NWC REU 2014
May 21 - July 30



Photo of author

Wind Farm Layout Optimization Problem by Modified Genetic Algorithm

Grant Williams, April Taylor, and Dr. Renee McPherson


What is already known:

  • Wind energy is the fastest sector of the energy field, but it has a lot of technical complications that need to be worked out.
  • The wake turbine interaction is a very difficult computation problem that can only be modeled with heuristic models.
  • Minimizing turbine-wake interactions is a difficult, but necessary part of planning a wind farm and a model to do so efficiently is needed.

What this study adds:

  • A modified genetic algorithm is presented that is capable of modeling a large number of turbines relatively quickly.
  • The model produces relatively optimal solutions to the wind farm layout optimization problem.
  • The model presented produced very promising results and is robust enough to incorporate more complex wake models or other parameters.


Wind energy is a rapidly growing source of energy for the United States, but there are still technical problems to resolve before it can become the major source of energy production. One of the biggest problems with land based wind farms is minimizing wake- turbine interactions within a constrained space and thus maximizing power. When wind blows through a turbine’s blades, a choppy, turbulent wake is created that interferes with the ability of nearby turbines to produce power. Research has already been done on finding ways to model wind farms and place the turbines in a way that minimizes wake-turbine interactions, but current methods are either computationally intensive or require proprietary software. I present a modified genetic algorithm that is able to pro- duce optimized results in a relatively short amount of computation time. The algorithm presented is able to make use of the computation power of graphical processing units and multiple processors and by doing so produces results much quicker than an algorithm run sequentially on a single processor.

Full Paper [PDF]