Session: 09-02-03 WEC performance analysis

Paper Number: 81155

81155 - Accurate WEC Power Estimation for Multi-Modal Wave Spectra 

The power-generation capability of wave energy converters (WECs) is generally dependent on the wave period and height. When the power performance of a WEC is analyzed via simulations, idealized wave spectra such as Bretschneider or JONSWAP (Joint North Sea Wave Project) are used to estimate the power output of the WEC under a range of sea states as determined by the wave period (usually energy period) and significant height. Then, from those results, a power characteristic matrix is formed, which can thereafter be used to estimate the power generation capability at other sites of interest using the site joint probability distribution of sea-state occurrences. However, it has been shown that the power performance of a WEC is also dependent upon the shape of the spectra. That is to say, that the power generated under a specific sea state with a Bretschneider spectrum would be different from the power the WEC would generate under the same sea state with a JONSWAP spectrum with a gamma of, for instance, 2.

The problem of providing power projections becomes even more complicated when we consider that, at sites in open oceans, waves may have swell and wind-generated components, each of which has its own spectrum. Under such a circumstance, considering only the wave period and significant height can result in erroneous power numbers. The larger the probability of occurrence of dual- or multi-hump spectra is, the larger the error in the power numbers can be.

In this paper, we propose that, for more accurate power predictions, wave spectra should be divided into swell (long) and wind-generated (short) components. The power-generation capability corresponding to each component can then be estimated separately, and the largest one can be taken as the power number associated with that sea state. To verify the efficacy of this approach on the OPT PB3 PowerBuoy, we picked a point in the Pacific Ocean offshore Chile, extracted the associated time history of wave spectra at that site for the entire 2013, one spectrum for every 3 hours, from the IFREMER (Institut français de recherche pour l'exploitation de la mer) WW3 (WaveWatch III) hindcast data. Each spectrum was decomposed to its swell and wind-generated wave components using functions we developed in MATLAB using the WAFO toolbox developed at Lund University, Sweden. The details of the spectra decomposition process will be discussed in the paper.

Up to 3 simulations—one for each of the undecomposed spectrum, swell component, and wind-generated component—were run for each 3-hour period on the operational model of the PB3 in time domain in OrcaFlex, which is a well-known commercially available software package for hydrodynamics modeling and simulation. In total, 7931 20-minute simulations were run, and the simulation results were postprocessed to extract the force and speed time histories, which were then put through OPT’s calibrated power take-off (PTO) mechanical-to-electrical power loss model.

The results show that, by applying our spectra decomposition technic to the problem of power projection of the PB3, the annual mean relative error incurred is only -0.57%, with the negative sign indicating an underestimation of power-generation capability, while the maximum monthly mean value of relative error in absolute value is under 2.5%. More specifically, monthly means of relative error are within -2.27% and 0.19%. Hence, the results show great overall similarity between the power estimate calculated from the undecomposed spectra and the maximum of the power estimates resulting from to the component spectra.

It is our expectation that the same approach can be applied to other types of WECs.

Presenting Author: Kourosh Parsa Ocean Power Technologies, Inc.

Authors:

Kourosh Parsa Ocean Power Technologies, Inc.
Mark Kim Ocean Power Technologies, Inc.
Neil Williams Ocean Power Technologies, Inc.