MAGLIFTER TRADEOFF STUDY AND SUBSCALE SYSTEM DEMONSTRATIONS

Transcription

1 NAS MAGLIFTER TRADEOFF STUDY AND SUBSCALE SYSTEM DEMONSTRATIONS J. Dill D. Meeker Foster-Miller, Inc. 350 Second Avenue Waltham, MA December 2000 Period Covered: 11/4/97 1/14/00 Approved Final Report Contract Number: NAS Contract Amount: $545, Competitively Awarded COTR: J. Howell/PS05 Prepared for: National Aeronautics & Space Administration George C. Marshall Space Flight Center Marshall Space Flight Center, AL 35812

2 Report No: NAS MAGLIFTER TRADEOFF STUDY AND SUBSCALE SYSTEM DEMONSTRATIONS J. Dill D. Meeker Foster-Miller, Inc. 350 Second Avenue Waltham, MA December 2000 Period Covered: 11/4/97 1/14/00 Approved Final Report Contract Number: NAS Contract Amount: $545, Competitively Awarded COTR: J. Howell/PS05 Prepared for: National Aeronautics & Space Administration George C. Marshall Space Flight Center Marshall Space Flight Center, AL 35812

3 Limited Rights Notice (June 1987) (As Modified by NASA FAR Supplement ) These data are submitted with limited rights under Government Contract Number NAS These data may be reproduced and used by the Government with the express limitation that they will not, without written permission of the Contractor, be used for purposes of manufacture nor disclosed outside the Government; except that the Government may disclose these data outside the Government for the following purposes, if any, provided that the Government makes such disclosure subject to prohibition against further use and disclosure. This Notice shall be marked on any reproduction of these data, in whole or in part. EXPIRATION: December 2005

4 CONTENTS Section Page 1. DEFINITION OF REQUIREMENTS Definition of System Requirements for Launching Vehicles Mass of the Shuttle Aerodynamic Drag Levitation Drag Motor Efficiency Transmission Line Losses Altitude of Launch Total Power, Energy and Force Comparison to Navy Aircraft Launcher Requirements TRADE-OFF STUDY SUMMARY AND IMPLICATIONS FOR FUTURE NASA SYSTEM Suspension System Air Bearings Wheels Slippers Passive Magnetic Suspension Active Magnetic Suspension Summary of Suspension Options Candidate Magnetic Suspension Options Propulsion Air Core Linear Synchronous Motor (LSM) Iron Core Linear Synchronous Motor (LSM) Linear Induction Motor (LIM) Summary of Propulsion System Trade-Offs Launch Carrier Vehicle Magnetics Considerations Energy Storage and Power Generation Flywheel Energy Storage Steam and Compressed Air Storage Superconducting Magnetic Energy Storage (SMES) Ultracapacitors Controls Conceptual Design for Full-Scale Systems v

5 Section Page 2.7 Relevance of Navy Catapult Interests to NASA Maglifter and Bantam Launcher Applications Navy Catapult Machine Topology Energy Storage Scheme Power Electronics Design Conclusions Electromagnetic Space Launch Assist System Development Strategy CONCEPTUAL DESIGN EXPERIMENTS SUBSCALE DEMONSTRATION SYSTEM System Design Initial Decisions and Design Constraints Configuration Lift or Levitation Coil Analysis Propulsion Subscale Test Rig Description Demonstration Testing REFERENCES APPENDIX A - MAGNET SYSTEM DESIGN OPTIONS AND SCALING APPENDIX B - HIGH SPEED WHEELS APPENDIX C - FLYWHEEL-GENERATOR SPECIFICATIONS vi

6 ILLUSTRATIONS Figure Page 1. Launch speed dependence of performance requirements Permanent magnet versus s.c. coil MMF scaling Pulse power capability of Westinghouse generator Overall launcher system controls Full-scale EARS arrestor Side view of rotary rig for evaluating drive coils and magnets Side view of rotary test rig Measured and predicted back-emf from rotating test rig Crumax 3714 (37 MGOe) magnet Maglifter scale model configuration Magnet about to pass figure-8 connected lift coil Displaced magnet passing across lift coil Stiffness coefficient versus velocity Damping coefficient versus speed Drag versus speed Lift coil geometry Top view of best drive configuration Propulsion coil geometry Drive structure, top view Decomposition of magnet MMF into a continuous current density Drive current and fundamental sinusoidal component D finite element solution for the propulsion Wire-frame and shaded representations of Amperes model Force/current parameter via Amperes for 1 coil, 1 magnet pair Current-to-force for an entire phase Current-to-force for a three-coil phase block Subscale Maglifter system Closer view of test vehicle and drive section of track Close-up of sensor system Optical sensor boards along the side of the track Braking section of track Drive coil sections Nine drive pulse width modulation power supplies Control board vii

7 Figure Page 35. Switching transistor block Test 1 - complete pass Test 2 - propulsion section Test 3 - initial levitation viii

8 TABLES Table Page 1. Trade-off study parameters Effect of altitude increase on energy requirements for system Total system requirements Comparison of the NASA and Navy system requirements Suspension technique trade-off Ratings of candidate motor designs versus requirements criteria Comparison of energy and power requirements with storage system capabilities Candidate Maglifter and Bantam system architectures Key NASA and Navy system performance requirements Key system element design recommendations Maglifter and subscale system specifications Lift coil properties Velocity versus distance along drive section of track ix

9 1. DEFINITION OF REQUIREMENTS Trade-off studies have been conducted to develop a conceptual design for the major elements of the full-scale Maglifter system. While these studies were planned to continue throughout the second year of the program as originally proposed, during the year of evaluation that has been completed, basic concepts have been developed for all of the key system elements. Further refining of the system concept would be needed for the broader range of possible NASA interests, but the system concepts that are defined here are a good first approximation to a workable system. The details of the system, particularly in the energy storage area may change as technology advances, so where possible the impacts of new technology on the system concept will also be discussed. This report is meant to summarize the trade-off studies and to describe the concept for the Maglifter system as seen today. Because issues have been raised about Bantam as a predecessor to a full-scale Maglifter, discussions of the requirements, and possible design differences in a linear motor launcher for that system are included in this report. A brief discussion will also be included of Navy aircraft launcher requirements to define where those stand relative to the NASA requirements. In those places where specific technologies have been shown to be the best approach they will be described. In other cases where specific technologies have been rejected, reasons will be given for their rejection. Finally, in the places where further studies are still needed, the primary issues and candidates will be discussed. The key elements of the Maglifter system which have been studied by Foster-Miller include the following: Linear motor propulsion system. Suspension system. Total power requirements. Energy storage. Controls. While the launch vehicle and its aerodynamics are an important part of the system, as is the vehicle carrier that remains on the track after launch, their designs are not included in the current study in detail. The goals of this program have been to: Define the basic requirements for motor force, power and energy storage for a Maglifter launch system. 1

10 Develop design concepts for the key elements of the system that will scale well to the full-size system. Demonstrate the key propulsion and suspension technologies at sub-scale and then at increasingly larger scales to show feasibility of the basic system concept. 1.1 Definition of System Requirements for Launching Vehicles In order to define the trade-off system requirements, it is useful to look at a few simple defining relationships based on the size and launch speed of the vehicle. Because of their relevance to NASA, two launch vehicle sizes are discussed in this report. The primary emphasis of the report is on Maglifter, a shuttle replacement size vehicle. This was the emphasis of the study based on the contract. The second vehicle system to be discussed is the Bantam system that is a smaller, 100,000 lb class launch vehicle, because of its relevance to NASA as a candidate for electromagnetic launch in the nearer term. Table 1 defines the relationships of the vehicle size and launch speed to critical design requirements that help define the system characteristics. Basically, the vehicle weight, launch speed and acceleration level can be set and then other characteristics such as power, energy and track length required to achieve these conditions can be calculated. The numbers given in Table 1 are those given by NASA as requirements for the Maglifter and Bantam systems. Two different acceleration levels are included in the table for the full-scale Maglifter to demonstrate how track length, motor force and power requirements change with acceleration level. Once these quantities have been calculated, they can be used to define the launch motor, energy storage system, total power requirements and other properties of the total system. In this way, a picture can be built up of candidate systems. Later in this report, the same quantities will Table 1. Trade-off study parameters Maglifter Maglifter Bantam Parameter (3g) (1g) (3g) M=mass(lb) 1,000,000 1,000, ,000 M=mass(kg) 454, ,000 45,400 a=acceleration(g s) v=velocity(m/sec) v=velocity(mph) l=length(m)=(v 2 )/2a F=Ma(N) 1.36E E E+06 P=Fv(W) 3.68E E E+08 U=1/2Mv 2 =Fl(J) 1.65E E E+09 t=sqrt(2l/a) (sec)

11 be used to define the subscale system which has been built at Foster-Miller to demonstrate the basic magnetic levitation and linear motor launch technologies. By defining the mass or weight of the launch vehicle, the desired maximum acceleration and the final launch velocity, it is possible to define other important parameters for the launcher. The quantities in the table can be used to define the various systems requirements such as the launch motor force, the power that the motor will require, and the energy that must be stored in the energy storage system in order to achieve a launch. These quantities are related by the simple equations shown in the table. In order to size the complete system, however, there are a number of other factors that must be included. The quantities included in Table 1 are the force, power and energy that must be input to the launched vehicle to achieve successful lift-off. The delivered useful energy requirement for the Maglifter system as shown in Table 1 is 16.5 GJ. This is the kinetic energy of the launched vehicle mass (10 6 lb) going at the exit speed of Mach 0.8 (600 mph at sea level). The motor force to input this energy at an acceleration rate of 3 g s is 13.6 MN. The power that must be delivered at the end of the track is 3.68 GW. The motor total force and energy requirements will be higher than this because of a number of factors including the following: Mass of the shuttle (supports the payload on the track and contains the magnets for propulsion and levitation in synchronous motor designs). Aerodynamic drag. Levitation system drag. Motor efficiency. Transmission line losses. Altitude of launch. The following sections provide brief summaries of these other factors and their impact on the motor sizing. Since factors like aerodynamic and levitation are speed dependent, the calculations in Table 1 have been redone as plots in Figure 1 to show how the different factors leading to an estimate of total system requirements depend on launch speed Mass of the Shuttle The shuttle is the device that supports the vehicle being launched (Bantam or Maglifter) on the linear motor track. The weight of the shuttle will be influenced significantly by the motor design selected and as such is one of the trade-off issues that must be considered in the design. For a number of reasons that will be discussed later, Foster-Miller feels that a linear synchronous 3

12 a) b) Figure 1. Launch speed dependence of performance requirements 4

13 c) d) Figure 1. Launch speed dependence of performance requirements (continued) 5

14 motor (LSM) is the preferred approach for a Maglifter type of application. The alternative would be a linear induction motor (LIM). Selection of a linear synchronous motor (LSM) has a significant impact on the shuttle weight. In a synchronous motor of the design recommended here, it is necessary to carry magnets on the moving shuttle. This leads to a significantly heavier carrier vehicle than would be required in an induction motor design where forces are produced by inducing Eddy currents in the shuttle. The weight impact on the shuttle can be controlled somewhat in a LSM by using superconducting magnets to produce the fields as will be discussed later under design details. However, the shuttle will still be heavier than with a LIM. The intent of the current program was to perform a more detailed design of the shuttle during the second year of the effort so information is not available at this time. However, while that detailed design of the shuttle is not available, it is possible to provide what is felt to be a reasonable order of magnitude estimate of the weight which will be accurate enough for the accuracy of the current trade-off study. For either size vehicle in Table 1, a shuttle mass or weight which is roughly equal to the dry weight of the launched vehicle is a reasonable estimate. As shown in (1), a vehicle dry weight of 10 percent of the fueled vehicle weight is reasonable for a Maglifter size vehicle. Using a shuttle weight comparable to this seems reasonable since the weight at this size will primarily be the structural weight required to support the launch vehicle. Assuming that superconducting magnets are used to provide the fields on the shuttle, only a fraction of the total shutttle weight will be required for the magnets. If permanent magnets were used instead of superconducting magnets, the total magnet weight would be much higher and consequently, the shuttle weight would be higher. The details of the trade-off in magnet selection are given in Appendix A. To be conservative in estimates, a value of 20 percent of launched vehicle weight has been assumed in the trade-off study for the shuttle weight. If the mass of the shuttle (M shut ) is assumed to be 20 percent of the launched mass (M payload ), or 200,000 lb, the combination of shuttle and payload, which can be given the name transporter, has mass of 1.2 million lb (M trans ). The power required to accelerate the transporter is shown in Figure 1 as P accel. The addition of the mass of the shuttle to the mass that must be accelerated increases the motor force requirement to 16.3 MN, the power at the end of the track increases to 4.41 GW, and the energy requirement to 19.8 GJ Aerodynamic Drag Aerodynamic drag will be a major consideration for the detailed system design as will lift from the wing structures that are expected to be part of the launch vehicle design. For our purposes, because of the scope of the study, only a rough estimate was made of aerodynamic drag by assuming a frontal area for the shuttle and launch vehicle. These were added to get the total frontal area for the transporter (vehicle + shuttle) (A trans ). Assuming a drag coefficient (C d ) 6

15 of 0.3, the following equation was used to determine aerodynamic power (P aero ) loss shown in Figure 1. Paero = 1 ρair Cd V 2 3 As can be seen from the figure, aerodynamic drag, even at 0.8 Mach, doesn t add significantly to motor force or energy requirements Levitation Drag To reduce system cost and complexity, the primary candidate for transporter support is passive magnetic levitation with wheels used at low speeds. This approach eliminates the need for separate power electronics and controls for levitation that can be a significant total system cost savings. In this design, the motor must make up the force and energy required to levitate the transporter. In providing this energy, there is an accompanying drag on the linear motor propulsion system. As will be shown in the section on subscale motor design, the levitation system drag is a function of the levitation system design details such as coil size and pole pitch, coil resistance and inductance, and vehicle speed. In a properly designed levitation system, the drag goes through a peak at relatively low speeds and then declines as speed increases farther. In an actual system, the levitation coils would only be included in the track once the drag has dropped below its peak value. Based on estimates made during the subscale motor design, and information from references (2), it should be possible to achieve levitation forces that are a minimum of 10 times the drag caused by the levitation coils. In a properly designed system, values of lift to drag can be as high as 80 or more making the total drag no more than a few percent of the motor force. In general, the magnetic drag will be small and comparable to the aerodynamic drag in magnitude for the speeds where it will be used. As such it doesn t have a significant impact on the motor design Motor Efficiency A motor efficiency of 90 percent was assumed. This includes the I 2 R losses in the shuttle (moving part) and stator (track side), as well as losses in the power electronics required for local commutation. The lost power (Q motor ) is determined by the following equation: Q motor =(P accel +P aero +P drag )(1-η motor ) The motor efficiency adds to the power and energy system requirements, but is implicit in the motor force estimates so that value is unaffected. As can be seen from Figure 1, this is a relatively small effect. 7

16 1.1.5 Transmission Line Losses If a single power source is located at the end of the track, the transmission line near the end of the track must transmit the maximum power, and the full energy for the launch. In contrast, the transmission line at the beginning of the track sees the minimum power for the least amount of time. Thus, the optimal transmission line should taper from thin at the beginning to thick at the end of the track. In an optimal design with modular power sources, it will probably be best to distribute the power sources along the track and minimize the transmission losses. To determine the dimensions of a copper DC transmission bus, it was first assumed that the voltage would be a constant (V line = 20 kv), and the current carrying capacity limited by the temperature rise of the line ( T = 40 C). For every segment of transmission line of length L the electrical heating causes a temperature rise: 2 I R dt = m c T Using the following equations for the resistance and mass of the conductor segment, the cross-sectional area of each segment can be found: R L = ρ e A m = A L ρ d ρel ρd A I 2 dt = A L c T A = ρ ρ e d I c 2 dt T where: ρ e = resistivity ρ d = density c = specific heat of copper Starting from the beginning of the track, the current I is determined by dividing the power by the line voltage (P motor /V line ). The model integrates I 2 as it works its way down the track; all other variables are known. Using a buss bar with a cross section like that shown in Figure 1 will minimize the cost of the buss bar while keeping the heating under control. 8

17 1.1.6 Altitude of Launch Since some of the systems scenarios for Maglifter consider launching the vehicle on a track that both accelerates the vehicle and raises it to an altitude on the order of 10,000 ft, the potential energy associated with this increase in altitude is also included in Table 2. Of all of the effects discussed in this section, increasing the launch altitude has the biggest effect on system design. If it is assumed that the track is run at a 45 deg elevation angle and the altitude increase is equal to the total length of a horizontal track, the energy associated with altitude increase is one-third of the kinetic energy required to accelerate the vehicle. This is easy to understand since the energy in both cases is related to the force applied times the distance it is applied over. In the case of the kinetic energy, the force to accelerate at 3 g s is just three times the weight of the vehicle. In the case of the potential energy to raise the vehicle in altitude, the force is just equal to the weight of the vehicle (since the acceleration is 1g). An altitude change equal to the horizontal track length is used just to show that the potential energy is on the same order of magnitude as the kinetic energy. Because of time and budget constraints, it has not been possible to do a detailed trade-off on launch altitude and its impact on system design. Clearly the more of the initial energy of launch that can be provided by the ground based EM launcher, the smaller the fuel and propulsion requirements will be for the launched vehicle. These issues will have to be considered as part of any detailed system design. In all probability, because of basing and safety considerations, initial implementations of EM launchers for rockets will be built near to the coast for launch over water at sites such as Cape Canaveral in Florida. Such basing will preclude launching at any altitude significantly above sea level. Because of this consideration, the emphasis in this trade-off study has been on system design for sea level launch. Even though the added energy requirements for launch at altitude significantly increases the system energy, at the levels being considered, factors of between 1.2 and 2 probably don t change trade-offs significantly. It is only with changes like those between a Bantam size and a full-size Maglifter, where launch vehicle mass is different by an order of magnitude, that significant changes are seen in system trade-offs. Table 2. Effect of altitude increase on energy requirements for system Maglifter Maglifter Parameter (3g) (1g) Bantam M=mass(lb) w/carrier 1,200,000 1,200, ,000 M=mass(kg) w/carrier 544, ,300 54,400 h=altitude of Launch(m) PE=Mgh 6.61E E E+08 U=1/2Mv2=Fl(J) 1.96E E E+09 Total Energy=U+PE 2.62E E E+09 9

18 1.2 Total Power, Energy and Force The total power required of the generator (P gen ) is the sum of the acceleration, aerodynamic, suspension drag, motor loss, and transmission loss power contributions. The power that must be delivered to the linear motor is the sum of the above components: P gen = P accel + P aero + P drag + Q motor Graphs of power versus time, power versus distance, and energy required versus time are shown in Figure 1. Also shown in Figure 1 is the cross-sectional area of one transmission line as a function of distance along the track (two lines are required for a DC bus). As discussed in the various subsections of subsection 1.1, the power and force requirements are basically determined by the mass of the launch vehicle and the shuttle that it rides on. The aerodynamic drag and the magnetic drag from the suspension may add another 10 percent to the power requirements. The motor efficiency probably adds approximately another 10 percent also. Table 3 summarizes the approximate motor force, power and energy requirements for launch at sea level and a speed of 0.8 Mach. The values in this table were used in considering system design options. As can be seen from Table 3, the length of the track increases as the square of the launch velocity for a given acceleration. If higher acceleration rates can be used for unmanned vehicles, than the 3 g s specified in the table, the length of the track will decrease in simple proportion to the increase in acceleration. The force the motor must produce is defined primarily by the mass of the vehicle and its carrier. As noted above, while important to the final detailed design, for an order of magnitude analysis like that being done here, the aerodynamic and magnetic drag forces are not significant at the speeds being considered when compared to the basic system force requirements. These Table 3. Total system requirements Maglifter Maglifter Parameter (3g) (1g) Bantam M=mass(lb) w/carrier 1,200,000 1,200, ,000 M=mass(kg) w/carrier 544, ,300 54,400 a=acceleration(g s) v=velocity(m/sec) v=velocity(mph) l=length(m)=(v2)/2a F=Ma(N) 1.96E E E+06 P=Fv(W) 5.29E E E+08 U=1/2Mv2=Fl(J) 2.38E E E+09 t=sqrt(2l/a) (sec)

19 issues will become much more important in the design when specific vehicle and final motor design details are being developed. For the same acceleration rates, the force required to launch Bantam will be roughly an order of magnitude lower than that required to launch the full-scale Maglifter because of the order of magnitude lower mass of the system. Since the power required to run the system is also proportional to the mass and acceleration of the launch vehicle and carrier, for the same launch velocity (taken as 0.8 Mach in the table), the power required for the full-scale Maglifter is an order of magnitude higher than that for the Bantam system. The energy required to be stored in the storage system, is also an order of magnitude higher for the full-scale system than for the Bantam scale system. This difference in required energy storage level may mean that the best energy storage options may be different for these two systems. Having defined the motor force required, the power requirements to run the motor, and the energy required for the launch, it is now possible to consider the different options for the key system components. Also included in the design options discussions are considerations for the system suspension of the launch vehicle carrier and the controls required for running the motor. The controls will be a major factor in the selection of the motor design and the cost of the system although they will be much the same for both the Bantam and the full-scale system. Controls won t have to be scaled up as the system is scaled up. This is one area in particular that work completed at the Bantam system size level will be directly applicable to the full-scale system. A further trade-off refinement that will play into any detailed system design is that between the propulsion energy provided by the launcher and that on the vehicle. In this study the vehicle weight has been assumed and other elements of the system have been sized accordingly. A maximum speed to be imparted to the vehicle of 0.8 Mach has been assumed. As part of a detailed design, one would want to optimize the energy provided by the launcher versus the vehicle size and fuel carrying capacity to arrive at an optimized overall system design. Aerodynamic factors will also have to be considered in much more detail than the rough estimates made here as part of a detailed system design. Again, to the precision with which the trade-offs have been considered here, these factors are expected to be second order effects that won t significantly change the basic designs of the different key elements of the system. 1.3 Comparison to Navy Aircraft Launcher Requirements Like NASA, the Navy has been considering the use of linear motor technology for vehicle launching for some time. The purpose of this section is to compare the basic Navy requirements and likely system design issues to those for NASA applications. Table 4 shows the Bantam and full-scale NASA systems requirements along with the Navy requirements. Because the Navy requirements include a range of vehicle weights and speeds, only approximate order of magnitude values are given for the three key quantities (force, power and energy), which will be needed to define the launcher systems. The impact of these issues on design selection for key system elements will be discussed in later sections. 11

20 Table 4. Comparison of the NASA and Navy system requirements Parameter Maglifter Bantam Navy Requirement M=mass(lb) w/carrier 1,200, ,000 M=mass(kg) w/carrier 544,300 54,400 a=acceleration(g s) 3 3 v=velocity(m/sec) v=velocity(mph) l=length(m)=(v2)/2a F=Ma(N) 1.96E E+06 1E+06 P=Fv(W) 5.29E E+08 1E+08 U=1/2Mv2=Fl(J) 2.38E E+09 1E+08 t=sqrt(2l/a) (sec) As can be seen in Table 4, the motor force requirements for a Navy system are comparable to NASA Bantam requirements. This similarity means that both systems will need to address many of the same issues in terms of motor coil size, forces, and current levels. The power requirements are also similar to Bantam indicating that power electronics devices capable of powering one system can be used to power the other. Just as with respect to Bantam, however, full-scale Maglifter has an order of magnitude higher requirement for these critical parameters than a Navy system. There are a number of significant differences in the Navy requirements that will lead to some differences in the system design. The Navy in general, is interested in lighter vehicles launched at lower speeds than NASA requires. These two factors lead to an order of magnitude lower total energy requirement even when compared to Bantam. This leads to significantly lower requirements for energy storage and total power supply to run the system. In addition to these basic requirements, there are a number of specific Navy requirements that influence design that aren t issues for NASA. These include limited motor length because of the limited deck space on an aircraft carrier, strict system weight restrictions, and higher repetition rates. The motor length is set by the length of the aircraft carrier flight deck where limited space is available. This leads to a higher acceleration requirement which in turn increases the force, power and energy requirements to make them closer to the Bantam requirements than they would be if the Navy could keep to a 3g acceleration level. There is a further length consideration for the Navy system that requires the armature to be stopped in a very short distance again because of available deck space. As will be discussed later, this favors an induction motor design with a light armature for the Navy application. The major issue with higher repetition rates in the Navy application is a thermal one. In either Bantam or Maglifter, there are expected to be hours or days between launches. This type 12

21 of cycle time means that heating from repetitive cycling of the motor is not an issue. A motor can therefore be designed in such a way that as long as it stays within thermal bounds on a single launch, the thermal management will be acceptable. In contrast, the Navy requires launch intervals on the order of minutes which means that there is a high probability that the motor will not have cooled completely from the previous launch when it is required to operate for the next launch. This will require a much more detailed thermal design and may also require a more substantial thermal management system than will be required for NASA applications. Another difference in the Navy requirement is the fact that the aircraft are attached to the launch motor armature directly eliminating the need for a separate suspension system. This simplifies the system design and eliminates one major consideration in the trade-off studies for the Navy application. There is also no track per se in the Navy system while a track appears to be needed for the NASA systems because of the high launch speed that makes use of a wheeled carrier vehicle difficult. The final difference is that the Navy system requires tight control of velocity profile and final launch velocity when compared to the NASA system. To limit aircraft stresses, the Navy is seeking a system with tight control on acceleration and velocity profile. There are also tight requirements on the final speed that must be reached. This means that the motor must have tight control on these parameters. As will be seen in the system trade-off discussions, this may affect the motor design preference and some of the other details of the system design. 13

22 2. TRADE-OFF STUDY SUMMARY AND IMPLICATIONS FOR FUTURE NASA SYSTEM In developing a system design to meet the requirements defined in Section 1, there is a close coupling between the selection of the suspension and drive systems. Once these two key elements of the system have been selected, the other subsystems including energy storage, power supply and conditioning, and controls are driven by those selections. There is some interplay in motor design selection with controls and power conditioning, but this isn t as fundamental as the interplay between the suspension and motor design selection. This section will describe the different possibilities for each of the key system elements. The criteria for selection will be discussed and the different possibilities will be ranked relative to each other based on those criteria. The suspension system will be discussed first since it has influenced the design in ways that favor a particular motor design. 2.1 Suspension System In considering possible suspension approaches for a Maglifter type of system, four candidate approaches were considered. These included wheels, slippers like those currently used on rocket test sleds, magnetic suspensions, and air bearings. Within the magnetic suspension techniques, the category was further divided down into passive and active suspension systems. For a man-rated system like Maglifter, criteria that must be evaluated in selecting a suspension design include the following: 1. Ability to work at the desired acceleration rates. 2. Ability to work at 0.8 Mach final velocity. 3. Overall load capacity of the suspension system. 4. Ability to provide adequate shuttle location for the selected motor design. 5. Control of dynamic forces on launch vehicle. 6. Suspension stability over the full range of system operation. 7. Drag losses due to suspension. 8. Cycle times for repeat launches. 9. Track cost and/or precision. 10. System complexity/auxiliary equipment required. 11. Reliability/robustness. 12. Technical Risk/Background experience. 14

23 This list of criteria is felt to be adequate for a trade-off analysis, but may not be as exhaustive as required if a detailed design were being done. In this list, items 1 through 7 are the basic performance criteria for any suspension system. Of these criteria, items 1 and 2 are the most basic requirements for the suspension. It must work at the desired acceleration rates and to the desired launch velocity. If these two criteria can t be met, meeting the other requirements is irrelevant. While meeting criteria 1 and 2, the suspension must control dynamic forces such as vibration transmitted to the launch vehicle and carrier to acceptable levels to prevent damage to system hardware and, for man-rated applications, provide an acceptable launch for the crew. Suspension stability over the complete range of operating conditions is closely related to control of dynamic forces. To be readily integrated into the carrier vehicle, the suspension system needs to have a reasonable load capacity in terms of load per unit area or volume of the support. For some motor designs, in particular induction motors, it is also important to maintain accurate positioning of the shuttle with respect to the stator to maintain acceptable motor performance. The final performance issue is that the suspension should have low drag so that it has a minimum impact on system thermal management and drive motor sizing. The other criteria are systems related issues. To be viable from an operational systems viewpoint, any suspension system must be able to meet the launch cycle times desired. To control costs, the auxiliary equipment required should be small. The precision to which the track must be built should be easily achievable with existing construction technology. It is important that the suspension be reliable and robust, especially for man-rated launches, but also for unmanned launches because of the cost of the launch vehicle and its cargo. The complexity of the suspension system, including any auxiliary equipment required, has a direct impact on both cost and reliability of the overall system. The simpler the system, the more reliable it is generally expected to be. Finally, to minimize development costs and risks, it is desirable that the suspension approach has background technology that can be built upon so that the technical risks are low. In rating the various candidate suspension techniques against these criteria, Table 5 was constructed. Ratings in the table are all relative with those areas where a given approach can meet the requirements, given a (+), and those areas of concern, given a (-). A discussion of each of the candidate suspensions in more detail is contained in the following sections Air Bearings Air bearings are commonly used in both linear and rotary load support. Applications like air bearing slides are common using hydrostatic techniques to provide high load capacity. In general, these air bearing slides are low speed devices. In rotating machines, air bearing spindles can be used to high surface speeds although the desired operating speed for Maglifter is about a factor of two higher in surface speed than would be considered normal operating speeds for a rotating hydrostatic gas bearing. 15

24 Table 5. Suspension technique trade-off Suspension Technique Candidates Rating Criteria Air Bearing Wheels Slippers Passive Magnetic Suspension Active Magnetic Suspension Acceleration Rates Final Velocity Load Capacity Shuttle Location Dynamic Forces Suspension Stability Drag Losses Cycle Times Track Cost/Precision Low Complexity Reliability Technical Risk Major concerns with an air bearing approach include their ability to work at the high surface speeds, the ability to maintain proper shuttle location relative to the motor stator, and control of dynamic forces. Hydrostatic gas bearings tend to have low damping unless some form of compensation or feedback control is introduced into the system. This type of control adds to bearing complexity and response times may not be adequate to provide stability over the complete speed range. Other areas of concern with air bearings include the cost and complexity of an air bearing track and the reliability of that track in an outdoor environment. Reliability issues include concerns about plugging of orifices and maintaining high precision in the track so that thin air films are properly maintained. Technical risk was also seen as an issue since air bearing technology on this scale would be developmental and thus would add to the overall system development risk Wheels Wheels are used at speeds up to Mach 1 on specialty vehicles such as the jet cars that have set the land speed record. These wheels are generally machined aluminum and are of a very special design. As such, wheels that could achieve the 0.8 Mach final speed of Maglifter were down rated on speed capability because of their special design requirements that make them far from a standard wheel. To achieve these speeds the wheels must also be relatively small meaning that they have limited load capacity. For the full-scale Maglifter with a million pound weight, wheel load capability would be an issue since a large number of wheels would be required to work in unison to carry the vehicle. Other issues include the precision required in a track to prevent wheel flyoff at the higher speeds. Appendix B contains estimates of wheel flyoff issues and track precision for Maglifter. For a 16

25 Bantam size vehicle, load capacity would be less of an issue, but speed capability of a wheel is still a concern along with track precision Slippers Slippers are the common method of support for rocket sleds for speeds well beyond Mach 1. As such they should be considered as a serious candidate for the suspension system, at least for a Bantam size vehicle. Slippers were down rated in the comparison because of concerns about load capacity, dynamic forces that may be transmitted to the launch vehicle, cycle time and track precision. On the other hand, it is known that they work successfully to these speeds on a routine basis on rocket sleds. The major disadvantage for a full-scale Maglifter is the large number of slippers that would have to work together to carry the vehicle loads. They would also add a considerable drag on the vehicle that might be a thermal issue in terms of track and slipper cooling. For a smaller vehicle such as Bantam, slippers might be a viable candidate, especially if there was interest in building a system in the short term. They would be very low in development risk since they are reasonably well established for rocket sled applications. The use of slippers in the Bantam application would need to be investigated further as part of a more detailed study of that specific application. Ultimately, they may suffer from issues with cycle time since they will need to be replaced frequently because of wear, and track maintenance will be higher than with other suspension techniques. However, they are a known technology and in a first demonstrator might simplify system design at least at the start Passive Magnetic Suspension A passive magnetic suspension is viewed as the best approach for both the full-scale Maglifter and a Bantam size launcher. It should be able to meet the acceleration and speed requirements as well as most of the systems requirements rated in the table. The major area of concern for a totally passive magnetic suspension is the stability of this type of suspension and its ability to provide adequate shuttle location under the dynamic forces that may be present in the system (including vehicle aerodynamics at least on the larger Maglifter vehicle). If stability problems can be avoided, because of the lower cost and complexity of the system, a passive magnetic suspension is favored over a dynamic magnetic suspension in this application. Any passive magnetic suspension will require wheels or another secondary suspension that supports the vehicle until adequate speeds are reached for the vehicle to levitate. Lift coils would only be used on the portion of the track where speeds are high enough for acceptable lift to drag ratios to be achieved. This is recommended because a passive suspension, like null flux coils, goes through a peak in drag force at low speeds that gives a poor lift to drag ratio (that speed depends on the ratio of the inductance and resistance in the coils among other factors). This would require the wheels to work up to speeds of perhaps 25 to 50 mph depending on the detailed magnetic suspension design. The liftoff speed can be quite low with proper system design, but since wheels are easily designed for these speeds, there might also be a cost savings by reducing the number of levitation coils needed along the track if wheels are used for some 17

26 reasonable portion of the track. Again the exact details of what liftoff speed is desirable would require a more detailed system design like that originally planned for year two of the design efforts as originally proposed Active Magnetic Suspension An active magnetic suspension can meet all of the performance requirements for this type of application including providing high precision on shuttle location and excellent dynamic stability. An active magnetic suspension can also lift a vehicle at zero speed if desired possibly eliminating the need for wheels. The major drawback in an active suspension is the increased cost and complexity associated with an active system. There are also concerns about reliability when an active system is compared to a passive one. The ultimate suspension for Maglifter might be a primarily passive suspension system like the null flux approach described above with a smaller active control system on the null flux coils that is used only to provide improved damping if it is required to control dynamic motions. This type of hybrid system might improve overall suspension performance at a lower cost than a fully active system and would also have better reliability for man-rated applications since it could be designed to provide adequate support even if the active system failed. Such a system might be designed to meet all of the design criteria with the active portion working, and provide a safe, but rougher ride than desirable if the active portion of the system were lost during a launch. This type of a suspension system could be investigated on the subscale launcher which has been built by Foster-Miller or evaluated during a detailed design phase for a demonstration system Summary of Suspension Options Of the suspension options considered, for Maglifter applications, either a passive or active magnetic suspension appears to be the best approach. More detailed analyses are needed to define the optimum design, but the primary candidate would be a passive magnetic suspension like null flux levitation. The second choice would be to add some form of active damping to a basically passive suspension if possible. If superconducting magnets are used because of their high magnetic flux, it will be important to find techniques to minimize the alternating fields experienced by the magnets. Use of high T c magnetic materials might provide magnets with reduced sensitivity to alternating fields, since these materials are generally less sensitive than low T c materials. For a smaller vehicle like Bantam, two suspension options appear viable. This includes magnetic suspension, which would be the favored approach at this time, and slippers. Magnetic suspension appears better for this application also because it has the potential to be long life and low maintenance when compared to a slipper type of arrangement. Developing a null flux suspension for Bantam would also lower the technical risk for a full-size Maglifter development. It also should transmit less dynamic forces into the structure of the carrier and launch vehicle. However, slippers are currently being used on rocket sleds to speeds significantly higher than those required and may provide a very low cost suspension if the track life is reasonable. This is 18

27 one area where further study is required to define the best trade-off between development cost, initial construction cost and life cycle cost before a final selection can be made Candidate Magnetic Suspension Options For the both the Maglifter and Bantam systems, the most viable way to support the vehicle with respect to the motor is via magnetic levitation. However, a magnetic levitation system must be chosen that is appropriate to the motor s requirements. The motor design and levitation system are closely linked in the overall system design. This section provides more detail on the design options for magnetic suspension systems. Possible methods of magnetic levitation include: Electrodynamic levitation with Null-Flux coils. Active magnetic suspension. Levitation obtained by exploiting the normal forces produced by a Linear Induction Motor (LIM). Null-Flux coils were used to levitate the scale model Maglifter vehicle and are felt to be the primary candidate for a full-scale system. With side to side linking of the null flux coils, they can provide lateral as well as vertical stability for the launcher system. Because of the null flux coil design with coils above and below the moving magnets, this type of suspension both levitates the vehicle and prevents it from rising too high above the track prematurely. This may be important for controlling lift forces during launch of aerodynamic vehicle designs like Maglifter. This method of levitation has the advantage of requiring no active control or power electronics, either on the track or on the vehicle. The use of active control for the levitation will have a significant impact on track costs since any track-mounted electronics must be duplicated many times to service the entire track length. The disadvantages of passive levitation are that the stiffness is low making it difficult to maintain small clearances and the intrinsic damping is very low (and perhaps even negative under some operating conditions). A null flux suspension was chosen for its simplicity in the scale model machine. It is also considered a strong candidate for the full-scale system because of its simplicity, lower cost and expected reliability. A low stiffness passive magnetic suspension will provide poorer positioning of the moving magnets relative to the stator magnets. This makes a passive magnetic suspension most compatible with an air core LSM since that motor design is the most tolerant to changes in moving magnet gap. A passive magnetic suspension wouldn t work well with either an iron core LSM or a LIM where tight clearances are required to maintain high performance and the presence of iron in the stator adds significant negative stiffness to the system. 19

28 Active magnetic suspensions are frequently used to suspend rotating machines with relatively small clearances (0.005 in. to in. air gap lengths are typical). They have the advantage that the closed-loop suspension dynamics can be prescribed relatively arbitrarily, permitting high stiffness and damping and easy maintenance of small clearances. The main disadvantage of active magnetic suspension for the Maglifter and Bantam launchers is that they require active control, either with controlled electromagnets on the vehicle or positioned along the track. If the active part of the suspension is located on the vehicle, the amount of power electronics is minimized, but power to run the suspension must be either stored on board the vehicle or transferred from the track. An active control of the suspension would also be an issue relative to using superconducting magnets to provide high flux and high performance in the motor design because of their susceptibility to AC fields. If the suspension stators are located on the track, power transfer problems are avoided, but a very large amount of electronics is required on the ground to operate the suspension over the entire track length increasing the overall system cost. To get an idea of the size of the required magnetic suspensions, a rule of thumb of about 100 lbf/in. 2 can be used to relate pole surface area to load capacity. This implies at least 1000 in. 2 of pole face area for a 100,000 lb load. The actual pole face area would probably be significantly larger when one considers the weight of the shuttle, suspensions, suspension electronics and controls, dynamic load requirements, and the need to produce bi-directional forces (implies an up suspension and a down suspension) in addition to the actual basic suspension load. While active magnetic suspensions in combination with a LSM or LIM may be a viable possibility, a significant development effort would be required to build an adequate active linear suspension. The larger the vehicle size, the harder it will be to develop an adequate active suspension since the volume and weight of the vehicle will increase as the cube of the linear dimensions while the available pole area will only increase as the square. A third option is to employ a linear induction motor that exploits the normal forces produced by the machine to provide levitation. While this is a viable configuration that is under consideration for many magnetic levitation designs, our evaluation of the motor design favors a LSM for this application. As such, this type of suspension wasn t evaluated in detail. One problem with designs that try to use the normal forces from the drive magnets in a LIM for suspension is that the magnetic design choices that make a good suspension are different from those that make a good motor. Because the two designs can t be optimized independently, the system tends to end up with inferior performance of both elements. The reasons for selecting a LSM over a LIM are discussed more completely in subsection 2.2. Another suspension option that might be considered is one using a continuous sheet conductor. This type of suspension has been investigated for Maglev trains. A sheet conductor suspension has higher drag at low speeds than a null flux design and only provides lift as opposed to the bi-directional vertical stabilization offered by a null flux suspension. For these reasons this design wasn t studied in detail here either. 20

29 2.2 Propulsion In considering the options for the propulsion or motor drive system for Maglifter and Bantam type of launcher applications, options to be considered include Linear Synchronous Motor (LSM) and Linear Induction Motor (LIM) configurations. These two major choices can be further divided into short stator and long stator classes for both types of motors. Short stator designs are those where the active motor windings are on the moving vehicle while long stator designs are those where the active windings are in the track. For transportation applications like people movers and trains, short stator designs are frequently preferred because they lead to lower track costs with the use of passive elements in the track. The disadvantage of short stator designs is that power must be carried on or transferred to the moving vehicle to power the system. For trains, such power transfer is feasible, either through direct electrical contact or inductive power transfer. However, as speeds increase, it becomes increasingly difficult to transfer power reliably. For Maglifter applications, therefore, with the high final speed (of 0.8 Mach) desired, short stator designs were quickly dropped from our considerations because it was felt that power transfer to the moving vehicle at a reasonable efficiency wouldn t be feasible. The elimination of short stator designs still leaves the issue of whether to favor a LSM or a LIM for the design. This issue can be further expanded into considering two classes of LSM, air core and iron core designs. Review of literature and experience of Foster-Miller team members indicated that each of these basic design classes has the potential to meet the requirements of this application. Motor selection was made by evaluating the relative merits of each motor type for this specific application. As with the suspension techniques, a set of criteria was developed for ranking the motors. Each candidate was then rated relative to its ability to fulfill that requirement. The design criteria identified for the propulsion system include: 1. Ability to provide desired 3g acceleration rate with reasonable sized motor. 2. Ability to achieve desired speeds. 3. High efficiency to minimize power and cooling requirements. 4. Minimum complexity to provide desired performance. 5. Control flexibility to launch a range of vehicle weights. 6. Graceful degradation/no single point failures. 7. Compatibility with magnetic suspension. 8. Track cost and ability to be built with conventional construction approaches. 9. Reasonable technical risk based on background experience. 10. Ability to meet cycle times on the order of days or hours. 11. High reliability/robust design especially for man-rated applications. Table 6 contains the evaluation of the three candidate approaches with respect to these technologies. 21

30 Table 6. Ratings of candidate motor designs versus requirements criteria Rating Criteria Candidate Motor Designs Linear Synchronous Linear Synchronous (LSM) Air Core (LSM) Iron Core Linear Induction (LIM) Acceleration Speed Efficiency Complexity Control Flexibility Graceful Degradation +* +* +* Suspension Compatibility Track Cost Technical Risk Cycle Time Reliability * In locally commutated or optically commutated designs. The following sections discuss each of the candidate choices and their ratings in more detail Air Core Linear Synchronous Motor (LSM) For the relatively long track and high acceleration rates required for a Maglifter design, an air core synchronous motor design with superconducting magnets on the shuttle is considered the primary candidate. With superconducting magnets and an air core design to eliminate problems with magnetic material saturation, it is possible to produce very high drive forces in a relatively compact motor design. There is experience with this type of motor design because of the Japanese developments in Maglev trains which use similar technology. Because of the higher acceleration levels, maximum speeds, and vehicle sizes, the motor required for Maglifter would be significantly larger than that needed for a Maglev train, but similar in basic design. By using a locally commutated design such as that demonstrated by Foster-Miller in the LCLSM program for DOT (3), or the optically commutated design demonstrated under this contract, a highly reliable and robust LSM design can be achieved. The number of switches and power electronics elements required can be kept down to a manageable number by block switching the locally commutated power supplies. In the Foster-Miller Maglifter test rig a total of nine power supplies, three for each of the three phases are used to power the entire track. This combination of power supplies permits three blocks of the track to be powered at any given time (in the subscale rig design as described in Section 4, the vehicle is typically being driven by three coil blocks at any one time). As the optical sensors detect that the vehicle is passing out of the first block being powered by power supplies 1, 2, and 3, those same power supplies leap frog the vehicle to turn on block 4 which is just ahead of the vehicle. In this manner, the total amount of power electronics required is kept low. At the same time, the size (i.e., current and voltage level) of the IGBTs that must be used is kept to a reasonable level because each power supply must only power three coils at any 22

31 one time rather than a block of nine or more coils as might be done in conventional block switching. An air core LSM with some form of local commutation such as optical commutation as used on the subscale system appears to meet all of the rating criteria for both the Maglifter and Bantam size applications. By using the local commutation, there is no concern about keeping the vehicle coupled to the travelling electrical wave in the track. The wave is kept in synchronism with the vehicle because it only moves down the track as the sensors detecting vehicle position tell the power supplies to switch to successive track elements. By setting the current limit and voltage on the power supply, the acceleration rates can be controlled in the system. For a full-scale system, another level of feedback control would be added to the system by adding either velocity or acceleration feedback into the system. By doing this, the current limit could be directly controlled to ensure that the desired velocity profile or acceleration rates are being maintained over the complete track length. This control strategy will provide a high reliability and insure that the desired launch velocity is achieved even if there are problems with portions of the system. By using local control, and block switching the power supplies on the track, it is possible to have a very robust system since if any individual control element is lost, only a small portion of the track is disabled. With proper sizing of system elements, and velocity or acceleration feedback, it will be possible for the remaining sections of the track to compensate for the failed portion and successfully achieve a launch. Of the possible drive systems, the air core LSM has the best compatability with a passive null flux levitation system. It is possible to couple a passive magnetic suspension with other types of motor designs but only by making design compromises that yield poorer performance. One approach would be to have magnets on the shuttle for just the levitation in a LIM motor design. This option would have to be evaluated in more detail in terms of benefits and potential problems to determine if it was viable. Because of the experience base from Maglev trains, the LSM is probably the closest to actual application at a scale similar to that needed for NASA applications. There are also expected to be less thermal problems with a LSM using superconducting magnets than with other motor designs because it will operate at higher efficiencies. This could potentially be a major advantage in the Maglifter design because of the extremely high power levels of that system. With a power of 5 GW and an energy of almost 25 GJ with losses of only 10 percent, the system will have to dissipate 2.5 GJ of energy Iron Core Linear Synchronous Motor (LSM) Technology for iron core LSM systems is also fairly well established. The German Transrapid system is an iron core LSM (4). The use of this motor configuration permits relatively high forces to be achieved with electromagnets or permanent magnets on the shuttle rather than superconducting magnets. However, in order to do this, the iron in the core is used to 23

32 focus the flux in the system. This affect of the iron in the system means that the motor must run at much tighter clearances than an air core LSM. The requirement for tighter clearances in an iron core LSM means that the track must be more precisely aligned and tighter control must be maintained on the shuttle. A gap of 7 to 10 mm can be common in this type of motor design as compared to a gap of as much as 70 to 100 mm in a large air core system. These tight gap clearance requirements raise significant questions about whether the required clearances are achievable and maintainable over the meters of a Maglifter acceleration track with the thermal effects of weather on the track, motion of the track due to ground shifting, etc., that the system will experience. This requirement for tighter control means that it is generally necessary to use an active magnetic suspension with the iron core LSM as is done in the German system. This approach has the disadvantage of increased cost associated with the more complex control and power electronics for the suspension system as compared to a null flux system which is basically passive. The iron core LSM itself produces high lift forces relative to the drive forces, at least as designed for the German system. These forces are undesirable in this case and generally can t be used to support the vehicle without the addition of a separate active suspension system. They are also only lift forces and thus provide no stabilization relative to any aerodynamic forces generated during the launch of the vehicle. The motor design for a Maglifter application would have to be significantly different from the German one because of the need for high drive forces and small lift forces making prior experience with this type of motor less relevant and the motor design more developmental. The use of iron in the motor focuses the flux to make more of it pass through the drive gap clearance making the performance for a given flux level very efficient. However, as in all iron core motor designs, it also limits the flux levels and hence the forces per unit area that can be achieved because of issues associated with saturation of the iron. This can have significant impact on the overall system design. The motor iron can be run heavily saturated, but this adds nonlinear effects into the control system which complicates control. Overall, the requirement to use an active suspension system, and the potentially high cost of the track related to the active suspension and the tight tolerances are seen as negative factors for the selection of an iron core LSM design. Also, the fact that design experience with this motor type has focused on high lift to drive designs means that more development would probably be needed to arrive at a satisfactory iron core LSM design for Maglifter. More evaluation could be used in terms of detailed cost trade-offs, but at this time, this type of motor is seen as less likely to be a good final solution for a Maglifter type of system Linear Induction Motor (LIM) As with the iron core LSM, an iron core LIM will provide high lift forces, but will probably also need a separate active magnetic suspension system to provide the required shuttle support. In general, the lift forces don t have very good stability and can vary significantly as the slip in 24

33 the motor changes. One common advantage of a LIM that makes it desirable in applications like the Navy aircraft launcher is the light weight of the moving shuttle because there are no magnets on it. In this case, since magnetic suspension is required, and the induction forces in the shuttle can t provide stable suspension, this weight advantage is lost because it will be necessary to put magnets on the shuttle for the suspension anyway. For the high power levels being considered here, there may also be significant thermal problems with the shuttle due to the induced currents. The force and energy levels required for the full-scale Maglifter may make thermal management a major difficulty with a LIM design. Much more detailed design calculations need to be done to address the thermal issues than have been completed in this one year study. For a Bantam size system, a LIM may be more feasible because of the order of magnitude lower power and energy levels. It would be important to consider, however, whether it makes sense to use a LIM for a Bantam size launcher if it can t be scaled up to a full Maglifter size by some logical progression of technology. A clear advantage of a LIM for a Bantam class vehicle would be the ability to take advantage of the significant development and demonstration efforts currently being conducted by the Navy. The vehicle sizes and required acceleration rates for a Naval aircraft are comparable to Bantam requirements although the final velocity requirements are significantly lower. Because a LIM depends on forces being produced by the currents induced in the moving shuttle by the stator magnetic fields, the flux is generally channeled with iron and clearances must be kept small. The small clearances are an issue in a LIM much as they are in an iron core LSM. A LIM drive system will require an active magnetic suspension for tight shuttle position control and a precisely aligned track over the m length of the launch portion of the system. Because forces are achieved with induced currents, higher stator currents are required to achieve a given level of force resulting in a need for a larger power system, higher current levels, and higher losses (or reduced efficiency). Because a LIM requires active suspension, it is also seen as more complex than an air core LSM. The cost of the track is also expected to be higher than with an air core LSM for much the same reasons as for the iron core LSM Summary of Propulsion System Trade-Offs Based on the trade-off analyses performed to date, an air core LSM with a passive null flux suspension is the best approach for a full-scale Maglifter type of application. The iron core LSM and the LIM both have the potential to work for this application, but appear higher risk and higher cost. More detailed design, thermal and cost analyses need to be done before making a final commitment to a system design. With any of these motor designs, a locally commutated design is recommended because of advantages in system robustness and because of possible cost savings that may be realizable by using a larger number of smaller IGBT devices in the power system. Foster-Miller has investigated these concepts both in the LCLSM system demonstrated for DOT as part of our studies of designs for magnetic levitation trains (3), and in the optically 25

34 commutated system with block switching of the locally commutated coils which has been built as part of this Maglifter study. For a Bantam size vehicle launcher, more detailed evaluation should be done for both air core LSM and LIM designs. While not specifically part of this study, preliminary evaluations of the requirements for this size system indicate distinct advantages for each approach. At a first look, an air core LSM appears more readily scalable to a full-scale Maglifter system than a LIM. On the other hand, a LIM design may be more readily buildable in the near term because it can take advantage of the Navy demonstration work on LIM launcher design. A LIM design also has the advantage of not requiring large superconducting magnets which is one of the most significant technical concerns about the LSM design. 2.3 Launch Carrier Vehicle Magnetics Considerations The design of the magnets that will be carried on the shuttle or carrier vehicle is a key issue in the system design. Since the same magnets provide the propulsion and lift forces, proper magnetic design is key to overall system performance. As will be discussed in Sections 3 and 4, the scale model system built by Foster-Miller used high flux NdFeB permanent magnets. This selection was made because of the scale of the system being designed. With the low weight target of the scale model vehicle, 6 kg total, it would be impossible to include superconducting magnets along with a cryogenic cooling system to maintain them at their operating temperatures. The other consideration that ruled out superconducting magnets was the cost of using them. As the system size increases because of scaling issues in the magnetics, the use of superconducting magnets becomes increasingly favorable in terms of performance even in systems which are only slightly larger than the one built under this program. A detailed analysis of the scaling issues associated with permanent magnets versus superconductors is given in Appendix A. That evaluation is summarized briefly in this section. In general, much greater fields can be achieved with superconducting coils than with permanent magnets as the scale increases. If the current density in the superconducting coil is assumed to be invariant of scale, the total number of Amp-Turns increases with the square of the scale factor for the coils. However, the number of Amp-Turns increases only linearly with the permanent magnets. The result is that superconducting coils perform better in large-scale devices, whereas permanent magnets outperform them in small devices. For a given design, the ability of a permanent magnet null flux suspension to provide adequate lift will decrease as size increases since the mass of the vehicle increases as the cube of the scale and the lift force with permanent magnets only increases as the square. Said another way, as a permanent magnet system is scaled up the lift to mass ratio (a measure of the suspension effectiveness) decreases inversely with the scale. This indicates that for a full-scale or even large subscale system, if a permanent magnet suspension can be made to work, it will operate at much larger displacements of the null flux system. Running farther from the null flux position will result in larger suspension induced magnetic drag losses which will increase the propulsion power required. A null flux suspension which is run at greater displacements to 26

35 achieve forces near the force peak will also be a less stable suspension because of the decrease in suspension stiffness as the levitation force peak is approached. (See Figure 2.) One way to evaluate the scale at which a cross-over occurs in scaling is to pick a configuration in which the fields are easy to evaluate. Perhaps the easiest to consider is an infinitely long cylinder of radius R alternately filled with strong PM material and superconductor. For the PM cylinder, the flux density is constant everywhere in the cylinder, and the total flux, φ, flowing down the cylinder is: φ pm = µ o H c πr 2 For a superconducting coil, the flux density is a function of radius, because different radii are surrounded by different levels of magnetomotive force (MMF). In this case, flux density as a function of radius is: B(r) = (R - r)µ o J sc Integrating over the volume of the coil yields: φ sc 1 = µ 3 ojsc πr 3 Figure 2. Permanent magnet versus s.c. coil MMF scaling 27

36 Setting φ pm and φ sc equal and solving for R yields: H R = 3 J sc c The strongest permanent magnets presently readily available (48 MGOe), have an H c of approximately 1.1x10 6 Amps/meter. For a superconducting coil, a maximum bulk current density of 200x10 6 A/m 2 is feasible. To give this number some frame of reference, about 5x10 6 A/m 2 is the maximum current level conforming to the code for residential wiring. Substituting these values into the last equation yields a radius of m (about 0.65 in.) for the cross-over point where superconducting coils become more effective than permanent magnets. The implication is that on length scales over about 1.3 in., there may be an advantage to using superconductors, rather than strong permanent magnets. It is important to note, however, that this analysis neglects the additional volume of the housing necessary to maintain the low temperature required for superconductivity. A typical thickness for this housing is about 0.05m (2 in.). This extra 2 in. of housing can be taken into account by substituting (R-0.05) for R in the calculations and resolving for the break-even point. The result is that the flux produced by the s.c. coil does not equal the flux produced by a PM of the same size as the coil and housing until R=0.11m (about 4-1/4 in.) This implies that practically, superconducting magnets are not advantageous until the diameter of the coil (including the housing) reaches more than 8-1/2 in. Since the force for both propulsion and levitation available from superconducting magnets increases faster than the mass, the fraction of the total mass required by the superconducting coils can go down for larger scale devices making the percentage of the mass available for other functions larger. The key result of this scaling analysis is that permanent magnets are appropriate for the scale model built under this effort. However, superconducting coils will be necessary in a significantly larger scale model or in a full-scale prototype so that adequate performance can be maintained without devoting too large a percentage of the carrier vehicle system weight to the magnetic field source. 2.4 Energy Storage and Power Generation Significant power generation and energy storage capabilities will be required to run the launcher system. In fact, for the full-scale Maglifter system, the power level and energy storage issues are probably the two biggest challenges that will need to be faced in system construction. As discussed in Section 1, the size of the energy storage system will be a function of: The total launch vehicle kinetic energy. The energy to accelerate the launch vehicle carrier or shuttle. Additions for losses in the system like aerodynamic and magnetic drag. Corrections for motor efficiency. 28

37 The maximum power capacity of the machine will be a function of the motor force and final speed at launch corrected for the system losses. Table 7 summarizes the requirements for Maglifter and Bantam in terms of energy storage and power. Also included in the table for comparison are capabilities of established pulse power systems that could form the basis for what would be required for a launcher system. On this basis, the Bantam system is clearly within the range where flywheels can be seriously considered as an energy storage device. A SMES is also a possibility, but is expected to be much more costly and difficult to maintain than a flywheel for a system on the scale of Bantam. On the other hand, for a full-scale Maglifter system, a total of 11 flywheels of the Princeton size or a total of 25 flywheels of the MIT size would be required to meet the system needs. At this level, a much more detailed analysis on reliability and system trade-offs needs to be done. While larger flywheel rotors can be considered than these two established systems, there are limits to rotor size that may favor using a larger number of smaller flywheels rather than a small number of very large systems. Because of the high energy levels required for Maglifter, there are only a few truly viable energy storage techniques that can be considered. These include flywheels, and superconducting magnetic energy storage (SMES). Steam and compressed air systems were also briefly considered, but don t seem like viable approaches because of the difficulty in recovering the energy quickly. For a smaller system, like a Bantam class vehicle, ultracapacitors might be added to the list, but they are still in the development stage so their true viability has yet to be shown Flywheel Energy Storage The simplest way to store and extract energy using a rotary generator is to couple a flywheel directly to the generator rotor. Since the power capacity of a generator declines with speed, there Table 7. Comparison of energy and power requirements with storage system capabilities Energy Level (x10 9 J) Maglifter 25 5 Power Level (x 10 9 W) Bantam MIT Flywheel Generator Princeton Flywheel Generator BWX SMES* UltraCapacitors 1.2 x x 10-4 *Private communication from BWX indicates that a SMES in the 2 GJ, 0.5 GW is buildable with current technology. The numbers given for SMES in the table are for a system currently under construction by BWX. 29

38 is a practical limit to how much of the initial rotary kinetic energy can be recovered. To avoid excessive over-capacity of the generator, most flywheel systems operate within 25 percent of peak speed, or 50 percent of kinetic energy. For the Maglifter system, the flywheel kinetic energy would need to be approximately double the delivered energy, or 50 GJ. While significant development has been underway in recent years in high speed composite flywheels for applications like electric vehicle surge power and satellite attitude control and energy storage, these wheels are generally in the kilowatt power level and kilojoule energy storage range. Even in a distributed energy storage and power system, these types of wheels appear to be too low in capacity to be viable. With the high technology nature of composite flywheels, such as magnetic suspension, vacuum sealing of the wheels, etc., they would also be expected to reduce overall system reliability and increase system cost when compared to a large steel wheel running at moderate speeds. An analysis of the stress in a solid cylindrical flywheel yields the following relationship of flywheel cost as a function of material (for a given energy requirement): Flywheel Cost Cost/Mass Tensile Strength/Density On this basis, the most cost-effective flywheel material for a large energy storage system is medium strength steel including the shaft inertia of the machine itself. Assuming an allowable tensile stress of 50,000 psi for a steel flywheel, the limiting tip speed is 326 m/s. At 1800 rpm this limits the diameter to 11.3 ft. Both Princeton and MIT have flywheel generators used for fusion research. Detailed specifications for these machines are given in Appendix C. MIT s machine is a modified steamturbine generator with an initial speed of 1800 rpm, 1 GJ energy delivered, and 225 MW power. Princeton uses two modified hydro-electric generators made by G.E. turning at a maximum speed of 375 rpm. Each system supplies 2.25 GJ at 500 MW. Coincidentally, the ratio of the deliverable energy to the peak power (minimum discharge time) is about the same for these two machines (4.5 sec), and very close to the requirement for Maglifter (25 GJ/ 5 GW = 5 sec). Maglifter would require at least 11 Princeton machines or 25 MIT machines without having any excessive generation or flywheel capacity. The power that must be delivered will ramp up essentially linearly over 9.3 sec to a peak power requirement of 5 GW. More details of the Princeton and MIT systems are given in Appendix B. The cost of the Princeton machine can be estimated from a quote they got from G.E. in the late 1980 s for an additional flywheel generator unit: $10 to 12 million. This translates to approximately $15 million in today s dollars or $163 million for 11 units. A direct request for quote from the division of G.E. responsible for the Princeton machine resulted in a quote of $135 million for six modified hydro-electric flywheel generators designed to meet the Maglifter requirements. 30

39 Westinghouse also estimated the cost for a system based on modified steam-turbine generators. Their preferred configurations were eight 850 MW, or five 1.2 GW machines. The approximate cost for either of these solutions was $120 to $150 million. Neither of the quotes from G.E. or Westinghouse takes significant advantage of the short duration of the power requirement. This could be conservatism on their part, or could be related to the Maglifter peak power occurring at the minimum flywheel speed. Pulse Power Advantage In sizing the power generation equipment for this type of system, it fortunately is not necessary to design the systems for steady-state operation at the required peak power levels. Fortunately, for short time periods, rotary generators can exceed their steady-state power ratings by a significant amount. This is because the resistance heating in the windings does not have to be continuously dissipated. If a generator were limited only by temperature rise in the windings and dielectric strength of the insulation, the power capability would follow the following formula: Power 1 Time In reality there is a continuum between steady-state (continuous cooling) and adiabatic operation. Peak power delivery is also limited by magnetic saturation of back-iron, and mechanical stress. To better quantify this effect, manufacturers of large generators were contacted. The graph in Figure 3 shows the pulse power capability of a Westinghouse 1 GW steam turbine generator. This curve indicates that it should be possible to deliver between 200 and 250 percent of the normal power level capacity of the 1 GW turbine for a short pulse duration on the order of 9 sec as required for Maglifter. By using this pulse power advantage, it is possible to significantly reduce the size of the generator equipment required. Location of Generators and Energy Storage Systems Though the energy requirement per unit length along the track is roughly constant, the design of the energy storage system is dominated by the power requirement (five times the output of the largest generators commercially available). The power that must be delivered is required to ramp up essentially linearly over 9.3 sec to a peak power requirement of 5 GW at the end of the drive section of the track. To minimize the cost of power transmission it may make sense to run all the generators in parallel at the end of the track, clustering them near where the power requirement is greatest. 31

40 Figure 3. Pulse power capability of Westinghouse generator Investigation of the power line requirement showed that a DC power bus (copper) operating at 20 kv would need to have a minimum combined cross section of 16 in. 2 at the end of the track, tapering down to 2 in. 2 at the beginning. This represents $490,000 worth of copper, which is insignificant compared to the cost of the generator(s) Steam and Compressed Air Storage Given the number of flywheels of reasonable size (1 to 2 GJ) required, alternative energy storage means were explored including steam and compressed air. The latter method is being studied as a way of storing energy from power plants during off-peak hours. While the problems of making and storing superheated steam or compressed air are relatively trivial, quickly converting the energy to electricity through rotary motion of a generator is not. Both steam and air turbines do not scale favorably for pulse power as does the generator for a flywheel system. A minimum of five of the largest existing steam turbines (1 GW) would be required. Furthermore, turbines are inherently delicate machines. According to designers of steam turbines at Westinghouse, a 10 sec pulse is out of the question; their turbines require tens of minutes to come up to their power rating to avoid mechanical and thermal stresses. With air turbines, the thermal transient problem is reduced, but there is still the problem of controlling the delivery of the air to match the load and the run-up time associated with the rotary inertia of the system. The supply of large air turbines is also limited with no current systems exceeding 200 MW in size. 32

41 2.4.3 Superconducting Magnetic Energy Storage (SMES) When the storage of significant amounts of energy is considered, superconducting magnetic energy storage or SMES is one of the candidate techniques generally mentioned. In a SMES system, electrical energy is stored in a superconducting coil as a persistent current. While such systems can potentially be very large, the BWX system listed in Table 7 with 100 MJ of energy stored at 100 MW is one of the largest systems to be built. A system one to two orders of magnitude larger is considered within the state-of-the-art, although such a system hasn t been built. There are several disadvantages of a SMES system as the energy storage system. These include the requirement for a large cryogenic system to maintain the superconducting state in the coil and the high cost of such a system relative to a flywheel. A major advantage is the fact that there are no moving parts in the primary energy storage device. However, there is significant rotating machinery in the refrigeration system. The current SMES systems use low temperature superconductors which means that they must be maintained at temperatures on the order of 4 to 10K which increases their operating costs. As high temperature superconducting materials advance, they may become viable candidates for large SMES systems. This type of system could be maintained in a superconducting state by liquid nitrogen (73K) instead of the helium or cryocooler systems required by low temperature materials. The availability of high temperature SMES systems could reduce the cost of this type of energy storage. At this time, however, a SMES system for either a Bantam or a Maglifter size system appears to have a significant cost and risk disadvantage versus a flywheel energy storage system. The disadvantage may be less for the larger Maglifter system because of the relatively large number of flywheels that would be required, but even then, the flywheel based system is more established and would require less development Ultracapacitors Ultracapacitors can store significantly more energy than standard capacitors, but still are orders of magnitude lower in energy storage than required to run a launcher system at either of the size levels considered here. One system concept that might work with ultracapacitors would be a totally distributed energy storage system. If the motor drive system utilizes concepts like local control of individual coils in the motor system like the LCLSM system demonstrated by Foster-Miller for DOT (3), it might be feasible to distribute the energy storage directly with the power electronics and conditioning hardware. In this type of system concept, smaller size storage devices like ultracapacitors or high performance composite flywheels might be worth reconsidering. However, smaller devices like these also tend to be lower voltage devices which means they cannot meet the power requirements to drive the system at the high speed end of the track without running a large number of systems in series. It might be possible with proper local power electronics to generate both the voltage and current to run the coils of the system even at the high speed end. Such a configuration would have to be further analyzed to determine if it is a realistic possibility. 33

42 2.5 Controls There are two aspects to the control system for an electromagnetic space launcher. One element is the control of the launch itself while the second is the control of the overall system functioning. The overall system control includes functions beyond the launch such as the braking of the launch vehicle carrier after launch, monitoring and diagnostics for the key elements of the system, and return of the carrier to the beginning of the track for the next launch. Figure 4 shows a high level block diagram of the overall control system for a launcher system. As can be seen in Figure 4, the main launch computer is the central processor for the entire launch system. It provides the central processing system through which the launch operator inputs the parameters for the desired launch and through which the launch is controlled. As part of this system there is a performance data storage system that maintains historical data on system performance for general archiving, and also as part of the system diagnostics process. As part of the detailed performance diagnostics, each new launch will be compared to prior launches under similar conditions. Any changes in performance will be evaluated for their possible significance in terms of overall system health. Utility Power Energy Storage PWM Control Power Electronics High-Speed Digital Signal Processor (direct motor control) Sensed Voltage & Current Linear Motor Position Velocity Acceleration Sensing Sensors for Launcher, and Crew Safety Power Storage Status Monitoring Signals Main Launch Computer Performance Data Storage Link to Main Launch Control Operator Display Operator Input 487-NAS Figure 4. Overall launcher system controls 34

43 Each of the key system elements, motor, energy storage, power generators, and power switching and conditioning electronics will have their own embedded diagnostics systems. To ensure high availability and reliability, the computer control system will be a distributed system design where any one of a number of distributed processors in the system will have the capability to continue a launch to successful completion even if one processor fails. The heart of the control system will be a high speed digital signal processor which will be responsible for the actual launch. This processor will track the launch velocity and acceleration profile as the launch vehicle proceeds down the track. The launch profile will be carefully tracked to ensure that the launch will be successfully completed. Upon completion of a launch, performance data from all key subsystems will be collected and analyzed as part of system condition monitoring. 2.6 Conceptual Design for Full-Scale Systems While the full-scale Maglifter system is an order of magnitude higher in both energy and power requirement than the Bantam system, the initial studies performed under this effort indicate that both systems are achievable with a common system architecture. The initial recommendations for each of the key system elements is given in Table 8. In both cases a locally controlled or locally commutated linear synchronous motor is seen as the best option at this time. This type of motor has a simpler control architecture and has been demonstrated at sizes similar to that required for a launcher. For both the Bantam and Maglifter size systems, flywheel energy storage appears to be the best option. Energy storage of the order of magnitude required is a well-proven technology with large steel flywheels used in applications like fusion experiments and particle accelerator systems. While multiple flywheels will be required, they are relatively known technology and can be made highly reliable. As discussed in subsection 2.5, a LIM design also needs to be further evaluated for launching Bantam size vehicles. During the detailed system design, it will be important to consider trade-offs in system reliability and availability with different numbers of flywheels for running the system. Having a larger number of devices means that the failure of any one device will have less impact on the launch. Power generation requirements are within the capabilities of utility type generators Table 8. Candidate Maglifter and Bantam system architectures System Element Maglifter Bantam Motor Synchronous Synchronous or induction Suspension System Null flux passive magnetic Null flux passive magnetic or active magnetic Energy Storage Multiple flywheels Multiple flywheels Power Generation Multiple utility generators uprated for pulse power Multiple utility generators uprated for pulse power Controls Distributed controls Distributed controls 35

44 especially with modifications to their design for pulse power operation. Power electronics for both size systems will be a major challenge requiring the use of some of the largest available IGBTs or similar devices to keep the number and cost of devices manageable. Because of the short time pulse nature of the operation, systems can be run at much higher power levels than would be possible for continuous operation. The controls will be a distributed controls architecture which will include all functions required to operate the system including user interface, system condition monitoring, and detailed control of each of the key elements of the system. 2.7 Relevance of Navy Catapult Interests to NASA Maglifter and Bantam Launcher Applications As discussed in subsection 1.2, the Navy has an interest in electromagnetic catapult technology for aircraft launching from carrier decks that is similar to NASA s interests in electromagnetic launchers. However, because of the lower launch speeds, smaller vehicles, and shorter track distance available, the optimum system configuration for a Navy launcher most likely won t be directly transferable to a NASA application without considerable redesign. Table 9 summarizes the key system performance requirements for Maglifter and Bantam launchers along with the order of magnitude estimates for the Navy requirements for comparison. While the force levels are comparable, the power and energy requirements are significantly higher for the NASA systems because of the higher desired launch speed and larger vehicle sizes. The desired Bantam launcher would accelerate a 100,000 lb vehicle approximately 0.8 Mach or 600 mph+. At a constant acceleration of 3 g s, a track length of approximately 1200m would be required to reach the target speed. As discussed in Section 1, this results in a requirement for just over 500 MW of mechanical output at the end of the track, and an average power output of 250 MW over the entire length of the track. The total mechanical energy transferred to the Bantam vehicle is approximately 2380 MJ. Although the payload weight is comparable in each case, the Bantam launcher ultimately requires much higher speed, power, and energy than the Navy Catapult. Since the performance specifications for the two machines are significantly different (e.g., over an order of magnitude Table 9. Key NASA and Navy system performance requirements Parameter Maglifter Bantam Navy Requirement F=Ma(N) 1.96E E+06 1E+06 P=Fv(W) 5.29E E+08 1E+08 U=1/2Mv2=Fl(J) 2.38E E+09 1E+08 t=sqrt(2l/a) (sec)

45 difference in total output energy), it is not possible to directly use the Navy Catapult design for the Bantam Launcher. One particular Navy requirement, i.e., the need to quickly stop the motor shuttle after a launch because of limited deck space, favors use of a LIM because of the light shuttle weight in an induction motor design. This is in contrast to the air core LSM design regarded as the best choice for the NASA launchers where the higher shuttle weight isn t as much of an issue because more space is available to stop the shuttle after a launch. However, there are some similarities in the two applications, and some of the technology developed in the creation of a Navy Catapult might be leveraged to cut the development effort of the Bantam Launcher. Specific technology areas to compare are: machine topology, power storage, and power electronics design Navy Catapult Machine Topology It is likely that the Navy Catapult will have a topology similar to that being investigated by Foster-Miller for the Electromagnetic Aircraft Recovery System (EARS). An artist s conception of the full-scale EARS motor is shown in Figure 5. Revised Linear Induction Machine Moving Shuttle Pole Pitch (Half Wavelength) 1' for 3 Coils Active Region of Shuttle Stator Windings External Lamination Stack Central Lamination Stack 463-NAV Figure 5. Full-scale EARS arrestor 37

46 This machine consists of a U-shaped aluminum shuttle straddling a long row of race trackshaped coils mounted on a laminated iron core. Two laminated iron cores are located outboard of the shuttle to act as flux return paths. This particular layout is good in that it allows large shear forces to be applied to the shuttle (>20 psi in the active region of the shuttle). The toothless structure allows large fluxes to be induced through the aluminum shuttle without concentrating the flux and saturating the laminated iron core. For comparison, typical linear induction motors for transportation applications have a maximum shear force of about 1 psi. This design is ideal for the Navy application. However, this machine does have disadvantages, as well. This machine tends to have a very large leakage inductance relative to most induction motors. The disadvantage of having a high leakage inductance is that at high shuttle speeds, high voltages are required to move the energy stored in leakage flux paths. The result is that the reactive power requirements might be much larger than actual mechanical output power when the shuttle is moving at high speeds. Since the power electronics must be sized to handle the reactive power as well as the active power, this machine becomes expensive to run at high speeds. While this design offers a number of advantages for the specific Navy requirements, at the much higher speeds required for NASA s applications, the high leakage inductance may be prohibitive making the LIM design less appropriate. Some aspects of the catapult s mechanical layout also make it unsuitable for use as a Bantam Launcher. Since the Navy catapult is expected to be basically an iron-cored machine, there is a sensitivity to air gap length. Air gaps must be kept acceptably small, and clearances must be maintained to obtain the desired performance from such a design. For the catapult, maintaining clearances is not a problem, because the launch speed is relatively low and the launcher is relatively short. In this case, conventional bearings or wheels can be used to accurately maintain the position of the shuttle with respect to the motor stator. In comparison, the Bantam launcher travels to much higher velocities (600 mph+) over much longer distances (1200m.). As discussed in Section 1, it is difficult to obtain wheels that can handle the stresses produced by the high load and high hoop stress at speeds approaching Mach 1 making precise positioning of the shuttle more difficult in a Maglifter application. For the NASA launchers, the most viable way to position the motor shuttle and launch vehicle with respect to the motor stator is seen as magnetic levitation. Null flux coils were used to levitate the scale model Maglifter machine and are recommended for the full-scale system also. This method of levitation has the advantages of requiring no active control or power electronics, either on the track or on the vehicle. This consideration is particularly important in linear motor applications, where any track-mounted electronics must be duplicated many times to service the entire track length. The disadvantages of this type of levitation are that the stiffness is low (making it impossible to maintain small clearances) and the intrinsic damping is very low making dynamic motions of the shuttle larger than in a well damped system. This system was chosen for its simplicity in the scale model machine and appears to have significant advantages in a full-scale Bantam or Maglifter system also. 38

47 This selection of null flux levitation makes an air cored LSM a better choice for the NASA application in contrast to the LIM being the most likely candidate for the Navy interests. In an air-cored linear synchronous motor, the motor is less strongly dependent on precise vehicle positioning making it a better fit with electrodynamic levitation. However, if a LIM were to be used, adequate positioning to maintain motor performance could not be maintained by virtue of the suspension system s low stiffness. An active suspension could be used with a LIM, but this would probably increase overall system cost because of track alignment and power electronics requirements for the suspension system which are eliminated with the electrodynamic null flux suspension. The German Transrapid system used an active magnetic suspension, but because of the high speeds required by either NASA launch system, there are concerns about the viability of that approach for this application. A much more detailed analysis of the tradeoffs and costs for a Bantam size system is required before a final selection is made Energy Storage Scheme Since the Bantam and the largest Navy aircraft are roughly the same weight and would be accelerated at roughly the same rate, the force produced by the linear machine is on the same order in both cases. Since work is the product of force and distance, the amount of mechanical energy transferred to the vehicles per unit of track length is also roughly the same in each case. This means that power transmission and motor power electronics for one system can draw on the experience gained in developing the other. Because the transient power levels for each machine are relatively very high, it is probably not possible to connect either machine directly to the local power grid. Instead, the energy needed for a launch must be stored up in some sort of reservoir, so that the energy can be accumulated slowly at a low, constant power drain on the grid. When an adequate amount of energy is stored, it can then be released over the short launch time span at high power, without drawing a large transient load from the local power system. For both applications, the method of energy storage is still under investigation. A likely candidate for energy storage for both machines is a large, relatively low speed flywheel. Energy storage systems in the required class for the Bantam Launcher are feasible, and in fact, have already been built as discussed in subsection 2.3. The MIT flywheel described in Appendix B is used to power a Tokamak fusion reactor. This reactor typically runs at a power level of 300 MW for a 3 sec run. Energy for each run is stored in a 120 ton alternator rotor connected to a 75 ton flywheel. The total energy stored in the system at 1800 rpm is approximately 2000 MJ. A 3 sec run at 300 MW would imply that about 900 MJ are drawn out of the energy storage system during each run. Three such flywheels might provide adequate energy storage for the entire Bantam launcher track; and a similar flywheel would be more than adequate for the Navy Catapult. 39

48 The design and experience from any energy storage scheme that is developed for the Navy Catapult should be directly applicable to the Bantam Launcher. However, a scale-up of the energy storage design will be necessary to accommodate the much higher energy requirements of the Maglifter launcher. This scale-up may consist simply of multiple storage devices to obtain the desired capacity, or a redesign of the smaller-scale catapult storage to the scale of the Bantam and Maglifter launchers Power Electronics Design Although the power levels (at the end of the track) in the Bantam launcher are at least five times as high as for the Navy catapult, some similar problems will have to be solved: for example, how to Pulse Width Modulate (PWM) large currents at high voltages. It has been proposed to do this in EARS via a number of large devices in parallel to obtain the desired currents. It might be possible for NASA to build on Navy experience by starting with the basic power electronics design for the catapult and increasing the number of devices as needed to obtain the desired power level Conclusions Although a Navy Catapult design could not be used as-is for a Bantam-sized launcher, a significant portion of the technology developed under a Navy catapult program might be directly applicable. The result is that development costs might be reduced, in light of the knowledge gained from this other work. Specific technologies that might be exploited for the Bantam Launcher are: Energy storage techniques. Power electronics design. 2.8 Electromagnetic Space Launch Assist System Development Strategy With the studies that have been carried out to this point, it has been possible to define a candidate system that will meet NASA s requirements for a linear motor space launch assist system. While the required system development efforts will be substantial, there appear to be no fundamental technology barriers that must be overcome to make such a system a reality. With other work that is ongoing in magnetic levitation and large linear motor applications, there is a substantial technology base which can make NASA s development task easier. This section will discuss how the system can be developed through logical steps in scale, and how other work can aid in the effort. While the full-scale Maglifter system is an order of magnitude higher in both energy and power requirement than the Bantam system, the initial studies performed under this effort indicate that both systems are achievable with a common system architecture. The initial recommendation for each of the key system elements is given in Table

49 Table 10. Key system element design recommendations System Element Maglifter Bantam Motor Synchronous Synchronous Energy Storage Flywheel Flywheel Power Generation Utility generator uprated for pulse power Utility size generator uprated for pulse power Controls Distributed controls Distributed controls In both cases a locally controlled or optically commutated linear synchronous motor is seen as the best option at this time. This type of motor has a simpler control architecture and has been demonstrated at sizes similar to that required for a launcher. For both the Bantam and Maglifter size systems, flywheel energy storage appears to be the best option at the current time. Energy storage of the order of magnitude required is a well-proven technology with large steel flywheels used in applications like particle accelerator systems. While multiple wheels will be required, they are relatively low technology devices and can be made highly reliable. During the detailed system design it will be important to consider trade-offs in system reliability and availability with different numbers of flywheels for running the system. Having a larger number of devices means that the failure of any one device will have less impact on the launch. Power generation requirements are within the capabilities of utility type generators especially with modifications to their design for pulse power operation. Because of the short time pulse nature of the operation, systems can be run at much higher power levels than would be possible for continuous operation. The controls will be a distributed controls architecture which will include all functions required to operate the system including user interface, system condition monitoring, and detailed control of each of the key elements of the system. The recent technology work funded by NASA to study Maglifter concepts can be built on to quickly complete the technology development work that is required before embarking on the construction of a 100,000 lb class launch system for a Bantam vehicle. The plan would involve continuing the ongoing work with a new focus on the design and construction of key electromagnetic and electrical building blocks for this size system. The work discussed below could be completed in two to three years and would provide realistic tests of full-scale propulsion and levitation coils in an interim size system. The power electronics would be sized for approximately 10 percent of a Bantam system with a shortened track on the order of 10 percent of the final track length. By using different weight test vehicles tests could be run over a relatively wide range of forces and energy levels with such a system. By using a full 100,000 lb test vehicle for instance, the full force of the machine could tested to a moderate speed. Using a 10,000 lb test vehicle would permit relatively high speeds to be achieved so that back EMF issues could be evaluated. 41

50 There are significant questions about performance of the scaled up system even in the 100,000 lb size range that need to be answered before committing to detailed design of that system. Designing and building an interim system would permit these questions to be answered. This interim hardware would be used to finalize the designs of the electromagnetic elements (propulsion and lift coils) and the power electronics building blocks for the Bantam system. Once these components have been optimized, the next phase of the project would focus on detailed design and construction of the 100,000 lb class launch system. In parallel with these demonstration efforts, the system trade-off studies should be continued to further optimize the system design choices. A detailed cost estimate has not been done for this type of effort, but a ROM estimate indicates that a system on the scale of the interim one discussed above could be built for on the order of $10M to 25M total. Costs could be $5 to 6M less if an available power source and energy storage system could be used. A strong reason for following this path would be to provide a direct test of the Bantam launch system coils and electromagnetics. While magnetically levitated trains have been built which are comparable in scale to the Bantam system (40,000 to 50,000 lb class), a train is a low acceleration device (0.5g max) which takes a reasonable amount of time to achieve full speed. In the launcher application, high acceleration rates are required to keep the track length and hence the track and power electronics costs manageable. The proposed interim system would evaluate essentially a section of high acceleration track and power electronics and provide significant new information that would be invaluable to the building of the full Bantam system. As described above, the interim system would test full-scale Bantam hardware on a shortened track segment that would form a building block for the full Bantam system. Results of the interim testing would be used to optimize the design of the electromagnetic elements, controls and power electronics before initiating the detailed design of the system. This would provide significant risk reduction and provide a high probability of success for the construction of the Bantam system. Based on the results of this testing, the full Bantam launch system could be designed and constructed. Once a Bantam scale system has been built, it would provide a strong basis for the design and construction of a full-scale Maglifter system. 42

51 3. CONCEPTUAL DESIGN EXPERIMENTS The coil designs for both propulsion and levitation systems were developed analytically using both 2-D and 3-D magnetic modelling. To finalize the coil designs the team felt that there was a need for proof-of-concept testing of the propulsion coils to ensure that the desired acceleration levels could be achieved. It was decided that is would be easier to build a new simple rig for this testing than to modify one of the existing Foster-Miller test facilities. The proof-of-concept rig shown in Figures 6 and 7 was designed and built in a matter of weeks. The power control electronics were also designed and constructed as part of this rig evaluation. For the overall track there are a total of 24 blocks of controls that control 9 coils each in an optically commutated control scheme. One of these controllers was built in prototype version and used to drive the coils in the rotary test rig. To give some scale to the rig in Figures 6 and 7, the lexan disks are 2 ft in diameter. Located near the outer edge of the disks are pairs of 2 x 2 in. NdFeB permanent magnets that are identical to the ones used on the test vehicle. Between the two disks are nine drive coils of the design used for the linear drive motor. Testing showed that the coils can easily drive the disk up to magnet passing speeds of at least 35 m/sec which is the planned maximum speed for the linear test rig. The protrusions from the disk are reflector elements used with the optical commutation switches to time the coil drive voltages to provide proper drive power sequencing for the synchronous motor design. Once the proper timing has been achieved, the drive currents are permanently synchronized with the motion of the reflectors regardless of the speed of the system. The rig also provided an evaluation of the permanent magnets selected and quantified the interactions between the drive coils and the permanent magnets. Measurements of back EMF were made with this test set up to define the power supply requirements and verify the design analyses that had been done to eliminate the back EMF. In Figure 7, the drive coils can be seen slightly better than in the view in Figure 6. The coils are 2 in. wide, 4 in. high and 0.67 in. thick. Testing of the coils and verification of the back EMF calculations has allowed us to trade-off the number of turns in the coils, the total coil inductance, and the back EMF to arrive at a design which operates in reasonable ranges of both current and voltage. As shown in Figure 8, the testing showed excellent agreement between calculated back EMF and measured values. Testing with this rig also permitted the controls engineers to optimize their controls design and test the optical sensors to be used to perform the optical commutation as the vehicle is driven down the track. As can be seen from Figure 8, the agreement is excellent 43

52 Figure 6. Side view of rotary rig for evaluating drive coils and magnets between the predicted force profile and the measured one. Based on these results, the sensors, permanent magnets, coils, power electronics circuits and controls were all verified before proceeding to the full subscale demonstration system. 44

53 Figure 7. Side view of rotary test rig 45

54 Figure 8. Measured and predicted back-emf from rotating test rig 46

55 4. SUBSCALE DEMONSTRATION SYSTEM The specifications for the proof-of-concept scale model system built under this effort and for the full-scale Maglifter and Bantam launchers are shown in Table 11. This section will discuss the system design, review the system as built, and discuss the preliminary testing which has been completed. 4.1 System Design The total length of the scale model is twice the length of the acceleration portion of the track that is shown in Table 11 to provide adequate distance to stop the test vehicle with the eddy current brake. A simple eddy current brake was selected for the system because it was inexpensive and reliable. Being a passive device, the brake works effectively even if the power to the system fails. For an actual launch system, it may be desirable to use an active brake (linear motor run as a brake) to reduce the stopping distance. However, because of its simplicity and failsafe nature, an eddy current brake may be useful even in a large system to provide an extra margin of safety in stopping the launch carrier vehicle once the launch vehicle has separated from it. Using the trade-offs discussed in Section 2 and the positive results of the component testing in Section 3, the scale model system was designed as an air core linear synchronous motor using optical commutation. Null flux coils are used to provide levitation. The permanent magnetic Table 11. Maglifter and subscale system specifications Bantam Maglifter Maglifter Scale Model Parameter 3(g) (3g) (1g) (As Built) M=mass(lb) 100,000 1,000,000 1,000,000 M=mass(kg) 45, , , a=acceleration(g s) v=velocity(m/sec) v=velocity(mph) l=length(m)=(v2)/2a F=Ma(N) 1.36E E E P=Fv(W) 3.68E E E E+04 U=1/2Mv2=Fl(J) 1.65E E E E+03 t=sqrt(2l/a) (sec)

56 field for both the drive and propulsion is provided by 2 in. x 2 in. NdFeB magnets mounted on the test vehicle with the same magnets being used for both propulsion and levitation Initial Decisions and Design Constraints Although superconducting coils will be used on the a full-sized system, scaling considerations, which are discussed in more detail in Appendix A, point to the use of strong permanent magnets in the small-scale model. Superconducting coils also require cryogenic cooling and an elaborate casing, which is not feasible on a 6 kg prototype vehicle. Due to considerations of availability and an appropriate scale, it was decided to build the vehicle around 2 in. x 2 in. x 1/2 in. NbFeB magnets with a 37 MGOe energy product as shown in Figure 9. These magnets are a stock item available from Crucible Magnetics. It was also decided based on the tradeoff study discussed in Section 2 that the vehicle should be magnetically levitated in all degrees of freedom via a null flux Maglev approach Configuration After considering several alternatives, a general configuration for the scale model Maglifter was defined. The device could be considered as built around the drive coils. The drive coils are racetrack-shaped coils positioned as if they were threaded on a very long bobbin with an oval cross section. The vehicle straddles the drive coils. There is a row of vehicle magnets on either side of the drive coils. The polarity of the magnets in each row alternates going down the length of the vehicle with a total of six magnets on each side of the drive coils. The 2 in. x 2 in. magnets are spaced 1 in. apart on the vehicle structure. Figure 9. Crumax 3714 (37 MGOe) magnet 48

57 On the outside of the vehicle magnets are a row of "Figure 8" connected levitation coils standing abreast. There is one row of levitation coils on each side of the vehicle. This configuration is pictured in Figure 10, with one of the rows of levitation coils removed to expose the magnets and drive coils. To obtain approximately the desired vehicle weight, it was determined that 12 magnets should be placed on the vehicle (six on each side). Since each magnet weight approximately 0.25 kg, the total magnet mass is approximately 3 kg. With a total mass goal of 6 kg for the system this left 3 kg for structure to hold the magnets in place between the drive and levitation coils Lift or Levitation Coil Analysis A three dimensional boundary analysis showed that the mutual inductance of adjacent figure- 8 lift coils is small compared to the self-inductance of the coil. Therefore, this analysis only considers a train of magnets passing over a single figure-8 lift coil, neglecting the interaction between adjacent coils. The figure-8 levitation coil is pictured in Figure 11. The levitation coil is composed of two roughly square coils connected in reverse series. The ends of the leads are shorted together, so that the only load driven by the coil is its own impedance, and the only source of voltage is the induced EMF from the motion of the magnet. Figure 10. Maglifter scale model configuration 49

58 Figure 11. Magnet about to pass figure-8 connected lift coil If the magnet is perfectly centered with respect to the coils as it passes over the coils, the net flux linking the coil is zero. Positive EMF from one side of the coil is cancelled out by negative EMF induced in the opposite side of the coil. Since there is no net induced voltage, no current flows in the coil. When the coil is centered, there is no lifting force, and no drag force. However, when the coil is displaced, more flux is linked by one-half of the lift coil, and less by the opposing half. This situation is pictured in Figure 12. A net voltage is induced, creating a current in the coil, and a magnetic field that acts upon the magnet. For small displacements, the direction of the induced currents is such that there is a net centering force on the permanent magnet as it passes over the lift coil. To get a tractable solution for the dynamics of the null-flux suspension, some simplifying assumptions must be made. First, it is assumed that the car is infinitely long and moving at constant speed. Second, higher harmonics than the fundamental harmonic of the field produced by the vehicle s magnets are ignored, observing that in general, higher harmonics contribute only a fraction of the total forces produced by electric machines. Lastly, it is assumed that the displacement of the vehicle from the centered position is relatively small, so that the flux linkage of the magnets with the lift coil can be approximated by a function that is linear in the vertical displacement, δ. 50

59 Figure 12. Displaced magnet passing across lift coil Under these assumptions, the net flux linking the coil has the form: Φ = αδ cos(ωt) where α is the magnitude of the fundamental component of the PM s flux linkage with the lift coil, and ω is the frequency at which the magnets pass the lift coil: ω π = v l (ν is the velocity of the magnet and l is the magnet pole pitch, m (3 in.)). In practice, α is relatively difficult to approximate analytically. However, this parameter can be obtained relatively easily via Amperes, a three-dimensional boundary element code. The induced voltage in the coil is the total derivative of the flux with respect to time: V d Φ = = δαω sin( ωt) + δα cos( ωt) dt An electric circuit equation can then be written for the coil: L dt dt + Ri = δαω sin( ωt) + δα cos( ωt) 51

60 where L and R are respectively the inductance and resistance of the lift coil. The instantaneous force generated on the magnet in the centered position is: f i d Φ = = ( αcos( ωt)) dδ i A convenient transformation to use in this case is: i(t) = z o (t)cos(ωt)-z 1 (t)sin(ωt) If z o and z 1 vary slowly compared to ω (or if the lift coils are located 120 deg apart electrically), the average force is: f = 1 z o 2 α Noting that di dt ( ) ( ) ( + ) ( ) = z ωz cos ωt z ωz sin ωt o one can substitute the transformation into the electric circuit equation for the coil. If parts associated with sine and cosine are separated out, one obtains two equations for the two unknowns, z o and z 1: Lz ( 1 + ωz 0 )+ Rz 1 = δαω Lz ( z Rz 0 ω 1)+ o = δα Applying the Laplace transform and solving for the z s yields: 2 2 ( ) s L sr L Zo = αδ + + ω sl+ s2lr+ R + ω L Z αδωr = sl+ s2lr+ R + ω L Substituting into the expression for average force yields: ( ) s L sr L f = 1 αδ + + ω 2 sl+ s2lr+ R + ω L

61 This transfer function can be used directly to describe the dynamics of the suspension. However, a more intuitive understanding of the suspension dynamics can be obtained by considering the first two terms of the suspension s Taylor expansion about s=0. The result is an equation for the force with terms that look like stiffness and damping coefficients: L f = αω 2 R + ω L RR L ( ) 1 α ω δ δ R + ω L 2 ( ) This expression can be used to shed some light on the nature of the suspension s dynamics. Figures 13 and 14 are non-dimensionalized plots of the stiffness and damping parameters versus vehicle speed based on this equation. They evidence some undesirable properties: the damping coefficient becomes negative at the same time that the stiffness coefficient is beginning to enter the zone in which the highest stiffness is obtained. Drag forces from the suspension can also be obtained from the solutions for z 0 and z 1. For some steady-state displacement, the average power dissipated in the coil is: 2 π / w ( 2 1 ) w 2 1 P = ir= Rz z 2 π 2 0 Figure 13. Stiffness coefficient versus velocity 53

62 Figure 14. Damping coefficient versus speed Employing conservation of energy, P R fdrag = v = π 2 2 αδ ω l R + ω L The drag force for each figure-8 lift coil is plotted in Figure 15. The result looks like a torque-slip curve for an induction motor. The curve has the desirable feature that drag force falls at higher speeds. In general, suspension dynamics for the other degrees of freedom work via the same mechanism; similar transfer functions should be available for all 6 deg of freedom. Several different candidate geometries were evaluated to get a design able to lift the vehicle at a relatively low speed. It was determined that having coils placed either 60 or 120 deg apart electrically is beneficial from the point of view of eliminating the lowest-frequency part of the ripple on the force. This spacing causes the 2ω component of the force ripple to sum to zero, leaving only the steady-state component plus higher harmonic (and presumably smaller magnitude) ripple. 54

63 Figure 15. Drag versus speed This desirable magnet spacing implies that the lift coils can be spaced either 3 lift coils per magnet pole pitch, or 1-1/2 lift coils per magnet pole pitch. Due to scaling considerations the 1-1/2 coils/magnet pole means that larger coils can be used that have a smaller R/L (implying lift-off at a lower speed). Since the inductance and coupling coefficient, α, were determined by deducing the values from a 3-D boundary element analysis, each candidate geometry is somewhat laborious to evaluate. Several geometries that looked reasonable were evaluated. The best geometry evaluated has lift coils where each half of the coil is a 2 in. x 2 in. square, with a 1 in. x 1 in. bore. The depth of the coil trades off increasing L (undesirable, in the sense of limiting maximum stiffness) with increasing α (desirable, because this increases the maximum stiffness). A depth of 1/2 in. was selected. This geometry is pictured in Figure 16. The relevant attributes of the coil are listed in Table 12. Note that the choice of wire gauge is somewhat arbitrary. The turn count ends up cancelling out in the equations for the coil s dynamics, so the dynamics are invariant to turn count. An 18 AWG wire was assumed, because this size is easily available and relatively easy to wind. For the magnets to interact properly with the lift coils, the magnets must be positioned with a 1.0 in. gap between magnets, obtaining the desired 1-1/2 coils/pole spacing. When the track was built, the vehicle height was set so that the magnets were placed approximately 0.25 in. below the null flux coil centerline so that sufficient lift force would be provided to raise the vehicle off the wheels at a speed on the order of 10 mph. 55

64 Figure 16. Lift coil geometry Propulsion Table 12. Lift coil properties Since the drive must apply high forces Lift Coil Attribute Value without the benefit of an iron core, the design α 1.44 Newtons/Amp L mh goal is to get as many current carrying coils R as close to the magnets as possible. This N 100 turns of 18 AWG consideration leads to the tube propulsion coil array, in which it is possible to pack all the space adjacent to one face of the permanent magnet with copper. From the point of view of drive efficiency, the best configuration is to position the magnets in direct contact with one another, and to use drive coils without any space between them. This configuration is pictured in Figure 17. However, there must be some space between coils and between magnets, due to mechanical considerations for adequately mounting the magnets and the coils. In addition, the most advantageous configuration levitation system has an inch space between the permanent magnets. For the subscale model, coils were selected with the geometry pictured in Figure 18. Each coil in the array is separated by a 0.3 in. gap, into which an alignment structure can be inserted. There is a 3/16 in. nominal air gap between the coil surface and magnet surface. Schematically, the assembled drive structure is pictured in Figure

65 Figure 17. Top view of best drive configuration Figure 18. Propulsion coil geometry 57

66 Figure 19. Drive structure, top view The idea in controlling the propulsion is to make the coils act as if they are a set of magnets with a pole pitch the same as the magnets on the vehicle, but leading the vehicle magnets by half a pole pitch. At this orientation, the most force per unit of current is obtained. This configuration can be achieved by a technique reminiscent of brushless DC motors. Plates are attached to the vehicle, which occlude optical sensors associated with each coil. Depending on the state of the optical sensors, the coil current is controlled to be either be i max, 0, or -i max. In this way, a traveling wave of current moves down the track, maintaining a constant position relative to the vehicle. The result is a constant drive force (or acceleration) applied to the vehicle, regardless of position or speed. The magnitude of the drive force is varied by modifying the magnitude of i max. The detailed analysis of the drive system is a complicated, three-dimensional open-circuit magnetics problem. However, with some simplifications, the system is amenable to a more tractible 2-D analysis. This 2-D model was used for design purposes, and then the final design was evaluated and adjusted based on information from a 3-D boundary element analysis and experimental data from a rotating test rig. To complete the design, it is necessary to analyze the two-dimensional geometry previously shown in Figure 19. The first step is to get a closed form solution for the field of the permanent magnets. NdBFe magnets can be thought of magnetically as volumes of air surrounded by current sheets. The density of the current in the current sheets is exactly equal in magnitude of 58

67 H c, the magnet s coercivity. In our 2-D geometry, the location of the magnet s current sheets is pictured in Figure 20. The currents can be decomposed into a Fourier series in the x-direction. Retaining just the fundamental component yields, over the region that the magnets lie, a current density distribution of: J mag Hc x = 2 3 cos π l l It is interesting to examine the magnitude of J mag. For Crumax 37, the value of H c is approximately 9.5 x 10 5 A/m 2. Substituting into J mag yields a current density of 43.2 x10 6 A/m 2. This is about eight times higher than the current that would be acceptable for a steady current in copper wire. However, it is over two orders of magnitude lower than might be realized with superconducing coils, which can support a sustained current density of 10 to 20 x 10 7 A/m 2. With this simpler, continuous expression for the magnet currents, the resulting field can be calculated closed-form. Since the field is sinusoidal in the x-direction, the equations for the magnetic field can be collapsed into an ordinary differential equation. The solution for the y-component (the component that produces a force on the drive coils in the x-direction) is: Figure 20. Decomposition of magnet MMF into a continuous current density 59

68 B y = ( ) 2 3µ ohc π g+ l/ 3 exp π l x y ( 1 exp ( π/ 6) π ) sin sinh π l l where g is the gap between the magnets and the coils, and l is the magnet pole pitch (3 in.). Assuming a gap of 1/8 in., the expression for B y simplifies to: B y = Cos[ x] Sinh[ y] (Note that y refers to the distance from the line of symmetry between coils.) The J x B force produced by the magnets on the coils can then be integrated over the volume of one 100-turn coil positioned at x=x o to yield: F= i Cos[ x o ] where i is the coil current. If the current in the coil is a cosine of magnitude I that matches the variation of the force with magnet position, the average force is: F avg =(1.269/2) I= I Since there are three coils per magnet, and six magnets on each side of the vehicle, the total force is obtained by multiplying the average force per coil by 18: F total =11.42 I Since it is desired to produce a constant force of 600N, the amplitude of the coil currents is: I=600 N/11.42 (N/A)=52.53 Amps However, the plan is to switch the coils between i max, 0, and i max, with the waveform pictured in Figure 21. Each coil is in a full-on state for 1/3 of the cycle, off for 1/3, and fullnegative on for 1/3 of the cycle. For this waveform, it turns out that the height of the fundamental is about 110 percent of i max. That is, I=1.1*i max The peak current predicted from the 2-D model is then i max = Amps A similar result was also obtained making a 2-D model of the device in a 2-D planar finite element program. A typical solution from the calculations is pictured in Figure

69 Figure 21. Drive current and fundamental sinusoidal component Figure D finite element solution for the propulsion 61

70 Once a design was arrived at by approximate 2-D methods, it was checked using the computational 3-D magnetics solver Amperes, by Integrated Engineering Software. This program allows the user to make a three-dimensional model of the coils and magnets. Once the model is created, the program can drive the magnets past the coils and compute the resulting relationships between coil current and force. The model used to obtain the forces on a single coil with Amperes is shown in Figure 23. If the coil is again assumed to have 100 turns, and the magnet H c =9.5 x 10 5 A/m, the profile shown in Figure 24 of the current-to-force relationship is obtained by driving one pair of magnets past a single coil: By shifting and superimposing multiple copies of this elemental curve, the current-to-force relationship can be obtained for the entire vehicle acting upon coils that are connected in an arbitrary way. The design decision was made to control the coils in three phases, since every third coil interacting with the vehicle has an identical current. Each phase is then broken into blocks of Figure 23. Wire-frame and shaded representations of Amperes model 62

71 Figure 24. Force/current parameter via Amperes for 1 coil, 1 magnet pair three adjacent coils, so that only those coils near the vehicle need be energized to create forces on the vehicle. However, in considering the current necessary to produce a 10g force on the vehicle, the coil currents could still be considered to be connected as three (non-segmented) phases. Assuming a long track and the previously described vehicle with six pairs of magnets, Amperes predicts that each phase will see the waveform pictured in Figure 25. Also pictured in Figure 25 is the coil current waveform that will be applied to the phase to produce force. By integrating the product of these two functions and multiplying by 3, a more exact estimate of the total current requirement can be obtained. The result of this procedure yields currents with i max = 52.8 Amps from the full 3-D model. Recall that the equivalent 2-D result was Amps, about a 10 percent difference from the 3-D model. The maximum back-emf can also be estimated as a result of Figure 25. The maximum backemf that each phase sees is the amplitude of Figure 25 multiplied by the maximum vehicle speed of 35 m/s. This procedure yields a back-emf of 220V. However, this result is for an entire phase. Since each phase is broken down into blocks of three coils, each block can be interacting with at most only half the magnets on the vehicle. Therefore the back-emf that any one block of three coils will see at maximum speed is 110V. Note that this back-emf scales linearly with vehicle speed. By superimposing the elemental waveform in Figure 24, the current-to-force versus position curve for an entire three-coil block can be obtained. This curve is pictured in Figure

72 Figure 25. Current-to-force for an entire phase Figure 26. Current-to-force for a three-coil phase block 64

73 In order to overcome the back EMF in the coils and provide adequate voltage and current to produce the drive forces desired at maximum speed, a 300V DC power supply was designed for the overall system. 4.2 Subscale Test Rig Description This section contains photographs and descriptions of the various elements that make up the subscale test track. The track has been run at high acceleration rates and the performance of the drive and levitation systems have been verified by these tests. Because the drive power is greater than the braking in the current design, all tests to date have been run at 150V on the DC buss to limit the average acceleration on the track. When run at 150V the motor provides design accelerations at lower speeds, but as the back EMF in the drive coils approaches the drive voltage the acceleration drops off until there is minimal acceleration over the last 5 to 6 ft of the drive. Figure 27 shows the complete subscale Maglifter system. The track is a total of 40 ft long and is mounted on a laminated beam which is supported on a series of six standard wooden top laboratory benches. The benches were used to place the system at a convenient working height Figure 27. Subscale Maglifter system 65

74 for assembly and testing. As seen in the picture, the 20 ft of the track starting at the far end contains the drive motor while the 20 ft at the near end is the braking section. Figure 28 shows a closer view of the test vehicle and the drive section of the track. The laminated beam which forms the base for the track has been raised off of the table tops to permit the wires for the sensors to be run under the track. The wires for the coil blocks are also brought out through the laminated beam and under it to make connections with the switching transistors. The high voltage DC power supply is also seen on the wall behind the track. Figure 29 shows a close-up of the sensor system for measuring vehicle speed and levitation height. The lower electronics board on the vehicle carries an optical sensor for detection of speed and levitation height using the triangular dark and light pattern that can be seen on the side of the track. The upper electronics board is a low power RF transmitter and battery pack. The signals from the optical sensor are transmitted by the RF system to a stationary receiver positioned along the side of the track. The output from the receiver is then input into a storage oscilloscope for analysis. By measuring the passing frequency of the leading edges of the dark pattern, the average speed can be calculated. Measurements of speed at different positions along Figure 28. Closer view of test vehicle and drive section of track 66

75 Figure 29. Close-up of sensor system the track are used to calculate acceleration rates. Individual measurements of the duration of the pulses generated by the dark and light pattern lengths are used to measure the vertical height of the vehicle. As the vehicle accelerates, it lifts off of its wheels and levitates above the guide rails. At full acceleration, the vehicle was found to lift off in the first 1 to 2 ft of the track depending on the acceleration rate at roughly 9 mph which is where the null flux coils were expected to provide adequate lift based on the design analysis. Figure 30 clearly shows the optical sensor boards along the side of the track and the timing tabs on the vehicle which are used to provide the optical commutation of the synchronous motor. By adjusting the position of the tabs on the vehicle relative to the permanent magnets, proper phasing is obtained between the flux wave in the motor stator and the magnets on the vehicle. Figure 31 shows the braking section of the track. The brake is a totally passive eddy current brake which has been designed to provide high drag at the higher vehicle speeds. The aluminum in the brake has been machined out to reduce the overall effective conductivity of the brake section. In a large-scale system, some combination of an active braking system and passive eddy 67

76 Figure 30. Optical sensor boards along the side of the track current braking would probably be used to reduce the length of the braking section. The brake as designed is somewhat less efficient than the drive motor in terms of acceleration rates since eddy current braking provides a force proportional to speed meaning the brake becomes less effective as the vehicle slows. Because the brake is less efficient than the drive, the drive power has been limited to date to keep the average accelerations over the length of the drive to approximately 6g s. This has been done by limiting the DC voltage on the drive to 150V instead of using the full 300V available. Operating the track in this manner gives full 10g accelerations in the early part of the track where speeds are low and consequently back EMF values in the coils are low. As speeds increase, the acceleration rates decrease as the back EMF increases in the coils. When running at 150V the back EMF in the coils nearly matches the drive voltage resulting in minimal acceleration over the last 5 to 6 ft of the motor. Using the track in this manner has permitted us to demonstrate that the desired acceleration rates can be achieved, and that null flux levitation does work in combination with the LSM drive. 68

77 Figure 31. Braking section of track The testing has been limited however, to maximum speeds on the order of 59 mph instead of the design goal of 78 mph. Since acceleration demonstration and verification of the coil design techniques to achieve the desired accelerations was a major goal of the demonstration, the existing speed limitation is not seen as a major drawback in the design. Figure 32 shows the drive coil sections. Two nine coil sections are seen in the middle of the track. The coils on both sides of the drive coils are the null flux levitation coils. Seen along the side of the track are the optical sensors for the LSM commutation. Figure 33 shows the nine drive pulse width modulation power supplies which are used to power the track. There are three power supplies for each phase of the three phase drive. At any one time three blocks of coils are powered by one of the three power supplies for each phase. Three blocks of coils are driven at one time since the vehicle is two blocks of coils long meaning that except when the vehicle is exactly centered in two blocks, it will span a portion of three coil blocks. When the optical sensors detect that the vehicle has moved out of one block, the three power supplies powering that block leap frog the next two blocks on the track to power the coils 69

78 Figure 32. Drive coil sections on the block that the vehicle is just entering. Each power supply is thus used to power one phase on nine coil blocks over the 27 block total length of the motor drive section. Figure 34 shows one of the control boards which are used to switch power to each block of the track. Figure 35 shows one of the switching transistor blocks. There is one control board and one set of switching transistors for each coil block. The control boards are distributed along the drive section of the track as are the switching transistors. 4.3 Demonstration Testing As originally proposed, the program plan was to construct the demonstration system in the first year of the program and only conduct limited testing to demonstrate the basic functioning of the system. During the second year, more extensive testing was to be conducted investigating the details of the system performance. Because of the limited funds available when the track was completed, only the simple performance tests have been completed. 70

79 Figure 33. Nine drive pulse width modulation power supplies Initial tests were run at increasing current levels to determine how well the drive and brake performed. This testing determined that at a drive voltage of 150V (half power), the motor had sufficient force to drive the vehicle nearly to the end of the brake section of the track. Because of this brake limitation, all tests have been performed to the lower power levels at this time. Figures 36 through 38 show oscilloscope traces of the system levitation measurements made using the onboard optical sensor and the levitation measuring system on the side of the track. These traces were taken with seven of the nine drive power supplies operating. At the top of these displays is a second trace which shows the position of the pulses from the track side optical commutation sensors relative to the levitation data. These signals were taken from the last optical sensor on the track and therefore effectively mark the end of the drive section of the track. In Figure 36 the total levitation event can be seen. Analysis of the details shows that the vehicle lifts off at approximately 9.5 mph at a distance of 1.5 ft from the beginning of the track. It remains levitated well into the brake section of the track and then slowly drops back onto the 71

80 Figure 34. Control board wheels at the vehicle stops. After reviewing a number of similar traces, there seems to be a consistent change in the levitation as the vehicle enters the brake section probably caused by the fact that the front of the vehicle is into the brake while the back of the vehicle is still being driven by the motor. This may cause the vehicle to pitch up as it enters the brake. Figure 37 focuses on the drive portion of the motor. It shows that once the system levitates it is relatively stable until the brake section is entered. Again the upper trace marks approximately where the end of the drive system is reached. On the far right of the levitation trace is seen the start of the upward pitching motion as the vehicle enters the brake. Figure 38 is an expansion of the early part of the vehicle motion. The jaggedness in the trace is the image of the triangle pattern which is distinguishable at low speeds. As the speed increases the individual pulses merge together at this time scale. The results of velocity measurements made at three places along the track using all nine of the power supplies are given in Table 13. The average acceleration over the whole length of the 72

81 Figure 35. Switching transistor block drive section of the track is 6 g s. At the 2m and 4m positions, the average acceleration is approximately 8.6 g s. The drop in the acceleration in the later end of the track is due to the fact that the system is currently being run at 150 VDC instead of the full 300 VDC. With the system operational, it could now be used to investigate the stability of the levitation, and several concepts for improving that stability if necessary. Concepts have also been developed for improving the braking capabilities of the system so that it could be run at full power to expand the range of operation. 73