Performance analysis of aluminium wound double fed induction generator for cost-efficient wind energy conversion systems

Currently, both limited fossil fuel resources and environmental factors have increased the use of renewable energy sources. Renewable energy resources, such as wind energy systems, are gaining popularity, resulting in increased competition among manufacturers. This study aims to achieve a cost-efficient wind energy conversion system by designing and analysing the performance of a 250 kVA aluminium wound double-fed induction generator (DFIG). The advantages and disadvantages of aluminium windings are compared with those of copper windings, and three DFIG models are created: Model-1 with a copper winding set, Model-2 with the same geometry as Model-1 but designed with an aluminium winding set, and Model-3 with an aluminium winding set and slightly different stator and rotor diameters. The three DFIG models were analysed using finite element analysis (FEA) in ANSYS Maxwell, and the simulation results were obtained. According to the FEA results, Model-1 with copper windings had a higher efficiency than Model-2 with aluminium windings, but Model-2 had better cost and weight performance than Model-1. Model-3 and Model-1 had similar efficiencies, but Model-3 had a slightly greater torque ripple compared to Model-1 because of a slightly different stator and rotor diameter. Although the total machine weight of the aluminium-wound DFIGs was slightly increased, the total manufacturing costs were less than those of the copper-wound DFIGs at the same efficiency levels.


Introduction
Energy demand is on the rise, leading to a need for more efficient energy production or utilization of energy resources.Renewable energy resources are increasingly important due to the limited availability of fossil fuels and their negative environmental effects.The manufacturing and installation costs of wind energy conversion systems (WECSs) are high, but they have significant potential in renewable energy production.The global wind electricity market is projected to grow from $89.66 billion in 2021 to $104.18 billion in 2022, with a compound annual growth rate (CAGR) of 16.2% [1].Researchers in academic and industrial fields are working to reduce these costs by utilizing materials with low density and high mechanical strength, such as those used in the tower, blade, and nacelle parts of wind turbines, to improve their weight and performance.These efforts are being accelerated by advances in material science and technology.
Comparable studies have also been conducted on wind generators, which are a crucial component of WECS.To reduce costs and weight, new aluminium-based winding structures have been designed as an alternative to copper winding structures [2][3][4].Aluminium-based winding structures are commonly used in distribution systems and high-frequency transformers.When compared at the same power factor parameter, a 2% lower efficiency was observed when using aluminium winding instead of copper winding on distribution transformers.However, for the same current density, aluminium winding is lighter than copper winding [5][6][7][8][9].At frequencies above 500 Hz, the ratio of AC/DC resistance for aluminium winding approaches the resistance ratio of copper winding [10].In addition, the use of multi-stranded winding structures reduces eddy current effects compared to solid-wound winding structures [11].
Generally, aluminium winding structures have been traditionally preferred for designing the rotors of squirrel cage induction motors.However, in recent years, studies have been conducted on the use of aluminium winding for different types of dynamic electric machines in various applications [12][13][14][15][16].A design of an aluminium-wound switched reluctance motor for the traction system in automotive applications has been addressed.To reduce winding losses and achieve a higher winding fill factor, a multi-strand concentrated winding structure is used instead of a solid winding structure.When considering total winding losses, aluminium winding has a 7% higher winding resistance compared to copper [14].In [15], the performances of two permanent magnet synchronous motors that had the same core geometry but different winding structures: one with aluminium winding and the other with copper winding has been investigated for automotive applications.The study found that the aluminium wound PMSM had 4.15% better efficiency performance than the copper wound PMSM, for the same torque and speed values.However, for the same winding loss, the aluminium wound PMSM showed 18% worse output torque performance compared to the copper wound PMSM.In terms of weight, the aluminium wound PMSM was 14.7% lighter than the copper wound PMSM.Moreover, the manufacturing cost of the aluminium winding structured PMSM was 67% lower than that of the copper winding structured PMSM [15].A similar study has been conducted on PMSM for direct drive elevator systems [16].The aluminium-wound PMSM, although 9% larger in volume compared to the conventional copper-wound PMSM, had a reduced total weight due to the lighter winding, and it was 12.33% cheaper to manufacture.
In the context of wind energy conversion systems, limited studies have been conducted on the materials used in generator windings.In [17] and [18], a wound-rotor synchronous machine (WRSM) with a power capacity of 300 kVA was designed with aluminium windings, and performance analysis was performed.By only changing the winding conductor material, the weight was reduced by nearly 12% compared to the traditional copper wound WRSM.However, the study also revealed that the copper winding structure exhibited 2% to 3% better efficiency performance compared to the aluminium winding structure [17].In addition, the aluminium winding structure was analysed in detail in terms of various aluminium alloys, and a comparison was made.The comparison results showed that the 1350, 2041, and 2017 types of aluminium alloys exhibited better efficiency performance than the conventional materials used in electrical machines [18].
Recently, there has been a growing interest in using carbon nanotube (CNT) fibers as winding materials in electrical machines due to advances in nanotechnology.In a study conducted on a permanent magnet synchronous motor (PMSM), a performance analysis was performed on a CNT-wound winding structure.The results showed that the slot area of the designed PMSM increased by 50%, while the thermal capacity was 40% better than that of a conventional copper wound PMSM.Additionally, the CNT-wound PMSM was six times lighter than the copper-wound PMSM.However, manufacturing CNT conductor material is a significant challenge, as isolated enamelled manufacturing of CNT fiber conductors is difficult, and the manufacturing costs of CNT fibers are higher than those of aluminium and copper-based conductors [19].
Considering the current manufacturing conditions, using aluminium-based winding conductors is a costeffective and readily available alternative to copper and CNT-based conductors [19].Literature suggests that aluminium is used as a winding material in transformers, induction motors, synchronous motors, and permanent magnet motors, particularly in research and development operations.In this study, a performance comparison between aluminium and copper winding structures was performed on a double-fed induction generator (DFIG), which is commonly used in wind energy conversion systems.The study involved designing 250 kVA aluminium and copper wound DFIGs and comparing different performance parameters such as manufacturing cost, total weight, and efficiency.The second section of the study provides a detailed explanation of wind energy conversion systems, while the third section delves into the aluminium winding structure.The results of the various analyses are explained in the last section.By using optimum winding parameters, comparable winding performance can be achieved on electrical machines while minimizing winding losses.

Wind turbine systems
Energy production systems that do not rely on fuel costs are essential to reduce the unit cost of electrical energy.Therefore, in areas with high wind potential, wind power is the most viable option for energy generation.Various types of wind turbines are available for different loads and grid connections, ranging from a few kilowatts to hundreds of megawatts in energy production scale.A wind generator is an electromechanical machine that helps the WECS convert kinetic energy to electrical energy.The main components of the WECS are electrical, control system, and mechanical components.Figure 1 shows the main components of the WECS [20].In figure 1, wind turbine rotor hub, blades and gearbox used to convert the low speed of the wind turbine to the high speed of the electric generator shaft.These components are known as mechanical components of the WECS.The wind turbine control cabinet and wind turbine converter components are also known as control units of the WECS.The main electromechanical component of the WECS is the electric generator that converts the wind energy to electrical energy [21,22].
WECSs are also categorized by implementation type.The main implementation areas of the WECSs are onshore and offshore regions.However, in these regions, WECSs are classified and constructed on-grid or offgrid connection types.Off-grid WECS connections are an alternative for customers who live in regions with expensive classical grid extension costs.WECSs are used together or integrated with other energy conversion systems because of variations in wind speed, wind generator output power, and customer load demand.Connections are implemented between the grid and WECS at different voltage levels for on-grid connected WECS [23].
One of the difficulties for operators who are working to provide a steady state connection between the WECSs, and grid is uncontrolled wind speed.In principle, WECSs are categorized according to generator speed into two main categories.One of these, fixed speed WECS, the other one, variable speed WECSs.Fixed speed WECSs are leading in the wind energy industry due to their low-cost manufacture and maintenance expenses and their highly reliable technology.
However, there are still challenges to efficient energy production with wind turbines.One of the most significant problems with wind turbines is their variable efficiency, which can be addressed by adjusting the generator speed with variable wind speed.This helps to reduce mechanical loads and improve generator efficiency [24].
The most common electric generator types used in wind energy conversion systems are squirrel cage induction generators (SCIG), wound rotor induction generators (WRIG), double fed induction generators (DFIG), and permanent magnet synchronous generators (PMSG).Identically, wind generators are designed, manufactured, and optimized to generate maximum electrical power for nominal wind speed conditions.For high wind speeds, safety systems are activated to protect the wind energy conversion system.The most commonly preferred wind generator type is a double fed induction generator [25].
Double feed induction generators (DFIG) have become a popular choice for variable speed wind turbines in recent years, thanks to their ability to operate within the ±30% speed range of synchronous speed and the fact that the power electronics converter power used is only about 25% of the generator power.Compared with conventional squirrel cage induction generators, DFIGs offer higher energy and power density, as well as greater control over energy and power flow.However, wind turbine output power must be fixed at the same frequency as the connected energy distribution grid frequency regardless of wind speed variations.Therefore, DFIGs are typically used in energy generation systems that require variable operating speeds and fixed grid frequency.
Various control techniques are available for managing DFIGs connected to the grid, including direct power control, direct voltage control, vector control, and fuzzy logic control.These techniques enable efficient management of the DFIG's power output and ensure optimal performance [26].
The grid connection of a DFIG is depicted in figure 2, where the stator windings are directly connected to the AC grid.This connection provides a constant voltage and frequency to the generator, as well as the necessary magnetic flux for energy production.The rotor windings, on the other hand, are connected to the AC grid via a back-to-back converter.This converter consists of two voltage-controlled converters, one located on the grid side and the other on the rotor side, which control the energy flow.The grid-side converter controls the energy flow to the grid, while the rotor-side converter controls the variable energy from the rotor windings, feeding the DC line.

Design of the double fed induction generator
In wind turbines, it is not possible to convert all the kinetic energy of the wind into mechanical energy.According to the Betz coefficient, theoretically, only 60% of the wind kinetic energy can be converted into mechanical energy [27].The wind power that can be obtained in a wind energy system can be calculated using equation (1), where Pm is the mechanical power transferred to the generator shaft from wind, ρ is the air density, A is swept area of turbine blades, υ is wind velocity, and C p is performance coefficient of turbine blades.C p coefficient depends on the values blade angle (β) and tip speed ratio (λ).Tip speed ratio can be expressed by equation (2) where, R is the radius of the blades, and ώ m is the angular velocity of rotation of the blades.
The mechanical power output that can be obtained from a wind turbine depends on the values of the power coefficient C , p blade angle , b and tip speed ratio , l as shown in equation (1).The power that can be obtained by considering the mechanical transmission gear and generator efficiencies can be expressed by equation (3).
Determining the design parameters of the DFIG Depending on the mechanical power value that can be obtained from the wind turbine, the basic dimensions of the DFIG are determined analytically using general design formulas.Analytical modelling can be expressed as the determination of the generator volume and then the calculation of physical parameters such as stator/rotor core diameters and lengths using the value of the generator volume, in principle [28].In the apparent power equation (4) of the DFIG, winding distribution coefficient K w ¢ ¢' magnetic loading value B, ̅ electrical loading value ¢ac , ¢ stator diameter ¢D , ¢ core length ¢L , ¢ the number of revolutions per second ¢n¢ can be used to calculate the dimensions of a DFIG.Accordingly, the value of D 2 L (machine volume) can be calculated depending on the apparent power value of the DFIG.

K S 1.11
BacD L n10 kVA 4 For stator and rotor cores, it is necessary to calculate the value of the amount of flux per pole to calculate the number of conductors per slot and therefore the cross-section of the conductor (by equation ( 5)).Then, depending on the determined stator and rotor slot numbers, the number of windings per phase is calculated by equation (6), and the number of slots per phase is determined by the number of conductors per slot.When calculating the amount of flux per pole, the magnetic loading value B ¥, pole step ′Y′, and core length ′L′, parameters are used.To calculate the number of windings per phase (by equation ( 6)), the value of the back EMF induced per phase, winding distribution coefficient K w ¢ ¢', grid frequency ′f′, and amount of flux per pole m ¢AE ¢

( ) = AE
Depending on the mechanical power of the wind turbine, the electrical and physical basic parameters of the wind turbine with a power of 250 kVA are shown in table 1.

Winding material specifications
The previous section provided the basic parameters of the generator, and this section focuses on the choice of winding material between copper and aluminium.Copper has higher electrical and thermal conductivity than aluminium, but aluminium has a lower density than copper.Choosing aluminium as the winding material can reduce the weight of the winding, leading to weight gain benefits.Additionally, using aluminium instead of copper can result in cost savings.To evaluate the impact of the choice of winding material, DFIG models were created with both copper and aluminium windings while maintaining the same output parameters.Table 2 presents the physical properties of the aluminium and copper winding materials.
It is necessary to create equal winding resistance in both winding models to make an accurate comparison.To achieve equal winding resistance, the winding length must first be kept constant.Therefore, both generator models should have equal stator/rotor tooth width on which the windings are wound.Equation (7) provides cross-sectional equality for the copper and aluminium windings, depending on the equal winding length and resistance.Equation (8) provides the wire diameter to be used, depending on the cross-sectional value.
When calculating using equation (7), it is observed that the cross-section of the aluminium winding wire is approximately 1.8 times that of the copper wire for equal winding resistance.Consequently, the wire diameter ratio is approximately 1.34 times.As a result, the slot area for the aluminium winding will increase at the same rate as that for the copper winding, causing an increase in the outer diameter of the stator.Considering the weight of the winding, the weight of the copper winding and the weight of the aluminium winding can be calculated using equation (9) When comparing the mass of the winding material using equation ( 9), the copper winding with equal winding resistance and length is 1.826 times heavier than the aluminium winding.In this case, the total weight of the winding can be reduced by using aluminium winding.

Copper and aluminium wound DFIG FEA models
The copper and aluminium winding models of the DFIG were created using the Ansys-Maxwell program in accordance with the basic dimensions, and their dimensional and electrical parameters were optimized.Figure 3 displays the FEA model that was created.Three models were analysed: DFIG-1 (copper-winding model), DFIG-2 (aluminium-winding model) with the same machine dimensions and cross-sectional value as the copper-winding model, and DFIG-3 (aluminium-winding model) with electrical parameters equal to those of the copper-winding model.Table 3 displays the electrical and physical parameters of the three designed DFI generators.To ensure an accurate comparison, the design was conducted such that the electrical and magnetic loading values were approximately equal in all three models.Moreover, the output powers and power coefficients were set equal in all three models.As shown in table 3, the stator/rotor windings in the DFIG-1 and DFIG-2 models have conductors with the same physical dimensions and cross-section, resulting in the aluminium winding resistance value being approximately 1.5 times the copper winding resistance.The high winding resistance value of the DFIG-2 model increased in winding losses.In contrast, in the DFIG-3 model, the aluminium winding resistance was designed to be close to that of the copper winding, resulting in winding losses approximately the same as those in DFIG-1.In terms of efficiency, the DFIG-1 and DFIG-3 models have similar efficiencies, whereas the DFIG-2 model has approximately 2% lower efficiency.The lower efficiency in the DFIG-2 model is due to higher winding losses compared with the other models, as expected.
One of the crucial stages in the design process is the verification of the determined core dimensions in terms of magnetic saturation.Potential regions of magnetic saturation on the stator or rotor cores negatively affect the output performance.The starting point of saturation on the B-H curve of the laminated steel material (NSSMC_50H400) used in the stator and rotor cores is approximately 2 T. Figure 4 illustrates the flux distribution generated during the transient and steady operation for the DFIG-1 and DFIG-3 models.In both models, magnetic saturation does not occur at any point on the core in either the transient or steady operating states.
Figure 5 and figure 6 depict the waveforms of the phase-back electromotive force (EMF) and phase currents for the DFIG-1 and DFIG-3 models.It can be observed that both models have the same magnitudes of phase currents and voltages.
Figure 7(a) displays the torque curve of the DFIG-1 model, which reaches a steady state in approximately 1.5 seconds.Figure 7(b) demonstrates the torque curve of the DFIG-3 model, which achieves a steady state in a shorter period than the DFIG-1 model, but with greater torque ripple.The torque ripple can be calculated using equation (10).In the equation, T , min T , max and T avg represent the minimum, maximum, and average torque values, respectively.The torque ripple percentage is around 2% in DFIG-1, while in DFIG-3 it is approximately 3%. Figure 8 shows the efficiency versus angular velocity variation for all three models.The speed variation range starts from the speed region of 1000 rpm, which is the sub-synchronous operation region, and reaches the speed region of 2000 rpm, which is the over-synchronous operation region.In figure 8(a), the DFIG-1, which has a copper winding structure, has about 2%-3% higher efficiency compared to the DFIG-2 model with an aluminium winding of the same geometry, as expected.Figure 8(b), however, shows the efficiency-oriented optimization result, after the stator and rotor slot dimensions are revised according to the appropriate  aluminium winding parameters.All models with a stator diameter of 640 mm have the same or slightly lower efficiency than the DFIG-1 model with a copper winding.Considering factors such as volume and weight, the stator diameter of 640 mm was selected, which has the efficiency value closest to the efficiency of the DFIG-1 model.The slot dimensions were revised, and the DFIG-3 model was created on the basis of a stator diameter of 640 mm. Figure 9 shows the relationship between output power and efficiency when all three generator models operate at the rated speed.Aluminium-winding DFIG-3 and copper-winding DFIG-1 models have close performance with each other in terms of efficiency.On the other hand, the DFIG-2 model with aluminium winding has a lower efficiency value of about 2%-3%.

Thermal analysis of the DFIG models
Another important step in the design of electrical machines is thermal analysis.It is also necessary to verify the design work from a thermal point of view.Considering that a large part of the power in electrical machines is lost in the windings, thermal analysis becomes even more important for this study.While the thermal conductivity of copper is 398 W mK −1 , that of aluminium is 210 W/mK.Therefore, even if the copper and aluminium coil power losses are close to each other, their different thermal conductivities will cause the winding temperatures to differ.
Thermal analysis of the designed machine models was prformed using thermal analysis software.Figure 10 shows the temperature distributions for all three models, under full load operating conditions.For the DFIG-1 model (figure 10  winding, and 140 °C in the rotor winding.For the same geometric values, it was observed that the winding temperatures in the generator increase between approximately 35 °C-40 °C with the use of aluminium winding. On the other hand, the DFIG-3 model (figure 10(c)) has average temperatures around 98 °C in the stator winding and 110 °C in the rotor winding.Looking at the comparison with the DFIG-1 model, an increase of approximately 5 °C on average is observed.The thermal analysis results show that the DFIG-3 and DFIG-1 models can be operated under full load in continuous mode, whereas the DFIG-2 model cannot be operated under full load for a long time in continuous mode.

Results and discussions
The simulation work shows that the output performances of the DFIG-1 and DFIG-3 models, such as winding losses, output power, efficiency, etc, are very close to each other, while the DFIG-2 model is slightly inferior in terms of efficiency.In the light of these performance data, a comparison of the physical dimensions of the models is detailed in this section.Table 4 shows the physical dimensions of the three models.The DFIG-1 and DFIG-2 models have identical physical properties; the only difference is the preference of copper and aluminium winding material, respectively.The winding weight of the DFIG-1 model is approximately 15% of the total machine weight.In the DFIG-2 model, however, the weight of the winding was reduced using aluminium as the winding material, which accounts for approximately 5.1% of the total weight.In the weight comparison of the DFIG-1 and DFIG-2 models, which have the same volume, the DFIG-2 is approximately 11.8% lighter.Although the DFIG-2 model has the lowest weight compared with other models, it has the lowest efficiency.The winding weights of the DFIG-1 and DFIG-3 models, which have approximately equal efficiencies, are 84.2 kg and 38.66 kg, respectively.Although the winding weight was reduced by about 55% in the DFIG-3 model, compared to the DFIG-1 model, the core weight increased due to the increase in the stator diameter because the slot area increased to achieve the same winding resistance.The stator core weight of the DFIG-3 model increased by 26%, whereas the rotor core weight decreased by about 7% with an increase in the rotor slot area.Comparing the total machine weight and volume, the DFIG-3 model is about 2% heavier and about 16% larger in volume than the DFIG-1 model.
To make a cost comparison, the unit prices in table 5 were used.The unit prices in the table are average values and, may differ depending on many parameters.The cost comparison was carried out with a focus on the winding and core material.The cost comparison of DFIGs is given in table 6.
Since the core dimensions of the DFIG-1 and DFIG-2 models are the same, only the winding costs vary.Considering that about 84.2 kg of copper wire is used in the DFIG-1 model, the average cost of winding is 843.68 USD, while the average cost of winding in the DFIG-2 model, consisting of 25.3 kg of aluminium wire, is only 63 USD.The cost of winding in the DFIG-3 model is about 96.26 USD.Compared with the DFIG-1 model, the DFIG-3 model is about 93% inexpensive in terms of winding cost.On the other hand, when the costs incurred due to the increase in core size are analysed, the cost of the core of the DFIG-1 model is about 417.16 USD, while the cost of the DFIG-3 becomes about 468.32 USD.Considering the total cost in terms of winding and core, the DFIG-3 model provides a cost gain of 696.26 USD.
The annual energy production is considered for performance comparison of the designed DFIGs in a wind energy conversion system, based on the annual wind speed data of a region.The fundamental parameters and coefficients of the wind turbine are detailed in [30].The probability distribution function is created using wind data collected over a year, and this function is utilized to determine the average annual power output, as depicted in figure 11.The annual energy production and their comparisons are shown in table 7.According to the results, while the wind energy conversion system including DFIG-1 and DFIG-3 models have nearly the same annual energy production value, the DFIG-2 model has a slightly lower annual energy production, approximately at the level of 2%.This difference could be even greater in regions with different wind profiles.

Conclusions
Today, wind energy conversion systems from kVA levels to MVA levels are implemented.Numerous studies are being conducted on the continuity of energy production and the development of more efficient cost-effective structures in wind energy conversion systems.the turbine structure constitutes a very large part of the installation cost of wind energy conversion systems, the cost gain to be obtained in each other system will contribute to the energy production costs.This study addresses the gains that can be obtained when an aluminium winding structure is used in a DFI generator with an output capacity of 250 kVA compared with a copper winding structure.Three separate DFIG models were created, which are the copper-winding model (DFIG Model-1), the aluminium-winding model with the same geometries as the copper-winding model (DFIG Model-2), and the aluminium-winding model with the same electrical parameters as the copper-winding model  (DFIG Model-3), and their FEA analyses were carried out using the Ansys-Maxwell program.According to the analysis results, an approximately 11.8% lighter structure can be obtained using aluminium winding instead of copper winding in the same stator and rotor geometry.However, the alternator efficiency decreases by 2%-3%.
In addition, considering the thermal conditions, additional cooling or the use of a larger volume is required in the aluminium-winding structure.In comparison with the model designed with the same electrical parameters, thermal efficiency and output were approximately the same under full load conditions, whereas the aluminiumwinding model was 2% heavier and had 16% greater volume.Considering the costs of the core and winding, the aluminium winding model proves to be more cost-effective, with a reduction of about 55% in cost compared to the copper winding.Consequently, despite the necessity of a partial increase in volume and weight to achieve the same output performance as the copper winding structure, this compromise leads to cost savings in DFIGs.In addition, by maintaining the same geometries as the copper-winding structure, it is possible to reduce both cost   and weight at the expense of slightly lower output power.On the other hand, in terms of annual energy production efficiency, a comparison reveals that the DFIG-1 and DFIG-3 models exhibit approximately 2% higher efficiency than the DFIG-2 model, depending on the generator efficiencies and wind profile of the region.
In the future, this study will also help the manufacturing of DFIGs that have high energy and power density depending on the winding materials manufactured from the alloys that have high conductivity regarding the aluminium alloys.In addition, it can be provided that increasing the slot fill factor rate depens on the studies such as those using rectangular winding conductor geometries instead of cylindrical winding conductor geometries.

Figure 2 .
Figure 2. Connection of the conventional DFIG interconnected to the grid.

Figure 4 .
Figure 4. Magnetic field distributions of DFIG 1 and DFIG 3 in the transient and steady state.
(a)), the average temperatures were around 95 °C in the stator winding, and 105 °C in the rotor winding.For the DFIG-2 model (figure 10(b)), the average temperature values were around 130 °C in the stator

Figure 9 .
Figure 9. Efficiency variations versus output power at 1500 rpm.

Figure 12 .
Figure 12.Power curves of the system with DFIGs.

Figure 13 .
Figure 13.Product of power and probability

Table 1 .
. Basic electrical and physical parameters of the DFIG.

Table 3 .
Electrical and physical design parameters of the DFIGs.

Table 4 .
Physical specs comparison of DFIG models.

Table 6 .
Cost comparison of the DFIGs.

Table 7 .
Annual energy production comparison.