Critical Review of Direct-Drive In-Wheel Motors in Electric Vehicles

Abstract

The primary challenge for electric vehicles in replacing oil-fueled vehicles today is their limited range, despite significant advancements in energy storage technology and alternative fuel vehicles over the past few decades. Direct-drive in-wheel motors (IWMs) can achieve higher efficiency by eliminating components such as gearboxes, differentials, and clutches, allowing for longer mileage with the same battery capacity. This positions them as a promising technology for the future of electric vehicles. This article primarily analyzes the key challenges that limit the widespread application of direct-drive IWMs in electric vehicles, including torque density, cost, reliability, efficiency, and ease of production. The article also investigates and compares the electromagnetic performance of the most representative motor topologies studied in direct-drive IWMs within both industrial and academic settings, and comprehensively evaluates the performance of these motor architectures with respect to the aforementioned performance requirements. Based on these investigations, this article aims to provide guidance and reference for the electromagnetic design and analysis of direct-drive IWMs.

1. Introduction

Energy shortages and the environmental crisis will be the primary concerns for society in recent and upcoming decades [1]. Electrifying vehicles in conjunction with a robust power grid is seen as a viable approach to harness renewable sources such as wind, solar, and hydroelectric power. Therefore, electric vehicles (EVs) stand out as a primary option for future transportation due to the finite nature of fossil fuels [2].

EVs have gained rapid progress over the past two decades, with all major automakers introducing their own EV models. In terms of their configuration and transmission system arrangement, EVs can be categorized as centralized drive systems or distributed drive systems [3]. The former is mass produced and widely accepted by consumers, with typical manufacturers include Tesla Inc, Volkswagen group, BYD, and so on. The latter, characterized by distributed drive, is considered the next-generation EV scheme due to its more compact structure and flexible maneuverability [4]. The direct-drive in-wheel motor system is a common example of a distributed drive system. Currently, the widespread adoption of in-wheel motors is on the horizon; for instance, Schaeffler supplied in-wheel motors to initial customers in 2023 [5].

Direct-drive in-wheel motor (IWM) systems show distinct strengths in many aspects when compared to conventional centralized drive systems. Unlike traditional indirect drive systems that require gearboxes, the direct-drive system of the IWM eliminates the need for a gearbox, resulting in higher integrated efficiency [4,6,7,8,9,10,11,12,13]. The research history of direct-drive IWM system spans over a century [1]. Utilizing IWMs in the drive system configuration can create significant free chassis space for other purposes [2], as exemplified by the GM Hy-wire concept vehicle [14,15].

Despite its advantages such as increased space and efficiency, direct-drive IWMs have not been extensively integrated into electric vehicles. One challenge lies in the need to incorporate mechanical brake disks into IWM drive systems, which must operate at relatively high heat loads due to the essential higher output torque required when the rotor and wheel rotate at the same angular speed. Presently, the benefits of direct-drive IWMs do not outweigh their drawbacks, including challenges related to brake disk installation and overheating, higher output torque demands, improved thermal design, and other factors [16].

Therefore, a comprehensive review of direct-drive IWMs is necessary, encompassing their technical requirements, current developments, and challenges, as well as future directions for exploration.

In light if this, this article conducts a review that focuses on the challenges of mass production and key performance requirements of direct-drive IWMs, discussing the applicability of different motor topologies, encompassing both industrially produced and scientifically researched motors. Specifically, this paper distinguishes itself from the existing literature reviews in the following aspects:

(1)

It systematically examines the technical requirements for IWM drive and provides a comprehensive overview of both the currently industrial-applied and under-researched types of IWMs.

(2)

It discusses the applicability of different IWMs based on key performance requirements, provides a technical evaluation, and offers insights into the future development directions of IWMs.

The rest of the article is constructed as follows: Section 2 introduces the primary challenges and design considerations specific to direct-drive IWMs, followed by the classifications of IWMs in Section 3. Section 4 reviews various direct-drive IWM topologies and offers a quantitative comparison for three selected IWMs. Moreover, Section 5 conducts comprehensive evaluations of promising IWM topologies, while Section 6 presents our concluding remarks.

2. Main Challenges and Design Specifications of Mass-Produced Direct-Drive IWMs

As a distributed drive system, the distinctive feature of IWM systems lies in their distribution across multiple wheels, powerful propulsion, and enhanced transmission efficiency compared to centralized drive systems. Figure 1 illustrates a centralized drive system, distributed indirect-drive system, and distributed direct-drive system for EVs. It can be observed that the centralized drive system, in addition to the essential components such as the battery, power electronics (PE), and motor, requires a gearbox and differential, leading to a decrease in transmission efficiency.

On the other hand, the distributed drive system situates the IWM and PE inside or near the wheels, allowing for more space to accommodate the battery, thereby enhancing the vehicle’s range effectively. Furthermore, the distributed direct-drive system eliminates the need for a reducer gear, further enhancing transmission efficiency and saving space.

2.1. Main Challenges of IWMs

The main challenges in mass-produced direct-drive IWMs for EVs are summarized in Figure 2, with detailed descriptions to follow in subsections.

(a)

High Torque Density

In EVs, the direct-drive IWM system lacks a gear reducer, necessitating higher output torque [17,18]. However, the torque density of IWMs is often insufficient to meet the requirements of various vehicle types [19]. Typically, the gear ratio of EV transmission systems is usually around 6 to amplify the output torque and obtain a minimum total mass [20], which gives a stringent output torque requirement for direct-drive systems.

In addition, the performance of electric machines is quite sensitive to overheating, and their efficiency is directly connected to the temperature of its power electronics and windings [21]. With more kinetic energy absorbed by the regenerative braking system, the temperature rise in the IWM’s rotor would be lowered, and then a larger margin is achieved for the thermal and electromagnetic design of IWMs. Hence, higher peak torque is essential for regenerative braking.

One successful product is the Protean Drive IWM, which can recover up to 85% of the kinetic energy available during braking [17]. The torque density of the IWM should be maximized to enhance driving and steering performance, including more robust torque characteristics and higher starting torque, all within the constraints of limited space and stringent reliability requirements [22].

The statistics of peak torque and power with respect to volume and weight in current mainstream IWM products are shown in Figure 3. It is evident that the torque, power, torque density, and power density of IWMs have progressed to achieve heightened levels.

Another critical aspect for enhancing torque density involves addressing the issue of unsprung weight in vehicles. The primary drawback of a high unsprung weight in EVs is the degradation of their dynamic characteristics. The driving comfort, a pivotal dynamic attribute, plays a crucial role in determining the viability of widespread EV adoption [23].

In order to study the effects of larger unsprung weight on IWMs’ riding, handling, and steering behaviors, different methods, including subjective assessment, objective measurements, and numerical analysis, are applied in [24]. This study is based on the Protean Electric PD18 IWM, of which the additional unsprung mass is 30 kg. For ride behavior, the results show that additional unsprung mass causes degraded behavior on pitch control, small impact feel, large impact feel, and unsprung shake. As for steering behavior, typically deficits appear due to additional unsprung mass, especially in effort linearity, effort build-up, and parking effort [24].

(b)

Low Cost

The cost of direct-drive IWMs are usually too high [25]. In a direct-drive EV, there must be a minimum of two IWM systems, leading to a dual-drive system. The drive motor and controller represent a significant portion of the overall powertrain budget. Consequently, the total cost of IWMs tends to exceed that of traditional centralized powertrain designs. Particularly in the context of electric commuter vehicles, the substantial cost associated with the high consumption of large permanent magnets (PM) in IWMs poses a challenge in reducing production expenses to levels comparable to those of centralized-drive electric vehicles [26].

Primarily, the economic benefits brought by direct-drive IWMs are because of the elimination of components engaged in torque transition, specifically the gear, differential, and other auxiliary parts. The outcome of all these eliminations would be a drive system that is distinctly more straightforward to design and manufacture, with fewer auxiliary parts, in a car that would give engineers far more noteworthy adaptability with various vehicle configurations [21]. Given that the main expenses of an in-wheel drive system are centered around the motors and power electronics, cost reductions in the motor component are especially critical.

(c)

Robustness

Another challenge of IWM design is the harsh operating circumstances, which is due to its mounting location [27]. Thus, the requirements on the mechanical design of IWMs is also higher. In addition to larger unsprung mass, the bumpy road surface could apply impact force on the motor and make the airgap of the motor deformed. The deformation of the airgap will generate high-order electromagnetic force harmonics, which becomes a significant source of noise excitation to the whole vehicle [28]. The bearing stiffness of the wheel hub in [29] is 5 MN/m, which is relatively high, but the peak deformation of the rear-left IWM could be 0.465 mm; while it is 0.19 mm, that of the rear-right IWM is under the step input operating condition [29]. The road surface roughness could cause an over 50% increment on the vertical load of the wheel assembly, which should be even larger considering the safety factor [30]. A typical IWM is powered by a battery located on the vehicle with cables. However, due to the harsh environments in which the power and signal cables of an IWM operate, there is a risk of disconnection caused by high acceleration or vibration.

Good sealing of the whole IWM is essential to operate smoothly, but it is hard to achieve, which is one major reason why the designers of the Ford F150 electric truck abandoned IWMs [31]. In order to overcome this problem, a wireless IWM (W-IWM) has been proposed. The risk of disconnection would disappear if the cables of the IWM are removed [32]. To fulfill the design specification requirements that the motor should be sealed to comply with IP68 and IP6k9k standards, a modular unit such as a motor controller and windings are fully sealed with waterproof material in Protean, but magnetic particles, salt, and water from outside could still enter into the IWM system, shortening the product’s lifespan [33].

(d)

High Efficiency

In addition to the inherent advantages of zero gearbox loss, the regenerative braking capability of a direct-drive system significantly enhances efficiency and extends mileage. The relatively shorter time constant of the electromagnetic system and high-frequency power electronics in EVs enable rapid and precise control, providing a clear edge over internal combustion engine (ICE) vehicles [34]. Consequently, implementing an electrical antiskid braking system and an improved regenerative braking system would be viable enhancements for the direct-drive system of IWMs.

The regenerative braking system in EVs is another strength compared to vehicles using ICE, as it eliminates the need for an additional hydraulic regenerative energy system. Higher regeneration efficiency recovers larger kinetic energy and thus alleviates the mileage anxiety. The mileage could be increased by 8–25% with the help of regenerative braking [35]. The regenerative energy of rule-based strategy, the model predictive control (MPC) velocity-tracking strategy, and the velocity optimization strategy are 294.86 KJ, 309.35 KJ, and 319.76 KJ, respectively, in [36], and the respective regenerative efficiency is 65.98%, 69.22%, and 71.55%. The MPC velocity-tracking strategy gains a 4.91% improvement in regenerative efficiency over the rule-based strategy. The regenerative ratio using the energy-efficient scheme in [37] could be 78.03%.

(e)

Ease of Production

For mass-produced IWM applications, ease of production is a critical metric. Ease of production translates to lower manufacturing time and costs, and it is essential for enabling mass production.

With advancements in motor technology, axial flux motors [38] and axial-radial flux motors have garnered significant attention due to their larger airgap flux area within the same volume, thus possessing greater torque output capabilities based on torque generation mechanisms. Therefore, these two types of motors are particularly well-suited for IWM applications. However, the reality is that these motor types involve significantly more complex manufacturing processes, especially in the fabrication of stator iron cores. To mitigate excessive core losses in these motors, current practices in axial flux motor stator core manufacturing involve wind-stamped steel tape [39], soft magnetic composite materials [40], or printed circuit board ironless stator technologies [41], which are notably more challenging compared to traditional radial flux motors.

Additionally, for radial flux motors, structures such as magnetic gear motors and other dual-rotor or dual-stator configurations also entail more intricate manufacturing processes, which deserve attention in mass production scenarios.

2.2. Design Specifications of IWMs

Based on the challenges mentioned above, an investigation of direct-drive IWMs from both the commercial and academic areas is conducted to illustrate their specific design requirements in this section. Due to its direct integration into the wheel rim and simple mechanical construction, outer-rotor topologies are more desirable for in-wheel traction applications. Additionally, the greater torque capacity brought by the larger rotor inertia and wider airgap diameter works to reduce torque pulsation and enable steady and smooth driving [42].

(a)

Dimensions

Both machine continuous and peak torque are largely proportional to the product of the machine axial length and square of the machine diameter, commonly denoted as D2l. The recent Protean/Brabus E-Class prototype vehicles, the Mercedes W212 E-class platform, can be purchased with a range of wheels from 16 in to 19 in as standard factory options [43]. The diameter of PD18 is 420 mm which is suitable for wheels of 18 in. Figure 4 showcases the outer diameter and axial length of commercial IWM products for reference.

(b)

Power, Torque, and Speed Range

There are both direct-drive systems and indirect drive systems utilized in EV IWMs. The first challenge to achieve direct-drive is to lift the total output torque of the IWM. The torque and power requirements for braking in direct-drive IWMs are much higher than that in motoring [43]. To illustrate this difference, the design requirements of a light vehicle with total weight of 1.5 ton is listed in

Table 1. Torque and power requirements.

	Pull Away at 30% Grade	Pull Away at 6% Grade	Braking at 1 g
Torque	1300 Nm	600 Nm	5000 Nm
Power	/	80 kW	700 kW

[43].

The maximal regenerative braking torque of a single IWM is around 80 Nm, which means that the regenerative braking torque from the four motors together can make the EGV reach a maximum deceleration of 1.2 m/s² [44].

According to the absence of a speed reducer, IWM systems can be classified into direct-drive systems and indirect-drive systems. The specifications of typical industrial products for these two systems are presented in

Table 2. Specifications of industrial indirect-drive IWM products.

Manufacture	Peak Torque (Nm)	Peak Power (kW)	Peak Speed, V (rpm)	Speed Ratio of Reducer, R	Speed Without Reducer V/R (rpm)
NTN	60	33	15,000	11	1363.6
Toyota	68	40	14,400	8.5	1694.1
ZF	44	40	21,600	16	1350
Schaeffler	132	31	4450	3.18	1399.4
YASA	390	320	8000	5	1600

and

Table 3. Specifications of industrial direct-drive IWM products.

Manufacture	Peak Torque (Nm)	Peak Power (kW)	Motor Weight (kg)	Torque Density (Nm/kg)	Power Density (kW/kg)
Shanghai Electric Drive	640	65	32	20.6	2.03
Protean	1250	75	36	34.7	2.08
Elpha	700	60	25.2	27.8	2.38
ENSRTOL ERMAX	240	80	12.3	19.5	6.50
GAC GROUP [45]	800	83	31	25.8	2.7
Elaphe M700 [46]	>700	75	23	>30.4	3.3
Elaphe M1100 [46]	1100	90	40	27.5	2.3
Elaphe L1500 [46]	1500	110	34.8	43.1	3.2

, respectively. It can be observed that the designed speed range of the direct-drive IWMs exhibits a lower speed range but higher torque requirements compared to indirect-drive systems.

(c)

Weight

Usually, the total unsprung mass should be less than one tenth of the sprung mass to ensure a smooth and steady driving experience. Despite the potential handling benefits from 4- or 2-wheel independent drive, it is a widely held belief that the addition of significant unsprung mass will have a major impact on the riding and handling behaviors of the vehicle. Whilst this belief is widely accepted, it has not been quantified to any large extent in published works [43]. Although the IWM increases the unsprung mass, this causes minimal steering and handling issues if the suspension system is designed suitably, while the removal of other components gives overall efficiency, weight, and complexity gains.

(d)

Thermal Design

A generally accepted rule of machine design is that the torque limit of a machine is constrained by its thermal limit eventually [13], but the limited size of a wheel raises the requirements of the thermal design [47]. The temperature level of the armature winding is usually higher, and the peak temperature usually appears on the end winding. The peak temperature of the end winding in [48] could be 229.49 °C during rapid acceleration operation, which distinctly surpasses the limit of the insulation level (155 °C), and the reliability of the IWM would be threatened in such a scenario. Hence, an advanced cooling system such as a spray cooling system should be employed in the thermal design considering the IWM’s drive cycles [48].

One of the most important aspects in the design of Protean’s IWM, and key to the success of achieving high continuous torque, is the thermal management of losses [33]. Since IWMs achieve high mechanical power outputs ranging above 110 kW, a large amount of heat is generated. With the objective of increased heat dissipation, active cooling is required for stator housing [49].

The advantages of a more compact integrated motor drive resulting from integrating the converter within the machine housing comes with certain challenges. The process of effectively integrating the relatively delicate converter within the machine housing is complicated by the intense vibration and shock loads the converter experiences in the enclosed space. The limited room within the enclosure means that the converter is placed in close proximity to the windings, necessitating a meticulous thermal analysis to ensure that the PE can endure the localized high temperatures. To address these issues, a liquid cooling system is implemented with both axial and orthogonal paths to facilitate heat transfer and enhance heat dissipation, cooling both the PE and windings concurrently [50].

In a related development, a spiral channel structure is employed in the study by [51], offering a theoretical benchmark for the research and development of heat dissipation systems for IWMs.

3. Classifications of IWMs

Given the diversity in the structures and principles of electric machines, the suitability of utilizing them as IWMs may vary. Thus, it is essential to categorize IWMs comprehensively to gain a macroscopic understanding of their distinct characteristics.

In terms of classification, IWMs can be classified based on (1) the direction of their airgap flux and (2) different motor topologies and principles.

3.1. Classification of IWMs Based on Airgap Flux Direction

Since there is relative movement between the stator and rotor in IWMs, an airgap must exist to realize electromechanical energy conversion [52]. The torque density of IWMs could be enhanced by increasing the total effective airgaps in the limited space of the wheel hub [53]. The motor configurations can be classified based on the differences in flux directions: radial flux (RF) motors, axial flux (AF) motors, and radial-axial flux (RAF) motors. Their schematic diagrams and three-dimensional (3D) views are illustrated in Figure 5.

(a)

Radial Flux IWMs

RF motors are the most common topology for IWMs. The Protean Electric PD18 IWM adopts a radial-flux outer rotor with thin yoke and large pole pair number, mounted in the compact powertrain system [54,55]. Currently, RF motors have a relatively complete theoretical model and mature manufacture techniques, which allows them to be a common choice for IWMs.

(b)

Axial Flux IWMs

The typical configuration of the simplest AF motor with a single rotor and single stator is demonstrated in Figure 5. However, this type of AF motor is not extensively used due to the significant attractive force between the rotor and stator [56], necessitating the presence of a thrust bearing and resulting in friction losses and wear [57]. To address these issues associated with AF PM machines, the coreless stator AF PM motor is considered a solution. Nevertheless, it suffers from drawbacks such as low flux density and low torque density, attributed to the coreless stator. Overcoming these limitations requires the use of a higher-speed reducer and a high-efficiency cooling method [3]. AF IWMs discussed in [58] adopt a coreless rotor and also need a gearbox to meet the torque requirements. The non-magnetic material used in AF PM motors is also used to achieve reduced mass and lower cogging effects [59,60].

Another method to reduce the axial attractive force between the rotor and stator is employing a double-sided or multi-disk AF configuration [57], of which the yokeless and segmented armature (YASA) PM motor is a typical instance. However, even in the case of YASA AF PM motors, tapered roller bearings or angular contact ball bearings are required to withstand the axial forces resulting from airgap offset, assembly errors, and motor asymmetry [59,61].

AF topology could gain higher torque in scenarios with small aspect ratios λ (La/Da) [3,62,63,64,65], which are ideal for IWMs. Also, with the elimination of thrust bearing, an AF PM motor with a sandwiched topology of double stators or double rotors is a much more feasible solution for IWMs. In fact, IWMs usually require a high pole number for smaller weight and a flat dimension (λ < 0.3); AF PM motors with double stators show better performances than RF motors [66].

(c)

Radial-Axial Flux IWMs

RAF motors have two axial parts to reduce axial force, which are also employed in [53,67,68]. The axial-radial configuration of one axial part and one radial part is also analyzed in [69], but this motor usually suffers from the high axial force and consequent severe bearing wear. An RAF motor with toroidal winding and an SMC core is used as an IWM in E-bikes [70]. The end-winding of RAF motors are usually reduced, which could improve the total efficiency of the motor.

The wedgy airgap motor is a special RAF motor which could be regarded as a superposition of both an AF motor and an RF motor. The Halbach array PM rotor is adopted in [71,72] to reduce total mass and lower torque ripple.

RAF motor has two axial parts and two radial parts, hence the end-winding of the stator is also exploited and the total airgap is maximized [73]. The output torque of this RAF motor is almost 2~3 times that of other motors [74]. Due to the much larger airgap region where the magnetic energy is stored, this motor’s torque density is greatly increased; however, this comes at the cost of a more complex construction and greater PM content [73].

3.2. Classification of IWMs Based on Machine Topology

Figure 6 presents a classification of IWMs based on their motor topologies and principles, showing that IWMs can be categorized into two main types: asynchronous machines and synchronous machines. Asynchronous machines primarily include induction machines (IMs) [75], while synchronous machines can be further subdivided based on the presence of PMs. Categories without PMs consist of switch reluctance machines (SRMs) [76] and synchronous reluctance machines (SynRMs) [77], whereas those with PMs can be additionally categorized into conventional permanent magnet synchronous machines (PMSMs) [78] and flux-modulated PM machines [79].

Within the realm of conventional PMSMs, there exists the surface-mounted PMSM (SPMSM) [80], interior PMSM (IPMSM) [81], consequent-pole PMSM (CPPMSM) [82], and PM-assisted SynRM (PMaSynRM) [77]. Furthermore, flux-modulated machines, an evolving category of electric machines, have garnered significant research attention in recent years. This class encompasses magnetic gear machines (MGMs) [83], PM vernier machines (PMVMs) [84], transverse flux PM machines (TFPMMs) [85], flux-switching PM machines (FSPMMs) [86], flux-reversal PM machines (FRPMMs) [87], and hybrid excited machines (HEMs) [88].

In Figure 6, the potential of various machine topologies in IWM applications has been studied in the literature. This paper will provide a detailed description in Section 4 of the potential IWM topologies and their applicability. Furthermore, it will meticulously analyze the torque characteristics of three representative structures, SPMSM, PMVM, and TFPMM, using the finite element method (FEM).

It should be noted that there is an overlap between the categories in the two classifications mentioned above, leading to the emergence of many potential IWM structures.

4. Overview and Comparative Study of Different IWM Topologies

4.1. Overview of Diverse IWM Topologies

Theoretically, IWMs could use all kinds of rotatory motors, regardless of the DC motor, induction motor, or synchronous motor used. But some kinds of motor are rarely used in IWMs currently, such as single-phase motors, DC motors, and traditional line-start synchronous motors. This section summarizes and analyzes the most representative motor structures used in direct-drive IWMs in industrial and academic settings.

4.1.1. Asynchronous IWMs

Induction machines (IMs) feature a robust rotor design that does not require the use of PM materials. Consequently, IMs offer advantages such as cost-effectiveness, the use of well-established technology, high reliability, easy maintenance, and a higher speed limit [89]. Additionally, IMs typically exhibit a relatively higher starting torque. Studies involving both experimental and numerical simulations have confirmed that the starting torque of IMs can be over 40% greater compared to PMSMs [90]. There are also many disadvantages of IMs, such as high energy consumption, the wires are complicated to revise, they have poor speed governing performance, and it is easy to damage the rotor bar. But the power factor of IMs is always inductive, since the equivalent circuits of IMs are composed of a resistor and an inductor, which is the same as synchronous reluctance machines. Usually, IWMs need a large pole pair number to reduce yoke thickness so as to lower the overall weight of the stator and rotor, but IMs need a small pole pair number to achieve good performance metrics such as higher torque [91].

For example, in commercialized EVs like the Tesla Model S, the IM has a pole pair number of only 2. Rotor copper loss in IMs is proportional to slip, typically ranging between 0.02 and 0.05, which can reduce the overall efficiency of the motor. Consequently, induction motors are not commonly chosen for in-wheel drive applications in EVs [92,93,94]. An example of an in-wheel induction motor with an outer rotor features a stator/rotor slot number of 108/124, as depicted in Figure 7 [93], with an efficiency of only 75% due to the high slot count.

4.1.2. Synchronous Non-PM IWMs

Similarly to SynRM, SRMs feature a rotor design that does not require the use of PM materials. This characteristic provides a distinctive advantage for IWMs due to the high level of robustness it offers. Thanks to the robust rotor design, SRMs also possess attributes such as low mass, high fault tolerance, and cost-effectiveness [12,95,96]. However, eccentricity in IWMs can lead to imbalanced magnetic forces, resulting in noise and potentially reducing the overall driving comfort [97].

Figure 8 shows one typical SRM design for the IWM investigated in [12]. The fundamental principle governing torque production in SRMs is akin to cogging torque, which can lead to higher torque ripple compared to machines driven by sinusoidal current sources. For instance, the average output torque of an SRM in the study by [98] is 61.4 Nm, while its torque ripple reaches 3.88 Nm, even after optimization. Therefore, careful consideration is necessary when designing IWMs that utilize SRMs, and a well-thought-out strategy must be adopted to enhance their performance from a control perspective [99].

Another drawback of SRMs is that the saturation of iron core must be analyzed in detail to obtain an optimal design solution [100].

4.1.3. Conventional PMSM Based IWMs

(a)

Fractional Slot Concentrated Winding PMSM-based IWMs

PMSM offer high efficiency compared to induction motors due to the elimination of field excitation losses, and they are easier to design for better heat dissipation [101]. Fractional slot concentrated winding (FSCW) PMSM, with its ability to accommodate a larger pole pair number, emerges as an ideal choice for IWMs [4]. Research by [102] highlights the advantages of FSCW PMSMs for in-wheel drive applications, including low torque ripple, high power density, wide speed range, and extended mileage, with the stator and rotor configurations depicted in Figure 9. As shown in Figure 9, the yoke is significantly thin because of the large pole pair number of FSCW PMSMs.

In practical IWM applications that demand high power density, a significant slot fill is necessary. However, the presence of insulation paper within the slots during manufacturing is essential, which can reduce the effective slot fill [16]. The numerical analysis in [16] shows that larger torque is achieved with more slots, but larger slot numbers result in smaller slots and lower slot fill [22], which lowers the torque density. Despite this paradox of slot fill and slot number, FSCW PMSMs remain an attractive option. Nevertheless, the drawbacks of FSCW PMSMs include increased core losses and PM losses at high speeds, possibly due to higher flux harmonics levels [103,104].

The reluctance torque of FSCW PMSMs is also much lower because of the unbalanced armature reaction flux passing through each rotor pole, and the ratio of reluctance torque to the total torque is smaller with smaller differences between the number of rotor poles and the number stator teeth [105,106]. Balancing the benefits of a small number of stator slots with enhancing reluctance torque in FSCW PM motors can be challenging. Strategic choices can significantly reduce the magnet volume in surface-mounted magnet rotors without compromising performance or risking demagnetization. For instance, replacing a surface-mounted motor with V-shaped magnets can save up to 44% of magnet mass and enhance demagnetization resistance [107].

(b)

Consequent Pole PMSM-based IWMs

Consequent-pole (CP) and dual-PM-excited topology also offer higher torque capability, on which many scholars have conducted multiple studies [108,109]. PM machines with consequent pole PM (CPPM) rotors could achieve similar continuous performance with only 67% of the PM usage of conventional PM rotors, which can lower the overall cost of the drive system significantly, but other problems such as magnetic unbalance and leakage flux also appear [110]. A SPMSM with a CP rotor for direct-drive IWMs is analyzed in [110], and the results show that this topology has low cogging torque, high field-weakening ability, and high efficiency.

The PMSM in [110] with a CP rotor has 24 stator slots and 20 rotor poles; its structure is shown in Figure 10, which has the same unit machine as the CP IWM in [111]. In addition to lower PM consumption, the inductance of the CP machine is also higher and thus the field-weakening capability is stronger [111].

CPPM rotors are frequently adopted in dual-PM machines, of which PMs are placed on both the rotor and stator so that the torque density can be increased by 20% [112].

4.1.4. Flux-Modulated PM IWMs

(a)

Magnetic Gear IWMs

Magnetic gearing is becoming attractive, since it offers the advantages of high efficiency, reduced acoustic noise, and being maintenance-free [113]. Due to the magnetic flux modulation of the stationary ring, the high-speed rotating field of the armature windings can be modulated to the low-speed rotating field of the PM outer rotor. Hence, self-decelerating is realized. The torque of the magnetic gear PM machine proposed in [114] (Figure 11) is 9.7% higher than that of the traditional SPMSM, but its torque ripple is larger due to the magnetic reluctance shifts in an electric cycle [114]. In addition, a novel MG machine consists of a flux-modulated MG machine and a SR machine, and both parts are integrated. This integrated MG machine shows the enhanced wide speed–torque profile [115].

The magnetic gear PM machine in [113] has the advantages of high efficiency, lower noise, and less maintenance, but the total PM consumption is large. The torque transmission experiences two stages, the IWM and magnetic gear, which could lower the total transmission efficiency.

(b)

Permanent Magnet Vernier IWMs

Based on the flux modulation principle, the PMVM drawing more and more interest due to its high output torque at low speed, which is considered to be a good choice for IWM drives [116,117,118,119,120].

Figure 12 shows a conventional configuration of the PMVM analyzed in [117]; high torque with low speed could be achieved due to its magnetic gearing principle. The pole pair number of the stator winding of a vernier motor is usually small, so the number of stator slots is consequently small, too.

The low power factor restricts the application of the PMVM [119]. Ref. [121] compares the PM magnetic flux distribution of PMVMs and traditional PMSMs with same-stator winding, of which the overall field has the same pole pairs. But the flux linkage of PMVMs is only one third of that of traditional PMSM motors, which is the main reason for the lower power factor for PMVMs [121].

(c)

Transverse Flux PM IWMs

TFPMMs also belong to a special kind of flux modulation machine [122], which also includes FSPMMs, FRPMMs, and PMVMs. The magnetic fields of TFMs are distributed in three dimensions, and the winding and the magnetic circuits are completely decoupled in structure. Their electric load and the magnetic load can be increased by adjusting the sectional area of the coil and the size of the magnetic circuit independently, so higher power density and higher torque density can be obtained [123], and a flat design with a small aspect ratio, such as an in-wheel motor, can be achieved [124]. TFPMMs are of greater importance due to their relatively high torque density and relatively simple structure, as well as short end-windings [18,125,126]. TFPMMs could have a large number of pole pairs without limited stator slots, thus achieving high torque at low speed [127,128], which is an ideal option for direct-drive IWMs [124].

But half of the winding of conventional TFPMMs serves as end winding, which causes copper loss, large flux leakage, and a low power factor [125]. The poor power factor is a major drawback of TFPMMs, but increases in torque usually incur a low power factor for TFPMMs [129,130]. The large number of individual components of TFPMMs also requires a specialized manufacturing process [50,131,132]. Special core material such as soft magnetic composite (SMC) is employed to manufacture TFPMMs because the magnetic paths of TFPMMs are three-dimensional. An SMC core could also reduce the core loss of TFPMMs, since its operating frequency is usually high [22]; however, it also has the disadvantages of low permeability and weak mechanical performance [133,134]. Huge numbers of new topologies about TFPMMs have also been considered and summarized in [124]. The TFPM machine in [135] adopts laminated stator teeth and an SMC stator core-back and SMC rotor pole pieces to obtain a better-rated torque and overload capability. Figure 13 shows the RF TFPMM configuration investigated in [85], which is only one-phase.

(d)

Flux-Switching Permanent-Magnet IWMs

FSPM machines usually have a robust rotor made of mere iron. A novel FSPM machine with a V-Shaped PM (Figure 14) in the stator and an outer-rotor designed for IWMs is analyzed in [136], which obtains high torque density and PM usage efficiency, with the disadvantages of high cogging torque and torque ripple. Also, FSPM machines have severe saturation problems due to PM placement at the stator squeezing the available space of the stator’s core [137]. The PM in the stator of FSPM motors could be replaced with DC excitation to enhance the field-weakening capability of IWMs [138].

The coefficient of PM utilization of FSPM machines is typically lower. A comparative analysis conducted in [139] examined a novel rotor-FSPM machine alongside stator-FSPM machines and a traditional Prius2004 Interior Permanent Magnet (IPM) machine. The coefficient of PM utilization of these three machines are 165.11 Nm/kg for the rotor-FSPM machine, 123.37 Nm/kg for the stator-FSPM machine, and 308.31 Nm/kg for the Prius2004 IPM machine. Also, a comparative analysis of a 50 kW (peak value) EV drive machine of the rotor-FSPM, stator-FSPM, and Prius2004-IPM machines shows that the amplitudes of flux density waveforms of representative positions in the magnetic path of the Prius2004-IPM machine is the lowest, which indicates the larger saturation and higher core loss of the FSPM machine.

Doubly Salient (DS) PM machines are also stator-PM motors. Two DS machines without PMs, namely the radial flux DSDC and axial flux DSDC, have been analyzed and quantitatively compared in [140]. The high torque density and low torque ripple value offered by the axial flux DSDC machine are particularly favorable for the in-wheel direct drive applications [141]. But the power factor is still low for the DS machine in [140].

(e)

Hybrid Excited IWMs

Recently, there is an increasing tendency to study HEMs, which combine the advantages of PM machines with the controllability of magnetic flux with auxiliary field windings [142]. The machine is essential for aerospace and EV applications which operate in a wide speed range.

But the auxiliary field-winding of HEM is usually located on the stator, which has a lower magnetic flux because of the poor excitation ability of the DC field coils and the worse extra DC saturation effect in the stator core. To address this issue, a new hybrid reluctance machine is proposed in [88,143,144], in which an integrated dual-layer PM source is introduced into stator slots, aiming to relieve DC saturation and, meanwhile, evoke the flux modulation effect. Figure 15 shows the configuration of HEM with the DC excitation proposed in [143]. Ref. [64] presents a novel axial-flux PM motor with adjustable field winding, which could have high field-weakening capability.

4.2. Comparative Study of Selected IWM Topologies

Usually, it is a challenging task to compare different machine topologies because each machine has a wide range of variables, so it is not easy to decide which variable to be kept constant when comparing, especially in cases where the design parameters are not specific [145]. One conventional solution is to keep a constant size based on Essen’s equation [91]. One helpful rule of machine design is that the torque limit of the AC machine is basically unrelated to the pole pair number of the rotor [91], but a large pole pair number could gain smaller weight as the yoke of both the stator and rotor could be thinner, which is ideal for applications in in-wheel motors. In this section, an FEA model of different motors is built and analyzed, based on the dimension constricts of Protean PD18 [54].

The design requirements of benchmark model are chosen as listed in

Table 4. Main design requirements of the selected IWM topologies.

Requirements	Rated Speed (rpm)	Peak Speed (rpm)	Rated Torque (Nm)	Peak Torque (Nm)
Value	480	1500	260	630

.

Based on the investigation in [24], the total mass of the benchmark model is chosen as 30 kg.

(a)

Fractional Slot Concentrated Winding PMSM-based IWM

FSCW PMSM is the most widely used choice for direct-drive in-wheel motors because of its strengths of short end-winding [146] and ease of manufacturing [147]. But the salient ratio of FSCW PMSM is relatively low [148], of which the reluctance torque comprise less than 15% of the peak torque in [107]. Choosing the proper slot and pole numbers is vital to the performance of a FSCW PMSM.

The LCM (least common multiple) of slot Z and pole 2P should be as large as possible to obtain a smaller cogging torque and torque ripple [149], and the winding factor could also be larger [150]. To maximize the LCM of Z and 2P, a slot and pole number combination of Z = 2P ± 1 is selected with the limitation of slot number and pole number, but this slot and pole combination could incur an UMF (unbalanced magnetic force) [150,151,152,153]. The combination of 51 slots and 46 poles is widely used in the direct-drive in-wheel motors of two-wheel electric bikes in China [150], and is applied in direct-drive in-wheel motors of EVs in [154]. Based on these studies, a FEA model of a FSCW PM motor with 51 slots and 46 poles is established to explore its performance, and the design parameters are shown in

Table 5. Main design parameters of the investigated FSCW PMSM.

	Parameters
	Rotor outer diameter (mm)	Rotor inner diameter (mm)	Core length (mm)	Airgap length (mm)	PM thickness (mm)
Value	420	390.44	115	0.776	4
	Rated voltage (V)	Rated Current (A)	Rated torque (Nm)	Rated speed (rpm)	Slot Fill
Value	150	20	290	800	0.43
	Turns per coil	Branches	Coil pitch	Core material	PM material
Value	15	1	1	M19-24G	N40UH

. The open-circuit flux density distribution map of the investigated FSCW PMSM is shown in Figure 16a, and the its torque characteristic related to current variation is shown in Figure 16b.

(b)

Permanent Magnet Vernier IWM

Based on the flux-modulation principle, the PMVM could acquire high torque at low speed [155,156,157], which is quite relevant to the application of direct-drive in-wheel motors.

The PM vernier split-tooth motor in [158] is used for direct-drive IWMs and exhibits higher torque than vernier motor with a single tooth. Split-tooth motors could shorten the pitch of each coil of vernier motor and reduce the copper loss of end-winding. Based on the topology in [158], a FEA model of a vernier motor was built to investigate its performance, and the structural parameters are presented in

Table 6. Main design parameters of the PMVM.

	Parameters
	Rotor outer diameter (mm)	Rotor inner diameter (mm)	Core length (mm)	Airgap length (mm)	PM thickness (mm)
Value	420	393.12	115	1	6.72
	Rated voltage (V)	Rated Current (A)	Rated torque (Nm)	Rated speed (rpm)	Slot Fill
Value	150	20	290	600	0.47
	Turns per coil	Branches	Coil pitch	Core material	PM material
Value	40	1	1	M19-24G	NdFe35

. The open-circuit flux density distribution map of the investigated PMVM is shown in Figure 17a, and the its torque characteristic related to current variation is shown in Figure 17b.

(c)

Transverse Flux PM IWM

Typical applications of the TFPMM are connected with the term ‘mobility’. They are considered as propulsion motor of ships, in aerospace applications, spacecraft application, in railway applications with rotating or linear principles, bus applications, and so on. This type of machine is considered to be in-wheel traction drive in electric vehicles [128,159,160,161,162,163]. The TFPM machine in [159] has rated torque of 366 Nm and a relative simple structure. The single-phase model, also known as decoupled model, in [159] is adopted because the difference between the coupled model and decoupled model is relatively negligible. The 3D model, open-circuit flux density distribution, and torque characteristics of the analyzed TFPMM are shown in Figure 18a, b, and c, respectively. The design parameters of this motor case are shown in

Table 7. Main design parameters of the TFPMM.

	Parameters
	Rotor outer diameter (mm)	Rotor inner diameter (mm)	Core length (mm)	Airgap length (mm)	PM thickness (mm)
Value	420	393.12	115	1	6.72
	Rated voltage (V)	Core material	PM material	Rated speed (rpm)	Rotor poles
Value	150	M19-24G	NdFe35	600	120

.

As shown in the torque curve in Figure 16, Figure 17 and Figure 18, FSCW PMSM exhibits larger torque potential because only slight saturation is observed. PMVM gains largest torque with double total PM volume and least torque ripple. TFPMM has a smaller output torque and larger torque ripple. Due to large flux leakage, TFPMM has severe saturation and lower overload capability.

5. Comprehensive Evaluations of the Different IWM Topologies

5.1. Comprehensive Evaluations

Based on the performance metrics of direct-drive IWMs for EVs listed in Section 3, this section comprehensively evaluates the applicability of various promising topologies in IWMs with performance radar plots. The corresponding evaluation results are presented in

Table 8. Performance evaluation with radial plots: different radial flux and axial flux IWM topologies.

RF-FSPMM	RF-PMSM	RF-CPPMSM
RF-PMVM	RF-SRM	TFPMM
RF-IM	RF-MGM	RF-PMaSynRM
AF-IM	AF-PMSM	AF-PMVM

.

It is important to note that this assessment includes the following criteria:

• Torque density is uniformly defined as the torque output per unit volume.
• Efficiency refers to motor efficiency and excludes the inverter section.
• Cost measurement encompasses both the materials and manufacturing.
• Reliability metrics include PM demagnetization risk, unbalanced magnetic pull and PM detachment risk, back electromotive force under faults, structural reliability, thermal constraint, torque ripple, etc.
• Ease of assembly is assessed based on existing manufacturing processes.

From the evaluation results, the widely utilized RF-PMSM topology exhibits a relatively balanced overall performance, which is a key reason for its widespread industrial adoption. In comparison, the RF-CPPMSM offers lower costs, but sacrifices torque density accordingly, making it more suitable for lightweight cost-effective EVs. A flux modulation motor topology, the RF-PMVM, demonstrates performance similar to RF-PMSM but with higher torque density, showing great potential in IWM applications. However, its lower power factor should be considered, as this would increase inverter capacity and raise system costs. In terms of torque density, the MGM also holds advantages, but its higher manufacturing costs and lower reliability due to the dual-layer PM rotor are limiting factors.

In contrast, other RF motor topologies exhibit poorer competitiveness for direct-drive IWM applications for specific reasons, including the following:

• Lower torque density and efficiency: Induction motors, switched reluctance motors, synchronous reluctance motors, and consequent-pole motors, despite their excellent reliability and relatively lower manufacturing costs, still lag in torque density, a crucial quality for IWMs. They may be more suitable for other application scenarios, like household appliances.
• Relatively complex manufacturing methods and lower cost-effectiveness: Transverse flux and magnetic gear machines, where industrial applications often prioritize ease of mass production, lack a competitive edge in this aspect, with other aspects also lacking strong persuasiveness.
• Lower power factor: This is prevalent in non-PMSM topologies, potentially exacerbating motor controller costs.

On the other hand, AF PM machines significantly feature higher torque densities, making them highly promising for direct-drive IWM applications. Despite current manufacturing limitations hindering their mass production, their superior torque densities indicate that they will likely become the primary type used in future IWM applications.

Furthermore, it is important to clarify that temperature rise and thermal management are also vital topics for in-wheel drive systems, given the challenges associated with heat dissipation within the wheel hub. However, due to the focus of this paper, we have not elaborated extensively on this aspect. Instead, as losses serve as the heat source within the hub, we hope to indirectly address the issue of heat using a comparative analysis of the losses of different motors.

5.2. Discussions

To further advance the development of IWMs and enhance their suitability for in-wheel drive EV applications, the following points merit attention:

• The flux modulation motor mechanism determines their high torque density. More forms of topologies warrant exploration, considering robustness, efficiency, power factors, and other performance metrics.
• Enhancing the manufacturing processes of AF motors and exploring motor topologies based on manufacturing constraints are crucial to fully leverage the inherent advantages of high torque density in AF motors. Additionally, the operational reliability of AF motors should be a focal point of consideration.
• In addition to motor topologies, new materials [164,165], heat dissipation methods [166], manufacturing processes (such as high-slot fill factor windings and stacking of ultra-thin silicon steel laminations [167]), superconducting windings [168], etc., can all contribute to enhancing motor efficiency, specific power, and torque density to meet the requirements of direct-drive in-wheel applications. This is worth further exploration and research.

6. Conclusions

This paper provides a general overview of the design considerations of direct-drive IWM in EVs. With the requirement of longer milage and better driving experience, direct-drive IWMs are the most potential technology for future EVs because of their high efficiency, larger chassis space, better controllability and high fault tolerance. But there are still challenges restricting the mass production of direct-drive IWMs.

Direct-drive IWMs systems must achieve high torque density and reliable thermal design, high total efficiency, and sufficient mechanical robustness within the limit of total cost and weight, which delivers high requirements for the design of IWMs. Massive novel topologies designed for IWMs have be proposed and researched in recent years.

Due to the limited volume inside the wheel, the topologies of IWMs could be classified as radial flux IWMs, axial flux IWMs, and axial-radial flux IWMs. In general, axial flux and radial-axial flux motors exhibit significantly higher torque densities, but they also come with markedly increased manufacturing complexity.

In terms of motor topology and mechanisms, FSPM, DSPM, RM, and IM motors have robust rotors, which are desired for IWMs, but other drawbacks such as lower torque density efficiency and power factor, which are yet to improve. MG machines could have reduced acoustic noise and less maintenance, but their PM consumption is large and their manufacturing cost and robustness is compromised. PMSM with CP could save the total usage of PM and decrease torque density, which is a favorable choice for lightweight IWMs. HEM could have better speed controllability due to the ability to adjust the excitation through DC excitation. However, their torque density still remains relatively low at the same current losses. FSCW PMSM is the most commonly used topology for direct-drive IWMs, which has the best balance between high torque density and low total weight. PMVM is a better choice compared to TFPM motors for IWMs. TFPMs could have low total weight, but relatively larger torque ripple and lower torque density.

The core advantages of IWMs (high efficiency, flexible control) position them as having tremendous potential in the wave of electrification. Currently, in-wheel drive systems are mature and widely used in two-wheel drive applications. However, there is still room for development in passenger vehicles, mainly due to challenges in unsprung mass, thermal management, and cost, which require ongoing technological iterations. The next 5–10 years will be a critical stage for in-wheel motors to transition from engineering validation to large-scale production.