Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Combining computational fluid dynamics and neural networks to characterize microclimate extremes : Learning the complex interactions between meso-climate and urban morphology

Javanroodi, Kavan LU ; Nik, Vahid M. LU orcid ; Giometto, Marco G. and Scartezzini, Jean Louis (2022) In Science of the Total Environment 829.
Abstract

The urban form and extreme microclimate events can have an important impact on the energy performance of buildings, urban comfort and human health. State-of-the-art building energy simulations require information on the urban microclimate, but typically rely on ad-hoc numerical simulations, expensive in-situ measurements, or data from nearby weather stations. As such, they do not account for the full range of possible urban microclimate variability and findings cannot be generalized across urban morphologies. To bridge this knowledge gap, this study proposes two data-driven models to downscale climate variables from the meso to the micro scale in arbitrary urban morphologies, with a focus on extreme climate conditions. The models are... (More)

The urban form and extreme microclimate events can have an important impact on the energy performance of buildings, urban comfort and human health. State-of-the-art building energy simulations require information on the urban microclimate, but typically rely on ad-hoc numerical simulations, expensive in-situ measurements, or data from nearby weather stations. As such, they do not account for the full range of possible urban microclimate variability and findings cannot be generalized across urban morphologies. To bridge this knowledge gap, this study proposes two data-driven models to downscale climate variables from the meso to the micro scale in arbitrary urban morphologies, with a focus on extreme climate conditions. The models are based on a feedforward and a deep neural network (NN) architecture, and are trained using results from computational fluid dynamics (CFD) simulations of flow over a series of idealized but representative urban environments, spanning a realistic range of urban morphologies. Both models feature a relatively good agreement with corresponding CFD training data, with a coefficient of determination R2 = 0.91 (R2 = 0.89) and R2 = 0.94 (R2 = 0.92) for spatially-distributed wind magnitude and air temperature for the deep NN (feedforward NN). The models generalize well for unseen urban morphologies and mesoscale input data that are within the training bounds in the parameter space, with a R2 = 0.74 (R2 = 0.69) and R2 = 0.81 (R2 = 0.74) for wind magnitude and air temperature for the deep NN (feedforward NN). The accuracy and efficiency of the proposed CFD-NN models makes them well suited for the design of climate-resilient buildings at the early design stage.

(Less)
Please use this url to cite or link to this publication:
author
; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Air temperature, CFD simulations, Extreme microclimate conditions, Neural networks, Urban morphology, Wind speed
in
Science of the Total Environment
volume
829
article number
154223
publisher
Elsevier
external identifiers
  • scopus:85126576613
  • pmid:35245539
ISSN
0048-9697
DOI
10.1016/j.scitotenv.2022.154223
project
Flexible energy system integration using concept development, demonstration and replication
language
English
LU publication?
yes
id
88858723-49e0-4213-bf8a-94f14d578060
date added to LUP
2022-05-20 10:24:09
date last changed
2024-06-13 12:31:44
@article{88858723-49e0-4213-bf8a-94f14d578060,
  abstract     = {{<p>The urban form and extreme microclimate events can have an important impact on the energy performance of buildings, urban comfort and human health. State-of-the-art building energy simulations require information on the urban microclimate, but typically rely on ad-hoc numerical simulations, expensive in-situ measurements, or data from nearby weather stations. As such, they do not account for the full range of possible urban microclimate variability and findings cannot be generalized across urban morphologies. To bridge this knowledge gap, this study proposes two data-driven models to downscale climate variables from the meso to the micro scale in arbitrary urban morphologies, with a focus on extreme climate conditions. The models are based on a feedforward and a deep neural network (NN) architecture, and are trained using results from computational fluid dynamics (CFD) simulations of flow over a series of idealized but representative urban environments, spanning a realistic range of urban morphologies. Both models feature a relatively good agreement with corresponding CFD training data, with a coefficient of determination R<sup>2</sup> = 0.91 (R<sup>2</sup> = 0.89) and R<sup>2</sup> = 0.94 (R<sup>2</sup> = 0.92) for spatially-distributed wind magnitude and air temperature for the deep NN (feedforward NN). The models generalize well for unseen urban morphologies and mesoscale input data that are within the training bounds in the parameter space, with a R<sup>2</sup> = 0.74 (R<sup>2</sup> = 0.69) and R<sup>2</sup> = 0.81 (R<sup>2</sup> = 0.74) for wind magnitude and air temperature for the deep NN (feedforward NN). The accuracy and efficiency of the proposed CFD-NN models makes them well suited for the design of climate-resilient buildings at the early design stage.</p>}},
  author       = {{Javanroodi, Kavan and Nik, Vahid M. and Giometto, Marco G. and Scartezzini, Jean Louis}},
  issn         = {{0048-9697}},
  keywords     = {{Air temperature; CFD simulations; Extreme microclimate conditions; Neural networks; Urban morphology; Wind speed}},
  language     = {{eng}},
  month        = {{07}},
  publisher    = {{Elsevier}},
  series       = {{Science of the Total Environment}},
  title        = {{Combining computational fluid dynamics and neural networks to characterize microclimate extremes : Learning the complex interactions between meso-climate and urban morphology}},
  url          = {{http://dx.doi.org/10.1016/j.scitotenv.2022.154223}},
  doi          = {{10.1016/j.scitotenv.2022.154223}},
  volume       = {{829}},
  year         = {{2022}},
}