Combining statistical modelling of degradation with indirect measures of the degradation

Abstract

The main idea of this thesis is to investigate the literature about the models that were developed to link the physical degradation with indirect measures like vibration. Wear is the chosen degradation mechanism for the project. Fast Fourier Transformation and Wavelet Transformation seem to be the most promising techniques in order to deal with the objective of finding the models. After a sufficient amount of literature review, it appears that no one has actually developed a physical degradation model based on vibration signals. There are a lot of papers discussing Fast Fourier Transform and Wavelet Transform to assess the fault diagnostics but they all lack in developing any physical model that can link the physical degradation with vibration signals. The other possibility is to set up a laboratory to develop the understanding of vibration signals based on conducting experimental analysis but it is beyond the scope of this project.

Taking the above into consideration, and due to experimental limitations, further objectives which depend upon the main idea have been changed accordingly. In order to deal with wear, Arhcard‘s wear law has been used extensively. But as far as wear in gearbox of offshore wind turbine is concerned, not much work has been done in relating Archard‘s wear law to the wear propagating in the gears. Due to the simplicity of Archard‘s wear law, it has been chosen to develop a wear model for gears of offshore wind turbine. Consequently, the wind load study has been performed in order to see the effect of wind load on the wear in a gearbox, because in offshore the wind load is comparatively larger as compared to onshore. The offshore wind has both the temporal and the spatial variation with the stochastic nature therefore modelling the wind speed is not an easy task. An attempt has been made to model wind speed based on the data obtained from National Data Buoy Center. The Archard‘s wear model has been analysed based upon the developed wind model to see the behaviour of wear in gears. Uncertainties such as epistemic and aleatory present in both, wind model and wear prediction model, have been discussed. Finally, maintenance strategies have been developed by applying Monte Carlo simulation to the proposed wear prediction model for gears.