On the Development of an Efficient Surrogate Model for Predicting Long-Term Extreme Loads on a Wave Energy Converter

Abstract

Accurate prediction of long-term extreme loads is essential for the design of wave energy converters (WECs), but it is also computationally demanding due to the low probabilities associated with their occurrence. While a full long-term probabilistic analysis using integration over all sea states or Monte Carlo Simulation (MCS) may be used, these methods can be prohibitively expensive when individual response simulations are complex and time-consuming. The application of polynomial chaos expansion (PCE) schemes to allow the propagation of uncertainty from the environment through the stochastic sea surface elevation process and ultimately to WEC extreme load response prediction is the focus in this study. A novel approach that recognizes the role of long-term ocean climate uncertainty (in sea state variables like significant wave height and spectral peak period) as well as short-term response uncertainty arising from irregular wave trains is presented and applied to a single-body point-absorber WEC device model. Stochastic simulation results in time series realizations of various response processes for the case-study WEC. We employ environmental data from a possible deployment site in Northern California (NDBC 46022) to assess long-term loads. MCS computations are also performed and represent the “truth” system against which the efficiency and accuracy of the PCE surrogate model is assessed. Results suggest that the PCE approach requires significantly less effort to obtain comparable estimates to MCS.