A parallel genetic algorithm for continuous and pattern-free heliostat field optimization

The heliostat field of a solar power tower system, considering both its deployment cost and potential energy loss at operation, must be carefully designed. This procedure implies facing a complex continuous, constrained and large-scale optimization problem. Hence, its resolution is generally wrapped by extra distribution patterns or layouts with a reduced set of parameters. Griding the available surface is also an useful strategy. However, those approaches limit the degrees of freedom at optimization. In this context, the authors of this work are working on a new meta-heuristic for heliostat field opti- mization by directly addressing the underlying problem. Attention is also given to the benefits of modern High-Performance Computing (HPC) to allow a wider exploration of the search-space. Thus, a parallel genetic optimizer has been designed for direct heliostat field optimization. It relies on elitism, uniform crossover, static penalization of infeasible solutions and tournament selection.