1. Introduction
The rate of wind power installation around the globe is increasing exponentially, with the total worldwide installed capacity reaching 1017 GW at the end of 2023 [
1]. Wind power forms a key pillar in the energy decarbonization strategy of most countries aiming to reduce their total emissions in line with the 1.5 °C scenario agreed to in the Paris Agreement [
2]. To meet this target, the necessary installed capacity of combined onshore and offshore wind needs to be 3040 GW and 7820 GW by 2030 and 2050, respectively [
3]. Offshore wind is projected to provide an increasingly large proportion of the total energy supplied by wind, increasing from just 6% in 2018 to 21% and 35% in 2030 and 2050, respectively [
4,
5]. Over the last twenty years, the technology used in offshore wind has continued to mature, and the rate of research and development of this technology continues to increase. From 2002 to 2022, there was an average annual increase of 18% in the number of patents filed related to offshore wind power, reaching over 17,000 total applications in 2022 [
6].
One of the key components of an offshore wind turbine is the generator, which converts mechanical energy in the wind to electrical energy for the grid. Presently, the most attractive choice for generator technology in offshore wind turbines is the direct-drive synchronous permanent magnet (DD-SPM) machine [
7]. For ease of manufacture, these machines often employ surface-mounted PMs (SPMs), which offer a robust design and structural stability [
8]. This requires the use of high-energy magnets like NdFeB to produce a large air-gap flux density [
9]. An alternative topology would be the use of buried or interior PMs (IPMs) that can achieve high air-gap flux densities through flux focusing. However, the added complexity and weight of these designs make them unattractive for low-speed wind power applications [
10]. These machines also employ two converters driving two separate three-phase winding sets, as this increases the fault tolerance of the overall system; in the event of a converter failure, the turbine can continue to operate, albeit at a reduced capacity [
11,
12,
13]. These machines are preferable in offshore wind applications to the alternative doubly fed induction generator. These machines have gearboxes that add transmission losses in addition to increasing the failure rate of the turbine when compared to DD-SPM machines [
14,
15].
The windings in offshore wind power DD-SPM machines are traditionally overlapping windings with an integral number of slots per pole per phase (
spp) [
15]. This requires a distributed winding layout with end windings that overlap. The distributed winding structure yields high efficiency owing to the low rotor and PM eddy current losses [
16]. However, distributed windings are difficult to manufacture, especially considering the size and power rating required in modern wind power generators. Furthermore, these distributed windings overlap, which leads to large end-winding losses [
16]. The integer-slot machine structure also suffers from large cogging torque, which has to be mitigated in wind power generators by skewing the stator laminates [
17], thus sacrificing power. Machines equipped with fractional-slot concentrated windings (FSCWs) have gained increasing popularity in industry as an alternative to this integer-slot winding (ISW) and could soon replace the conventional ISW topology in wind power [
18]. Machines equipped with FSCWs have a fractional
spp number that is ½ lower such that each coil is wound about only a single tooth. This results in concentrated, or non-overlapping, windings. These machines are simpler to wind and so can have higher slot fill factors [
19], and the shorter end windings reduce the relative copper loss of this winding type with ISWs [
16]. These machines also do not suffer from large cogging torque, so skewing is not necessary [
20]. Furthermore, the concentrated windings reduce the mutual inductance between phases and so improve the fault tolerance of these machines [
21]; this is of greatest effect when the coils are wound about only alternate teeth (single layer) as opposed to every tooth (double layer) [
22]. However, these machines come with a substantial disadvantage in that the simple winding structure results in a large number of unwanted magneto-motive force (MMF) harmonics in the air-gap [
23]. These MMF harmonics are loss-causing and induce eddy currents primarily in the rotor and PMs [
24,
25]. The PM eddy current losses can lead to thermal demagnetization, exacerbated by heat in the rotor caused by the increased eddy current loss there. This is particularly problematic for offshore wind turbine generators, as large current and magnetic loadings are often required, so any added thermal demagnetization risk substantially impacts machine performance.
Work on the use of FSCWs in offshore wind turbines has been carried out before, including work on the design and performance analysis of a modular dual three-phase 3.3 MW offshore wind turbine generator with an FSCW [
26]. In this work, Xia et al. investigated different
spp numbers and their relative performance comparison when looking at onload torque, loss-causing armature MMF harmonics, and cogging torque. They found that an
spp number of 2/5 or where the slot-pole number follows the relationship
(where
Ns and
p are the slot number and pole pair number, respectively) results in an FSCW machine that is best for an offshore wind turbine generator. Furthermore, both these choices result in machines that can be operated with two converters as split-phase machines for unwanted MMF harmonic elimination. Xia et al. extended this work to show how the modular structure employed also allows for improved radial cooling through gaps in the stator segments [
27]. These works both used measurements from a full-scale 3.3 MW generator and so demonstrate the efficacy of the FSCW design at such a high-power rating. However, they did not compare the chosen topology with an ISW machine topology and so did not show how this winding structure has the potential to replace conventional machines. Work has been carried out on the comparison of FSCWs with ISWs for small-scale wind power applications rated to a few kWs. For 4 kW IPM machines, the FSCWs are found to perform better in terms of cogging torque, power density, torque ripple, and copper loss than similar ISW machines [
28]. A comparison of single-layer, double-layer, conventional overlapping, and toroidal windings for 15 kW machines for wind power application was carried out in [
29]. After consideration of active mass, rotor and PM eddy current losses, torque ripple, and manufacturing simplicity, the double-layer concentrated winding structure is the most favourable topology. Various works have investigated the impact of the
spp number on important factors such as demagnetization risk and vibrations in large-scale machines, suggesting that FSCW machines are more susceptible to demagnetization but suffer from fewer vibrations [
30,
31]. Various other works have included FSCWs in the design of novel wind turbine generators, including axial-flux [
32], double-stator [
33], and vernier machines [
34]. However, the robustness and simplicity of SPMs mean that these alternatives are unlikely to be competitive for offshore wind power applications. Two separate 20 MW wind turbine generators with distributed and concentrated windings are compared in [
35], but the authors do not compare the machines’ torque performance or the comparative power of these two machines.
Arguably, the most important metric for the comparison of wind turbine generators is the amount of energy that they can produce over their lifetime, which is directly related to the power generated by the machine during operation. Wind turbines are currently designed to last for over 25 years, and so even an incremental increase in the amount of power that the generator can produce would yield a substantial energy increase over the lifetime of the turbine. Therefore, in this paper, a key metric for machine comparison is the stator power, which combines both the torque performance and efficiency of the generator design. The other primary concern, particularly when investigating FSCW machines, is the rotor and PM eddy current losses, which increase the thermal demagnetization risk. There exists a wealth of recent work investigating the application of the hybrid star–delta winding connection to eliminate unwanted armature MMF harmonics and thereby reduce the rotor and PM eddy current losses in FSCW machines [
36,
37]. In particular, a dual three-phase machine that employs a hybrid star–delta winding connection in each winding set has been shown to exhibit excellent harmonic elimination properties [
38,
39,
40]. To the authors’ best knowledge, no investigations have been conducted into the application of star–delta windings in large-scale wind power generators equipped with FSCWs despite the substantial benefits they could offer to machine performance.
In this paper, a range of 3 MW FSCW machines have been designed and compared with a conventional 3 MW ISW machine. As a second converter is already the convention in offshore wind power, all the machines investigated at the 3 MW scale are also dual three-phase ones. It has been demonstrated extensively in the literature that single-layer FSCW machines suffer from much higher rotor and PM eddy current losses than double-layer FSCW machines [
22,
25]. As the principal aim of this work is to reduce the rotor and PM eddy current losses, FSCW machines with single-layer windings are not considered. The baseline ISW machine used for this investigation is a 480s/160p machine, and so a suitable starting FSCW machine was deemed to be 192s/160p. This maintains the pole number and, therefore, the electrical frequency and is also a multiple of the common 12s/10p machine that has been studied extensively in the literature. The machines proposed in [
38,
39] are rated to about 200 W, and the researchers investigated a dual star–delta connection in both 24s/10p and 24s/22p machines. The former yields a machine that no longer has a purely concentrated winding but exhibits excellent MMF harmonic cancellation properties, which substantially reduces the rotor and PM eddy current losses. The latter maintains a concentrated winding and can eliminate all non-torque-producing harmonics but still has notable rotor and PM eddy current losses. In this paper, these two machines are scaled to 3 MW and yield both a 384s/160p (multiple of 24s/10p) machine and a 192s/176p (multiple of 24s/22p) machine. These two slot-pole multiples are studied both as dual three-phase machines with a 30 elec. deg. phase shift between the converters, as well as novel dual star–delta winding machines with a 15 elec. deg. phase shift between the converters.
This work shows that the 192s/160p FSCW dual three-phase machine is not able to match the performance of the baseline ISW machine in terms of stator power or rotor and PM eddy current losses. However, the 384s/160p dual three-phase machine achieves 0.75% higher stator power than the ISW machine with equivalent rotor and PM eddy current losses. By utilizing star–delta windings in the 384s/160p machine, it can achieve a stator power 1.2% higher than that of the ISW machine, although with a notable increase in torque ripple. The 192s/176p dual three-phase machine maintains the advantage of concentrated windings whilst also being an improvement over the 192s/160p machine but still falls behind the ISW in terms of stator power. Again, the dual star–delta winding improves machine performance such that its torque capability is greater than that of the ISW machine, but it still suffers from large rotor and PM eddy current losses with reduced stator power.
2. Machine Winding Layouts
For this investigation into 3 MW FSCW offshore wind power generators, the generator parameters for a 3 MW ISW machine were used as a baseline [
15]. These baseline ISW parameters can be seen in
Table 1.
The common 12s/10p machine was selected as the initial FSCW machine for investigation at the 3 MW level. It is evident from the literature that a double-layer winding exhibits a much larger reduction in armature MMF harmonics and so reduced rotor loss when compared with a single-layer winding [
22]. As this is a principal concern with the move to an FSCW, a double-layer winding was chosen. All geometric properties in
Table 1 were kept the same for the FSCW comparison; the only modifications were those made to the circuit and are summarized in
Table 2. To maintain the same number of total turns per phase for the FSCW machine, the number of turns per coil had to be increased to 15. This meant that the number of parallel strings had to be reduced to 16, with 4 coils being in each parallel string. Finally, for a fair comparison of the onload performance it was decided to operate the machines with equal copper losses. One of the principal advantages of an FSCW is the reduced end-winding length, which will allow a higher current to be used while maintaining equal copper loss. In [
41], a method for the calculation of coil length for both distributed and concentrated windings is given as
with
where
L is the active length of the machine,
τs is the slot pitch,
Ns is the number of stator slots, and
Rs-mid is the radius of the middle of the stator slot. The copper losses arising from DC resistance can be calculated using
where
Ns is the number of slots,
Nc is the number of conductors per slot,
ρ (Ωm) is the resistivity of copper at room temperature,
Lw (m) is the sum of both the active length and end-winding length,
S (m
2) is the slot area,
kb is the slot packing factor, and
Irms (A) is the phase RMS current. The results of (1) for the 480s/160p and 192s/160p machines can be used in (3) to calculate the copper losses, and then the current of the 192s/160p can be adjusted until the copper losses are equal to those of the ISW machine. This yields a slightly larger rated current for the FSCW machine. The updated circuit properties for the 192s/160p dual three-phase machine can be seen in
Table 2.
Using the dimensions given in
Table 1 and
Table 2, the machine structure and winding layouts for the ISW and FSCW were generated and can be seen in
Figure 1.
The star–delta connection works by connecting the ends of the star coils to the terminal nodes of the delta-wound coils. The 30° phase shift between currents in star and delta coils mimics the behaviour of a dual three-phase machine and can be utilized for MMF harmonic elimination. The difference between the star and hybrid star–delta winding connections, as well as the voltage phasors for the star and delta winding sets, can be seen in
Figure 2.
In [
38], a novel dual three-phase machine was proposed that combined stator shifting and star–delta windings on a baseline 12s/10p machine to produce a 24s/10p machine. The stator shifting employed in the baseline 12-slot machine yielded a 24-slot machine, and applying the same method to the 192s/160p FSCW machine in this study yields a 384s/160p FSW machine. Much like the 192s/160p machine, this topology can use a second converter operating at a 30° elec. deg. phase shift to eliminate the first sub-harmonic and improve machine efficiency. As the number of slots has been doubled, so has the total number of coils, and so the number of turns per coil is reduced to 7.5. By moving to a coil pitch of 2, this machine adopts a semi-overlapping winding structure, and so (1) is used once again for the calculation of the updated end windings. However, this machine is not a fully distributed winding like the ISW machine. Principally, the 384s/160p machine has double-layer windings as opposed to the single-layer windings in the ISW machine. The key difference this has is that the centre of each coil is shifted by a quarter of the slot width on each side. This reduces the end-winding length by a factor of
and leads to the following equation for winding length in the semi-overlapping 384s/160p machine:
where
β is the slot–tooth ratio. By following the same process of calculating the rated current for equal copper loss, the rated current is found to be 163.29 A, which lies between the distributed winding and concentrated winding machines, as expected. The winding schematic for such a dual three-phase machine can be seen in
Figure 3a. In [
39], a second dual three-phase machine was proposed that used star–delta windings to eliminate unwanted harmonics while maintaining a coil pitch of 1. This 24s/22p machine was compared with a 24s/20p dual three-phase machine and was found to have improved electromagnetic performance. Scaling up the 24s/22p machine such that it matched the baseline 3 MW FSCW yielded a 192s/176p dual three-phase machine. Once again, this machine can operate its second converter at a 30 elec. deg. phase shift to eliminate unwanted harmonics and improve machine performance. Since this machine has the same number of slots and therefore coils as the 192s/160p machine, it also operates at the same rated current. The winding schematic for such a dual three-phase machine can be seen in
Figure 3b.
As detailed in [
39], 24s/10p and 24s/22p are both slot-pole multiples that are feasible solutions for a dual three-phase star–delta-wound machine. In each case, the star–delta windings can be used to artificially create a 30 elec. deg. phase shift between two winding sets in the machine. If star–delta windings are utilized in this way, then the second converter can instead operate a different set of windings at a 15 elec. deg. phase shift for even greater harmonic performance. In [
36], it is noted that the number of turns in the delta coils must be
more than the star coils to achieve an equivalent MMF in the air-gap. The winding layouts for the 384s/160p and 192s/176p dual three-phase machines with star–delta windings can be seen in
Figure 4 and
Figure 5, respectively. In each case, a schematic is given that shows which coils are wound in star or delta, and a second schematic that shows the interconnections between the delta windings for one of the three-phase sets.
The number of turns for these coils can be seen in
Table 3 for the 384s/160p and 192s/176p dual three-phase machines with star–delta windings. It is also explained in [
36] that the copper losses remain the same for star and delta coils as the total copper area has been kept constant within the slot. Therefore, these machines are operated at the same rated current as the previous FSW machines. The winding layouts for the 384s/160p and 192s/176p dual three-phase machines with star–delta windings can be seen in
Figure 4 and
Figure 5, respectively. In each case, a schematic is given that shows which coils are wound in star and which in delta, and a second schematic shows the interconnections between the delta windings for one of the three-phase sets.
In summary, the following machines have been selected for this investigation:
A 480s/160p dual three-phase ISW machine that serves as a baseline conventional distributed winding offshore wind turbine generator [
15].
A 192s/160p dual three-phase FSCW machine that serves as a baseline existing FSCW design for an offshore wind turbine generator [
26].
A 384s/160p dual three-phase FSW machine that shows how the baseline FSCW can be improved by stator shifting.
A 192s/176p dual three-phase FSCW machine that is an alternative slot-pole multiple to the baseline FSCW machine.
A 384s/160p dual three-phase FSW machine with star–delta windings that shows how the 384s/160p dual three-phase machine can be further improved with the star–delta connection.
A 192s/176p dual three-phase FSCW machine with star–delta windings that shows how the 192s/176p dual three-phase machine can be further improved with the star–delta connection.
5. Conclusions
This paper investigated the use of machines equipped with FSWs for use in offshore wind turbine generators. The main objective was to identify an FSCW machine that could produce a similar or even greater amount of power to a conventional ISW machine whilst also keeping rotor and PM eddy current losses to a minimum. A 480s/160p dual three-phase machine was used as an ISW baseline, and the common 12s/10p FSCW machine was multiplied to give a 192s/160p dual three-phase machine that served as an initial FSCW comparison. Techniques developed in previous work were then used to produce a stator shifted 384s/160p dual three-phase machine, a 384s/160p dual star–delta winding machine, a 192s/176p dual three-phase machine, and a 192s/176p dual star–delta winding machine. These machines were then compared extensively, including sections on air-gap flux density harmonics, open-circuit line EMF, torque performance, and efficiency and power.
Unfortunately, an FSCW machine was not identified that could match the stator power of the baseline ISW machine. However, the 192s/176p dual star–delta winding machine proposed in this work was found to have the highest stator power of all FSCW machines (only 2.6% less than the ISW machine and 2.4% more than the 192s/160p dual three-phase FSCW machine). These machines, of course, still benefit from the other key advantages of concentrated windings, such as the ease of manufacture and higher fault tolerance. Furthermore, the same slot filling factor was used for all machines investigated. Concentrated windings can achieve a higher slot filling factor than distributed windings, and future work should include this in the analysis, as it would greatly improve the comparative performance of the concentrated winding machines. Furthermore, if a laminated rotor core structure is used, in addition to conventional PM segmentation methods, the rotor and PM eddy current losses in these concentrated winding machines could be greatly reduced.
The two 384s/160p machines were able to produce more power than the baseline ISW machine with equivalent rotor and PM eddy current losses. The 384s/160p dual star–delta winding machine proposed in this work showed the best performance and was able to produce 1.2% more power than the presently employed ISW with only a 0.53% increase in the torque ripple coefficient. This work demonstrates that dual star–delta windings can be utilized to improve the efficacy of FSW machines in offshore wind turbine generators and therefore proposes two prospective machine topologies:
If annual energy production is the most critical consideration, then the 384s/160p dual star–delta winding machine is the best solution, with a 1.2% increase in stator power compared with the existing ISW baseline machine.
If a concentrated winding structure is desired, then the 192s/176p dual star–delta winding machine is the best solution, with only a 2.6% reduction in stator power compared to the existing ISW baseline machine.