Flexible cable routing framework for wind farm collection system optimization

The wind farm cable routing problem is an important element of offshore wind power plant design optimization. The prohibition of cable crossings places it in a category of its own compared to related combinatorial optimization problems. A novel framework that systematizes implementing this constraint with benefits to solution quality and computational cost is presented. It reframes the problem with separate topological and embedding layers and relies on the insight that tree graphs are always embeddable on the plane. This reduces the number of constraints related to crossings and enables the use of non-straight connections without changing the problem size. Gains in solution quality are attained because the search space retains near-optimal solutions commonly excluded by crossing constraints in state-of-the-art approaches. Reductions in computational cost are obtained by concomitantly pruning unlikely connections from the search space. Moreover, the framework facilitates the application of vehicle-routing heuristics (unaware of crossings) to the cable routing problem by leveraging the empirical observation that most crossings involve feeder cables. Meta-heuristic- and branch-and-cut-based optimizers are used to demonstrate the framework, which is optimizer-agnostic. Comparative results with solutions from the literature show decreases in total cable length of 1.6 to 17%, while run times are at least 60-fold reduced. The framework's implementation is made fully available through an open-source license. The improved computational efficiency and solution quality achieved with the developed framework make it a valuable tool for optimizing offshore wind farm collection systems, potentially reducing cost, increasing feasibility, and allowing for faster design iterations.