Permanent Magnet Motor: High Power Density Brushless Motors (part 1)

Home | Articles | Forum | Glossary | Books

1. Design considerations

A utilization of active materials of an electric motor can be characterized by:

• power density, i.e., output (shaft) power-to-mass or output power-to- volume ratio;

• torque density, i.e., shaft torque-to-mass or shaft torque-to-volume ratio

Torque density is a preferred parameter to power density when comparing low speed motors, e.g., gearless electromechanical drives, hoisting machinery, rotary actuators, etc. The utilization of active materials increases with the intensity of the cooling system, increase of the service temperature of insulation and PMs, increase of the rated power, rated speed and electromagnetic loading, i.e., [eqn. not shown].

All types of electrical machines and electromagnetic devices show a lower ratio of energy losses-to-output power with an increase in output power. This means that the efficiency of electric motors increases with an increase in the rated power. Large PM brushless motors can achieve a higher possible efficiency than any other electric motor (except those with superconducting excitation windings). The limitation, however, is the high price of PM materials. NdFeB magnets offer the highest energy density at reasonable costs.

Their major drawback, compared to SmCo, is temperature sensitivity. Performance deteriorates with increased temperature which has to be taken into account when the motor is designed. Above a certain temperature, the PM is irreversibly demagnetized. Therefore, the motor's temperature must be kept below the service temperature (1800 for most NdFEB PMs, maximum 2000C) when using NdFeB magnets. A natural air cooling system is sometimes not efficient and the stator must be cooled by water or oil circulating through the stator housing. The rotor losses in PM synchronous motors are small so most PM machines employ passive cooling of their rotors.

FIG. 1. Magnetic flux distribution in a 4- and 16-pole motor.

FIG. 2. PM motor mass versus number of poles for constant stator inner diameter.

The main dimensions (inner stator diameter D1in and effective length Li of core) of an electric motor are determined by its rated power Pout ? Selm, speed ns, air gap magnetic flux density Bmg, and armature line current density Am. Magnetic flux density in the air gap is limited by the remanent magnetic flux density of PMs and saturation magnetic flux density of ferromagnetic core. The line current density can be increased if the cooling is intensified.

For a given stator inner diameter the mass of the motor can be reduced by using more poles. FIG. 1 illustrates this effect in which cross sections of 4-pole and 16-pole motors are compared. The magnetic flux per pole is decreased in proportion to the inverse of the number of poles. Therefore, the outer diameter of the stator core is smaller for a motor with a large number of poles at the same magnetic flux density maintained in the air gap.

PM motor mass as a function of the number of poles for the same stator inner diameter is shown in FIG. 2. However, the electromagnetic power decreases as the number of poles increases, i.e.,

Selm =0.5p2kw1D2 1inLiBmgAmf/p.

The absence of the exciter in large PM brushless motors reduces the motor drive volume significantly. For example, in a 3.8 MW PM synchronous motor, about 15% of its volume can be saved.

2 Requirements

PM brushless motors in the megawatt range tend to replace the conventional d.c. motors in those drives in which a commutator is not acceptable. This can be both in high speed (compressors, pumps, blowers) and low speed applications (mills, winders, electrical vehicles, marine electromechanical drives).

FIG. 3. Typical torque-speed characteristic for ship propulsion.

Ships have been propelled and maneuvered by electrical motors since the late 1970s. Recently, rare earth PMs allow the design of brushless motors with very high efficiency over a wide speed range. This is the most important factor in ship and road vehicle propulsion technology. For ship propulsion the typical torque-speed characteristic is shown in FIG. 3. The motor and converter has to be designed with the highest efficiency to meet the rated point N ("corner power" point). It is advantageous to use constant flux motors here in contrast to the hyperbolic characteristics of road vehicles. Point N represents the "worst case conditions" for the efficiency because the core losses as well as the winding losses reach their maximum values. In addition, the efficiency should not decrease significantly at a partial load down to 20% of rated speed (0.2 nr) as this is the speed of long distance journeys.

The solid state converter should operate in such a way as to obtain the lowest possible winding and switching losses. The last feature requires a speed dependent rearrangement of the winding and inverter components. A subdivision of the stator winding and converter into modules is necessary due to reliability reasons. In case of failure of the drive system the best solution is a modular concept where the damaged module can be quickly replaced by a new one. Armature windings with more than three phases are a promising option.

Like most PM brushless motors, a large motor should be controlled by the shaft position angle to obtain the electromagnetic torque directly proportional to the armature currents. The current waveform coincides with the induced voltage and the stator losses reach their minimum value.

3. Multiphase motors

For some large PM brushless motors the number of armature phases m1 > 3 is recommended. The armature phase current is inversely proportional to the number of phases, i.e., Ia = Pout/(m1V1? cos f). For the constant output power Pout = const, constant input phase voltage V1 = const and approximately the same power factor cos f and efficiency ?, the armature current is lower for the greater number of phases. This means that a multiphase a.c. motor of the same dimensions as a three-phase motor has similar mechanical characteristics but draws lower phase currents.

Another distinguishing feature is the possibility of a step change in the speed by changing the supply phase voltage sequence. The synchronous speed is inversely proportional to the number of phase voltage sequences k, i.e., ns = f kp (eqn. 1)

...where f is the input frequency. For the three-phase motor, only two voltage sequences are possible and the switching only causes the reversal of the motor speed. For the m1-phase motor with an odd number of phases m1, the first k =(m1 - 1)/2 sequences change the synchronous speed. For a nine-phase motor it will be k =1, 2, 3, 4. The remaining frequencies excluding the zero sequence produce rotation in the opposite direction (for a nine-phase motor k =5, 6, 7, 8). For an even number of armature phases, the number of sequences changing the speed is k = m1/2. This is due to harmonic fields. In addition, a synchronous motor must meet the requirement that the number of rotor poles must be adjusted to match each speed.

A multiphase system (m1 phases) of voltages can be expressed by the following equation:

vl = v 2cos[2pft - (l - 1)k2p/m1] (eqn. 2)

where l =1, 2, 3, ...m1 or l = A,B,C, ...m1. The voltages vl create a star connected m1 phase voltage source. Changing the sequence of phase voltages is done by selection of the appropriate value of k. The sources vl or vA,vB,vC,... can be replaced by a VSI or voltage sources modeling the inverter output voltages.

As in three-phase systems, multiphase systems can be star or polygon connected. There are the following relationships between line and phase voltages and armature currents for the star connection:

V1L =2V1 sin

_3600 2m1

_ ,IaL = Ia (eqn. 3) For the polygon connection IaL =2Ia sin

_3600 2m1

_ ,V1L = V1 (eqn. 4)

The operation of the motor can be described by the following voltage equation in the stator co-ordinate system:

[v]=[Ra][ia]+ d dt

{[?a]+[?f ]} =[Ra][ia]+[Ls] d[ia] dt

+[ef ] (eqn. 5)

where [Ra] is the matrix of armature resistances, [fla]=[Ls][ia] is the matrix of the armature reaction fluxes caused by the armature currents [ia], and [?f ]is the matrix of PM flux linkages which create the no-load EMFs [ef ]= d/dt[?f ].

The synchronous inductance matrix [Ls] contains the leakage and mutual inductances.

FIG. 4. Nine-phase windings located in 18 stator slots: (a) symmetrical, (b) asymmetrical.

The instantaneous electromagnetic torque developed by a multiphase synchronous motor is

Td = pelm 2pns

= 1 2pns l=m1 _ l=1 eflial (eqn. 6)

where the instantaneous electromagnetic power is:

pelm = l=m1 _ l=1 eflial (eqn. 7)

Nine-phase motors are easy to design since their windings can be placed in stator cores of most standard three-phase motors. As has been mentioned, a nine-phase motor has four different synchronous speeds for four phase volt age sequences. Values of these speeds depend on the type of stator winding, which can be symmetrical or asymmetrical (FIG. 4). These windings can be distinguished from each other by the Fourier spectrum of the MMF produced by each of them. The symmetrical winding produces MMF harmonics which are odd multiples of the number of pole pairs p. With the asymmetrical winding the number of pole pairs cannot be distinguished.

4. Fault-tolerant PM brushless machines

First prototypes of fault-tolerant PM brushless machines of modular construction were designed and tested in Germany in the late 1980s. There are many potential faults that can occur in a motor and associated power electronics; however, the most common faults are:

• stator winding open-circuit

• stator winding short-circuit

• inverter switch open-circuit (similar to winding open-circuit)

• inverter switch short-circuit (similar to winding short-circuit)

• d.c. link capacitor failure

The most successful approach is a multiphase machine - solid state converter system in which each phase may be regarded as a single module. The machine must produce rated torque with any single phase fault. Thus, the machine must be overrated by a fault-tolerant rating factor kfault that depends on the number of stator independent phase windings m1, i.e.,

kfault = m1 m1 - 1 (eqn. 8)

Eqn (eqn. 8) shows that a three-phase machine (m1 = 3) must be overrated by 50%, if the two remaining phases are to make up the short fall from one lost phase. The higher the number of stator phases, the lower the penalty for fault tolerance. Differences between standard PM brushless machine and fault-tolerant PM brushless machine are summarized in Table 1.

Table 1. Differences between standard and fault-tolerant PM brushless machine



Number of phases

Construction of stator

Construction of winding

Phase reactance per unit

Mutual inductance

Short circuit current

Participation of harmonics in torque production

Interaction of phase currents

Slot-opening permeance harmonics in the air gap

Losses in PMs


Standard Machine


one cylindrical unit distributed in slots

less than 1 up to 50% of the phase self-inductance

higher than rated current fundamental harmonic only

current in one phase affects currents in remaining phases low harmonics content (large number of slots)

problematic only at high speeds


Fault-Tolerant Machine

more than 3

modular construction

concentrated around teeth (one coil per slot)


less than 5% of the phase self-inductance

the same as rated current

higher harmonics can be engaged

current in one phase has almost no influence on other phases

high harmonic content (concentrated coils)

significant eddy current losses in PMs at low speeds


8.5 Surface PM versus salient-pole rotor

The PM rotor design has a fundamental influence on the output power-to- volume ratio.

FIG. 5. Cross sections of large PM motors: (a) surface PM rotor, (b) salient-pole rotor: 1 - damper bar, 2 - bandage (retaining sleeve), 3 - air gap, 4 - PM, 5 - rotor hub, 6 - rotor shaft, 7 - gap between poles, 8 - pole shoe, 9 - axial bolt.

Two motors of surface PM and salient-pole construction, as shown in Fig. 5, have been investigated. As the rare earth PMs are rather expensive the power density must be maximized.

Tests made on two 50-kW, 200-V, 200-Hz, 6000-rpm motors designed show the following:

• the salient pole motor causes greater space harmonic content in the air gap than the surface PM rotor motor,

• the synchronous reactances of the surface PM rotor motor (Xsd = Xsq = 0.56 Ohm) are smaller than those of the salient rotor motor (Xsd =1.05 Ohm and Xsq =1.96 Ohm),

• the subtransient reactances in the d-axis are X__ sd =0.248 Ohm for the surface PM rotor and X__ sd =0.497 Ohm for the salient pole rotor motor, which results in different commutation angles (21.00 versus 29.80),

• the rated load angle of the surface PM rotor motor is smaller than that of the salient pole motor (14.40 versus 36.60),

• the relatively large load angle of the salient pole motor produces large torque oscillations of about 70% of the average torque as compared with only 35% for the surface PM rotor motor,

• the output power is 42.9 kW for the salient pole motor and 57.4 kW for the surface PM rotor motor,

• the surface PM rotor motor has better efficiency than the salient pole motor (95.3% versus 94.4%),

• the volume of PMs is proportional to the output power and is 445 cm3 for the salient pole motor and 638 cm3 for the surface PM rotor motor.

The stator dimensions and the apparent power have been kept the same for the two tested motors.

The air gap field of the salient pole motor is of rectangular rather than sinusoidal shape and produces additional higher harmonic core losses in the stator teeth. The air gap field harmonics due to the stator slots and current carrying winding produce eddy current losses in the rotor pole faces and stator core inner surface. In the surface PM motor the field harmonics induce high frequency eddy current losses in the damper. The damper is usually made from a copper cylinder and has axial slots to reduce the eddy current effect.

6. Electromagnetic effects

6.1 Armature reaction

The action of the armature currents in the phase windings causes a cross field in the air gap. This implies a distortion of the PM excitation field. The resulting flux induces a proportional EMF in the armature phase conductors.

At some points in the air gap the difference between the d.c. link voltage and the induced EMF may decrease significantly, thus reducing the rate of increase of the armature current in the corresponding phase.

The armature reaction also shifts the magnetic neutral line of the resultant flux distribution by a distance dependent on the armature current. The displacement between the current and flux distribution contributes to the de crease in the electromagnetic torque. Moreover, the distorted excitation field and flux in the d-axis produces noise, vibration and torque ripple.

The armature reaction together with the commutation effect can produce dips in the phase current waveforms, which in turn reduce the mean value of the armature current and electromagnetic torque.

FIG. 6. Rotor of a large a.c. motor with surface PMs and nonferromagnetic parts for reducing the armature reaction according to Siemens: 1 - ferromagnetic core, 2 - nonferromagnetic core.

The influence of the armature reaction on the electromagnetic torque can be minimized by an increase in the air gap or using an anisotropic material with a large reluctance in the q-axis of the magnetic circuit. An interesting construction of the rotor magnetic circuit shown in FIG. 6 has been pro posed by Siemens. The nonmagnetic parts suppress the cross-field of the armature currents and thus reduce the effect of armature reaction. Consequently, the inverter can better be utilized because the EMF waveform is less distorted.

6.2 Damper

As in synchronous machines with electromagnetic excitation, the damper reduces the flux pulsation, torque pulsation, core losses and noise. On the other hand, the damper increases the losses due to higher harmonic induced currents as it is designed in the form of a cage winding or high conductivity cylinder.

A damper can minimize dips in the current waveform due to armature reaction and commutation and increase the electromagnetic torque. The damper bars reduce the inductance of the armature coil with which it is aligned and reduces the effects of the ?ring of the next phase. Thus, the addition of the complete damper (cage with several bars per pole) can increase significantly the output power.

It is known from circuit theory that the self-inductance of a magnetically coupled circuit decreases when a short-circuited coil is magnetically coupled to it. The same happens to the mutual inductance between two circuits magnetically coupled. Obviously, if the coupling between the short-circuited coil and the magnetic circuit changes, the self and mutual inductances also change.

FIG. 7. Self and mutual inductances of a 6-phase, 75-kW, 60-Hz, 900-rpm, 8-pole PM motor as functions of the rotor position angle between the axis of each phase and the d-axis of the motor.

FIG. 7 shows the test results on a 6-phase, 75-kW, 60-Hz, 900-rpm PM motor prototype. The angle between the d and q-axis is equal to 22.50 geometrical (8 poles), which corresponds to 900 electrical. The self inductance L11 and mutual inductances M12, M13,..., M16 have been plotted against the angle between the axis of each phase and the d-axis of the motor.

The mutual inductance M14 between two phases shifted by 90 degree electrical with no damper would be nearly zero, since the magnetic circuit is almost isotropic. However, the damper introduces a magnetic coupling between the phases 1 and 4, and M14 varies with the angle. The voltage induced in the damper by one of the phases produces a magnetic flux that links this phase with the other phases so that the mutual inductance is greater than zero.

All the inductances plotted in FIG. 7 would also vary with the rotor angle if the magnetic circuit was made of anisotropic material and the damper was removed. In this case the self-inductance L11 would have its maximum value at the zero angle, i.e., when the center axis of the phase 1 and the d-axis coincide.

6.3 Winding losses in large motors

Suppose that the phase armature current has a square-wave shape with a flat topped value I (sq) a . Such a rectangular function can be resolved into Fourier series:

• for a 1200 square wave


6.4 Minimization of losses

The large volume of ferromagnetic cores, supporting steel elements as bolts, fasteners, clamps, bars, ledges, spiders, etc., high magnetic flux density in the air gap and intensity of leakage fields including higher harmonic fields contribute to large power losses in magnetic circuits and structural ferromagnetic components. The core losses are minimized by using very thin stator laminations (0.1 mm) of good quality electrotechnical steel. It is also recommended to use laminations for the rotor spider instead of solid steel. To avoid local saturation, careful attention must be given to uniform distribution of the magnetic flux.

FIG. 8. Concentrated non-overlapping stator winding (with one slot coil pitch).

For multiphase modular motors the armature winding losses can be reduced by rearranging the stator winding connections. To avoid excessive higher harmonic losses, both MMF space distribution and armature current waveforms should be close to sinusoidal, or at least higher harmonics of low frequency should be reduced.

Multiphase modular motors can also be designed with stator coil span equal to one slot instead of one full pitch. Such a winding is called winding with concentrated non-overlapping coils (FIG. 8). The concentrated non overlapping winding must meet condition. The end connections are very short, the winding losses are significantly reduced and the slot fill factor is high.

7. Cooling

The degree of utilization of active materials is directly proportional to the intensity of cooling. The efficiency of PM motors is high, for 2p> 4 the stator yoke is thin and the rotor losses are small, so that PM brushless motors can achieve good performance with indirect conduction cooling, i.e., the stator winding heat is conductively transferred throughout the stator yoke to the external surface of frame. Direct liquid cooling of stator conductors allows for significant increase of the power density due to high heat transfer rates. On the other hand, it requires hollow stator conductors and hydraulic installation support.

The rotor gets hot due to heat rejected from the stator through the air gap and eddy current losses in PMs and ferromagnetic core. Although the rotor cooling systems are simpler, careful attention must be given to the tempera ture of PMs as their remanent flux density Br and coercivity Hc decreases, diminishing the machine performance.

Table 2 summarizes some typical approaches to PM motors cooling.

Table 2. Large PM motor cooling options.


Approach | Description | Comments

Stator (armature)


Indirect conduction

Direct conduction

Direct cooling


Passive air

Forced air

Open-loop spray

Closed loop internal


Armature heat is conductively transported through the stator yoke to heat exchange surface;

external surface can be cooled by surrounding air, water or oil

Winding is flooded with dielectric fluid;

Heat conduction across winding insulation;

Winding is internally cooled; direct fluid contact

Rotor cavity air circulation driven by rotor motion

Rotor cavity air circulation driven by separate blower

Liquid spray on rotor surface

Liquid (oil) circulated internal to rotor


• coolant is completely isolated from conductor

• high heat transfer rates can be achieved due to thin yoke of some PM motors

• pod motors (ship propulsion) cooled by water

• very high heat transfer rates are achieved

• dielectric fluid must be used

• slot fill factor compromised

• highest heat transfer

• dielectric fluid must be used

• slot fill factor compromised

• complex fluid manifolding

Simplest, requires shaft rotation, good match for quadratic loads

Extra motor with associated manifolding requested

Wet rotor, requires pump system

Highest heat transfer, requires rotating shaft seals


8. Construction of motors with cylindrical rotors

First prototypes of rare-earth PM motors rated at more than 1 MW for ship propulsion were built in the early eighties. This section contains a review of constructions and associated power electronics converters for large PM motors designed in Germany. Although some construction technologies and control techniques are now outdated, the experience and data base obtained have become a foundation for design of modern large PM brushless motors.

8.1 Motor with reduced armature reaction

One of the first large PM brushless motors rated at 1.1 MW, 230 rpm was built by Siemens in Nuremberg, Germany. The rotor construction with 2p = 32 poles is shown in FIG. 6. There are nonferromagnetic poles with large reluctance in the q-axis to reduce the armature reaction. To minimize the cogging torque, the rotor was made with skewed poles. The stator (inner diameter 1.25 m, stack length 0.54 m) with a six-phase detachable winding consists of 8 replaceable modules (each module covering four pole pitches) to provide on-board repairing. A single module contains 2×6 coils located in 24 slots of a single-layer, full pitch winding. The one-module winding is shown in FIG. 9.

FIG. 9 Stator winding of one module of a 6-phase, 1.1-MW large PM motor manufactured by Siemens.

FIG. 10 Large variable-speed drive with six-phase PM motor manufactured by Siemens: (a) power circuit, (b) inverter module for phase control current.

The six phase has been chosen since an even number of phases enables the division into two equal redundant systems (two line converters) and the duration of the trapezoidal current can be maximized to 1500 or even 1800 with respect to the achievable commutation time. Each of the phase windings consists of two halves of identical coils being connected in series (switch S2 closed in FIG. 10b) at low speeds, i.e., less than 55% of the rated speed and in parallel (switch S1 closed in FIG. 10b) at high speeds (reduction of synchronous reactance).

A separate PWM inverter is assigned per phase as shown in FIG. 10a.

Three inverters in parallel are connected to a constant d.c. link. Two line converters are fed from 660-V three-phase on-board supply. The inverter module consists of a four-quadrant bridge which controls the phase current according to the set value which is dependent on the rotor angle. GTO thyristors and diodes with blocking capacity of 1.6 kV and a permissible turn-off current of 1.2 kA have been used. At 230 rpm and f =61.33 Hz the commutation angle is rather large. Through the application of PWM, the armature cur rent is kept quasi-constant within approximately 1200 and coincides with the induced voltage ef . The total current flow duration is about 1600. A water cooling system with axial ducts is used for the stator.

The instantaneous electromagnetic power and developed torque per phase can be found as a product of EMFs and armature current according to eqns (6) and (7).

The large PM brushless motor shows about 40% less mass, frame length and volume, and 20 to 40% less power loss as compared with a d.c. commutator motor.

FIG. 11 Cross section of a large PM motor built by ABB and Magnet Motor GmbH: 1 - stator frame section, 2 - ledges clamping the stator sections, 3 - bolt for mounting the stator element, 4 - stator core, 5 - stator winding,6-cooling duct, 7 - fiberglass bandage, 8 - PMs, 9 - rotor yoke, 10 - rotor ring (spider).

8.2 Motors with modular stators

A large PM motor built by ABB in Mannheim, Germany, in co-operation with Magnet Motor GmbH at Starnberg is shown in FIG. The stator consists of Z = Nc = 88 identical elements and Z/2 stator sections (two teeth or two elements screwed together). The number of stator slots is s1 = Z.


FIG. 12 Cross section of simple multiphase PM motor with m1 =5,2p = 12, and Z = 10: 1 - PM, 2 - stator element (tooth and winding), 3 - stator yoke.

FIG. 13. Basic control unit: 1 - stator element, 2 - winding sub-element, 3 - inverter module, 4 - microprocessor.

FIG. 14. Longitudinal section of a 1.5-MWPM motor for ship propulsion designed by ABB: 1 - rotor, 2 - armature, 3 - electronics, 4 - housing.

The large PM motor has a simple construction, compact stator winding with short overhangs, and reduced magnetic couplings between phases. The last enables each phase winding to be energized individually with no influence of switching of the adjacent phases.

Each stator element can be fed from its own inverter; therefore the inverters can be small. To obtain a better modular drive system, the winding of each stator element has been divided into three identical sub-elements as shown in FIG. 13. Each sub-element is fed from a small 5-kVA PWM inverter module IM and, in total, 88 × 3 = 264 inverter modules are used. Using this advanced modular technology a highly reliable drive has been achieved. Owing to symmetry, the failure of one inverter module causes the switch-off of seven further modules. These modules are in a stand-by mode for substituting other failing inverter modules. In this way 97% of the rated output power can be obtained with the failure of up to eight inverter modules. The basic control unit (FIG. 13) is assigned to two adjacent stator elements. It contains six inverter modules and a command device with a microprocessor.

The motor efficiency at rated speed is as high as 96% and exceeds 94% in the speed range from rated speed to 20% of the rated speed. At low speed, the winding losses can be minimized by connecting the stator winding sub-elements in series.

A similar 1.5-MW PM brushless motor ( FIG. 14), the so-called "MEP motor" (multiple electronic PM motor) developed for submarine and surface ship propulsion. The rotor consists of a steel spider construction carrying a laminated core with surface SmCo PMs mounted on it. Both rotor and stator stacks consist of 0.1 mm thick laminations. The stator is water cooled and the rotor is cooled by natural convection. All rotor parts including PMs are protected against corrosion by fiberglass bandages and epoxy paint. The motor insulating system is designed according to class F. The power electronic modules are mechanically integrated in the motor housing and connected to the corresponding stator units. Hall sensors are used to provide speed and position signals to the power electronics converters. To improve the motor self-starting capability, the numbers of stator and rotor poles are different. The speed is in the range of 0 to 180 rpm, the d.c. inverter input voltage is in the range of 285 to 650 V, the torque is 76 kNm, number of armature poles (teeth) Z = 112, number of rotor PM poles is 2p = 114, number of power electronics modules is 28, diameter of machine is 2.25 m, length 2.3 m, air gap (mechanical clearance) 4 mm and mass 22,000 kg.

Very high efficiency can be obtained in the speed range from 10 to 100% of the rated speed.

FIG. 15. Rotors of large synchronous motors studied by AEG: (a) with surface PMs without damper, (b) with surface magnets and damper cylinder, (c) with salient poles and cage winding.

8.3 Study of large PM motors with different rotor configurations

A 3.8-MW, 4-pole, SmCo PM, load commutated, inverter-fed synchronous motor has been investigated by AEG, Berlin, Germany.

The first rotor is designed with surface PMs according to FIG. 15a. It does not have a damper. A retaining sleeve (bandage and epoxy resin) is used to protect the PMs against centrifugal forces. The air gap is limited by mechanical constraints and rotor surface losses.

The calculated subtransient inductance is equal to that of the synchronous inductance. Both amount to 40%. This high value of subtransient inductance is due to the absence of a damper. A PM synchronous motor with such a rotor design and parameters is not suitable for operation as a load commutated PM synchronous motor (large subtransient inductance). A forced commutated converter (PWM inverter) is the most suitable power source in this case. An increase in the air gap of 1.5 times causes a decrease in both synchronous and subtransient inductances to 30%. On the other hand, this design requires 30% more SmCo material to keep the air gap flux density at the previous value.

To feed the PM synchronous motor from a load commutated CSI, the subtransient inductance must be reduced to about 10%. This can be achieved with the use of a damper. FIG. 15b shows the cross section of the designed PM rotor with an outer damper conductive cylinder. The air gap g is twice as much as that in FIG. 15a in order to minimize the losses in the damping cylinder. The required PM material amounts to twice that required in the first design without the damper. A calculated subtransient inductance equal to about 6% is very suitable for load commutated converters. To decrease the surface losses in the damping cylinder, grooves could be made on the conductive cylinder surface. This requires a thicker cylinder, which increases the PM material volume required to obtain the same air gap magnetic flux density.

Table 3. Comparison of large variable speed PM motor drives

A reduction in the amount of PM material with a low subtransient inductance is achieved using the configuration shown in FIG. 15c. Here, internal PMs magnetized radially and pole shoes are used. The pole shoes are not only used to protect the PMs against demagnetization, but also to carry the bars of the damper windings. The volume of PM material is reduced to about 70% of that required for a damping cylinder, but it is still more than in the case of surface PMs without dampers. The subtransient inductance amounts to 6%.

The comparison of discussed above large variable speed PM motor drives is summarized in Table 3.

FIG. 16. Double disk PM brushless motor: (a) cutaway isometric view. 1 - PMs, 2 - stator assembly, 3 - housing, 4 - shock snubber, 5 - shock mount, 6 - rotor shaft, 7 - rotor disk clamp, 8 - shaft seal assembly,9-bearing retainer, 10 - stator segment, 11 - center frame housing, 12 - spacer housing, 13 - rotor disk, 14 - bearing assembly, 15 - rotor seal runner, 16 - rotor seal assembly; (b) stator segment. 1 - cold plate, 2 - slotted core, 3 - winding terminals, 4 - end connection cooling block, 5 - slot with conductors. Courtesy of Kaman Aerospace EDC, Hudson, MA, U.S.A.

Table 4. Design data of large axial flux PM brushless motors manufactured by Kaman Aerospace EDC, Hudson, MA, U.S.A.

cont. to part 2 >>

Top of Page PREV.   NEXT Article Index HOME