Reproduction of five degree-of-freedom loads for wind turbine using equispaced electro-hydraulic actuators

Highlights

• Five degree-of-freedom turbine loads are defined in blade and hub coordinate systems.

• Using equispaced electro-hydraulic actuators to replicate the defined turbine loads.

• Using a loading decomposition strategy to determine the reference force for each actuator.

• Designing the decomposition strategy by incorporating additional dummy loads.

• Each actuator features two intrinsic pressure-sensed closed loops and two pressure modes.

Abstract

Several improvements are proposed to enhance the capability of conventional wind turbine simulators since such simulators can only simulate the turbine shaft torque, or can partially generate the other five degree-of-freedom turbine loads in addition to the turbine shaft torque. In the improved wind turbine simulator, the real wind rotor and blades are represented by an equivalent rotating disc to replicate the turbine inertia effects. Twenty-four electro-hydraulic loading actuators are symmetrically distributed around this rotating disc and are controlled independently to reproduce the additional five degree-of-freedom turbine loads that can be described in the blade and hub coordinate systems. A loading decomposition strategy is also proposed to decompose such five-degree-of-freedom turbine loads into reference loading forces for each loading actuator by incorporating additional seven-degree-of-freedom dummy loads. A typical axial loading actuator is taken as an example for in-depth dynamic modeling and analysis. Two intrinsic pressure-sensed closed control loops and different pressure modes of this actuator are presented and analyzed in detail. The dynamic model of this typical loading actuator has been thoroughly validated by using experimental measurements. Simulation results have also demonstrated that the five degree-of-freedom turbine loads can be accurately reproduced by using such electro-hydraulic loading actuators. Therefore, the improved wind turbine simulator can fully simulate steady-state and dynamic turbine loads with good accuracy and can expedite the system-level loading tests that are critical in increasing turbine reliability and offering high confidence system design.

Introduction

Although being presently one of the fastest growing renewable energy sources around the world, wind turbines still frequently experience catastrophic failures and reliability issues such as gearbox faults, widespread premature generator failure, wind rotor failure and main shaft/bearing failure [1], [2]. Such faults and failures stand out as the leading contributors that cause the most turbine downtime and pose a challenge to wind turbine reliability and durability [3], [4]. These failures and the associated down time also elevate the operation and maintenance costs and subsequently increase the overall cost of wind power. Major reasons for such reliability issues are the detrimental effects from continuously turbulent wind loads induced by aerodynamic variations and control actions [5], [6].

In the light of such increasingly adverse loading influences on wind turbines, wind turbine simulators become indispensable for evaluating the reliability and durability of turbine subsystems at the pre-production prototype stage. These simulators can accurately reproduce the specified wind loads and hence can create well-controlled loading conditions for performance testing of different turbine subsystems. Therefore, high-confidence and reliable designs for wind turbines can be verified by using these simulators.

There are primarily two types of conventional wind turbine simulators: wind turbine torque simulators and wind turbine loading simulators. Wind turbine torque simulators enable the accurate simulation of aerodynamic torque in a controlled environment. These torque simulators can be typically configured into a torque-controlled back-to-back set-up with “wind & rotor” and “gear & generator” sides [7], [8]. The “wind & rotor” side is always represented by an electrical motor and an optional speed reducer, while the “gear & generator” side mainly consists of a wind turbine gearbox and an electrical generator. The “gear & generator” side is driven by this electrical motor at a given wind speed and thus experiences the applied torque conditions very similar to the in-field conditions of a real wind turbine. A wind turbine torque simulator was developed in Ref. [9] to simulate the rotational speed and shaft torque of a real wind turbine. In that simulator, an inverter-controlled induction motor was controlled to supply the shaft torque and drive a generator by regulating the current and frequency of the inverter. However, the simulator cannot accurately reproduce the dynamic characteristics for a given wind turbine. A DC motor based wind turbine torque simulator was developed in Ref. [10] to create a controlled test environment for drive train of wind turbines. However, the simulator cannot be employed to simulate dynamic behaviors of large scale wind turbines. In Refs. [11], a speed-driven hardware-in-the-loop physical wind turbine torque simulator was proposed to replicate the dynamic behavior of a stand-alone low-power wind energy system. However, the simulator suffers from the problems of reduced frequency bandwidth and inherent physical disturbances that could not be fully alleviated. A 20 kW torque simulator that has similarity with a 3 MW wind turbine was developed in Ref. [12] to verify advanced diagnosis algorithms by adopting a torque control method. However, the similarity only existed at the entire system level and there is no detailed information regarding component-level similarity. Thus, the quantitative comparisons between the torque simulator and the actual 3 MW wind turbine do not provide any meaningful information. A wind turbine emulator was presented in Ref. [13] to simulate wind turbine power curves for a given wind speed. However, the simulation accuracy of the emulator could not be well guaranteed since the emulator only works in open loop without using controllers. In general, wind turbine torque simulators can only simulate the turbine shaft torque and cannot reproduce other degree-of-freedom turbine loads such as thrust load, bending load and shear load.

Wind turbine loading simulators were primarily developed to conduct performance and reliability tests on wind turbine drivetrain prototypes. Apart from the turbine shaft torque load, these loading simulators can also be capable of reproducing off-axis turbine loads by using non-torque loading actuators. A typical dynamometer test facility located at National Renewable Energy Laboratory (NREL) could not only generate turbine shaft torque load, but also could apply more off-axis turbine loads such as shear load, bending load and independent thrust load to the test article. Such additional loads were produced by using servo hydraulic cylinders in the dynamometer [14], [15]. However, the dynamometer was not able to apply high-frequency reversing loads for different test articles. Other full scale test-rigs [16], [17] commonly employed non-torque hydraulic actuators to generate off-axis turbine loads for conducting static and dynamic tests of the entire turbine system. However, the system design and modeling issues were not sufficiently addressed and hence the dynamics and capability of these hydraulic actuators were not thoroughly investigated. Further, the system synergy between mechanical design and practical electrical controller was not detailed. Although wind turbine loading simulators can reproduce more off-axis turbine loads in the other five degrees of freedom, they cannot fully generate the five degree-of-freedom turbine loads. Further, such loading simulators fail to include dynamic effects of relatively large turbine inertia that can induce large vibrations and elastic deflections of the gearbox and will cause gear tooth misalignment, leading to uneven tooth load distribution and increased bearing vibrations [18].

In this paper, significant improvements are presented to enhance the loading reproduction capability of conventional wind turbine simulators. The major contributions of this paper include the following:

• Since the turbine inertia is an important source for premature gearbox failures, a rotating disc is also proposed to represent the real turbine rotor and turbine inertia effects.

• A set of equispaced electro-hydraulic loading actuators are symmetrically placed around this rotating disc to fully replicate the five degree-of-freedom turbine loads.

• A loading decomposition strategy is proposed to decompose the five-degree-of-freedom turbine loads into reference loading forces for each loading actuator by involving additional dummy loads.

• The in-depth modeling and analysis of a typical loading actuator are presented. Intrinsic pressure-sensed closed control loops and two different pressure modes are also defined to thoroughly analyze the operating characteristics of the loading actuator.