Performance and Condition Monitoring of Tidal Stream Turbines

Roger I. Grosvenor, Paul W. Prickett, Carwyn Frost, and Matthew Allmark
Submission Type: 
Full Paper
phmce_14_072.pdf1.32 MBMay 29, 2014 - 5:13am

Research within the Cardiff Marine Energy Research Group (CMERG) has established a series of generic design guidelines for the developing commercial deployment of Tidal Stream Turbines (TST). This paper briefly reviews the contributions of the mathematical modelling studies to this field. The mathematical models combine Computational Fluid Dynamics (CFD), structural Finite Element Analysis (FEA) to provide Fluid-Structure-Interaction (FSI) results. Non-dimesionalised power and thrust curves, along with flow visualisations, are produced for a variety of configurations and flow conditions. In this paper the total axial thrust and, in particular, any cyclic variations in this measurable parameter are analysed and discussed.
The mathematical models are validated via the testing of 0.5 m diameter turbines in a water flume facility at Liverpool University. A dataset of results was available for the reported tests, for which a three blade turbine was used, with a constant plug flow of 0.94 m.s-1 and at a range of controlled conditions within the power curve profiles. The recorded signals were angular velocity, servo motor current (used to oppose flow generated motion and hence to estimate generated power) and the overall axial thrust. The tests reported in this paper are split between an ‘optimum’ setup (three identical blade angles) and an ‘offset’ setup. For the latter tests one of the three blades was deliberately set at other than its optimum pitch angle.
The analysis approach reported here aims to extract any rotational effects from the small cyclic variations observed on the axial thrust signals. These are derived from a strain-gauged force block at the attachment point for the turbine support tube. The preliminary results analysis demonstrates that the captured signals are far from ideal for frequency-based methods. The acquired sample rate is low, at 47.6 Hz, and for any individual set of results some A/D conversion quantisation effects are observable. The recorded angular velocity signals have an even lower sample rate and typical results show ± 2.5% velocity variations during a 90 second test duration.
Initially, traditional frequency spectrum plots obtained are presented with components, for any given set of test conditions, observed at the rotational frequency (ωr) and at either or both 2.ωr and 3.ωr. In this paper alternative time-frequency methods are compared. In particular synchro-squeezing methods, implemented within Matlab, are assessed. The results demonstrate the flexibility of this method. Within a range of selectable parameters for the algorithm there is shown to be a balance between averaged spectrum results and time varying thrust values. The latter are correlated to the ± 2.5% angular velocity variations.
Mathematical modelling results are also presented to confirm the expectations for these frequency components. The potential for using the results obtained at these frequencies in a condition monitoring regime is discussed. This conclusion is made in light of the signal deficiencies. Anticipated refinements to the frequency analysis methods are discussed along with an extension to include prognostic predictions. The operational parameters that apply uniquely to each full-scale turbine installation are also discussed.
It is observed that the measurements from the turbine supporting structure are more easily made, and are expected to form a constituent part of integrated monitoring systems. The paper concludes with a brief description of the next generation of scale model TST that is about to be deployed for further testing. The Liverpool University water flume now facilitates profile flow and/or surface wave testing.

Publication Year: 
Publication Volume: 
Publication Control Number: 
Page Count: 
Submission Keywords: 
renewable energy
condition monitoring
Tidal Stream Turbines
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed