Proceedings of the 2008 International Conference on Electrical Machines

Paper ID 1083

Linear Eddy Current Brake for Railway Vehicles Using Dynamic Braking
*Yasuaki Sakamoto, *Takayuki Kashiwagi, *Takashi Sasakawa, **Nobuo Fujii, Member IEEE *Electromagnetic Applications Laboratory, Railway Technical Research Institute 2-8-38, Hikari-cho, Kokubunji, Tokyo, 185-8540, Japan Tel: (+81)-42-573-7299, Fax: (+81)-42-573-7300, e-mail: ysakamo@rtri.or.jp ** Dept. of Electrical and Electronic Systems Engineering, Kyushu University 744, Motooka, Nishi-ku, Fukuoka, 819-0395, Japan Tel: (+81)-92-802-3694, Fax: (+81)-92-802-3694, e-mail: fujii@ees.kyushu-u.ac.jp
Abstract-One type of braking system for railway vehicles is the eddy current brake. Because this type of brake has the problem of rail heating, it has not been put to practical use in Japan. Therefore we proposed using a linear induction motor (LIM) for dynamic braking in eddy current brake systems. It reduces rail heating and uses an inverter for the self-excitation function. In this paper, we estimated the performance of an LIM from experimental results of a fundamental test machine and confirmed that the LIM under constant current excitation has an approximately constant braking force regardless of frequency for relatively low frequencies, and the reduction ratio of rail heating is approximately proportional to the frequency. These characteristics are convenient for the realization of the LIM rail brake system.

I.

INTRODUCTION

Fig. 1. Image of LIM rail brake.

One type of braking system for railway vehicles is the eddy current brake. It generates braking force by inducing eddy current in the rails while the vehicle runs. Using the electromagnetic force between the bogie and rails, the eddy current brake has the advantage of stable braking force that does not depend on the adhesion between the wheels and rails. But the eddy current brake cannot avoid rail heating because the braking principle is based on converting the kinetic energy of the vehicles into eddy current loss in the rails. The possibility of rail temperature rise affecting buckling stability is one of the reasons why eddy current brakes have not been put to practical use in Japan. Therefore we proposed a method that reduces rail heating and are investigating it [1]. The technique of reducing rail heating converts a portion of the kinetic energy, which should be consumed while braking, into electric power, using dynamic braking by applying a linear induction motor (LIM) to the excitation poles of the eddy current brake. There are some references on such a LIM with a rail as the secondary (for example [2]), but there are no thorough investigations that focused on dynamic braking. This paper reports on characteristics of the LIM rail brake. II. CONCEPT The proposed LIM rail brake is mounted on to the bogie as shown in Fig. 1. To install the armature in the limited space 1

between the wheels similar to a conventional eddy current brake, it is assumed that the armature length is about 1.2 m in the narrow gauge lines of Japan. The overhang of the armature coil in the lateral direction must be within the limits of the rolling stock gauge of the vehicles. The advantages of applying the LIM using dynamic braking to the eddy current brake system are: 1) For the same braking force, the LIM rail brake does not raise the rail temperature as much as a conventional eddy current brake. 2) Since the generated electric power can be allotted to the armature copper loss, an on-board energy source becomes unnecessary to maintain excitation of the LIM. The former is a clear solution to a problem with the conventional eddy current brake, and the latter is an advantage to the eddy current brake system itself. Since the LIM with a rail as the secondary is low in efficiency, it is accompanied by a large excitation loss to generate the prescribed braking force. However, because the generated electric power which should be consumed for vehicle braking can be allotted to the armature copper loss, excitation power of the LIM is not serious problem for use in braking. Moreover, the generated electric power can serve as an independent power supply system for the LIM, which is not related to the malfunctioning of the feeder and main circuit for the traction. Fig. 2 shows an example of the composition of the power

978-1-4244-1736-0/08/$25.00 2008 IEEE

Proceedings of the 2008 International Conference on Electrical Machines

Armature

Rotor (Solid steel: SS400)

Fig. 3. Fundamental test machine (rotary type).

Fig. 2. Example of a power supply system for a LIM.

supply system. By using an on-board auxiliary power supply as an initiating excitation power source for the LIM drive, the power supply system for the LIM can be completely independent of the main circuit systems. Such a composition can improve the redundancy of the entire braking system. Incidentally, when the reduction ratio of rail heating is not set to a high value, a composition without a braking resistor and DC chopper can also be selected. III. FUNDAMENTAL TEST MACHINE To investigate dynamic braking characteristics of the LIM with a rail as the secondary, we fabricated a fundamental test machine with a solid steel rotor of narrow width by remodeling a conventional squirrel cage induction motor. Fig. 3 shows the axial sectional view of the test machine, Fig. 4 the photograph of a solid steel rotor of narrow width, and Table I the main specifications. Since the test machine has a pole pitch and rotor width similar to that of a LIM rail brake, the influence of the transverse edge effect can be evaluated but the longitudinal end effect cannot. As shown in [3], the transverse edge effect in the eddy current brake is more significant than the longitudinal end effect. Fig. 5 shows the experimental apparatus for the fundamental test machine. An arbitrary rotational speed is able to be set as a test condition by using a directly connected drive motor. A. Slip characteristics The slip characteristic of the test machine under constant current excitation at 33.4 Hz, corresponding to a synchronous speed of 50 km/h, is shown in Fig. 6. The calculated values were computed by a theoretical analysis method which uses Fourier series [4]. Though a two dimensional analysis model was used, a method that considers transverse edge effects [5] was applied, with the rail (rotor) lateral surfaces considered to be an overhang of the secondary conductor. As for magnetic saturation, a multilayer theory [6] was applied to the modeling of the secondary. Fig. 6(a) shows that a slip larger than s = -0.4 is required to generate sufficient braking torque, because the secondary resistance of solid steel is large. The braking torque does not change very much for large changes in the slip, in the large
Fig. 4. Solid steel rotor of narrow width TABLE I SPECIFICATIONS OF FUNDAMENTAL TEST MACHINE Number of poles Rated output Inner diameter of stator Mechanical clearance Rotor width 4 255 kW (before remodeling) 270 mm 5.0 mm 65 mm

Clutch Drive motor

Torque meter

Fundamental test machine

Fig. 5. Experimental apparatus for the fundamental test machine.

slip range. The reduction ratio of rail heating is expressed as the ratio of the energy which is not converted into the rail heating to the kinetic energy consumed while braking, and the peak ratio is 55%. Since the secondary material is solid steel, the power factor is very low. In the range excluding small slip, the calculated values agree moderately with measured values.

Proceedings of the 2008 International Conference on Electrical Machines

Braking torque T (Nm)
30
I1 = 48 A

20 10
Measured Calculated

0 -1

-0.8

-0.6

-0.4

-0.2

Slip s (a)
100

Reduction ratio of rail heating r (%)

80 60 40 20 0 -1 -0.8 -0.6 -0.4 -0.2 0

B. Equivalent circuit The parameters of the equivalent circuit of the test machine (Fig. 7) were ascertained by experiments. The circuit constants depend on the secondary frequency because the skin effect is large. Therefore, the circuit constants can be expressed by approximate curves with parameters of the frequency f and the slip s. The concrete calculation method for the parameters is described below. At each frequency f, a synchronous operation test was performed with the drive motor. The phase voltage V1, current I1, and power factor cos of the test machine were measured. The impedance Zm of the excitation circuit can be calculated as follows:
m = 1 a
1 = V1 2 cos + j 1 cos I1 a = r1 + j l1

(1) (2) (3)

Slip s (b)
0.6

Power factor cos

where r1 is the armature resistance, l1 is the primary leakage inductance, and is the angular frequency. The resistance rm and excitation inductance lm are expressed as follows:
1 rm = Re 1 Zm 1 1 lm = Im 1 Zm

0.4 0.2 0 -1

(4) (5)

-0.8

-0.6

-0.4 Slip s (c)

-0.2

Generated power Pout (kW)

3 2 1 0 -1

where Re-1 and Im-1 express the reciprocal of the real and imaginary parts, respectively. Then, the load test of slip s was performed at each frequency f. The secondary voltage V2 can be calculated as,
V2 = V1 cos + j 1 cos 2 a I1 .

(6)

-0.8

-0.6

-0.4 Slip s

-0.2

The secondary current I2 can be derived with the resistance rm and excitation inductance lm corresponding to each frequency as,
2 = I1

(d)
Apparent power Pa (kVA)
15 10 5 0 -1

l jrm V2 V2 . = I1 m rm lm Zm

(7)

The secondary resistance r2/s and inductance l2 are given as follows:

V r2 = Re 2 s I2
-0.8 -0.6 -0.4 Slip s (e) -0.2 0

secondary leakage

(8) (9)

l2 =

V Im 2 . I2

Fig. 6. Slip characteristics of the fundamental test machine under constant current excitation at 33.4 Hz. (a) Braking torque. (b) Reduction ratio of rail heating. (c) Power factor. (d) Generated power. (e) Apparent power.

Fig. 8 shows the equivalent circuit constants of the test machine which were obtained by the calculations shown above.

Proceedings of the 2008 International Conference on Electrical Machines

r1 V1 I1
x1 (=l1) x2 (=l2)

r2

Resistance rm ()

30
Approximate curve

V2 xm (=lm)

I0

I2 rm

20 10

Excitation inductance lm (mH)

Fig. 7. Equivalent circuit of the test machine. V1 : Terminal voltage. I1 : Primary current. V2 : Secondary voltage. I2 : Secondary current. I0 : Excitation current. r1 : Armature resistance. l1 : Primary leakage inductance. x1 : Primary leakage reactance. r2 : Secondary resistance. l2 : Secondary leakage inductance. x2 : Secondary leakage reactance. rm : Resistance. lm : Exciting inductance. xm : Exciting reactance. : Angular frequency. s : Slip.

20

Frequency f (Hz)

40

60

(a)
10

The secondary resistance r2 and secondary leakage inductance l2 greatly depend on the slip frequency, and the influence of the skin effect can be confirmed. These circuit constants can be expressed by approximate curves with parameters of the frequency f and slip s as follows:
rm = 9.30 + 6.90 log( f ) lm = 6.47 r2 = 0.566 + 0.0108 s f

20

Frequency f (Hz)

40

60

( mH) ( )
0.491

(11) (12) (13)

Secondary resistance r2 ()

( )

(10)

(b)
1.5 1 0.5

l2 = 22.6 (s f )

( mH) .

Next, the excitation impedance Zm, the secondary impedance Z2, and the equivalent secondary impedance Z2e can be expressed recursively as follows:
j rm lm Zm = rm + j lm r Z 2 = 2 + j l2 s Z Z Z 2e = m 2 . Zm + Z2

10

20

30

40

50

60

Slip frequency sf (Hz)

Secondary leakage inductance l2 (mH)

(c)
10

(14) (15) (16)

Power (kW)

Although these impedances are of the test machine, they have the fundamental characteristic of an induction motor with a rail as the secondary. To confirm the validity of the obtained circuit constants (10)-(16), the calculated value of the power was compared with the measured value with a prescribed voltage impressed into the equivalent circuit (Fig. 9). The validity of the equivalent circuit was confirmed. IV. PERFORMANCE OF THE LIM RAIL BRAKE The performance of the LIM rail brake was estimated by using the equivalent circuit derived by virtually opening and unrolling the fundamental test machine into a linear motor. Since the circumference of the air gap of the test machine and the length of the LIM are different, the estimated circuit constants of the LIM were calculated by adding components of the test machine circuit in series. That is, the impedance of the LIM was

10

20

30

40

50

60

Slip frequency sf (Hz)

(d)
Fig. 8. Equivalent circuit constants of the test machine. (a) Resistance. (b) Exciting inductance. (c) Secondary resistance. (d) Secondary leakage inductance.

6 4 2 0 -1
Mechanical input (measured) Mechanical input (calculated) Electrical output (measured) Electrical output (calculated)

-0.8

-0.6

-0.4 Slip s

-0.2

Fig. 9. Comparison of the equivalent circuit and test machine. The frequency f is 33.4 Hz. The primary current I1 is 48 A.

Proceedings of the 2008 International Conference on Electrical Machines

Braking force Fx (kN)

derived by multiplying the impedance of the test machine by a coefficient according to the armature length ratio. However, the longitudinal end effect is not considered in this calculation. As shown in the result of analyses [1], the longitudinal end effect reduces braking force by about 10% in the large slip range. When the coefficient of the armature length ratio is k, the impedances of the LIM are assumed as follows:
Z2 = k Z 2 Z2 e = k Z 2e .

8
Electric loading 1,490 A/cm

6 4 2 0
80
v = 160 km/h 130 km/h 100 km/h 70 km/h

(17) (18)
Reduction ratio of rail heating r (%)

20

40

60

80

100

Frequency (Hz) (a)

The braking force Fx, the reduction ratio of rail heating r , the power factor of the equivalent secondary circuit cos2 , the generated power (electrical output) Pout , and the apparent power of the equivalent secondary circuit of the LIM can be calculated as follows:
2 3 (1 s ) k r2 I1 Z2e Fx = Z2 sv 2

60 40 20 0 20 40 60 80 100

(19)
2

Power factor cos2

r =

Re[Z2e ] Z2 s (1 s ) Z2e r2 Re[Z2e ] cos 2 = Z2e

(20) (21) (22) (23)

0.8 0.6 0.4 0.2 0 20

Frequency (Hz) (b)

Pout = 3I1 ( k r1 + Re[Z2e ] )

2

Pa2 = 3 Z2e I1

40

60

80

100

where v is the velocity of the vehicle using the LIM rail brake. Fig. 10 shows the estimated performance of the LIM rail brake with an armature length of 1.2 m. The primary current I1 is set to be 250 A, which corresponds to the electric loading of 1,490 A/cm. It is assumed that the LIM is used in a relatively low frequency range in consideration of the inverter capacity. In the low frequency range, a constant braking force (of about 5.4 kN) is generated, which does not depend on the velocity. On the other hand, the reduction ratio of rail heating is approximately proportional to the frequency up to a threshold frequency, and the peak value is 60%. It is characteristic that the plot of the generated power follows a line which depends only on the frequency and is independent of the velocity for frequencies lower than the threshold frequency. The power factor is relatively low (about 0.4-0.5) in the low frequency range, because the secondary current I2 is small in comparison with the excitation current I0 (shown in Fig. 11). The generated power is balanced with the armature copper loss at about 10 Hz. In this balanced state, the apparent power is relatively small (about 50 kVA) and no braking resistor is necessary in the power supply system for the LIM. However, the reduction ratio of rail heating is not high. There is a tradeoff between the reduction ratio of rail heating and the magnitude of apparent power. Fig. 12 shows characteristics of the LIM under the condition of constant apparent power operation and Fig. 13 shows

Frequency (Hz) (c)

Generated power Pout (kW) 200 150 100 50 0
800 600 400 200 0 20 40 60 80 100

20

40

Frequency (Hz) (d)

60

80

100

Apparent power Pa2 (kVA)

Frequency (Hz) (e)

Fig. 10. Estimated performance of the LIM rail brake with an armature length of 1.2 m, electric loading 1,490 A/cm, air gap 5 mm, and longitudinal end effect not considered. (a) Braking torque. (b) Reduction ratio of rail heating. (c) Power factor. (d) Generated power. (e) Apparent power.

Proceedings of the 2008 International Conference on Electrical Machines

2

20

Frequency f (Hz) Braking force Fx (kN)

I2 / I0

v = 160 km/h 130 km/h 100 km/h 70 km/h

15 10 5

Electric loading 1,489 A/cm Pout = 11 kW

Fx
50 100 150

20

40

60

80

100

Frequency (Hz)
Fig. 11. Ratio of the excitation current I0 and secondary current I2.
20

Velocity v (km/h)
Reduction ratio of rail heating r (%) Apparent power Pa2 (kVA)

(a)
80 60 40 20 0 50 100 150

100

Frequency f (Hz) Braking force Fx (kN)

15 10 5

Electric loading 1,489 A/cm Pa2 = 75 kVA

Pa2

Fx
50 100 150

Velocity v (km/h)

(b)
Fig. 13. Velocity characteristic curves of the LIM rail brake under the condition of constant current and constant power generation of 11 kW, without considering the longitudinal end effect. (a) Frequency and braking torque. (b) Reduction ratio of rail heating and apparent power.

Velocity v (km/h)
Reduction ratio of rail heating r (%) Generated power Pout (kW)

(a)
50 40 30 20 10 0 50

Pout
100 150

Velocity v (km/h)

(b)
Fig. 12. Velocity characteristic curves of the LIM rail brake under the condition of constant current and constant apparent power operation of 75 kVA, without considering the longitudinal end effect. (a) Frequency and braking torque. (b) Reduction ratio of rail heating and generated power.

relatively low frequencies. The LIM generates electric power which is determined by only the frequency. The peak value of the reduction ratio of rail heating is 60%. These characteristics are convenient for designing a controller for the LIM rail brake system. However, there is a trade-off between the reduction ratio of rail heating and the magnitude of apparent power. Therefore, careful studies are necessary for the arrangement of the inverter capacity and the operation frequency of the LIM. ACKNOWLEDGMENT The authors would like to acknowledge Mr. Erimitsu Suzuki of Railway Technical Research Institute for his extremely helpful suggestions. REFERENCES
[1] [2] [3] [4] Y. Sakamoto, T. Murai, and T. Sasakawa, The Basic Study of A.C. Excited Eddy-Current Rail Brake, The papers of Technical Meetings on Linear Drives, IEE Japan, LD-04-34, July 2004 (in Japanese) M. Hofmann, Th. Werle, A. Binder, R. Pfeiffer, Asynchronous Linear Machine for Railway Systems, Conference Proceedings of the ICEM 2000, pp. 223-227, 28-30 August 2000, Espoo, Finland M. Hofmann, Th. Werle, R. Pfeiffer, and A. Binder, 2D and 3D Numerical Field Computation of Eddy-Current Brakes for Traction, IEEE Trans. on Magnetics, Vol. 36, No. 4, pp. 1758-1763, April 2000 S. Nonaka and N. Fujii, Three-Dimensional Analysis of High-Speed Linear Induction Motor with Primary Iron Core of Finite Dimension, Electrical Eng. In Japan, Vol. 100-B, pp. 413-420, July 1980 (in Japanese) R. L. Russell and K. H. Norsworthy, Eddy Current and Wall Losses in Screened-rotor Induction Motors, Proc. IEE, Vol. 105A, pp. 163-175, 1958 I. Boldea, M. Babescu, Multilayer Theory of D. C. Linear Brakes with Solid-Iron Secondary, Proc. IEE, Vol. 123, pp. 220-222, March 1976

characteristics of constant power generation. The calculations confirm that the frequency needs to be varied accordingly with the velocity for constant apparent power operation, while the frequency virtually does not need to be varied for constant power generation. And the calculations also confirm that the braking force almost becomes a constant value at any velocity, just by setting the electric loading constant. V. CONCLUSIONS We estimated the performance of the eddy current brake using an LIM for dynamic braking from experimental results of a fundamental test machine. We confirmed that the LIM under constant current excitation has an approximately constant braking force regardless of both the velocity and frequency, for

[5] [6]