A turbine simulator allowed testing and defining a promising self-sensing, pitch-control scheme against a wide range of blade and turbine conditions.
Tobias Rösmann / Moog Wind Pitch Control Systems / www.moog.com
Control systems generally ensure the safe operation of wind turbines in adverse conditions. A more recent self-sensing, closed-loop pitch control system will let synchronous motors adjust blade pitch, even if there is a failure in the motor position feedback. In traditional pitch-control approaches, non-synchronized pitch speeds can cause high turbine loads. But a study by the Fraunhofer Institute for Wind Energy and Energy System Technology, and Moog Inc. found that a self-sensing, closed loop, motion-control concept can properly control the feathering speed, even under grid loss condition.
Recently, Fraunhofer and Moog conducted a load-simulation-based case study on the effect of non-synchronized pitch speeds during feathering. We simulated a 7-MW turbine to examine the effect of pitch position incoherence on the turbine structural loads during feathering.
Pitch control’s two functions
A wind turbine, pitch control performs two essential functions. First, it controls the turbine rotor speed when the wind exceeds the turbine’s rated values by adjusting the blade pitch. Second, it acts as the actuator for the wind turbine braking system.
The schematic shows the general control layout for one blade.
To stop a wind turbine, all three blades must move into the feathering position and distribute and balance the loads on all structural parts during the procedure. All three pitch axes must move out of the wind synchronously.
Various open-loop control schemes are available for the pitch actuator feathering drive if the pitch axis position feedback signal is lost. These include hydraulic actuators, dc motor pitch actuators, and ac induction pitch actuators. All three have one disadvantage: The speed of each axis depends on the load applied by the blade. The self-sensing, closed-loop pitch control scheme overcomes this disadvantage. Here’s why: The self-sensing, closed loop, pitch-control scheme prevents blades from asynchronous movement in the event of an encoder failure.
Compared to state-of-the-art self-sensing control schemes for ac induction or synchronous motors, the new closed-loop speed control scheme provides up to three times the rated torque from standstill to ensure a safe and high-performance movement into the feathering position. The principle electrical design of a pitch control system is shown in Electrical layout for the pitch axis.
Pitch incoherence during an emergency stop
As part of our case study, Fraunhofer and Moog simulated the effect of pitch incoherence during an emergency stop using three scenarios. By definition, pitch-position incoherence appears when the blades have different positions. The first scenario describes the emergency stop with synchronized feathering speed of 6°/s for all three blades.
The graphs are from three simulated tests. Emergency feathering starts at the 30 sec mark. The scenario with one blade stuck or unfeathered created the greatest moments.
In a second scenario, one blade pitches at 3°/s, while the other blades pitch at 6°/s. This scenario represents an open-loop controlled feathering drive of one axis.
In the last scenario, one blade is stuck. We analyzed the effect of blade seizure with GH Bladed, software for wind turbine design and analysis. Results showed that a stuck blade generates the worst-case load on the turbine structure. One reason a blade can stick is an encoder failure in one axis.
Tilt moment during an emergency stop shows a comparison of the tilt moment at the yaw bearing. We normalized the moments relative to the maximum absolute tilt moment at the synchronized feathering speed. The emergency feathering command starts at t = 30s. The maximum moment at the first oscillation of the moments without synchronized feathering speed is 40%, and 140% greater than with synchronized feathering speed. Depending on rotor position when the feathering initiates, the unsymmetrical load may also be induced into the yaw moment MZK.
The graphs are for the three test scenarios used previously.
The amplitude of the yaw bearing, tilt-moment oscillation is also equivalent to the corresponding mainframe load. Our case study shows tower foot bending moments for each scenario in Bending moments at the tower bottom. These moments also show increased amplitudes during the three scenarios of unsynchronized pitch feathering speed.
The impact of pitch system faults, such as the discussed encoder failure, is reported in “Identification of pitch system failure modes and their effects on wind turbine design,” by T. Delouvrié, for the Centre of Renewable Energy Systems Technology, Electronic and Electrical Engineering Department. Researchers identified the stuck-blade scenario as a likely “design driver.” In other words, it is critical to prevent the stuck-blade scenario. The research performed by Fraunhofer and Moog found that pitch drive capabilities, which prevent asynchronous pitch feathering speeds, will mitigate increased loads on main turbine components especially in terms of cumulated lifetime fatigue.
Self-sensing speed control algorithm
In response to a simulated encoder failure, we applied the self-sensing speed control algorithm described in a paper presented at the IEEE’s International Symposium on Sensorless Control for Electrical Drives titled “Implementation and Applications of Sensorless Control for Synchronous Machines in Industrial Inverters.” The self-sensing control in our study reduced turbine fatigue loads, which would arise from a stuck blade due to an encoder failure in one pitch axis. This approach let the pitch system drive the blade into the feathering position without depending on the encoder signal.
The self-sensing control algorithm taken from the IEEE proceedings uses an extended Kalman filter to estimate the motor states such as d-, q-current, rotor position, and speed. The model of the extended Kalman filter accounts for the saturation of the q inductance (Lq), with increasing q-current, by using a look-up table, which is part of the parameter set. We also considered the d-inductance saturation. The output-estimated-position angle is used within the dq-transformation of the motor currents and the estimated speed is fed to the speed controller. At low motor speeds, we injected an additional d-current signal to improve the estimate and guarantee full performance especially to accelerate the blade from zero speed to run into the feather position. The illustration Measurements of self-sensing control provides some details.
Self-sensing control scheme
To express the drive parameters for a self-sensing control as robustly as possible, we carried out enhanced simulations to investigate the influence of variations. For one, we used the adapted real-time firmware code from the drive to simulate the control and a Kalman environment, and for another, developed a more realistic motor model. This motor model included saturation and cross-coupling effects in the machine. From these simulations, we created a parameter data set for the drive which also helped commission a self-sensing driven, Moog Interior Permanent Magnet Synchronous Motors (IPMSM) on a test bench. Measurements of self sensing control provides an example of the successful implementation of self-sensing control with IPMSM for pitch systems. We accelerated the motors from a standstill, while a load machine applied up to 300% of the nominal torque.
The graphs provide an example of the successful implementation of self sensing control.
Qualifying the performance of a new pitch axis servo system
We qualified the performance of a new Moog Wind Turbine Pitch Axis Servo (PAS) System by running a comprehensive application load analysis and configuring the new system. By thoroughly expressing the parameters for this system, the team was able to fine tune the system’s performance. The last step we took was qualifying the new system vis-à-vis the requirements defined by the load analysis.
Analyzing load and designing pitch axis
The main components of the Moog Wind Turbine Pitch Axis Servo system are a:
• Motor and holding brake
• Frequency inverter drive
• Motor-drive cable harness, and
• Grid-filter inductor
Other components that must be adjusted relative to application
loads are a:
• Backup power supply, and
• Brake resistor.
To configure the PAS system, Moog developed an automated load-analysis tool to import and analyze clients’ load simulations. Moog conducted these load simulations for different load cases defined in the certification guidelines in GL 2010 and respectively IEC 61400. The time series for the pitch design must include all relevant load components for the pitch system, such as aerodynamic, inertial, and gravitational loads, as well as friction and efficiencies.
Not all design load cases are relevant for the pitch design. In the table, Design load cases, we give an overview of the DLC groups and their typical relevance for the pitch system. The DLC groups marked by ‘o’ in the table typically do not represent the worst-case scenarios for the pitch system.
Defining PAS system duty cycles
To match the results of the application load analysis with the capabilities of a certain PAS system configuration, Moog’s team defined a scope of duty cycles for rotating electrical machines according to IEC standard 60034-1. The duty cycle describing the performance of the PAS system during normal operation DLCs is defined in accordance with the IEC S9 duty cycle, characterized by non-periodic load and speed variations within a permissible operating range including frequent overloading.
We displayed the capability of the PAS system of the introduced duty cycles with torque-speed curves. Wind-turbine owners can operate the PAS system at any point on or below one of the torque-speed curves shown in the table without exceeding the thermal or other limits, such as electrical, mechanical, or insulation.
What we learned
From our findings, we learned that one critical design goal for a pitch system is to ensure the wind turbine avoids a blade seizure situation. The simulation work with Fraunhofer also showed that synchronized feathering of all three blades further reduces the loads on a wind turbine. The data gained from these simulations became one of the primary drivers for Moog’s new self-sensing, closed-loop-control algorithm. By employing the Moog IPMSM, we used the new control scheme to provide high peak performance even under stall, grid loss, or both conditions.
We also found that the general qualification design process for the pitch-axis servo system is crucial for safety because load requirements must be fulfilled for 20 years, under all environmental conditions. This is why Moog developed a standardized tool to analyze the time series generated by the simulation software. Its output provides the performance requirements for each pitch axis servo system across the required duty cycles. WPE&D
Filed Under: Featured
