Large-eddy Simulation of flow and dispersion around an isolated building: Analysis of influencing factors

Zhengtao Ai, C. M. Mak

Research output: Contribution to journalJournal articleResearchpeer-review


CFD simulations are known to be very sensitive to many physical and numerical parameters. Many previous studies have analyzed the factors that influence RANS modeling, but few have considered how these factors affect LES modeling. This study investigates the factors that influence the LES modeling of flow and dispersion around an isolated building. The LES model is validated against two wind tunnel experimental datasets. The factors analyzed are upstream distance of computational domain, mesh resolution, length of sampling period, inflow fluctuating algorithm, subgrid-scale (SGS) Schmidt number, SGS model, and Smagorinsky constant. Based on the experimental data, the influences of these factors on the numerical results are discussed and appropriate selections are recommended. Specifically, this study finds that the fluctuating intensity of the velocity components provided by a Spectral synthesizer (an inflow fluctuating algorithm) is less than half of that provided by the Vortex method (an inflow fluctuating algorithm). In addition, the velocity field is less sensitive to some of the influencing factors than the concentration field. Therefore, different optimum values for specific influencing factors are suggested to predict velocity and concentration fields. The findings of this study are expected to improve the quality of LES modeling of flow and dispersion in built environments. (C) 2015 Elsevier Ltd. All rights reserved.
Original languageEnglish
JournalComputers & Fluids
Pages (from-to)89-100
Number of pages12
Publication statusPublished - 2015
Externally publishedYes


  • CFD
  • Dispersion
  • Flow
  • LES
  • Sensitivity study


Dive into the research topics of 'Large-eddy Simulation of flow and dispersion around an isolated building: Analysis of influencing factors'. Together they form a unique fingerprint.

Cite this