Flywheel energy storage using superconducting magnetic bearings

Ap/h,d

0964-1807(94)00037-9

Pergamon

Sup”““ndu~~r,~~ir)~ Vol. 2. No. 7:X. pp. 449-455, 1994 Copyright (‘ 1995 Elsewer Science Ltd Printed in Great Britain. All rights reserved 0964-1807,‘95 $9.50 + 0.00

FLYWHEEL ENERGY STORAGE USING SUPERCONDUCTING MAGNETIC BEARINGS JOHN R. HULL,

THOMAS

M.

MULCAHY,

and

ROBERT

KENNETH L. UHFRKA, ROBERT G.

A.

ERCK

ABBOI’D’

Energy Technology Division, Argonne National Laboratory, 9700 South C‘ass Avenue, Argonne. IL 60439. and ‘Commonwealth Research Corporation, Chicago. IL 60623. U.S.A.

Abstract-The ability of high-temperature superconducting (HTS) bearings to exhibit low rotatlonal loss makes possible high-efficiency flywheel energy storage (FES). In this paper, we discuss the general benefit of high-efficiency FES and a possible route to develop the HTS bearings required to achieve it. We describe a method to characterize superconducting bearings in terms of an effective coefficient of friction and present an expression for the coefficient that is general for most geometries. We also report a coefficient of friction of 3 x IO-’ for an 0.32 kg rotor. Initial experiments with a 12 kg rotor are also described.

INTRODUCTION

Storage of electrical energy on a utility scale is currently not practicable for most utilities; thus, full use of available base-load capacity is not possible. Existing storage methods are either too expensive or too inefficient, or require special geography [I]. Without the capacity to store energy, electric utilities are forced to cycle base load power plants to meet load swings in customer demand. Demand can change by as much as 30% over a 12-h period, resulting in significant costs to utilities as powerplant output is adjusted to meet these changes. The transmission system also experiences these changes in load. Thus, an additional advantage of a successful storage technology is the utilization of available nighttime transmission capacity to reduce the need to construct new lines. The development of HTS bearings with low rotational loss [2] led to the realization that FES incorporating HTS bearings has the potential for very high (>90%) diurnal storage efficiencies. By incorporating HTS bearings with inertial rims made from high-strength composite materials into ultra-high-speed flywheels and combining these with new-technology motor/generators, energy can be stored and released with very low loss [l]. Such high efficiencies cannot be achieved by FES that uses existing bearing technologies, including active magnetic bearings. The frictional losses are so high that flywheel stored energy is lost at rates on the order 1%/h [3,4], but HTS bearings should result in idle losses of only 0.1 %/h. The most practical FES option assumes arrays of many individual flywheel units. Although FES units with stored energy as low as 2-10 kWh may have some applicability, electric utilities are most likely to desire units with b 5 MWh storage capacity. Units with such capacity are approximately the largest that can be factory-assembled and truck-mounted for delivery to substations or energy storage facilities. FES may also be coupled with alternative generation sources, such as wind and photovoltaic. With a fleet of FES units handling real-time automatic grid control, a utility could manage generation variability as well as load variability. Most designs for a superconducting bearing include an HTS stator and a permanent magnet (PM) rotor. Current manufacturing capability places an upper limit on the size of the components for both the HTS and the PM. To develop HTS bearings for 5-MWh flywheels, it presently seems necessary to consider bearings in which the stator consists of an assembly of HTS components and the rotor consists of an assembly of PM pieces. 449 APSUP2:7/8-B

JOHN R. HULL et d

4.50

COEFFICIENT

The figure of merit for the losses associated friction p, defined as

OF FRICTION

with a superconducting

bearing

is the coefficient

P = F,IF,

of

(1)

where F, is the drag force and FL is the lift force. The lift force is the weight to be levitated F, = Mg, where M is the mass and g is the acceleration torque

(2)

of gravity.

tD = F,R,

Drag force is calculated

from the drag

= --la,

(3)

where R, is the mean radius at which the drag force acts, I is the total moment rotating object, and u is the angular acceleration,

of inertia

u = 2z(df/dr), where f is the frequency of the radius of gyration

of rotation. It is convenient R, as follows:

(4) to describe

I = MR;. We combine

equations

(l-5)

of the

the moment

of inertia

in terms

(5)

to obtain P = -(271R;

dfldN(gR,).

(6)

In equation (6), all terms are easily measured or calculated except R,. There is some ambiguity in its value because, a priori, we do not know the bearing loss mechanism nor how it is distributed over the radius of the bearing. For definiteness, we always take R, to be the outer radius of the bearing part of the rotor. In our experiments, this corresponds to the outer radius of the levitated permanent magnet. To determine the value of p required to achieve the low FES idle loss of 0.1 %/h, we simplify the discussion by assuming that all of the mass is located in a rim at radius R (R, = R, = R). The stored energy E is E = 1/2Mv2,

(7)

where v = 27cfR is the rim speed. The power loss P is P = F,v = -dE/dt, where t is time. By combining

equation

(7) and (8), we obtain

-(dE/dt)/E and by using equation

(8)

= F,v/(1/2Mv2),

(9)

(1) and (2) P = -

CWlWElvl(2g).

(10)

Typically, v is at least 1000 m/s in order to achieve high energy densities for useful energy storage, and the desired idle loss is IdE/dtI/E < 0.1%/h. This requires that p < 1.4 x 10m5. This is the maximum coefficient of friction that will give the desired idle losses, if the losses occur at room temperature. However, if HTS bearings are used, an additional factor must be considered because the HTS mateial must be maintained at cryogenic temperature. Assuming a refrigeration efficiency of -30% of the theoretical maximum, - 14 units of mechanical energy are needed to remove 1 unit of heat at 77 K. Thus, the required value for ~1must be reduced by a factor of 14 so that /J < 1.0 x 10-6. It should be noted that (dE/dt)/E is the desired goal in FES, not p itself. From equation (6) it is apparent that a higher value of p can be tolerated if R, > R,. Typically, for a conventional roller bearing, p = 0.001.

High-efficiency

SMALL

ROTOR

451

FES using HTS bearings

(< 1 kg) EXPERIMENTS

We have conducted several spin-down tests in a vacuum chamber consisting of a glass belljar chamber evacuated with an oil diffusion pump. This experimental arrangement is illustrated in Fig. 1. The chamber is -27 cm in diameter and contains a liquid nitrogen cryogenic chamber within the vacuum. One or more HTS elements can be placed either inside or on top of the liquid nitrogen chamber. The vacuum is < IO-” Pa when the cryogenic chamber is cold. This apparatus has been used to measure spin-down of flywheel rotors levitated above HTS stator arrays. The rotation rate is measured with a tachometer, and the flywheel position is determined by a traveling telescope. Spin-up has been accomplished with both a gasjet and an induction motor. The rotational-loss mechanisms of HTS bearings are still not well understood, although one of the major mechanisms is associated with inhomogeneities in the magnetic field as a function of azimuthal angle. One important loss from these inhomogeneities is hysteresis loss in the HTS elements. According to the critical-state model of superconductors, the energy loss AE per unit area AA of superconductor is given by AE!AA = IY(AE)~~J, where K is a geometry factor, AB is the magnetic density of the superconductor.

(11)

field inhomogeneity,

and ,I, is the critical current

To Vacuum Pump

Fig. 1. Schematic

representation

of belljar spin-down

vacuum

chamber.

JOHN R. HULL et al.

452

1200 s?

:

1000

800 z ‘E 600 2

CC

400 200 OL 0

5000

20000

15000

10000

25000

Time (s) Fig. 2. Rotational

rate vs time for vacuum

spin-down

of 11.3-g rotor.

For FES applications, the losses as a function of velocity will be especially important. At present, we have high-speed data only for rotors with small monolithic disk magnets. A typical magnet in this case had a diameter of 19 mm, a height of 6.4 mm, and a mass of 11.3 g. The magnet was levitated over a single HTS elements. Figure 2 shows rotation rate as a function of time for one of the experiments with this magment. To shorten the time required for the experiment, we slowed the magnet at intervals by either letting dry nitrogen into the chamber or by introducing mechanical friction against the spinning rotor. These actions are responsible for the sharp decreases in rotation rate shown in Fig. 2, e.g. at 7500 and 10,000 s. To calculate p for any rotation rate, we fitted a straight line through a series of consecutive data points and used the slope to calculate df/dt. We used this value to calculate 11 by using equation (6). In Fig. 3, we show p for this 19-mm-diameter disk magnet as a function of magnet rim velocity for two levitation heights. The results shown in Fig. 3 illustrate several general features of HTS bearing loss. First, p is smaller when the levitation height is larger. This is understandable from equation (11) because AB will be smaller at a larger distance from the magnet. Second, the rotor passes through a resonance (at u - 1 m/s) and then the loss becomes relatively constant. The resonance in p corresponds to vibrations of the rotor that are visible to the eye. Typically, with all of our experiments, the velocity at which the resonance occurs increases as the levitation height decreases. Because the stiffness of the levitation force behaves similarly, we identify this resonance with the vertical magnetic stiffness. Fortunately, for practical systems, this resonance occurs at very low velocities.

.OOOl -

0

10-6?

0

001

0 00 00 tb..

0

0

0

0

q n 0

10-6 r

0

i :

Al-2, 1.9mm

.

height

0 0 10-7 0.01

Al-1,2.8mm

0

.

. ...*..’

. 0.1

* ’ *.**.’

1

.

* .*m***’

10

.

* * mm-

100

Velocity(m/s) Fig. 3. Coefficient

of friction as a function of magnet rim velocity for an 11.3-g rotor.

High-efficiency

FES using HTS bearings

453

As shown in Fig. 3, it appears possible to construct HTS bearings in which 11 is independent of velocity at high speed. Although the coefficient of friction was independent of velocity at high speed. it was much higher than desirable. In our bell-jar spin-down experiments, we have also used PM rotors that contain an NdFeB ring that has 88-mm OD, 64-mm ID, and is 13 mm high. The rotor has a mass of 0.32 kg. An early version of this rotor is shown in Fig. 4. The PM is polarized axially with an internal magnetization of h 1.1 T. Using this PM, we have obtained a coefficient of friction of I_(= 3 x IO ‘. The HTS array in these experiments consisted of 10 HTS cylinders, each - I8 mm in diameter and I I mm high. The HTS cylinders were equally spaced around the mean diameter of the PM ring. Each HTS element consisted essentially of a single domain of melt-textured Y--Ba-Cu0 with the c-axis aligned predominantly along the vertical. Our experience in using different HTS arrays with this

Fig. 4. Photograph

of ring-magnet

rotor

used in belljar chamber

JOHN R. HULL et a/.

454

rotor is that samples with higher J, have lower ~1,which is consistent with equation (11). For these purposes, J, for differing HTS elements may be relatively determined by measuring the levitation force at a fixed height with a standard reference magnet. MULTI-COMPONENT

ROTOR

(2-15

kg) EXPERIMENTS

Because monolithic PMs are not available commercially in sizes greater than about 100 mm OD, development of HTS bearings large enough for practical FES will required the rotors to be levitated with permanent magnets constructed from an assembly of components. For this purpose, we have designed and constructed an FES test apparatus (FTA) to investigate rotors with mass up to - 1OCkg. The main FTA test chamber is a cylindrical stainless steel vessel constructed of stainless steel with ID of 914 mm and height of 864 mm. A cryochamber housing the HTS bearing stator elements is placed within the FTA vacuum chamber. The FTA includes a 1000 l/s turbomo lecular vacuum pump and two rotary vacuum pumps. These pumps can create a vacuum of < lo4 Pa when the cryogenic chamber is cold. A schematic diagram of the chamber is shown in Fig. 5. We have successfully levitated and spun rotors of 2.5 and 12.25 kg in this chamber. The 2.5-kg rotor consists of a PM structure with a plastic turbine mounted on top. A photograph of this rotor is shown in Fig. 6. The larger rotor uses the 2.5 kg magnet for levitation and adds a graphite/epoxy composite ring with a 376 mm OD. Coefficients offriction for these rotors are still under analysis.

/

Spin-up Motor

Top flange J

f

Therm1 Feedth

I

Scale: 1 cm = 5 cm Fig. 5. Schematic

To Vacuum System

diagram

+ of spin-down

I vacuum

chamber.

Sensing Probe Feedth rough

High-efficrency

Fig. 6. Photograph

of magnet

435

FES using HTS hearings

and cyrochamber

used 111large vacuum

chamber

CONCLUSIONS

With the advent of HTS materials and advances in composite materials and power electronics, economical deployment of FES may now be possible. A series of small-scale tests on HTS bearings have been performed using a belljar vacuum cleaner. Hysteresis in the HTS stator caused by azimuthal inhomogeneities in the magnetic field of the magnet rotor has been identified as a major mechanism of rotational loss. We have defined a coefficient of friction as a figure of merit to describe the rotational Losses from HTS bearings. The lowest coefficient of friction in our experiments was 3 x 10m7, obtained with a 0.32-kg rotor. Present experiments involve rotors of - 10 kg. Ackno,c,(ed~~Pmenrs-This work IS partially supported by the lJ.S. Department of Energy, Energy Ethciency and Renewable Energy, as part of a program to develop electric power technology. under Contract W-31-10%Eng-38. The authors are indebted to U. Baiachandran, S. Dorris, D. Shi. W. Zhong. and W. Gawalek for providing HTS superconductors used in the experiments and to Z. Yang for providing useful comments on the manuscrrpt.

REFERENCES R. Abboud, J. Hull, K. Uherka and T. Mulcahy, Flywheel energy storage using superconducttng magnetic bearings, Prtx. Am. Powrr Co@, Chicago (April 19941. B. R. Weinberger, L. Lynds, J. R. Hull and LJ. Balachand~n. Low friction in high temperature supercon[~uct(~r bearings, App!. Phys.Lrrr. 59, 1132 1I34 ( 1991 i. P. 0. Jarvinen, B. L. Brench, R. D. Hay and N. E. Rasmussen, Testing and evaluation of a solar photovoltaic flywheel energy storage system, Proc. Inrrrsoc. Enei-ql, Conr. Enyn~q Conj,. Atlanta, pp. X98-903 (August 19X1). J. A Kirk and P. A. Studor. Flywheel energy storage II: magnetically suspended flywheel. Int. J. Mcch .%i. 19. 233 245 (I 977).