Session: 09-05-04 Wave Energy: Power Take Off Systems

Submission Number: 180640

Uncertainty and Sensitivity in Wave Energy Converter Arrays: Sparse Polynomial Chaos Expansion of Layout–PTO–Wave Effects 

Designing wave energy converter (WEC) arrays demands simultaneous choices about spatial configuration, power-take-off (PTO) settings, and the incident wave climate. This paper quantifies the relative contribution of these three factor groups to array performance for a seven-device system using non-intrusive Polynomial Chaos Expansion (PCE). A linear hydrodynamic model with full device–device coupling provides training data for total absorbed power, array interaction factor, and representative structural load metrics under directional JONSWAP seas. Uncertain inputs include layout parameters (longitudinal spacing, transverse offset, and global yaw), PTO parameters (linear damping and stiffness about the dominant mode), and wave parameters (significant height, peak period, mean direction, and spreading). We construct sparse PCE surrogates (up to third order, selected via LARS) and derive variance-based global sensitivity indices (first-order, total-order, and selected second-order Sobol’ terms).

Results for the 7-WEC case study show that wave parameters—especially HsH_sHs​, TpT_pTp​, and incidence angle—dominate variability in annual energy, while PTO damping exerts the leading controllable influence on both power and loads. Spatial configuration contributes materially through direction-dependent interaction effects: longitudinal spacing and global yaw govern constructive/destructive interference, and their interaction with wave direction is one of the largest second-order effects. The PCE surrogates reproduce high-fidelity responses with low error, enabling rapid “what-if” exploration and robust design screening.

The proposed framework provides a transparent attribution of performance drivers across layout, PTO, and wave climate for multi-device arrays. Practically, it indicates where to focus tuning effort (PTO), which layout levers carry the most value under directional uncertainty (spacing and yaw), and how to reduce design space by prioritizing the variables that explain most of the variance.

Presenting Author: Avni Jain Delft University of Technology

Presenting Author Biography: A Post-doctoral researcher at Delft University of Technology utilizing machine learning methods to study wave energy converter arrays.

Authors:

Avni Jain Delft University of Technology
Jian Tan Delft University of Technology
George Lavidas Delft University of Technology