Bilateral Gaussian Wake Model Formulation for Wind Farms: A Forecasting based approach

Harsh S. Dhiman*, Dipankar Deb, Aoife M. Foley

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)
62 Downloads (Pure)


Optimal placement of turbines in a wind farm is a major challenge where the wake effect reduces the effective wind power capture. Wind speed prediction is essential from a reliability point of view. In this article, a bilateral wake model which is derived from two benchmark models, namely, Jensen's and Frandsen's variation is used for studying the performance of far-end wakes. A prediction based approach is formulated wherein the inputs to the classical SVR model are based on the two benchmark models and the proposed bilateral Gaussian wake model. Wind speed is predicted for upstream turbines of two wind farm layouts (5-turbine and 15-turbine). Further, to observe the impact of input dimensionality, two techniques: (i) Grey relational analysis (GRA) and (ii) Neighborhood component analysis (NCA), are considered. Results reveal that for a wind site WBZ tower, NCA outperforms GRA by 36.48%, 34.0% and 7.03% for Jensen's, Frandsen's and bilateral wake model respectively. When compared to the two benchmark models for both the techniques (GRA and NCA), the prediction performance of bilateral wake model is superior. Overall, it is observed that the feature selection tools like GRA and NCA improve the wind speed prediction accuracy in the presence of wind wakes.

Original languageEnglish
Article number109873
JournalRenewable and Sustainable Energy Reviews
Early online date03 May 2020
Publication statusPublished - Jul 2020


  • Feature selection
  • Gaussian model
  • Neighborhood component analysis
  • Wake effect
  • Wind speed prediction

ASJC Scopus subject areas

  • Renewable Energy, Sustainability and the Environment


Dive into the research topics of 'Bilateral Gaussian Wake Model Formulation for Wind Farms: A Forecasting based approach'. Together they form a unique fingerprint.

Cite this