Daming Wang

Abstract

Wave tank model testing is a commonly used method to assess the performance of Wave Energy Converters (WECs). Wave data collected for testing can be obtained by different instruments (e.g. buoys, ADCP, and HF radar). Currently, the widely accepted way to re-create the wave conditions for WEC model testing is to obtain wave parameters from the wave data collected and apply them to a suitable generic wave spectrum, such as the JONSWAP spectrum or Pierson-Moskowitz spectrum, then reproduce it in the wave tank. By using this method, each wave condition is simplified to several wave parameters such as the significant wave height, wave peak period, etc. However, the parametric wave spectrum obtained by this method is just a simplified mathematical model that omits much useful wave information, such as the details of the wave spectrum and the wave directional information. At later development stages, there is a need to use site-specific complex wave conditions representative of the potential prototype deployment sites for model testing of the WECs. Today, with the development of advanced wave measurement instruments, such as the high-frequency radar system, the site-specific hourly/ half-hourly wave spectra can be obtained to provide the information to recreate the wave conditions in the wave tank in a much more accurate way. After obtaining numerous hourly/ half-hourly wave spectra, it is necessary to determine a certain number of sea states that can best represent the ocean environment of interest for WEC model testing. This thesis compared ten regrouping methods to obtain a small number of representative sea states from a large data set. It was found that the method based on the non-directional wave spectrum K-means clustering technique obtained the sea states with the highest representativeness regardless of the total data set used. The representative sea states were tested both numerically and physically using two different WEC designs, a point absorber and a 1:25 hinged-raft device for their power output performance. The results have shown that the representative sea states obtained from the method not only represented well the ocean environments but also represented the annual power output conditions of the WECs well regardless of the non-linearity. The error in the annual energy output predicted using representative sea states from different regrouping methods was within 1%. The error in the annual energy output predicted using the same regrouping method with a different number of groups was less than 5%.

Keywords

WEC, WEC-Sim, K-means method, HF radar, nonlinear

Document Type

Thesis

Publication Date

2022

Download

Share

COinS