IET Electrical Systems in Transportation

Review Article

Thermal management of electric machines ISSN 2042-9738

Received on 4th December 2015
Revised 16th May 2016
Accepted on 27th June 2016
doi: 10.1049/iet-est.2015.0050
www.ietdl.org

Yinye Yang1 , Berker Bilgin1, Michael Kasprzak1, Shamsuddeen Nalakath1, Hossam Sadek1, Matthias
Preindl1, James Cotton1, Nigel Schofield1, Ali Emadi1
1McMaster Institute for Automotive Research and Technology (MacAUTO), Faculty of Engineering, McMaster University, Hamilton, ON L8P 0A6,

Canada
E-mail: [email protected]

Abstract: Electric machines have broadly been used in many industries including the transportation industry. With the evolving
trend of electrification in transportation, electric machines with higher power density and higher efficiency are demanded and,
thus, more stringent thermal management requirements are needed for electrified vehicle applications. This study
comprehensively presents various important aspects of thermal management in electric machines with the main focus on
transportation applications. Design considerations, challenges, and methods for enhanced thermal management are discussed.
Fundamental thermal properties of common materials are presented and sources of losses in various parts of machines are
explained. Furthermore, typical cooling techniques and thermal analysis approaches for electric machines are reviewed in
detail. This study will serve as a reference guideline for machine designers, who are interested in thermal management, and for
thermal researchers working on electric machines.

1 Introduction thermal management requirements and integrated cooling

techniques [13, 14].
Electric machines are used in many industries including home Thermal management has been applied as a general practice in
appliance, medical and health, food, energy, mining, machining, the manufacturing of electric machines. However, it has not been
manufacturing, and transportation including automotive and comprehensively documented in academic research papers. The
aerospace applications. They generate almost all electric energy majority of the papers focus on one particular aspect of thermal
(with the exception of fuel cells, photovoltaic panels, and batteries) analysis applying to one specific machine topology. Liao et al. [15]
and provide about 60% of total energy conversion from electric discussed thermal analysis for design of high performance motors.
form to mechanical form [1]. Throughout the technology Howey et al. [16] analysed air-gap convection in rotating electrical
development for more than a century, electric machines have machines in an effort to predict machine temperature distribution.
evolved into various configurations under both DC and AC Bennion and Cousineau [17] conducted a sensitivity study on
machine categories. Brushed DC machines were among the most traction motor cooling. It is difficult to find a paper inclusively
popular machines until the 1970s and the advent of brushless DC covering all the aspects of thermal management in electric
machines with power electronic devices took place. The removal of machines. Only until recently, Boglietti et al. [18] published a
mechanical brushes and commutators enables higher efficiency and survey describing the evolution and modern approaches to thermal
reliability, contributing to the extensive applications of brushless analysis of electric machines. It reviewed various thermal analysis
DC machines in a wide range of fields [2]. On the other hand, AC approaches in great detail. However, it focused on the analytical
induction machines (IMs) have widely been used in industrial level without giving too much information about practical design
drives such as fans, pumps, and compressors, while synchronous and manufacturing considerations. In addition, Moreno et al. [19]
AC machines are extensively used in generating electricity in presented thermal management and reliability of automotive
power plants and automotive alternators [2]. In addition, various electric traction-drive systems. However, they focused on limited
other machine configurations exist including switched reluctance parts of thermal management in electric machines. Thus, this paper
machines (SRMs), synchronous reluctance machines, transverse aims to comprehensively discuss different aspects of electric
flux machines, axial flux machines and so on [3]. machine thermal management and general techniques applied.
Thermal management of electric machines is critical in order to Fundamental loss generation principles, material properties,
improve machine performance and ensure reliability. Historically, analysis techniques, cooling methods, and design considerations
machine thermal management relies on design experience and will be covered extensively. The rest of the paper is structured as
simple sizing methods such as winding current density and follows: Section 2 discusses the compound design considerations
operating conditions to a large extent [4]. These methods typically for electric machine thermal management, challenges, and
apply large safety factors to avoid the worst heating scenarios and approaches to enhance the thermal performance. Section 3
result in over-design of the cooling system or over-sizing of the discusses typical material thermal properties and Section 4 explains
machine. With the evolving trend of electrification in fundamental loss generation mechanism in different parts of
transportation, there is growing demand for electric machines in electric machines. Typical cooling techniques are reviewed in
automobiles [5], locomotives [6], agriculture machines [7, 8], and Section 5 and various thermal analysis approaches are then
aerospace [9] ranging from starter-generators to traction propellers presented in Section 6. Finally, Section 7 concludes the paper with
[10]. Machines with higher efficiency, wider speed range, and findings and contributions. The authors hope this paper will serve
better power density are in demand while smaller size, lighter as a reference guideline for machine designers who are interested
weight, and lower cost are desired [11, 12]. All these targets in thermal management and for thermal researchers who are
require optimised machine designs and, thus, better thermal interested in machine technologies.
management is of exceptional importance. In addition, machines
with variable load and speed instead of steady-state operations are
broadly applied. Transient heat profiles lead to more stringent

IET Electr. Syst. Transp. 1

© The Institution of Engineering and Technology 2016
 may occur around the magnets. High temperature can affect the
magnet performance and result in potential demagnetisation [14].
For brushed DC machines, maximum load and speed should be
controlled to prevent excessive heat generated on the brushes and
commutators.
To enhance thermal management in electric machines, it is
essential to resolve challenges from all aspects including the heat
sources, the heat dissipation paths, and the heat removal sinks, as
shown in Fig. 1. Heat generated from iron losses can be reduced by
thinner-cut laminations, advanced steel materials such as cobalt–
iron (CoFe) alloys, and well-controlled annealing method [24, 25].
Decreasing the amount of alloy materials can increase the thermal
conductivity and magnetic performance of laminations, though at
the expense of increased resistivity and eddy currents losses in the
laminations [26]. In addition, magnet losses can be reduced by
segmenting magnets in the axial direction. Lower copper losses can
be achieved by lower winding temperature. On the other hand,
thermal management can be enhanced by facilitating the heat
dissipation along the thermal path. Advanced thermal conductive
Fig. 1 Thermal management in electric machines
materials can be incorporated to reduce contact resistance. Thermal
fillings can be used to increase the heat transfer coefficient. Larger
2 Thermal management considerations heat dissipation surface areas can be considered for better heat
Thermal management is a complex, but crucial aspect in electric dissipation. Cooling paths can be integrated to target at local areas
machine design. For example, high temperature results in high of heat concentration and so on [27]. Finally, thermal sink is also
copper resistance and hence high losses in the windings. Excessive critical for electric machine thermal management. Both active
temperature reduces the remanence and coercivity of permanent cooling and passive cooling can be applied to remove the heat from
magnets (PMs), affecting machine torque and demagnetising the the machine. Increasing coolant flow rate and jet cooling velocity
magnets [20]. Inefficient cooling system leads to faster insulation typically increases the heat transfer coefficients [19]. The
degradation and increased winding failure rates due to increased configurations of water jackets and housing fins also determine the
thermal cycles [2]. It is a rule of thumb that 10°C temperature rise cooling efficiency. More details will be presented in Section 5.
halves the life expectancy of the electric machine [21]. In addition, In summary, good thermal management of electric machines
lamination stacks, coils, and end turns all have a direction- requires careful design considerations from various perspectives.
dependent thermal conductivity [22]. Non-thermally conductive The following sections will describe the issues in detail based on
materials such as magnet adhesives and winding insulations may fundamental material properties, heat generations, cooling
also result in local heat concentration. Therefore, effective thermal methods, and typical thermal analysis used for electric machine
management improves electric machine performance and ensures thermal management.
reliable machine operation, while poor thermal management results
in performance degradation, accelerated machine wear, and even 3 Material thermal properties
may cause machine breakdown.
Thermal management requirements in electric machines vary The temperature rise in an electric machine is usually limited by
depending on the application. For small mechatronic drives in the thermal properties of the materials. Maximum temperature
automotive, medical applications, and home appliances, motors specification of the machine, manufacturing techniques and also
typically work under light load within a short period of time. Thus the machine topology play an important role in material selection.
thermal management is primarily targeted to enable efficient heat Material thermal properties should be carefully evaluated in a
removal at transient conditions. For fixed speed industrial drives thermal management system design process.
where there is typically no space or environment temperature limit,
natural convection and forced convection are often applied. This is 3.1 Conductors
a relatively inexpensive and sufficient solution for steady-state The largest amount of losses in an electric machine is usually due
applications. For power dense wind turbine generators, where to the losses in the conductors. This includes the copper losses in
space is limited and maintenance access is difficult, thermal the stator conductors and, especially in IMs, followed by the losses
management is critical to ensure the continuous operation and meet in the rotor conductors [28]. Magnet wires are widely used in
the service life. Liquid cooling channels in both the generator and electric machines to create the stator magnetic field. The insulation
the gearbox are typically implemented, and a heat exchanger is around the copper conductor enables the contact between the wires
utilised to dissipate heat to the ambient. For traction motor without causing any electrical short circuit. Magnet wire insulation,
applications, separate cooling systems may be required such that a in general, is made of organic material, which softens at a lower
lower-temperature cooling loop is used for electric machine and temperature than copper or electrical steel. The mechanical
power electronics, while a higher-temperature loop is used for the strength of the insulation is also much lower than copper or steel
engine [19]. In electrified powertrain applications, electric traction [29]. Therefore, electrical insulation around the conductors
motors are usually integrated into the transmission. Thus, generally creates the limit on the lifetime of the stator winding.
transmission oil can be utilised for both the transmission cooling American National Standards Association standardises magnet
and electric machine cooling. wires [30]. The selection of the correct magnet wire is highly
Different thermal management requirements are necessary for dependent on the machine operating condition, such as the thermal
different types of electric machines. In general, rotor is more class, insulating (or coating) material, current density, frequency,
difficult to cool during operation due to lack of accessibility. SRM voltage and so on. The lifetime of the insulation reduces by half for
has the simplest machine configuration where there is no magnet or every 10 °C increase in the temperature over its thermal class. The
windings on its rotor, and therefore SRM has the highest capacity insulating layers, the so-called enamels or films, are bonded on the
in terms of high temperature resistance [23]. In comparison, IM copper or aluminium conductor. They are made up of different
generates significant a amount of heat in the rotor as current is types of resins and define the thermal class of the magnet wire
induced in square-cage bars or wounded rotor windings. Without [31]. Table 1 lists the thermal classes of round magnet wires and
proper thermal management, excessive heat can build up and the insulation materials based on some of the widely used ANSI/
transfer to the neighbouring components, leading to negative NEMA MW1000-2015 standards [30]. Round magnet wires form a
impacts on the bearing life. Similarly, special thermal concerns random-wound machine winding, which is generally applied below
should be given to the PM machines as local heat concentration 690 V phase-to-phase voltage. Beyond this point, rectangular wires
2 IET Electr. Syst. Transp.
© The Institution of Engineering and Technology 2016
 high overload capabilities, high temperature operation, automated
winding process and so on [32].
Magnet wires have standardised dimensions, which are
categorised with American Wire Gauge (AWG). Different
standards offer different AWG values, but for the same AWG
number, the wire size should be the same even though the
insulating material is different. Magnet wires have an insulation
layer (or build), which defines the grade of the wire. Generally four
different grades are available: single-, heavy-, triple-, and
quadruple-build, in the order of increasing insulation thickness. For
different grades, the magnet wire size is still standardised. The
dielectric breakdown voltage of the magnet wire is dependent on
the insulation grade: the higher the grade, the higher the breakdown
voltage. However, the insulation material itself also has an impact
on the breakdown voltage [34].
It should be noted that cost is an important factor in selecting
the proper magnet wire. Magnet wires have the same dimensions
for the same AWG value for different thermal classes, but the cost
generally increases as the thermal class increases.
Fig. 2 Demagnetisation and characteristic curves of a PM (TDI In addition to the conductors on the stator, squirrel-cage IMs
Neorec53B iron-based rare-earth magnet) [36] have rotor conductors, which are conventionally manufactured by
die-casting aluminium in the rotor slots. At room temperature,
are used and they are formed, so that the turn-to-turn voltage does copper has 70% higher conductivity than that of aluminium.
not exceed the dielectric breakdown of the insulation system [29]. Therefore, aluminium die-casted IMs would naturally end up with
It should be noted that polyester (amide) (imide) in Table 1 means higher rotor copper losses. For the same power density, this might
that the insulating thermoplastic has all three functional groups of require a more aggressive thermal management system. However,
ester, amide, and imide in one polymer. besides the electrical conductivity, copper has higher coefficient of
For most of the magnet wire thermal classes, different thermal expansion (CTE) when compared to aluminium. The
insulation materials are available. The choice of the correct difference between the CTE of copper and steel would apply
insulation material is dependent on many factors, such as oil fatigue stress at the copper–steel interface [35]. Thermal
resistance, flexibility, resistance to common solvents, solderability, management system should be properly designed in a copper-cage
thermal stability, heat shock capability, windability, resistance to IM to avoid cracks on the conducting bars throughout the thermal
moisture, salt water resistance and so on. For example, cycling of the machine. It should be also noted that die-casting of
polyurethane is not used for high temperature and high current copper is a challenging process due to the higher melting
applications. Polyester (amide) (imide) underlying coating with temperature and density of copper [35].
polyamideimide superimposed coating (e.g. MW35-C) provides
high resistance to chemicals and moisture [32]. Glass fibres can 3.2 Permanent magnet
provide flexibility and they are available in larger diameters [33].
These characteristics help in defining the proper magnet wire for The magnetic properties of PMs are highly dependent on the
the given requirements, such as oil-cooling, high winding stresses, operating temperature. In electric machines, maximum operating
temperature should be defined to maintain stable operation of the
magnet.
Table 1 Thermal classes and insulating materials of round Iron based rare-earth magnets (NdFeB) are widely used in
copper magnet wires according to ANSI/NEMA traction machines due to their high-energy product and high
MW1000-2015 [30] coercivity. However, the performance of NdFeB magnet is very
Magnet Standard Insulating material sensitive to temperature. Normal and intrinsic curves in Fig. 2
wire Underlying Superimposed show that the flux density and coercivity of NdFeB magnet reduces
thermal coating coating as the temperature increases. This means that, the output torque
class and the capability of tolerating the negative field from the stator
105 MW15-C polyvinyl acetal-phenolic windings are highly dependent on the magnet temperature [3].
130 MW28-C polyurethane polyamide From Fig. 2, it can be observed that at high temperatures, the
155 MW79-C polyurethane
intrinsic coercivity, Hci of the magnet decreases. Depending on the
MW80-C polyurethane polyamide
external field and the permeance coefficient, if the load line crosses
the magnetisation curve below the knee point, magnet loses some
MW41-C glass fibre covered
of its magnetisation [37]. This is an irreversible loss and it is
180 MW76-C polyester (amide) polyamide caused by the reversal of domains, which were initially aligned in
(imide) the direction of magnetisation. Irreversible losses are dependent on
MW77-C polyester (imide) the operating conditions, temperature and how long magnet is
MW78-C polyester (imide) polyamide exposed to that temperature. Irreversible losses can be only
MW82-C polyurethane recovered by re-magnetising the magnet. In PM machines, the
MW83-C polyurethane polyamide temperature, loading conditions and irreversible-loss data provided
by the magnet manufacturer should be carefully evaluated to
MW50-C glass fibre covered
design proper thermal management system [38].
200 MW74-C polyamide (amide) (imide) Cobalt-based rare-earth magnets (SmCo) are also used in
MW35-C polyester (amide) polyamideimide electric machine applications. SmCo has higher Curie temperature
(imide) as compared to NdFeB; therefore it can operate at temperatures up
MW44-C glass fibre covered to 250 °C. However, SmCo has lower coercivity and energy
220 MW61-C aromatic polyamide product when compared to NdFeB, and is more expensive.
MW37-C polyester (amide) polyamideimide Aluminium–nickel–cobalt (AlNiCo) magnets can also operate at
(imide) high temperatures, but they have lower energy product when
MW81-C polyamideimide compared to NdFeB and SmCo. Unlike other magnets, ferrite
magnets have positive temperature coefficient (PTC) for the
240 MW16-C aromatic polyimide
coercivity [38]. The coercivity of ferrite magnets increases as the

IET Electr. Syst. Transp. 3

© The Institution of Engineering and Technology 2016
 of the insulation might be affected leading to higher eddy current
losses.

3.4 Insulation materials

In electric machines, slot liner and slot wedges are used to prevent
possible winding-to-ground short circuit faults. These insulation
materials are usually made of aramid or Mylar layered paper.
Nomex is an insulation paper supplied by DuPont for slot
insulation. It has higher mechanical strength, better flexibility,
higher chemical and moisture resistance, and high dielectric
breakdown strength at high temperatures [43]. Nomex paper is also
utilised in magnet wire film insulation, for thermal class 220. The
thickness of the slot insulation material ranges between 0.1 and
0.65 mm.
Encapsulation of the stator via vacuum pressure impregnation
process helps to improve the transmission of heat from the copper
Fig. 3 Loss fraction in typical four-pole IMs at 50Hz to the stator core by replacing the air pockets with varnish or
thermally conductive epoxy [39]. It also provides higher dielectric
temperature increases. Therefore, both low and high temperature and mechanical strength, chemical and heat resistance [29].
performances should be evaluated carefully when designing a
thermal management system for electric machines with ferrite
3.5 Coolant
magnets.
It should be also noted that CTE of PMs is twice as high as that The heat transfer coefficient of the thermal management system
of silicon steel. Without proper temperature control, the shear depends on many parameters, including the geometry, flow rate,
stress on the magnets can cause deterioration on the machine and also the type of the coolant. For the same operating conditions,
performance and potentially cracks on the magnet [39]. such as flow rate and temperature, air cooling has a much lower
heat transfer rate when compared to liquid cooling. However, it is
3.3 Silicon steel the preferred choice in small to medium power applications. Very
high heat transfer coefficients can still be obtained with advanced
In a thermal management system design, the thermal class of the air cooling techniques, such as air jet impingement cooling [39].
coils and the maximum allowable temperature of the PMs Thermophysical properties of air change with temperature and
generally define the maximum operating temperature of the pressure. The thermal conductivity and specific heat of air increase
machine. The magnetic properties of electrical steels usually with temperature. In addition, the density of air decreases with
change drastically near the Curie temperature (770 °C for iron). At increasing temperature. Air density is an important parameter in
any magnetic field intensity, the magnetic flux density of the steel defining the mass flow rate and pressure loss of the cooling system
reduces as the temperature approaches to Curie point. In [40], it [44].
was shown that, in case of non-oriented steel, the permeability of Liquid cooling has been acknowledged as a more effective
electrical steel did not change much under about 500 °C. This method to achieve higher heat transfer coefficients. When selecting
temperature is much higher than the maximum temperature PMs the coolant for the liquid cooling system, thermophysical
and magnet wires usually handled. properties, dielectric strength, and chemical compatibility of the
It is also important to note that the surface insulation of coolant play important role. Table 2 shows the properties of some
electrical steels should handle the annealing temperatures during of the liquid coolants used in electric machine applications. With
the manufacturing process. Electrical steels have surface insulation the highest specific heat, water can transfer the highest amount of
to increase the resistivity between laminations and, hence, reduce heat with the same flow rate. It also has low viscosity, which
the eddy currents. Annealing is a heat treatment process to maintains lower fluid flow resistance [39]. However, water cannot
eliminate the stress due to the plastic deformation on the be used for direct cooling, since it is electrically conductive and
laminations and helps to return the magnetic properties of the steel may cause corrosion over time.
back to the stress-free conditions [41]. American Society for
Testing and Materials (ASTM) standardises the classification of
insulating coatings for electrical steels [42]. The insulation 4 Loss generation in electric machines
material, annealing temperature, duration, and pressure during the The losses in electric machines are generated by different physical
annealing process should be carefully defined not to deteriorate the mechanisms. They are typically classified as copper, iron core,
surface insulation of the electrical steel. Otherwise, the resistivity magnet, mechanical, and stray losses. The copper loss results from
Joule heating due to the resistivity of the conductors. The iron
Table 2 Properties of liquid coolants for electric machine losses are caused by the time-varying magnetic field that yields
applications [39] hysteresis and eddy current losses. To fully capture the
Coolant type Specific Density, Therm. Dynamic electromagnetic losses, additional loss mechanisms need to be
heat, K/kg kg/m3 at 1 cond., viscosity (Pa taken into account, viz. the skin effect, proximity effect, magnet
K atm W/m K s) × 10−3 at eddy current, excess core loss, and stray loss. Electric machines
20 °C feature also mechanical losses, in particular friction and windage
losses. Fig. 3 shows the typical losses of different rated-power
water 4181 997 0.613 1.002
four-pole IMs at 50 Hz [45]. The loss mechanisms of electrical
methanol/water 3560 935 0.4 2.0 machines are discussed in the next subsections and computational
(40/60 by methods to estimate losses are introduced.
weight)
ethanol/water 3500 927 0.38 3.0
4.1 Copper loss
(44/56 by
weight) Copper loss is a major loss component in all electric machines,
ethylene glycol/ 3350 1069 0.39 3.4 especially at high torque operation. Most machines feature copper
water (50/50 by losses in both the stator and rotor, except for PM and reluctance
volume) machines. In addition, frequency dependent skin and proximity
engine oil 1765 887 0.138 81.3 effects add to the overall copper losses. The skin effect is caused
by eddy currents induced by a conductor's own flux linkage. The

4 IET Electr. Syst. Transp.

© The Institution of Engineering and Technology 2016
 components are either found analytically [46, 47] or estimated by
using finite element (FE) software. The magnetic field strength is
low in the overhang region as compared to the active region and,
therefore, the proximity effect in the overhang region is typically
negligible [46]. The skin effect is approximated as constant
throughout overall length of the conductor [46]. The impact of the
frequency on the phase resistance of the active and overhang
regions is shown in Fig. 4. It should be noted that the phase
resistance is also marginally influenced by excitation current
especially when the core is saturated [46].
The copper loss increases as the resistance increases with the
temperature. The temperature at overhang conductors is higher
than the active conductors. The conductors in the slots are
surrounded by electrical steel, which has higher thermal
conductivity [48]. This provides better heat dissipation in the active
region of the coils. Therefore, the heat source by the copper loss is
Fig. 4 Frequency impact on the phase resistance in an example PM modelled separately for active and overhang regions. The effect of
machine temperature in copper loss model is determined by updating
resistivity as a function of temperature [48]

� T = � � 1 + � � − �� , (2)

where ρo is the resistivity at initial temperature To, α is the

temperature coefficient, and T is the final temperature. The updated
resistivity is used to compute the resistance components in (1).

4.2 Iron losses

The iron loss is the second major loss component in electric
machines. It is typically dominant when a machine operates at
higher frequency, i.e. at high speed. Hysteresis and eddy current
losses are the main iron loss components. The hysteresis loss is
caused by a time-varying magnetic field primarily in major loops.
Eddy current loss is caused by the current induced in the core
(typically laminations). The difference between classical eddy
current and total eddy current losses is defined as excess loss [49].
The harmonic content due to slotting, winding distribution, non-
sinusoidal and pulse-width modulation (PWM) excitations cause
Fig. 5 Manufacture's loss data for 50CS350 China steel core losses at higher order frequencies [50, 51].
The Steinmentz equation is primarily used to quantify the core
loss. Its coefficients are generally treated as constants. However,
variable coefficients are required for predicting core loss for a wide
operating region. The coefficients vary with flux density and
frequency. Loss data from a core manufacturer for different
frequency and flux density is shown in Fig. 5. This data is used to
actually find the coefficients. The variation of hysteresis and eddy
current coefficients with respect to flux density is shown in Fig. 6.
Modified equations are used to account for PWM effects and
arbitrary waveforms [52]. The loci of the flux rotation are used to
account for rotational loss [53]. The Steinmentz equation that takes
harmonic components into account is [54]

�core = ∑� �1.6
h�
2 2
� �� + �e��� �� , (3)
�
Fig. 6 Hysteresis and eddy current loss coefficients
where n is the harmonic index, Kh and Ke are the hysteresis and
proximity effect is caused by the flux from neighbouring eddy current loss coefficients, respectively, B is the flux density,
conductors [46]. The skin and proximity effects cause a non- and f is the frequency.
uniform current distribution in conductors, which becomes more The iron loss is distributed throughout the core depending on
extreme as frequency increases. In practice, the copper loss is the amplitude and frequency of local flux density waveform. In AC
calculated using the winding resistance parameter. At low machines, the main iron losses happen on the stator where the
frequency, the copper losses are dominated by the Joule effect that stator teeth and back iron are the major core loss regions. The loss
is modelled with the DC resistance. Frequency dependent effects in tooth tips, or pole shoes, is relatively low and it can be neglected
are captured introducing equivalent AC resistances [47]: while considering thermal design [48]. In comparison, the rotor
core loss is typically less since the frequency of the magnetic flux
�cu = � �2p �dc + �skin � + �proximity � (1) density variation in the rotor core is lower compared to that of the
stator. It should be noted that air gap and current harmonics yield
where m is the number of phases, Ip is the rms phase current, f is alternating rotor flux harmonics causing additional iron losses. For
the frequency, Rdc, �skin � and �proximity � are the DC, skin, and example, in wound field synchronous machines (SMs), the rotor
lamination tends to be made from thick steel to reduce
proximity components of phase resistance, respectively.
manufacturing costs. In this case, the specific rotor core loss can be
The DC component of phase resistance is found through the
high in proximity of the air gap [55]. However, these effects can be
resistance equation based on the conductor geometry. The AC
ignored in some other machines. For example, the large effective

IET Electr. Syst. Transp. 5

© The Institution of Engineering and Technology 2016
 �m�2�
�magnet =
24 �m ∑� � � ,
2
�
2 2
(4)
�

where Vm is the volume of the magnet, Wm is the width, B is the

amplitude of flux density for a particular harmonic order, w is the
frequency, ρm is the resistivity of the magnet, and n is the harmonic
index. The value of the flux density is either found analytically or
by static FE analysis (FEA).
Equation (4) is valid for smaller magnets and low frequency
operating machines in which skin effect is negligible. There are
many methods reported in the literature that consider the eddy
current reaction field in analytical models [56, 57]. The
commercial FE software packages also have advanced magnet loss
Fig. 7 Rated losses in an IPM machine calculation methods. However, magnet segmentation adds
complexity to the FEA. Also, the inclusion of fringing-effects gives
better magnet loss calculation results, and thus, 3D models are
typically preferable [57].

4.4 Mechanical losses

Mechanical loss in electric machines consists of friction and
windage losses. Friction losses are mainly caused by the bearings
and are thermally relevant. The resulting heat and the heat passing
to the ambient via bearing increase the local temperature. This
tends to degrade the lubricant and reduce the life of the bearing.
The aerodynamic drag experienced by rotor periphery and cooling
fan causes windage loss. These losses can be minimised using high
quality bearings, lubricants and high-performance fan designs. A
general expression for windage and friction losses is [58]

�f� = 2 �3� �310−6 + �f�� � 10−3 (5)

where D, L and G are outer diameter, length and weight of rotor,

respectively, n is the rotational speed, and Kfb is the frictional loss
coefficient.
Fig. 8 Cooling technologies depending on cooling target

air gap in surface PM machines shields the rotor back iron from 5 Cooling techniques for electric machines
flux harmonics [52]. Cooling technologies for electric machines can be classified
A simple approach to include the core loss in thermal analysis is according to the mode of heat transfer: conduction, natural
to consider two heat sources, one is in the stator and other is in the convection, forced convection, radiation, and evaporative cooling;
rotor [48]. A detailed approach is to consider many heat sources or according to cooling fluid: water, air, oil, and phase change
corresponding to regions classified based on the same core loss materials; or according to the parts being targeted for cooling:
density [54]. The values of core losses can be found from stator core, stator windings, end windings and so on, as shown in
analytical equations. Many of the commercial software also Fig. 8. Some of these technologies can be incorporated
provide core loss calculation as part of their post processing. simultaneously to minimise the temperature of the various critical
locations, i.e. hot spots. This section discusses various commonly-
4.3 Magnet losses used cooling technologies for electric machines.
Dissipating the heat generated within an electric machine to an
Magnet loss, if applicable, is usually the lowest when compared to external heat sink depends on various factors: the mode of heat
other losses in most electric machines. Fig. 7 shows the different transfer, the effective heat transfer area and geometry, the working
losses at the rated condition of an example IPM machine. Despite fluid used for cooling, the flow rate and temperature of the cooling
the low amount, the magnet loss is very critical in thermal aspect. media and so on. The simplest cooling technique is dissipating the
The local magnet temperature can increase significantly as the heat heat to the ambient by natural convection. Typically, the heat
generated by magnets takes a long path to reach the cooling sink dissipation can be improved by increasing the heat transfer surface
causing a large equivalent thermal resistance. Magnet over- area with added fins on the housing. More complicated forced air-
temperature can cause irreversible partial or full demagnetisation. cooling system can be used to further increase the heat dissipation.
In addition, surface mounted magnets are more prone to harmonics For example, a shaft mounted fan can be employed to enhance the
than embedded magnets [52]. Typically, magnet segmentation is heat transfer from the housing fins, the end windings, and rotor
used to reduce the magnet loss. However, too many layers of surfaces. However, for high current densities, using air as the
segmentation may adversely affect the heat conduction and cooling fluid may not be sufficient and some form of liquid cooling
increase cost. may be required for better cooling efficiency. Typical rules of
Eddy currents that are induced in the magnets due to rotor flux thumb for cooling techniques and associated heat transfer
variation cause the magnet loss. At no-load, only the air gap coefficients are listed in Table 3 [4, 59]. It is apparent that higher
reluctance variation causes rotor flux harmonics. At loaded cooling efficiency enables higher current density and hence higher
condition, flux harmonics due to non-sinusoidal excitation, machine output power, however, at the expense of higher system
distribution, and PWM have to be taken into account. The loss complexity and energy cost.
generated by slot harmonics is typically the major contributor to In terms of the cooling location, stator core cooling is the most
the magnet losses [31]. Assuming that the flux density across the commonly used technology in electric machines. It can be divided
magnet is uniform and neglecting skin effect due to the reaction into direct cooling where the heat is removed directly from the
field, the magnet loss can be calculated according to [56] stator core [60, 61], or indirect cooling where the heat is transferred
radially and/or axially to the external surfaces and dissipated by the

6 IET Electr. Syst. Transp.

© The Institution of Engineering and Technology 2016
Fig. 9 Cooling technologies depending on cooling locations

Fig. 10 Various water jacket cooling configurations and paths

(a) Helical channels, (b) Circumferential channels, (c) Meander axial channgels, (d) Axial Serpentine channels

cooling fluid [62, 63], or it can be the combination of the both [64, requirement for an auxiliary system to provide coolant, risk of
65]. Fig. 9 illustrates the typical cooling technologies depending on corrosion inside the water circuit, risk of leaks, and more
cooling locations in electric machines. precautions with maintenance.
Water jacket cooling is a commonly-used indirect cooling Direct stator core cooling applies where water jacket cooling is
technique. It enables an effective heat transfer from the active part not sufficient and a significant temperature gradient might exist
of stator to the coolant. The advantages of using water jacket between the outer stator surface and the inner stator core. Direct
include higher power-to-frame size ratio, lower noise level, higher cooling channels can be placed at the stator outer diameter adjacent
efficiency, and completely enclosed environment. Also, the to the housing to minimise the electro-magnetic impact on machine
removed heat is not directly dissipated into the environment [66, performance, or inside stator back iron to form direct contact
67]. Different configurations of water jacket channels and frame between the coolant and the hot stator core. For the same pressure
structures have been exploited including helical ducts, drop in the channels, the machine with direct stator core cooling
circumferential channels, meander shape of the ducts, and axial channels has significantly lower average temperature in the stator
serpentine channels, as shown in Fig. 10. The configuration and the back iron compared to the ones with indirect cooling technologies
number of cooling paths determine the cooling efficiency as well as [68].
the pressure drop from the inlet to the outlet. Disadvantages of Placing the cooling channels in the stator slots with direct
using water jacket for cooling include higher manufacturing cost, contact to the windings can also achieve good cooling effect [69,

Table 3 Rules of thumb for cooling type and heat transfer coefficients [4]
Current density range, Heat transfer coefficient, W/ Cooling efficiency System complexity Energy cost
A/mm2 (m2·K)
natural convection 1.5–5 5–10 low simple none
forced convection 5–10 10–300 medium medium low
liquid cool 10–30 50–20,000 high complex high

IET Electr. Syst. Transp. 7

© The Institution of Engineering and Technology 2016
Fig. 11 Example LPTN for an electric machine [18]

70]. Furthermore, filling materials with high thermal conductivity, thermal analysis include lumped parameter thermal network
but low electromagnetic conductivity such as resin, epoxy, (LPTN), FEA, and computational fluid dynamics (CFD).
polymers, and thermoplastic can be applied to fill the winding gaps
and allow better heat dissipation [71–73]. Direct winding heat 6.1 LPTN analysis
exchanger utilising micro-feature enhanced meso-channels can be
placed between two neighbouring winding coils in one slot to The LPTN analysis offers a quick method of determining the
axially dissipate the heat generated from the windings, while the temperature distribution within electric machines where the user
meso-channels allow for low pressure drop and high heat transfer can rapidly determine the changes resulting from different input
coefficients [74]. parameters. Many publications have proven its application in
Another critical location that sometimes requires additional various types of machine topologies including PM, induction, and
cooling efforts in electric machines is the end winding. End SRMs. Commercially available software is capable of performing a
winding cooling techniques include spray cooling, liquid jets and thermal analysis using a LPTN approach.
using thermally conductive material between the end windings and The LPTN models usually consist of different components
the frame. Due to the superior heat transfer characteristics of oil being lumped into simplified areas to represent more complex
compared to water, oil may also be used as a cooling medium for geometry [18, 82, 83]. The components can be lumped in different
spray cooling and liquid jets [75, 76]. partial models such as radially in a 2D cross section of the
Evaporative cooling is another effective cooling technique that machine, or a 3D layout taking into account the axial direction
uses two phase flow cooling systems to enhance heat dissipation of effects of heat distribution. Kim et al. [82] proposed a 2D LPTN
electric machine, which has been successfully applied in large which has three parallel paths for the heat to flow to the air. 2D
electrical machines. This method has the perfect cooling capability LPTNs are used by El-Refaie et al. [84] and Lindström [83] where
and high reliability, but has the inconvenience of requiring very heat transfer is possible between the interior components, including
high quantity of coolant which makes it impractical for small size interaction with the motor frame, and leading to the coolant
machines [77]. channel. A similar network is given by Mellor et al. [85], but
Other special cooling techniques have been developed. without forced cooling the heat leads from the frame to the ambient
Piezoelectric fans are implemented to replace liquid cooling temperature. Alternatively, Nategh et al. [86] and Sali et al. [87]
systems to improve fault-tolerance and reliability [78]. used 3D networks which allow heat transfer axially through the
Temperature sensitive ferrofluids are used as cooling and dielectric motor components instead of just through lumped cross sections.
medium for end winding to avoid extra complexity and costs in The 3D networks contain many additional nodes to encompass the
liquid cooling systems including deionised water, pumping system, different axial positions of components, as well as the front and
filters, instrumentation and so on [79]. Totally immersed electrical back ends of the motor. One 3D LPTN of an electric motor from
machines are investigated such that the rotor and stator surfaces are Boglietti et al. [18] is displayed in Fig. 11. The nodes located
directly flushed by hydraulic oil to increase the heat dissipation within each machine component represent the temperature of a
[80]. New machine geometry designs using stator flux-barriers for lumped area. These temperature nodes are then connected with
cooling are proposed to enhance the cooling area and, therefore, thermal resistances due to conduction, convection, or radiation.
increase cooling efficiency [81]. Thermal capacitance values are used to connect temperature nodes
and ambient heat in the cases of transient thermal storage
simulations [83]. Heat generation losses are found through
6 Thermal analysis approaches analytical calculations, electromagnetic FEA studies, or
A key branch in the thermal management of electric machines is experimental testing, and are then input into these nodal locations
the ability to predict the temperature distribution during the design [18].
stage using thermal analysis approaches. Common methods for

8 IET Electr. Syst. Transp.

© The Institution of Engineering and Technology 2016
Fig. 12 Examples of electric machine configurations
(a) Brushless PM machine, (b) Brushed PM machine, (c) IM, (d) SRM, (e) Salient-pole SM, (f) Non-salient-pole SM

A LPTN is similar to that of an electrical circuit network, where 6.1.1 Synchronous machine: This type of machine features a
voltage represents temperature, current represents heat flow, and distributed armature winding that is located on the stator and
electrical resistance and capacitance represent thermal resistance distributed or concentrated field windings on the rotor. A DC
and capacitance, respectively [18]. Heat transfer coefficients, i.e. h- current is driven in the field coils to generate DC magnetic field in
values, can be used to calculate the thermal resistances due to the rotor. The presence of slotting, saliency, and the armature
convection. Empirical formulas to calculate heat transfer reaction (with current harmonics) add higher order flux harmonics
coefficients have been developed for different areas of the motor that can result in high iron loss densities in proximity of the air gap
such as across the air gap and heat transfer through the internal end [55]. The stator current produces copper losses that are similar to
space air [83, 88]. The h-values change based on the pattern, shape, other AC machines. A DC/DC converter typically generates the
rotational speed, and orientation of the applied cooling method and rotor DC current using rotor inductance to smooth the current [89].
the surface area, which is in contact with. The resulting harmonics produce copper losses at the resulting
The LPTN model can then predict the steady-state or transient frequencies. Some synchronous motors and grid-connected
heat flow within the electric machine, and the temperatures of each synchronous generators feature a third ‘damper’ winding that adds
component. The more detailed the network is, the more accurately line-starting capabilities and prevents oscillations in the case of
it can determine the temperatures of different areas within the varying loads. During transients, this winding behaves similar to an
machine. The temperatures predicted by the lumped parameter IM rotor winding and results in similar loss mechanisms. In steady-
networks have been found to closely represent the same state operation, current harmonics, e.g. due to an inverter, can
distribution as is found in FEA and CFD simulations as long as the cause significant copper losses in the damper winding [55]. The
machine is not overly complex. The results of changes in user thermal analysis of SM using LPTN is an ongoing research topic
inputs in the network are quickly found since the calculations are with recent results in [90–92].
analytical and usually much less complex than that of FEA and
CFD. 6.1.2 PM synchronous machine: In comparison to other types
A downside to the LPTN method of thermal analysis is that it of machines, the PM machine consists of a rotor which contains
involves substantial effort to create an accurate model [18]. The embedded or surface mounted PMs. Magnetic losses are generated
LPTN provides a tool that can calculate the heat transfer and in the magnets which are applied to the LPTN. A LPTN which
temperature distribution much quicker than CFD or FEA methods, represents a PM machine will incorporate the quantity and
but it can be very time consuming to create the thermal network dimensions of the magnets as well as its position within the rotor.
[82–84]. Further efforts are required through validation and The addition of the bonding or insulating material surrounding the
correction from either simulation or experimental measurement, magnets also changes the thermal resistance between magnets and
and it is challenging to modify the LPTN to be applicable to the rotor due to conduction. Two examples of PM machines can be
various types of machines. In addition, the heat transfer between viewed in Figs. 12a and b. It should be noted that the brushes and
nodes is calculated based on the thermal properties of the materials commutators are placed at the end of the machine and thus not
and fluids, and with the temperature affecting the material shown in Fig. 12b for the brushed PM DC machine. Many
properties, calibration is required to modify the thermal network to publications have demonstrated performing a thermal analysis of a
apply it to different operating conditions. PM machine using a LPTN [83, 93].
The geometry dimensions such as the number of poles, stator
teeth, rotor teeth, and magnets, if applicable, will all affect the 6.1.3 Induction machine: An IM incorporates a rotor core which
thermal circuit and resistances between connected nodes in the either has squirrel-cage rotor bars or slots that contain coil
LPTN. The LPTN models for different machines vary depending windings, similar to those in the stator core. In the case of a
on different stator and rotor configurations, which have key telltale squirrel-cage type rotor, the aluminium or copper bars are
difference and lead to different network path when building LPTN connected by end rings which short circuit the connections and
models. generate copper losses as current is induced by the rotating stator

IET Electr. Syst. Transp. 9

© The Institution of Engineering and Technology 2016
 Heyns and Wang [58] apply an FEA partial model which observes
1/6 of the motor geometry. Zhang et al. [97] use a 3D geometry
FEA electromagnetic and thermal linked model focusing
specifically on the stator core to determine the temperature
distribution. Marignetti et al. [98] use a 3D full model FEA
coupled electromagnetic and thermal model to find the heat
distribution through interior components including the windings,
shaft, magnets, core and so on. Wu et al. [99] design a 1/8 partial
model electromagnetic and thermal coupled FEA simulation for a
switched reluctance motor, modelling the stator, windings, rotor,
shaft, and magnets. Heat transfer coefficient values and heat
resistances are required as inputs to calculate the convective heat
transfer between surfaces and air or fluids. Losses and heat transfer
boundary values are also inputs and are determined through
analytical calculations, numerical simulations, or experimental
testing [58]. Material properties of components and the geometrical
dimensions determine the thermal resistance due to conduction
within the machine. The more detailed the mesh that is applied to
the model, the more accurate the temperature prediction will be,
however the computational time will also increase with complexity.
FEA thermal analysis can also be used to validate a LPTN model
under development [58, 86]. An example of a partial model FEA
Fig. 13 Example of a partial model FEA thermal analysis thermal analysis for comparison with a LPTN model is displayed
in Fig. 13.
magnetic field. Meanwhile, eddy current and hysteresis losses are
induced in the electrical steel laminations that are axially aligned 6.2.2 CFD analysis: CFD is usually the most accurate
through the rotor bars. Similarly with the wound-rotor, the coils in temperature prediction method when compared with the LPTN and
the rotor core generate secondary copper losses as current is FEA methods. It can be used to determine heat transfer boundaries
induced. A LPTN designed for an IM will incorporate the addition for inputs into LPTN and FEA methods, and has the ability to
of either the copper/aluminium bar or the coil material. Insulation perform simulations to predict the fluid flow characteristics and
around the coils and sleeves inserted in the rotor slots would also optimise cooling methodologies (e.g. water jacket, air cooled etc.)
be incorporated and changes would be made to the thermal [100].
resistances due to conduction through the different material Conventional CFD simulations only solve for the fluid flow and
components. An example of an IM machine can be viewed in do not cover the solid domains. The losses values evaluated from
Fig. 12c. Many publications have demonstrated performing a electromagnetic calculations are incorporated into the CFD
thermal analysis of an IM using a LPTN [18, 88, 94]. simulations as surface heat flux or constant surface temperatures
boundary conditions. Several CFD thermal analyses with different
6.1.4 Switched reluctance machine: Unlike the previous levels of complexity were implemented to simulate the air flow
machine types, the rotor core of an SRM only consists of electrical around different components of the machine such as in the stator–
steel material and does not contain any magnet or coil winding rotor air gap [101, 102], end windings [103], or in cooling passages
components. This allows for a more simplified LPTN as the radial [104, 105]. Huang et al. [106] used steady-state CFD simulations
thermal resistance due to conduction only passes through the rotor characterise the heat transfer performance of three different shaped
core material of certain dimensions. An example of an SRM cooling channels for forced cooling in traction motors of hybrid
machine can be viewed in Fig. 12d. Many publications have electric vehicles and zero emission vehicles. The cooling channels
demonstrated performing a thermal analysis of an SRM using a were embedded in the stator core with oil as cooling fluid. The
LPTN [95, 96]. authors reported that the dimensional parameters developed can
easily be applied to 2D or 3D FEA for thermal analysis or lumped
6.2 Numerical modelling methods parameter model analysis for the complete motor.
The simulation results can be improved by applying conjugate
Numerical modelling methods exist which are also proven tools for heat transfer method [105, 107], where the solid domains have to
predicting the temperature within electric machines when be modelled in addition to the fluid domains and the heat losses
compared with experimental results. The two most common can be defined in the solid domains as volumetric sources obtained
examples of numerical modelling methods are FEA and CFD. by electromagnetic simulations. Due to the advancement of the
While these methods can be more accurate in predicting computational power in the recent years, using CFD conjugate
temperatures within electric machines with a low comparative thermal analysis for electric machines became a focus of several
error, they generally are more time consuming due to higher researchers [107, 108]. Previous research using conjugate CFD
complexity and more detailed input components. Each thermal thermal analysis focused on air cooled systems [109–111], water or
analysis type has its associated advantages and weaknesses. oil cooled systems [80, 112, 113] and hybrid systems [114]. In the
recent study performed by Schrittwieser et al. [111], the authors
6.2.1 FE analysis: FEA involves dividing the components within compared between the simulation results based on a conjugate heat
a model into many tiny meshed nodes and elements, to analyse the transfer model and those obtained by conventional heat transfer
changes across those dimensions. Many examples of commercially model (no conduction). The results showed that the temperature
available software are capable of performing 2D or 3D thermal deviations between the two models are relatively small and that the
FEA in both steady-state and transient simulations. Results found conventional heat transfer model are generally applicable for
through thermal FEA are similar to that of LPTN models since calculating stator temperatures with the benefit of shorter
common inputs are applied; however, the processing time is a calculation times.
downside of FEA. FEA thermal analysis has an advantage when CFD thermal analysis has the advantage in scenarios including
the geometry becomes too complex to model using a LPTN [18]. when focusing on simultaneous air flows and mass transfer with a
2D simulations can be used to represent the temperatures within mix of laminar and turbulent states within a machine, and when
a machine when there is assumed to be little temperature difference designing channels and patterns for coolant flow [100]. Similar to
across the axial components. FEA models are usually simplified to how materials are input into geometry components in FEA, fluid
be partial fractional models, representative of the full geometry by properties and specifications are input in CFD, including volume
applying periodic patterns, in order to speed up processing time. flow rate, inlet temperatures, and pressure [18, 100]. Temperatures

10 IET Electr. Syst. Transp.

© The Institution of Engineering and Technology 2016
of fluids can be resultant of change over time as opposed to being interested in thermal management and thermal researchers who are
approximated during calculations for the initial conditions. interested in machine technologies.
However, a drawback to CFD thermal analysis is that simulation
takes much longer to process depending on complexity of the 8 Acknowledgments
geometry, degree of meshing, and the computational processing
power available. This limits the user as they are not able to quickly This research was undertaken, in part, thanks to funding from the
observe the result of changes to input parameters in the design on Canada Excellence Research Chairs program.
the thermal distribution within the machine.
9 References
6.3 Experimental validation [1] Boldea, I., Tutelea, L.N.: ‘Electric machines: steady state, transients, and
design with MATLAB’ (CRC Press, Boca Raton, FL, 2009)
Both lumped parameter thermal network analysis and numerical [2] Chapman, S.J.: ‘Electric machinery fundamentals’ (McGraw-Hill, New York,
modelling methods provide profound insights and guidance in NY, 2005, 4th edn.)
temperature prediction and thermal management of electric [3] Bilgin, B., Sathyan, A.: ‘Fundamentals of electric machines’, in Advanced
machines. However, certain degrees of error or inaccuracy always electric drive vehicles’ in Emadi, A. (Ed.), (CRC Press, Boca Raton, FL,
2014), pp. 107–186
exist due to model limitation, parameter deviation, various [4] Staton, D.: ‘Thermal Analysis of Traction Motors’. Transportation
assumptions, simplified boundary conditions and so on. Thus, it is Electrification Conf. and Expo (ITEC), Dearborn, MI, 2014
often necessary to calibrate and verify the thermal analysis through [5] Yang, Y., Arshad-Ali, K., Roeleveld, J., et al.: ‘State-of-the-art electrified
experimental tests. In principle, two types of experiments are powertrains: hybrid, plug-in hybrid, and electric vehicles’, Int. J. Powertrains,
2016, 5, (1), pp. 1–28
conducted. (i) Experiments that are designed to determine machine [6] Hou, R., Yang, Y., Emadi, A.: ‘Hybrid electric locomotive powertrains’. IEEE
thermal parameters including thermal capacitance, thermal Transportation Electrification Asia-Pacific (ITEC Asia-Pacific), Beijing, 2014
resistance, thermal conductivity, convection heat transfer [7] Yang, Y., Emadi, A.: ‘Integrated electro-mechanical transmission systems in
coefficients, radiation heat transfer coefficients and so on. These hybrid electric vehicles’. 2011 IEEE Vehicle Power and Propulsion Conf.,
Chicago, IL, September 2011
can either be performed on complete machines to draw conclusions [8] Yang, Y., Schofield, N., Emadi, A.: ‘Integrated electro-mechanical double-
on similar machines and structures [115, 116], or on part of rotor compound hybrid transmissions for hybrid electric vehicles’, IEEE
machines such as segmented stator structures [117] and end Trans. Veh. Technol., 2016, 65, (6), pp. 4687–4699
windings [118]. (ii) Experiments that are conducted to verify the [9] Emadi, A., Ehsani, M., Miller, J.M.: ‘Vehicular electric power systems: land,
sea, air, and space vehicles’ (Marcel Dekker Inc., New York, NY, 2004)
analysis or simulation results, typically in the form of prototype [10] Bilgin, B., Magne, P., Malysz, P., et al.: ‘Making the case for electrified
machines. transportation’, IEEE Trans. Transp. Electrification, 2015, 1, (1), pp. 4–17
In terms of temperature monitoring, direct thermal sensor [11] Rogers, S.: ‘EV everywhere grand challenge, electric drive status and
measurements are widely applied on industrial applications and challenges’ (U. S. Department of Energy, Energy Efficiency & Renewable
Energy, 2012)
laboratory instrumentations at machine locations such as on the [12] Zhu, Z.Q., Howe, D.: ‘Electrical machines and drives for electric, hybrid, and
windings, end caps, housing, cooling jacket and so on [119, 120]. fuel cell vehicles’, Proc. IEEE, 2007, 95, (4), pp. 746–765
Typically used thermal sensors include thermocouples, thermistors, [13] Bilgin, B., Emadi, A.: ‘Electric motors in electrified transportation: a step
resistance thermometers and so on. Thermocouples are based on toward achieving a sustainable and highly efficient transportation system’,
IEEE Power Electron. Mag., 2014, 1, (2), pp. 10–17
the principle that the junction of two dissimilar metals generates a [14] Bennion, K.: ‘Electric Motor Thermal Management’. National Renewable
voltage which increases with temperature. They have the Energy Laboratory, NREL Report No. PR-5400-50510, 2011
advantages of high measurement range, fast response time, and [15] Liao, C.M., Chen, C.L., Katcher, T.: ‘Thermal analysis for design of high
simple configuration. However, they only have medium accuracy performance motors’. Int. Conf. on Thermal and Thermomechanical
Phenomena in Electronic Systems, Seattle, WA, 1998
and sensitivity, and are prone to electromagnetic interference due [16] Howey, D.A., Childs, P.R.N., Holmes, A.S.: ‘Air-gap convection in rotating
to low signal strength. Thermistors are made of metal oxides whose electrical machines’, IEEE Trans. Ind. Electron., 2012, 59, (3), pp. 1367–1375
resistances increase or decrease with increasing temperature, hence [17] Bennion, K., Cousineau, J.: ‘Sensitivity study of traction drive motor
also called positive temperature coefficient or negative temperature cooling’. IEEE Transportation Electrification Conf. and Expo (ITEC),
Dearborn, MI, 2012
coefficient sensors, respectively. Compared to thermocouples, [18] Boglietti, A., Cavagnino, A., Staton, D.A., et al.: ‘Evolution and modern
thermistors provide more accurate, sensitive, robust signal output, approaches for thermal analysis of electrical machines’, IEEE Trans. Ind.
and they have the advantages of simple construction, small size, Electron., 2009, 56, (3), pp. 871–882
low thermal mass, low cost, and being ineffective to [19] Moreno, G., Narumanchi, S., Bennion, K., et al.: ‘Gaining traction: thermal
management and reliability of automotive electric traction-drive systems’,
electromagnetic noise. However, due to their narrow linear range, IEEE Electrification Mag., 2014, 2, (2), pp. 42–49
thermistors have limited useful temperature span [121]. Resistance [20] Yang, R., Yang, Y., Bilgin, B., et al.: ‘Analysis of interior permanent magnet
thermometers, or resistance temperature detectors (RTDs), are machines with grain boundary diffusion processed magnet’. IET Int. Conf. on
similar to thermistors where they utilise resistances to measure Power Electronics, Machines, and Drives (PEMD 2016), Glasgow, Scotland,
2016
temperature. The resistances of RTDs increase with temperature [21] Toliyat, H.A., Kliman, G.B.: ‘Handbook of electric motors’ (CRC Press, Boca
hence they are PTC sensors. RTDs are the most accurate, sensitive, Raton, FL, 2004)
and stable temperature sensors for industrial applications, although [22] Bennion, K., Cousineau, J., Moreno, G.: ‘Electric motor thermal management
they are higher priced and require external power sources. In some for electric traction drives’. SAE Thermal Management Systems Symp.,
Denver, CO, 2014
cases, where direct sensor is difficult to access such as rotor and [23] Krishnamurthy, M., Edrington, C.S., Emadi, A., et al.: ‘Making the case for
magnets, indirect temperature measurement is implemented by applications of switched reluctance motor technology in automotive
using infrared thermography, wireless communication, or slip-ring products’, IEEE Trans. Power Electron., 2006, 21, (3), pp. 659–675
signal transmission [122–124]. Temperature estimation techniques [24] Senda, K., Namikawa, M., Hayakawa, Y.: ‘Electrical steels for advanced
automobiles – core materials for motors, generators and high-frequency
based on flux observer, signal injection, lumped-parameter thermal reactors’. JFE Steel Tech. Rep., Tokyo, Japan, 2004
networks have also been developed to ensure design redundancy in [25] Gerada, D., Mebarki, A., Brown, N.L., et al.: ‘High-speed electrical
case of sensor failures [125–127]. machines: technologies, trends, and developments’, IEEE Trans. Ind.
Electron., 2014, 61, (6), pp. 2946–2959
[26] Nategh, S., Krings, A., Huang, Z., et al.: ‘Evaluation of stator and rotor
7 Conclusion lamination materials for thermal management of a PMaSRM’. Int. Conf. on
Electrical Machines (ICEM), Marseille, 2012
Thermal management in electric machines plays a critical role in [27] Narumanchi, S.: ‘Thermal management of power electronics and electric
determining the machine performance and reliability. This paper motors for electric-drive vehicles’. IEEE Energy Conversion Congress and
comprehensively examines different aspects related to electric Exposition (ECCE), Pittsburgh, PA, 2014
[28] Emadi, A.: ‘Energy-efficient motors’ (CRC Press, Boca Raton, FL, 2004)
machine thermal management. Fundamental material thermal [29] Stone, G.C., Culbert, I., Boulter, W.A., et al.: ‘Rotating machine insulation
properties and loss generations have been discussed, diverse systems’ ‘Electrical Insulation for Rotating Machines’ (IEEE Press, Hoboken,
approaches for thermal analysis and cooling techniques have been New Jersey, USA, 2014)
presented, and various design considerations for thermal [30] ‘National Electrical Manufacturers Association Standard’. ANSI/NEMA MW
1000-2015, 2015
management have been provided. It is the purpose of this paper to
serve as a reference guideline for machine designers who are
IET Electr. Syst. Transp. 11
© The Institution of Engineering and Technology 2016
[31] Marquez, E., Nelson, D.: ‘Development of current squared – time curves for [65] Mizuno, S., Noda, S., Matsushita, M., et al.: ‘Development of a totally
simplified wire size selection for electric traction systems in automotive enclosed fan-cooled traction motor’, IEEE Trans. Ind. Appl., 2013, 49, (4),
applications’. IEEE Transportation Electrification Conf. and Expo, Dearborn, pp. 1508–1514
MI, June 2015 [66] Karim, Z.A.A., Yusoff, A.H.M.: ‘Cooling system for electric motor of an
[32] ‘Magnet Wire/Winding Wire Product Application and Packaging Data’. electric vehicle propulsion’, Adv. Mater. Res., 2014, 903, pp. 209–214
Superior Essex, 1999. Available at: http://www.superioressex.com/ [67] Borges, S.S., Cezario, C.A., Kunz, T.T.: ‘Design of water cooled electric
[33] Beeckman, R.: ‘NEMA magnet wire thermal class ratings’. Essex Group, Inc. motors using CFD and thermography techniques’. Int. Conf. on Electrical
Available at: http://www.superioressex.com/. [Accessed August 2015] Machines, Vilamoura, Portugal, 2008
[34] ‘Magnet wire insulation guide’. MWS Wire Industries. Available at: http:// [68] Huang, Z., Nategh, S., Lassila, V., et al.: ‘Direct oil cooling of traction motors
mwswire.com/, [Accessed 2016] in hybrid drives’. IEEE Int. Electric Vehicle Conf. (IEVC), Greenville, SC,
[35] Mechler, G.C.: ‘Manufacturing and cost analysis for aluminum and copper die 2012
cast induction motors for GM's powertrain and R&D divisions’. M.S. thesis, [69] Vlach, R., Grepl, R., Krejci, P.: ‘Control of stator winding slot cooling by
Department of Materials Science and Engineering, Massachusetts Institute of water using prediction of heating’. IEEE Int. Conf. Mechatronics (ICM),
Technology, Cambridge, MA, 2010 Kumamoto, 2007
[36] ‘NEOREC series neodymium iron boron magnet datasheet’. TDK [70] Galea, M., Gerada, C., Raminosoa, T., et al.: ‘A thermal improvement
Corporation, May 2011. Available at: http://tdk.co.jp/ technique for the phase windings of electrical machines’, IEEE Trans. Ind.
[37] Constantinides, S.: ‘The demand for rare earth materials in permanent Appl., 2012, 48, (1), pp. 79–87
magnets’. Arnold Magnetic Technologies. Available at: http:// [71] Yang, Y.: ‘Double-rotor switched reluctance machine for integrated electro-
arnoldmagnetics.com/, [Accessed 15 July 2015] mechanical transmission in hybrid electric vehicles’. PhD dissertation,
[38] Trout, S.R.: ‘Material selection of permanent magnets, considering the McMaster University, Hamilton, ON, Canada, February 2014
thermal properties correctly’. Proc. the Electric Manufacturing and Coil [72] Yang, Y., Schofield, N., Emadi, A.: ‘Double-rotor switched reluctance
Winding Conf., Cincinnati, OH, October 2001 machine (DRSRM)’, IEEE Trans. Energy Convers., 2015, 30, (2), pp. 671–
[39] Tong, W.: ‘Mechanical design of electric motors’ (CRC Press, Boca Raton, 680
FL, 2014) [73] Rhebergen, C., Bilgin, B., Emadi, A., et al.: ‘Enhancement of electric motor
[40] Takahashi, N., Morishita, M., Miyagi, D., et al.: ‘Examination of magnetic thermal management through axial cooling methods: a materials approach’.
properties of magnetic materials at high temperature using a ring specimen’, IEEE Energy Conversion Congress and Exposition (ECCE), Montreal,
IEEE Trans. Magn., 2010, 46, (2), pp. 548–551 Canada, 2015
[41] ‘Selection of Electrical Steels for Magnetic Cores’. AK Steel. Available at: [74] Semidey, S.A., Mayor, J.R.: ‘Experimentation of an electric machine
http://aksteel.com/, [Accessed 16 July 2015] technology demonstrator incorporating direct winding heat exchangers’, IEEE
[42] ‘Standard Classification of Insulating Coatings for Electrical Steels by Trans. Ind. Electron., 2014, 61, (10), pp. 5771–5778
Composition, Relative Insulating Ability and Application’. American Society [75] Boglietti, A., Cavagnino, A.: ‘Analysis of the endwinding cooling effects in
of Testing and Materials, ASTM A 976-13, 2015 TEFC induction motors’, IEEE Trans. Ind. Appl., 2007, 43, (5), pp. 1214–
[43] ‘DuPONT NOMEX Paper Type 410’, DuPONT. Available at: http:// 1222
www.dupont.com, [Accessed August 2015] [76] Staton, D.A., Popescu, M., Hawkins, D., et al.: ‘Influence of different end
[44] Chang, C.C., Cheng, C.H., Ke, M.T., et al.: ‘Experimental and numerical region cooling arrangements on end-winding heat transfer coefficients in
investigations of air cooling for a large-scale motor’, Int. J. Rotating Mach., electrical machines’. IEEE Energy Conversion Congress and Exposition,
2009, 2009, pp. 1–7 Atlanta, GA, 2010
[45] De Almeida, A.T., Ferreira, F.J.T.E.T.E., Fong, J.A.C.: ‘Standards for [77] Guechi, M.R., Desevaux, P., Baucour, P., et al.: ‘On the improvement of the
efficiency of electric motors’, IEEE Ind. Appl. Mag., 2011, 17, (1), pp. 12–19 thermal behavior of electric motors’. IEEE Energy Conversion Congress and
[46] Nalakath, S., Preindl, M., Bilgin, B., et al.: ‘Modeling and analysis of AC Exposition, Denver, CO, 2013
resistance of a permanent magnet machine for online estimation purposes’. [78] Gilson, G.M., Pickering, S.J., Hann, D.B., et al.: ‘Piezoelectric fan cooling: a
IEEE Energy Conversion Congress and Exposition, Montreal, QC, 2015 novel high reliability electric machine thermal management solution’, IEEE
[47] Sudhoff, S.: ‘AC conductor losses in power magnetic devices: a multi- Trans. Ind. Electron., 2013, 60, (11), pp. 4841–4851
objective design approach’ (Wiley-IEEE Press, 2014) [79] Karimi-Moghaddam, G., Gould, R.D., Bhattacharya, S., et al.:
[48] Demetriades, G.D., De La Parra, H.Z., Andersson, E., et al.: ‘A real-time ‘Thermomagnetic liquid cooling: A novel electric machine thermal
thermal model of a permanent-magnet synchronous motor’, IEEE Trans. management solution’. IEEE Energy Conversion Congress and Exposition
Power Elctron., 2010, 25, (2), pp. 463–474 (ECCE), Pittsburgh, PA, 2014
[49] Zhang, Y., Cheng, M.C., Pillay, P.: ‘Magnetic characteristics and excess eddy [80] Ponomarev, P., Polikarpova, M., Pyrhönen, J.: ‘Thermal modeling of directly-
current losses’. Industry Applications Society Annual Meeting, Houston, TX, oil-cooled permanent magnet synchronous machine’. Int. Conf. on Electrical
2009 Machines (ICEM), Marseille, 2012
[50] Nalakath, S., Preindl, M., Yang, Y., et al.: ‘Modeling and analysis of core [81] Nollau, A., Gerling, D.: ‘Novel cooling methods using flux-barriers’. Int.
losses of an IPM magnet machine for online estimation purposes’. Annual Conf. on Electrical Machines (ICEM), Berlin, 2014
Conf. of IEEE Industrial Electronics Society, Yokohama, Japan, 2015 [82] Kim, D.J., Jung, J.W., Kwon, S.O., et al.: ‘Thermal Analysis using Equivalent
[51] Mthombeni, T.L., Pillay, P.: ‘Physical basis for the variation of lamination Thermal Network in IPMSM’. 2008 Int. Conf. on Electric Machines and
core loss coefficients as a function of frequency and flux density’. Annual Systems, Wuhan, China, 2008
Conf. on IEEE Industrial Electronics, Paris, November 2006 [83] Lindström, J.: ‘Development of an experimental permanent-magnet motor
[52] Han, S., Jahns, T.M., Zhu, Z.Q.: ‘Analysis of rotor core eddy-current losses in drive’ (Department of Electric Power Engineering, Göteborg, 1999)
interior permanent magnet synchronous machines’. Industry Applications [84] El-Refaie, A.M.H.N.C.a.J.T.M.: ‘Thermal analysis of multibarrier interior PM
Society Annual Meeting, Edmonton, Alta, October 2008 synchronous machine using lumped parameter model’, IEEE Trans. Energy
[53] Ma, L., Sanada, M., Morimoto, S., et al.: ‘Prediction of iron loss in rotating Convers., 2004, 19, (2), pp. 303–309
machines with rotational loss included’, IEEE Trans. Magn., 2003, 39, (4), [85] Mellor, P.H., Roberts, D., Turner, D.: ‘Lumped parameter thermal model for
pp. 2036–2041 electrical machines of TEFC design’. IEEE Proc. B-Electr. Power Appl.,
[54] Ridge, A., McMahon, R., Kelly, H.P.: ‘Detailed thermal modelling of a 1991, 138, (5), pp. 205–218
tubular linear machine for marine renewable generation’. IEEE Int. Conf. on [86] Nategh, S., Wallmark, O., Leksell, M., et al.: ‘Thermal analysis of a
Industrial Technology, Cape Town, 2013 PMaSRM using partial FEA and lumped parameter modeling’, IEEE Trans.
[55] Rasilo, P., Belahcen, A., Arkkio, A.: ‘Importance of iron-loss modeling in Energy Convers., 2012, 27, (2), pp. 477–488
modelling in simulation of wound-field synchronous machines’, IEEE Trans. [87] Sali, N.V.a.K.P.: ‘Lumped parameter analysis of SMPM synchronous electric
Magn., 2012, 48, (9), pp. 2495–2504 motor used for hybrid electric vehicle traction drive’, IOSR J. Mechan. Civil
[56] Deeb, R., Janda, M., Makki, Z.: ‘Prediction of eddy current losses of surface Eng., 2014, 2014, pp. 42–47
mounted permanent magnet servo motor’. Int. Conf. on Electrical Machines, [88] Staton, D., Boglietti, A., Cavagnino, A.: ‘Solving the more difficult aspects of
Marseille, September 2012 electric motor thermal analysis’. IEEE Int. Electric Machines and Drives
[57] Zhu, Z.Q., Ng, K., Schofield, N., et al.: ‘Improved analytical modelling of Conf., 2003. IEMDC'03, Madison, 2003
rotor eddy current loss in brushless machines equipped with surface-mounted [89] Illiano, E.: ‘Design of a highly efficient brushless current excited synchronous
permanent magnets’, IEE Proc. Electr. Power Appl., 2004, 151, (6), pp. 641– motor for automotive purposes’. PhD thesis, ETH, Zurich, 2014
650 [90] Mejuto, C.: ‘Improved lumped parameter thermal modelling of synchronous
[58] Heyns, G.C., Wang, R.-J.: ‘Thermal analysis of a water-cooled interior generators’. PhD thesis, The University of Edinburgh, 2010
permanent magnet traction machine’. IEEE Int. Conf. on Industrial [91] Nategh, S.: ‘Thermal analysis and management of high-performance electrical
Technology (ICIT), Cape Town, February 2013, pp. 416–421 machines’. PhD thesis, KTH, 2013
[59] Miller, T.J.E.: ‘SPEED's electric machines’ (CD-Adapco, 2002) [92] Shafaie, R., Kalantar, M., Gholami, A.: ‘Thermal analysis of 10-MW-class
[60] Dessouky, Y.G., Williams, B.W., Fletcher, J.E.: ‘Cooling enhancement of wind turbine HTS synchronous generator’, IEEE Trans. Appl. Supercond.,
electric motors’, IEE Proc. Electr. Power Appl., 1998, 145, (1), pp. 57–60 2014, 24, (2), pp. 90–98
[61] Lee, Y., Hahn, S.Y., Kauh, S.K.: ‘Thermal analysis of induction motor with [93] Junak, J., Ombach, G., Staton, D.A.: ‘Permanent magnet DC motor brush
forced cooling channels’, IEEE Trans. Magn., 2000, 36, (4), pp. 1394–1397, transient thermal analysis’. 2008 Int. Conf. on Electrical Machines,
Part 1 Vilamoura, 2008
[62] Farsane, K., Desevaux, P., Panday, P.K.: ‘Experimental study of the cooling of [94] Cossar, C., McGilp, M., Omori, S., et al.: ‘Analytical thermal models for
a closed type electric motor’, Appl. Therm. Eng., 2000, 20, (14), pp. 1321– small induction motors’. 18th Int. Conf. on Electrical Machines, Vilamoura,
1334 2008
[63] Li, H.: ‘Cooling of a permanent magnet electric motor with a centrifugal [95] Tang, Y.: ‘Characterization, numerical analysis, and design of switched
impeller’, Int. J. Heat Mass Transf., 2010, 53, (4), pp. 797–810 reluctance motors’, IEEE Trans. Ind. Appl., 1997, 33, (6), pp. 1544–1552
[64] Nakahama, T., Suzuki, K., Hashidume, S., et al.: ‘Cooling airflow in [96] Faiz, J., Ganji, B., Carstensen, C.E., et al.: ‘Temperature rise analysis of
unidirectional ventilated open-type motor for electric vehicles’, IEEE Trans. switched reluctance motors due to electromagnetic losses’, IEEE Trans.
Energy Convers., 2006, 21, (3), pp. 645–651 Magn., 2009, 45, (7), pp. 2927–2934

12 IET Electr. Syst. Transp.

© The Institution of Engineering and Technology 2016
[97] Zhang, Y., Ruan, J., Huang, T., et al.: ‘Calculation of temperature rise in air- [112] Jungreuthmayer, C., Bäuml, T., Winter, O., et al.: ‘A detailed heat and fluid
cooled induction motors through 3-D coupled electromagnetic fluid- flow analysis of an internal permanent magnet synchronous machine by
dynamical and thermal finite-element analysis’, IEEE Trans. Magn., 2012, 48, means of computational fluid dynamics’, IEEE Trans. Ind. Electron., 2012,
(2), pp. 1047–1050 59, (12), pp. 4568–4578
[98] Marignetti, F., Colli, V.D., Coia, Y.: ‘Design of axial flux PM synchronous [113] Pechanek, R., Bouzek, L.: ‘Analyzing of two types water cooling electric
machines through 3-D coupled electromagnetic thermal and fluid-dynamical motors using computational fluid dynamics’. Int. Power Electronics and
finite-element analysis’, IEEE Trans. Ind. Electron., 2008, 55, (10), pp. 3591– Motion Control Conf. (EPE/PEMC), 2012
3601 [114] Polikarpova, M., Ponomarev, P., Lindh, P., et al.: ‘Hybrid cooling method of
[99] Wu, W., Dunlop, J.B., Collocott, S.J., et al.: ‘Design optimization of switched axial-flux permanent-magnet machines for vehicle applications’, IEEE Trans.
reluctance motor by electromagnetic and thermal finite element analysis’. Ind. Electron, 2015, 62, (12), pp. 7382–7390
Digest of INTERMAG 2003. Int. Magnetics Conf., vol. 39, no. 5, 2003, pp. [115] Boglietti, A., Cavagnino, A., Staton, D.A.: ‘Determination of critical
3334–3336 parameters in electrical machine thermal models’, IEEE Trans. Ind. Appl.,
[100] Srinivas, K.N., Arumugam, R.: ‘Analysis and characterization of switched 2008, 44, (4), pp. 1150–1159
reluctance motors: part ii - flow, thermal and vibration analyses’, IEEE Trans. [116] Yoheswaran, B., Pullen, K.R.: ‘Flow and convective heat transfer in disk-type
Magn., 2005, 41, (4), pp. 1321–1332 electric machines with coolant flow’. Int. Conf. on Electrical Machines
[101] Pickering, S.J., Lampard, D., Shanel, M.: ‘Ventilation and heat transfer in a (ICEM), 2014
symmetrically ventilated salient pole synchronous machine’. Int. Conf. Power [117] Wrobel, R., Mellor, P.H., Holliday, D.: ‘Thermal modeling of a segmented
Electronics, Machines and Drives, June 2002 stator winding design’, IEEE Trans. Ind. Appl., 2011, 47, (5), pp. 2023–2030
[102] Howey, D.A., Holmes, A.S., Pullen, K.R.: ‘Measurement and CFD prediction [118] Boglietti, A., Cavagnino, A., Staton, D.A., et al.: ‘End space heat transfer
of heat transfer in air-cooled disc-type electrical machines’, IEEE Trans. Ind. coefficient determination for different induction motor enclosure types’, IEEE
Appl., 2011, 47, (4), pp. 1716–1723 Trans. Ind. Appl., 2009, 45, (3), pp. 929–937
[103] Connor, P.H., Pickering, S.J., Gerada, C., et al.: ‘Computational fluid [119] Yang, Y., Schofield, N., Emadi, A.: ‘Double-rotor switched reluctance
dynamics modelling of an entire synchronous generator for improved thermal machine design, simulations, and validations’, IET Electr. Syst. Transp., 2016,
management’, IET Electr. Power Appl., 2013, 7, (3), pp. 231–236 6, (2), pp. 171–175
[104] Wrobel, R., Vainel, G., Copeland, C., et al.: ‘Investigation of mechanical loss [120] Cavagnino, A., Tenconi, A., Vaschetto, S.: ‘Experimental characterization of a
components and heat transfer in an axial-flux pm machine’, IEEE Trans. Ind. belt-driven multiphase induction machine for 48-V automotive applications:
Appl., 2015, 51, (4), pp. 3000–3011 losses and temperatures assessments’, IEEE Trans. Ind. Appl., 2016, 52, (2),
[105] Schrittwieser, M., Marn, A., Farnleitner, E., et al.: ‘Numerical analysis of heat pp. 1321–1330
transfer and flow of stator duct models’, IEEE Trans. Ind. Appl., 2014, 50, [121] Michalski, L.: ‘Temperature measurement’ (John Wiley & Sons, New York,
(1), pp. 226–233 2001)
[106] Huang, Z., Marquez, F., Alakula, M., et al.: ‘Characterization and application [122] Ganchev, M., Kubicek, B., Kappeler, H.: ‘Rotor temperature monitoring
of forced cooling channels for traction motors in HEVs’. 20th Int. Conf. on system’. Int. Conf. on Electrical Machines (ICEM), 2010
Electrical Machines (ICEM), 2012 [123] Mejuto, C., Mueller, M., Shanel, M., et al.: ‘Improved synchronous machine
[107] Shanel, M., Pickering, S.J., Lampard, D.: ‘Conjugate heat transfer analysis of thermalmodelling’. Int. Conf. on Electrical Machines (ICEM), 2008
a salient pole rotor in an air cooled synchronous generator’. IEEE Int. Conf. [124] Stipetic, S., Kovacic, M., Hanic, Z., et al.: ‘Measurement of excitation
of Machines and Drives, 2003 winding temperature on synchronous generator in rotation using infrared
[108] Maynes, B.D.J., Kee, R.J., Tindall, C.E., et al.: ‘Simulation of airflow and thermography’, IEEE Trans. Ind. Electron., 2012, 59, (5), pp. 2288–2298
heat transfer in small alternators using CFD’. IEEE Proc. Electrical Power [125] Specht, A., Wallscheid, O., Böcker, J.: ‘Determination of rotor temperature
Applications, 2003 for an interior permanent magnet synchronous machine using a precise flux
[109] Ujiie, R., Arlitt, R., Etoh, H.: ‘Application of computational fluid dynamics observer’. Int. Power Electronics Conf. (ECCE-Asia), Hiroshima, Japan, 2014
(CFD) on ventilation-cooling optimization of electrical machines’. Review [126] Reigosa, D.D., Briz, F., Garcia, P., et al.: ‘Magnet temperature estimation in
Energy Technologies Generation, Transmission and Distribution of Electric surface PM machines using high-frequency signal injection’, IEEE Trans. Ind.
and Thermal Energy, 2006 Appl., 2010, 46, (4), pp. 1468–1475
[110] Nollau, A., Gerling, D.: ‘A new cooling approach for traction motors in [127] Wallscheid, O., Böcker, J.: ‘Global identification of a low-order lumped-
hybrid drives’. IEEE Int. Electric Machines and Drives Conf. (IEMDC), 2013 parameter thermal network for permanent magnet synchronous motors’, IEEE
[111] Schrittwieser, M., Bíró, O., Farnleitner, E., et al.: ‘Analysis of temperature Trans. Energy Convers., 2016, 36, (1), pp. 354–365
distribution in the stator of large synchronous machines considering heat
conduction and heat convection’, IEEE Trans. Magn., 2015, 51, (3), pp. 49–
53

IET Electr. Syst. Transp. 13

© The Institution of Engineering and Technology 2016