Topology optimization of wind farm layouts

TL;DR

This paper introduces a density-based topology optimization method for wind farm layout design, maximizing energy production while respecting turbine spacing and count constraints, and efficiently finding innovative layouts.

Contribution

It presents a novel application of topology optimization with explicit gradient calculation for wind farm layout design, enabling efficient and effective layout optimization.

Findings

Successfully optimized wind farm layouts with minimal computational resources.

Achieved layouts that are significantly different from intuitive configurations.

Demonstrated the method's efficiency and effectiveness through numerical results.

Abstract

A novel approach for the solution of the wind farm layout optimization problem is presented. The annual energy production is maximized with constraints on the minimum and maximum number of wind turbines placed, and on the minimum spacing between the wind turbines. The proposed approach relies on a density-based topology optimization method, where continuous density variables varying between zero and one are assigned to each potential wind turbine location. A wind turbine exists if its associated variable equals one, otherwise it does not exist if the associated variable is zero. Intermediate values of the density variables are penalized with interpolation schemes traditionally used in the context of multi-material structural topology optimization. The penalized intermediate values of the design variables become uneconomical and the optimization algorithm is implicitly pushed towards a preference of crisp 0-1 final values. The optimization problem is solved with a gradient-based algorithm based on first-order information. Because of the proposed problem formulation, the functions involved are formulated explicitly in terms of the design variables and their analytical gradients can be calculated directly. The numerical results highlight the capability of the proposed approach in finding far from intuitive wind farm layouts with small computational resources and time.