2014 17th International Conference on Electrical Machines and Systems (ICEMS),Oct.

22-25, 2014, Hangzhou, China

Effect of Saturation on Iron Loss in PMSM

Xiaofeng Ding1,*, Hong Guo1, Min Du1, Binbin Li2, Guanliang Liu1, Linlin Zhao1
1
School of Automation Science & Electrical Engineering, BeiHang University, Beijing 100191, China
2
Mechanical and Electrical Institute, Hangzhou Wahaha Group Co., Hangzhou 310020, China
*
E-mail:͑dingxiaofeng@buaa.edu.cn

Abstract —In order to improve the efficiency of permanent According to the reference [4], the AC excitation frequency
magnet synchronous motor (PMSM) during all operation load and amplitude directly affect the core loss. The most used
conditions, this paper analyzes the iron loss in stator core taking equation that characterizes core losses is the Steinmetz
into account magnetic saturation. Firstly, the flux density
equation
waveforms at five different positions of stator core have been
studied respectively by finite element analysis (FEA) when the P kf D BE (1)
motor are operated in different loads. Fast Fourier transforms
(FFT) of the flux density waveforms gives the explanation for the
changes of the waveforms. Then the losses of five different parts Where, f and B are the frequency and amplitude of flux
are predicted through FEA, which are consistent with the rate of density respectively. Parameters k , D , E are curve fitted
changes of the flux density seriously. In the end, the conclusion is
given that the degree of saturation affects the iron loss in the with measurement data. D is typically a number between 1
stator core obviously. and 2; E is typically between 1.5 and 3. These coefficients
vary material to material and because of the nonlinearity of
core loss, the Steinmetz parameters are valid for a limited
I. INTRODUCTION range of frequencies and flux densities. When the frequency
PMSMs are supplied by solar/fuel cells with limited energy range is wide, different sets of Steinmetz parameters are
usually used in electric propulsion of near-space airship and defined to maintain accuracy all over the range. Some
wheel hub motor systems of lunar vehicle. In order to utilize literatures reported different ways to use unified Steinmetz
the limited energy sufficiently, the efficiency of motors should parameters to model the core loss over a wide range [5-6].
be improved. Meanwhile, servo drive motors would be These parameters are still derived by fitting the measurement
operation under different speeds and loads according to the data.
different tasks of airship or lunar vehicle. Therefore, it’s
important that PMSM should work with high efficiency during
all operation load conditions.
The iron loss is accounted for a larger proportion of the total
losses in PMSM. Therefore, the pursuit of ultra-efficient
motors relies heavily on accurate grasp of the motor iron loss
mechanism and precise calculations. And the iron loss
depends on the flux waveform, frequency and amplitude [1].
When the motors are running during full load conditions, the
magnetic would exist unsaturated, saturated and local severe
saturated [2]. Different degree of saturation will affect
magnetization process in the silicon steel, thereby different
iron loss will be induced. Fig. 1 Core loss vs. peak flux density and frequency
Therefore, alternating, rotating, fundamental and harmonic (3F3 from Ferroxcube) [7]
magnetization [3] are analyzing separately under different
degree of saturation in this paper firstly. Then fast Fourier Unfortunately, the Steinmetz equation is only valid for
transforms (FFT) of the flux density waveforms has been done sinusoidal excitation. This is a huge drawback, because the
core of stator is mostly exposed to non-sinusoidal flux
and give the explanation for the changes of the waveforms.
waveforms in motors. Waveform has strong impact to the core
Furthermore, the losses of five different parts are predicted
loss. Instead of sinusoidal waveform, the SPWM and SVPWM
through FEA, which are consistent with the rate of changes of
are the most commonly seen waveforms in motors. Many
the flux density seriously. In the end, the conclusion is given literatures have reported that the core loss under rectangular
that the degree of saturation affects the iron loss in the stator excitation is different from sinusoidal excitation, and the duty
core obviously. cycle plays an important role to the loss [8]. This Fourier
analysis helped the qualitative understanding of the core loss
II. MANY ASPECTS EFFECT CORE LOSS under non-sinusoidal excitations, but it is not quantitatively
The calculation of core losses in motors is difficult and has accurate [9-10].
not yet been entirely solved. Particularly, the influence of One of the approaches based on an improved Steinmetz
frequency, amplitude, waveform, DC bias and temperature on equation is derived in [11-13]. The approach leads to the
the losses is not entirely clarified. improved generalized Steinmetz equation (iGSE)
3356

978-1-4799-5162-8/14/$31.00 2014 IEEE

 D 1.5
1 T dB
³
10A
('B) E D dt

Radial Flux Density/T

P ki (2) 1
T 0 dt 25A
where 'B is peak-to-peak flux density and 0.5 50A

k 0 75A
ki 2S D
(3) 109A
(2S )D 1 ³
-0.5
cos T 2E D dT 150A
0 -1
The parameters N, ¢, and £are the same parameters as used 200A
-1.5
inthe Steinmetz equation (1). The iGSE is capable of
0 5 10 15 20
calculating losses of any flux waveform, without requiring Time/ms
extra characterization of material parameters beyond the ˄a˅
parameters for the Steinmetz equation. This approach is
1.5
widely applied [14-15].

Tangential Flux Density/T

10A
1
25A
III. MODES OF MAGNETIZATION IN STATOR CORE
0.5 50A
In this paper, the Toyota Prius HEV traction motor is chose
as the analyzing model. The basic parameters are shown in 0 75A

Table 1. The flux distribution is complex in stator core of -0.5 109A

PMSM, as shown in Fig. 2. This motor is driven by sinusoidal 150A
power supply and the peak value of current is 109A. There are -1
200A
both tangential and radial fluxes at five points, which should -1.5
be called rotational flux. If the small tangential flux is     
Time/ms
neglected, it could be considered only alternating flux exists at
the point 3. And the amplitude of tangential component is ˄b˅
bigger than the radial component both at point 4 and 5. The Fig. 3 Flux density under different currents at point 1
teeth tip near the air gap is exposed to both big tangential and
big radial flux density, as shown at point 1 and point 2. 2
According to the spatial harmonics of the motor and time 10A
Radial Flux Density/T

1.5
harmonics of the current [16], there are abundant harmonics in 25A
1
flux density at every point. When the motor is supplied by
PWM inverter, the harmonics of current will further increase 0.5 50A

the harmonics of flux. 0 75A

-0.5 109A
Table 1 Basic parameters of the motor -1
150A
Power 50 kW Slots 48 -1.5
200A
-2
Poles 8 Out radius of stator 134.95mm
    
Velocity 900rpm Inner radius of stator 80.9mm Time/ms
˄a˅
Slot fill factor 0.84 Length of stator 83.8mm
1.5
Tangential Flux Density/T

10A
1
25A
0.5
50A
0
75A
-0.5 109A
-1 150A
-1.5 200A
    

Time/ms
˄b˅
Fig. 4 Flux density under different currents at point 2

Fig.2 Flux density distribution in stator core when the motor is driven by
sinusoidal power supply, Ipk=109A. (a) the points in stator core, (b) radial
flux density, (c) tangential flux density.

3357
 2 0.6
10A 10A

Radial Flux Density/T

1.5
Radial Flux Density/T
25A 0.4 25A
1
0.5 50A 0.2 50A
0 75A 0 75A
-0.5 109A 109A
-0.2
-1 150A
150A
-1.5 200A -0.4
200A
-2 -0.6
0 5 10 15 20 0 5 10 15 20
Time/ms
Time/ms
˄a˅
˄a˅
2

Tangential Flux Density/T

Tangential Flux Density/T

0.2
10A 10A
0.15 1.5
25A 1 25A
0.1
0.05 50A 0.5 50A
0 75A 0 75A
-0.05 109A -0.5
109A
-0.1 150A -1
150A
-0.15 200A -1.5
-2 200A
-0.2
0 5 10 15 20 0 5 10 15 20
Time/ms Time/ms
˄b˅ ˄b˅
Fig. 5 Flux density under different currents at point 3 Fig. 7 Flux density under different currents at point 5

0.5 In order to analyze the effect of magnetic saturation on flux

10A
density, different amplitude of currents has been injected into
Radial Flux Density/T

0.4
0.3 25A the motor, as shown in Fig.3-Fig.7. The peak amplitudes of
0.2 current are 10A, 25A, 50A, 75A, 109A, 150A, 200A
50A
0.1
respectively.
0 75A
-0.1 Both radial component and tangential component have more
-0.2
109A than 1T flux density and apparent dB/dt at point 1 and point 2
-0.3 150A as shown in Fig.3 and Fig.4 respectively, which are near the
-0.4 air gap. And the waveforms of flux density vary obviously
200A
-0.5 from 10A to 200A. The amplitude of tangential component is
     less than 0.5T at point 3, which is similar with the radial one at
Time/ms both point 4 and 5 as shown in Fig. 5, 6, 7. However, the
(a) waveforms of these three components have high rate of
change dB/dt. The residual three components of the point 3, 4,
1.5 5, have high amplitude and almost steady waveforms when the
Tangential Flux Density/T

10A
1 currents change from 10A to 200A.
25A Fig. 8 and Fig. 9 give the explanation for big change rate of
0.5 50A flux density at point 1, which are FFT results of radial and
0 75A tangential flux density. According to the increase of current,
109A the fundamental amplitude (60Hz) of flux density increase
-0.5
150A gradually, and five times, seven times and 11 times harmonics
-1
200A increase significantly as shown in Fig.8. This explains the
-1.5 tangential flux density waveform changes according to the
     vary currents as shown in Fig. 3b. Similarly, the radial flux
density waveform has the similar change with the tangential
Time/ms
one as shown in Fig.9.
(b)
Fig. 6 Flux density under different currents at point 4

3358
Fig. 8 Harmonics of tangential flux density at point 1 Fig. 9 Harmonics of radial flux density at point 1

IV. LOSS CALCULATION higher change rate of flux density dB/dt induces more iron
In order to calculate the losses for the corresponding points, loss. The loss also increases with the amplitude of flux density
five small circles with radius of 0.2mm have been put on the enlargement according the enhancement of the current. Both
five points and the positions of the points are set as centers of of these could be explained by equation 2.
the five circles respectively. Fig.10 shows the transient losses Meanwhile, the average losses for every circle with seven
at point 1 when the peak amplitudes of motor phase current different currents have been calculated presented in Fig.14.
are 10A, 109A and 200A respectively. And Fig.11, Fig.12 and The most increment is given by point 1 when the current is
Fig.13 give comparisons between transient loss and the from 10A to 200A, which have already been explained in part
corresponding flux density, which present that the loss III of this paper in detail. The core losses of point 1 and 2 is
depends on dB/dt and amplitude of flux density directly. The more than other three points due to the losses are induced by
both radial and tangential flux density.

Fig. 10 Losses at point 1 when the peak amplitudes of phase current are 10A, Fig. 11 Core loss and corresponding flux density at point 1 when hen the
109A and 200A respectively peak amplitude of phase current is 10A
3359
 VI. REFERENCES
[1] Mu M, "High Frequency Magnetic Core Loss Study" [D],Virginia
Polytechnic Institute and State University, 2013.
[2] Zhu, Z.Q.; Ng, K.; Howe, D., "Analytical prediction of stator flux
density waveforms and iron losses in brushless DC machines,
accounting for load condition," [J], Electrical Machines and Systems,
2001. ICEMS 2001. Proceedings of the Fifth International Conference,
pp.814,817 vol.2, Aug 2001
[3] Huang Y, Dong J, Zhu J G, et al, "Core loss modeling for permanent-
magnet motor based on flux variation locus and finite-element method "
[J], Magnetics, IEEE Transactions on, Vol.48,No.2,1023-1026, 2012
[4] Chas P. Steinmetz, “On the law of hysteresis” [J], Proceedings of IEEE,
Vol,72, Issue 2, pp. 197-221, Feb. 1984.
Fig. 12 Core loss and corresponding flux density at point 1 when hen the
[5] Ray Ridley, Art Nace, “Modeling ferrite core losses”,[J], Switching
Power Magazine, vol. 3, no. 1, pp. 6–13, 2002.
peak amplitude of phase current is 109A
[6] Christopher Oliver, “A New Core Loss Model",[J], Switching Power
Magazine, Volume 3, Issue 2, 2002.
[7] Datasheet of 3F3 MnZn ferrite, Ferroxcube, 2008 Sept. 01.
[8] Rudy Severns, “HF core losses for non-sinusoidal waveforms”,[C],
Proc. HFPC 1991.
[9] Albach, M. Durbaum, T., Brockmeyer, A. , "Calculating core losses in
transformers for arbitrary magnetizing currents a comparison of
different approaches,"[C], Power Electronics Specialists Conference,
1996.
[10] Wasekura M, Wang C M, Maeda Y, et al, A transient core loss
calculation algorithm for soft magnetic composite material[C], ECCE,
3719-3725, Denver, USA, 2013.9.15-19
[11] J. Reinert, A. Brockmeyer, and R. De Doncker, “Calculation of losses
in ferro- and ferrimagnetic materials based on the modified Steinmetz
equation,” [J], IEEE Trans. Ind. Appl., vol. 37, no. 4, pp. 1055–1061,
Jul./Aug. 2001.
Fig. 13 Core loss and corresponding flux density at point 1 when hen the [12] J. Li, T. Abdallah, and C. R. Sullivan, “Improved calculation of core
peak amplitude of phase current is 200A loss with nonsinusoidal waveforms”[C], in Ind. Appl. Conf., 36th IEEE
IAS Annu. Meeting., 2001, vol. 4, pp. 2203–2210.
[13] K. Venkatachalam, C. R. Sullivan, T. Abdallah, and H. Tacca,
“Accurate prediction of ferrite core loss with nonsinusoidal waveforms
using only Steinmetz parameters” [J], in Proc. IEEE Workshop Comput.
Power Electron., 2002, pp. 36–41.
[14] J. Biela, U. Badstuebner, and J. W. Kolar, “Impact of power density
maximization on efficiency of DC–DC converter systems” [J], IEEE
Trans. Power Electron., vol. 24, no. 1, pp. 288–300, Jan. 2009.
[15] I.Villar,U.Viscarret, I. Etxeberria-Otadui, and A. Rufer, “Global loss
evaluation methods for nonsinusoidally fed medium-frequency power
transformers” [J], IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4132–
4140, Oct. 2009.
[16] Xiaofeng Ding, Chris Mi, “Modeling of eddy current loss and
temperature of the magnets in permanent magnet machines” [J], Journal
of circuits, systems, and computers, Vol. 20, No.7, 1287-1301ୈ2011

Fig. 14 Core losses at five positions with different currents

V. CONCLUSION
This paper presented alternating, rotating, fundamental and
harmonic magnetization at five different positions in stator
core of PMSM. The effect of magnetic saturation on flux
density and iron loss were analyzed through FE model. The
bigger change rate and amplitude of flux density occurs
depending on deeper degree of magnetic saturation, which will
induce more core loss.

ACKNOWLEDGMENT
This work was supported by the Aeronautical Science
Foundation of China (2013ZC51031).

3360