Solver-free heuristics to retrieve feasible points for offshore wind farm collection system

Abstract

A set of solver-free heuristics for the offshore wind collection system problem is presented. Currently, algorithms of this type are not able to cope with relevant constraints when explicitly minimizing for investment cost. The first algorithm is a two-step decision process of which the output is a tree network satisfying cable thermal limits, but vulnerable to violate planarity constraints. A cable crossings repair heuristic is then introduced in order to fix infeasible points from the first heuristic. Finally, the negative cycle cancelling refining heuristic takes over to improve feasible points. The latter iteratively swaps connections, intending to find cycles with negative costs to reduce investment expenditure. The applicability of the workflow is empirically demonstrated through a set of large-scale real-world offshore farms. Results indicate that: (i) feasible points can be retrieved in a computing time of seconds; and (ii) warm starting can help global optimization solvers to converge significantly faster for hard problems.