Advances in flywheel performance for space power applications M. Olszewski and D. U. O’Kain Power derived from stored energy rather than directly from the primary thermal source is an attractive alternative in certain space applications. To meet the stringent requirements for space needs will, however, require that flywheel performance be advanced from its current levels. Previous flywheel development programmes focused on the use of composite materials and resulted in storage densities, at ultimate speed, of 286 kJ/kg being attained. More recent work, directed at the space application, has advanced flywheel storage densities at ultimate speed to 878 kJ/kg.
Kinetic energy stored in a flywheel is a relatively old technology. The earliest known application, for instance, is the potter’s wheel. A more recent example is found in the modern automobile. The internal combustion engine operates smoothly because a flywheel incorporated in the drivetrain stores the energy that is supplied intermittently by explosions in the cyclinders. The action of the flywheel thus smooths out the variable power input. The basic principle of flywheel energy storage is that a rotating wheel represents stored mechanical, or kinetic, energy. The amount of energy stored depends upon the inertia of the wheel and its rotational speed. Energy is added by increasing wheel speed. Energy is withdrawn when the wheel speed is decreased, typically by coupling the wheel to a load.
Space power needs Energy storage may prove to be a critical component of the energy supply system for space applications. The specific role played by energy storage is determined by the characteristics of the mission needs. In applications where the primary energy source is solar, energy storage will be required to maintain operation of the space platform (such as NASA’s planned space station) during the dark portion of the orbit. In applications related to Strategic Defense Initiative (SDI) missions, the duty cycle may involve periods of high power demands for short times interspersed with much longer periods in which power demands are at a much lower level. An energy storage system interposed between the prime power system and the energy user would mitigate problems associated with rapid ramping to full power. Flywheels offer a number of advantages that make them attractive for energy storage applications in space Mitchell Olszewski, B.S., MS., Ph.D. power systems. They possessrelatively Completed undergraduate and graduate work high energy storage densities, which at Rutgers University and received M.S. degree results in relatively low storage system in Mechanical Engineering in 1972. Received a Ph.D. degree in Mechanical Engineering from mass. Also, the power density of the the University of California, Berkeley in 1975. flywheel is independent of the energy He is now in the Engineering Technology density and is fixed by the output Division of the Oak Ridge National Laboratory device rather than the flywheel itself. and has conducted research in various energy fields including beneficial uses of reject heat The flywheel can be charged slowly and and energy conservation using thermal and discharged rapidly (or vice versa) mechanical storage techniques. His primary without adversely affecting storage or interest is the development of.energy storage power density. Thus, the requirements and conversion technologies for space power applications. of a variety of duty cycles can be met using a standard flywheel design. David U. O’Kain, B.S., MS. Because of these operating characReceived his B.S. degree in Engineering teristics, flywheels have been given Physics and his M.S. degree in Engineering serious consideration for space misMechanics from the University of Tennessee. He has primarily been involved in the design sions over the past several years. and development of high-speed rotating Several workshops have been sponequipment. He currently serves as the Technolsored by NASA [l, 21 and they are ogy Manager for energy storage activities interested in a concept that would be within the ETAC organization. used on the space station that inteEnduvour, New Swin. Volumr 11, No. 2,1987. grates the energy storage and attitude 01ao-a327/a7 00.00 + 30. @1887.PW~WIIOII JOWIAS ~td. mintedin Great Britsincontrol functions [S]. In addition, 58
flywheels have been proposed for multimegawatt SD1 applications [4, 5,
61.
Flywheel storage modules proposed for space applications consist of several major components. Energy is stored in two counter-rotating flywheel rotors aligned on a common axis. In this configuration the momentum vectors of the two flywheels counterbalance, and the platform experiences a zero net momentum vector. The flywheels are magnetically suspended to isolate them vibrationally from the platform and allow for dynamic control of the bearing stiffness. As a baseline, it is assumed that a tandem configuration is used. In this system the flywheel shaft has on one end a motor that is used to charge the flywheel. On the other end of the shaft is a generator that is used to extract energy from the flywheel. This configuration gives the designer flexibility since various combinations of input and output devices can be used with the same flywheel. An alternative configuration, of interest for the NASA integrated function concept, is a concentric configuration. In this design a thick rim rotor is used and the suspension system and the motor/ generator are placed in the flywheel bore. Although this system concept is less flexible, in terms of design options, than the tandem design, it is volumetrically efficient. Early development work Flywheel storage density is a function of geometry and material properties. As shown in Table 1, it is necessary to use flywheels made of composite materials to achieve energy storage densities that are suitable for space applications. Composite flywheel technology was initially established in the Mechanical Energy Storage Technology (MEST) Program conducted by the Oak Ridge National Laboratory [7]. Performance
testing during the first phase of the programme concentrated on ultimate speed evaluations. The purpose of this testing regime was to obtain energy density data at the maximum wheel speed and determine the failure mode that acted as the limiting factor for the design. The flywheels were generally of the disk and rim configuration. Representative examples are shown in figures 1 and 2. The rim design illustrated in figure 1 (designated as (G-G) consists of a subcircular rim of Kevlar and S-glass mounted on a graphite/epoxy composite crucible hub. The rim is composed of concentric unbonded rings. The design shown in figure 2 (and designated as GEB) is a constant thickness disk surrounded by a circumferentially wound graphite/epoxy rim. The disk uses an alpha-ply layup of
Figure 1
Subcircular
Figure 3
Overwrap
rim flywheel
flywheel
design.
design.
TABLE 1
CHARACTERISTICS
OF MATERIALS
USED IN FLYWHEELS
Ultimate tensile strength (a) MPa
Density (p) g/cm3
Steels 4340 18 Ni (300)
1517 2070
7.7 8.0
Composites E-glass/epoxy S-glass/epoxy Kevlaf epoxy Graphite epoxy
1379 2069 1930 1586
1.9 1.9 1.4 1.5
726 1089 1379 1057
(201.6) (302.5) (382.9) (293.7)
Other METGLASSb
2627
8.0
328
f 91.1)
Material
a Kevlar is a trademark of Du Pont. b METGLASS is a registered trademark
of the Allied
Figure 2
Hybrid disk/rim
Figure 4
Composite
material
197 25.9
Corporation,
flywheel
a [kJ/kg (Wh/kg)]
( 54.7) ( 71.8)
Morristown,
design.
spoke system design.
N.J.
TABLE 2
Manufacturer
PERFORMANCE
RESULTS FOR INITIAL ULTIMATE
Wheel type
Materiala
SPEED CONFIGURATION
TESTS
Energy density at Maximum speed
Energy stored (MJ)
(KJW ORNL Brobeck Garrett AiResearch Rocketdyne APL-Metglass Hercules AVCO LLNL LLNL GE Owens/Lord
a Material
Overwrap Rim Rim
K49 SGIK49 K49iK29lSG
178 229 286
2.02. 2.55 4.43
Overwrap Rim Rim Disk (Contoured Pierced) Disk (Pierced) Disc (Tapered) Disk (Flat) Disk (Solid/ Ring) Disk Disk/Ring
G
143 81 135
7.67 0.14 3.06
EG SGIG
158 225 242 198
1.44 1.12 0.58 1.01
SMC SMCIG SMCIG SMCIG
63 90 100 132
0.61 1.01 1.30 1.44
legend is: SG = S-glass; compounding.
SG
K49 = Levlar 49; K29 = Kevlar 29; G = Graphite;
uniaxial SZglass preimpregnated with resin (prepreg). Other more complex designs were also examined. These included concepts such as over-wrap designs (figure 3) and rim configurations employing complex spoke systems (figure 4). The results of these initial ultimate speed tests are presented in Table 2. The energy density and stored energy figures represent the wheel’s capability at the maximum speed attained in the test. Actual operational values would be lower. As shown in the table, a variety of design concepts and materials of construction were employed. The highest ultimate energy density achieved was 286 kJ/kg with the subcircular rim design. The highest stored energy, 2.13 kWh, was achieved by the overwrap configuration. For the next phase of work in the MEST Program, the field of candidate rotors was narrowed and the testing TABLE 3
G”
FATIGUE AND ULTIMATE
were resumed by the Enrichment Technology Applications Center (ETAC) in Oak Ridge in 1985. Significant advances in flywheel technology have been achieved through precision fabrication of carbon fibrelepoxy material. A series of spin tests was performed using sections of a carbon/ epoxy cylinder which was constructed by filament-winding circumferentially oriented fibres. The fibres used in the fabrication were Hercules IM6 and AS6, and the resin system was ERL2258. The fibres were wet-wound and cured on a 0.61-m (24-inch) diameter mandrel. The outside diameter of the fabricated cyclinder was 0.69 m (27 inch). The ring was then cut into
Advances
TABLE 4
in flywheel
60
performance
The MEST Program was phased out in 1983 and flywheel technology was essentially frozen at these performance levels. Flywheel development activities
SPEED TESTS FOR ADVANCED
ROTORS
Disk/rim
SMC Yes
SMCIG Yes
K49 Noa
K49 -b
175=
229
237
134
1.86 40 638
2.32 47 056
2.24 30012
1.50 27 575
cyclic test.
Demo unit 1A 1B 1c
DIMENSIONS
OF TEST RIMS
Axial length (in)
Radial thickness (in)
Rim weight (kg)
Rim inertia (kg-m’)
4.0 1.9 1.9
1.5 1.5 1.5
12.5 5.8 5.8
1.34 0.63 0.63
design
Disk
a Rotor failed at 2586 cycles. b Rotor was not cycle tested. c Rotor had previously completed
SMC = S-glass sheet molding
regime expanded to include cyclic fatigue tests. Results for these tests are given in Table 3. The disk and disk/rim (illustrated in figure 2) concepts completed the full 10 000 cycle test. Subsequent ultimate speed tests of these rotors yielded energy densities of 175 and 229 kJ/kg, respectively. The disk/rim results were of particular interest because ultimate speed data were obtained before and after cycling the design. Test results for the rotor that had not been spun showed a density of 233 kJ/kg. Thus, for this concept (design and materials) there is no degradation in performance through 10 000 cycles.
Flywheel
Material Completed 10 000 Cycle Test Ultimate Energy Density fkJ/kg) Total stored energy (MJ) Speed at failure (rpm)
M = Metglass;
Subcircular rim
Bidirectional weave
lengths to provide spin test samples. The dimensions of the test rims are shown in Table 4. A sketch of one of the test units is shown in figure 5. Spin testing was performed in a vacuum test chamber with a vertical axis of rotation. The spin test facility is located in the ETAC laboratories at the Oak Ridge Gaseous Diffusion Plant. The flywheel test units were driven by an air turbine drive system. The test results are summarized in Table 5, and the results are compared with MEST program results in figure 6. The purpose of the tests was to
,-IMG
Figure 5
Typical
rim configuration
,o-
for Demo 1 series.
-
o-
Dec.9’ 1985 5-
IC R&l 1405 m/s o-
‘0 oNov.& 1985 18 RIM II73 m/s A
‘5o-
Ott 17, 1985 IA RIM 1055 m/s .
,oo-
‘5 -
Nov 8.1985 I B RIM + WEB I I73 m/s A
o-
‘0 -
o-
‘5-
Garrett . Brobeck .
LLNL . *GE
o-
,o -
l
LLNL Rocketdyne .
Hercules .
o-
o-
0
LLN&Owens Corning/ Lord F\PL .
I 2
I 4 Klnetac
Figure 6
Comparison
Ott 17,1985 IA RIM + WEB 1055 m/s .
LLNL ’
‘S-
Dee 9, I985 lCRIM+WEB 1405 m/s .
of rotor performance.
energy
I 6 ot foiiure
I e ( MJl
determine the flvwheel rim canabilities. The web structures were &relatively heavy; consequently, the data points for ‘rim + web’ do not reflect a typical design. Normally the web structure will be lightweight. The resulting rotor specific energy will then be near the ‘rim only’ data points. The most dramatic result was obtained with Demo lC, which was intentionally accelerated until failure occurred at a peripheral velocity of 1405 m/s and a frequency of 38 657 rpm. At this speed the specific energy of the rim was 878 kJ/kg. The kinetic energy of the unit was 7.28 MJ. The 1405 m/s peripheral velocity represents an increase of -40 per cent over previous spin tests. After the failure, the component fragments were examined. The web central hub section was intact, but the remainder of the web was broken into small pieces. The carbon/epoxy rim material was broken into extremely small dust-like particles. It is not known whether the failure initiated in the rim or in the outer portion of the web. The high speed failure provided valuable information with respect to containment requirements. Significant axial loads were present as a result of the failure as well as the expected radial loads. The demonstrated failure speed of 1405 m/s provides firm experimental support for flywheel design operating speeds of 1100-1200 m/s. Another series of demonstrations is planned incorporating thicker rim sections. The rim will be constructed by assembling multiple concentric rings with static interference fit in order to minimize radial tensile stresses at high speed. The planned configuration of Demo 2 is shown in figure 7.
Flywheel module
maintenance
power
A flywheel energy storage module can be used for sprint power applications, but it is also an attractive alternative to other forms of energy storage where the discharge time requirement is a few minutes up to a few hours. A flywheel power module could be used to provide the maintenance power needs of a spacecraft. A schematic illustration of a flywheel energy storage module is shown in figure 8: the concept incorporates a highly efficient axial gap permanent magnet motor/generator. It is anticipated that magnetic suspension systems would be used to achieve very low drag in space applications. A pair of counter-rotating flywheels would be used to minimize any effects on spacecraft attitude control during charging and discharging. The module shown in figure 8 is based on the Demo 2 planned configuration. The specific 61
TABLE 5
1985 date
Demo unit
FLYWHEEL DEMONSTRATION
Velocity m/s
energy of the module is estimated to be 216 to 252 k.I/kg at an operating speed of 1100-1200 m/s. The weight of the rim will be approximately one-half the weight of the total system. A lightweight vacuum housing (that is, no containment) is shown. The characteristics of the module are volume = 0.15 cubic metres, mass = 200 kg, and energy = 43-47 MI. The high specific energy and small volume of this module make it an attractive alternative to batteries for small, unmanned satellites.
TEST RESULTS
Rim specific energy kJ/kg (Whlkg)
Result
Oct. 17
IA
1055
485 (138)
Web failure, small crack. No rim damage.
Nov. 8
1B
1173
885 (188)
Stopped for inspection. No damage.
Nov. 12
1c
1221
883 (184)
Stopped for inspection. No damage.
Dec. 9
1c
1405
878 (244)
Intentional test.
Carbon fibre technology The carbon/epoxy composite rims fabricated by ETAC have an ultimate hoop strength of 3206 MPa. In the near future it should be possible to fabricate carbon/epoxy composite material with a strength of 3620 MPa. It is not unreasonable to expect fibre strength improvements of up to an additional 40 per cent over the next five years. The resulting composite strength of 5068 MPa would permit a flywheel operating speed of -1600 m/s, yielding a 78 per cent improvement in specific energy when compared to the 1200 m/s operating speeds suggested for Demo
failure
2.
References IM6 I
[l] Keckler, C. R., et al. ‘Integrated Flywheel Technology 1983’, Proceed-
I
ings of Workshop held at Greenbelt, MD, August 2-3, 1983, NASA Confer-
ence Publication 2290. [2] Keckler, C. R., et al. An Assessment of Integrated Flywheel System Technology, Proceedings of Workshop held at Huntsville, AL, February 7-9, 1984, NASA Conference Publication 2346. [3] Van Tassel, K. E. and Simon, W. E. Inertial Energy Storage for Advanced Space Station Applications, ProceedFigure 7
Planned configuration
ings of the 20th Intersociety Energy Conversion Engineering Conference,
for Demo 2 flywheels.
I
Miami Beach, Fla, August U&23,1985. [4] Hoffman, H. W., Martin, J. F. and Olszewski, M. Energy Options for Space Power, Proceedings of Second
Suspension (typic01 2 places)
Symposium on Space Nuclear Power Systems, Albequerque, NM, January,
1985. [5] Jones, J. E., MacPherson, R. E. and Nichols, J. P. Multimegawatt Space Nuclear Power Concept, Proceedings
Housing -
of First Symposium on Space Nuclear Power Systems, Albequerque, NM,
Composite flywheel system
January 11-13, 1984. [6] Olszewski, M. Development of Regenerable Energy Storage for Space Multi-Megawatt Applications, Proceedings of 21st Intersociety Energy Conversion Engineering Conference, San
I Figure 8 62
Schematic
illustration
of flywheel
module
Diego, Ca, August 25-29, 1986. [7] Olszewski, M. ‘Flywheel Performance: Current State-of-the-Art, Proceedings
I
Motor/generator
for maintenance
of 19th Zntersociety Energy Conversion Engineerign Conference, San Franc& power
application.
co, Ca, August 19-24, 1984.