Fault Diagnosis of Floating Wind Turbine Drivetrain- Methodologies and Applications
MetadataVis full innførsel
- Institutt for marin teknikk 
Increasing demand for renewable energy is driving innovation and attracting many research resources all over the world. Wind energy is one of the most promising sources of renewable energy. Wind turbine has been in use for several centuries. In recent decades, the renewable energy, wind and solar, had the highest growth rate comparing to other sources mostly due to environmental effects. The trend in wind energy is to go toward floating wind turbine to access more robust and strong wind profiles. However, application of the floating offshore wind turbine is not mature yet, and there are still many challenges and questions need to be answered. The primary goal is to have a safe and optimal operation during the turbine lifetime and to reduce the cost of energy to be compatible with other sources. Operation and maintenance cost plays a significant role in this regard particularly for offshore wind. Among different components of a wind turbine, gearbox failures cause the highest wind turbine downtime. Therefore, this thesis tried to address the floating wind turbine gearbox fault diagnosis by simulation-based case studies. The wind turbine is a 5-MW, spar type floating, horizontal axis, 3-bladed up-wind, variable-speed and pitch regulated turbine. The global analysis was conducted using the aero-hydro-servo-elastic code SIMORIFLEX-AeroDyn. The obtained forces and moments from global analysis were then applied to a high fidelity gearbox model in SIMPACK to investigate the dynamic of the gearbox. The gearbox is a 5-MW high speed and has four points support. The main bearing and one bearing inside the gearbox were investigated using both time and frequency-based analysis for fault detection, estimation and isolation. The choice of the bearings was based on their fatigue life as well as their effect on other gearbox components. It was demonstrated how the statistical hypothesis tests can effectively detect, estimate the wear (a gradual change) in the bearings. Moreover, two new theories related to statistical/stochastic analysis have been extended in this thesis. The first one is the derivation of the joint distribution of excursion duration and amplitude of a narrow-band Gaussian process, and related marginal distributions. The work was validated using both ideal narrow-band time domain simulations and also some real sea states. The second work is the extension of the statistical hypothesis test; that is, to integrate the frequency information into time-domain statistical hypotheses tests. It should be noted that the applications of the developed theoretical works are not limited to wind turbines. It ranges from signal processing, fault diagnosis, renewable energy, to marine applications. The dissertation is based on the articles published (4 papers) or to be submitted (1 paper) to scientific journals. As a brief summery, the wind turbine modeling and condition monitoring explained; then, the main finding of simulation-based case studies and the theoretical developments are presented, and finally conclusion is drawn.