Heuristics-based design and optimization of offshore wind farms collection systems

A parallel-based algorithmic framework for automated design of Offshore Wind Farms (OWF) collection systems is proposed in this paper. The framework consists basically on five algorithms executed simultaneously and independently, followed by a combined analysis aiming to generate the best results in terms of different objective functions. The main inputs of the framework are the location coordinates of the Wind Turbines (WT) and the Offshore Substation (OSS), wind power production time series, and the set cables considered for the collection system design. Four heuristics and one metaheuristic algorithm are considered. The heuristics are based on modified versions of well-known graph-theory algorithms: Kruskal (KR), Prim (PR), Esau-Williams (EW), and Vogel's Approximation Method (VAM); all of them coded in a unified framework with quartic time complexity. The metaheuristic is built upon a Genetic Algorithm (GA) designed using a hierarchical-restricted penalization system. Comparisons between all of these methods are performed from different perspectives, taking into consideration the particular constraints treated for OWF practical applications. In general, primals from heuristics lead to faster and better results when only a single cable is available, and provide collection systems with lower electrical power losses for multiple cables choice, whilst the GA shows better results when the initial investment is prioritized and several cable types are considered.