Research on the Application of Power Semiconductor Devices in Wind Turbine Generators

Harnessing solar energy through wind power is not a novel concept, yet today’s power semiconductor devices and control systems have made this energy source far more viable and efficient. Among existing solar energy utilization technologies, wind turbine generators have emerged as the forerunner in large-scale production of green electrical energy. Currently, both the US government and European governments are vigorously supporting the production of sustainable energy. In 2003, the total installed value of wind power plants in the United States reached 1.6 billion US dollars; it is projected that by 2020, the installed capacity will increase by an additional 100,000 MW, meeting 6% of America’s electricity demand. The US will also construct the world’s largest onshore wind farm in Tehachapi, located in the Mojave Desert. However, data from 2002 shows that 90% of the world’s newly added wind power capacity was still in Europe. Variable Energy Input: A Challenge for Designers It is instructive to assess how early pioneers addressed many of the challenges that still perplex designers today. Chief among these challenges is the variability of energy supply. Conventional steam turbine power plants employ four key mechanisms to regulate generator speed and electrical output: the primary energy consumption rate for steam generation; the rate of steam delivery to the turbine; the electrical excitation level of the generator; and changes in the rotor load angle. Such generators are synchronous generators, where the rotor rotates at a multiple of the grid frequency and is synchronized with it. Adjusting the rotor’s angle relative to the zero-phase "no-load" position increases or decreases the electrical energy supplied to or drawn from the grid, enabling the generator to operate as a power producer or a motor, respectively. In typical generator operation, the rotor leads the grid by approximately 30°. Since electrical output is directly coupled to the grid, the generator shaft torque provided by the robust grid conditions controls its speed, maintaining a constant grid frequency. So, how much power can wind energy generate? Theory dictates that for a known air density, the available energy in watts per square meter varies with the cube of the airflow velocity. Thus, rotor performance is critical to every aspect of wind turbine generator design. One of the most vital parameters is the tip speed ratio (TSR), defined as the ratio of the rotor blade tip speed to the free-stream wind velocity. This parameter describes the rotor’s power coefficient (Cp); in 1919, German physicist Albert Betz postulated that this coefficient could never exceed 0.593 (the Betz limit). In practice, the typical rotor power coefficient rarely exceeds 0.4 at a tip speed ratio of 7 (Figure 1). Assuming a fixed rotor speed and negligible efficiency losses, the power output of a wind turbine generator can be calculated using the following formula: Power = Cp × (ρ/2) × Vw³ × A Where: Cp = rotor power coefficient ρ = air density (kg/m³) Vw = wind speed (m/s) A = rotor swept area (m²) It is therefore useful to evaluate wind turbine generators based on their rotor swept area and power generation in kilowatts per hour. The designer’s task is to find the optimal combination of rotor structure and generator principle to achieve the maximum overall power coefficient, at a reasonable cost for mass production. Practical wind turbine generators have an output power range from 20 kW to 30 kW, with the current state-of-the-art reaching 4.5 MW. Most utilize three rotor blades, as experiments have shown this configuration delivers the best balance between efficiency, dynamic performance, and structural economy. The core components typically include a rotor, a gearbox to increase the generator shaft speed, a generator, a circuit interface, and a control loop (Figure 2). The greatest ongoing challenge has been stabilizing rotor speed to maximize power generation. While a wind turbine generator is a mechatronic system where key components cannot be isolated, rotor control principles are a decisive factor. The control system must protect the machine’s operation under conditions ranging from calm still air to extreme gusts with multi-directional, variable-speed changes that may occur only once a century. As an indicator of the associated mass, the rotor assembly of Vestas’ V90 series 3 MW wind turbine generator weighs 40 tons, despite the use of numerous high-cost carbon fiber composites. The Simplicity of Stall Control Masks Underlying Problems One method to limit power capture is to rotate the rotor assembly out of the wind. A yaw system is generally used to keep the rotor facing into the wind, consisting of a wind speed sensor, a wind direction sensor, an electric or hydraulic drive unit, interface circuitry, and gears and bearings to rotate the generator nacelle. The sensor assembly is often mounted at the rear of the nacelle, typically a three-cup anemometer with a wind vane. Other technologies include ultrasonic devices, such as the pair used on Vestas’ V90-3.0 MW turbine. In practice, the wind speed behind the rotor is slightly lower than the actual free-stream speed, a result of the local low-pressure effect created by the rotating blades. While this difference is not significant, characterization can compensate for such errors. However, experience has shown that speed control using a yaw system yields poor results, so designs generally either maintain the rotor in the maximum power windward position or rotate the nacelle to the minimum wind energy direction for shutdown. The simplest aerodynamic method to stabilize energy capture is passive stall control, which uses a rotor with a fixed pitch angle. At a given rotor speed, an increase in wind speed causes airflow separation over the blade surface, creating a stall effect. This airflow separation automatically limits energy capture but is dependent on air density and blade surface finish quality. This method also requires robust grid conditions and a high-power generator to maintain stability. If grid connection is lost or a power fault occurs, rotor overspeed must be prevented, necessitating aerodynamic brakes on the rotor and conventional disc-type mechanical brakes on the input shaft. Since the rotor has a fixed pitch angle and cannot be rotated to the maximum torque position to facilitate starting, it is sometimes necessary to operate the generator in motor mode to accelerate the rotor to grid-synchronized speed. Finally, the structure must be sufficiently robust to withstand the large dynamic loads inherent to stall control. Nevertheless, several successful wind turbine generators employ this principle. Nordic WindPower’s Model 1000 1 MW wind turbine generator is a simple, lightweight design featuring a two-bladed stall-controlled rotor with a swept area of 2290 m². This turbine is self-starting, with stall strips on the blades to flatten the peak power curve seen in some early stall-controlled turbines, resulting in a flat-topped power curve. The rotor uses a glass-fiber reinforced polyester structure, which offers excellent aeroelasticity—this "soft" or "flexible" structure readily absorbs large dynamic loads. Other helicopter-derived components include a teetering hub, whose elastic bearings allow ±2° of relative movement between the blades and the input shaft, reducing wind shear forces between them. Additional damping in the generator and yaw control systems further enhances structural flexibility. The generator, manufactured by Weier Electronics, is a four-pole single-speed induction generator, where the rotor rotates slightly faster than the rotating magnetic field. This slip provides a damping effect that helps suppress electromechanical oscillations. The slip value can be varied between 1% and 10% by switching resistors in the generator rotor circuit to control the excitation current. Since the torque of an induction generator is proportional to slip, this method enables speed control—a function that is difficult to achieve with asynchronous generators. At 0% slip, the generator is synchronized with the grid frequency and produces no electrical power (excluding reactive power consumed by the rotor), nor does it draw power. Similarly, if the generator speed is lower than the grid frequency, it operates in motor mode and draws current from the grid. To limit this current draw, the input shaft disc brake typically prevents rotor rotation when wind speed is below approximately 4 m/s to 5 m/s (the turbine’s so-called cut-in speed). Vestas has also applied slip control technology in its OptiSlip system, with optical coupling between electronic circuitry on the rotor and a controller on the stator. In this design, the control slip value is approximately 10% with a response time of about 10 ms, enabling smooth power output under turbulent conditions and reducing structural loads. Slip value also affects power generation efficiency; megawatt-class generators typically operate with a slip value of 1% and an efficiency of approximately 95%. Because the rotor circuit consumes reactive power, the power factor is generally low (approximately 0.87). For this reason, switched capacitor banks are an integral part of traditional systems, but power circuits are increasingly being used to control the power factor actively. For Nordic’s Model 1000 turbine, switched capacitors maintain the output power factor at 1 across the entire operating range of the turbine. By introducing damping factors into the yaw system’s control loop, it is possible to allow a degree of blade oscillation around the tower axis, thereby absorbing turbulence. As a result, the Model 1000 turbine’s structure can withstand wind speeds of 55 m/s, start operation at 4 m/s, and shut down at 25 m/s. At a rotor speed of 25 rpm and a blade tip speed of 71 m/s, the generator delivers its maximum power of 1 MW at a wind speed of 17 m/s. When the rotor begins to overspeed, centrifugal force actuates a hydraulically released valve, rotating the blade tips to a braking position. Mita-Teknik, a specialist in wind power control systems, produces a SCADA (Supervisory Control and Data Acquisition) system that also activates the aerodynamic and mechanical brakes. The generator outputs 690 V three-phase alternating current to the tower base via flexible cables, which the SCADA system retracts to prevent tangling. Communication between the SCADA system and central equipment is via a modem and telephone line, with a personal computer used for independent monitoring and recording of turbine operation. Control Systems Simplify Power Capture Many wind turbine designers prefer rotor pitch control technology, as it significantly mitigates speed variation and system power capture issues. Contemporary designs employ two distinct pitch control methods: the first gradually reduces the blade’s angle of attack to the airflow from the maximum position for full power to a cyclic pitch position for minimum power capture; the second increases the angle of attack to the aerodynamic stall point. Danish engineers MB Pedersen and P Nielsen tested both methods in the experimental Nibe-A and Nibe-B turbines in 1980 (Reference 1). Their results showed that full blade pitch control produces a smoother output characteristic and has the potential to reduce rotor thrust at high wind speeds (Figure 3). Today, more advanced blade aerodynamic and control algorithms help minimize the differences between the two methods. Bonus Energy’s products are a prime example of active stall design branded as CombiStall. Its "Danish concept" turbines include a fixed-speed three-bladed rotor, a generator that supplies power directly to the grid, and a fail-safe system. The company’s largest product is the Model B40 2.3 MW turbine, with a rotor swept area of 5330 m². The glass-fiber reinforced epoxy blades can be rotated by 80° to a shutdown position. During normal operation, a microprocessor-controlled servo loop continuously adjusts the blades to the stall position. A dual-generator design enables two-speed operation (11 rpm or 17 rpm), increasing efficiency at partial loads. By connecting a six-pole generator winding at low wind speeds, the generator can produce power at two-thirds of its rated speed; at higher wind speeds, it switches to the four-pole main winding and operates at the normal speed. The turbine is self-starting at an average wind speed of approximately 5 m/s to 6 m/s. When a thyristor soft-start circuit connects the generator to the grid, the rotor accelerates to grid-synchronized speed. After a few seconds of steady operation, the main contactor bypasses the thyristor circuit to eliminate semiconductor losses. The turbine’s electrical output then increases approximately linearly with wind speed up to a maximum of about 14 m/s to 15 m/s, at which point the control loop engages to maintain constant power output and prevent generator overload. If the average wind speed exceeds the turbine’s operating limit, the control system cyclically pitches the blades and applies the brakes to shut down the turbine. When wind speed drops below the restart limit, the safety system resets automatically and the turbine restarts—unless a fault occurs, in which case the turbine remains offline. A backup system provides fail-safe operation, as it can deactivate the turbine control system using a centrifugal device in the event of a severe fault. Frequency Converters Simplify Operation The most flexible power capture and control capabilities come from variable-speed operation, as the turbine rotor can ideally operate at the maximum blade tip speed ratio. Early attempts to replace fixed-speed step planetary gearboxes with automatic gearboxes failed due to cost and reliability issues. Since slip control methods only provide limited speed control for induction generators, many modern turbines adopt an alternative approach first trialed in the 1980s with the 3 MW Growian wind turbine: the DFIG (Doubly Fed Induction Generator). The Growian design included a synchronous generator with a three-phase slip-ring fed rotor, acting as a rotor-wound induction generator. This configuration allows a cycloconverter to inject alternating current into the rotor (Figure 4a). A cycloconverter is an AC-AC frequency converter manufactured with a thyristor array that samples the three-phase line frequency to generate a low-frequency control waveform (Figure 4b). Superimposing this control waveform on the rotor’s electric field helps stabilize the generator’s output frequency; controlling the amplitude and phase of this waveform regulates the generator’s power coefficient, thereby simulating the ability of a synchronous generator to supply real and reactive power. This configuration still has drawbacks, one of which is greater vulnerability to grid faults compared to other designs. A relatively simple variable-speed technology uses an AC-DC-AC link as a frequency converter, which first rectifies the generator’s "rough AC" output and then inverts it to line frequency. This technology decouples the generator from the load, enabling the use of more efficient synchronous generators and maintaining generator torque control by adjusting the DC link state. Vestas’ V90-3 MW wind turbine is a product example, employing full blade pitch control and the company’s OptiSpeed technology to control its rotor with a swept area of 6362 m². The OptiSpeed system allows a 60% variation in rotor and generator speed, minimizing fluctuations in power output to the grid and reducing structural stress. At the core of this system is the company’s VMP-Top controller and frequency converter, which form the power electronic circuitry to control the generator and its output to the grid transformer. The turbine is otherwise conventional, retaining a gearbox to increase generator speed (the generator’s original speed range is 9 rpm to 19 rpm). However, in a conceptually simplest approach, Enercon has pioneered a range of gearless direct-drive wind turbine generators, with a current rated power output of up to 4.5 MW. In this design, the rotor is mounted directly on the generator, reducing the number of drivetrain bearings to only two low-speed rotating components. The challenges lie in generating sufficient power at low speeds and converting it to grid frequency in the most efficient manner. Enercon’s solution to the generator problem is to use an electrically excited synchronous generator with a large number of poles—for example, the 4.8 m diameter 84-pole electrically excited synchronous generator in the company’s Model E-40 600 kW wind turbine. Here, the rotor speed varies from 18 rpm to 34 rpm with a swept area of 1521 m². Leveraging its extensive expertise in industrial variable-frequency drive design, Enercon uses its own proprietary electronic circuitry. In contrast, Zephyros’ newly launched Model Z72 2 MW wind turbine also features a direct-drive generator but uses an upgraded ACS 1000 variable-speed motor drive controller from ABB. A drive shaft bearing supports a permanent magnet generator (PMG), also manufactured by ABB. Zephyros highlights the benefits of PMGs, citing reduced generator losses, excellent partial-load efficiency, and a lower failure rate. The main disadvantages of PMGs are their high cost due to the use of high-permeability magnetic materials (such as neodymium-iron-boron and samarium-cobalt) and poor power factor characteristics, which must be compensated for by the frequency conversion circuitry. Nevertheless, many experts consider PMGs to be the future direction, especially for large direct-drive designs. Adrian Wilson, an electrical technology specialist at the UK’s NaREC (National Renewable Energy Centre), states that this approach is the core of an ongoing research project focused primarily on weight reduction. Since the theoretical power output of a wind turbine generator increases with the cube of the air volume it captures, the weight of structural components increases proportionally. Wilson notes that current design methods cannot simply be scaled up to the 10 MW level—let alone the 20 MW or 30 MW required in the future—so his department is investigating a direct-drive design that eliminates the weight of the gearbox. This approach also requires a large-diameter generator. At the scale involved in this project, a counterintuitive design is possible: a bicycle wheel-like structure with spokes supporting the generator’s electrode pairs. Grid output connection requires a full-power AC-DC-AC frequency converter link, which in turn requires multiple parallel converters. IGBTs Replace Thyristors The power semiconductor devices required for wind turbine generators are unfamiliar to those working in microelectronics. Instead of submicron line widths, one considers the European standard printed circuit board area occupied by a single device module (ranging from 34 mm×94 mm to 140 mm×190 mm). Such devices can withstand kiloampere-level currents at several kilovolts, and advancements in this technology over the past few decades have been the single greatest contribution to the development of wind turbine generators. In the Growian era, thyristor technology could handle high-power applications but suffered from high conduction losses and poor switching performance, often with switching times in the 100 ms range. Correspondingly, frequency converter stages used 6-step or 12-step waveforms to approximate a sinusoidal energy distribution, resulting in particularly strong odd harmonics (e.g., the 5th and 11th harmonics). These limitations necessitated the use of harmonic frequency filters. Replacing the first-generation thyristors of the Growian era with IGBTs (Insulated Gate Bipolar Transistors) allows the use of PWM (Pulse Width Modulation) to overcome poor harmonic performance. This technology also enables more convenient control of real and reactive power. While traditional thyristors are highly robust—modern thyristors such as Mitsubishi’s FT1500AU-240 can switch 1.5 kA of current at 12 kV with a switching time of 15 ms—conventional thyristors cannot be turned off when the conduction current exceeds the holding current value. GTO (Gate Turn-Off) thyristors (e.g., Mitsubishi’s FG6000AU-120D) can continuously supply 6 kV of voltage and 1.5 kA of current, with turn-off control achievable in 30 ms, but they are difficult to drive. Worse still, all thyristors are challenging to parallelize—a necessity to achieve the power levels required for wind turbine generators. High-power IGBTs combine the easy drivability and current-sharing characteristics of MOSFETs with a switching time of 1 μs. Although the PWM frequency required for converting line frequency is low (only a few kilohertz), this fast switching reduces conduction losses as the IGBT traverses the linear operating region. Devices such as Eupec’s FZ600R65KF1 have a turn-on time of less than 1 μs and a turn-off time of less than 6 μs, capable of controlling 1.2 kA of current at 6 kV; low-voltage devices such as the company’s FZ3600R12KE3 can switch 3.6 kA of current at 1.2 kV. IGBTs are therefore used in high-power frequency converters and soft-start controllers. Other companies specializing in high-power semiconductor devices include ABB, Dynex, Fujitsu Electronics, Powerex, and Semikron. Gamesa Eólica’s wind turbine generator series, with an output power range of 660 kW to 2 MW, extensively uses IGBT technology to achieve variable-speed and variable-frequency control. Variable pitch rotor blade control allows continuous adjustment to capture maximum power and can be coupled to a DFIG system with a generator speed range of 900 rpm to 1900 rpm. This control technology minimizes peaks, flicker, and harmonics, simplifying grid connection approval. A vector control system can generate or consume reactive power, enabling precise adjustment of the power factor and improving grid voltage stability. Gamesa Eólica’s power circuitry also allows its turbines to remain online during power outages elsewhere in the grid. Economically, these features are critical in Spain, where additional tariffs are levied for high-quality grid connections. Ivan Novikoff, Head of the Wind Energy Division at France’s Cegele, notes that the choice of wind turbine generator and its technology depends primarily on the location and characteristics of the local infrastructure. Novikoff states that issues such as cable laying, starting current during commissioning, and short-circuit current are all dependent on the system configuration. When specifying wind turbine generators for a known application, the company considers numerous secondary but essential factors, ranging from permitted rotor height and noise emission to the quality of the manufacturer’s on-site service. From an investor’s perspective, Novikoff explains, the key economic factors for the machine include the reliability of wind supply, machine reliability and maintenance costs, and differences in electricity production tariffs.