论文

Design and Implementation of a Wheel Hub Motor

for an Electric Scooter

Wolfgang Gruber, Wolfgang Bäck and Wolfgang Amrhein

Institute of Electrical Drives and Power Electronics, Johannes Kepler University of Linz

Altenberger Str. 69, A-4040 Linz, Austria

wolfgang.gruber@jku.at

Abstract-This paper describes the optimization, design, buildup and measurements of a wheel hub motor for an electric scooter, which was designed to replace the hub motor of a commercial E-max scooter bike. To be able to replace the old drive by the new one without any structural alteration works on the scooters frame the size of the motor and shaft are given by the 13-inch wheel. In comparison to the existing system the new concept features far higher power, torque, speed range and efficiency.

I. INTRODUCTION

Electric scooters are increasingly demanded as alternative method of transportation for small distance travel. This trend is boosted by the increasing awareness to avoid polluting and harmful emissions from gasoline and diesel engines. A scooter is especially useful when traffic is heavy and parking space limited. Additionally, the costs per distance are

comparatively low. Recently, also advanced battery packages and charger have extended the range per charge [1], [2].

However, in spite of their advantages some drawbacks

remain. Most such systems are designed for low speed only; the drive efficiency, reliability and energy density should be

improved, whereas torque-to-weight ratio and price have to be further reduced. To reach these goals a lot of research has been achieved concerning the motor design [3], [4], [5]. This

paper explains the design choices that have been made in order to increase the performance (especially speed, torque

and power density) of an existing commercial electrical scooter. 1. Commercial E-max Scooter

The considered electric scooter is the commercial available E-max model 90S [6]. The features of the 13-inch maintenance free in-wheel hub rear motor are a rated power of 2,5 kW, a maximum motor power of 4 kW and a maximum speed of 45 km/h. The cage frame of the scooter is very strong, light and offers a lot of space to fit up the batteries end electronics. Furthermore hydraulic floating disc brakes are located in the front and as well in the rear.

2. Commercial Hub Motor

The brushless DC hub motor components are depicted in Fig. 1. The surface mounted permanent rotor magnets build up 20 pole pairs. The laminated stator features 36 slots holding a three phase two layer winding. The DC current link is 48 V. Three Hall switches are integrated to measure the

978-1-61284-247-9/11/$26.00 ©2011 IEEE

a)

b) c)

Fig. 1: Commercial E-max wheel hub drive: a) installed system; b) rotor; c) stator

permanent magnet flux and are used to commutate the phases. The mechanical air gap between rotor and stator is 0,65 mm.

The useable wheel diameter is 300 mm and the magnets axial height is 35 mm. The drive was examined and the effective copper filling factor was measured to be about 0,25. The cooling of the motor is achieved over the stator plate of the wheel; no additional cooling system is implemented. II.

SPECIFICATION OF THE NEW DRIVE

It is the major goal of this work to reach up to 120 km/h and significantly improve the motor torque without increasing the motor size. Due to air friction the demand for driving power features a cubic relation with the speed of the scooter. For the considered scooter the drag coefficient was measured experimentally and a necessary drive power of 13 kW was computed. To enable such a high power density in a 13-inch wheel an additional cooling system is needed.

Due to the fact that a high acceleration is demanded (even at quite high rotational speeds) the battery DC voltage was increased up to 100 V. Therefore it becomes possible to exert the maximum motor torque of 300

Nm for superior acceleration performance even at driving speeds up to 35 km/h. Furthermore faster driving speeds also lead to increased energy consumption. There is a cubic dependency between velocity and air friction power loss.

However, due to the high power density demand it is necessary to exploit the axial dimension with axially longer magnets. Compared to the off-the-shelf motor a duplication of the axial size of the magnets is possible. Furthermore the effective cooper filling factor must be increased significantly. TABLE I

FIXED DESIGN PARAMETERS FOR ROTOR AND STATOR

Symbol and Description

roaxszrbhgap – radial air gap size

rrValue Unit 0,5 mm TABLE II

VARIABLE DESIGN PARAMETERS FOR ROTOR AND STATOR

Symbol and Description

Range Unit Another improvement can be achieved utilizing a proper vector oriented control instead of the Hall sensor commutation [7]. Because of the high speed range the implementation of field weakening becomes requisite.

III.

OPTIMIZATION PROCEDURE

Motors with 36 slots, 40 magnetic poles and tree phases are very promising. They feature an extremely high winding factor regarding the fundamental wave and produce nearly no lateral forces. Due to the necessary field weakening capability a high inductance in the direct-axis is important. Therefore and due to the increased mechanical robustness, a interior permanent magnet design for the rotor was chosen. Fig. 2 shows the geometry of the principal stator and rotor design with their parameters. Despite the fact that some parameters are fixed due to the geometric boundaries a lot of values still remain for the optimization. The fixed parameters are syssstspbss – breadth of the stator slot (related to maximal value) 2,5 – 50 % rmc(related to maximal value) 55 – 95 % hm – magnet height

2 – 4

mm

parameters for the optimization together with their reasonable ranges. These parameters are varied in the optimization run to get the best combination concerning the drive efficiency, the direct-axis inductivity (representing the field weakening capability), the rotor mass, the cogging torque and the motor constant. For this reason a multidimensional generic optimization was achieved using a 2D finite element program. By the help of the strength Pareto evolutionary algorithm (SPEA) [8] the best possible motor designs are searched for.

One genetic generation consists of 25 individuals (motor variants). The time to compute such a generation lasts about three hours in the available personal computer cluster. The best design variants build up the next generation by mutation and crossing methods. All in all 75 such generations were built, simulated and evaluated, taking roughly one week computation time. Fig. 3 exemplarily depicts one of the resulting Pareto distributions. There the motor constants are plotted over the efficiencies (for the nominal point of operation). Every dot in this graph reflects a simulated motor design variant. It is clear that there is no drive design that represents an optimal solution for all design criteria. Therefore the different design criteria have to be assessed concerning their importance. Thus, the best solution can be chosen out of the different Pareto distributions. Table III exemplarily comprises some characteristic data of two different drive variants. The first one shows a quite high direct-axis inductance; it is therefore well capable for field weakening and thus can reach high speeds with relatively low field weakening losses. On the other hand, the motor constant is not too high, leading to relatively high losses when generating torque. The second motor variant behaves quite differently: A high motor constant leads to lower copper losses for a torque generation

TABLE III

CHARACTERISTIC DATA OF TWO POSSIBLE DESIGN VARIANTS Magnetic active mass 17,4 15,4 kg mΩ % Eddy current loss constant 313 421 µW/rpm² Copper losses (at 120 km/h)

364 1816 W

of 300 Nm but increased field weakening losses at 120 km/h.

It was decided to manufacture the first design, capable to reach high speeds with relatively low losses. The computed efficiency plot of this finally chosen hub motor is depicted in Fig 4.

IV.

PROTOTYPE SYSTEM

After the finite element optimization process a prototype motor was designed, built and put into operation. This section briefly deals with the implemented drive, cooling system, power electronics and control scheme.

1. In-Wheel Motor

Fig. 5 and 6 show the rotor and stator of the prototype system. The stator consists of the quill shaft, a stator joist and the laminated stator yoke. The rotor is composed of the rotor housing (split in a front and back housing) and the rotor back yoke holding the embedded magnets. The star connected 3 phase stator coil system is sealed by a heat resistant epoxy resin. To guarantee the mechanical durability of the parts (especially the quill shaft) finite element stress simulations have been achieved. The overall weight of the propulsion system is only about 20 kg, what is favorable for a comfortable driving behavior [9].

2. Liquid Cooling System A very important issue is the temperature increase of the stator coils and permanent magnets. For this reason an

necessary cooling capability. After these calculations a proper water cooling system was implemented. The stator is cooled down by a cooling spiral, which is located as close as possible to the stator coils. The necessary flow rate of the liquid is provided by the SHURflo 8000-433-136 diaphragm pump. An additional plate type heat exchanger with a switchable DC fan from Olaer [11] releases the heat. With this system a motor loss of up to 2 kW can be dissipated, still ensuring an acceptable coil temperature. The cooling spiral is depicted in Fig. 6 c).

3. Power Electronics

To energize the motor coils proper power electronics is

needed. A DC link voltage of 110 V and a rated effective current of at least 120 A per phase is necessary. Furthermore a vector oriented control scheme must be implemented. A

内容需要下载文档才能查看 内容需要下载文档才能查看

a) b)

c) d)

e) f)

Fig. 6: Manufactured drive components: a) laminated stator yoke; b) quill shaft; c) stator joist with cooling spiral; d) sealed stator; e) rotor housing with back yoke with magnets; f) completely assembled hub drive

possible solution for this demanding effort constitutes the

novel Semikron SKAI

40 A2 MD15-W 3-phase power

module [10]. It incorporates three MOSFET-half bridges with

integrated driver, sensor and safety circuits. Furthermore the embedded (free programmable) digital signal processor TMS320F28335 is capable to implement the demanded

vector oriented control scheme. The integrated analog to digital converter is thereby utilized to measure all phase voltages and two phase currents (up to 400

A). Additional interfaces e. g. for temperature sensors, encoder, potentio-meter or CAN bus alleviate the handling and programming. Furthermore, all control and sensor signals are monitored and

in case of a failure some components or the whole electronic is powered down.

4.

Control Scheme

A state of the art field oriented control scheme for permanent magnet synchronous machines in the rotating frame is utilized. Motor torque is created by the quadratic-axis current component is,q only. The reluctance torque component is small and therefore negligible. To enhance the speed range, which is limited by the battery voltage Ubatt, a negative current component in direct-axis direction is,q is used. The necessary direct-axis current component is computed from the stator voltage equation

us,d+jus,d=Rsis,d?ωrLs,qis,q

+j(Ri (1)

ss,q+ωrLm,dIf+ωrLs,dis,d

)with the stator coil resistance Rs, the stator coil self inductances Ls,d and Ls,q, the direct-axis coupling inductance Lm,d, the permanent magnet substitution-current If and the angular rotor speed ωr. Starting from

Ubatt= (2)

the needed direct-axis current is computed by

is,d=

2A

(3)

with

A=ω22rLs,d+R2

s

B=2(Rsis,q+ωrLm,dIf)ωrLs,d?2ωrLs,qis,qRs (4) C=(ω2

2

rLs,qis,q)+(Rsis,q+ωrLm,dIf

)

2

+Ubatt

.

This negative direct-axis current component is only energized into the coils when the stator voltage would exceed Ubatt. Additionally space vector modulation is used to drive the B6-inverter bridges for the best utilization of the battery voltage.

V. MEASUREMENTS

To compare results from the finite element optimization, the prototype motor was mounted on a test bench to measure some characteristic data. The test bench, equipped with a load machine and a torque sensor, is depicted in Fig. 7. Additionally a power analyzer is available.

1. Torque-Current-Ratio Exemplarily the measurement of the motor constant ki is shown in Fig. 8. This is achieved by measuring the induced voltage uind at certain rotor speeds. The correlation

Table IV summarizes the measurement data. Comparing these results with Fig. 4, it is cognizable that the measured efficiency is below the theoretical one. This originates mainly in two reasons. Firstly, in the measurement also the power electronics losses are included, because the battery voltage and current are measured to compute the electrical power. Secondly, the reduced motor constant increases the copper losses in the drive significantly. The 10% reduction in the motor constant increases the needed stator currents. For this reason the copper losses grow by 24% and therefore worsen the efficiency.

VI. CONCLUSION

ind

ωr

=

N

3is

=

3

ki

(5)

allows the computation of the torque to current ratio, also representing the motor constant. Thereby N stands for the effective number of windings of one phase in the considered 3-phase stator coil system. There is a difference of about 10% between the 2D-finite element simulation results and the measured values. The reasons for this deviation from the 2D-FE simulation results are axial end effect fluxes (which are increased due to saturation effects) and it is assumed that the magnetization of the permanent magnets lies below its nominal rated value.

2. Efficiency

At certain points of operation the efficiency was determined. For first measurements the load was only about 90 Nm and rotor speeds only up to 750 rpm were taken into account. Concerning both, the torque and speed, these values are well below the rated ones, due to the relatively weak load machine. The mechanical power is computed as product of motor angular speed and torque, whereas the electrical power is derived from the battery voltage and the battery current.

TABLE IV

EFFICIENCY MEASUREMENT DATA

This paper describes the optimization, design and build-up of a 13-inch wheel hub motor for an electric scooter. The rated power of 13 kW is extraordinary high for such a propulsion system. Therefore it is suitable to accelerate the vehicle up to 120 km/h, featuring a similar performance to a 125 ccm motorcycle. Other alternative applications for such a drive systems are electric utility vehicles.

The drive was designed to replace the standard hub motor of the E-max scooter. Table V summarizes the characteristic data of the novel and conventional drive system. It can be seen that especially the power and torque density have been improved significantly.

TABLE V

COMPARISON BETWEEN THE PROTOTYPE AND E-MAX STANDARD DRIVE

Used constructional volume 3,9 7,07 dm³ Continuous max. torque 75 ~210 Nm Continuous max. power ~2,8 ~18 kW ACKNOWLEDGMENT Speed Torque Mech. Power El. Power Efficiency

(in rpm) (in Nm) (in W) (in W) (in %)

This work was conducted in the research program of the

Austrian Center of Competence in Mechatronics (ACCM GmbH), which is a part of the COMET K2 program of the Austrian government. The authors thank the Austrian and Upper Austrian government for their support.

REFERENCES [1] Affanni, A.; Bellini, A.; Franceschini, G.; Guglielmi, P.; Tassoni, C.; “Battery choice and management for new-generation electric vehicles,” IEEE Transactions on Industrial Electronics, vol. 52, no. 5, pp. 1343- 1349, Oct. 2005 [2] Pellegrino, G.; Armando, E.; Guglielmi, P.; “An Integral Battery Charger With Power Factor Correction for Electric Scooter,” IEEE