Tidal Steam Turbine Blade Fault Diagnosis Using Time-Frequency Analyses

Tidal Stream Turbines are developing renewable energy devices, for which proof of concept commercial devices are been deployed. The optimisation of such devices is supported by research activities. Operation within selected marine environments will lead to extreme dynamic loading and other problems. Further, such environments emphasise the need for condition monitoring and prognostics to support difficult maintenance activities. This paper considers flow and structural simulation research and condition monitoring evaluations. In particular, reduced turbine blade functionality will result in reduced energy production, long down times and potential damage to other critical turbine sub-assemblies. Local sea conditions and cyclic tidal variations along with shorter timescale dynamic fluctuations lead to the consideration of time-frequency methods. This paper initially reports on simulation and scale-model experimental testing of blade-structure interactions observed in the total axial thrust signal. The assessment is then extended to monitoring turbine blade and rotor condition, via drive shaft torque measurements. Parametric models are utilised and reported and a motor-drive train-generator test rig is described. The parametric models allow the generation of realistic time series used to drive this test rig and hence to evaluate the applicability of various time-frequency algorithms to the diagnosis of blade faults.