Time-Domain modeling of saturation in conformal mapping ... · [3] D. C. Hanselman, Brushless...

2. QUASI-STATIC FIELDS , 11. ELECTRIC MACHINES AND DRIVES

Time-Domain modeling of saturation in conformal mapping approach for PermanentMagnet Synchronous Machines

Martin Hafner, David Franck and Kay HameyerInstitute of Electrical Machines – RWTH Aachen University

Schinkelstraße 4, D-52062 Aachen, GermanyE-mail: Martin.Hafner@IEM.RWTH-Aachen.de

Abstract— In the electromagnetic field simulation of modernservo drives, the computation of higher time and space harmonicsis essential to predict torque pulsations, radial forces, rippletorques and cogging torque. Field computation by conformalmapping (CM) techniques is a time-effective method to computethe radial and tangential field components. In the standard CMapproach, computational results of cogging torque simulationsas well as overload operations observe deviations to nonlinearfinite element simulations due to the neglection of slot leakageand saturation effects. This paper presents an extension of theclassical CM. Additional CM parameters are computed fromsingle FE computations so as to consider both effects listedabove in the model over a wide operation range of the electricaldrive. The proposed approach is applied to a surface permanentmagnet synchronous machine, and compared to numerical resultsobtained by FEA. An accuracy similar to that of FE simulationsis obtained with however the low computation time that ischaracteristic for analytical models.

I. STANDARD CONFORMAL MAPPING

The air gap field computation by conformal mapping isgenerally obtained from the solution of a linear Laplace prob-lem, assuming the magnetic core has an infinite permeability.Since this system is linear, the field excitation by magnets andcoils, as well the influence of the slotting, can be modeledindividually.

Assuming a slotless stator, the field ~B (Θ) at a certaincoordinate angle Θ in the air gap, Θ ∈ [0, 2π[, consists of aradial flux density Br (Θ) and a tangential flux density Bϕ (Θ)

~B (Θ) = Br (Θ) · ~er +Bϕ (Θ) · ~eϕ. (1)

The angle dependent quantities Br (Θ) and Bϕ (Θ) can beexpanded into a Fourier Series

~B (Θ) =

∞∑n=0

(Br,n · ~er +Bϕ,n · ~eϕ) enpΘ, (2)

where n is the frequency order and p the number of polepairs. In this representation of the air gap field, the Fouriercoefficients Br,n and Bϕ,n are the solution of a linear Laplaceproblem with magnets and a slotless stator depending on themagnetization configuration [1], [2], [3]. The field at a certaininstance of time t due to rotor movement is given by

~B (t) = ~B · eωrt (3)

where ωr is the angular speed of the rotor.Stator slotting significantly influences the magnetic field

distribution. It is modeled by "permeance functions". Thesepermeance functions ~λ represent the impact of slotting on theslotless field distribution and can be obtained by Schwarz-Christoffel transformations [4], [5]. Correlating the field dis-tribution with slotting, s ~B (t), with the field without slotting

(3), yields the vectorial permeance ~λs ~B (t) = ~λ∗ · ~B (t) (4)

~λ∗ =

(λr λϕ−λϕ λr

). (5)

The air gap field excited by the armature current is givenby

a ~B (t, I, ψ) =

p ~B(√

2Ie(ωst+0◦))

p ~B(√

2Ie(ωst+120◦))

p ~B(√

2Ie(ωst+240◦)) · e(φq+ψ+0◦)

e(φq+ψ+120◦)

e(φq+ψ+240◦)

,

(6)where p ~B is the normalized field due to one phase and alsoobtained by conformal mapping [6]. In (6) the angle φqdefines the relative phase orientation to the quadrature axisof the machine, ψ is the flux weakening angle and ωs isthe stator current angular frequency. We shall in the sequelsystematically omit the coordinate Θ and the time argument t,and only indicate the dependency in I and ψ when applicable.We shall also label the quantities obtained by the conformalmapping approach with a CM exponent. The overall air gapfield is thus defined as

g ~BCM (I, ψ) = a ~BCM (I, ψ) + s ~BCM . (7)

II. LEAKAGE AND NONLINEARITY

The main idea of rewriting the CM governing equationssketched in section I is to obtain a CM formulation in whicheach parameter represents a physically motivated quantity inorder to distinguish their origin within the electromagneticfield computation. The definition of the armature field a ~B in(6) includes an implicit formulation of ~λ in the ansatz functionof the field p ~B. For further purposes, ~λ∗ must be factorized

g ~BCM (I, ψ) = ~λ∗ ·(~BCM + a

u~BCM (I, ψ)

), (8)

where an arbitrary permeance state ~λ is identical for bothfield fraction of g ~BCM (I, ψ). In (8) the auxiliary fieldau~BCM (I, ψ) is defined by

au~BCM (I, ψ) =

(~λ∗)−1

· a ~BCM (I, ψ). (9)

A. Ideal CaseThe comparative study [7] for modern analytical models

for the electromagnetic field prediction concludes that therelative error of the air gap flux density obtained by CMwith respect to FE prediction increases strongly in functionof the slot-opening to slot-pitch-ratio. In order to analyze thiswith CM field computation, a PMSM with a slot-opening-factor of 43% is studied. The cross-section of the motor is

U

ZV

XWY

Fig. 1: PMSM Cross-Section

TABLE I: PMSM Parameters

3 Number of Pole Pairs18 Number of Stator Teeth

0.73 Pole Pitch Factor3 mm Permanent Magnet Height

24.5 mm Rotor Radius (inc. PM)0.8 mm Air Gap Height

54.2 mm Outer Stator Radius43 % Slot Opening Factor

1.35 T Remanence Flux Density101 mm Length4 mm Yoke Width

depicted in Fig 1. All parameters of the geometry and theelectromagnetic evaluation are given in Table I. The torquecomputation by Maxwell stress theory in case of CM is givenin [8]. Figure 2(a) compares the cogging torque in functionof rotor position and the corresponding time step obtained by(8) and by a FE computation where the stator is modeled asinfinite permeable iron (IFEA), utilizing Neumann boundarycondition. One observes that in several rotor positions thetorque results are close to each other, but in other angle rangesCM overestimates the integral quantity torque, which is inagreement to the statements of [7]. The maximum deviationoccurs at step three, where the edge of the magnet is alignedwith the center of the slot. The corresponding flux densitydistribution, depicted in Fig. 3, shows a significant increase inits tangential component evoked by a flux path deformationin direction of the stator teeth. This effect is also known asslot leakage [9], and cannot be covered by the scalar vectorquantity ~λ in (4). Adding this effect as optional correction, (8)becomes

g ~BCM−Mod (I, ψ) = ~λ∗ ·(~BCM + a

u~BCM (I, ψ) + ∆ ~B

),

(10)where ∆ ~B is the slot leakage of the rotor field. ∆ ~B can bedetermined by a single IFEA in no-load case. A comparisonof the torque results in rated operation in case of IFEA andthe application of (8) and (10) shows that the deviation ofstandard CM vanishes, see Fig. 2(b) and Fig. 3.

-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

0 2 4 6 8 10

0 10 20 30 40 50 60

Torq

ue [N

m]

Computational Step

Rotor Position in Electrical Degree

IFEACM

CM-Mod

(a) Cogging Torque

22

23

24

25

26

27

28

29

30

0 2 4 6 8 10

0 10 20 30 40 50 60

Torq

ue [N

m]

Computational Step

Rotor Position in Electrical Degree

IFEACM

CM-Mod

(b) Torque for J = 6 Amm2 .

Fig. 2: Comparison of cogging and rated torque obtained byIFEA, CM and CM-Mod.

B. Nonlinear Case

The formalism in II-A to obtain (10) can also be appliedin order to represent saturation. In this case, the permeancestate ~λ turns from a constant quantity into a function of themagnitude of the current I and the flux weakening angle ψ,

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

90 100 110 120 130 140 150

Ra

dia

l F

lux D

en

sity [

T]

Circumferential Position (degree)

IFEACM

CM-ModRotor

Slot

(a) Br

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

90 100 110 120 130 140 150

Ra

dia

l F

lux D

en

sity [

T]

Circumferential Position (degree)

IFEACM

CM-ModRotor

Slot

(b) Bϕ

Fig. 3: Comparison of radial and tangential flux densityobtained by IFEA, CM and CM-Mod in time step three.(Rotor/Slot denote the relative geometry position)

yieldingg ~BCM (I, ψ) = ~λ∗ (I, ψ) ·

(~BCM + a

u~BCM (I, ψ) + ∆ ~B

).

(11)For single points of interest, e.g. rated operation, ~λ (I, ψ)can be identified from a single nonlinear FE simulationg ~BFEA (I, ψ) by

~λ (I, ψ) =(~BCM + a

u~BCM (I, ψ) + ∆ ~B

)−∗· g ~BFEA (I, ψ)

(12)to further improve the analytic field computation.

III. CONCLUSION

In this paper, the standard CM approach is extended toobtain a set of ansatz functions which can each be related tomeaningful design quantities, modeling the effect of slottingand slot leakage on the rotor and armature field fractionrespectively. This parameters can be re-parameterized by idealand nonlinear FEA to account for leakage and saturationphenomena. A first demonstration on a load and no-load caseof a PMSM yields results which are in good agreement toideal FEA. The extension procedure to capture nonlinear phe-nomena is also shown and will be demonstrated on saturatedoperation points in the full paper.

REFERENCES

[1] Z. Zhu and D. Howe, “Instantaneous magnetic field distribution inbrushless permanent magnet DC motors. III. effect of stator slotting,”Magnetics, IEEE Transactions on, vol. 29, no. 1, pp. 143–151, 1993.

[2] Z. Zhu, D. Howe, and C. Chan, “Improved analytical model for predictingthe magnetic field distribution in brushless permanent-magnet machines,”Magnetics, IEEE Transactions on, vol. 38, no. 1, pp. 229–238, 2002.

[3] D. C. Hanselman, Brushless Permanent Magnet Motor Design, 2nd ed.The Writers’ Collective, Mar. 2003.

[4] D. Zarko, D. Ban, and T. Lipo, “Analytical calculation of magnetic fielddistribution in the slotted air gap of a surface permanent-magnet motorusing complex relative air-gap permeance,” Magnetics, IEEE Transactionson, vol. 42, no. 7, pp. 1828–1837, 2006.

[5] ——, “Analytical solution for cogging torque in surface Permanent-Magnet motors using conformal mapping,” Magnetics, IEEE Transactionson, vol. 44, no. 1, pp. 52–65, 2008.

[6] K. J. Binns, Analysis and computaion of electric and magnetic fieldproblems,. Pergamon Press; [distributed in the Western Hemisphereby Macmillan, New York], 1963.

[7] L. Wu, Z. Zhu, D. Staton, M. Popescu, and D. Hawkins, “Comparisonof analytical models for predicting cogging torque in surface-mountedPM machines,” in Electrical Machines (ICEM), 2010 XIX InternationalConference on, 2010, pp. 1–6.

[8] M. Hafner, D. Franck, and K. Hameyer, “Static electromagnetic fieldcomputation by conformal mapping in permanent magnet synchronousmachines,” Magnetics, IEEE Transactions on, vol. 46, no. 8, pp.3105–3108, 2010. [Online]. Available: 10.1109/TMAG.2010.2043930

[9] K. Oberretl, “Die Berechnung des Streuflusses im Luftspalt von elek-trischen Maschinen mit Käfig- oder Dämpferwicklungen Teil 1: Theorieund Berechnungsmethoden,” Archiv fur Elektrotechnik Berlin, vol. 69,no. 1, pp. 11–22, 1986, compendex.