X International Symposium on Industrial Electronics INDEL 2014, Banja Luka, November 0608, 2014

IPMSM Inductances Calculation Using FEA

Dejan G. Jerkan, Marko A. Gecić and Darko P. Marčetić

Department for Power, Electronic and Telecommunication Engineering
Faculty of Technical Sciences, University of Novi Sad
Novi Sad, Serbia
[email protected]

Abstract – Accurate determination of interior permanent magnet reluctant component. In recent years, the magnetic and
synchronous machine’s (IPMSM) inductances is very important thermal capabilities of the PM have been enhanced by
issue, especially in areas of high-performance drives and systems. employing permanent magnets with high coercitivity [1].
This paper presents the method for calculation of the direct and IPMSMs are used in more and more applications because of
quadrature inductances of permanent magnet synchronous
their small volume, very good efficiency, lower moment of
machine using finite element analysis (FEA), where the
calculation of these parameters is based on the determination of inertia, rotor without heat problem, etc. [2]. Because of
flux linkages. Two types of IPMSMs are investigated, with demands of high-performance drives it is very important to
tangentially and radially magnetized permanent magnets. The calculate as accurate as possible the values of the parameters
results of the calculated inductances are presented by diagrams of the IPMSM. Of the most important significance are the
and they are discussed and compared with those obtained by direct- and the quadrature- axis inductances, as they are
measurements. determining corresponding synchronous reactances [3]. Also,
Keywords – Finite element analysis, direct and quadrature axis they are the most important parameters when steady state and
inductances, magnetic flux density, interior permanent magnet dynamic models of IPMSM are developed [4]. Unlike surface
synchronous machine PM motors, which have the same value of inductance in direct
and quadrature axes and where all the torque is produced by
NOMENCLATURE the magnet flux, interior permanent magnet motors have
vd ,vq – stator d- and q- axis voltages different direct and quadrature inductances which results in an
i d, i q
– stator d- and q- axis currents additional torque component called reluctance torque [5]. The
ia,b,c – stator phase currents conventional methods of testing for determination of
Rs – stator phase resistance synchronous machine parameters are often inappropriate in the
m – permanent magnet flux case of permanent magnet machines, because magnetic field
Ld ,Lq – stator d- and q- axis self inductances produced by the permanent magnets cannot be canceled
ds – stator d-axis flux during measurements, and its field affects the total level of
qs – stator q-axis flux saturation of iron during experiments. On the other hand, finite
ω – actual rotor angular speed element method provides great opportunities for accurate
mel – electromagnetic torque numerical analysis of IPMSM, because using FEA algorithms
mm – load torque allows calculations of fundamental field quantities (such as
J – motor inertia flux linkages and stored magnetic energy), and also the fields
p – number of pole pairs
produced by PMs can be canceled easily [6]. Standard
– saliency ratio (Lq/Ld)
experiments are reconstructing parameters of the machine
Js – current density vector
Az – z component of the magnetic vector potential
based on quantities which can be measured through electrical
µ – permeability of material connection of the machine, without knowledge of field
distribution inside of the machine. During the last two decades
the finite element method proved to be the most appropriate
I. INTRODUCTION numerical method in terms of modeling, flexibility and
The interior permanent magnet synchronous motors (IPMSM) accuracy to solve the nonlinear Poisson’s equation governing
have many advantages, such as high power density and the magnetic field who's concerned a principal element in
possibility for speed regulation in wide range of speeds [1], calculation of machine parameters [6].
[7], [9]. IPMSMs are widely used in high-performance drives
such as industrial robots and high-performance machine tools
because of their advantages on high-torque with additional

134
 II. EQUIVALENT CIRCUIT AND BASIC EQUATIONS Knowing the value of variable of interest in every node of the
Fig. 1 shows the d- and q-axis equivalent circuits of region, combined with the usage of interpolation functions
IPMSM in which magnetic losses due to variable magnetic allows complete definition of the behavior of the variable field
field in stator core are neglected. Based on Fig. 1 the on each element. The precision of the method depends not
mathematical equations of the equivalent dq axis steady state only on the dimensions of elements and their number but also
model of IPMSM in the rotor reference frame are given with: on the type of the interpolation function. As for the numerical
method, the FEA algorithm converges to the exact solution
v d  i d   0 − ωLq  i d   0 
 v  = Rs  i  +   +
provided to increase the number of subdivisions of the
 q  q  ωLd 0   iq  ωψ m  solution domain and to ensure continuity of the interpolation
(1) function of its first derivatives along the borders of adjacent
The electromagnetic torque of the IPMSM has two elements [6]. FEA algorithms used in the problems of
components: fundamental magnetic torque (which is electrical machines analysis are based on the evaluation of the
proportional to the product of the magnet flux and q-axis stator magnetic vector potential A. As already mentioned, 2-D FEA
current), and the reluctance torque (which is dependent on the algorithm will be used in this research, and the starting
saliency ratio and to the product of dq-axis stator current equation for FEA is given by:
components). It is essential to determine IPMSM inductances
to predict reluctance torque as an additional torque component.
Based on Fig.1 torque can be expressed as: δ 1 δAz δ 1 δA
− ( ( )) − ( ( z )) = J s
δx µ δx δy µ δy
T=
3
2
(
p Ψmiq + (1 − ρ )Ld id iq ) (2)
(3)

Equation (3) is used for magneto-static solutions, so there

is no time varying of currents or motion of rotor.
Adequate usage of FEA algorithms demands definition of
proper boundary conditions, which determine how lines of
magnetic potential vector pass through areas that separate
regions with different magnetic properties. Analyzed machines
have small area of 2-D cross-section (please see Table 1 for
machine’s dimensions), so only one boundary condition needs
to be defined for this type of FEA, so-called Dirichlet
boundary condition, and the most common use of it is to
define Az=0 along outer stator surface. This means that with
this boundary condition we force the magnetic field to stay
inside boundary defined by stator outer diameter. For larger
geometries it is useful to define additional boundary
conditions which allow usage of only one slice of machine’s
geometry, and the width of that slice is defined by pole width,
which means that for large machines with large number of
pole pairs these boundary conditions can speed-up the
Figure 1. d- and q-axis equivalent circuits of IPMSM a) d-axis equivalent simulations considerably [6].
circuit, b) q-axis equivalent circuit FEA algorithm solves equation (3) for every node of the
mesh created by division of region of interest with finite
elements using some of well-known numerical algorithms,
III. FINITE ELEMENT ANALYSIS OF IPMSM such as Newton-Raphson’s, for instance. After finding values
of Az in all nodes of meshed region, calculating flux linkages
The finite element analysis is used in many areas of of specific phase windings can be done easily. The flux
technical sciences, such as magnetics, electrostatic problems, linkage Ψij of the j-th winding when the i-th winding is
heat transfer, fluid dynamics etc. All FEA algorithms are
based on solution of field equations over domain of interest supplied with current can be expressed by [7]:
using division of that domain with small segments of simple
geometric shapes, called finite elements (in 2-D FEA the most  
1  
common are triangular shapes), in order to reconstruct the
field of the entire domain. Values of field variables inside any
Ψij = p ∫
Sj  + ∫
 Az dΩ − Az dS  N j l j

Ω j Ω −j 
of the elements are represented using 2-D interpolation (4)
functions [6]. These functions are defined on each element where Nj is the number of turns of j-th winding, l j is the
using the values of the calculated variable in each node. length of the stator core and Sj is the cross-section of the coil

135
region. Ω +j and Ω −j represent the positive and negative trace that the 2-D FEA simulation neglects the leakage flux in the
end region, because it cannot be included without usage of 3-D
of the winding in (x,y) plane. The flux vector in d-q domain FEA algorithm [8], [9]. Analyzed machines have concentrated
can be formed as follows: winding, which means that they have very short end
2 connections, especially in comparison with distributed
Ψ = Ψds + jΨqs = (Ψa + aΨb + a 2 Ψc ) (5) windings. Because of that, one may expect very small end turn
3 leakage inductances. However, their influence can be included
2 1 1 by adding some analytical expressions [7] on calculated
Ψds = (Ψa − Ψb − Ψc ) (6)
inductances, or by comparing the 2-D FEA results with
3 2 2
measurements, because measured inductances include those
1
Ψqs = (Ψb − Ψc ) (7) effects. In this paper end connections leakage inductances have
3 been neglected.
The inductances Lq are calculated in a similar manner, only
IV. INDUCTANCE CALCULATION
this time the current vector needs to be aligned with the q axis.
Accurate inductance calculation of PMSM is a relevant The phase currents are then given as:
topic, since the inductances determine large part of the
electrical machine behavior [1]. The inductances estimation in 3
the d-q axis is crucial not only for determination of the torque ia = 0 A , ib = −ic = A (10)
2
and flux weakening capability but also for designing control
systems in order to optimize the efficiency, power factor, etc. The d and q components of the current vector are then defined
[3]. In this paper two types of IPMSM geometries will be as:
analyzed. Both types have identical stator (three-phase four-
pole concentrated winding), and the rotors are chosen in such a id = 0 A , iq = 1A (11)
way to represent two most significant types of PM orientation
used in IPMSMs. First type, here called IPMSM-T (Figure 2a) Figures 6 and 8 show the field solution for this case, for both
is with tangentially magnetized PMs, and the second one, here types of IPMSM. After calculating the flux components, the
called IPMSM-R (Figure 2b) is with radially magnetized PMs. inductances for one turn per coil are given by:
Figure 3 and 4 respectively show PMs air-gap flux distribution Ψqs
Ψds
for the two mentioned types of motors. It is well known that in Ld = , Lq = (12)
the type IPMSM-R quadrature inductance Lq is greater than id iq
direct inductance Ld, because there is more iron along q axis
(Figure 2b). However, such conclusion is not so
straightforward for the type IPMSM-T, because PMs flux path
is between two neighboring magnets, which means through
iron (which increases the inductance), but the flux path is much
longer (which decreases the inductances).
The first FEA simulation will be used to calculate Ld. The
magnet flux is turned off by setting the magnet remanence Br
to zero. The current vector must be aligned with the d axis.
Variation of inductance as a function of current amplitude is of
interest, so series of magneto-static simulations with different
levels of current excitation will be conducted in order to obtain
these relationships. For example, if the magnitude of stator
current is chosen to be 1A, then to align the current axis with
phase a axis the instantaneous phase currents have to be
q
defined as: d
ia
ia = 1A , ib = ic = − = 0 .5 A (8)
2
The d and q components of the current vector are then defined
as:
id = 1A , iq = 0 A (9)
b)
Figures 5 and 7 respectively show the field solution for this
case, for both types of IPMSM. The flux linkages of phases a,
Figure 2. Two types of IPMSM: a) IPMSM with tangentially magnetized
b and c are then calculated using equation (4). The flux vector PMs (IPMSM-T), b) IPMSM with radially magnetized PMs (IPMSM-R)
is constructed and it’s direct and quadrature components are
calculated using expressions (6) and (7). It is important to state

136
 1
B[T] 0.5

-0.5

-1
0 60 120 180
angle [deg]

Figure 3. Flux distribution of permanent magnets in air gap of IPMSM-T

Figure 7. Distribution of flux lines in IPMSM-R when there is only d axis

1
current in stator windings
0.5
B[T]

-0.5

-1
0 60 120 180
angle [deg]

Figure 4. Flux distribution of permanent magnets in air gap of IPMSM-R

Figure 8. Distribution of flux lines in IPMSM-R when there is only q axis

current in stator windings

V. RESULTS OF INDUCTANCE CALCULATIONS

Rated parameters for IPMSM-T machine are given in
Table 1, and IPMSM-R machine was generated for the sake of
comparison with IPMSM-T configuration. As mentioned
Figure 5. Distribution of flux lines in IPMSM-T when there is only d axis above, both machines have the same stator. Results of
current in stator windings calculations described in previous chapter are shown in
Figures 9 (for inductance Ld) and 10 (for inductance Lq). It is
interesting to notice that for both types of IPMSMs quadrature
inductance Lq is approximately 50% greater than direct
inductance Ld, and that inductance Ld is more influenced by
saturation. Results are showing good match with experimental
results (existent only for type IPMSM-T), but there are some
difference between simulations and experiments (both Ld and
Lq calculated with FEA are approx. 10% larger from those
obtained with experiments). This can be explained by the fact
that PMs cannot be excluded from experiments without
destroying the rotor, and their field is added on time-varying
field produced by excitation during experiments, which has
Figure 6. Distribution of flux lines in IPMSM-T when there is only q axis influence on saturation level during measurement. Also, for
current in stator windings
the purpose of FEA simulations authors did not have precise
data for steel used for stator stack (relative permeability as a
function of field strength).

137
 0.03 IPMSM-T FEA
for both types of motors. Further research using FEA software
IPMSM-R FEA combined with measurements may be useful for determination
IPMSM-T measured of end connection leakage inductance influence, and also to
Ld[H]

0.02 investigate saturation of q axis flux path because of the

presence of permanent magnets. Results for type IPMSM-T are
showing good match with the results from the manufacturer of
0.01 the machine. Presented method can be useful for detail analysis
1 2 3 3.5 4 5
Irms[A] of different constructions of IPMSM with possibility to
separate influence of excitation from influence of permanent
Figure 9. Results of FEA simulations and measurements for direct axis magnets field, which cannot be done easily during
inductance Ld
measurements.

0.05 ACKNOWLEDGMENT
This work was supported by the Ministry of Education,
Lq[H]

0.03 Science and Technological Development of the Republic of

IPMSM-T FEA Serbia under project III42004.
IPMSM-R FEA
IPMSM-T measured
0.01 VII. REFERENCES
1 2 3 3.5 4 5
Irms[A]
F. Shiferl, T.A. Lipo, “Power Capability of Salient Pole Permanent Magnet
Synchronous Motors in Variable Speed Drive Applications”, IEEE
Figure 10. Results of FEA simulations and measurments for quadrature axis Trans. Ind. Appl., vol. 26, pp. 115–123, Jan./Feb. 1990.
inductance Lq Y. Li, X.P. Yan, “The perspective and status of PMSM electrical servo
system,” Micromotors Servo Technique, vol. 4, pp. 30-33, 2001.
M. Pastorelli, J. Bottomley, P. Giangrande and C. Gerada, “Sensorless control
TABLE I. IPMSM-T NAMEPLATE DATA of PM motor drives — A technology status review,” IEEE Work. Elec.
Mach. Design Contr. Diag. (WEMDCD), vol. 48, no. 4, pp. 168–182,
IPMSM-T rated data Values March 11-12, 2013.
K. Jang-Mok and S. Seung-Ki, “Speed control of interior permanent magnet
Number of stator slots 12 synchronous motor drive for the flux weakening operation,” IEEE Trans.
Number of poles 8 Ind. Appl., vol. 33, no. 1, pp. 43–48, January/February 1997.
Stator core outside dimension [mm] 102 D. Zarko, “A Systematic Approach to Optimized Design of Permanent
Magnet Motors With Reduced Torque Pulsations”, Pd.D. thesis,
Stator core stack thikness [mm] 42 University of Winsconsin-Madison, USA, 2004.
Stator core inside diameter [mm] 60.6 P. Silvester, “Finite Elements for Electrical Engineers”, Cambridge University
Air gap width [mm] 0.3 Press, 1990.
Winding specification φ0.65×84T, Y ,Al [7] J. Cros and P. Viarouge, “Synthesis of High Performance PM Motors
Resistance (2 phase) 20°C 7.5Ω With Concentrated Windings”, IEEE Trans. Energy. Conv., vol. 17, no.
2, pp. 248-253, Jun 2002.
Direct and quadrature inductances [mH] Ld=20.8 Lq=30.1
[8] T. Cox, F. Eastham, and J. Proverbs, “End Turn Leakage Reactance of
Rated speed [rpm] 500 Concentrated Modular Winding Stators,” IEEE Transactions on
Rated torque [Nm] 1.00 Magnetics, vol. 44, no. 11, pp. 4057-4061, November 2008.
[9] A. M. EL-Refaie, “Fractional - Slot Concentrated-Windings
VI. CONCLUSION Synchronous Permanent Magnet Machines: Opportunities and
Challenges”, IEEE Trans.Ind. El., vol. 57, no. 1, January 2010.
In this paper, the method for calculation of inductances for
two types of IPMSMs was presented. The method is based on
the FEA calculation of flux linkages. It was shown that the
quadrature inductance Lq is greater than direct inductance Ld

138