Comparison of Wind Farm Layout Optimization Results Using a Simple Wake Model and Gradient-Based Optimization to Large-Eddy Simulations: Preprint

Paul Fleming, Jared Thomas, Andrew Ning

Research output: Contribution to conferencePaper

Abstract

Comparison of wind farm layout optimization results with simulations or wind tunnel tests is important because the models used during optimization use simplifying assumptions that can alter the design space. Some of the characteristics of the simple models may negatively influence the resulting layouts. In this paper we perform wind farm layout optimization and compare the resulting improvements to large eddy simulation (LES) results. We begin by describing the models used, including changes specific for use with gradient-based optimization. We then compare our version of the models with the original versions and large eddy simulation LES) data from previous publications. Using the models described, we performed gradient-based wind farm layout optimization, using exact gradients, and compare the results to LES simulations. The models predicted an improvement in annual energy production (AEP) of 7.4%, while the LES reported an AEP improvement of 9:9%. We conclude that the improvements found by optimizing with the simple models are not just an artifact of the models, but are real improvements.
Original languageAmerican English
Number of pages22
StatePublished - 2019
EventAmerican Institute of Aeronautics and Astronautics (AIAA) SciTech Forum - San Diego, California
Duration: 7 Jan 201911 Jan 2019

Conference

ConferenceAmerican Institute of Aeronautics and Astronautics (AIAA) SciTech Forum
CitySan Diego, California
Period7/01/1911/01/19

Bibliographical note

See NREL/CP-5000-74503 for paper as published in AIAA proceedings

NREL Publication Number

  • NREL/CP-5000-72924

Keywords

  • gradient-based optimization
  • large-eddy simulations
  • modeling
  • wakes
  • wind energy

Fingerprint

Dive into the research topics of 'Comparison of Wind Farm Layout Optimization Results Using a Simple Wake Model and Gradient-Based Optimization to Large-Eddy Simulations: Preprint'. Together they form a unique fingerprint.

Cite this