Elsevier

Atmospheric Environment

Volume 45, Issue 2, January 2011, Pages 428-438
Atmospheric Environment

CFD simulation of near-field pollutant dispersion on a high-resolution grid: A case study by LES and RANS for a building group in downtown Montreal

https://doi.org/10.1016/j.atmosenv.2010.09.065Get rights and content

Abstract

Turbulence modeling and validation by experiments are key issues in the simulation of micro-scale atmospheric dispersion. This study evaluates the performance of two different modeling approaches (RANS standard k-ε and LES) applied to pollutant dispersion in an actual urban environment: downtown Montreal. The focus of the study is on near-field dispersion, i.e. both on the prediction of pollutant concentrations in the surrounding streets (for pedestrian outdoor air quality) and on building surfaces (for ventilation system inlets and indoor air quality). The high-resolution CFD simulations are performed for neutral atmospheric conditions and are validated by detailed wind-tunnel experiments. A suitable resolution of the computational grid is determined by grid-sensitivity analysis. It is shown that the performance of the standard k-ε model strongly depends on the turbulent Schmidt number, whose optimum value is case-dependent and a priori unknown. In contrast, LES with the dynamic subgrid-scale model shows a better performance without requiring any parameter input to solve the dispersion equation.

Graphical abstract

Research highlights

► Near-field pollutant dispersion in urban areas is particularly challenging to predict. ► The accuracy of RANS models depends strongly on the value of the turbulent Schmidt number. ► The optimum value of Sct depends on the configuration under study. ► With LES, the dispersion equation can be solved without any parameter input. ► Prediction accuracy depends strongly on urban geometry and reference wind direction.

Introduction

Outdoor air pollution is associated with a broad spectrum of acute and chronic health effects (Brunekreef and Holgate, 2002). The pollutants that are brought into the atmosphere by various sources are dispersed (advected and diffused) over a wide range of horizontal length scales. Micro-scale dispersion refers to processes acting within horizontal length scales below about 5 km. It can be studied in detail by wind-tunnel modeling and by numerical simulation with Computational Fluid Dynamics (CFD). Wind-tunnel modeling is widely recognized as a valuable tool in wind flow and gas dispersion analysis but it generally only provides data at a limited number of discrete positions and it can suffer from incompatible similarity requirements. CFD does not have these two disadvantages; it provides “whole flow-field” data and it can be performed at full-scale. Furthermore, it is very suitable for parametric studies for various physical flow and dispersion processes. On the other hand, the accuracy of CFD is a main concern, and grid-sensitivity analysis and experimental validation studies are imperative.

In the past decades, CFD has been used extensively in micro-scale pollutant dispersion studies. A distinction can be made between generic studies and applied studies. Generic studies include configurations such as idealized isolated buildings (e.g. Leitl et al., 1997, Li and Stathopoulos, 1997, Meroney et al., 1999, Blocken et al., 2008, Tominaga and Stathopoulos, 2009, Santos et al., 2009), idealized isolated street canyons (e.g. Leitl and Meroney, 1997, Chan et al., 2002, Gromke et al., 2008) or regular building groups (e.g. Kim and Baik, 2004, Shi et al., 2008, Wang et al., 2009, Buccolieri et al., 2010, Dejoan et al., 2010). Applied studies refer to actual (isolated) buildings or actual building groups (urban areas) (e.g. Hanna et al., 2006, Patnaik et al., 2007, Baik et al., 2009, Pontiggia et al., 2010).

Many previous studies have indicated that CFD simulations based on the steady Reynolds-Averaged Navier–Stokes (RANS) equations are deficient in reproducing the wind-flow patterns (e.g. Murakami et al., 1992) and near-field pollutant dispersion concentrations around buildings (e.g. Leitl et al., 1997, Meroney et al., 1999, Blocken et al., 2008, Tominaga and Stathopoulos, 2010), which motivates the use of Large Eddy Simulation (LES) for micro-scale pollutant dispersion. A number of authors have applied LES to dispersion around isolated buildings (e.g. Tominaga et al., 1997, Sada and Sato, 2002) and in street canyons (e.g. Li et al., 2008, Hu et al., 2009). One of the main concerns in micro-scale atmospheric dispersion modeling, however, is determining the spread of pollutants from sources in actual urban environments. During the past decade, the continuous progress in computational power has allowed us to also apply LES to this kind of street-scale dispersion problems. An overview of previous LES studies in actual urban areas is provided in Table 1. For every study, the city name and location, the spatial extent of the urban study (near-field or far-field) and the subgrid-closure scheme are listed. It is also indicated whether RANS simulations were performed and whether validation by comparison with experiments was conducted. Finally, also the cell type and the grid resolution are reported. The present study aims at expanding the current state of the art in LES dispersion modeling, as discussed below.

The previous studies all involved a large group of buildings (13 or more) with the primary intention to determine the far-field spread of contaminants released from a source through the network of city streets and over buildings. This type of studies is called “far-field” dispersion studies in the framework of this paper. Given the extent of the computational domains involved, the grid resolutions in these far-field studies are generally relatively low, with a minimum cell size of the order of 1 m. An exception to this is the study by Camelli et al. (2005), who used cell sizes down to 0.22 m. Although the results provided by LES are generally promising, comparison with experimental data was only performed in two studies. For dispersion in actual urban areas, the relative performance of LES compared to RANS is not well known, as this was not addressed in previous studies.

Up to now, to the knowledge of the authors, no high-resolution CFD studies of near-field gas dispersion for relatively large building groups which are accompanied by grid-sensitivity analysis and validation by comparison with experiments have been performed. The aim of this paper is to present this kind of study for pollutant dispersion around a building group in downtown Montreal. The focus is both on the prediction of pollutant concentrations in the surrounding streets (for pedestrian outdoor air quality) and on the prediction of concentrations on building surfaces (for ventilation system inlets placement and indoor air quality), i.e. two zones close to the source where the computation of the concentration distribution is known to be particularly challenging. The CFD simulations are validated by detailed wind-tunnel experiments performed earlier by Stathopoulos et al. (2004), in which sulfur-hexafluoride (SF6) tracer gas was released from a stack on the roof of a three-storey building and concentrations were measured at several locations on this roof and on the facade of a neighboring high-rise building. Note that earlier CFD studies for the same case included none or only one of the neighboring buildings (Blocken et al., 2008, Lateb et al., 2010), while in the present study, surrounding buildings are included up to a distance of 300 m. For this purpose, a high-resolution grid with minimum cell sizes down to a few centimeters (full-scale) is used. The grids are obtained based on detailed grid-sensitivity analysis. Both LES and RANS simulations are performed.

Section snippets

Description of the experiments

Experiments of pollutant dispersion in downtown Montreal were conducted in 2004 by Concordia University and IRSST1 (Stathopoulos et al., 2004). Two types of experiments were conducted: on-site and in the Concordia University boundary layer wind tunnel (Stathopoulos, 1984), with a scale factor of 1/200. SF6 was used as tracer gas and released from a stack located on the roof of the BE building, which is a three-storey

Governing equations

RANS turbulence models can provide accurate solutions for a wide range of industrial flow problems while requiring relatively low computational resources. The basic principle of this turbulence modeling approach is the application of the Reynolds-averaging operator to the Navier–Stokes equations, resulting in the appearance of new unknowns: the Reynolds stresses. These stresses can be linked to the flow variables in different ways, which defines the type of turbulence model.

With LES, a spatial

Domain

Two computational domains have been created, one for each wind direction (see Figs. 2b and 3b). The inlet and outlet planes are perpendicular to the flow direction, as required by the vortex method (Mathey et al., 2006) used to generate a time-dependent velocity profile at the inlet (more details in Section 4.4). The streamwise, spanwise and vertical coordinates are denoted by x, y and z, respectively. The BE building is modeled in detail, including the roof-top structures. The other buildings

Grid-sensitivity analysis for case SW

The results of the simulations performed on the grids SW-c, SW-m and SW-f with SKE and Sct = 0.5 are shown in Fig. 5a, where the concentration values obtained with the coarse and fine grids (vertical axis) are compared to those on the medium grid (horizontal axis). A slight change in the results can be observed from SW-c to SW-m, whereas the results obtained with SW-m and SW-f are similar.

In the case of LES with implicit filtering, the local filter width is equal to the computational cell size.

Summary and conclusions

High-resolution CFD simulation of near-field pollutant dispersion in a building group in downtown Montreal was performed with two different turbulence modeling approaches: RANS standard k-ε and LES with the dynamic Smagorinsky SGS model. Contrary to most of the previous CFD studies of urban dispersion which focused on the far-field spread of contaminants, the present simulations focused on the concentration values close to the source (on the building surfaces and in the surrounding streets) and

Acknowledgements

The numerical simulations reported in this paper were supported by the laboratory of the Unit Building Physics and Systems (BPS) of Eindhoven University of Technology.

References (48)

  • M. Lateb et al.

    Numerical simulation of pollutant dispersion around a building complex

    Building and Environment

    (2010)
  • B.E. Launder et al.

    The numerical computation of turbulent flows

    Computer Methods in Applied Mechanics and Engineering

    (1974)
  • B.M. Leitl et al.

    Car exhaust dispersion in a street canyon. Numerical critique of a wind tunnel experiment

    Journal of Wind Engineering and Industrial Aerodynamics

    (1997)
  • B.M. Leitl et al.

    Concentration and flow distributions in the vicinity of U-shaped buildings: wind-tunnel and computational data

    Journal of Wind Engineering and Industrial Aerodynamics

    (1997)
  • Y. Li et al.

    Numerical evaluation of wind-induced dispersion of pollutants around a building

    Journal of Wind Engineering and Industrial Aerodynamics

    (1997)
  • S. Murakami et al.

    Numerical study on velocity-pressure field and wind forces for bluff bodies by k-ε, ASM and LES

    Journal of Wind Engineering and Industrial Aerodynamics

    (1992)
  • M. Pontiggia et al.

    Hazardous gas releases in urban areas: assessment of consequences through CFD modelling

    Journal of Hazardous Materials

    (2010)
  • K. Sada et al.

    Numerical calculation of flow and stack-gas concentration fluctuation around a cubical building

    Atmospheric Environment

    (2002)
  • J.M. Santos et al.

    Numerical simulation of flow and dispersion around an isolated cubical building: the effect of the atmospheric stratification

    Atmospheric Environment

    (2009)
  • R.F. Shi et al.

    Large eddy simulation of wind field and plume dispersion in building array

    Atmospheric Environment

    (2008)
  • T. Stathopoulos

    Design and fabrication of a wind tunnel for building aerodynamics

    Journal of Wind Engineering and Industrial Aerodynamics

    (1984)
  • T. Tamura

    Towards practical use of LES in wind engineering

    Journal of Wind Engineering and Industrial Aerodynamics

    (2008)
  • Y. Tominaga et al.

    Turbulent Schmidt numbers for CFD analysis with various types of flowfield

    Atmospheric Environment

    (2007)
  • Y. Tominaga et al.

    Numerical simulation of dispersion around an isolated cubic building: comparison of various types of k-ɛ models

    Atmospheric Environment

    (2009)
  • Cited by (317)

    • Large-eddy simulation-based wind and thermal comfort assessment in urban environments

      2024, Journal of Wind Engineering and Industrial Aerodynamics
    View all citing articles on Scopus
    View full text