This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
1

Rotor Design Optimization of Squirrel Cage

Induction Motor - Part I: Problem Statement
Alessandro Marfoli, Mauro Di Nardo, Michele Degano, Chris Gerada, Wenliang Chen

Abstract—Squirrel cage induction motor is the most widely

adopted electrical machine in applications directly fed by the
main grid. The analysis, design and optimization of this machine
topology has been addressed by a considerable amount of
literature over the last century. Although its wide adoption,
the induction motor design, especially when carried out in an
automatic fashion, still presents significant challenges because the
accurate prediction of the performance requires time-consuming
finite element analysis. This work proposes a systematic approach Fig. 1: a) General torque-slip curve with the most important op-
to perform the design optimization of a squirrel cage induction erating points highlighted (starting, pull-up, breakdown, rated). b)
motor focusing on the rotor slot geometry, being this the major Torque-slip curves classification of NEMA.
player in defining the torque-speed characteristic. Structured as a
two-parts companion papers, this first part presents an innovative one does not define any limit and it is employed for higher
performance evaluation methodology which allows a very fast efficiency and breakdown torque applications with respect to
estimation of the torque and efficiency behaviour preserving
the results’ accuracy. The proposed performance estimation the design B. Such motors (class A) typically require the use of
technique is assessed against experimental tests carried out on an a reduced starting voltage strategy. They are typically designed
off-the-shelf induction motor. The selection of the performance with simple rotor slot geometries while class B motors are
indexes to be optimized is justified in detail along with the realized with double or deep rotor slot configurations. Design
description of a generalized rotor parametrization which allows a class C imposes a higher starting torque requirement than
comprehensive exploration of the research space. The presented
optimization procedure is then applied to a case study and the design class B with the same limitation in term of starting
preliminary results are commented, highlighting benefits and current, at the cost of a lower breakdown and pull-up torque
drawbacks of the proposed methodology. and a higher rated slip and a lower rated efficiency. They
Index Terms—Efficiency Improvement, Finite Element Anal- are typically realized with double rotor slot configuration
ysis, Fast Performance Computation, Induction Motor, Multi- with rotor resistance higher than motors design class B. It is
Objective Optimization, Squirrel Cage, Rotor Slot Design. used whenever starting capability under high load is required.
Motor’s design of class D, usually realized with single rotor
slot geometries, guarantee a really high starting torque at the
I. I NTRODUCTION
cost of a very high rated slip and low rated efficiency. They
quirrel cage induction motor (SCIM) is the most
S widespread electrical machine for historical reasons, man-
ufacturing simplicity, robustness and low cost. Typical in-
are used in applications with high peak loads, high inertia
and very intermittent operation [3]. Table I reports the range
of variation of each requirement divided in classes; while in
dustrial applications for directly fed SCIMs include fans, Fig.2 is reported the minimum starting, breakdown and pull-up
pumps, compressors, conveyors, presses, elevators, extractors, torques as a function of the power rating of the motor, along
etc, with the first three covering the majority of the mar- with the maximum starting current for a 4-poles SCIM class
ket [1]. National and international standards provide a set B (which covers about 90% of all general purpose induction
of minimum performance requirements for general purpose motors [4]). Similar classification and minimum performance
motors. The National Electrical Manufacturers Association in requirements in terms of torque-speed characteristics are de-
US (NEMA) categorises SCIMs, having the same power, in fined by the IEC 60034-12 however, for the sake of brevity,
different typologies [2] according to the maximum value of the are not reported in this paper [5].
starting current and the minimum values of the starting, pull- Before the energy crises in the 1970s, the majority of SCIMs
up and breakdown torque (as shown in Fig.1a)). An overview for general-purpose applications were designed to provide the
of the NEMA classification of small-medium SCIM (power
rating: 0.367-367 kW) is reported in Table I, where torques TABLE I: NEMA classification of SCIM
and currents are expressed in percentage of the respective rated
Class A Class B Class C Class D
values. In Fig.1b), a qualitative representation of the torque-
slip curves of each class is reported. Design classes A and B Starting torque 70-275 70-275 200-285 275
Starting current n.d. 600-800 600-800 600-800
exhibit a similar trend in terms of performance requirements Pull-up torque 65-190 65-190 140-195 n.d.
and they are used whenever the starting torque demanded by Breakdown torque 175-300 175-300 190-225 275
the application is relatively low. Classes A and B differ from Rated slip 0.5-5 0.5-5 1-5 >5
Rated efficiency high medium medium low
each other in terms of the starting current requirement: the first

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
2

posed works lacks in generality, either due to the selected rotor

slot parametrization or the chosen objective function or the
adopted performance evaluation method (which sometimes not
include all the involved non-linearities or it is computationally
too expensive to allow a complete exploration of the research
space).
This work, structured as two-parts companion papers, pro-
poses an automatic design methodology for SCIMs. Although
the procedure is general and potentially suitable to design the
whole machine, in these papers it is applied to optimize only
the rotor slot geometry, being the most critical part in defining
the torque and efficiency performance.
In this first part of the work the optimization problem is
Fig. 2: Torque, current and efficiency requirements defined by
NEMA for a 4-poles, 50Hz SCIM class B.
defined as follows. First, a detailed description of an inno-
vative mixed analytical-FE performance evaluation method is
presented in Sec. II, along with the experimental validation
rated power and operating characteristics at the minimum in Sec. III. It will be demonstrated that the selected approach
cost; the efficiency was not a strict requirement, but rather a guarantees obtaining a good trade-off between the conflicting
consequence of the maximum temperature limit. As the energy requirements of high accuracy and short computational time.
price began rising, motor manufacturers started promoting Afterwards, section IV describes the optimization process and
motors with higher efficiency. Regulatory authorities begun the selection of the objectives and constrains. Finally the
introducing minimum efficiency requirements, providing clas- iterative procedure implemented to relieve the computational
sification of electric motors also according to this figure of burden of the performance identification is presented. A gen-
merit [6]. The NEMA defines two categories of efficiency: eralized rotor slot parametrization, enabling to explore a wide
the energy efficient and the premium efficiency. In Fig. 2, the variety of rotor slot shapes during the optimization, is then
minimum requirement for both the classifications is reported. described in details. Sec. V and VI report the preliminary
Similarly, the IEC 60034-30-1 defines four different classes of optimization results and the final considerations highlighting
efficiency (they are not reported for the sake of brevity), where the benefits of the proposed optimization procedure.
two of them are essentially equivalent to NEMA levels: IE2 The second part of the work will outline a detailed description
(Energy Efficient) and IE3 (Premium Efficiency). In Fig. 2, the of the various optimization results. Different optimum designs
gaps between efficiency classes are quite narrows, therefore obtained considering several constraints and rotor cage materi-
the evaluation of the SCIM performance during the design als will be commented considering also the thermal behaviour
process requires high accuracy to appreciate the improvements. of the machine.
The most accurate evaluation of the SCIM performance can be
achieved with computational expensive time step finite element
II. P ERFORMANCE EVALUATION METHOD
analysis (FEA). However, a very precise and time consuming
FEAs is not suitable when an iterative design procedure is In general, the performance calculation methods of a SCIM
adopted [7]. On the other hand, a fast performance prediction can be divided in two main families. The first one relies
based on analytical formulation might lack in accuracy due to on brute force time-step finite element analysis (FEA), while
the inevitable assumptions of the latter. Although both stator the second is based on the resolution of the single-phase
and rotor geometries affect the SCIM performance, the shape equivalent circuit (EC) [7], [13], [14], reported in Fig. 4. The
of the slot hosting the rotor bars, plays the most important role first method tends to be more accurate but computationally
in determining the behaviour of the torque-speed characteristic expensive, while the results quality of the EC-based methods
of the machine as well as its efficiency as a function of the depends on the accuracy in estimating the unknown lumped
load. Traditionally, the rotor slot design process starts from parameters. The latter can be calculated analytically [15],
a limited set of well-known slot shapes (trapezoidal, rectan- [16] disregarding the parameters’ dependency with respect to
gular, rounded, Boucherot) and geometrical modification are current and frequency or alternatively from a set of FEAs.
then applied until the considered performance specifications Several procedures have been proposed in literature to estimate
are met [8], [9]. By doing so, the quality of the obtained the EC parameters as a function of current and frequency
design largely depends on the engineer experience and on the with less expensive magneto-static or time-harmonic (TH)
tools used to predict the SCIM performance. An attempt to FE simulations [7]. The most common method consists
ensure that the obtained design is not a local optimum, given in FE-replicating the indirect experimental tests, i.e. no-load
the performance indexes and constraints, is to use a formal and locked-rotor tests. In particular, the no-load test, carried
optimization algorithm. Several SCIMs’ optimization studies out by magneto-static or time-harmonic FEAs for different
have been proposed in literature. Some of them [8], [10] current values, provides the equivalent no-load inductance and
employ deterministic optimization algorithm using aggregated iron losses resistance. The subdivision between magnetizing
cost function while others [3], [11], [12] utilize stochastic and leakage components of the inductance can be performed
search methods targeting specific applications. All the pro- through the calculation of the first harmonic of the airgap

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
3

as possible the real working scenario. This method, named

contextual identification, relies on the knowledge of both
amplitude and phase of the equivalent transformer ratio K̄,
which can be determined analytically or from a single no-load
TH-FE simulation [17]. For each operating point, simulated
with a current fed standstill TH-FEA, the magnetizing current
I˙m is the sum of the imposed stator current I˙s and the
Fig. 4: Equivalent circuit. FE calculated current flowing in the reference rotor bar I˙r
divided by the complex ratio K̄. The stator leakage Lls and
magnetizing M inductance can be then determined as:
flux density or via a second set of simulations, supplying ˙ − 6 ψ˙s )
|ψ̇s | sin(6 Im
properly the rotor bars [17]. The locked-rotor tests, executed Lls = (1)
˙
|Is | sin( Im
6 ˙ − 6 I˙s )
via standstill TH simulations for different value of the fre-
quency in the rotor reference frame, aims at determining the |ψ̇s | sin(6 ψ˙s − 6 I˙s )
rotor impedance. The equivalent impedance resulting from M= (2)
this test can be determined either from the FE-calculated |I˙m | sin(6 Im
˙ − 6 I˙s )
rotor joule losses and the total magnetic energy or applying where ψ̇s is the FE-calculated stator flux phasor, while the
Kirchhoff’s circuit laws with the FE-calculated voltage and equivalent iron loss resistance is given by:
current phasors [18]. The rotor impedance is then obtained
2
from simple circuital considerations knowing the magnetizing Riron = 3Vms /Pf e (3)
and the stator leakage inductance. This classical approach
where Pf e are the iron losses resulting from the TH-FEA (the
presents two critical issues, strictly correlated with its implicit
stator part of the iron losses is multiplied by 1/s). Once the
assumptions:
parameters on the primary part of circuit have been calculated,
• the EC parameters determined with the FE replicated indi- knowing the complex ratio K̄, the voltage induced on the
rect tests do not change in the real load scenario; reference rotor bar can be calculated as follows:
• the parameters calculated from the no-load tests depends
only from the current while the one determined from the V̇mr = V̇ms /K̄ ∗ = jwM I˙m /K̄ ∗ (4)
locked-rotor test depends only from the frequency; this is
Finally, from the secondary voltage phasor V̇mr , the rotor
not true in the case of closed rotor slot geometries.
parameters can be easily computed as in (5) and (6):
Although the last point can be overcome by identifying the
EC parameters with FEAs as function of both current and Rr = Re {−Vmr
˙ /I˙r } (5)
frequency, at the cost of an increased computational burden
Lrl = Im {−Vmr
˙ /(wI˙r )} (6)
(performing non-linear TH-FEAs), the first issue is an intrinsic
limitation of this procedure. Fig. 3 reports the EC parameters calculated with the above
Recently, a different method to calculate the EC parameters identification method as function of both current and frequency
has been presented [17], [19] with the aim of solving both for a SCIM featuring closed rotor slot; clearly all four pa-
limitations. This alternative approach employs one standstill rameters depends on both current and frequency. Once such
(non-linear) TH-FE simulation, carried out for a single load parameters are identified in the current-frequency plane, and
condition, i.e. imposing a certain current and frequency in the 3D parameters (Lew , Lring , Rring ) have been analytically
the rotor reference frame, to determine all the EC param- calculated [15], [16], the non-linear EC can be solved for
eters related with such operating point. By doing so, all any slip value in order to fully determine the steady state
the EC parameters corresponding to one operating point are performance of the SCIM. Although the contextual identifi-
simultaneously determined from one FEA, replicating as close cation method allows a more accurate estimation of the EC

a) b)

Fig. 3: EC parameters as function of current and frequency: a) stator leakage inductance Ls (continuous lines) and magnetizing inductance
Lm (dashed lines), b) rotor resistance Rr and rotor leakage inductance Lr .

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
4

parameters of any SCIM, including designs featuring closed

rotor slot geometries, it is computationally expensive and
therefore not suitable to be embedded in an automatic design
procedure. The computational time is directly proportional to
the number of currents chosen to identify the behaviour of the
EC parameters for a given frequency, i.e. slip. The choice of
the interpolating functions, used to approximate the parameters
in the resolution of the EC, is important because it can lead to a
reduction of the number of currents to be simulated. However
the selection of these interpolating function is crucial and Fig. 6: The experimental platform.
becomes critical when very similar machine geometries are TABLE II: Machine specifications and parameters
compared, as in the case of automatic design procedure. With
the twofold aim of reducing the computational burden while Parameter Value Unit
ensuring high accuracy during the EC parameters calculation, Rated torque 75 Nm
an iterative procedure is proposed in this work. The latter, Rated frequency 50 Hz
Rated voltage 400 Vrms
shown in Fig. 5, for a given slip, consists of: Number of poles 4 n.d.
• carrying out the time-harmonic FE simulation (with a sup- Number of stator slots 36 n.d.
posed stator current Is,0 at the first iteration); Number of rotor bars 28 n.d.
• identifying the EC parameters via the Eq. (1)-(6);
• solving the non-linear EC and so determine the effective behaviour of the machine. Fig. 6 shows the experimental test
stator current Is,k+1 . rig set up, consisting in: the SCIM under test on the right side
If the relative error between the FE simulated stator current of the picture connected through the torque transducer to a
Is,k and the one calculated solving the EC Is,k+1 is below DC motor acting as a prime mover. The DC motor operate
a certain limit I , the iterative procedure ends, otherwise a in speed control while the SCIM is fed with nominal voltage
new FEA is carried out with the new value of the stator and frequency. Once the steady state temperature of the motor
current Is,k+1 . This procedure permits to reduce the number at rated operating point is achieved, different slip values are
of currents to be simulated at a certain operating point and investigated in order to identify the steady state performance
to avoid the use of interpolating functions required during of the motor under test.
the EC resolution. Implementing this iterative method permits Since the SCIM features a skewed rotor layout, a modification
to reduce the number of simulated currents per slip and the of the EC is required. The approach presented in [20] is
overall computational burden, ensuring high accuracy of the adopted and the modified circuit is reported in Fig. 7 for the
evaluated performance. sake of clarity. The coefficient Ksk compute as in (7) is used
to weight the magnetizing inductance and to split part of it on
the stator and rotor branches of the circuit:

sin(θsk /2)
Ksk = (7)
θsk /2

where θsk is the electrical skewing angle.

Fig. 5: Flow chart of the iterative procedure aimed at evaluating the

machine performance for a given slip. Fig. 7: Veinott modified EC.

The EC lumped parameters are computed using 2D FE

III. E XPERIMENTAL VALIDATION OF THE PERFORMANCE model with straight rotor cage. Some of the parameters are
EVALUATION METHOD then modified according to Fig. 7 and finally the performance
For validation purpose, the performance of an off-the-shelf in term of steady state torque and stator current are computed
SCIM are computed by means of the proposed methodol- by solving the non-linear EC as described in Section II.
ogy and compared against experimental results. The main Fig. 8 shows a comparison between the experimental and
characteristics and geometrical parameters of the motor are the predicted results in per unit. The compared quantities are
reported in Table II. The validation exercise aim to asses showing a good agreement, thus confirming the accuracy of
the accuracy of the approach in predicting the steady state the adopted method.

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
5

1.2 B. Evaluation of the starting and rated performance

Exp test EC-FEA
1
Torque [p.u.]

The starting performance are determined carrying out a

0.8
0.6
single voltage-fed TH-FEA. In fact, in such operating point,
0.4 the iterative procedure based on current-fed time harmonic
0.2 FE simulation, described in Fig. 5, is not needed. At starting
0 condition, s = 1, the rotor is at standstill and so the pulsations
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 of stator and rotor quantities are equal. Consequently, a single
Slip [p.u.] TH-FE simulation supplying the stator windings with symmet-
1.2 ric three-phase voltages, with rated amplitude and frequency,
Stator current [p.u.]

Exp test EC-FEA

1 allows the direct evaluation of the performance. The rated
0.8 operating point is hereafter defined as the condition in which
0.6 the motor provides the rated torque Trated . However, the rated
0.4 slip srated is not a-priori known as it strongly depends on
0.2
the rotor slot geometry. To identify the effective operating
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 point, an iterative procedure, summarized in Fig. 9, has been
Slip [p.u.] implemented, consisting in the following steps.
Fig. 8: Experimental vs EC-FE results: a) Steady state torque-slip i) The values of the rated slip, rated current and the slopes
characteristic and b) stator current-slip characteristic. of both current and torque function at the rated condition
(s, Is0 , mT , mIs ) are initially supposed or determined
following the procedure described at the end of this
IV. O PTIMIZATION PROBLEM DEFINITION section.
ii) The machine performance are evaluated using the iterative
The following three sub-paragraphs detail A) the selection
procedure reported in Fig. 5 hereafter called TH-FE
of the objectives and constrains of the optimization problem,
current iteration.
B) the iterative procedure implemented to relieve the com-
iii) The procedure ends if the obtained torque T (i) is equal
putational cost of their identification, and C) the rotor slot
to the rated torque within the limit defined by T such as
parametrization based on a generalized Boucherot bar.
|T (i) − Trated | < T .
iv) Otherwise, the slip and current values are updated accord-
A. Objectives and constraints ing to (9) and (10) using the results of the last simulation
Starting torque and current, pull-up and breakdown torques, and the slopes mT and mIs :
rated efficiency and power factor can be all considered possible
objectives or constraints of the optimization problem under s(i) = [Trated − T (i − 1) + mT · s(i − 1)]/mT (9)
study. In fact, all of them are deeply influenced by the rotor Is0 (i) = mIs · [s(i) − s(i − 1)] + Is (i − 1) (10)
slot geometry and their values have to comply with the bound-
aries defined by the national or international standard (except v) The machine performance are reevaluated by means of
the power factor). If all the above mentioned performance the TH-FE current iteration.
indexes are determined, the subdivision between objectives vi) If |T (i) − Ttarget | < T , the rated operating point is
and constraints depends on the specific application under identified.
study. For example, applications where the rated efficiency vii) Otherwise the slope of both current and torque are up-
is the most important requirement, the latter can be treated as dated by means of (11) and (12):
the only objective of the optimization problem while all the mT = [T (i) + T (i − 1)]/[s(i) − s(i − 1)] (11)
remaining indexes can be considered as constraints. However,
the accurate evaluation of all of them is computationally mIs = [Is (i) + Is (i − 1)]/[s(i) − s(i − 1)] (12)
expensive because it implies the determination of the full
and the new current and slip are determined according to
torque-speed characteristic. In this work, with the aim of
(9) and (10).
reducing the computational burden and the complexity of
viii) The machine performance are reevaluated again.
the problem, only the rated efficiency, the starting torque
ix) The iteration stops if the inequality |T (i) − Ttarget | < T
and current are determined during the optimization process.
is satisfied, otherwise the procedure restart from step vii.
The first two (ηrated , Tstart ) are selected as objectives while
the ratio between the starting and the rated current (kI ) At the end of every iteration the rated performance
is considered as a constraint; the optimization problem is (s, Is , m T , mIs ) corresponding to a certain rotor geometry,
therefore defined as: described by the vector x, are stored. During the optimization,
when evaluating the performance of the geometry x∗ , with
max (ηrated , Tstart )
(8) the aim of reducing the number of FE simulations carried
s.t. kI ≤ kI,max out during the search of the rated current and slip, the initial
Selecting the current ratio as a constraint and not as the third values of the variables used to start such iterative procedure
objective reduces the complexity of the optimization problem are imposed equal to the corresponding p
variables of the closest
and so increase the quality of the results [21]. geometry previously evaluated x min . In other words, before

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
6

the distance between the outer and central circles hout and
the central and the inner circles hin and the external bridge
thickness b. All these 7 variables are defined in per unit of
their respective maximum values. In particular the following
drawing sequence has been adopted:
i) the bridge thickness b is calculated knowing its per unit
value bpu and its upper bmax and lower limits bmin :
b = bmin + bpu · (bmax − bmin ) (14)
ii) the outer circle radius can be defined once its maximum
value has been calculated from geometrical consideration:
(
Rout−max = [sin α/(1 + sin α)] · (Rr − b)
(15)
Rout = Rout−max · Rout−pu
iii) the distance between the outer and central circles follows
after the calculation of its maximum allowed value:
(
hmax = Rr − Rout /2 − Rs
(16)
hout = hmax · hout−pu
iv) while the distance between the central and inner circle is:
hin = (hmax − hout ) · hin−pu (17)
v) once the center of the central xm and inner circles xin are
defined, the maximum value of their radius are calculated
imposing geometrical constraints:
(
Rm−max = min(xm · sin α, xm − Rs , Rr − b − xm )
Rin−max = min(xin · sin α, xin − Rs , Rr − b − xin )
(18)
thus allowing to calculate the radii Rm , Rin from their
per unit values;
vi) finally the neck width can be calculated as:
Fig. 9: Flow chart of the iterative procedure aimed at finding the
rated slip and the corresponding performance. d = dpu · min(Rout , Rm ). (19)
starting the evaluation of the geometry x∗ , the euclidean Adopting this per unit parametrization allows obtaining always
distances between the latter and all the previously evaluated geometrically feasible solutions, and to explore a wide variety
solutions xp are determined: of bar geometries including Boucherot types, drop shapes
v
u nv and deep bars as shown in Fig. 11. Varying without any
uX constraint all the 7 variables describing the geometry implies
d(x∗ , xp ) = t (x∗i − xpi )2 (13)
i=1
that the slot surface Aslot is not constant. If the latter has
to be kept constant (Aslot = Atarget ), the variables are not
where nv is the number of variables defining the rotor geome- anymore independent and a correlation among them has to
try. Once all these distances are calculated, the nearest solution be established. Given the difficulty of finding a closed form
xpmin to the current one, can be easily identified as well as its of such relationship, the variables defining the slot geometry
rated performance (which have been opportunely stored). In
h_max
the next section, it will be shown that adopting this ”learning”
ploy the number of FE simulations needed to identify the rated hm hout
performance of a certain geometry is greatly reduced. This is
Rr
because the evaluation starts from the current and slip values
of the most similar geometry previously evaluated.
Rs α b
d
C. Rotor bar parametrization
A generalized parametrization of the Boucherot bar has been Rin
implemented in order to widen the shape variety explored Rin_max
Rm_max Rm Rout
during the optimization. The rotor slot geometry is described
Rout_max
by 7 variables, as shown in Fig. 10, i.e. the neck width d, the
radii of the outer, central and inner circles Rout , Rm and Rin , Fig. 10: Parametrization of rotor bar.

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
7

• both optimizations lead to almost the same results in terms

of rated efficiency and starting performance;
• the real current ratio of the optimal solutions is substantially
equal to the maximum value allowed during the optimization
process.
Analysing means (µ) and standard deviations (σ) of the
geometrical variables of the optimal solutions, obtained con-
sidering the whole set of parameters during the optimization,
Fig. 11: Examples of rotor bar geometries that can be obtained using shown in Table. III, it is evident that the bridge thickness tends
the implemented parametrization. to converge to the minimum value while the distributions of
other variables have a much higher variance. The number of
(all except b) are updated proportionally to the square root geometrical variables to optimize can then be reduced, keeping
of
p the ratio between the target slot area and the actual one the bridge thickness equal to the minimum value, simplifying
Atarget /Aslot . By doing so, the area of the slot equals the the optimization problem without affecting the results (as
target value and its shape is preserved respect to the initial shown in Fig. 12).
geometry.
TABLE III: Mean and variance of the distribution of the optimized
variables in p.u.
V. P RELIMINARY O PTIMIZATIONS RESULTS
b Rout hout Rm d hm Rin
A multi-objective stochastic optimization algorithm µ 0.026 0.543 0.312 0.59 0.082 0.638 0.658
(NSGA-II embedded in Matlab) has been chosen to carry σ 0.005 0.031 0.082 0.064 0.078 0.083 0.081
out the rotor slot design of the SCIM with a population
size of 100 elements evolving for a maximum of 100 The Pareto fronts have been obtained evaluating a total
generations. As previously mentioned, the stator geometry number of 10000 geometries. Each solution requires the as-
has been kept constant during the optimization process and sessment of the starting performance via a single voltage-fed
its main geometrical parameters are reported in Table. II. TH-FEA and a variable number of current-fed TH-FEAs to
Figure 12 reports the results in terms of Pareto fronts determine the rated performance. Fig.13 reports the evolution
of two optimizations carried out considering a maximum of the average (per generation) number of the simulated slips
current ratio kI,max of 7.4 and a constant slot area A∗slot . and the total TH simulations performed during the search of
The first optimization considers the full set of geometrical the rated operating point. It can be observed that the number
variables during the design process (called variable bridge in of simulations steeply decrease at the fifth generation when the
Fig. 12), while in the second one the bridge thickness is kept learning ploy, described in Sec. IV, is activated. Once enabled,
constant and equal to the minimum allowed by manufacturing the iterative procedure to identify the rated performance of
considerations (0.5 mm). The analysis of the results is leading the solution under assessment starts from the current and slip
to the following considerations: values of the most similar geometry previously evaluated. The
• rated efficiency and starting torque show a clear competitive number of simulations continues to reduce as the optimization
behaviour, i.e. the improvement of one of them implies the algorithm proceeds (as shown in the inset of the same figure)
worsening of the other; because more and more results are available to be compared
• the range of variability of the starting torque is much higher with the geometry under evaluation and so the probability
than the respective range of the rated efficiency; e.g. an of having similar slots increases. After few generations, each
increment of 100% of the starting torque (from 140 to 280 solution requires on average 1.25 slips to be simulated in order
Nm) implies a reduction of the rated efficiency of only 1.1%; to find the rated condition. Each of them requires on average
less than 1.5 TH-FE simulations to identify the rated current.

Fig. 12: Pareto fronts obtained considering a current ratio limit

kI,max = 7.4 and constant slot area with and without including Fig. 13: Evolution during the optimization of the average number of
the bridge thickness bpu in the optimization process; with the color slips and simulations executed during the search of the rated operating
scale is shown the real current ratio kI . point.

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2020.3019934, IEEE
Transactions on Energy Conversion
8

VI. C ONCLUSION [2] National Electrical Manufacturers Association. Motors and generators
mg-1, 2011.
This work, first part of two companion papers, has proposed [3] D. Zhang, C. S. Park, and C. S. Koh. A new optimal design method of
an automatic design procedure of squirrel cage induction rotor slot of three-phase squirrel cage induction motor for nema class d
motors including the following novelties. speed-torque characteristic using multi-objective optimization algorithm.
IEEE Transactions on Magnetics, 48(2):879–882, Feb 2012.
• First an original performance evaluation method has been [4] K. Jeon, T. Chung, and S. Hahn. Nema class a slot shape optimization
presented. The latter is based on an iterative resolution of the of induction motor for electric vehicle using response surface method.
In 2011 International Conference on Electrical Machines and Systems,
classical single-phase equivalent circuit featuring parameters pages 1–4, Aug 2011.
determined via current-fed finite element time harmonic [5] IEC. Iec 60034-12: Rotating electrical machines - part 12: Starting
simulations in the real load scenario. The presented method performance of single-speed three-phase cage induction motors, 2016.
[6] A. T. De Almeida, F. J. T. E. T. E. Ferreira, and J. A. C. Fong. Standards
allows a really fast prediction of both torque and efficiency for efficiency of electric motors. IEEE Industry Applications Magazine,
without loosing accuracy as verified by the experimental 17(1):12–19, Jan 2011.
tests carried out on an off-the-shelf induction motor. [7] L. Alberti, N. Bianchi, and S. Bolognani. A very rapid prediction of
im performance combining analytical and finite-element analysis. IEEE
• Albeit the adopted performance evaluation method is general
Transactions on Industry Applications, 44(5):1505–1512, Sep. 2008.
and so suitable to analyse and design the whole machine in [8] S. Williamson and C. I. McClay. Optimization of the geometry of closed
a very time-efficient manner, as a vessel to investigate its rotor slots for cage induction motors. IEEE Transactions on Industry
Applications, 32(3):560–568, May 1996.
capabilities, it has been applied to optimize the rotor slot [9] A. C. Smith. Integrating fe into induction motor design-a marriage
geometry. A generalized parametrization of a Boucherot bar of inconvenience? In IEE Seminar on Current Trends in the Use of
has been then introduced which allows exploring the most Finite Elements (FE) in Electromechanical Design and Analysis (Ref.
No. 2000/013), pages 4/1–4/7, Jan 2000.
common slot shapes (double cages, drop-like, deep bar, etc.) [10] A. Daidone, F. Parsasiliti, M. Villani, and S. Lucidi. A new method
with the minimum number of geometrical variables. for the design optimization of three-phase induction motors. IEEE
• Starting and rated performance in terms of torque and Transactions on Magnetics, 34(5):2932–2935, Sep. 1998.
[11] G. Lee, S. Min, and J. Hong. Optimal shape design of rotor slot in
efficiency are the selected objectives to be optimized during squirrel-cage induction motor considering torque characteristics. IEEE
the proposed automatic design procedure, having considered Transactions on Magnetics, 49(5):2197–2200, May 2013.
only applications directly fed by the main. Being the rated [12] M. De Martin, F. Luise, S. Pieri, A. Tessarolo, and C. Poloni. Numerical
multi-objective optimization of a squirrel cage induction motor for indus-
operating point dependent from the slot geometry, a new trial application. In 2015 Intl Aegean Conference on Electrical Machines
iterative algorithm has been implemented for its identifi- Power Electronics (ACEMP), 2015 Intl Conference on Optimization
cation. The latter, essentially based on the secant method, of Electrical Electronic Equipment (OPTIM) 2015 Intl Symposium
on Advanced Electromechanical Motion Systems (ELECTROMOTION),
starts the identification using the rated values of both slip and pages 170–175, Sep. 2015.
current of the most similar geometry previously evaluated. [13] G. R. Slemon. Modelling of induction machines for electric drives. IEEE
Adopting this approach, the number of FE simulations Transactions on Industry Applications, 25(6):1126–1131, Nov 1989.
[14] N. Bianchi, S. Bolognani, and G. Comelato. Finite element analysis of
needed to identify the rated performance of the machine three-phase induction motors: comparison of two different approaches.
under assessment is greatly reduced. It has also been shown IEEE Transactions on Energy Conversion, 14(4):1523–1528, Dec 1999.
that this computational burden continues to decreases as the [15] A. Boglietti, A. Cavagnino, and M. Lazzari. Computational algorithms
for induction-motor equivalent circuit parameter determination—part i:
optimization algorithm proceeds because the likelihood of Resistances and leakage reactances. IEEE Transactions on Industrial
finding similar geometries increases. Electronics, 58(9):3723–3733, Sep. 2011.
Thanks to this method, the evaluation of the objectives [16] A. Boglietti, A. Cavagnino, and M. Lazzari. Computational algorithms
for induction motor equivalent circuit parameter determination—part
for each induction motor requires 1 TH-FEA for the starting ii: Skin effect and magnetizing characteristics. IEEE Transactions on
performance calculation and about 1.3 TH-FEAs for the identi- Industrial Electronics, 58(9):3734–3740, Sep. 2011.
fication of the rated condition. Considering that each TH-FEA [17] D. Genovese, P. Bolognesi, M. De Martin, and F. Luise. A contextual
parameter identification method for the equivalent circuit of induction
takes approximately 4s on a standard-spec PC (Intel Core i3, machine. In 2016 XXII International Conference on Electrical Machines
3.4GHz, 8GB RAM), the whole optimization process, which (ICEM), pages 25–31, Sept 2016.
evaluate 10000 geometries, takes less than 32 hours. [18] M. De Martin, M. Bailoni, A. Tessarolo, M. Bortolozzi, D. Giulivo,
F. Agnolet, and R. Santarossa. Investigation into induction motor equiv-
Part II of these companion papers will analyse: alent circuit parameter dependency on current and frequency variations.
• several optimization results obtained considering different In 2014 International Conference on Electrical Machines (ICEM), pages
constraints in terms of maximum starting current, rotor 196–202, Sep. 2014.
[19] A. Marfoli, L. Papini, P. Bolognesi, D. Genovese, and C. Gerada.
slot area and cage materials; Analysis of induction machine: Comparison of modelling techniques.
• the accuracy of the efficiency evaluation method imple- In 2017 IEEE International Electric Machines and Drives Conference
mented during the optimization, which does not consider (IEMDC), pages 1–7, May 2017.
[20] O. I. Butler and T. S. Birch. Comparison of alternative skew-effect
the influence of the spatial field harmonics in the losses parameters of cage induction motors. Proceedings of the Institution of
calculation; Electrical Engineers, 118(7):879–883, July 1971.
• the thermal behaviour of the optimized machines and its [21] Deb Kalyanmoy. Multi-Objective Optimization Using Evolutionary
Algorithms. John Wiley Sons, Inc., USA, 2001.
influence on the optimization results, which have been
obtained considering a constant temperature distribution.

R EFERENCES
[1] A. T. de Almeida, F. J. T. E. Ferreira, and G. Baoming. Beyond induction
motors—technology trends to move up efficiency. IEEE Transactions
on Industry Applications, 50(3):2103–2114, May 2014.

0885-8969 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Cornell University Library. Downloaded on September 08,2020 at 21:59:50 UTC from IEEE Xplore. Restrictions apply.