1 Introduction

Since the second coming of electrical vehicles (EVs) by GM EV1 in 1996, the automotive industry has recognized the potential of such technology as a replacement for the conventional internal combustion engine (ICM). Nevertheless, the immaturity of the electrical motor and battery technologies at that time rendered the EV1 to be incomparable to the ICM vehicles in both performance and cost. However, transportation electrification became a necessity rather than an option due to alarming levels of climate change and the consequent legislation. Therefore, the automotive industry has taken partial steps toward clean vehicles by introducing engine electrification and hybrid electrical vehicles (HEVs) to avoid a dramatic change and to allow the development and evolution of the motor and battery technologies.

Looking into the products of EV and HEV from the 1996 inception to the late-2010s, all the know topologies, i.e., DC field rotor (DC), induction machine (IM), switched reluctance (SR) and permanent magnet (PM) have made a presence in automotive [1]. To the author’s knowledge, the first topology to leave the race is the SR, beside Holden-ECO, SR has not been used for this application. This is since the main challenges of SR machines are low power density and torque ripple which are both essential requirements for automotive as well as the need for an unconventional converter which rises the total cost of the drive. The IM and DC have also declined, their current presences are only in Tesla and Renault, respectively. The PM has dominated the automotive market having most of the electric motors including the Tesla’s latest 2016 models S. The different PM machine topologies including surface-mounted (SPM), interior (IPM), and PM-assisted synchronous reluctance (SynRel-PM) have all been seen in automotive [1]. Despite the heavy presence of PM machines in automotive, the fact that PM material is needed makes a great challenge. Therefore, reduced PM design techniques have been proposed and used in the latest technology. Other types of permanent magnets, such as ferrite and AlNiCo have also been fairly explored for automotive applications [2].

Therefore, in this paper, ten different electrical machine designs based on machines used in different EVs and HEVs are selected for a comparative study. The study aims to evaluate the different electrical machines topologies, structure, operation conditions and performance. In addition to electromagnetic performance, mechanical and thermal results have been compared. Furthermore, different state-of-art design techniques have been explored and studied by employing them in the compared machines. This study explores the possibility to push the current limits of electrical machines to meet the future automotive targets.

2 Analysis of different types of electrical machines based on market research

Table 1 presents ten vehicles, both hybrid and full electric, which their electric motors have been selected for comparison in this section.

Table 1 Electric motors of hybrid and electric vehicles used in the study
Full size table

Figure 1 presents the cross section and winding layout of the ten machines. The machines have been selected based on the availability of the information of the machines in the literature [3,4,5,6,7,8,9,10,11,12,13,14,15]. The selected machines have DC, IM and PM topologies, PM machines having SPM, IPM and SynRel-PM rotor topologies with mostly NdFeB magnets and one design (based on Chevrolet Volt 2) with ferrite. The analyses are made using ANSYS package.

Fig. 1
figure 1

Cross sections of the studied machines. (a) HA (b) HC (c) TP1 (d) TP2 (e) CV1 (f) CV2 (g) BI (h) NL (i) RZ (j) TS

Full size image

Honda Accord (HA) and Civic (HC), having a conventional SPM structure with a concentrated winding. Mechanically, the PMs are buried in the rotor core frame in HA, whereas a retaining sleeve is used in the HC. The designs merit techniques are segmented stator teeth with concentrated winding, therefore, automated winding with high impact factor is an advantage [3]. However, the SPM structure consumes high PM quantity, in addition to the complex aforementioned design features, i.e., buried PM and sleeve, that are required for mechanical safety.

2010 Toyota Prius (TP1) is one of the most compared machines in the field of electrical machines, it has been used as a benchmark for automotive design research in many publications [4]. This is due to the success of the vehicle and the widely available information of the motor. The IPM structure offers high torque per magnet volume and, reluctance torque and consequently wide power-speed range [5, 6]. Similarly, the 2014 Toyota Prius (TP2) have an IPM rotor structure with a new magnet set near the rotor surface. The mechanical design of the rotor has complex-shaped gaps to maintain good mechanical performance and allow for rotor cooling channels.

The Chevrolet Volt two motors (C1) and (C2) having different rotors and the same stator, however, the axial length is smaller in CV2. Both motors are SynRel-PM, CV1 designed with a conservative quantity of NdFeB, whereas, CV2 is one of a very few automotive motors with ferrite. The design utilizes the low cost and mass density of ferrite by employing large arc-shape quantities [7,8,9].

BMW i3 (BI) and Nissan Leaf (NL) are both EVs in the compared machines in this study. Both motors are SynRel-PM structure with reduced PM quantity. BI with slitting design techniques focusing on utilizing the reluctance torque for the benefit of low magnetic quantity [10, 11]. NL having an unorthodox rotor design with a thicker magnet near the outer rotor radius and V-shaped thinner magnets near the inner radius [12, 13].

Finally, the non-PM machines of Renault Zoe (RZ) and Tesla Model S (TS), having a traditional DC rotor (field wound, FW) and IM structure, respectively [12, 13]. RZ rotor consists of a conventional commutated field winding rotor, it is worth noting that the entire Renault fleet of HEVs and EVs are DC rotor machines [14]. TS IM has tackled the rotor heating issue with a complex but innovative rotor cooling [15].

Table 2 lists the main design parameters of the ten machines. The packing factor of all concentrated winding (CW) and distributed winding (DW) are assumed to be 0.4 and 0.6, respectively. Table 3 presents a summary of the open circuit and electromagnetic load results of the ten machines.

Table 2 Design parameters and specifications of the studied machines
Full size table
Table 3 Performance summary of the studied machines
Full size table

Similar torque density is observed in the PM topologies which is around twice of the non-PM topologies. Power density has a similar trend with the PM topologies having around 3 to 4 times the value of non-PM topologies. Standing out is the CV2 the topology with ferrite, this is due to the large quantity of the low mass density ferrite and the multi-flux paths to allow more reluctance torque.

Torque and power per magnet weight are factors used to estimate the cost of the machine. Beside the SPM topologies which have the best utilization of the PM material, SynRel topologies have overall higher torque and power per magnet weight compared to the IPM. Moreover, utilizing the inverter kVA, i.e., torque per kVA and power per kVA, shows HA and HC in top due to their low voltage input, i.e., high number of turns. However, looking at the utilization of current density, SynRel topologies have the highest power per current density similar to the non-PM topologies values. Higher torque per current density is also found in the SynRel topologies. This shows how the SynRel differs from the IPM by higher utilization of the reluctance torque.

A summary of the performance of the compared machines:

  • NL has the highest power density and moderate current density, however, it lacks in size and power per volume.

  • CV2 has a high power density and power per volume, CV2 is the only topology employing ferrite magnets, the low mass density of the ferrite leading to a small total weight. Additionally, the high current density contributes to the higher power capability.

  • Small and consequently light-weighted machines are the TP1 and TP2. This leads to high power and torque density as well as high torque per volume. However, small size with high power density makes these machines prone to high temperature rise.

  • PM-assisted SynRel topologies have the highest power density and power per volume as well as torque density and torque per volume. The performances come with moderate current density and therefore manageable temperature rise. Additionally, these topologies have good utilization of the current density, i.e., power and torque per current density.

  • Magnet-less topologies, i.e., RZ and TS, can achieve good power and torque densities. However, rotor temperature is expected to be an issue for such machines due to the active copper in the rotors. Nevertheless, RZ wound field machine has the advantage of variable field current and therefore better field weakening capability at the expense of commutator complexity and losses.

  • Naturally cooled machines, i.e., HC and HA, have low power and torque density. Additionally, SPM topologies suffer from low field weakening capability making them not suitable for a wide speed range application. However, the lack of active cooling reduces the system’s complexity and weight.

  • BI machine has low current density and therefore possibly does not require a cooling system. However, high power and torque density have been achieved by additional active material in the machine. Similar to the HC and HA, the lack of active cooling reduces the system’s complexity and weight.

3 Results summery

A summary of the performance of the compared machines:

  • NL has the highest power density and relatively moderate current density and consequently temperature rise, however, it lacks in torque density and torque per volume compared to other PM machines.

  • CV2 has a high power density and power per volume, CV2 is the only topology employing ferrite magnets, the low mass density of the ferrite leading to a small total weight. Additionally, the high current density and the required intensive cooling of CV2 contribute to the higher power capability. However, mechanically, high stress due to the rotor geometry limits the speed capability.

  • Small and consequently light-weighted machines are the TP1 and TP2. This leads to high power and torque density as well as high power and torque per volume. However, small size with high power density makes these machines prone to high temperature rise. The small rotor diameter reflects well on the mechanical performance making the stress very low and allows for higher speed operation.

  • PM topologies with IPM and PMSR rotor configurations have the highest power density and power per volume as well as torque density and torque per volume. The performances come with moderate to high current density and therefore manageable temperature rise with intensive cooling methods. Additionally, these topologies have good utilization of the current density, i.e., power and torque per current density. However, the rotor geometry in some of the PM machines results in high mechanical stress and hence limits the speed range.

  • Magnet-less topologies, i.e., RZ and TS, can achieve good power and torque densities. However, rotor temperature is an issue for such machines due to the active copper in the rotors. Nevertheless, RZ wound field machine has the advantage of variable field current and therefore better field weakening capability at the expected expense of commutator complexity and losses.

  • Naturally cooled machine, i.e., HC, and less complex cooled, i.e., HA, have low power and torque density. Additionally, SPM topologies suffer from low field weakening capability making them not suitable for a wide speed range application.

  • BI machine has low current density and therefore less complex cooled. However, the high power and torque have been achieved by additional active material in the machine.

Furthermore, to summarize the outcome of the comparative study, Fig. 2 presents the trend of the torque density, power density, the rotor peripheral speed, rotor mechanical stress, and the required cooling methods under conventional and hairpin winding arrangements. The peripheral speed is calculated at the machine’s base speed and the mechanical stress is predicted at speed of 15krpm.

Fig. 2
figure 2

The trend of electrical machines in EVs and HEVs. (a) Torque density (b) Power density (c) Rotor peripheral speed at machine’s base speed (d) Rotor mechanical stress at 15 krpm (e) Required cooling method (conventional winding) (f) Required cooling method (hairpin winding)

Full size image

From Fig. 2, it can be seen that the current limit of the torque and power densities are 13Nm/kg and 4.5 kW/kg, respectively. Moreover, along the timeline, the increase in the torque and power density are in-line with the increase in the more aggressive cooling methods. Similarly, the increase in the rotor mechanical stress and peripheral speed along the timeline is in-line with the torque and power densities.

4 New design criteria for future automotive

By the end of 2019, light-duty vehicles registered in the EU, Switzerland, Norway, Iceland and Turkey must comply with the worldwide harmonized light vehicle test procedure (WLTP) driving cycle as a replacement for the new European driving cycle (NEDC). Figure 3 presents the class 3 WLTP drive cycle [16].

Fig. 3
figure 3

WLTP driving cycle

Full size image

The UK advance propulsion center (APC) has summarized the requirements for electric motors in automotive to meet the legislative requirements of 2025 and 2035 as shown in Table 4 [17]. Similarly, the USA equivalent, FreedomCar (FC), published the criteria for 2020 [18]. Considering a maximum output power of 100 kW, the required torque-speed characteristic to meet the requirements have been predicted for different gearbox ratios and presented in Fig. 4. This summarizes the targets for electrical machine design for automotive to meet the 2035 performance requirement. Cost requirements and material suitability, on the other hand, are more difficult to quantify, however, rare-earth-free, low-cost steel grade and low copper consumption are targets to be considered in the design of electric machines for automotive.

Table 4 Electrical machine targets by the APC and FC [17, 18]
Full size table
Fig. 4
figure 4

Predicted torque-speed characteristics for future electrical machines in automotive

Full size image

5 Designing techniques

This section explores and presents different electrical machine design techniques that can help electrical machine designers to reach the future requirements for electric motors in automotive.

5.1 Cooling system

To explore the different electrical machine cooling systems and their heat removal capability, a study has been made using the ten compared machines. The study consists of employing different cooling systems on the machines and observe the temperature in the winding, steel and PM of the machine.

The temperature rise has been predicted using a combined LPC model and mathematical thermal calculations. Different cooling methods have been evaluated on all the machines as shown in Table 5. This is to have a fair comparison of the thermal capability of the different topologies as cooling specifications of some of the machines are not available in the literature. Neutral cooling, stator jacket, stator jacket with spray cooling and stator slot channels have been investigated. Figure 5 presents and illustrates the four methods. The calculation is made at peak conditions with a transient of 1200 s and 3000 rpm, and a flow rate of 6 l/min of EGW50/50 cooling oil.

Table 5 Temperatures (°C) in different motor parts under different cooling methods
Full size table
Fig.5
figure 5

Illustration of the cooling methods. (a) Natural convection, 10 mm thick aluminum case (b) Stator jacket, 10 mm thick spiral aluminum jacket with 2 mm inner wall and 5 mm channels thickness (c) Stator jacket and spray cooling (d) Stator slot cooling, 2X 1 mm diameter channels

Full size image

Both HA and HC operate with low current density, therefore, natural air cooling is suitable to maintain the machine temperature around 200 °C. Therefore, with a high temperature magnet grade and winding with insulation class 4 the machine can operate with natural convection. Similarly, the low current density of BI makes it operational with only natural convection.

TP1 and TP2 have the highest current density and consequently copper loss making the temperature rise extremely high. A stator slot channels cooling is therefore required to bring down the winding temperature. Similarly, CV1, CV2 and NL have high current density leading to the need for excessive cooling of both spray and stator jacket or stator slot channels cooling.

Magnet-less topologies in RZ and TS have high current density and therefore stator jacket or stator slot channels cooling is required. However, the active field winding and induction cage in the rotor leading to the need for rotor cooling. Therefore, spray cooling is a suitable solution. However, TS uses active shaft cooling [14].

5.2 Hairpin winding

The technology of hairpin winding allows for an automated winding process of the machine and the large rectangular coils achieve a high winding filling factor and therefore allow for more current or lower loss density. However, low number of turns and high AC loss are drawbacks of hairpin winding that are not very effective in automotive since relatively low number of turns are used and the moderate speed renders AC loss as insignificant.

The hairpin winding has been studied through employing such windings in the studied machines and compare them with their original counterparts. Original stator and parallel slot of each machine have been studied. Figure 6 presents an illustration of the conventional and hairpin winding. The fill factor and winding temperature rise are presented and compared in Table 6. Hairpin winding can be made with a low number of turns, current products are made with up to 6 turns [19], therefore, TP2 turns has changed from 11 to 4 turns and consequently the current changed from 235A to 646.25A to maintain the same current density and ampere-turn. Similarly, HA turns changed to 2 turns and the current to 5850A, HC turns changed to 2 turns and the current to 6875A.

Fig. 6
figure 6

Conventional and hairpin winding illustration. (a) Conventional winding (b) Hairpin winding (original stator) (c) Hairpin winding (parallel slot)

Full size image
Table 6 Comparison of conventional and hairpin winding in different machines, DC loss (W), Temperature (°C)
Full size table

5.3 Mechanical rotor design

To achieve high performance, in particular power density, high speed and a large rotor diameter are both used. This places a high demand on innovative rotor geometry to keep the mechanical stresses with the acceptable limits of the used material. To examine the mechanical capability of different rotor topologies, the radii of the investigated machine’s rotors have been changed to 100 mm. Table 7 lists the maximum stress and deformation of each of the rotors.

Table 7 Maximum stress and deformation of the investigated machines with 100 mm radius and at 15 krpm
Full size table

From the comparison, the highest stresses are exhibited by NL, this is due to the location of a thick magnet near the rotor surface. The rotor core that holds this magnet is under high stress due to the high deformation at the center as shown in Fig. 7. This can be avoided by multi-layer magnets having the smallest near the surface like in CV1, CV2 and BI. In addition, these machines have inner ribs between or at the sides of the larger magnets to reduce the deformation along the center. However, magnetically these ribs are undesirable due to leakage. Therefore, adding gaps between the magnets and ribs introduces a compromise between mechanical integrity and electromagnetic performance. Finally, the lowest stresses are found in the TP2. This is since the magnet near the surface is placed further inside the core with bridges to withstand deformation. The design provides a balance between mechanical and electromagnetic performance. The side magnets are held with notches instead of full bridges, whereas the core is held by bridges away from the magnets to provide a balance. HA machine encloses the magnets in core pockets with a middle bridge making the mechanical design suitable for maintaining stress low. The sleeve in the HC machine maintain a low stress level in the rotor and the sleeve, however, the process of applying retaining sleeve is considered complex and costly. Finally, both the RZ and TS rotors have low stresses. This is since the large RZ rotor teeth hold the winding. Similarly, the TS rotor core has a large core frame surrounding the copper cage.

Fig. 7
figure 7

Mechanical stress and deformation distribution in the studied machines

Full size image

On the other hand, the magnets are brittle with low yield/tensile strength, therefore, they need to be carefully retained. Enclosing the magnets by the rotor core and end-caps ensures mechanical safety since no space is allowed for magnet deformation. Additionally, in the case of breakage, the magnet fragments will remain in their location and this will not inhibit the machine’s performance. Gaps between the magnet and core can be filled with epoxy to maintain a full magnet enclosure. Other techniques to reduce magnet stress is corner gaps as highlighted in Fig. 8. Such gaps allow magnet deformation at the point of maximum impact with the rotor core under rotational forces.

Fig. 8
figure 8

Magnet-core gaps to reduce the stress on the magnet

Full size image

5.4 Reduced rare-earth, rare-earth-free and magnet-less topologies

From the comparison summary, it has been seen that magnet-less topologies with an active rotor suffer from high weight and therefore low power density and the need for rotor cooling. On the other hand, magnet-less topologies with a passive rotor, i.e., synchronous and switched reluctance have inertly low power density. Figure 9 presents the cross section and torque-speed curve for a switched reluctance (SR) machine designed for an automotive application, while Table 8 lists key specifications. It can be seen that the power density is low. Additionally, the need for a high number of slot leads to concentrating the copper loss across a larger area making it difficult to extract the heat.

Fig. 9
figure 9

SR Machine cross section and torque-speed curve

Full size image
Table 8 Key specifications of SR machine
Full size table

Reduced-rare-earth (RRE) and rare-earth-free (REF) are a positional solution to the cost and power density requirements. RRE is the topologies with a combination of NdFeB and ferrite [20]. This allows for a relatively high magnetic loading and avoids demagnetization of ferrite magnets while reducing the consumption of the rare-earth material. However, the need to replace rare-earth with sustainable materials makes this approach a short term solution.

REF CV2 machine has an identical stator to the CV1 with shorter axial length and less power. To investigate the capability of REF machine, the CV2 axial length has been changed to compare the performance of both machines at the same size and same output torque, i.e., increased axial length to match the torque of the CV1 machine. Figure 10 presents the torque-speed and power speed curves of the CV1 (51.5 mm axial length), CV2 (31.5 mm axial length), CV2 with 51.5 mm axial length and CV2 with 71.5 mm axial length.

Fig. 10
figure 10

Torque-speed characteristics of CV1 and CV2 machines with different axial lengths

Full size image

5.5 Aggressive cooling

The key to high power density is a smaller machine with a higher current density. A small, lightweight machine needs a high current density to match the automotive target, therefore, high loss density and consequently heat is expected. Mechanically, the small size allow for higher speeds due to the small rotor diameter.

The answer for excessive cooling is fully flooded and semi-flooded methods. The first being undesirable due to the high wind age loss. Semi-flooded cooling has been used in the aerospace applications and the research has matured this technology to a practical level [21]. The challenges with semi-flooded cooling are mechanical and assembly rather than theoretical design issues. Elimination of fluid leakage and sealing of the system are the main challenges.

To examine the effectiveness of the semi-flooded cooling, a study on the compared machines have been conducted. The theoretical design of a semi-flooded cooling has been illustrated in [22]. A separation sleeve between the stator and rotor, i.e., placed in the airgap is required. The thickness of the sleeve is determined based on the flow rate of the fluid and the slot opening thickness. Similar to retaining sleeves in high speed machines, the sleeve needs to be made of a non-magnetic, non-electrically conductive material to avoid electromagnetic and losses issues. A closed-slot machine could be a solution to avoid the sleeve and consequently increasing the airgap length. Closed-slot technique has been proposed and employed in high-speed machines to reduce the space harmonics [22]. However, closed-slot stator suffers from high voltage spikes as stated in [23], which affect the torque-speed characteristics. Coincidentally, a higher number of stator slots diminishes this effect as stated in [23]. Additionally, flux leakage through the closed-slot tip leads to lower torque and influences the torque-speed characteristics. This can be avoided by using heat treatment electrical steel, i.e., dual phase steel [24], however, the low magnetic properties of such steel can be a challenge.

Theoretically, the most electromagnetically suitable design for a semi-flooded machine is a conventional stator with open slots and no sleeve in the airgap, i.e., small electromagnetic airgap and no stator flux leakage. This can be achieved assuming sealing techniques are applied to close the stator opening and prevents any fluid leakage. Semi-flooded cooling has been studied on the compared machines using sleeve, closed-slot and sealed slot stator. A sleeve of 1 mm made of carbon fiber is used in the machines, therefore the airgap increased by 1 mm at the expense of the rotor. Similarly, a closed-slot of 1 mm tooth-tip has been used. Figure 11 presents the illustration of the sleeved stator, closed-slot and sealed slot. Table 9 presents the results.

Fig. 11
figure 11

Illustration of stator slot opening for semi-flooded cooling. (a) Sleeved stator (b) Closed-slot stator (c) Sealed slot stator

Full size image
Table 9 Comparison of torque (Nm) and temperature (°C) in semi-flooded machines with sleeved stator, closed-slot stator and sealed slot stator
Full size table

5.6 Design to recycle

Although this aspect has not been introduced as a mandatory requirement for EV and HEV, the latest legislative requires more than + 80% of the vehicle to be recyclable. This has been specified in from of recyclable materials. The recycling of electric machines and consequently design-to-recycle (DtR) has been briefly explored in the literature of electrical machine design [25]. The recycling aspect of an electrical machine is divided into recyclable materials and recyclable designs. The first refers to using materials that are recyclable and the second refers to using design techniques to avoid high integration of the different parts and materials of the electrical machine and easy disassembly.

Recyclable materials: Aluminum versus copper winding; aluminum being the most widely recyclable material making aluminum windings a main DtR technique. Aluminum windings have been proposed and studied for high-speed machines due to the better AC/DC loss ratio. However, for moderate speed machines like in automotive applications aluminum winding means higher winding loss and temperature rise. Aluminum hairpin winding can help introducing aluminum into automotive electrical machines since the higher filling factor can compensate for the higher winding loss. Similar to ferrite, the low cost and low mass density of aluminum are additional merits.

Silicon-contained laminated steel versus solid steel; according to [25], silicon-contained laminated steel became unrecyclable due to the contamination with adhesive. Therefore, soft magnetic composite (SMC) and solid steel are suitable solutions for electrical machine core recyclability. However, in addition to the low magnetic properties, the SMC is a brittle material with low mechanical strength. This makes the SMC unsuitable automotive. On the other hand, solid steel has main challenges of low magnetic properties and high eddy current losses which require several techniques to overcome [26]. The non-coated Ferrite/SmCo versus the coated NdFeB; due to the need for coating, NdFeB magnets are very difficult to recycle due to the high coating contamination [25]. A recycled NdFeB has half the magnetic performance of newly made magnets [2]. Therefore, non-coated SmCo and ferrite are easier to recycle. As shown in Section C, Ferrite has shown good potential for automotive. The recyclability is additional merit to the low cost and weight. On the other hand, SmCo has very similar magnetic capability to the NdFeB and higher capability to withstand higher temperature. However, the high cost of SmCo magnet compared to NdFeB is the main disadvantage.

Recyclable designs: Coating, adhesive and varnish; to recycle a material, the purity of the material is the main object. Therefore, the use of coating, adhesive and varnish would contaminate the materials and removing them and separating them from the other materials is a very difficult process that might not be even possible. Therefore, a DtR machine must contain a reduced or eliminated coating, adhesive and varnish.

Disassembly: an easy to disassemble design is required for recycling. The ability to assemble the different parts without high integration and therefore separate the different components easily.

To explore the DtR aspects, aluminum winding and solid steel have been employed in the studied machines. Table 10 compares the original machines with the same machine albeit of aluminum winding in conventional and hairpin (parallel slot) configurations. The higher power density due to the low mass density of aluminum is the main advantage, the increase in winding loss and consequently the temperature can be avoided by using hairpin winding. Aluminum hairpin winding show the advantages of low AC loss which brings the total winding loss close to its counterpart copper hairpin. On the other hand, solid steel suffers from low B-H curve which results in lower torque. It is worth noting the solid steel suffers from large eddy currents which consequently lead to high loss and disturbance in the magnetic performance, especially if used in the stator. However, the large eddy current can be mitigated using axial segmentation or slitting [26].

Table 10 Comparison of copper and aluminum winding, (AC loss at 5000 rpm), Cu/Al: conventional copper/aluminum winding and CuH/Alh: hairpin copper/aluminum winding. DC loss (W), AC loss (W), Temperature (C), weight (kg)
Full size table

5.7 Multi-objective optimization

Although this aspect has not been introduced as a mandatory requirement for EVs, the optimization of electric vehicle seeks to reduce the amount of energy used while simultaneously enhancing their driving capabilities and keeping their costs and weight to a minimum. Because there is no one optimal solution that can satisfy all of these competing objectives, it is essential that the issue be formulated to take into account all of them at the same time. The term "multi-objective optimization," refers to a technology that helps find the optimum solutions that compromise many goals [27, 28].

The multi-objective optimization is another research hotspot. The most common and widely used method for optimizing electrical machines is performed using finite element analysis (FEA) coupled with optimization algorithms such as genetic algorithm (GA) [29,30,31], particle swarm optimization (PSO) algorithm [32,33,34], nondominated sorting genetic algorithm-II (NSGA-II) [35, 36], fuzzy method, and sequential Taguchi method [37, 38]. In addition, the application of multilevel optimization strategy is a promising method for efficient multi-objective optimization, in particular for the optimization methods that need to effectively solve the multi-parameter and multi-objective optimization problems [39].

The aforementioned methodologies have the potential to be used for the purpose of effective multi-objective optimization of electrical machines with high dimensions for applications in the automotive industry.

6 Conclusion

This paper presents a comparison of the state-of-art electrical machines and design techniques used in automotive. Ten designs based on the market available literature have been compared, these machines have different topologies and specifications. From this comparison machines with PM-assisted SynRel topologies have the highest power density and power per volume as well as torque density and torque per volume.

The future targets for electrical motors in automotive applications are illustrated by highlighting the advanced propulsion center (APC) targets for 2035, predicting a torque-speed profile and working toward the WLTP driving cycle. Furthermore, state-of-art design techniques have been studied through employing them in the various topologies to highlight the influence of such techniques and their ability to bring the designs to future targets. The investigated techniques are hairpin winding, rotor mechanicals, complex cooling, recyclable design, and reduced and rare-earth free machines. Hairpin winding provides a significant reduction in loss density of the winding and higher thermal dissipation. Therefore, a lower temperature rise and hence smaller size and higher power density can be achieved with hairpin winding. Complex cooling such semi-flooded allows for high heat removal from the stator winding allowing higher current density and consequently higher power density. However, issues of airgap sleeve and sealing reduce the performance of the machine and add assembly complexity. Rare-earth free machines, i.e., ferrite, have a potential for automotive. The low mass density allows for less weight, however, the weak coercivity of the ferrite make it easier to field weakening and hence a field weakening higher than 1 is normally found, this reduces the wide power capability over the speed range. Finally, aluminum hairpin winding is a promising for design recyclability in addition to the lower AC loss, lower weight and cost benefits.