Hybrid metaheuristics for the multiobjective offshore wind farm layout optimization problem
OWFLOP, Multiobjective Optimization, Hybrid Metaheuristics, Evolutionary Computation.
Given the growing concern over the climate crisis and the international urgency for the transition to renewable energy sources, advances in academic and industrial research on wind farms have intensified and in recent years, most notably in offshore wind farms, due to their higher energy potential compared to onshore alternatives. As arrays with increasing numbers of turbines and more costly equipment must be designed, the need arises to consider space optimization in order to minimize foundation, installation and maintenance costs of farm components (which scale up to hundreds of millions of euros), to avoid obstacles, and to better exploit the regional wind profile. Thus, the Offshore Wind Farm Layout Optimization Problem (OWFLOP) is defined. In addition to being NP-hard, this problem is notoriously complex due to its high number of modeling variables and the computational cost of evaluating its objective functions. Due to these limitations, the OWFLOP is typically addressed through heuristic and metaheuristic techniques, some of which have already been established in the literature as more efficient. In this work, the biobjective OWFLOP is addressed, considering the minimization of total turbine installation cost and the maximization of annual energy production. To this end, some classical multi-objective metaheuristic approaches (NSGA-II, SPEA2, PLS) were evaluated against a local search approach and two hybridizations proposed with mechanisms adapted to the specific characteristics of the problem, on a new benchmark set proposed by Cazzaro and Pisinger.