A classification system for global wave energy resources based on multivariate clustering. (15th March 2020)

• Record Type: Journal Article
• Authors: Fairley, Iain; Lewis, Matthew; Robertson, Bryson; Hemer, Mark; Masters, Ian; Horrillo-Caraballo, Jose; Karunarathna, Harshinie; Reeve, Dominic E.
• Is Part Of: Applied energy. Volume 262(2020)
• Publication Date: 2020-03-15
• Subjects: Wave energy -- Resource assessment -- Global -- Numerical model -- K-means clustering; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2020.114515
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: A device-agnostic classification of global wave energy resources is presented. Classification is conducted by applying the k -means algorithm to ECMWF ERA5 data. Six classes are returned ranging from enclosed seas to high energy open coasts. Geographic and parameter space distributions match past regional scale assessments. New devices should be optimised for moderate energy, low variability areas. Abstract: Better understanding of the global wave climate is required to inform wave energy device design and large-scale deployment. Spatial variability in the global wave climate is analysed here to provide a range of characteristic design wave climates. K-means clustering was used to split the global wave resource into 6 classes in a device agnostic, data-driven method using data from the ECMWF ERA5 reanalysis product. Classification using two sets of input data were considered: a simple set (based on significant wave height and peak wave period) and a comprehensive set including a wide range of relevant wave climate parameters. Both classifications gave resource classes with similar characteristics; 55% of tested locations were assigned to the same class. Two classes were low energy, found in enclosed seas and sheltered regions. Two classes were moderate wave energy classes; one swell dominated and the other in areas with wave action often generated by more local storms. Of the two higher energy classes; one was more often found in the northern hemisphere and the other, most energetic, predominantly on the tips of continents in the southern hemisphere. These classes match existing regional understanding of resource. Consideration of publicly available device power matrices showed good performance was primarily realised for the two highest energy resource classes (25–30% of potential deployment locations); it is suggested that effort should focus on optimising devices for additional resource classes. The authors hypothesise that the low-risk, low variability, swell dominated moderate wave energy class would be most suitable for future exploitation.