Hairpin Windings: An Opportunity for Next-Generation E-Motors in Transportation

Advances in electrical machines and power electronics (PEs) are helping to achieve the power density and efficiency required by the transportation sector. However, the reliability of components and production processes is a challenge. This is especially true for electrical machines, whose winding processes are far from the high levels of automation, programmability, and repeatability that are required. This article looks into hairpin windings and outlines a number of future actions to address challenges and eventually enable the complete penetration of hairpin windings in transportation.


dvances in electrical machines
and power electronics (PEs) are helping to achieve the power density and efficiency required by the transportation sector. However, the reliability of components and production processes is a challenge. This is especially true for electrical machines, whose winding processes are far from the high levels of automation, programmability, and repeatability that are required. This article looks into hairpin windings and outlines a number of future actions to address challenges and eventually enable the complete penetration of hairpin windings in transportation.

Objectives
In the coming years, mobility will rely ever more on electrification [1].
Performance is very sensitive to vehicle components' mass. Hence, maximizing power and torque densities represents a key objective when designing electric drives for transport applications. In this context, electric motors (EMs) and PEs play a key role. On the one hand, at the PE level, wide-bandgap (WBG) devices have paved the way for lighter and more efficient power converters [2]. On the other hand, at the EM level, increased power densities are enabled by innovative cooling systems [3], new materials [4], and form-wound technologies, such as hairpin windings [5]. Nevertheless, such solutions often result in increased losses in cores and windings, with limited flexibility to suit transportation production requirements. Hence, new winding concepts and flexible and automatized manufacturing processes are required to mitigate these challenges, as EMs intended for transport applications require compact windings with a high slot fill factor, low losses, high reliability, and so on. This article elaborates on the topics discussed in [5], first reviewing hairpin technologies, with focus on the fill factor as a key enabler for power density, and then discussing their impact on ac losses, voltage distribution, thermal management, and manufacturing processes. Compared to [5], this work provides our vision beyond the state of the art to eventually provide solutions and consider power density, reliability, and automation requirements in a comprehensive manner.

Background: Slot Fill Factor
The choice of an appropriate winding layout represents a key design phase of electrical machines. In general, there does not exist a single, optimal winding structure; rather, it depends on the application [6]. The most common winding type is the randomwound one, consisting of insulated, round conductors that are wound inside stator or rotor slots to form a coil. The active sides of any coil turn are placed "randomly" against one another [7]. One of the main advantages of such a winding is flexibility. Random-wound windings range from single-layer to multiple-layer structures, from integer to fractional slot windings, from full-to short-pitched layouts, and from distributed to concentrated configurations [8]- [10]. However, random-wound windings generally feature a low slot fill factor.
The latter is an indicator of the amount of conductive material located in a slot. Therefore, for a given slot area, maximizing it would permit increasing the number of conductors per slot for a given wire cross section or increasing the wire cross section (and thus the electric loading) for a given number of conductors per slot. Hence, it is straightforward to envision such a parameter as a key enabler to increase power density in EMs. A number of techniques have been proposed to increase the fill factor [11]- [13]. However, these are often limited to concentrated windings; the process repeatability is compromised, the risk of damaging the insulation is high, and the production cost is usually elevated [14].
Form-wound windings, such as hairpins, feature higher fill factors than round-wound ones [5]. This winding type is constructed from preformed elements made of enameled, flat wires. Their rectangular shape inherently fits that of a stator slot with parallel sides and reduces the gaps between conductors. Figure 1 reviews fill factor enhancement methods, highlighting how hairpin windings may be the way forward.

Hairpin Manufacturing: State of the Art
From the manufacturing point of view, hairpin windings enable automatic production at a large scale, which is needed to sustain the green transportation revolution. Preliminary investigations have shown that, considering 1 million units per year, hairpin windings' production costs can be reduced up to 27% compared to random windings [15]. The production of stators equipped with hairpin windings is divided into a number of steps, as illustrated in Figure 2. The four major steps are reviewed in the following.

Shaping
A straight conductor is stripped and cut to a suitable length, then bent into a "hairpin." The specific shape has to fit the winding diagram requirements and overall dimensions of the stator at hand. There are basically two consequent major bends, aimed at obtaining a U shape and shifting one of the two legs according to the coil pitch and position in the slot. This process has to be performed accurately and with low tolerances to facilitate insertion and avoid stresses on the conductor and its insulation.

Arranging
After elementary hairpins have been made, they are inserted in the stator slots. This is usually done manually, and the ease of insertion is an indicator of the quality of the shaping phase.

Twisting
A gear element twists the free ends of the hairpins according to the coil pitch and places them in the most suitable position for the contacting step. Twisting is as sensitive as shaping in terms of conductor and insulation stress.

Contacting
Depending on the total stator slots and layers, the number of contacting spots is significantly higher than in a conventional random-wound winding. Hence, reliable contacting technologies with short cycle times are required. To establish an electrical connection, the stripping operation (see Figure 2) is performed first, consisting of removing the insulating material from the conductors. A fatigue-endurable mechanical connection and low contact resistances have to be ensured [16]. Laser welding seems to be the most promising solution for the contacting operation [17], as the materials involved receive little damage. Additionally, from  a mass production perspective, laser welding ensures a low cycle time, and the process is easily automatable.

AC Losses
Currently, the application of hairpin windings in EMs for transportation is mainly limited by inherent challenges in high-frequency operations, where skin and proximity effects take place. These phenomena make the total current distribution inside the conductors uneven, and consequently, the effective cross section through which the current flows is a fraction of the conductor's whole cross section. In round conductors, the solutions to mitigate these effects include the use of stranded conductors [6], Litz wires [18] and twisting/ transposition methods [19]. The disadvantage of stranded and twisted conductors lies in the high connecting effort related to the parallel strands and the reduction of the allowed conductor stress during the winding process [20]. The fill factor is also compromised.
In hairpins, the stranded conductor concept cannot be applied as flexibly as in round conductors, due to the manufacturing limitations inherent to such technology. To reduce losses in hairpin windings, methods such as removing the closest conductor to the slot opening and reducing the number of conductors and their height have been described. However, the first solution lowers the fill factor since part of the slot is left empty [21], whereas the second option increases manufacturing complexity, as more conductors need to be bent and welded [22].

Thermal Considerations
As reported in the previous section, the losses produced in hairpin windings during high-frequency operations can be significant and limit applications within a certain frequency range, i.e., typically below 1 kHz. On the other hand, the cooling of the conductor can be optimized when hairpins are used. In fact, the area of the conductor directly in contact with the lamination is much higher than that of round wires (i.e., theoretically, a single spot), thus increasing the heat exchange between the winding and lamination significantly. Additionally, the thermal conductivity grows within the slot since the amount of insulating material is reduced compared to random-wound windings [23]- [25].
In [23], this concept was experimentally proved by comparing the thermal performance of a hairpin winding machine against a random-winding one, both designed for the Chevrolet Voltec electric propulsion system. In [24], it was shown that the hairpin motor designed for the Chevrolet Bolt battery-electric vehicle achieves a peak efficiency of 97%, whereas in [25], a 96.5% peak efficiency was obtained even though the studied hairpin machine had a wound field rotor, which typically results in less efficiency. A review of various solutions based on actual vehicles is presented in [26], where torque and efficiency are investigated when moving from random to hairpin designs. Here, the main conclusion is that, if the same slot area (not the copper area) is maintained, the reduction of dc copper losses can overcome the increase in ac losses, resulting in improved performance.

Voltage Distribution Within Windings
WBG-based converters are enabling higher operating frequencies for machines. However, their inherently faster commutations and voltage gradients [the instantaneous rate of voltage change through time (dv/dt)] may trigger voltage overshoots, uneven voltage distributions, partial discharges (PDs), and faster insulation degradation [27], [28]. With a pure sinusoidal supply voltage, the voltage distribution is even; i.e., the potential difference between two consecutive turns in a coil is the applied voltage divided by the number of turns. Using inverters, the turn-to-turn voltage can be as much as 90% of the applied voltage, considering rise times of 50 ns [29].
In this context, random-and formwound windings may experience different levels of stress. In a random-wound stator, the turns near the motor terminals could be positioned against turns near the winding neutral point. This means that the insulation system should be sized to withstand the maximum supply  voltage of the motor. On the contrary, in form-wound windings, the insertion process is carefully carried out, ensuring that two adjacent turns are located against each other in the slots, thus resulting in the smallest possible voltage difference. The impact of a higher dv/ dt on the reliability of EMs is twofold, depending whether the turn-to-turn or turn-to-ground voltages incept PDs within the insulation system [30] or not. In the former case, breakdown can occur in a few hours. Below PD inception, large slew rates speed up intrinsic aging [31].

Progress Beyond the State of the Art
As described in the previous sections, several challenges need to be overcome to enable the widespread use of hairpin windings in transportation. To address the difficulties, the scientific community is pushing toward innovative winding geometries, updated analysis tools, new cooling concepts, accurate lifetime models, and novel manufacturing lines permitting reliable, flexible, and fully programmable coil forming setups. In the following, our vision relative to these developments is reported.

Programmable, Flexible, and Reliable Coil Forming and Joining Setup
Within the Clean Sky 2 project Automated Manufacturing of Wound Components for Next-Generation Electrical Machines (AUTO-MEA) [32], we see a number of technological improvements in manufacturing lines for hairpin stators. These include three main areas/stations.

Forming and Shaping
The primary units of this station need to be updated to achieve flexibility and process reliability, as summarized in the following: ■ The first unit is the stripping device, with laser stripping being the most suitable solution to ensure automated, reliable, and accurate removal of the enamel. ■ Next is the coining and cutting unit, where the conductors, once stripped, are segmented and cut. To facilitate insertion and welding, the conductors' ends are coined into chamfers, then cut by plastic deformation and separated by a clamp. This reduces the probability of insulation damage, enhancing the reliability of the process. ■ Finally, there is the bending unit, where the conductors are hosted to obtain the desired 3D shape. A drawback of this process relates to flexibility; i.e., several tools need to be reconfigured when manufacturing different stator designs. Two degrees of flexibility will be considered in the near future: hardware and software.
The bending device will be equipped with interchangeable accessories and tools (hardware flexibility) to accommodate conductors of different geometries and sizes. Additionally, updated control algorithms will permit realizing a certain number of different geometries per tool (software flexibility).

Insertion and Twisting
Future research will focus on providing tools for accurate and automatic insertion and twisting of the windings. In particular, within AUTO-MEA, a semiautomatic, three-axis twisting machine will be developed. The adoption of specific tools will reduce the possibility of insulation damage during this phase.

Welding
As described, laser technologies seem to be the best candidates for welding. However, contacting processes can be improved by further investigating aspects such as the resilience of the melting pot, physical dimension increase of the wire cross section, and repeatability of the joining procedure. For the sake of completeness, the graphical concepts of the bending unit, coil insertion station, and welding station are presented in Figure 3 New Winding Concepts Alternative winding concepts will be envisioned in the upcoming years to reduce ac losses and thus increase the speed/frequency current limits of hairpin windings. Some examples have been already reported in [5].

Hairpins With Variable Cross Sections
The basic idea consists of placing conductors with increasing cross sections along the slot radial direction, as reported in Figure 4(a), for a four-layer layout. This solution is already under study [33], and the results in terms of loss reduction are promising. However, the proposed  variable cross-section concept must be applied to half the hairpins located in a slot to minimize the number of elementary hairpins needed. The flexible forming and joining setup proposed in the preceding matches this concept perfectly, enabling automatic tooling reconfiguration.

Hairpins With Segmented Cross Sections
This method consists of "segmenting" the hairpins near the slot opening in subconductors, as done for the final two conductors in Figure 4(b). Losses drop significantly with such a technique [33]. However, a suitable transposition is required. Also, particular attention should be given to ensure reliable bending, twisting, and welding processes. The relevant complexities will depend on the number of segmented (SEG) conductors, which will be somewhat limited. Therefore, ways to mitigate this challenge should be investigated.
As a preliminary validation of SEG hairpins, motorettes emulating stator portions of a realistic traction application [34] have been built. These are shown in Figure 5(a), where four configurations can be observed, with q being the number of slots per pole per phase and k the number of layers per slot. One of these motorettes implements the SEG concept in Figure 4(b). In Figure 5(b), there are two motorettes equipping a random winding; their full details can be found in [34]. The experimental comparison between hairpin and random windings in terms of the dc-to-ac resistance ratio Kac versus frequency is plotted in Figure 5(c). For the random winding, the minimum and maximum Kac values are found by considering the 66% probability of having a value inside a certain range between a mean value and a standard deviation. Most importantly, Figure 5(c) highlights that the conventional eight-layer (Q4, eight layer) and SEG (Q4, four-layer SEG) hairpin layouts achieve lower losses than the 66% of the possible solutions that can occur in random windings up to 900/1,000 Hz. These results, besides proving that some hairpin layouts (including the SEG one) can lead to lower losses than random windings, even at high frequency operations, demonstrate that the segmented layout can compete against the eight-layer hairpin configuration.

Hairpins With a Reinforced Insulation Layer
Compared to random-wound ones, hairpin windings are less sensitive to the elevated dv/dt generated by the WBG converters feeding them and to the ensuing uneven voltage distribution. However, the first turn can be somewhat impacted by such phenomena; thus, it could be equipped with a reinforced insulation layer to make the voltage distribution more uniform.
This will not have a significant impact on manufacturing.

Continuous Hairpins
Continuous hairpin windings consist of preformed elements constituting more than one coil. This reduces the number of welding points. The cycle time is also shortened, and the winding process is easier compared to conventional hairpins. However, open slot structures or novel stator layouts are necessary to enable continuous insertion inside the slots. Such concepts come at the cost of introducing new electromagnetic and manufacturing challenges. A comparative summary highlighting the benefits and drawbacks of classical (i.e., random windings and standard hairpins) and new winding concepts for nextgeneration EMs is provided in Table 1, considering an equivalent copper area for the winding and conductor types.

Thermal Management
As reported in the "Thermal Considerations" section, improved thermal management is achieved using hairpins. In [35], hollow conductors are proposed to mitigate copper losses, as these enable axial coolants to flow through the conductor itself, although the fill factor is compromised. Nevertheless, thanks to better cooling, a higher current density is achieved but with lower efficiency. Further research is expected in this direction, aiming to find the best design tradeoff that maximizes both torque capability and efficiency. Another interesting research path relates to the machine topology to combine with hairpin technologies since it seems to play a role, as demonstrated in [26]. Additionally, future studies will need to focus on the cooling of the end windings. In light of this, suitable end winding shapes will be designed in such a way to host dedicated liquidcarrying pipes for direct cooling. This is a more cost-effective methodology compared to spray cooling and flooded stator technologies that are typically adopted today and do not suffer from reliability and robustness issues.

AC Loss Prediction
To analyze the new winding concepts, the classical analytical model for ac loss evaluation [6] can no longer be used. In [33], techniques to update the model have been proposed. Future studies will provide an experimental validation of those concepts.

Stress-Strain Evaluation
One of the technological limits of preformed elements is due to the stress that insulating and conductive materials experience during bending and twisting. This results in predefined   width-to-height ratios [36] for hairpins, which represents a bottleneck.
To enable more flexible designs, detailed analyses through advanced numerical structural tools are suggested to evaluate the maximum deformations. Besides easing the stress on insulating and conductive materials, these studies will potentially permit reducing the number of bends, shortening the end winding lengths, and creating innovative winding shapes for better thermal management.

Lifetime Estimation
Prioritizing reliability as a design objective for EMs equipping hairpin windings will result in significant progress. This approach was proposed in [37] for low-voltage, random-wound machines. Based on [27], a comprehensive model to predict the voltage distribution within hairpin windings will be developed, where the converter, cable, and motor will be taken into account. The model will target electrical stress within the winding and be tailored to the winding structures envisioned in the "New Winding Concepts" section. Figure 6 is a block scheme of the model, where the system geometrical and physical characteristics along with all the electrical parameters represent inputs of a computational algorithm that will define the estimated lifetime of the windings against electrical stress and loss distribution. The model, working in conjunction with the flexible coil forming and joining setup, will provide a powerful tool to assess the behavior and identify the weak points of windings in transport applications. These studies will enhance the reliability of electric drives and short the design process of new products since system weaknesses will be identified during early stages.

Conclusion
Besides aiming at maximizing efficiency and power density, next-generation EMs for transport applications should be oriented toward an easier, faster, and more reliable manufacturing process. While sophisticated thermal management techniques, innovative cooling systems, and new materials are enabling increased efficiency and power density levels, EM manufacturing represents a major bottleneck, and a paradigm shift should soon occur to cope with the high levels of automation, capacity, and repeatability required by the transportation sector. In this context, hairpin windings seem to be a promising technology, although they present several challenges. To address those issues, we propose a number of actions, mainly consisting of developing fully programmable, flexible, and reliable forming and joining machinery; envisioning innovative winding and core layouts; and implementing new electromagnetic, thermal, and structural modeling methodologies as well as reliability and lifetime models.