Detection of blade damage and ice accretion for health monitoring of wind turbines using integrated blade sensors
Abstract
Predictive maintenance of wind turbines could provide an estimated 50% reduction in cost when compared to reactive maintenance, which is the current industry approach. The ability to identify damage using wind turbine operating data is necessary for implementing a predictive maintenance approach. Wind turbine blade damage is the most costly type of damage to repair and requires significant down time. Furthermore, blade damage often leads to secondary component damage, such as the low speed shaft bearings and gearbox. It is proposed that rotor mounted inertial sensors can be used to measure static and dynamic rotor blade responses to monitor blade health to facilitate improved maintenance scheduling for individual turbines. In this thesis, free and forced dynamic response measurements on a 2.1 meter horizontal axis wind turbine are used to characterize the rotor structural dynamics of a turbine operating in steady wind flow. Several blade damage mechanisms were initiated to demonstrate that rotor mounted inertial sensors are an effective means of detecting and characterizing blade damage based on the blade structural dynamic response in operation. It was shown that for a particular damage condition, trends are observed in the forced response linear spectra. Notable, the lead-lag measurement degree of freedom was sensitive to blade root damage; however, tip delamination was most readily identified using the span measurement degree of freedom. For the case of blade ice accretion, aerodynamic and power output indicators confirmed changes that were observed in the forced response spectra for the flap measurement degree of freedom. Additionally, operational modal analysis was used to verify the observed changes in the forced response linear spectra.
Degree
M.S.M.E.
Advisors
Adams, Purdue University.
Subject Area
Alternative Energy|Mechanical engineering|Energy
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.