Currently, urbanization is proliferating globally, which engenders climate and environmental predicaments like urban heat islands and air pollution, menacing human well-being and health. In this scenario, augmenting urban ventilation and enabling cities to respire better is regarded as an efficacious strategy to alleviate these issues. Creating a propitious urban wind environment can not only enhance air circulation within the city, augment urban permeability, and improve human living comfort but also reduce the accumulation of pollutants and heat, and promote high-quality urban development.

With the pressing demand for optimizing urban wind environments, this inquiry endeavors to gain a profound comprehension of the attributes of urban wind environments through the amalgamation of dive...[ Read more ]

Currently, urbanization is proliferating globally, which engenders climate and environmental predicaments like urban heat islands and air pollution, menacing human well-being and health. In this scenario, augmenting urban ventilation and enabling cities to respire better is regarded as an efficacious strategy to alleviate these issues. Creating a propitious urban wind environment can not only enhance air circulation within the city, augment urban permeability, and improve human living comfort but also reduce the accumulation of pollutants and heat, and promote high-quality urban development.

With the pressing demand for optimizing urban wind environments, this inquiry endeavors to gain a profound comprehension of the attributes of urban wind environments through the amalgamation of diverse models, including large-scale complex terrain cities, block-scale urban spaces, and building-scale structures. Through the utilization of computational fluid dynamics (CFD) techniques, a multi-scale research framework for urban wind environment characteristics has been established, methodically scrutinizing the features of urban wind environments at varying scales. Furthermore, this study delves into the dynamic turbulent characteristics of building flow via the utilization of physical-informed data-driven analysis methodologies such as Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD), exploring the dynamic vortex structure as well as the dispersion mechanism of pollutants in intricate urban wind environments. The details are as follows:

Firstly, a numerical physio-chemical model of the NO_x-O₃ photochemical cycle is established via CFD, to investigate the near-wake region of an isolated residential/office building. The investigation delves into the dispersion of reactive air pollutants through the lens of fluid phenomenology and its impact on chemical reactivity, formation, transport, deposition, and removal. The simulations were conducted for different scenarios (emission source, wind direction, incoming flow field, concentration field e.g.). A non-dimensional number — the Damköhler number (Da) — is introduced to quantify pollutants' physical dynamics and chemical processes. The results showed that Da displays a strong inverse proportionality with the magnitude and spread of NO – increasing Da reduces human exposure to the toxic NO and NO₂ substantially.

The research is also extended to the role of turbulence models and different building aspect ratios, offering worthy references for future numerical investigations and the design of building complexes. Large eddy simulation with near-wall resolution (LES-NWR) and three viscosity models (i.e., RLZ, RNG, and STK) of the Reynolds-averaged Navier-Stokes (RANS) was assessed regarding the predictive capability of the mean pollutant concentration and velocity fields. Results show that the RLZ provides the most comparable results to the LES-NWR, whereas the RNG and STK underestimate the reverse flow in the building wake and produce subpar predictions at the roof. The RNG also yields the weakest prediction by overestimating the lateral separation bubbles. Furthermore, changing the aspect ratio has negligible effects on the stream and spanwise dispersion of the highly concentrated pollutant. The building width played a more decisive role in constraining the streamwise dispersion than the building height.

Thirdly, the objective is centered on achieving an ideal urban environment. A hybrid numerical intelligence model, encompassing wind tunnel data, simulation outcomes, and machine learning algorithms, is established, accompanied by the same physio-chemo coupling. To this end, 60 simulations were conducted under varied inflow speed and pollutant concentration conditions, altering the Damköhler number for NO (Da_NO) from 0.0031 to 0.0252 and the Damköhler number for O3 (Da_O3) from 0.252 to 1.259. An artificial neural network (ANN) was trained with the back-propagation (BP) algorithm to predict the wind velocity field and pollutant diffusion pattern. The ANN model exhibited a strong capability to characterize the intricate nonlinear spatial diffusion and the reaction of the air species, producing acceptably accurate predictions with significantly reduced computational and time expenditures compared to CFD simulation.

Lastly, we developed a data analysis procedure, namely a Proper Orthogonal Decomposition (POD)-Dynamic Mode Decomposition (DMD) augmented analysis, to isolate the energy- and evolution-wise dominant features of flow field in the urban area. This combination aims to extract modes imposing critical influence on pollutant dispersion from both energetic and dynamic perspectives. The two techniques were first conducted based on large-eddy simulation (LES) results. Subsequently, based on the POD and DMD ranking, the extracted modes were classified into three types: (1) type 1: energetically & dynamically significant mode; (2) type 2: energetically significant & dynamically insignificant mode; (3) type 3: energetically insignificant & dynamically significant mode. This technique can provide a systematic analysis of the flow field within the city scale; it can also provide help for potential applications, such as solving pollutant dispersion issues in urban areas.