig driveline, these signals would still exhibit this behaviour. Furthermore, the digital signal from the rotary encoder within the torque sensor became sufficiently noisy to indicate that the shaft was turning when it was not. These issues presented a major hurdle to obtaining high quality experimental data, and considerable effort has been expended trying to mitigate these problems. The first step taken to try and deal with these issues was the installation of load reactors. These are large inductors placed between the motor and the motor drive. They are used to suppress high frequency noise produced by the motor drive. These made no noticeable improvement to the problem. 32 The next step involved a complete review of the instrumentation wiring, grounding and shielding arrangement. A diagram detailing this is included in appendix B. The nature of the signal noise, whereby a constant offset is observed after switching on the motor, indicated that the problem was probably related to grounding issues. Possible sources included ground loops, which could create induced currents when exposed to the large magnetic field of the motor. Also, the ground references of the instrumentation might be moving due to the large currents drawn by the motor. To deal with these issues the instrumentation grounding was carefully examined. A l l grounding connections were thoroughly cleaned (in particular paint and anodizing was scraped off of casing grounding points) and carefully connected to a single point ground. In order to do this the computer and instrumentation power supplies were all powered from a power supply with an isolated ground. Additionally the thrust load cell had to be electrically insulated from the test rig to prevent a ground loop being formed between the load cell's ground and the test rigs ground. Conversely, the torque sensor, which has its case and shaft grounded together and therefore to the entire test rig (as the torque sensor shaft must be coupled to the motor shaft, which in turn is grounded to the test rig), could not be grounded to the same grounding point as the other instrumentation. Instead it was grounded to the test rig. The shielding of the instrumentation cabling was also carefully checked. It was found that much of the shielding had been done inappropriately, and so this was remedied. The overall effect of this work was to very slightly reduce the amount by which the signals shifted when the motor was switched on. 33 B y this stage extensive work had been carried out to improve the instrumentation grounding, with little progress being made to resolve the problem. The next logical step was to improve the motor and drive grounding. This was done by running one inch diameter copper pipe-work to the motor drive, the line and load reactors and the test rig frame. Each of these was grounded to the copper pipe via heavy braided copper wire. The copper pipe was then soldered to the steel walls and structural steelwork of the laboratory to provide a very stable ground. Copper pipe was chosen for its very low electrical impedance. Grounding of this nature can generally be described as excessive. The addition of the pipe-work reduced the torque and thrust signal shift to within acceptable levels. The torque and thrust signals were found to vary roughly sinusoidally with the rotation of the shaft. This is caused by slight misalignments of the three shafts in the driveline. Misalignments of the two shafts mounted in air bushings results in the torque sensor having a bending moment applied to it. The torque sensor uses strain gauges on one point on the shaft to measure torque. The applied bending moment stays stationary whilst the strain gauges rotate. A s a result, the torque sensor interprets this as a sinusoidally varying torque. The exact mechanism causing the thrust variation is a little less clear. Possibly the bellows coupling isn't perfectly true, and as such may place a varying load upon the load cell as the shaft rotates. Alternatively, the thrust bearings may not run true, and similarly apply a varying load to the load cell with rotation. Dealing with these variations mechanically has not been possible. A t typical test speeds the low pass filter averages out these sinusoidal variations. The point at which they become 34 problematic is when obtaining the zero reference from which measurements are taken. This has been overcome by taking the zero reference at a rotational speed of lOOrpm for a period of 15 seconds. At this low speed the torque and thrust remain essentially zero, whilst their sinusoidal variation can be averaged out to provide a consistent zero reference. Additionally, a sampling period of 15 seconds has been used in taking all the test measurements to allow signal variations to be averaged out. Early in the commissioning process it was noticed that the torque and thrust signals had a tendency to drift significantly over time when the rig was running. A series of tests were run to investigate this phenomenon, with the rig being run at a set speed for up to two hours whilst thrust and torque data was collected. Analysis of this data revealed that significant drift did occur. Both the torque and thrust signals drifted in a manner consistent with some type of thermal heating process, with the drift being initially rapid before it would taper off as it asymptotically approached some equilibrium value. Through this study it was established that after half an hour running at the given test speed, the signals stabilized sufficiently to allow data to be collected. Whilst some drift did continue after this time, it could reasonably be assumed to be small and linear, and so by measuring the zero reference at the start and end of the test run, an appropriate zero reference could be determined for each test point. As such, it was intended that tests would be carried out after a half hour warm-up period. The thermal drift investigation carried out was performed without the propeller attached, just with the shafts spinning. It was later discovered that the cooling affect of the spinning propeller reduced the thermal drift to within acceptable levels without the need for the warm-up period. 35 2.3.3 Other Commissioning Issues Because of the safety issues associated with running the test rig, a failure effects, modes and criticality analysis ( F E M C A ) was carried out. This analysis is a systematic method of considering the various ways in which a piece of equipment could fail, and establishes the likelihood and potential consequences of each failure. This analysis is included in Appendix C. A s part of this analysis it was determined that the wind tunnel floor needed to be reinforced for the dynamic loadings that might be created by the running test rig. The wind tunnel floor was reinforced beneath the test rig using four jacking posts located at the corners of the test rig, as well as some additional timber supports to more widely distribute the load across the tunnel floor. Regular calibration of the thrust and torque measurement systems has been necessary to ensure the validity of the results. The thrust has been calibrated by applying known masses to the propeller shaft via a pulley and cable system. Similarly, the torque calibration was performed using a moment arm with known masses applied via a pulley and cable system. Typical calibration plots are provided in appendix D . 36 3 Experimental Results and Discussion The overall performance of the test rig has been assessed through a series of three experimental investigations. Firstly, the repeatability with which results can be produced has been analysed. This has allowed a quantitative measure of the test rig's performance in this regard to be established. Once the repeatability testing was completed, an investigation into the effect of Reynolds number variation was carried out. This involved performing essentially the same test, at a range of Reynolds numbers, to establish whether the Reynolds number significantly influences the results. Finally, a comparative analysis of two identically specified propellers was carried out. The purpose of this was to determine if the two propellers were indeed identical, and to establish how effectively the test rig could be used for such a comparative analysis. A detailed explanation and discussion of each of these investigations follows. 3.1 Repeatability A set of tests has been carried out to examine the repeatability of results produced using the test rig. Degree of repeatability was assessed at four different test speeds. At each test speed six identical tests were carried out over three days (ie: two tests per day), with complete calibration being carried out at the start of each day of testing. Previous experience with the nature of the errors encountered, indicated that if two tests carried out upon the same day were significantly different, then at least one of these data sets was of poor quality. This was done to remove gross errors that occurred on occasion, which were of the order of 25% or greater. As such, if significant discrepancies were found 37 between two such tests, then both tests would be discarded, and another set of two tests would be carried out. The test results are shown in figures 3.1, 3.2 and 3.3 for a test speed of 1500rpm. Results for the remaining test speeds of 500rpm, lOOOrpm and 2000rpm are contained in appendix E. It should be noted that for the test speed of 2000rpm data has not been obtained for the full range of advance ratio due the limitations of the wind tunnel maximum speed. The figures show the six sets of raw data points collected, an average curve fitted to these points, and error bars determined to account for the variation of these points from the fitted curve. The curve fits for these data sets have been generated by applying a least squares fit to a sixth order polynomial. A sixth order polynomial was chosen as it was of a suitably high order to produce a fit which intuitively appeared of high quality. The error bars have been sized based upon a simple statistical analysis. This analysis was done by firstly once again fitting a sixth order polynomial to each data set. These polynomials were then evaluated at a range of fifty advance ratios across the test range. The standard deviations of each of these were then determined for each advance ratio. The error bars have been sized according to the magnitude of these standard deviations. This statistical analysis is by no means rigorous. Other more formal statistical techniques were investigated, however none were found which produced more meaningful results, or appeared more appropriate than that performed. A significant weakness in this analysis is that the 38 5 Efficiency co p p p p p p p • • • • > CO CO CO CO CO 0) < CD CD CD CD CD CD ^ o> oi w Fo ->• 0) to CD ro o o • • • • • > co GO CO GO GO GO < CD CD CD CD CD CD CD —i & i—t- S w 03 CO CD polynomials fitted to each data set incur a certain amount of artificial smoothing to the data, thus reducing the magnitude of the errors calculated. It could be argued however, that the knowledge that these curves should be smooth represents addition of information to the data sets, allowing some increase in accuracy to be justified. Inspection of the figures indicates that the error bars describe the data distributions very well, at least at the qualitative level, which provides some validation to the analysis performed. One of the key features of the plots is that in general, the torque and thrust coefficient absolute errors remain reasonably constant over the advance ratio range. Conversely, the efficiency errors tend to gradually increase with increasing advance ratio. This occurs because whilst the absolute errors of the torque and thrust coefficients remain constant, their relative errors increase with advance ratio, because the parameters themselves decrease with advance ratio. Consequently, the efficiency, which is calculated based upon the torque and thrust coefficients, has an increasing error with advance ratio. This behaviour was predicted during the test rig design process and was seen as a major obstacle to generating accurate data at higher advance ratios. A secondary reason for the considerably larger error in efficiency at higher advance ratios is that in this region the efficiency changes rapidly with advance ratio. The statistical analysis used attributes all errors to the y-axis, however, in this region very small variations in advance ratio (x-axis), lead to large changes in efficiency. Consequently, the errors in this region are artificially inflated to some extent. This problem could be avoided by performing a statistical analysis which simultaneously determines the x and y-42 axis error bars. Some investigation into statistical methods for performing this analysis has been carried out, but with no success. The significant work required to perform such an analysis does not seem justified, particularly in this region of the plot, which is of lesser importance. Furthermore, comparison of the relative graph slopes and error bars for low and high advance ratios shows that the error is certainly not proportional to the slope of the graph. This indicates that this slope effect is very much a secondary phenomenon. Whilst how repeatability varies with changing advance ratio is important, possibly the key repeatability characteristic is the degree of repeatability at peak efficiency. As such, this has been tabulated for the efficiency, torque and thrust coefficients in table 3.1. Propeller Speed Efficiency Torque Coefficient Thrust Coefficient 500rpm 1.7% 1.4% 2.5% lOOOrpm 1.5% 0.6% 1.9% 1500rpm 1.3% 0.4% 1.3% 2000rpm 1.8% 2.3% 1.0% Table 3.1: Performance parameter repeatability at peak efficiency The general trend shown in this table is that the repeatability error decreases with increasing propeller rotational speed. This is a very similar effect to that seen earlier. Essentially with increasing propeller speed the torque and thrusts being measured are increasing, whilst the absolute errors on these remain reasonably constant. Consequently the relative errors are smaller at higher propeller speeds. The errors for 2000rpm do not follow this pattern. The torque coefficient error is considerably large for 2000rpm then 1500rpm, or any other slower speed for that matter. This has lead to the efficiency error behaving similarly. The exact cause of this is 43 unclear. As it only occurs with the torque coefficient and not the thrust coefficient it cannot be attributed to problems in measurement of the wind tunnel or propeller rotational speed, and as such must be caused by the torque measurement itself. Beyond this, it is difficult to comment on the source of this increased error. 3.2 Reynolds Number Dependence Matching of the Reynolds numbers between aerodynamic and hydrodynamic flow regimes has not been possible. An investigation has been carried out to determine the impact of varying Reynolds number upon the measured performance of the propeller. Complete propeller performance tests have been carried out at six different test speeds. The chosen speeds were 500rpm, lOOOrpm, 1500rpm, 2000rpm, 2400rpm and 3000rpm (2500rpm was initially chosen instead of 2400rpm, the reasons for this change are given in section 3.5). The Reynolds number is determined based upon both the propeller rotational speed and the free stream velocity (refer to section 1.1 for Reynolds number definition). The propeller speed dominates this, with the Reynolds number typically varying by less than 10% across the advance ratio range for a given propeller test speed. As such, propeller rotational speed has essentially been considered equivalent to the Reynolds number for this investigation. The data produced by the repeatability analysis has been used for test speeds 500-2000rpm. Tests were carried out at 2400rpm and 3000rpm to generate this additional 44 data. Only two runs were carried out at these additional test speeds (unlike the repeatability data which involved six runs). The Reynolds number ranges covered by each test are shown in table 3.2. As can be seen from the table, each test speed represents a distinct range of Reynolds numbers. The results of the tests are shown in figures 3.4, 3.5 and 3.6. Once again measurements at test speeds above 1500rpm do not cover the full range of advance ratio due to the limitations of the wind tunnel maximum speed. Test Speed (rpm) Reynolds Number Range 500 186,000-202,000 1000 372,000 - 406,000 1500 557,000-616,000 2000 742,000 - 798,000 2400 891,000-938,000 3000 1,115,000- 1,153,000 Table 3.2: Test Reynolds Number Ranges Using blade element theory, we can explain the behaviour of propeller torque and thrust coefficients based on knowledge of aerofoil behaviour. Figure 3.4 shows how the lift and drag generated by a blade element contributes to propeller thrust and torque. As can be seen, most of the lift generated is directed in the thrust direction, and so produces thrust. A portion of the lift is directed in the torque direction and so also contributes to the torque required to turn the propeller. In terms of the section drag, most of this is directed in the torque direction, although a small portion is in the negative thrust direction. The overall effect of this is that increasing lift coefficient will increase the propeller thrust coefficient, and slightly increase the torque coefficient. In addition, increasing drag coefficient will increase the propeller torque coefficient, and slightly decrease the thrust coefficient. 45 In very general terms, increasing Reynolds number is expected to result in an increase in lift coefficient, and decrease in drag coefficient, both of which gradually approach some asymptotic value. This implies that for these propeller tests, as the Reynolds number is increased, the thrust coefficient should also increase, the torque coefficient should decrease, and as a result the efficiency should increase. Furthermore, these changes should behave in an asymptotic manner. V R Velocity of flow relative to blade V A Propeller advance velocity n Propeller rotational speed r Section radius LQ Component of Lift in torque direction L T Component of Lift in thrust direction D Q Component of Drag in torque direction D T Component of Drag in thrust direction 0 Section pitch angle a Local section angle of attack Figure 3.4: Blade element velocity and force diagram To some extent, within experimental error, trends of this nature do exist within the data. However, certain deviations from these trends do exist. The plots shown in figures 3.5 -3.7 are (and necessarily so) cluttered, and difficult to use for detailed analysis. A set of more spatially generous plots comparing the efficiency, torque and thrust coefficients between successive test speeds is provided in appendix F. From these additional plots table 3.3 has been produced, which summarises the changes in performance characteristics between successive test speeds. Many of the changes in performance characteristics between test speeds are within the experimental error of the measurements. As such these have been categorised as no clear change. 46 Speed Change (rpm) Change in 7] Change in Kn \ Change in Kj 500-1000 1000-1500 1500-2000 2000-2400 2400-3000 Increase Increase No clear change Increase No clear change Decrease No clear change Increase Decrease Increase Increase No clear change No clear change No clear change No clear change Table 3.3: Performance parameter trends with successive speed change (roughly proportional to Re) As earlier mentioned, based on a simple blade element theory analysis, with increasing Reynolds number, the thrust coefficient is expected to increase, the torque coefficient is expected to decrease, and the efficiency is expected to increase. The results shown in table 3.3 of the thrust coefficient and efficiency display this trend (note that the no significant change result does not indicate a deviation from the theory, but instead indicates that any change that might have been noticed was too small to be considered significant, as it was within the experimental error). The torque coefficient variation does not follow this trend, as from 1500-2000rpm and 2400-3000rpm a clear increase in the coefficient can be seen. The cause of this is unknown. It is interesting to note however, that these increases in torque coefficient were not sufficient to result in an associated decrease in efficiency between these test speeds. One of the major aims of the Reynolds number dependence investigation was to determine whether tests could be performed at suitably high Reynolds numbers for the results to be Reynolds number independent. The thrust coefficient behaves very well in this regard. From 500-1000rpm it shows a significant increase, and then from 1000-1500rpm another smaller increase (within the bounds of experimental error) can be seen, and yet again from 1500-2000rpm another still smaller increase is displayed. Then from 2000-3000rpm the line stays essentially constant. This is the exact asymptotic behaviour 47 o o o C D C - i CD C J o -1 JO c CD O O CD o C D ' zs XI CD o Q. C/> Z3 C 3 C T CD - i o o 3 T3 0) 22 cn' O CD H O "5 Si c CD O o CD 3 o CD* 7J CD Q . CD C 3 cr CD O o 3 T3 fi) -i to" o that was expected. Once again the torque coefficient does not behave as desired. It displays an almost oscillatory motion with increasing Reynolds numbers, with the changes between speeds not getting obviously smaller. These results indicate that the torque coefficient does not display Reynolds number independence over the Reynolds number test range. The efficiency plots show a general trend of increasing toward an asymptotic value with increasing Reynolds number within experimental error. From 500-lOOOrpm a large increase in efficiency can be seen, and then from 1000-1500rpm a smaller increase is displayed. From 1500-2000rpm the efficiency stays essentially constant within experimental error. The next speed step, between 2000 and 2400rpm displays a significant increase, which doesn't follow the expected pattern. From 2400-3000rpm there is no change within experimental error. These changes generally resemble the asymptotic nature suggested earlier, although the significant change between 2000 and 2400rpm is a departure from the expected trend. Overall there is some agreement between the expected behaviour and that shown, however, it is not a strong agreement, suggesting that some other phenomenon is occurring here. It is conceivable that other systematic sources unaccounted for by the repeatability analysis might lead to some variation with Reynolds number. Of particular concern is the parasitic thrust and torque measured due to air resistance on exposed sections of the drive train. This includes torsional resistance in the air bushings, on the surface of the shaft, and drag on the shaft, load cell, and all other elements which will reduce the measured thrust, due to drag created by the free stream velocity. The magnitudes of these parasitic loadings will vary with wind tunnel and propeller rotational speed, and therefore with 51 Reynolds number, and hence might lead to some apparent Reynolds number dependent behaviour. Calculation of these resistance forces and torques has been carried out. Details of the calculations are contained in appendix G. The calculations were performed for the two extreme cases of 500rpm and 3000rpm, and were specifically investigated at peak efficiency. The parasitic torque was found to be 0.2% of the measured torque at 500rpm, and 0.05% at 3000rpm. These values are small enough to be considered negligible. The parasitic thrust was found to be 5% of the measured value at 500rpm, and 4.5% at 3000rpm. The main contributor to this was the drag due to the free stream on the torque sensor and the flexure rods which restrain the torque and thrust sensors from rotating. These figures were surprisingly high, so the test speed of 2000rpm was also investigated, yielding an error of 8.5%. It should be noted that these calculations were only rough analyses, with an accuracy of perhaps ±50%. The magnitude of these results is surprising. These figures are significantly greater than errors associated with the repeatability, indicating that they will have a noticeable impact upon the results. The expected effect of the three test points taken is that the thrust coefficients at peak efficiency will be reduced by 5%, 8.5% and 4.5% at the speeds of 500, 2000, and 3000rpm. An additional complicating factor is that the magnitude by which each of these points would change is not only dependent upon the propeller test speed, but also the advance ratio. A detailed quantitative analysis of this affect has not been carried out across the results manifold, as the parasitic load calculations are necessarily very approximate, due to the relatively complex flow field, and range of geometries within the driveline, downstream of the propeller. The only solid conclusion 52 that can be obtained here is that parasitic drag will have noticeably reduced the thrust coefficient (and therefore also the efficiency) results for this Reynolds number analysis test. It is ironic that whilst this is the case, the thrust coefficient and efficiency both behave in relatively expected manners irrespective of this, and that it is the torque coefficient's behaviour which is not well explained. The goal of this investigation was to find some minimum test speed at which Reynolds number independence occurred. This has not been possible, as Reynolds number independence was not found across any speed range. The expectation that Reynolds number independence would be found was based upon some incorrect assumptions. Open water tests do not generally show Reynolds number dependence. Furthermore, open water test results are not usually quoted at a specific Reynolds number. Based on this information it was assumed Reynolds number independence was assured, provided tests were carried out above the minimum recommended open water test Reynolds number of 200,000. Further investigation has shown this to be incorrect. Open water tests do not show Reynolds number dependence because the range of Reynolds number which can be tested across, is too small for it to be observed. In determining full-scale performance, semi-empirical scaling methods, such as Lerbs equivalent profile method (Lewis 1998), must be applied to account for Reynolds number dependence. Consequently, the presence of continued Reynolds number dependence within the experimental results, is not unusual, and should have been expected. 5 3 3.3 Comparative Performance Test Two identically specified test propellers were acquired as part of the larger research project. This was to allow one propeller to be modified with the earlier mentioned ducted tips, and the other propeller to be kept as a standard for comparison. As part of this thesis a comparative performance test between the two propellers has been conducted. The goal of the test was to both determine if there were significant differences between the propellers, and importantly also to evaluate the test rig as a comparative tool. The test speed was chosen as 1500rpm. This was selected based upon the experience gained from the repeatability and Reynolds number tests. The repeatability analysis had shown that the 1500rpm test speed produced the most accurate data, and was a speed at which the full range of advance ratio could be covered. Furthermore, in terms of the Reynolds number dependence, the highest available test speed was preferred, as this should at least limit the Reynolds number dependent behaviour. The data produced by the repeatability analysis was once again reused for this test. 1 During the repeatability analysis six runs were carried out on propeller A , and these form the standard against which propeller B has been compared. The test upon propeller B involved two separate runs being carried out, with the results of these being averaged and then fitted by a sixth order polynomial, in the same manner as for the Reynolds number analysis. The results are shown in figures 3.8, 3.9 and 3.10. 54 CO c CD CO CO m — o" CD' o >< o o 3 TJ CD •j cn' O D Efficiency o b o o o b o o o o o Ko o o o co o o o o o o cn o o o b o o o o e_ co cn cn O "O (0 T l CD 3 &> 2 O CD o o 3 "O fi) w o 3 i m o CD' o ro o o to' In terms of error analysis, the error bars determined by the repeatability investigation at this speed have been applied to the propeller A data. Error bars have not been added to the propeller B data, because for this comparative analysis we just wish to see if the propeller B curves lie within experimental error of the propeller A data. The plot of the efficiencies shows very little difference in the two propellers performance beyond experimental error. The peak efficiencies are both identical. It should be noted that propeller B's curve is located slightly to the left of the propeller A curve. The plots of torque and thrust coefficient both exhibit very similar behaviour. In both cases the propeller B curves are significantly lower than that of propeller A . An alternative way of describing this behaviour would be to say that both of propeller B's curves are shifted to the left of propeller A 's . Another notable feature of these plots is that at higher advance ratios the difference between the two curves gets steadily smaller. The conclusion that can be drawn from this is that the two propellers appear to perform noticeably differently. Whilst the efficiency curves are very similar, the differences between the torque and thrust coefficients indicate a real difference between the propellers. The simplest explanation of this difference is that there is a slight difference in the pitch of the two propellers. Based upon the three plots, in which there is a general trend (particularly at low advance ratios) for the propeller B curves to be positioned to the left of the propeller A curves, it would seem that propeller B probably has a slightly lower pitch than propeller A . To further investigate this behaviour, measurements of the 58 two propellers shape distributions have been made. These are discussed in the following section. 3.4 Propeller Geometry Investigation The slope distribution on the suction side of the two propellers was measured. This was to determine if the observed differences in experimental results could be seen in the propeller geometry. A roughly circumferential line was drawn across the back of each blade at 70% of the propeller's radius (for simplicity, these were in fact straight lines drawn between the 70% radius points on the leading and trailing edges of the blade). The x, y and z co-ordinates of 13 equidistant points along these lines were then measured. This data was reduced to a chord-wise measurement by summing the squares of the x and y co-ordinates, and a depth measurement, z. The results are plotted in figure 3.11. For each propeller, the data points along each blade have been plotted, and a 4 t h order polynomial has been fitted to each data set, to produce an average propeller blade section profile. The plot shows that the 'pitch angle' (ie: the angle formed by a straight line between the leading edge and trailing edge points), at this particular radius, of the two propellers are identical. It was expected that a difference would be found which would explain the measured differences in performance, however, this is not the case. Aside from the identical 'pitch angle', a slight difference between the blade sections is apparent. This is a difference in either the section camber or thickness, or possibly both. These differences are highlighted by figure 3.12, which shows the section distributions with the blade section 'pitch angle' removed. 5 9 C O c CD O o 3 O O U —* o CD CD Q CD O 3 z (mm) cn O TJ CD 9 O <D O 3 CD CD 03 V) c CD 3 <D o O 3 •o &) In' O 3 o 1 > > < < CD CD 3 3 CQ CO CD CD CD CD CB CU Q . C L CD CD O O • *"t ' .""1 Tl T) O O "O -a CD cp_ CD CD" - i CO > —* T -o o •o -a CD CD CD CD The difference is small; although the consistency and limited spread of the data points show that this is beyond experimental error (error bars have not been added as these provide little more information than the spread of the data points themselves). To provide some insight into the impact of a difference of this magnitude we shall assume (with no justification) that the difference in the profiles is entirely due to camber, and not due to thickness. From thin airfoil theory the section lift coefficient of a parabolically cambered airfoil is given as: C L = 27c (a + 2s) where: a angle of attack (radians) £ maximum camber ratio e = m / c m maximum camber c chord length At peak efficiency, the section angle of attack will be approximately 5°. This gives the following section lift coefficients for each propeller profile: C L a = 2TC (a + 2sA) = 2TC ( 5X2TC / 360 + 2x14.8 / 214) = 1.420 C L r = 2TC (a + 2eR) = 2TC ( 5X2TC / 360 + 2x15.8 / 214) = 1.477 This shows that the small difference in section profile observed produces a 4% change on the section lift coefficient. The difference in thrust coefficient seen in figure 3.9 at peak efficiency was approximately 2 .2%. These are of a similar magnitude, however, the thrust coefficient of propeller A is higher than that of B, whilst the section lift coefficient 62 of A is lower than B. Hence, the results of this analysis do not agree with the experimental results. Whilst the analysis performed did not explain the experimental results, the assumptions made regarding the attributation of profde differences entirely to camber, and the implicit assumptions associated with thin airfoil theory were not necessarily valid. The only real conclusion that can be taken from this analysis is that very small differences in propeller geometries, of the order of that measured, can lead to large differences in propeller performance. 3.5 Error Discussion The repeatability analysis completed in section 3.1 has addressed the question of how accurately the same results can be reproduced. This does not account for the systematic errors associated with the experimental equipment and procedure, which can greatly affect the results, but in a repeatable manner. The parasitic drag and torsional resistance exerted upon the drive train is a good example of this. These loadings will reduce the measured efficiency significantly, but in a repeatable manner not allowed for by the repeatability error bars used throughout the results analyses. Errors are also incurred by the general arrangement of the test rig. The relatively close proximity of the rig frame to the propeller will affect the measured performance of the propeller. The measurement of the propeller speed also incurs some unknown systematic error. The speed is set and controller by the motor controller, and measured by an independent encoder. These two measurements have been found to differ by some small amount. It is expected that the encoder measurement should be highly accurate, however, evidently there is a possibility 63 of error being incurred here. The thermal drift of sensors is another potential source of systematic error. This is because the tests were usually carried out at about the same ambient temperature, and as the test would usually take about the same amount of time to carryout, repeatable errors could easily be produced here. The magnitude of these systematic errors is very difficult to determine with any degree of confidence, and so has not been accounted for by quantitative analysis. The repeatability analysis has been used to account for random errors associated with instrumentation etc. These include the limitations of the individual sensors, including the torque, thrust and pressure sensors. Additionally, the measurement of air density, which is calculated based upon barometric pressure and air temperature is important. This is particularly the case as the air temperature was found to vary significantly during testing, as more cold air was drawn into the wind tunnel from outside. Another important error is sourced from the velocity measurement technique. This utilises a pressure transducer connected across the contraction section of the wind tunnel. Viscous effects cause this method of measurement to be not perfectly accurate. Furthermore, the velocity measurement was made approximately ten metres upstream of the test rig, which will be slightly different from the velocity at the propeller itself. Calibration also incurs some degree of error, as no calibration is ever perfect. This varied assortment of error sources cannot effectively be quantified, however, it is important that they are identified. The peak efficiency measured for the two propellers was approximately 55%. The expected peak efficiency of propellers of this type is between 65% and 70%. The source 64 of this significant discrepancy is unclear. A reasonable portion of this could be attributed to the parasitic drag mentioned in section 3.2 which would reduce the peak efficiency by up to 5% (corresponding to 2.75% of actual efficiency). The remainder could be simply due to the propellers being less efficient than expected. During testing close attention was paid to the variation of the torque, thrust and wind tunnel speed over time. It was found that at certain test points these parameters were very steady, whilst at others they varied greatly. At certain test conditions, particularly at higher advance ratios and propeller speeds the measured parameters would continue to change indefinitely. Large settling times, sampling times and extensive filtering was utilised to deal with this, however significant errors were certainly caused by this phenomenon, particularly at certain operating conditions which exhibited highly unstable characteristics. The propeller test speed of 2500rpm was an extreme case of these variations. 2500rpm was a test speed initially chosen for the Reynolds number analysis. It was found that the data produced at this speed was of too poor quality to be used. Instead tests were carried out at 2400rpm, which behaved much more reasonably. 2600rpm and 2750rpm were also tried as alternatives, and were found to behave much like the 2500rpm results. Plots comparing the results of the 2400rpm and 2500rpm are shown in figures 3.13-3.15. Inspection of the efficiency and torque coefficient plots quickly reveals that something unusual is happening with the 2500rpm data. The efficiency plot shows the 2500rpm efficiency climbs to in excess of 75%, which is unrealistically high. From the torque and 65 CO c —* CD CO co N> o o "O 3 09 13 Q. N) cn o o —i T> 3 a o" CD' o *< o o 3 cn' O =J Efficiency 0 5 m — K ^ < o CD' 3 o »< O o 3 XJ 0) V) o 3 ro o o —i "o 3 ro cn o o -i T3 3 ro II ro cn o o o o "3 "3 3 3 to' o CO o o CO c —* CD 9° Ol ro 4* o o -. XJ 3 Q. ro O l o o —* 3 I—* IT —* C cn . — . o o CD —+i g' CD' 3 cn' O 3 t o o o K T o CO o o o o o o O l o o o o o C D CO O o o O l o o CT> O O c c/> O o ro ro o O 3 0) co" o ro o o 3 CD 3 Q . IO cn o o a " D 3 ro ro O l Ji-o o o o ~-v —i 3 3 thrust coefficient plots it becomes apparent that the torque coefficient is much lower than it should be, resulting in the elevated efficiency seen. During the 2500rpm test the torque measurement was found to vary wildly, although it did not appear to go beyond full-scale. The unusually low torque coefficient is thought to be a result of this variation. It appears that what has happened, is that as the torque coefficient has varied, at certain periods it has gone beyond the full-scale measurement. However, this was not observed at the output, as the instrumentation amplifier would clip all signals above full-scale, before the filter card would basically average this now low biased signal. This is a concerning phenomenon, as not only does it exclude 2500rpm, 2600rpm and 2750rpm as useful test speeds, it also raises questions about the possibility of this happening to a lesser (and possibly un-noticed) extent at other test speeds. The exact mechanism causing these large fluctuations is not well understood. During the 2500rpm test, the torque measurement was completely unstable and would seemingly randomly vary over time. This behaviour did not occur at other test speeds (aside from 2600rpm and 2750rpm), and none of the other measured variables behaved unusually. The degree of torque variation observed would be expected to cause the propeller speed to also change, however, this did not occur. Two possible causes of this problem have been identified. The motor controller has been highly problematic. Considerable effort was required to get it to control the propeller speed accurately. It is possible, although unlikely, that at this speed whilst the controller is able to accurately maintain the required speed, it does so in a manner which results in large torque variations. Alternatively, torsional vibration natural frequencies were 69 identified within the testing speed range during the design phase. Possibly some torsional vibration is being excited leading to this variation. This would seem unlikely however, as the time scale of torsional vibration would be much smaller than that of the torque variations observed. In addition, vibration would be expected to behave cyclically, whilst the torque variation observed did not. 70 4 Conclusions and Recommendations for Future Work 4.1 Conclusions A facility was developed for the aerodynamic testing of marine propellers. This involved the complete design, fabrication, building and commissioning of a test rig suitable for this purpose. Three sets of experimental investigations were carried out on the apparatus to evaluate its ability to reliably and accurately test the performance of marine propellers. The first investigation looked at the repeatability of the results produced. It was found that this varied significantly depending upon both the rotational speed of the propeller and the advance ratio. In particular the degree of repeatability was observed to diminish with increasing advance ratio. The propeller test speed of 1500rpm produced results with the highest degree of repeatability of the four speeds tested. At peak efficiency this was quantified as ±1.3%. The dependence of the measured propeller performance upon Reynolds number has been examined. It was expected that at higher Reynolds numbers the propeller performance would become independent of Reynolds number. The peak efficiency increased by 6.5% between the Reynolds numbers of 200k-400k, by 3% from 400k-600k, decreased by 0.5% from 600k-800k, and increased again by 2.2% from 800k to 900k. The results generally showed that with increasing Reynolds number the Reynolds number dependence became smaller, however complete Reynolds number independence was not observed. 71 A comparative study of two identically specified propellers was performed to verify the degree of similarity between them, as well as to evaluate the performance of the test rig as a comparative device. The results show that whilst the two propeller efficiency curves were practically identical, their torque and thrust coefficient curves were noticeably different. Measurements of the two propellers' geometries revealed subtle differences between them. It was concluded that subtle geometry differences (on the order of that measured) were sufficient to cause the observed performance variations. Additionally, the test indicated that the apparatus was capable of detecting relatively small differences in comparative propeller testing. 4.2 Recommendations for Future Work 4.2.1 Performance Improvements to Experimental Apparatus To further refine the experimental apparatus performance, a series of enhancements could be made. To reduce the parasitic drag upon the driveline, fairings should be added to shield the torque sensor, thrust bearings, load cell and vertical flexure rods. The current method of velocity measurement is less than ideal. During testing it was apparent that the distance between the velocity measurement and test rig location incurred significant error in the measurements. It is recommended that a pitot tube connected to the existing pressure transducer, located on the centreline of the shaft just upstream of the propeller, be tried in place of the existing system. Additionally, low 72 pass filtering of the pressure transducer's signal by interfacing it to the DBK18 low-pass filter card, might improve measurement accuracy and reduce the required sampling times. Further reduction to the signal noise might be achieved by relocating the computer and data acquisition equipment to a position much closer to the test rig, such as in the tunnel itself. This would reduce the lengths of signal cables and therefore also signal noise. Consideration should be given to replacing the current torque and thrust sensors with a custom made combined device such as that used by Hordnes (1998). Such a device could most easily be inserted in place of the existing torque sensor. This would most likely remove the temperature dependent signal drift and variation of the signals with angular position of the shaft. 4.2.2 Further Research In terms of the research project, several further experimental investigations are considered to be of importance. Comparison of the aerodynamic results with some equivalent hydrodynamic testing would be very useful to analyse the overall accuracy of the testing method and equipment. The influence of the test rig frame downstream of the propeller is not well understood. Potentially this could significantly affect results. An investigation into the impact of downstream objects on the propeller performance would be very useful, particularly when comparing the aerodynamic results to those of standard open water tests. 7 3 An investigation into the performance of a ducted tip propeller should be carried out. This would involve modification of one of the existing propellers. Ideally this would lead to an experimental optimisation of the ducted tip's geometry. Performance testing of a propeller with much greater pitch than those used to date would be of interest. This is because complete testing across the propeller's operational advance ratio range could be carried out at much higher test speeds than was possible for the current propellers. This would allow additional data to be obtained at higher Reynolds numbers. 74 References Arakeri, V . H . , Sharma, S.D., Mani, K. , 1985, " A technique to delay the inception of tip vortex cavitation from marine propellers", A S M E Cavitation and Multiphase Flow Forum, pp. 28-30. Crump, S.F., 1948, "The effects of bulbous tips on the development of tip vortex cavitation on model marine propellers", Report C-99, David Taylor Naval Ship Research and Development Centre. Goodman, T.R. and Breslin, J,P,. 1980, "Feasibility study of the effectiveness of tip sails on propeller performance", Report MA-RD-940-81006, Department of Ocean Engineering, Stevens Institute of Technology. Hordnes, I. and Green S.I., 1998, "Sea trials of the ducted tip propeller", A S M E Journal of Fluids Engineering, Vol . 120, pp. 808-817. Hydrocomp, 2001, "Commercial propeller specifications", http://www.hydrocompinc.com/knowledge/publications/PropSpecs2001.pdf, webpage. Itoh, S., Ishii, N . , Tagori, T., and Ide T., 1987, "Study of the propeller with small blades on the blade tips (1 s t Report)", Journal of the Society of Naval Architects of Japan, Vol . 159, pp. 82-90. Published in Japanese with English Abstract. Johnsson C A . and Rutgersson, O., 1991, "Leading edge roughness a way to improve propeller tip vortex cavitation", Propellers and Shafting Symposium, Paper no 12. Kotb, M.A. , 1984, "Experimental investigation of 3-D turbulent free shear flow past propellers and windmills", Ph.D. Dissertation, Virginia Polytechnic Institute and State University. Lewis, Edward V. , ed., 1988, "Principles of Naval Architecture, Vol II: Resistance, Propulsion and Vibration", The Society of Naval Architects and Marine Engineers, Jersey City. Mani, K. , Sharma, S.D. and Arakeri, V . H . , 1988, "Effect on propeller blade modification and cavitation induced noise", A S M E FED Vol . 64, pp. 64-67. Platzer, G.P. and Souders, W.G., 1979, "Tip vortex cavitation delay with application to marine lifting surfaces. A literature survey", Report 79/051, David Taylor Naval Ship Research and Development Centre. Straver, M.C. , 2002, "Experimental and computational studies of a ducted tip propeller", M.A.Sc. Thesis, University of British Columbia. 75 Appendix A - Engineering Drawings 76 J J C O C O O cz C D <r~~> C D O x C O m C > —I o H- ""So? s £ i 82°= O = — C m Z d O g ^ z I " 9 r n GO oo z <z> O 1 / 1 C O M z 1 cz> < C D C O D > —I co m co c?) C Z ? C O m — i C D ^ O X C O m o —I o > 00 -I s i C D m ~ o m O m co m O m r o 3 > I I I I I cj> <j> cn 4 ^ r o r o co — J — CD r o o O O — J C o C o O O O ' 3 > o Co C O C O r o r o O — O - n 3 > ^ O —I • - O cz •^r C D 0 _ —I o < :> — • co o c© a> 3 > m 3 > C O r o c o O C o C O co O x C O m C O o H- H-8§ O = O m ? n r o m l r j Z j o rn — e>> ^005 Q 0 0 7 Z Z Lo 0 o ^ z n 1 / 1 3 2 C D N r o I O z r o C D C O U > —I C O m zxj -n C O O c o —1 3 > m 3 > C D O x co m ro o m C 5 ° Zoo 51 r- c/> z n H l> C D m ' ~ o m C O m c o m • O cz m r o C n <C5 C n C o PH c~> 3 > Z C o o r o O C O m C D O r o ro 3 > O -C O C O m <^> O O z o r o O n m —I z o C D O r o cz> O D> z — i - - O c z C D o_ -o < 3 > I — • C O | Z — I m o <z> co r o co O c r C O o~> —I zc c n O x C O m O C o C n C o C o c o ' ^ O - u n ^ 8s°= Q o = 7 p o g z ^ c ^ i H rri 6 =£ £ z z fioo m & z K o •• o ^ z • 2 en CO F3 <z> r o I O ZXJ <3> o I z cz> C o C O C O C Z 1 o — C O C O C O 5 ° Z CD |— cn m l z n s i l> CO m ~o m o m C O m O cz m Co r o <r> I C O 3 > -cr> r o <z> O -co co cz> O c z CO o 5> ool r o o c r o 3 > - E H z o 3 > <z> o> CP cz> CP m X r o cr> CP m m ~O ro C O m r> — i o H-O b o o rH O b o 73 rn ^ ^ C m on -7 r o I O CD M ZO CP z PO' 4^-ro co D Z 1 Z ^ z o m C O C O C O 3 > CP o o l > -n Zoo Si r- cn m X z n 2 0 1| m ^ Z > CD m z o ~ o m ro m c o z o m O c z z o m CP r o cz> f i l l c n u i r o <Z5 o> <o o o o o o 11 r o ~ o ~ o m CP O ~ o ro O r o r o C P O r o ~ o ro o 4 ^ C o <Z> <Z> ro -O -nl 3 > z | ro —I CP ... 0 czz ^z C D o _ -Q < 3 > — - C O c o Z v < © <t> I oo r n 7X3 3> cz> co c o C D 3> C O C O C D H-o a o = on X X X * 8§ O = d 9 z ^ ^3 m —• ^ C rn on - , 7 5 -3 Z o Z m GO C D R CD ro O 7X3 Z —I C D c r > ro 1 n c o 3 > —1 C O 1 7X3 m — 7x3 Co '<:' co O —I —1 C o ~ o 3 > 32 > -n Z 0 3 I - c o m l si s i C D m Z3D ~ o m r~> m co 7X3 m O c r : m c ? C o c r > <r> cn O —n O O o r o C D O C o r o C n ~~o 1— — m I — 3 > O r > - t ^ r e O O 7X3 4 ^ C D o ro o 3 > 2 j •—I <3? • • O cr ^— C D CJ_ —I < 3 > I — - C O I <~r> ~ n> C 5 " m zo 00 r o 3 > H-8§ O = X X h-h" O 5 h ° U m g on T I S c . . is m on Z ICO F5 <z> r o O z o c z 1 D I z z o ' cz> 3> o o 1 zo zo CO cz CO ZLZ z^ : GO o CO CO O r o GO 0 n > Z > 09 70 zi CO I 0 CD m z o ' m m r o m c o z o m O z o m c z ro r ——i — 1 Cn cz> <Z5 cz> —-1 L—-* —1 cn cn cz> O r o cn C Z 5 c p m <z ::> m —I c o m — c o O o o z ^ D> 3 > m z z : CP r o • — i O O cz : z o n> 3 > — l O — I —c 3 > co Cn <Z> 2 CO > — I i — ZO ZJZ — I ' ~H -— CO m D> m CP r o < CP m m — — O • r n O ^ ~O ~~H CD CO z o O — O — l —I o m CD O c z i o c o m r o n> O 7^ Z CD m CP m CO CO CO ZJZ CD CO 3 > 3 > O —I ZO I 1 zo r o 3 > z o r o o o z o -Pi-co o CO O O r o — I o -D> Z r o — I CP • • O cz : zxr CD C L . -< 3 > — • CO I c o z^ ; i c © 3 o> >^ cr 00 3 > —I m PO 3 > x * O Q 0 8 0 = n -U w = o = 09 ^ m m Z £ £ & Z K O ^ Q -i z • 3 0 0 'CD N C D ro O p ^ PO <3> I Z , ,t7l PO ' —I C D Co I— 00 O —I cz? PO O Co m n I > z 3 0 > r— m Z 00 d on I n 5 0 IE i l ro> cn C D - < O -o ' _ 3> ro O —I D> m ^ 1 — C ^ oop m 0 0 0 ^ r™| , M l | r***^-/ 2: > > PO <~> O O oo CD o CD m PO ~o m • n m co PO m Q PO m cz? Cn r o CD C D CD C D O -3=> ; o —I CZ? • • o cz 7*r- CD o_ -o < D> j CO j c© 3 0 ~ Q <rr> ro CP — — s : ~T> — CP ~o CZ> =0 m • 3» i — • rn ro • O oo <=> 3> ro -o m ro m - cr> —I c_n —i O cn m m = zc C3 — r-> OO —1 • 3> OO = — 3> -n S -n GO O TJ O 5* O O 3 OO Z= "O m ro -< ro ZZL — O CP 1— 3* rz: cp —I03> m i— zz: CO ro m (— O CP zc ZC oo cz i— cz ro m O O — ro — — cz —( oo ~n i— —\ z m O ~0 —1 m m zc CO O 2 s: >- —1 oo m ^ C S m oo cr> zc > n n zz m m 3> zc i— m zsi —i ro O oo ro m 1 CP —1 in cr. m m m -o co CO —1 -o O —< rc —i zc z O c D=> z ; ^ J> O —1 3: ro i— oo CO oo rn rc- -a 1  GO CO O z r o 3> ~n CD m — i i — oo —I ra -n m O rn zc m m m cz SL" "O ^ OO ro rO Z= ZZL OO cz m ro o m 1— ro ST m m oo cp (— O m ~o m O ro — OO cz m — CP oo zc -n m — CO ro z : m — ro ro o zc — oo oo 3: ro —i m —\ ~n CO cz — m — CP zc — ro ro rz: O O m o -n C70 CZ SL —1 T> —I CP cz D> oo O ro zc o ro >- o m m -n m 2T -O s: oo CP cz ro m (33 ro co - oo o oo 3s» OO ~n ro O —i <: -n ro 1 o O ~n s: o O H X < CD CP -n m j » -n 3C — m m m ro ro cz ro m 0> "O >-ro —i O rs m O zc 1 QI— cz ro ro ro cz m m i  —1 cz 3> ro cp ro m ~o -o — zc 2* O S > m O 3T O 3> n z — O s: co Z D O "O CZ m n> m rn oo zc ro s : 3D x ro — 1— —I rn —| z : — >• CO O O ro — oo s: cz O = =S O — rs: OO OO CO m oo rn —( ra rn m O ZC CO I— OO O s: > oo > -< cz CO D> ~n —1 CZ CO ro —1 ro OO ro m —I -n ro — i <rr 2: -n ZC 3> -n cc zc ro C > TJ m ro O ro — oo ro r— m 3> OO — i m > oo ro ro oo —1 5» ~n — i -< m CD m s: o zc m O m ro m =s O ~n O oo ~n -n m zc CO o o 73 rn is £ z U O 5 -3 z CD CO 0> 1 CO m z o ~o CO i — 3> Sc o £ ^ JO m z? IS 51 r- »/> m l Z f O 2 o H <?> T T I Cn i l - • S L o6 o ---oo no co m zc m >- T3 O O oo O O C O O < 3z c : cr O PD ~n ~n ~n < z > r o m > O cn i > i— <7> > m ro oo o> = > x s . ra o m m 33 r— - O O { n 0 C , c z > zc 0 O 00 w n t -3S0 S z z 2 S O 0 » 2 D I 33 ^ - a O 33 -T3 ^ 73 m ~J r>-< >-n Z o o r- t/> m l 18 I E S i CO m CO " r - O o « * TD > - n > 3> 3 0 m D> CO 0 CO — r— r = r o r o c o = 0 0 —1 0 0 x i - s C_ — - 0 ~o r^ > m — CD O a s m - n 1— 0 2 m 00 * n —1 -< z z m ~D z r O —1 —1 —1 > 3» c s - CO O =u =0 1— 50 O CO 1— CO m —1 ~H OJ == 2= z c X CO ~H O 3> - 0 0 0 m OO —1 r ~ 0 = z== = CO 1— CO O m r « r o —1 r o O n z m > W 2 CD —) m 2= m - 0 - 0 m z r - 0 S r — — m c o n c n =S 3> CZ? CO —1 m z O —1 m m m - n m z c r — =0 O CP O CD —< m i — m cz? CO m c o m c o CD 0 c o 1— —c H > r O s : > x i m —1 — c~) S 1— —< zr: z > m - <D 3> CO CZ —1 m r o m —1 ZC CC? — n — —t m O —H m c o z c O m 0 —I O x z c === ~o CD 3> _, -H O c o O =2: ZC ZC — r n z r n m rz: <-) —1 r n 1— 00 —1 m — zrr <=? O O 00 2 : 0 z s m m m m r > CZ> CO CO —1 m t7> zrr — z s m i> -1-1 0 e? 3> Z2T O r -m > CD A ^ m c z W c_n TD ° o 0 0 r o c z CZ —1 00 CO tZ3 -n —1 3» o - c ^ r— m r— O o =r 3= -D m O f>§2 g o o !S3 T £ — C m m m Z £ Q ^ s 2 o Sc 3? 0 m ^ 7 0 23 >-n Z 03 51 r- to m l Z n cjo mC S i r o >-O r — < r - 3Z Co o S O r n r o c z 0 — = 1 Ze£ —1 — e ? — O Z= Z^  == CD m —- 1— O > o ~ n —| co 3* —1 - n t ) i r - 0 — c z > CO CO r o ZC •x* m -H Z S c z cz? — CO —1 O r c < z= 3> -< O — "O C*J —- -—1 CO Co 0 c r = r o Co r— 3> , Ji. O r— r-s _< Cn 1— O r n ~o * — I S cz> z c O r > 0 O " n - n >• 1 1— 1— m c z 0 00 z c ZC 3> s z zz: 3> O — z n o cz> >- = >• m — 1— z s I— v O — —n 1— O O c z 2 : r o a — r e m r n —1 O c o - r o — 3Z CD >-68 cn co "is o "O c o 8 > •o •o CD 3 C L CD to !-+• T c 3 CD fi) ^+ o' 3 ca W O 0) o 3 fi) 3 a (/) 2! C L 3 ' CQ 0) (O 0) 3 Appendix C - Failure Effects, Modes and Criticality Analysis A failure modes, effects and criticality analysis (FMECA) has been carried out to consider the performance of the test rig in terms of safety and reliability. Each of the rigs components has been considered with its probable failure modes listed, the causes of these failures and the effect that these failures might have. The criticality (low, medium or high) has been established qualitatively based upon an assessment of safety, and damage to equipment. Failure which might create safety hazards have been assigned high criticality, ones creating damage to expensive or difficult to replace parts have been assigned medium, and those creating only minor damage or less have been given low criticality. Usually for a F M E C A probabilities are determined based upon statistical analysis of repeated testing or data obtained from previous life cycle analysis of similar equipment. Such information has not been available for this analysis, and so the failure probability has been determined based upon the assessors experience, as well as engineering calculations where necessary. The key result of the analysis is that all failure modes evaluated as having a high criticality were found to have a low probability. This is by no means a mere coincidence, but a result of a careful design process. This does not imply that a failure with high criticality cannot occur, it merely demonstrates that the risks associated with failure of the rigs components were indeed carefully considered. The analysis table follows. 91 CO Oi -*• b CO W 3 3 SL w > > CD CD c c CO CO z r z r Zi' Zl" IO <o CO Oi CO CD z> CD CO o o 3 > 03 > CO < Tl D m CD a > 3 s la I | ? o > S l S o o' 5 3= 3 sr B 3. §• 2. 5' 3 S 5 3 E M 2. 5T » a. ] 0 I 1 ! c 3 O o 3 g 03 CD Cp_ 3 : CD CO CD , sT m 3 Q. 0 ) T 3 5 *< ft §5 •g. CD o =? — . CD 0> = • 8- §• £ l i s ? 9 7 » w 2. ° o Q. 3 « ! a. c co CD 2.3 a a. •5 S o o ! l i s i n -1 -3 w sr a $ c TJ a 5 o c I 5 8 | « s 2, -a CD 5 3 3 CD i!; zi S CD = ta f 1 i. = CD a. c CD S . f s § °-8 I 3 CL 3 3 = » CD -< - I 3 • CD > * 8 o 2-R « CD o 5' 3-. i C 7 « CO 5 : 2. o CD CD • CD fl) 2>€ ? 2. S <" =f 0} CD 5>" I T 3 CD O S i s CD W . C= X =* S 2 ST §' CO . CD 5 ¥ T3 CD 0 •» a. o. 1 ° fl) 0 ) > I P I f ! 0 3. _ - w 3 » | « .if « 3 5' o 1 = ? CD oj 3 CD ° -=; w Q. c w c « < CD =f 2"' o " - ~R - ° 8 at 9>. 3' f a CD -3" 1 I s ! s ° ""• CD O-0 — 2 s 0 8 CD 01 g a-D CD 1 O oi 3" 3 • i = 8 i t CO CD » 3: CD o O CD =1. JU S Q 6T a v> 2. o -= •D CD 3. w o E l I s-CD -0 ? Q. CD O i c l o -o 5 CD 3 CD Qi Hi — O CO O Q. Q. = CD C JT CD CD a C L ^ 0 - 0 ° 5} & s § a 8 2 w « O zt. o CD -5 « 2. (B CD =1 CD 57S a a _ i l 0 ) CD =• f s. •< 5 = CD T 3 ' 1 3 3 I I £ « ' Dj T3 a. 3 3 CD ° - CD T l SU E" -. CO o a. o 01 3 CL o 0) c V) CD IP 8 o cr cr o Q. CD m 3! CD O S3 CL O > CD 3 "O CD > CD 3 0) I o' H CD 33 CQ' 5 i- E? 5 f o CD a n o- ^  Cfl •< S CO a l l s i s 5' ~ 2 co a 3 $ 9 3 ^ . CD CD W CD T J » CO CD u crt5 O CD CD 0 • CD CD S m 5 S 3 r g - C Q co cr _. o 3 ? s I CO CD S <° CO CO JS « f° CD c r z r o c? r CD 3" 3 33 CD 3 0) CD ro Fo b CO CO in W zr zr o> to as =* c 5T w O 3 CD O o 3 i S > D » I ^ 3 r H i l t to S CD CD 5' <2 o. ct 3 o> g. — co 3 i. 3 -CD 9 CD -9 ~ - CD CD " S o . 3 <P C CD CD Q CD § 5 3 c CT 3 O (D s-' (D Q) M 5 3 ? 3 111 f I I 2 » » CD CD Bi. 3 I 3 ; ° > St 3 =r 3 to =? co O Or O >< to < . Oi CD 1 I I I < ? CD CD " C L • O C CD CD CD CD CD 3se » s » 3 S. J2. CD cr to = O i < « CD "° 9 CD -j 3 S 3 § a §••? e>. I 5 § i Q. pj 3 1* 3 J - C IQ *^ co or o Z) CD <^ O . = r r CD Q. Q. Z3 CD ST IU S =5' -o CD =J. Q. o i f S- CD o w -2 to „ a & _r CD J2- 0 CD" O CD CD 3 o" j? o zr CD pij — Q) CD — aj =* 7 3 w &> CD = CD CD a §• 3 » 8 » CD CD CD CD = I — cu -0 g-i • S a g CD = sr a §• 3 CD ^ W m CD a m CD a = 8 o 1 o 5 M 3 & 8 £ 0> z+ CO CD 3" CD X CD q 8 ? | 8 § 3 <' M CO *• o to o 77 ^ 3 = 3 CD CO 8 - 1 8 i 3 5" to w <° ^ to • -s 00° m CD a m CD a o E" N " 8 o cn O =T to = ED E » s 3 w CD | S | z « ^ = o as CD I? ^ 8 m CD ! to !. x I to" « _ CO -o" S to « & =* to o CA a) CO ^ CT =j" 3 N 1.8 CD g; 8" 5 or cn s §• a i u, » CD " § 3 O c -^  CD s o a CD 0) a . o 0) c V) CD |mo a. CD m =s CD D) 3 Q. O > CD. >< CD TJ O •a CD. CD" > CD IO K •o o or co tr 3 o' 5 cr o H CD & < 80S Iff I t i . S 8 I 3 CD « CO CD CD ft •g o_ 0 =r. cr co 5"' CD 1 |. CD 3 CO 0) p f i <4 — cn 2 CO W CD CO 4^ 111 r"°. s o S =• o i 9 P g-• co j? s. ™ g" ' O S . CO o - K c± <P_ 3. 93 t j 3. (J. B q 7 5 3 tu (Q » s w « S 5 2, CD CD „ a a "- i SL °* 2. . z!. w cr o CD "S "S CD = CO CD 3 "o SL s O ~ o £ o < : 3 H CD •o S E l CD O ~" Er T) J Z • 7 CO 3. CO c S3. CD S co c r co 2 o I "O o ^ co » c c r • Q 3 CD 1~ ra 3 ' CD _ § 3 g S. § I 9 * ° I c : » 5 1 = 1 a. 3 5 ffl = n cut — a » 7 3 w ID *• P r> = CD = CD | £ 1 ? == CD -O .< CD CD — ' • C D o S 5 5? 2 1 " * = ® ^ 3 =i o < 8 § i 8 cS s I t s . o 3 If! ^ 5 ~ o DL §f O a? ? ? ! 1 CO" T=- CD ID O ? =• i CD ~ : » CD , ± . CO I 3 C : <o zz • 1 2j (5 I f ! Tl C 3 O o 3 c CD 3. o CL CD 0) 3 a o ai c 0) CD C 0) 2 2-a- =r £ Oi 5" =* CD Q 8 2. E. o 5 l CO CD 3' 3. o 2 IO s o Q. CD m 3 : CD o > 3 CD 5' CD TJ O "O CD > CD O Q. >< 3 0) o cr 0) cr 5 ca' 3J CD a> (A Appendix D - Typical Thrust and Torque Calibration Charts 95 Voltage CD CD • r. < => 2. SP nt =; CQ ^ CD < o I—*-Q> CQ CD Voltage -f=> CO b cn o o o o CO ro o cn o o o o IV) b o o cn o o o o o o cn o o o o o CD ~D ->• M CD II N> " CD p r: CD CD CD X CO I 1 0 O •^ 1 • CD ro co *>. oo CO I J ro as oo H «< "O O -H O CD O cu 6" 3 o" r. < => o =; CQ CD < o »—I-D3 CQ CD Appendix E - Repeatability Results at 500,1000 and 2000rpm 98 Efficiency o o o o o o o o o o O b b ro ro co CO cn o cn o cn o cn o cn o cn o o o o o o o o o o o o o o o o o ro -o o CO CD O -o o ro o o Kn O CO o XI c CD O o CD ^ * ^ « o CD" 33 CD "O CD Oi Oi «< fi) o o o -1 T3 3 K T O CD —i O Si C CD o O CD O CD" 3 3] CD •n CD 0) 03 CT «< 0) IS5 O o o "D 3 Appendix F - Reynolds Number Dependence Comparison Plots between Successive Test Speeds 108 cn o o Efficiency p p p o p o p Efficiency p p p p p p o o ro o o o o o "3 •3 3 3 Efficiency o bo o o o CD o o ro ro 4^ o Efficiency KQ p o p p p p p o o o o b ro o o C O o o b o o b cn o o o o b o o b oo o o CD o .O C CD o O CD o CD' 3 s ® 9 •< ro 3 o o II 3 <T CD -t o O 3 T 3 0) w o 1 : o cn o o o o •3 T3 3 3 KQ o o o o o o o o o o b b b b b b b b b b o — L ro co cn CD CO CD o o o o o o o o o o O KQ o o o o o o o o o o b b b b b b b b b b o ro co cn oo co o o o o o o o o o o o o o o 03 o o o ro o o o o c c/> o O ro o o CD o o r o bo o o O CD =1 . i—*• O 33 0 CD © «< 1 3 cn 2_ © CO o o o c 3 CT CD - i O O 3 "D 0) o o b o o o b o o o o ro o o KT o co o o o o o cn o o CD o o o ro o o o b o o o CO o o o o o ro o o 1 I GO O o CD —h o CD' c n jj © CD i i o a O 00 M or CD —i O o 3 T3 0) -t Lo" o ro o cn o o o p "3 3 3 3 Appendix G - Calculation of parasitic thrust and torque Propeller Shaft Torsional Resistance Calculation Calculations of the resistance torque produced by the three large air bushings, and by the atmosphere surrounding the shaft are performed below. The extreme shaft speeds of 500rpm and 3000rpm have been considered Large Air Bushings: x = p co r /1 Q = x A r ^>Q = 2 7 i c o p B r 3 / t B = 3.5" 0 3.008" 500rpm: 3000rpm: Shaft Surface: Q = 289.8 x 10"6 Nm Q = 1739 x IO"6 Nm Air Bushing Housing Air Gap Shaft Q = x A r x = 2 p co (assuming laminar flow for simplicity) A = 27r rL L = 1.2m ,co ^ Q = 40 7 t r 2 L c o p (adding a 10 x factor of safety for turbulent flow) 500rpm: Q = 1 5 0 x l 0 " 6 N m 3000rpm: Q = 900 x 10"6 Nm Total Torsional Resistance: 500rpm: Q T = 3 x289.8 x 10-6 + 150 x 10-6 = l.ONram (factor of 3 added for 3 air bushings) 3000rpm: Q T = 3 X1739 x IO"6 + 900 x IO"6 = 6.1 Nmm 124 Consider Relative Error Incurred: At maximum efficiency the measured torques are: 500rpm: Q = 0.347 Nm 3000rpm: Q=11.8Nm Relative Errors incurred: 500rpm: 1.0 x 10~3 / 0.347 = 0.2% 3000rpm: 6.1 x 10"3 / 11.8 = 0.05% Driveline Drag Calculation The drag on various elements of the driveline, including the shaft, torque sensor and flexure rods has been calculated. The extreme shaft speeds of 500 and 3000rpm have each been considered, as has the additional speed of 2000rpm. Free Stream Velocities: Measured Thrusts: 500rpm: Uoo = 3.4 m/s 500rpm: T = 2.4 N 3000rpm: Uoo = 19.6 m/s 3000rpm: T = 100.6 N 2000rpm: Uoo = 15.7 m/s 2000rpm: T = 33.4 N Propeller Shaft: F = Vi p Uo,2 C D A (assume shaft surface behaves as a rolled up flat plat) A = 27 irL CD = 0.01 (based upon flat plate at this Re number) 500rpm: F = 0.55 x 10"3 N 3000rpm: F = 1 8 . 4 x l O " 3 N 2000rpm: F = 1 1 . 8 x l O " 3 N 125 Torque Sensor: F = >/2 p Uoo2 C D A A = w x h C D = 2.0 (bluff body drag coefficient) 500rpm: F = 0.07 I N 3000rpm: F = 2.36 N 2000rpm: F - 1 . 5 N Flexure Rods: F = 2 x Vi p U„o2 C D A (factor of 2 for 2 rods) A = L x 0 Ua 0 6mm (drag coefficient on cylinder at this Re) Q C D = 2.0 500rpm: F = 0.06 N 3000rpm: F = 2 .1N 2000rpm: F - 1 . 3 3 N Total Driveline Drag: 500rpm: F T = 0.13N 3000rpm: F T = 4.48 N 2000rpm: F T = 2.84 N Relative Error Incurred: 500rpm: 0.13/2.4 = 5% 3000rpm: 4.48 / 100.6 = 4.45% 2000rpm: 2.84 / 33.4 = 8.5% Q L=0.75m 126 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0080786/manifest

Comment

Related Items