Date of Award

May 2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mechanical Engineering

Committee Member

John R Wagner

Committee Member

Richard S Figliola

Committee Member

Gregory M Mocko

Committee Member

Edward R Collins

Abstract

Wind turbine drivetrain test facilities are impressive laboratories that offer a controlled environment to test the response of drivetrains under design conditions. The stochastic nature of the wind results in highly dynamic loads and this is reflected in the design standards and certification process of wind turbines. A wide range of wind conditions and turbine operating states are prescribed as design load cases. The design process of a wind turbine yields thousands of time series of fluctuating forces, bending moments and speed to cover these design load cases. The capability of a test bench to replicate such dynamic loads has been the subject of this research at Clemson University in cooperation with GE Renewable Energy.

This collaborative research project used the 7.5-MW test bench of Clemson University to test two multi-MW drivetrain designs used on GE onshore wind turbines. This testing provided the first demonstration that the design load cases that typically drive the design of wind turbine drivetrains can be replicated on a test bench with an acceptable accuracy. The acceptance threshold for the accuracy was found to vary depending on the specific loads. The yaw and nodding bending moments are most critical to replicate accurately due to being generally much large than the forces, and the measurement error of the load application unit of the test bench was demonstrated to be an appropriate threshold based on multibody simulations. The dynamic response of one of the drivetrains tested was simulated using inputs to the model with and without the measured tracking error. This is the error between the loads commanded to the test bench and the loads that are measured at the point of application. For the drivetrain displacements considered, the multibody simulations quantified the impact of the tracking error on the displacements to be within 11% of the peak displacement. Most displacements were within 2.6%-3.1% of the peak displacement on average.

Multibody simulations were also used to quantify the impact of a cross-coupling effect between forces and bending moments that occurs when the load application unit of the test bench applies loads dynamically. The impact on the dynamic response of the drivetrain from the cross-coupling was found to be generally small and acceptable despite significant tracking error on the forces. The testing also served as experimental verification of a novel method for the early assessment of the capability of a test bench to replicate dynamic loads. The identification of load time series that should be replicated with an acceptable level of accuracy and those that are likely beyond the capability of the test bench was validated.

A new avenue of research in the wind energy sector has been initiated, and several recommendations are proposed for growing the knowledge base and the role of test benches for design certification purposes.

Share

COinS