Date post: | 16-Apr-2015 |
Category: |
Documents |
Upload: | edu-blumer |
View: | 51 times |
Download: | 4 times |
American Institute of Aeronautics and Astronautics
1
Control of Dual-Opposed Stirling Convertors with Active Power Factor Correction Controllers
Timothy F. Regan* and Edward J. Lewandowski† Sest, Inc. Middleburg Heights, Ohio 44130
Jeffrey G. Schreiber‡ NASA Glenn Research Center, Cleveland, Ohio 44135
When using recently-developed active power factor correction (APFC) controllers in power systems comprised of dual-opposed free-piston Stirling convertors, a variety of configurations of the convertors and controller(s) can be considered, with configuration ultimately selected based on benefits of efficiency, reliability, and robust operation. The configuration must not only achieve stable control of the two convertors, but also synchronize and regulate motion of the pistons to minimize net dynamic forces. The NASA Glenn Research Center (GRC) System Dynamic Model (SDM) was used to study ten configurations of dual-opposed convertor systems. These configurations considered one controller with the alternators connected in series or in parallel, and two controllers with the alternators not connected (isolated). For the configurations where the alternators were not connected, several different approaches were evaluated to synchronize the two convertors. In addition, two thermodynamic configurations were considered: two convertors with isolated working spaces and convertors with a shared expansion space. Of the ten configurations studied, stable operating modes were found for four. Three of those four had a common expansion space. One stable configuration was found for the dual-opposed convertors with separate working spaces. That configuration required isochronous control of both convertors, and two APFC controllers were used to accomplish this. A frequency/phase control loop was necessary to allow each APFC controller to synchronize its associated convertor with a common frequency.
Nomenclature APFC = active power factor correction ASC = Advanced Stirling Convertor EMF = electromotive force GRC = Glenn Research Center SDM = System Dynamic Model SMD = Science Mission Directorate SPDE = Space Power Demonstrator Engine SSR = subsynchronous resonance TDC = Technology Demonstration Convertor
I. Introduction UAL-OPPOSED Stirling convertors have typically been operated with passive power factor correction using tuning capacitors. The tuning capacitors correct the power factors and operating frequencies when the two
convertors are controlled by a single analog controller. This was used with the Space Power Demonstrator Engine (SPDE) that had a common expansion space shared by the two convertors, and has also been used with dual-
* Sr. Aerospace Engineer, 18000 Jefferson Park Rd, Middleburg Heights, OH 44130, AIAA Member. † Sr. Aerospace Engineer, 18000 Jefferson Park Rd, Middleburg Heights, OH 44130, AIAA Member. ‡ Mechanical Engineer, Thermal Energy Conversion Branch, M/S 301-2, 21000 Brookpark Rd, Cleveland, OH 44135, AIAA Member
D
American Institute of Aeronautics and Astronautics
2
opposed Technology Demonstration Convertors (TDC) used for the SRG110 flight development project operating with thermodynamic cycles isolated from one another. In both cases, stable, synchronous operation was achieved.
There is a desire to use Active Power Factor Correction (APFC) controllers that eliminate the need for the tuning capacitors.2-5 The capacitors occupy volume which translates into mass in a space power generator, and impact reliability. As passive elements, they cannot be adjusted over time to account for changes in load impedance or to optimize system efficiency and performance. APFC controllers typically measure or estimate the piston position, calculate the piston velocity, and then switch various voltages at high frequency to force current to flow in proportion to the velocity. The force applied by the current appears as pure damping to the convertor if the force is in phase with the velocity. This approach minimizes electrical losses that arise from flowing current out of phase with the velocity.
APFC controllers have been developed by several labs, including GRC, and have successfully operated single Stirling convertors. Control of dual-opposed convertors with APFC control has proven to be more difficult than originally anticipated. Issues have been encountered in terms of synchronization and stability. In preparation for operating dual-opposed convertors with APFC control at GRC, a study was undertaken to determine which configurations of convertors and controllers could achieve stable, synchronous operation, and which could not.
The study investigated stability and synchronization of the various configurations. Convertors were considered stable when piston amplitude and other parameters were maintained within desired ranges. Stability was evaluated by observing the steady-state response of the system with convertors with slightly different characteristics, and also the transient response of the system to a disturbance of some form. In cases where the system was unstable, often the piston amplitude of one convertor decreased while the amplitude of the other piston grew without bound.
While it is understood that a good control system will result in stable operation, it is also a requirement that the dual-opposed convertor controller achieve synchronization of the two convertors. Synchronization is defined as both convertors operating at a common frequency with their pistons oscillating in phase with each other. By maintaining synchronization, dynamic forces generated by the internal oscillating masses are reduced.
This study used the GRC System Dynamic Model (SDM)6-7 to evaluate the candidate configurations. Two convertor models were used: the Technology Demonstration Convertor (TDC),8 manufactured by Infinia, and the Advanced Stirling Convertor (ASC),9 manufactured by Sunpower. While the study did not evaluate all possible free-piston configurations, it is felt that these results apply to many free-piston Stirling convertors when connected in the configurations studied, and are not specific to the two convertors used in this study.
The SDM simulations contain many of the nonlinearities associated with Stirling systems, including the nonlinear thermodynamics, gas dynamics, and spring rates, as well as the nonlinearities of some of the electronic components in the controller. Each simulation used an SDM model for each of the two convertors. In order to make the cases realistic, a slight difference was introduced into one of the parameters. In all the simulations performed for this study, the alternator motor constants of the two convertors differed by 0.5% to 2%. External disturbances were simulated by applying a 25-N impact on the case of one of the convertors or by varying the hot-end or cold-end temperature of one of the convertors by 2°C. The hardware realization of two Stirling convertors will not be exact copies one another. The differences must be dealt with in the system design. SDM used the slightly different motor constants to mimic the hardware case and present a realistically dissimilar pair to the APFC controller.
A. Related work The interconnection and operation of multiple free-piston Stirling convertors was studied by White in 1999.10
The study investigated means of synchronizing convertors operating with tuning capacitors and with an analog controller. Various dual-opposed configurations were simulated with a linear model and the results were compared with test data. All of these configurations included tuning capacitors and had either one or two Zener diode controllers with the alternators connected electrically in parallel either before or after the tuning capacitors or after the controllers. Series connection of the alternators was not considered.
Both the simulation studies and the test data included not only steady-state data but also data describing transient responses due to making and breaking the parallel electrical connection between the two alternators. Operation with significant amounts of unbalance in the convertor fill pressures and in operating temperatures was studied and the effects described in the report. Each configuration was studied with the convertor cases rigidly connected to each other as well as with convertor cases mechanically isolated. With isolated cases, the mechanical connection to transmit dynamic forces was broken.
The report identified the two-convertor/one-controller system with alternators connected in parallel after tuning capacitors as a robust, stable configuration. This configuration has been used for a significant amount of testing at GRC.
American Institute of Aeronautics and Astronautics
3
Another contribution to the body of knowledge related to control of multiple free-piston convertor systems came from the SPDE project in which two 12.5kW convertors were coupled through a common expansion space as well as through the alternator electrical connections. The alternators and electrical load were connected in series with the system’s two linear alternators through a tuning capacitor located after the two alternators. Testing showed that the SPDE operated stably under a wide range of operating conditions.
In addition, there is nearly 100 years’ worth of engineering literature describing the development of interconnected power systems. Among the concepts potentially relevant to the dual-opposed convertor control problem is the phenomenon of subsynchronous resonance (SSR). SSR is an electric power system condition where the electric network exchanges energy with a generator at a natural frequency of the combined system below the synchronous frequency of the system.11 Methods of analyzing and eliminating the effects of subsynchronous resonance in interconnected power systems have been discussed in the literature in relation to large 3-phase rotating systems, but the concepts have been found useful in studying dual-opposed Stirling control systems. The significant contribution from the literature concerns the manipulation of generator frequency to control SSR.
Common expansion space
Isolated working space Common expansion space
Isolated working space
Isol
ated
ele
ctric
al s
yste
ms
Ser
ies
elec
trica
l co
nnec
tion
Par
alle
l el
ectri
cal
conn
ectio
n
APFC
P
APFC
PP
APFC
P
APFC
PP
One APFC controller,Position feedback from one convertor
APFCMaster
Slave
P
APFC
Master
APFCAPFCMaster Slave
P
Master Slave
P
APFCAPFC
SlaveMaster
APFCAPFCMaster Slave
P
Master Slave
PP
APFCAPFC APFC
P P
APFC
Two APFC controllers,Position feedback from one convertor,Current synchronized to other convertor
Two APFC controllers,Position feedback on one convertor,Voltage synchronized to other convertor
Two APFC controllers,Position feedback on both convertors
10.
1. 2.
3. 4.
5. 6.
7. 8.
9.
One APFC controller,Position feedback from one convertor
Isol
ated
ele
ctric
al s
yste
ms
Ser
ies
elec
trica
l co
nnec
tion
Par
alle
l el
ectri
cal
conn
ectio
n
APFC
P
APFC
PPP
APFC
PP
APFC
PP
APFC
P
APFC
PPPP
APFC
PP
APFC
PP
One APFC controller,Position feedback from one convertor
APFCMaster
Slave
P
APFC
Master
APFCMaster
Slave
PPPPP
APFC
Master
APFCAPFCMaster Slave
P
APFCAPFCMaster Slave
PPPP
Master Slave
P
APFCAPFC
SlaveMasterMaster Slave
PPPPP
APFCAPFC
SlaveMaster
APFCAPFCMaster Slave
P
APFCAPFCMaster Slave
PPPP
Master Slave
PP PP PPPPPP
APFCAPFC APFC
P PP PPP PP
APFC
Two APFC controllers,Position feedback from one convertor,Current synchronized to other convertor
Two APFC controllers,Position feedback on one convertor,Voltage synchronized to other convertor
Two APFC controllers,Position feedback on both convertors
10.
1. 2.
3. 4.
5. 6.
7. 8.
9.
One APFC controller,Position feedback from one convertor
Figure 1. Ten candidate configurations of dual-opposed convertors with APFC controllers.
American Institute of Aeronautics and Astronautics
4
II. Configurations of dual-opposed convertors with APFC controllers Ten configurations of convertors and controllers were identified for this study. These ten configurations are
shown in Fig. 1. They use parallel or series electrical connection of the linear alternators, shared or isolated thermodynamic cycles, and single or dual APFC controllers. The first four have a single APFC controller and the last six have two controllers.
All configurations had the cases of the two convertors rigidly connected, so that the cases of the two convertors act dynamically as one in the range of the operating frequency. The rigid case connection provides a direct path for the opposing dynamic forces of the two convertors. The amplitude of the case is generally small and can be influenced by the mounting dynamics between the case and the external system. Stability of the system can be influenced in situations where case amplitude increases significantly. In all configurations simulated the convertor case was connected to ground via a reasonably stiff spring and damper.
A. Thermodynamic configurations Isolated working space.
It is common to configure two convertors with isolated working spaces, meaning that neither the expansion space nor compression space is shared between the two convertors. Common expansion space.
In the common expansion space configuration, the working spaces of the two convertors are connected. This configuration has the rigid case connection discussed previously. Both convertors see the same pressure wave acting on the power pistons, providing a strong coupling between the two convertors. The SPDE is an example of dual-opposed convertors connected with a common expansion space.12 This configuration is shown schematically in Fig. 2.
B. Electrical configurations
1. Series electrical connection With the series electrical connection of the linear alternators, the same current flows through each alternator.
The APFC controller for this configuration commands a single current to the series alternators based on feedback from one or both of the convertors’ piston position signals. With a single controller, it is not possible to independently control the voltage to each convertor. 2. Parallel electrical connection
With the parallel electrical connection, both alternators see the same voltage, but may have different currents. In this case, it is not possible to independently control the current to each convertor. 3. Isolated electrical systems
Electrically isolated convertors were considered where the only connection between the alternators was through two APFC controllers, and there is no direct electrical connection. The controllers must act in such a way so as to maintain synchronization of the two convertors. There are three ways considered in which to maintain synchronization: 1) current synchronization, 2) voltage synchronization, and 3) position feedback on both convertors.
pist
on
1
bounce space 1
linear alternator 1
compression space 1
common expansion
space
cooler 1 heaters 1 & 2regenerator 1
displacer 1 displacer 2pi
ston
2
bounce space 2
regenerator 2cooler 2
linear alternator 2
compression space 2
pist
on
1
bounce space 1
linear alternator 1
compression space 1
common expansion
space
cooler 1 heaters 1 & 2regenerator 1
displacer 1 displacer 2pi
ston
2
bounce space 2
regenerator 2cooler 2
linear alternator 2
compression space 2
Figure 2. Cross-section schematic of dual-opposed convertors with a common expansion space.
American Institute of Aeronautics and Astronautics
5
Configurations #5 through #8 attempted to achieve synchronization by synthesizing the effect of either series-wired or parallel-wired operation of the second convertor. The electrically isolated configurations require two active bridge circuits; one for each convertor. The current-synchronized configurations and the voltage-synchronized configurations require only one set of control electronics for the two active bridges. In these configurations, the current or voltage calculated by the control circuitry was based on the master convertor’s performance and piston position feedback and was used by both active bridges to control both convertors.
Configurations #9 and #10 sense piston position on both convertors. This allows independent control of each convertor based on the separate feedback.
III. Simulation of configurations GRC’s Stirling convertor SDM was used to simulate the ten configurations. The simulations began with
convertors going through start up to full power operation. If the simulations reached steady-state operation stably and synchronized, a 25-N impulse force was applied to the convertor case to determine robustness of the system in response to an outside disturbance.
A. Convertor Model The Stirling convertor models used in this study included the TDC manufactured by Infinia and the ASC
manufactured by Sunpower. These models included the essential dynamics of the convertor, including Stirling cycle thermodynamics; heat flow; gas, mechanical, and mounting dynamics. The SDM includes nonlinearities and subcycle dynamics. The linear alternator was modeled as a voltage source in series with an inductor and resistor.
The common expansion space convertor model used in this study was based on the layout of the SPDE, although the model parameters were based on the TDC. The following assumptions were made:
• Expansion space volume was twice that of a single TDC’s expansion space volume • Heater temperature Th was identical for both convertors • Working fluid was shared between the two convertors. Temperatures in the various convertor volumes
determined actual distribution of the working fluid.
B. APFC Controller Model The APFC controller model used in this study was a simplified representation of an APFC controller. It did not
include actual power electronic components and was not based on an exact APFC controller circuit design. While it would be more accurate to use a detailed APFC controller design for the analysis, the simplified controller was used to reduce computation time. Use of the simplified controller should not influence the conclusions of the study.
The model of the APFC controller used in SDM for this study consisted of a switching/energy storage module that switches positive and negative voltages onto the alternator leads to force the alternator current to a predetermined value. The predetermined value was based on a reference current that was a synthesized sine wave whose amplitude was determined by a control loop designed to maintain a set point of piston amplitude. The amplitude of the current was determined in this study by a control loop to keep piston amplitude at its set point value. If the piston amplitude was greater than the set point, the control loop raised the current amplitude to increase the load and bring the piston amplitude back to its set point. If the piston amplitude was less than the set point, the control loop decreased the current amplitude.
Specification of the phase angle of the reference current was the means whereby the power factor was controlled. To correct the power factor to unity, a phase angle of zero degrees between the current and the velocity was chosen. With a phase angle of zero degrees, the reference current was synthesized to be in phase with the generated electromotive force (EMF). The EMF cannot be measured, but it is known to be in phase with the velocity of the alternator and thus in phase with the piston velocity. The controller model managed the phase relationships by observing the times at which the piston position achieved a cycle minimum and identifying that time as the positive zero-crossing time of piston velocity, designated as T0. The phase angle of the reference current was adjusted each cycle. The reference current synthesizer used in the model was also able to synthesize non-zero phase angles. Such phase angles yield non-unity power factors. In such cases, the reference current was time-shifted from the piston velocity; the current led or lagged the alternator EMF by a known value.
The frequency used by the reference current synthesizer was also determined from observations of the zero crossings of piston position. The reference current frequency was corrected once per cycle.
American Institute of Aeronautics and Astronautics
6
IV. Results of the Study Table II summarizes the ten dual-opposed convertor-APFC controller configurations analyzed using SDM and
identifies which configurations resulted in stable operation. Configurations 2, 4, 9, and 10 showed stable operation while the other six were not found to be stable. In the cases of unstable operation, the piston amplitudes diverged from the nominal equal value with the amplitude of one convertor decreasing and the amplitude of the other convertor growing. The piston amplitude that grew appeared to grow without bound until the physical limits of the hardware were reached. After the piston amplitudes diverged, there did not appear to be any tendency for the piston amplitudes to come back in range.
Three of the four stable configurations were convertors with a common expansion space, while there was only one stable configuration with isolated working spaces. Comparing configurations #1 and #3 versus configurations #2 and #4, the only difference is that the former have isolated working spaces, while the latter have a common expansion space. This suggests that the stabilizing and synchronizing effect of the common expansion space was significant. With a common expansion space, both convertors essentially see the same pressure wave. The same pressure wave acts on both pistons, largely independent of differences in convertor parameters, which tends to synchronize their motions. A similar stabilizing effect is expected with convertors with a common compression space, but that was not simulated.
The stable configuration with isolated working spaces requires two controllers and two position sensors with the slave side of the system essentially being directed to mimic the piston position of the master side of the system. This configuration is more complex in that it requires two APFC controllers and two position sensors.
Table II. Summary of simulation results.
Configuration no.
Thermodynamic configuration
Electrical configuration
Controller Simulation results
1 Isolated working space Unstable
2 Common expansion space
Series electrical connection
One APFC controller, position feedback from one convertor Stable
3 Isolated working space Unstable
4 Common expansion space
Parallel electrical
connection
One APFC controller, position feedback from one convertor Stable
5 Isolated working space Unstable
6 Common expansion space
Two APFC controllers, position feedback from one convertor, current synchronized to other convertor Unstable
7 Isolated working space Unstable
8 Common expansion space
Two APFC controllers, position feedback from one convertor, voltage synchronized to other convertor Unstable
9 Isolated working space Stable
10 Common expansion space
Isolated electrical systems
Two APFC controllers, position feedback on both convertors Stable
A. Configurations #1 and #2 with series electrical connection The SPDE system was configured with the two alternators wired in series, along with a passive controller and
tuning capacitor. This may have fostered a notion that series wiring is a viable configuration to consider for other free-piston convertors where the tuning capacitor and passive controller are replaced by an APFC controller. However, SDM simulations suggest that an APFC controller and alternators connected in series will not stabilize convertors with isolated working spaces (configuration #1). Convertors with common expansion space like the SPDE appear stable when connected in series with an APFC controller (configuration #2).
B. Configurations #3 and #4 with parallel electrical connection As mentioned earlier, many dual-opposed convertor systems are currently tested at GRC connected in parallel
with tuning capacitors and passive controllers. Simulations of configuration #3 showed that convertors with isolated
American Institute of Aeronautics and Astronautics
7
working spaces were unstable when connected in parallel with an APFC controller. The similar results have been observed in the test cell when operation was attempted without tuning capacitors.
When tuning capacitors are removed from the parallel circuit, it appears that the current to the two convertors is split in such a way that it tends to destabilize the system instead of stabilizing the system. With real hardware one alternator’s EMF will be slightly larger than the other’s EMF because no two convertors can be perfectly matched. With the alternators directly connected to the controller in parallel, the larger portion of the total current goes to the
alternator with the smaller voltage. The higher current tends to apply more damping to that alternator, which further reduces that convertor piston’s amplitude. That in turn reduces the alternator’s EMF, making subsequent amplitudes even smaller. The alternator with the larger amplitude sees less damping, allowing its amplitude to grow. This is theorized to be the cause of instability observed in some of the parallel configurations studied.
Configuration #4 was the same as configuration #3, but with a common expansion space. This configuration was found to be stable. In some simulations a small limit cycle was observed, as shown in Fig. 3. Further investigation is needed to understand the source of the limit cycle and identify methods to eliminate it.
C. Configurations #5 through #10 with electrically isolated alternators Simulations of the configurations with electrically isolated alternators (configurations #5 through #10) identified
only configurations #9 and #10 as being stable. Current-synchronized electrically-isolated configurations #5 and #6 essentially mimic series-wired
configurations #1 and #2, since the alternators arranged in current-synchronized configurations see a similar current as if they were actually wired in series. No stable mode was found in simulation with either the common expansion space configuration #5 or the isolated working space configuration #6, even though configuration #2 was found to be stable. Further investigation is needed to understand why configuration #6 was unstable, and if there are cases for which it might be stable.
Figure 4 is a typical case position versus time plot of a simulation of configuration #5. Typically in the simulations one of the piston amplitudes was unable to be controlled and the piston hit the end stop, indicating instability.
91.00
100.00
95.00
97.50
0 660.00m250.00m 500.00m
Displ Phase Angles
Conv.... Conv....
-16.00u
15.80u
0
0 660.00m250.00m 500.00m
Case Motion
Case...
Figure 3. Case motion and displacer phase angles from a simulation of Configuration #4. Single APFC controller parallel connected common expansion space convertors.
American Institute of Aeronautics and Astronautics
8
The case motion shown in Fig. 4 did play a role in how fast the system became unstable and when the piston of
one convertor hit the end stop. Depending on how the convertors were mismatched and whether the natural frequency of the mounting system was above or below the convertor operating frequency, the case motion would either provide positive feedback or negative feedback to the convertor with the increasing piston amplitude. Regardless of the nature of the feedback, the system was still unstable, as the case motion plays only a small role in the system dynamics.
Voltage-synchronized electrically-isolated configurations #7 and #8 are similar to the parallel-connected configurations #3 and #4. As with configuration #3, configuration #7 was also found to be unstable. Configuration #8 was also unstable, although configuration #4 was stable. Further investigation is recommended here.
Stable modes were found for configurations #9 and #10. In these configurations, the two active bridges were controlled independently. Feedback of piston position from both convertors allowed the controllers to determine the current phase relative to piston velocity for each convertor, as well as the relative piston phasing between the two convertors. By independently adjusting the current phasing to each convertor, the desired piston phasing and amplitude can be achieved, resulting in stable and synchronized operation of the dual-opposed convertors.
The APFC controllers can provide more flexibility than tuning capacitors to account for changes in load impedance over time or to optimize system efficiency and performance based on various criteria such as power, efficiency, or vibration. These options are being developed at GRC.
V. Conclusions The stability of ten configurations of dual-opposed Stirling convertors controlled by APFC controllers was
studied through dynamic simulation. Different thermodynamic, electrical, and controller configurations were studied. Four configurations were found to stabilize and synchronize the two convertors.
Of the five configurations with isolated working spaces, only one configuration was stable, configuration #9, with isolated electrical systems, two APFC controllers, and position feedback from both convertors. Configurations that relied on position feedback from just one convertor or connected the convertors electrically in series or in parallel were not able to be stabilized.
Three of the configurations with a common expansion space were stable. This suggests that the stabilizing and synchronizing effect of the common expansion space is significant. A similar stabilizing effect is expected with convertors with a common compression space. Further investigation is recommended to understand why stable modes were not identified for the other two configurations with a common expansion space and to determine if stable modes for these configurations do exist. Analysis has shown that operation can be affected to influence power, efficiency, or vibration. Many control algorithms have been tested analytically with SDM to find the simplest and most robust control strategies.
Acknowledgments This work was performed for NASA Headquarters, Science Mission Directorate (SMD) under the Radioisotope
Power Systems (RPS) Program and was supported by the NASA Glenn Research Center Contract NAS3-03064. Any opinions, findings, and conclusions or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of the National Aeronautics and Space Administration.
Figure 4. Increasing case motion from Configuration #5, indicating system instability.
-46.50u
47.50u
-25.00u
0
25.00u
0 1.501.00
Case Motion
Mount...
American Institute of Aeronautics and Astronautics
9
References 1Brown, A. T., “Space Power Demonstrator Engine”, Phase I Final Report, Mechanical Technology, Inc., Latham NY,
NASA Contractor Report 179555, May 1987. 2Gerber, S.S., Jamison, M., Regan, T.F., and Roth, M.E., “Advanced Controller for the Free-Piston Stirling Convertor,”
Proceedings of the Second International Energy Conversion Engineering Conference (IECEC 2004), AIAA, Providence, RI, 2004.
3Daboussi, Z., “An Inverter-Based Sensorless Controller for Free-Piston Stirling Engines,” Proceedings of the 35th Annual IEEE Power Electronics Specialists Conference, Aachen, Germany, 2004, pp. 1707-1710.
4Lynch, T.H. and Koch, B., “Thermal Cycle Engine Boost Bridge Power Interface,” U.S. Patent 6,871,495, Mar. 29, 2005. 5Holliday, E. and Keiter, D.E., “Control Electronics for Palm Power 35W Free-Piston Stirling Engine,” Proceedings of the
Third International Energy Conversion Engineering Conference (IECEC 2005), AIAA, San Francisco, CA, 2005. 6Lewandowski, E.J. and Regan, T.F., “Overview of the GRC Stirling Convertor System Dynamic Model,” Proceedings of the
Second International Energy Conversion Engineering Conference (IECEC 2004) AIAA, Providence, RI, 2004. 7Regan, T.F., Lewandowski, E.J., “Application of the GRC Stirling Convertor System Dynamic Model,” Proceedings of the
Second International Energy Conversion Engineering Conference (IECEC 2004) AIAA, Providence, RI, 2004. 8Qiu, S., Augenblick, J.E., White, M.A., Peterson, A.A., Redinger, D.L. and Peterson, S.L., Developing a Free-Piston Stirling
Convertor for Advanced Radioisotope Space Power Systems,” in Proceedings of the Space Technology and Applications International Forum, (STAIF 2002), American Physical Society.
9Wood, J.G., Carroll, C., Matejczyk, D., and Penswick, L.B., “Advanced 80 We Stirling Convertor Phase II Development Progress,” Proceedings of the Third International Energy Conversion Engineering Conference (IECEC 2005), AIAA, San Francisco, CA, 2005.
10White, M. A. and Qiu, S., “Innovative Integration of Long-Life High Efficiency Thermal Convertors Using Proven Free-Piston Stirling Machines”, SBIR Phase II Final Technical Report, Stirling Technology Co. (currently known as Infinia Corp.) Kennewick, WA, 1999.
11IEEE SSR Working Group, "Proposed Terms and Definitions for Subsynchronous Resonance," IEEE Symposium on Countermeasures for Subsynchronous Resonance, IEEE Pub. 81TH0086-9-PWR, 1981, pp. 92-97.
12Dochat, G.R., “Free piston space Stirling technology program,” Proceedings of the 24th Intersociety Energy Conversion Engineering Conference (IECEC-89) IEEE, Washington, DC, Aug 6-11, 1989.
American Institute of Aeronautics and Astronautics
1
Control of Dual-Opposed Stirling Convertors with Active Power Factor Correction Controllers
Timothy F. Regan* and Edward J. Lewandowski† Sest, Inc. Middleburg Heights, Ohio 44130
Jeffrey G. Schreiber‡ NASA Glenn Research Center, Cleveland, Ohio 44135
When using recently-developed active power factor correction (APFC) controllers in power systems comprised of dual-opposed free-piston Stirling convertors, a variety of configurations of the convertors and controller(s) can be considered, with configuration ultimately selected based on benefits of efficiency, reliability, and robust operation. The configuration must not only achieve stable control of the two convertors, but also synchronize and regulate motion of the pistons to minimize net dynamic forces. The NASA Glenn Research Center (GRC) System Dynamic Model (SDM) was used to study ten configurations of dual-opposed convertor systems. These configurations considered one controller with the alternators connected in series or in parallel, and two controllers with the alternators not connected (isolated). For the configurations where the alternators were not connected, several different approaches were evaluated to synchronize the two convertors. In addition, two thermodynamic configurations were considered: two convertors with isolated working spaces and convertors with a shared expansion space. Of the ten configurations studied, stable operating modes were found for four. Three of those four had a common expansion space. One stable configuration was found for the dual-opposed convertors with separate working spaces. That configuration required isochronous control of both convertors, and two APFC controllers were used to accomplish this. A frequency/phase control loop was necessary to allow each APFC controller to synchronize its associated convertor with a common frequency.
* Sr. Aerospace Engineer, 18000 Jefferson Park Rd, Middleburg Heights, OH 44130, AIAA Member. † Sr. Aerospace Engineer, 18000 Jefferson Park Rd, Middleburg Heights, OH 44130, AIAA Member. ‡ Mechanical Engineer, Thermal Energy Conversion Branch, M/S 301-2, 21000 Brookpark Rd, Cleveland, OH 44135, AIAA Member
1
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Con
trol
of D
ual-O
ppos
ed S
tirlin
g C
onve
rtor
s w
ith A
ctiv
e Po
wer
Fac
tor
Cor
rect
ion
Con
trol
lers
4th
Inte
rnat
iona
l Ene
rgy
Con
vers
ion
Engi
neer
ing
Con
fere
nce
(IEC
EC).
26 -
29 J
un 2
006
San
Die
go, C
alifo
rnia
Tim
othy
F. R
egan
and
Edw
ard
J. L
ewan
dow
ski
Sest
, Inc
. Mid
dleb
urg
Hei
ghts
, Ohi
o 44
130
Jeffr
ey G
. Sch
reib
erN
ASA
Gle
nn R
esea
rch
Cen
ter,
Cle
vela
nd, O
hio
4413
5
2
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Out
line:
AP
FC C
ontro
l of D
ual-O
ppos
ed C
onve
rtors
•B
ackg
roun
d–
Why
impo
rtant
to e
valu
ate
•AP
FC c
ontro
llers
cou
ld re
plac
e tu
ning
cap
acito
rs in
mul
ti-co
nver
tor
syst
ems
•m
ass
redu
ctio
n•
relia
bilit
y–
Rel
ated
wor
k•
ME
GS
•SP
DE
•S
DM
Stu
dy–
Gle
nn R
esea
rch
Cen
ter’s
Sys
tem
Dyn
amic
Mod
el --
SD
M–
10 c
andi
date
arc
hite
ctur
es•
Res
ults
•C
oncl
usio
ns•
Ack
now
ledg
men
ts
3
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Bac
kgro
und
•A
pplic
atio
n: S
pace
pow
er s
yste
m–
Hig
h sp
ecifi
c po
wer
–
Hig
h re
liabi
lity
•A
PFC
con
trolle
r res
ults
com
pare
d to
tuni
ng c
apac
itor
–O
utpu
t pow
er–
Effi
cien
cy–
Mas
s–
Rel
iabi
lity
4
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Rel
ated
Wor
k
•S
PD
E M
TI–
Des
ign
and
oper
atio
n of
inte
grat
ed m
ultip
le fr
ee-p
isto
n S
tirlin
g co
nver
tors
.•
SP
DE
is a
25
kW d
ual-o
ppos
ed c
onve
rtor w
ith a
com
mon
w
orki
ng s
pace
.•
Tuni
ng c
apac
itor w
as u
sed
in s
erie
s al
tern
ator
con
nect
ion.
•M
EG
S (M
. Whi
te, I
nfin
ia C
orp,
199
9)–
Sim
ulat
ion
and
test
ing
of th
erm
odyn
amic
ally
inde
pend
ent
mul
tiple
free
-pis
ton
Stir
ling
conv
erto
rs.
•S
tudy
was
bas
ed o
n du
al-o
ppos
ed 3
50 W
con
verto
rs.
•Fi
ndin
g: V
iabl
e co
nfig
urat
ion
of d
ual o
ppos
ed s
yste
m w
ith
tuni
ng c
apac
itors
was
iden
tifie
d.
•In
dust
rial p
ower
sys
tem
s–
Des
ign,
sim
ulat
ion,
ope
ratio
n of
pow
er s
yste
ms
with
m
ultip
le, p
aral
lele
d 3-
phas
e en
gine
gen
erat
or s
ets.
5
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
onfig
urat
ion
Stu
dy u
sing
SD
M
•S
DM
is th
e fre
e-pi
ston
con
verto
r sim
ulat
ion
envi
ronm
ent d
evel
oped
at G
RC
–D
ynam
ics
–Th
erm
odyn
amic
s–
Ele
ctro
nics
–M
ultip
le c
onve
rtor s
yste
m s
tudi
es•
SD
M c
onve
rtor m
odel
–Th
erm
odyn
amic
ally
cou
pled
(mod
el b
asis
: SP
DE
co
nfig
urat
ion
with
110
W o
utpu
t)–
Ther
mod
ynam
ical
ly is
olat
ed (2
TD
C’s
and
2 A
SC
’s)
•S
DM
con
trolle
r mod
el–
Cur
rent
mod
e co
ntro
l–
DC
bus
bul
k st
orag
e ca
paci
tor i
s m
odel
ed a
s D
C v
olta
ge
sour
ce
6
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
onfig
urat
ion
Stu
dy u
sing
SD
M
Dual SCA
Dual SCA
Ver. 3.00Common exp.
Dual SCA
Xpfv2
.0
|Xp| Xpf
AP
FC
TDC
55v1
.698
TDC
55v1
.698
5, 6
, 7, 8
•TD
C
•A
SC
•D
ual e
xpan
sion
spa
ce “T
DC
”
•A
PFC
v2.
0
piston 1
boun
ce
spac
e 1
linea
r al
tern
ator
1co
mpr
essi
on
spac
e 1
com
mon
ex
pans
ion
spac
e
cool
er 1
heat
ers
1 &
2re
gene
rato
r 1
disp
lace
r 1di
spla
cer 2
piston 2
boun
ce
spac
e 2
rege
nera
tor 2
cool
er 2
linea
r al
tern
ator
2co
mpr
essi
on
spac
e 2
piston 1
boun
ce
spac
e 1
linea
r al
tern
ator
1co
mpr
essi
on
spac
e 1
com
mon
ex
pans
ion
spac
e
cool
er 1
heat
ers
1 &
2re
gene
rato
r 1
disp
lace
r 1di
spla
cer 2
piston 2
boun
ce
spac
e 2
rege
nera
tor 2
cool
er 2
linea
r al
tern
ator
2co
mpr
essi
on
spac
e 2
AS
C V
er 1
.2
11.5
Vol
t
3594
draw
ings
7
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Syn
chro
niza
tion
•S
ingl
e C
ontro
ller
–P
aral
lel w
iring
–S
erie
s w
iring
•D
ual C
ontro
ller
–C
urre
nt s
ourc
e th
at s
ynth
esiz
es id
entic
al c
urre
nt–
Vol
tage
sou
rce
that
syn
thes
izes
iden
tical
vol
tage
–C
ontro
l eac
h co
nver
tor i
ndiv
idua
lly
8
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
onfig
urat
ion
Stu
dy u
sing
SD
M
Com
mon
exp
ansi
on
spac
eIs
olat
ed w
orki
ng s
pace
Com
mon
exp
ansi
on
spac
eIs
olat
ed w
orki
ng s
pace
Isolated electrical systemsSeries electrical connection
Parallel electrical connection
APFC
P
APFC
P P
APFC
P
APFC
P P
One
APF
C c
ontro
ller,
Posi
tion
feed
back
fr
om o
ne c
onve
rtor
APFC
Mas
ter
Slav
e
P
APFC
Mas
ter
APFC
APFC
Mas
ter
Slav
e
P
Mas
ter
Slav
e
P
APFC
APFC
Slav
eM
aste
r
APFC
APFC
Mas
ter
Slav
e
P
Mas
ter
Slav
e PP
APFC
APFC
APFC
PP
APFC
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck fr
om
one
conv
erto
r,C
urre
nt sy
nchr
oniz
ed
to o
ther
con
verto
r
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck o
n on
e co
nver
tor,
Vol
tage
sync
hron
ized
to
oth
er c
onve
rtor
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck o
n bo
th c
onve
rtors
10.
1.2.
3.4.
5.6.
7.8.
9.
One
APF
C c
ontro
ller,
Posi
tion
feed
back
fr
om o
ne c
onve
rtor
Isolated electrical systemsSeries electrical connection
Parallel electrical connection
APFC
P
APFC
PPP
APFC
P P
APFC
P P
APFC
P
APFC
PPPP
APFC
P P
APFC
P P
One
APF
C c
ontro
ller,
Posi
tion
feed
back
fr
om o
ne c
onve
rtor
APFC
Mas
ter
Slav
e
P
APFC
Mas
ter
APFC
Mas
ter
Slav
e
PPPPP
APFC
Mas
ter
APFC
APFC
Mas
ter
Slav
e
P
APFC
APFC
Mas
ter
Slav
e
PPPP
Mas
ter
Slav
e
P
APFC
APFC
Slav
eM
aste
rM
aste
rSl
ave
PPPPP
APFC
APFC
Slav
eM
aste
r
APFC
APFC
Mas
ter
Slav
e
P
APFC
APFC
Mas
ter
Slav
e
PPPP
Mas
ter
Slav
e PP
PP
PPPPPP
APFC
APFC
APFC
PP
PP
PPPP
APFC
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck fr
om
one
conv
erto
r,C
urre
nt sy
nchr
oniz
ed
to o
ther
con
verto
r
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck o
n on
e co
nver
tor,
Vol
tage
sync
hron
ized
to
oth
er c
onve
rtor
Two
APF
C c
ontro
llers
,Po
sitio
n fe
edba
ck o
n bo
th c
onve
rtors
10.
1.2.
3.4.
5.6.
7.8.
9.
One
APF
C c
ontro
ller,
Posi
tion
feed
back
fr
om o
ne c
onve
rtor
9
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
1 –
Sin
gle
AP
FC–
Ser
ies
elec
trica
l con
nect
ion
–Th
e cu
rren
t fai
ls to
kee
p th
e co
nver
tors
syn
chro
nize
d.–
The
refe
renc
e cu
rren
t m
aint
ains
a fi
xed
phas
e re
latio
nshi
p w
ith o
ne
conv
erto
r’s v
eloc
ity.
The
rela
tions
hip
to th
e ot
her i
s un
defin
ed.
APFC
P
APFC
PPPP
10
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
2 (S
tabl
e)
–S
ingl
e A
PFC
–S
erie
s el
ectri
cal c
onne
ctio
n–
The
pres
sure
wav
e pr
ovid
es
stro
ng e
noug
h co
uplin
g be
twee
n th
e pi
ston
s th
at th
e co
nver
tors
syn
chro
nize
d.
APFC
P
11
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
3 –
Sin
gle
AP
FC–
Par
alle
l ele
ctric
al c
onne
ctio
n–
Uns
tabl
e af
ter 0
.4s.
–Th
e co
nver
tor w
ith p
ositi
on
info
rmat
ion
rem
ains
con
stan
t am
plitu
de.
The
conv
erto
r with
no
pos
ition
info
rmat
ion
is
driv
en in
to th
e en
d st
op.
APFC
P
12
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
4 (S
tabl
e)–
Sin
gle
AP
FC–
Par
alle
l ele
ctric
al c
onne
ctio
n–
Bes
t per
form
ance
ach
ieve
d in
si
mul
atio
n fe
atur
ed li
mit-
cycl
e st
abili
ty a
s sh
own.
APFC
P
-16.
00u
15.8
0u 0
066
0.00
m25
0.00
m50
0.00
m
Cas
e M
otio
n
C
ase.
..
91.0
0
100.
00
95.0
0
97.5
0
066
0.00
m25
0.00
m50
0.00
m
Dis
pl P
hase
Ang
les
C
onv..
..
Con
v....
13
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
5–
Dua
l AP
FC (C
urre
nt
Sou
rce
Sla
ve)
–Is
olat
ed e
lect
rical
–S
tabl
e re
sults
with
ba
lanc
ed c
onve
rtors
and
fix
ed m
ount
ing
–U
nsta
ble
with
the
slig
htes
t im
bala
nce
APFC
Mas
ter
Sla
ve
P
APFC
14
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
6–
Dua
l AP
FC (C
urre
nt
Sou
rce
Sla
ve)
–Is
olat
ed e
lect
rical
–E
xpec
ted
resu
lts s
imila
r to
Con
figur
atio
n 2,
(sta
ble,
se
ries-
conn
ecte
d si
ngle
A
PFC
con
trolle
r)–
Una
ble
to s
tabi
lize
the
syst
em.
APFC
APFC
Mas
ter
Sla
ve
P
15
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
7–
Dua
l AP
FC (V
olta
ge S
ourc
e S
lave
)–
Isol
ated
ele
ctric
al–
Sta
ble
resu
lts fo
r per
fect
ly
bala
nced
con
verto
rs.
–U
nsta
ble
for
•3˚
diffe
renc
e in
hot
-end
tem
pera
ture
•2%
diff
eren
ce in
mot
or c
onst
ant
Mas
ter
Sla
ve
P
APFC
APFC
16
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
8–
Dua
l AP
FC (V
olta
ge S
ourc
e S
lave
)–
Isol
ated
ele
ctric
al–
Exp
ecte
d re
sults
sim
ilar t
o C
onfig
urat
ion
4 (s
tabl
e,
para
llel-c
onne
cted
sin
gle
AP
FC)
–U
nabl
e to
sta
biliz
e th
is
syst
em.
–S
ame
limit-
cycl
e be
havi
or b
ut
the
case
mot
ion
ampl
itude
gr
ows
until
the
syst
em p
ulls
ap
art.
APFC
APFC
Mas
ter
Sla
ve
P
17
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
9 (S
tabl
e)–
Dua
l AP
FC–
Isol
ated
ele
ctric
al–
Inde
pend
ent A
PFC
con
trol
–U
sed
frequ
ency
con
trol
loop
to in
sure
iden
tical
op
erat
ing
frequ
ency
for
both
con
verto
rs.
–P
isto
ns n
ot a
lway
s in
180
-de
gree
opp
ositi
on.
Res
ulte
d in
som
e ca
se
mot
ion.
–A
dditi
onal
con
trol n
eede
d fo
r 180
-deg
ree
pist
on
oppo
sitio
n.
PP
APFC
APFC
18
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
AP
FC C
ontro
ller A
rchi
tect
ure
Stu
dies
•C
onfig
urat
ion
10 (S
tabl
e)–
Dua
l AP
FC–
Isol
ated
ele
ctric
al–
Inde
pend
ent A
PFC
con
trol
–S
ame
resu
lts a
s C
onfig
urat
ion
9
APFC
PP
APFC
19
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Res
ults
Con
figur
atio
n no
. T
herm
odyn
amic
co
nfig
urat
ion
Ele
ctri
cal
conf
igur
atio
n C
ontr
olle
r Si
mul
atio
n re
sults
1 Is
olat
ed w
orki
ng
spac
e U
nsta
ble
2 C
omm
on e
xpan
sion
sp
ace
Serie
s ele
ctric
al
conn
ectio
n O
ne A
PFC
con
trolle
r, po
sitio
n fe
edba
ck fr
om o
ne c
onve
rtor
Stab
le
3 Is
olat
ed w
orki
ng
spac
e U
nsta
ble
4 C
omm
on e
xpan
sion
sp
ace
Para
llel
elec
trica
l co
nnec
tion
One
APF
C c
ontro
ller,
posi
tion
feed
back
from
one
con
verto
r St
able
5 Is
olat
ed w
orki
ng
spac
e U
nsta
ble
6 C
omm
on e
xpan
sion
sp
ace
Two
APF
C c
ontro
llers
, pos
ition
fe
edba
ck fr
om o
ne c
onve
rtor,
curr
ent s
ynch
roni
zed
to o
ther
co
nver
tor
Uns
tabl
e
7 Is
olat
ed w
orki
ng
spac
e U
nsta
ble
8 C
omm
on e
xpan
sion
sp
ace
Two
APF
C c
ontro
llers
, pos
ition
fe
edba
ck fr
om o
ne c
onve
rtor,
volta
ge s
ynch
roni
zed
to o
ther
co
nver
tor
Uns
tabl
e
9 Is
olat
ed w
orki
ng
spac
e St
able
10
Com
mon
exp
ansi
on
spac
e
Isol
ated
el
ectri
cal
syst
ems
Two
APF
C c
ontro
llers
, pos
ition
fe
edba
ck o
n bo
th c
onve
rtors
St
able
20
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Con
clus
ions
•Te
n co
nfig
urat
ions
for d
ual-o
ppos
ed A
PFC
con
trolle
d co
nver
tor
syst
ems
have
bee
n st
udie
d th
roug
h si
mul
atio
n.
•O
ne c
onfig
urat
ion
has
been
iden
tifie
d as
a w
orki
ng c
onfig
urat
ion
for t
herm
odyn
amic
ally
-isol
ated
con
verto
rs.
•S
tabi
lizin
g an
d sy
nchr
oniz
ing
effe
ct o
f the
com
mon
wor
king
sp
ace
is s
igni
fican
t•
Furth
er in
vest
igat
ion
is re
com
men
ded
to u
nder
stan
d w
hy s
tabl
e m
odes
wer
e no
t fou
nd fo
r “se
ries-
like”
and
“par
alle
l-lik
e”co
nfig
urat
ions
of t
he c
omm
on e
xpan
sion
spa
ce c
onve
rtor.
•C
onve
rtor o
pera
tion
can
be a
ffect
ed b
y A
PFC
con
trol s
trate
gies
to
influ
ence
pow
er, e
ffici
ency
, or v
ibra
tion.
21
Nat
iona
l Aer
onau
tics
and
Spac
e Ad
min
istra
tion
ww
w.n
asa.
gov
Ack
now
ledg
emen
ts
This
wor
k w
as p
erfo
rmed
for N
AS
A H
eadq
uarte
rs,
Sci
ence
Mis
sion
Dire
ctor
ate
(SM
D) u
nder
the
Rad
iois
otop
e P
ower
Sys
tem
s (R
PS
) Pro
gram
and
w
as s
uppo
rted
by th
e N
AS
A G
lenn
Res
earc
h C
ente
r C
ontra
ct N
AS
3-03
064.
Any
opi
nion
s, fi
ndin
gs, a
nd
conc
lusi
ons
or re
com
men
datio
ns e
xpre
ssed
in th
is
artic
le a
re th
ose
of th
e au
thor
s an
d do
not
ne
cess
arily
refle
ct th
e vi
ews
of th
e N
atio
nal
Aer
onau
tics
and
Spa
ce A
dmin
istra
tion.