Status of Pulse Tube Cryocooler Development at Sunpower, Inc.

Transcription

1 89 Status of Pulse Tube Cryocooler Development at Sunpower, Inc. K. B. Wilson Sunpower, Inc. Athens, OH D. R. Gedeon Gedeon Associates Athens, OH ABSTRACT Sunpower, Inc. and Gedeon Associates continue to develop linear-compressor driven pulse tube cryocooler technology under SBIR funding from NASA Goddard Space Flight Center. The first Phase II development effort between November 1999 and November 2001 established that a single-stage pulse tube driven by Sunpower s manufactured linear compressor technology achieves efficient performance and promises to reach mass production costs similar to Sunpower s commercial Stirling cryocoolers. In addition, efficient cooling with a two-stage cold head was demonstrated at 30K. Sunpower and Gedeon Associates recently completed another SBIR program from NASA Goddard Space Flight Center to develop a three-stage, high frequency, linear-compressor driven pulse tube cryocooler for cooling below 10K. This paper reviews the development history of pulse tube technology at Sunpower Inc., updates the status of the single-stage pulse tube cryocooler, and presents the results of the three-stage pulse tube cryocooler development program. INTRODUCTION Sunpower, Inc. has been developing Stirling cycle cryocoolers for nearly fifteen years and currently manufactures the CryoTel TM commercial line of low-cost, high-efficiency Stirling cryocoolers. The Stirling cryocoolers make use of a linear compressor which is well suited for driving a pulse tube cryocooler. Sunpower s linear compressor technology can also be found in domestic household refrigerators and cryosurgical devices. Gedeon Associates has developed modeling and optimization software for twenty years, and introduced the commercial Sage software ten years ago. The Sage graphical interface allows the user to assemble a single or multi-stage pulse tube cryocooler model from interconnected component pieces, then to optimize that model interactively. 1 Copyright 2004 by Kluwer Academic/Plenum Press, New York 29 March - 1 April 2004 International Cryocooler Conference, Cambridge, Massachusetts

2 Combining the relative experience in linear compressors and pulse tube modeling and design, Sunpower and Gedeon Associates have worked together for over four years developing high-efficiency, linear-compressor-driven pulse tube cryocoolers (PTC s). To date, NASA Goddard Space Flight Center in Greenbelt, MD has funded two collaborative Sunpower/Gedeon SBIR Phase II programs. The goal of the first research effort was to demonstrate high efficiency in a singlestage PTC driven by Sunpower s linear compressor technology, which at that time was being manufactured for Sunpower s model M87 Stirling cryocooler 1. The single-stage approach was to build a u-tube configuration after gaining experience with an in-line configuration. As a secondary goal, a second-stage u-tube cold head was to be constructed as an add-on to the single-stage u-tube configuration for a two-stage PTC. Each build incorporated inertance-tubes functioning as acoustic tuning devices between the pulse tube and reservoir volume. Performance for each configuration (in-line, u-tube and two-stage) was demonstrated at 100W electrical input power. The ultimate objectives were to establish the commercial potential of linear-compressor driven PTC s and whether they could be cost-competitive with the Sunpower M87 Stirling cryocooler. The recently-completed SBIR Phase II program focused on the development of a high-frequency, linear-compressor driven three-stage PTC for cooling below 10K. Building on the multi-stage construction experience gained in the previous program, a new cryocooler was designed, fabricated and tested. Testing of particulate regenerators was conducted on an oscillating-flow regenerator test rig located at Sunpower on loan from NASA Glenn Research Center. Empirical data from this testing was used in the design and optimization of the second and third-stage cold heads which employed particulate regenerators. A 30 Hz, 300W e dual-opposed linear compressor was fabricated early in the program while the regenerator testing was underway. The cold head was built and tested sequentially: the first-stage cold head was tested as a stand alone single-stage PTC, then the second-stage cold head was added for a two-stage cryocooler. Finally the third-stage cold head was fabricated and tested to conclude the program. SINGLE AND TWO-STAGE PULSE TUBE DEVELOPMENT REVIEW/UPDATE Figures 1 through 3 show the pulse tube configurations that were constructed and tested. Detailed results from this effort were presented previously 1 and Table 1 summarizes the findings of the single and two-stage pulse tube development program. At the beginning of the three-stage PTC development program, the hardware from the single and two-stage PTC program was used as a test bed for design improvements. One such design improvement increased the performance of the single-stage hardware. The original design of the heat exchangers used woven copper screens soldered to a copper housing. Dr. Ray Radebaugh at NIST suggested that we try diffusion bonding to join the screens and their housing. We wanted to try the new bonding process simply to allow greater flexibility in the assembly technique of the three-stage cold head. However, we found that this new bonding procedure also increased the performance of the in-line PTC as shown in Figure 4. The in-line configuration now lifts 5.8W at 77K with 100W e input. Diffusion-bonded heat exchangers were never implemented into the u-tube and two-stage configurations from the original SBIR program. Table 1. Summary of single and two-stage PTC development program (100W e input for all tests) In-line Configuration U-Tube Configuration 2-Stage Configuration 1st Stage Lift K K ~80K 2nd Stage Lift K 2

3 Figure 1. In-line PTC configuration. Figure 2. U-tube configuration. 3

4 Figure 3. Two-stage PTC configuration. Since the ultimate objective of the program was to establish the commercial potential of the single-stage u-tube PTC versus the production M87 Stirling cryocooler, we performed a preliminary cost study on a production design of the u-tube prototype. The component cost of the single-stage PTC was very similar (within a few percent) to that of the M87 Stirling cryocooler. Assembly costs should be similar to, if not lower than, those of the M87. So the preliminary analysis suggested that a single-stage u-tube PTC should be cost competitive with the Sunpower Stirling cryocooler. THREE-STAGE PULSE TUBE DEVELOPMENT As a result of the success of the first SBIR program, NASA Goddard Space Flight Center awarded a second SBIR Phase II contract to develop a three-stage PTC for cooling below 10K. The two-year contract, beginning in January 2002, involved the design, fabrication and testing of a three-stage cold head driven by a high-frequency, dual-opposed linear compressor. We employed the multi-staging u-tube design and construction techniques derived in the first SBIR program, with improvements in the bonding of heat exchangers as discussed earlier. Additionally, there were new challenges in the design and fabrication of low-temperature particulate regenerators. The cryocooler was constructed sequentially, allowing us to test the first stage as a stand-alone single-stage cryocooler to compare to the simulated results. Then we added the second-stage cold head and tested a two-stage PTC, followed by the three-stage PTC. Linear Compressor (Pressure Wave Generator) After evaluating design trades between performance of the cold head and compressor requirements relating to its size and mass, we decided on a 200W e, 30Hz, dual-opposedpiston linear compressor. Actually, the linear motors are 150W e per side, 300W e total for the dual-opposed unit but the design point was targeted at 200W e. The 200W e level was more acceptable for the normal operating point of a space-flight cryocooler and also 4

5 After heat exchanger improvements Before heat exchanger improvements W e Input 20 C Reject 7 6 Lift (W) Cold Temperature (K) Figure 4. Performance of in-line PTC, before and after heat exchanger improvements. Table 2. Features of the dual-opposed linear compressor. Mass 11 kg Physical Envelope 141φ mm x 335 mm OAL Frequency 30 Hz Electrical Power 300 W (150 W per side) Total Swept Volume cc Charge Pressure 25 Bar allowed for some reserve power during cooldown. The prototype linear compressor is shown in Figure 5, with its features listed in Table 2. Low Temperature Regenerators Sunpower had no prior experience with packed-particle regenerator matrices that would be needed for the second and third stages of the cryocooler, so the program began by re-commissioning an existing oscillating-flow test rig for use in characterizing regenerator flow resistance and heat transfer. Oscillating-flow testing gave us a way to directly compare the pressure-drop and heat-transfer correlations used in the Sage simulation with experiments. The test rig was originally developed by Sunpower and Gedeon Associates under NASA funding. It was designed to measure thermal energy flow down a regenerator sample subject to oscillating helium flow, similar to that which occurs in actual cryocooler operation. Data reduction software takes the results from a large number of data points at varying frequency, pressure, flow amplitude, and other data points, and derives general-purpose correlations for friction factor and Nusselt number using parameter estimation techniques. The hardware and data-reduction software are documented in a NASA contractor report 2. The rig was last operated in 1993 and it was necessary to spend some time bringing the various pieces up to date and learning how to use them again. Once the hardware and software were functional, a few shake-down pressure-drop tests were run to verify that the rig was operational. After studying construction designs for particulate regenerators, we established an epoxy-bonding technique similar to the process described by E. Luo, et al 3. Initial testing 5

6 Figure 5. Dual-opposed linear compressor for three-stage PTC. of regenerator samples in the oscillating flow test rig required us to modify our overall regenerator design and bonding technique until we arrived at a robust design that could withstand substantial pressure amplitudes across the structure. To optimize the epoxy technique and address the effects of epoxy bonding on gas flow we first ran some nitrogen steady-flow pressure-drop tests on a number of matrices. Steady-flow tests are much simpler to conduct than an oscillating-flow test and quickly allowed us to evaluate the quality of a particular process. After several design and test iterations we had a satisfactory design and construction technique for the epoxy-bonded particulate regenerator matrix. The oscillating-flow testing was completed and the results were used for correlation of the Sage modeling software. Once this was done we were able to model and optimize the second and third stages of the cryocooler. Spherical lead particles were used in the second stage regenerator and Er 3 Ni rare-earth particles were used for the third-stage regenerator. Due to the expense of the Er 3 Ni, no oscillating-flow testing was conducted with this material, only with the lead material. Cold Head Modeling and Design Initial studies of the effect of operating frequency suggested that lower frequencies would yield better results. For example, Figure 6 shows simulated third-stage cooling power as a function of frequency. Each data point represents a completely optimized machine. At 6K, cooling power goes up from about 66 mw to 82 mw as frequency goes from 40 Hz down to 30 Hz. Below 30 Hz the size and mass of the compressor start to become issues, so we chose 30 Hz as our design-point frequency. These optimizations took place before we had oscillating flow characteristics for particulate regenerators, so only general trends were valid, not absolute cooling powers. For subsequent optimizations we lowered third-stage temperature down to 4.5K and found that there was an even larger dependence on frequency as the low temperature decreased. These preliminary cold-head 6

7 rd Stage 6K Frequency (Hz) Figure 6. Simulated third-stage heat lift as a function of frequency. optimizations also highlighted the need for better understanding of packed-particle regenerators. Our approach was to optimize the three-stage cold head for some design point and then investigate the performance at selected off-design conditions. We felt that the most important design condition was with the pulse tube orientation other than in the favorable cold-end-down position. Sage has the capability to simulate pulse tube orientation by way of a gravitational-stabilization factor included in the calculation of the manifold-induced pulse tube convection loss. Without the gravitational-stabilization factor the manifold convection loss corresponds to weightless operation. It also roughly corresponds to terrestrial horizontal orientation, except for an unknown amount of free-convection with the gravitational field acting perpendicular to the axis of the pulse tube. Originally we optimized the cold head for cold-end down performance, then checked the performance in a weightless environment. We found the off-design performance was drastically less than the optimized performance. So the gravitational-stabilization factor was turned off during cold head optimization, making the design point a weightless environment. The hope was that a cold head designed for weightless operation would not perform too much worse than one optimized for cold-end-down orientation. This did indeed turn out to be the case. The main difference between the two optimizations was that the pulse tubes are somewhat thinner in the weightless optimization and the turning-manifold flow areas somewhat larger. The second-stage regenerator was designed as an entirely lead matrix. In the thirdstage regenerator we considered two options: 1) a layered structure of lead and Er 3 Ni, with lead for the first 30% of volume then Er 3 Ni for the remaining 70%, and 2) a monolithic packing of only Er 3 Ni. Due to the expense of the Er 3 Ni we decided on the layered regenerator. Table 3 shows the final simulated results of the performance of the optimized three stage cold head as well as the simulated design of stand-alone two and single-stage cold heads, all with 200W PV power. Table 3. Final predicted performance results of Sage simulation (200 W PV power). 3 Stage Cold Head 2 Stage Cold Head 1 Stage Cold Head 1st Stage Lift K K K 2nd Stage Lift K K 3rd Stage Lift K 7

8 Figure 7. Assembled three-stage cold head without inertance assemblies. Fabrication The machining of the manifold components of each stage were outsourced to a precision CNC machine shop specializing in the use of CAD/CAM technology. Other than those components, all parts were machined at Sunpower, Inc. Most joining processes, such as the diffusion bonding of the heat exchangers, brazing and welding were also outsourced. However, some specialized joining techniques of the inertance assemblies were developed in-house. Figure 7 shows the assembled three-stage cold head without the inertance assemblies. Figure 8 shows the entire pulse tube cryocooler test setup. The first-stage inertance tube is wrapped around the first-stage reservoir mounted perpendicular to the linear compressor. Figure 8. Experimental setup of three-stage PTC. 8

9 7 6 5 Lift/PV (%) First-Stage Testing Frequency (Hz) Figure 9. Frequency-dependent performance of single-stage cold head. The first stage consistently achieved about 9.2W of lift at 80K with 200W e input power. Compressor PV power was also measured and the compressor was found to be operating very efficiently, typically from 87% to 90%. With equivalent piston amplitude between Sage and experiment, Sage predicted compressor PV power quite closely, but predicted a higher cooling power. Testing suggested that efficiency would increase by running at a higher operating frequency. Figure 9 shows the single-stage experimental performance for fixed piston amplitude versus varying operating frequency. Later testing of the first-stage inertance assembly revealed some insight into the performance by showing that the inertance assembly was optimally tuned around 35 Hz. This means that for a given piston amplitude, the largest amount of PV power produced was at 35 Hz. Of course the inertance assembly is only one piece of the system, so as the frequency increases to gain performance in the inertance assembly, other factors can change such as the mechanical tuning of the linear compressor. Second Stage Testing Figure 10 shows the experimental measurements and Sage predictions for secondstage cooling power as a function of temperature. During the testing the first-stage temperature remained close to 85 K, the compressor piston amplitude was about 11.3 mm (corresponding to about 160 W PV power input) and the first-stage cooling power ranged from about 0.5 W down to 0.1 W. The plot shows that Sage and experiment are similar. Part of the discrepancy is due to Sage over-predicting the pressure amplitude in the compression space. When the pressure amplitude in Sage is decreased, simply by decreasing the piston amplitude, the agreement to the experimental load curve is much better. Conversely, increasing the experimental compressor pressure amplitude should bring the experimental cooling power close to the Sage prediction. 9

10 Experimental Data Sage Prediction 3.5 Second-Stage Lift (W) T 1 ~ 85K X p ~ 11.5 mm Second-Stage Temperature (K) Figure 10. Comparison of experimental data and Sage-predicted performance on second-stage cooling power. Third-Stage Testing Unfortunately by the time we had the three-stage PTC assembled and ready to test, our funding for the program was virtually depleted. Therefore we were only able to run one three-stage PTC performance test and the performance was less than the target performance for the cryocooler. Our most plausible explanation of what happened is that we had underestimated the load imposed on the second stage by the presence of the third stage. As a result the second stage stalled at a temperature higher than 27 K as designed. This left the third-stage inertance assembly out of tune with the result that the third stage also stopped cooling at a temperature higher than anticipated. We feel the resolution to the problem is simply a matter of diverting less PV power from the second stage by suitable modification to the third stage regenerator, possibly something as simple as increasing its flow resistance. The result of that should be to allow the second stage to continue cooling to its intended operating temperature, at which time the third stage should also begin to perform as intended. Conclusions Sunpower, Inc. and Gedeon Associates have completed two SBIR Phase II development programs under funding from NASA Goddard Space Flight Center. The first SBIR program demonstrated that single and two-stage pulse tube cryocoolers driven by the same commercial linear compressor technology used in Sunpower s Stirling cryocoolers have high efficiency and are commercially viable. In the second SBIR program a threestage pulse tube cryocooler driven by a dual-opposed linear compressor was constructed and tested. There were many benefits to this development program including the modeling and fabrication of low-temperature regenerators and the design and construction techniques for multi-stage cold heads. Performance of the two-stage cold head was verified to be in close agreement with simulation. Available program resources were 10

11 exhausted prior to completion of the technical goals. The next steps in the troubleshooting of the three-stage PTC have been identified and further funding is being pursued. Acknowledgements We would like to thank NASA Goddard Space Flight Center for their continual support in the development of pulse tube technology at Sunpower, Inc. End Notes 1 K. B. Wilson and D. R. Gedeon, Development of Single and Two-Stage Pulse Tube Cryocoolers with Commercial Linear Compressor, Cryocoolers 12 (2003). 2 D. Gedeon and J.G. Wood, Oscillating Flow Regenerator Test Rig: Hardware and Theory with Derived Correlations for Screens and Felts, NASA CR , Feb E. Luo et al., Fabrication and testing on monolithic single bin bed by low temperature epoxy bonding, Proceedings of the CEC, Vol. 47, pp (2002). 11