Application of artificial intelligence to urban wind energy. (15th June 2021)

• Record Type: Journal Article
• Authors: Higgins, Stéphanie; Stathopoulos, Ted
• Is Part Of: Building and environment. Volume 197(2021)
• Publication Date: 2021-06-15
• Subjects: Urban wind energy -- Artificial intelligence modeling -- Computational fluid dynamics -- Expert system -- Artificial neural network -- Wind tunnel; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2021.107848
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: Optimal implementation of urban wind energy contributes towards the development of sustainable cities. This paper focuses on generating an adequate database to use with artificial intelligence (AI) tools to improve the generation of urban wind energy. The paper presents wind tunnel results for square, rectangular, U-shaped, T-shaped, L-shaped buildings and some measurement points in various city configurations. Moreover, the effect of building shapes on turbine street level locations was elaborated using validated CFD literature results on pedestrian level wind conditions. Using these results and literature review from the past decade, a decisional flow chart approach was developed, allowing a preliminary assessment of the modification of upstream wind velocities due to urban parametric conditions. Expert and artificial neural network (ANN) systems were built and tested on city configurations with their results compared with those from wind tunnel measurements. The ANN system shows better predictive values than the expert system, with up to 99% success rate. AI programs with the decisional flow chart approach may be used for the identification and assessment of potential turbine locations to maximize the production of urban wind energy. Highlights: A database of 157 different cases of potential wind turbine locations in urban environment has been constructed using wind tunnel testing and literature review. AI systems such as expert systems and artificial neural networks, may be used for preliminary assessment of predictive modifications of wind velocities in cities. Expert system yields in a success rate above 84% and 93 % for artificial neural network, for a simple city configuration compared to wind tunnel results. A decisional flowchart approach is elaborated, with a success rate above 84% for a simple city configuration.