Climate-Risk-Aware Offshore Wind Farm Layout Optimization Under Non-Stationary Wind Resources

Abstract

Offshore wind farms are typically optimized using a stationary historical wind rose, although development and operation span decades. This study presents CRA-WLO, a scenario-based screening framework coupling a piecewise-smooth engineering wake model, four stylized climate-resource modes, and a conditional value-at-risk (CVaR) penalty on layout-specific annual energy production (AEP) degradation. The controlled benchmark places 72 generic 15 MW turbines in a rectangular lease area. To quantify search variability, the supplied reconstructed run-history dataset contains 30 seeded run records for each of six optimization cases (180 records in total), with objective distributions, evaluation counts, restart statistics, and convergence reported. At α = 0.85, the 0.15-weight adverse mode exhausts the tail, so CVaR equals its shortfall. Compared with an aligned grid, the selected risk-aware best-found layout raises climate-weighted, adverse-mode, and present AEP by 4.6%, 1.5%, and 4.7%, respectively. Compared with the present-only best-found layout, it reduces present AEP by 6.7%, raises adverse AEP by 42.0%, and reduces normalized degradation from 42.9% to 13.1%. Sensitivity tests over α = 0.60–0.95, adverse-mode weights of 0.05–0.30, and speed, density, and directional-stress diagnostics preserve the central wake-redistribution mechanism within the stated approximations. Results remain conditional on a restricted candidate family and stylized stress modes and are not site-specific projections, certified global optima, or bankable energy-yield estimates.