Modelling the Wind Farm Layout Optimization Problem Using Integer Programming

The wind farm layout optimization problem is addressed in this manuscript through the perspective of integer programming. A number of intrinsic modelling benefits are obtained by means of this discrete optimization approach, as for example, possibility to include turbines number and type selection within the program, inclusion of terrain-dependent cost functions, incorporation of any shape for available and forbidden areas or unified optimization with electrical and control systems, among others. Three integer linear programs are presented and compared respect to a highly non-linear true annual energy production metric. All the formulations are wrapped in a neighborhood search heuristic for both accelerating convergence and to correct the optimization search. Two of the models support explicit power curve modelling following discrete sampling techniques of this function, while the last model focuses on minimizing a measure closely related to total wind speed deficit, ignoring the power curves. Numerical results on a benchmark problem indicate that the power-curve-free model outperforms the power-based models in terms of final solution quality.