Abstract
When dealing with the estimation of the power production of Wave Energy Converters (WECs), from either physical laboratory or numerical models, or measured data from real sea testing of prototype, an accurate representation of the wave spectra at the target location is important for accurate power production estimates. This due to the typically highly frequency-dependent response of the WECs. Depending on the location and other factors, real sea spectral shapes may deviate significantly from standard spectra such as the JONSWAP, often assumed in such estimations. Here, an alternative approach for defining representative spectra at a particular location, based on real sea wave measurements, is described, which allows a more accurate characterisation of the WEC performance.
The analysis described here has been performed on wave data from a buoy at the location of DanWEC, approximately 1 nautical mile north of the Port of Hanstholm in Denmark. Principal Component Analysis (PCA) is used as a tool for both the analyses of the variance of the wave spectra, and for dimensionality reduction, before further analyses to find representative wave spectra. In this paper a PCA on approximately five years of 1D wave spectra records shows that approximately 90% of the variance can be explained with only five parameters. The reduced number of spectral parameters from this method are used to similar spectra together, with the k-means algorithm. The mean spectrum of each group can be used as representative spectra for physical laboratory testing or numerical modelling. An advantage of this approach is that no assumption is made about the shape of the spectra.
The analysis described here has been performed on wave data from a buoy at the location of DanWEC, approximately 1 nautical mile north of the Port of Hanstholm in Denmark. Principal Component Analysis (PCA) is used as a tool for both the analyses of the variance of the wave spectra, and for dimensionality reduction, before further analyses to find representative wave spectra. In this paper a PCA on approximately five years of 1D wave spectra records shows that approximately 90% of the variance can be explained with only five parameters. The reduced number of spectral parameters from this method are used to similar spectra together, with the k-means algorithm. The mean spectrum of each group can be used as representative spectra for physical laboratory testing or numerical modelling. An advantage of this approach is that no assumption is made about the shape of the spectra.
Original language | English |
---|---|
Title of host publication | 10th ewtec 2013 European Wave and Tidal Energy Conference Series : Proceedings of the 10th European Wave and Tidal Energy Conference |
Editors | Peter Frigaard, Jens Peter Kofoed, AbuBakr S. Bahaj, Lars Bergdahl, Alain Clément, Daniel Conley, Antonio F. O. Falcão, Cameron MacLeod Johnstone, Lucia Margheritini, Ian Masters, António José Sarmento, Diego Vicinanza |
Number of pages | 5 |
Place of Publication | Aalborg |
Publisher | Technical Committee of the European Wave and Tidal Energy Conference |
Publication date | 2013 |
Publication status | Published - 2013 |
Event | European Wave and Tidal Energy Conference - Aalborg, Denmark Duration: 2 Sept 2013 → 5 Sept 2013 Conference number: 10 |
Conference
Conference | European Wave and Tidal Energy Conference |
---|---|
Number | 10 |
Country/Territory | Denmark |
City | Aalborg |
Period | 02/09/2013 → 05/09/2013 |
Series | European Wave and Tidal Energy Conference Series |
---|---|
Number | 10 |
Bibliographical note
The proceedings is published on a usb.Keywords
- Wave energy converters
- WECs
- Power production
- Wave measurements