Ocean Power Technologies (OPT) designs and builds wave energy converters (WECs) under the brand name PowerBuoy® to generate electrical power from ocean waves. An important component of as-sessing a candidate application is predicting the amount of generated power, which requires information about the deployment site wave climate. Historically, OPT has used a joint probability distribution of wave height and period based on an idealized spectral shape with the statistics for the bin center, but a more realistic spectral description has been shown to improve power predictions. While OPT can estimate power by running a simulation for every hourly spectrum in a multi-year timeseries, a more compact spectral representation is sought. Here measured spectra with similar statistics are averaged and used as pre-diction inputs; the results are compared to predictions based on ideal-ized spectra. To gain insight into how prediction quality might depend on device resonant period or site dominant period, examples will be shown for various locations and device sizes using National Data Buoy Center wave spectra measurements.
Three PowerBuoy designs are considered here: a smaller design called APB-350-A2 (or the A2 for short), a midsize PB10, and a larger PB40. Since all these designs are axisymmetric, wave direction does not affect power generation and is not addressed here. Three locations are used: Oregon (high energy, long period), New Jersey (moderate energy and period), and the Gulf of Mexico (low energy, short period). For all three locations, wave measurements have been collected for many years from National Data Buoy Center (NDBC) buoys. The study metric is the ratio of (a) predicted power if an average of measured spectra is used as the model input to (b) predicted power if idealized spectra are used. The input for (a) is an average of all NDBC hourly spectra which have similar significant wave height (Hs) and average (mean) wave period (Ta = 2πm0/m1)values, so that they fall within the same Hs and Ta bins (Figure 1). Note that the resulting average spectrum can have a different Hs than the bin center: in Figure 1, it is 0.65m, which would result in higher predicted power output. The input for (b) is the ideal Bretschneider spectra (Tucker, 1991) with Hs and Tα correspond-ing to the bin center. The Bretschneider spectra is chosen because it is fully defined by the Hs and Tα values, which are directly available from the NDBC data.