Browsing by Subject "Probabilistic analysis"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Development of self-powered wireless structural health monitoring (SHM) for wind turbine blades
    (2015-01) Lim, Dongwon
    Wind turbine blade failure can lead to unexpected power interruptions. Monitoring wind turbine blades is important to ensure seamless electricity delivery from power generation to consumers. Structural health monitoring (SHM) enables early recognition of structural problems so that the safety and reliability of operation can be enhanced. This dissertation focuses on the development of a wireless SHM system for wind turbine blades.The sensor is comprised of a piezoelectric energy harvester (EH) and a telemetry unit. The sensor node is mounted on the blade surface. As the blade rotates, the blade flexes, and the energy harvester captures the strain energy on the blade surface. Once sufficient electricity is captured, a pulse is sent from the sensing node to a gateway. Then, a central monitoring algorithm processes a series of pulses received from all three blades. This wireless SHM, which uses commercially available components, can be retrofitted to existing turbines.The harvested energy for sensing can be estimated in terms of two factors: the available strain energy and conversion efficiency. The available strain energy was evaluated using the FAST (Fatigue, Aerodynamics, Structures, and Turbulence) simulator. The conversion efficiency was studied analytically and experimentally. An experimental set-up was designed to mimic the expected strain frequency and amplitude for rotor blades. From a series of experiments, the efficiency of a piezoelectric EH at a typical rotor speed (0.2 Hz) was approximately 0.5%. The power requirement for sending one measurement (280 $\mu$J) can be achieved in 10 minutes. Designing a detection algorithm is challenging due to this low sampling rate. A new sensing approach-the timing of pulses from the transmitter-was introduced. This pulse timing, which is tied to the charging time, is indicative of the structural health. The SHM system exploits the inherent triple redundancy of the three blades. The timing data of the three blades are compared to discern an outlier, corresponding to a damaged blade. Two types of post-processing of pulses were investigated: (1) comparing the ratios of signal timings (i.e. transmission ratio); and (2) comparing the difference between signal timings (i.e. residuals). For either method, damage is indicated when the energy ratio or residual exceeds a threshold level. When residuals are used to detect damage, performance measures such as the false alarm rate and detection probability can also be imposed. The SHM algorithms were evaluated using strain energy data from a 2.5 MW wind turbine.