A comparison of optimizers in a unified standard for optimization on wind farm layout optimization. (1st February 2021)

• Record Type: Journal Article
• Authors: Croonenbroeck, Carsten; Hennecke, David
• Is Part Of: Energy. Volume 216(2021)
• Publication Date: 2021-02-01
• Subjects: Wind energy -- WFLO -- Wind farm layout optimization -- Optimization -- NP-Hard -- R package -- Wake models -- Open data -- Genetic algorithm -- Benchmark; C61 -- C65 -- Q42; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2020.119244
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Wind Farm Layout Optimization (WFLO) is a vivid field of research dealing with the difficult problem of optimally arranging a given number of wind turbines inside a local area (wind farm). There are several types of objective functions, varying optimization strategies, different sets of underlying data, assumptions and simplifications to the problem, among other issues making fair comparisons of problem solving techniques challenging. We discuss a new unified framework that provides highly accurate data, a modular approach to economically driven objective functions, and a unified and fair benchmark for mathematical optimizers that are easily plugged into that framework. Finally, we provide an exemplary work flow and use it to show a comparison study of optimizing techniques within the framework, focussing on types of optimizers frequently used in this field. Highlights: Using R package "wflo", we show a typical WFLO work flow. We run a selection of optimizers on a standardized WFLO environment. Genetic algorithms and simulated annealing perform best, but slow. WFLO researchers worldwide should use WFLO to have their contributions comparable.