Validation of a wind turbine gearbox strain simulation model in service to virtual sensing

Abstract

Over the last decades, the use of wind power as a source of renewable power has increased, while the Levelized Cost of Electricity (LCoE) of wind power has fallen. In order to further drive down the LCoE, there is an increasing interest to monitor key turbine quantities in order to improve maintenance procedures and decrease downtime. In this paper, we focus on the indirect detection of one such key quantity, the torque load on the gearbox. In order to avoid expensive direct torque sensors, we study the potential of strain gauges installed on the gearbox housing as virtual torque sensors. Our results verify that the strain response is repeatable, torque-driven and predictable; which are the most important conditions for using the sensors in virtual load sensing. Next, we compare the measured strain response to simulated strain results obtained from a physics-based model. The model consists of a static FE model of the gearbox housing and a torsional model of the shafts and gears in the drivetrain. An optimisation approach selects the FE model node where the simulated strain best matches the measured strain. We conclude that the studied strain gauges are promising for model-based virtual sensing of the torque on the gearbox.