A Tight Upper Bound for Grid-Based Wind Farm Layout Optimization

This paper uses the method developed by Billionnet et al. (1999) to obtain tight upper bounds on the optimal values of mixed integer linear programming (MILP) formulations in grid-based wind farm layout optimization. The MILP formulations in grid-based wind farm layout optimization can be seen as linearized versions of the 0-1 quadratic knapsack problem (QKP) in combinatorial optimization. The QKP is NP-hard, which means the MILP formulations remain difficult problems to solve, especially for large problems with grid sizes of more than 500 points. The upper bound method proposed by Billionnet et al. is particularly well-suited for grid-based wind farm layout optimization problems, and was able to provide tight optimality gaps for a range of numerical experiments with up to 1296 grid points. The results of the numerical experiments also suggest that the greedy algorithm is a promising solution method for large MILP formulations in grid-based layout optimization that cannot be solved using standard branch and bound solvers.