REVERSE TURBO BRAYTON CYCLE CRYOCOOLER DEVELOPMENT FOR LIQUID HYDROGEN SYSTEMS PI: L. Chow (University of Central Florida) J. Kapat (University of Central Florida) T. Wu (University of Central Florida) K. B. Sundaram (University of Central Florida) C. Ham (Florida Space Institute) Description Spaceport operations of the near future are one of the prominent applications for usage of large quantities of cryogenic propellants. Efficient storage and transfer of these fluids is necessary for reducing the launch costs. Also, for future manned and unmanned deep space missions and other missions to Mars, NASA is planning for extended cryogenic propellant storage durations of the order of several months as opposed to a few days or weeks. The goal of this task is to develop a reverse turbo Brayton cycle cryocooler for zero boil-off (ZBO) of cryogenic propellants. This cryocooler will be capable of extracting a few tens of watts (20-30 W) of heat at liquid hydrogen temperature (~18 K). Objective To develop reliable, compact, lightweight, affordable and highly efficient in their class cryocoolers for distributed cooling of liquid hydrogen systems for spaceport applications. Benefit to NASA All the previous attempts of cryocoolers by NASA for zero boil-off (ZBO) of cryogenic propellants in space have cooling powers of less than 2W at liquid hydrogen temperature. These versions of flight-like cryocoolers would be more appropriate for cooling of sensors and super conducting materials in a spacecraft. The present cryocooler under development with a few tens of watts of cooling power at liquid hydrogen temperature would be ideal for ZBO of cryogenic propellants in NASA s future robotic missions to Mars and for other human space missions. Work Accomplished This report focuses on the work accomplished during phase III of the project. Accordingly, the following discussion would be on the scaled down (with respect to power and speed) miniature centrifugal compressor design verification, integration and preliminary testing of the scaled motor/compressor assembly, and designs of 5.4 kw motor and two-stage compressor that would be used in future in the unscaled version of the cryocooler. Miniature Centrifugal Compressor Numerical Simulation, Integration with Motor and Preliminary Testing of the Assembly (with appropriate scaling) Numerical study of the miniature centrifugal compressor provided the results with air at the design condition and was presented in previous reports. However, the compressor was designed to compress helium, and air is used only for experimental convenience.
The performance of the compressor using helium as working fluid was predicted by dimensional analysis and numerical studies were conducted to evaluate the predictions. The similarity principle (dimensional analysis) can be expressed as: Pˆ 2 m& RT 00 ND ND prtt, η tt, = f,,, γ 3 5 2 ρ 00 N D 1442443 p D γ γrt ν 00 00 1444 44 244 4443 Similarity Variables Performance Variables Where, pr tt = total-to-total pressure ratio, η tt = total-to-total efficiency Pˆ = dimensionless power, m& RT00 = dimensionless mass flow rate 3 5 ρ N D 2 00 ND γ RT 00 p 00 D = ratio of impeller blade tip speed to the speed of sound, and, γ = specific heat ratio γ ND 2 = Reynolds number ν This equation predicts the performance of a scaled version of an original compressor. It also predicts the performance of a compressor if we change its inlet conditions or working fluid. For high Reynolds number flows, the performance of the compressor is a weak function of the Reynolds number. In order to better understand the theory of similitude, neon, which has the same γ (specific heat ratio) as helium but molecular weight close to air, is also considered. Through dimensional analysis, the compressor speeds at design point are found to be 313,000 RPM and 141,000 RPM, when the working gases are helium and neon respectively. Table 1 shows the design and test speeds (about 10% off-design speeds) for different working fluids obtained using the above principle. Table 1. Design and Test Speeds for Different Gases Working Fluid Neon Helium Air Design speed (RPM) 141,000 313,000 108,000 Test Speed (RPM) - 280,000 350,000 95,000 120,000 Performance of the compressor is plotted in figure 1. The performance curves of air, neon, and helium at design condition are in a very close region. Helium and neon results have almost the same values of efficiency at the same pressure ratio. CFD results thus verify the dimensional analysis. The integrated compressor motor rotor is mounted on a structure as shown in figure 2. Testing of the miniature centrifugal compressor is currently under progress. The initial phase of the testing comprised of the blank shaft test. As a continuation of the work reported in previous reports, the rotor structure is built in three pieces. The main shaft is made of stainless steel. It consists of the shaft and plug. The permanent magnet is inserted into the shaft. The plug is press fitted and then
electron beam welded to the main shaft. The aluminum impeller is mounted on the steel shaft by threading. Figure 1. Performance of the compressor at design speed (with Air, Neon, and Helium) The rotor is supported on two bearings, which are housed on the two ends of the structure. Figure 2. Rotor mounted on two bearings housed in the brackets
Stator of the motor is housed inside the motor cooling jacket. The cooling jacket is cooled with water. Part of the rotor and stator can be seen in figure 3. Figure 3. Stator and Rotor housed inside the cooling jacket The motor control software installed in a computer controls the DSP board (figure 4) connected through an emulator. The DSP board is connected to the controller, which supplies power to the stator of the motor by a low pass filter. Figure 4. Emulator, DSP board and controller board Temperatures and pressures have to be measured at various places in the experimental setup to get sufficient and accurate data. This will help for better optimization of the performance parameters and thus will enable for easy and successful implementation of future tasks involving systematic optimization. An Innova-Vortex mass flow meter will be used for mass flow rate measurements. Setra P- Transducers supported by multimeters will be used for pressure measurements. Figure 5 shows the accessories used in the test setup. The rotor is run up to a speed of 100,000 RPM. Because of some problems with the bearings used and the shaft balancing issues, it was not possible to stably run the rotor at this speed for enough time so that data can be collected sufficiently to evaluate the compressor performance. From a preliminary analysis, it is realized that the problem is
with the bearing radial play provided (0.0002 in) and the maximum interference fit specified by the manufacturer (0.0002 in). Since both the values are same, the bearing does not provide any provision for thermal and centrifugal growth of the shaft. Since it is customary to provide a tight interference fit between high speed rotating shafts and bearings, a value of 0.0002 in was chosen with the presumption that it provides good rigidity (when the motor alone was tested to 200,000 RPM in phase II of the project, the fit provided was lower than 0.0002 in because of short length of the shaft and a different material used, which does not need enough rigidity from the bearings). But it is now planned to provide a fit of 0.0000 in, so that there exists some radial play even after the fit. Further troubleshooting regarding balancing of the rotor is going on at this point. As the design speed is above the first rigid body mode, the rotor is running in flexible mode. Balancing of flexible rotors is currently being pursued for immediate implementation. Figure 5. Mass Flow Meter, Thermocouple units, Pressure Transducers and Multimeters A new rotor is planned to be built for which the material selection will be done after the troubleshooting phase. This rotor will be machined out of two pieces, and is expected to provide a better performance compared to the current specimen under test made of three pieces. The impeller on the rotor will comprise a single piece and plug the other. Magnet will be inserted into the rotor and will be contained inside with the help of plug. Diagnosing the current rotor for any other possible problems and testing the new rotor with a goal of reaching the speed of 108,000 RPM is the immediate key objective of the project. 5.4 kw Permanent Magnet Synchronous Motor Design This part summarizes the design of a 313,000 RPM permanent magnet synchronous motor with 5.4 kw shaft output power. The design work includes analytical analysis, numerical simulation, mechanical design, components selection, thermal design, rotordynamic analysis, optimization, and final verification. All these tasks have been performed and the design has been completed. Figure 6 shows the designed rotor structure. Table 2 shows the optimized key dimensions of the PMSM design. This new 5.4 kw, 313,000 RPM PMSM design is adopted from the already developed 2 kw, 200,000 RPM PMSM since the 200,000 RPM PMSM design was very successful and it has been demonstrated in previous phases of the project that this kind of structure is ideal for super high-speed motor design. The same low loss ceramic ball bearings will be used. Based on manufacturer s report, these bearings can run up to 500,000 RPM.
Since no slip between magnet and the shaft inner surface was found in our previous motor/generator test, round permanent magnet will be used to reduce machining cost and increase assembling feasibility. High energy density permanent magnet of Samarium-Cobalt (SmCo) will be used in the design. SmCo can work at high temperatures, which will prevent it to get demagnetized when the operating temperatures go high during testing. 240 strands of Litz-wire will be used to reduce eddy current loss and ohmic loss in the stator winding. Shaft diameter will be increased to meet the new rotordynamic requirements. (a) (b) Figure 6. (a) Integrated rotor, and (b) the cross section of 5.4 kw PMSM Table 2. Optimized dimensions of the 313,000 RPM PMSM design Physical gap length 1 mm Stator inner diameter 32 mm Stator outer diameter 48 mm Motor active length 36 mm (35.8 mm) Litz-wire 240 strands @AWG 36 Turns/phase/pole 5 Winding pitch 12/15 Shaft diameter 17 mm PM diameter 13 mm Shaft thickness 2 mm Permanent magnet SmCo 2:17-31 More extensive FEM simulations with rotation, including eddy current loss in the permanent magnet and stainless steel shaft were performed and the simulation results are shown in Table 3. The PMSM simulation also included a connection with external
rectifier circuit. Figure 7 shows the simulated eddy current distribution in the shaft and permanent magnet, when the PMSM is working as a generator and rotating at 313,000 RPM with 5.4 kw of rectified DC output power. The simulated eddy current loss is found to be 20 W as shown in Table 3. Table 3. Simulated and/or calculated PMSM parameters and losses Back peak EMF (V) 64.1 Nominal phase current (A) 40.5 Required minimum DC bus (V) 112 Torque (N.m) 0.165 Output Power (W) 5400 Copper DC loss (W) 42 Copper AC loss (W) 16 Stator iron loss (W) 25.8 Rotor eddy current loss (W) 20 Bearing loss (W) 210 Windage loss (W) 44.6 Low pass filters loss (W) 7 Total loss (W) 365.4 Efficiency (%) 93.7 Figure 7. Simulated rotor eddy current when rotating at 313,000 RPM Detailed thermal/cooling system design and rotordynamic analyses are reported in the next section. The controller design for the 5.4 kw PMSM is based on space vector PWM method and switch mode method to realize high efficiency real-time control and power supply. Figure 8 shows the block chart of the proposed controller for this design.
f * acceleration deceleration f f θ V Programmable SVPWM signal generator Interface 3- phase switch mode power supply PMSM Motor DSP Programmable dead time generator Figure 8. Block diagram of the proposed controller design Two-stage Centrifugal Compressor Design The structure and design of the 313,000 RPM two-stage miniature helium centrifugal compressor incorporated the proven aerodynamic design of its scaled version, the 108,000 RPM single-stage air compressor. Due to the current limitation of bearing selection, the same ceramic ball bearings will be used (as mentioned in the 5.4 kw PMSM design section). The 1/4 ball bearing is rated higher than 300,000 RPM, while it limits the shaft size. With the use of bearing support at both sides, the radial IGV, radial impeller and axial diffuser design will be used to minimize the overall system size. Manufacturability is another issue that needs to be considered in this design. A base-cover structure to ease manufacturing and assembling will be used as is the case with single-stage compressor assembly. Some components, including second stage diffuser, compressor end-board and motor cooling jacket may also be split for easy assembling. The overall cross-sectional drawing of the two-stage compressor/motor assembly is shown in figure 9. Based on the ease of manufacturability, one of the two similar configurations (figure 10) will be considered for the rotor. The second configuration will not have a tie-shaft structure but has threading on the plug projection. Alignment for high speed rotating machine is a big issue. The base will be machined from a single stainless steel block to ensure the perfect alignment of the bearing holes. All other stationary components will be installed in the base to ensure the relative positional accuracy. Left side bearing has an adjustable axial spring loader to ensure stable running at high speeds. An FEA model was built for this integral shaft system to perform rotordynamic simulation accurately. The rotordynamic equation of motion for a coupled, flexible rotor/casing system with conventional bearings is: [ M ]{ q& } + [ C]{ q& } + [ K]{ q} = { f }, where q represents the physical coordinate degrees of freedom, f represents external forces, M represents mass matrix, K represents the bearing stiffness matrix, and C represents damping matrix. The simulated results for the structure with two impellers are shown in Table 4. The critical speed for rigid body modes are well below the nominal operating speed of 313,000 rpm. And the critical speed for first bending mode is much higher than the operating speed. So there is enough room for a safe operation.
stator Motor cooling jackect magnet Bearing 2 nd stage diffuser 1 st stage diffuser Rotor Figure 9. Solid model of the compressor/motor assembly 93 45 φ45 2 35.8 Magnet φ18 φ36 φ6.35 Thread surface Washer/ Locker Figure 10. Rotor design for two-stage helium compressor Table 4. Simulated critical speed for the rotor structure with two impellers Shaft Material Titanium Stainless Steel 1 st rigid body mode (rpm) 44609 35153 2 nd rigid body mode (rpm) 92813 72150 1 st bending mode (rpm) 508389 512075 The thermal management forms a necessary aspect in this design because the motor will generate 5.4 kw shaft power at full load. Supposing it has an efficiency of 0.9 (including controller efficiency), the heat generated will be higher than half-a-kw. Because of small size, it is imperative to have an efficient cooling system design to
transfer this heat and to guarantee the safe and efficient running of motor. An aluminum water cooling jacket is designed to handle this problem. Simulation shows that with water speed in the jacket as 2 m/s and inlet water temperature as 20 0 C, the inner surface of the water jacket can be below 60 0 C at full load (figure 11). This will ensure the temperature in winding and magnet is less than 100 0 C. This water jacket can also be such that it can be split into two half pieces axially for an easy installation. Figure 11. Thermal management of the rotor Water jacket simulation Since the development of compliant surface gas bearings is de-scoped from the project (as mentioned in previous reports), the rotor design is highly constrained by the available bearing technology. In case the air/air foil bearing is available, the shaft can be simplified to a straight and shorter one to enhance its rotordynamic stability and ease its manufacturing and installation. The bearing support can be placed after impellers on either side, which will reduce the complexity of the structure and eliminate the use of the radial IGV. A possible improvement in future could be to use an axial IGV with tiltable vanes to ensure perfect aerodynamics in compressor. Future Work Efforts in the past three years have been focused on the design, fabrication and testing of a compressor and electric motor two key components in a RTBC cryocooler. Both the compressor and electric motor are very compact, highly effective and reliable. Future work continues the development, testing, evaluation and improvement of an integrated compressor/motor. Another effort in phase IV is to perform thermodynamic cycle analysis to see how the current compressor/motor can be utilized in the storage of methane a potential application of the current effort in view of NASA new exploration initiative. Other potential NASA mission needs for cryocoolers will also be researched. March 2006