Improvements in Numerical Modeling of Airflows Around Multiple Buildings

Open Access
Davidovic, Danko
Graduate Program:
Architectural Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
November 23, 2009
Committee Members:
  • Jelena Srebric, Committee Chair
  • Dennis K Mc Laughlin, Committee Member
  • Stanley Allan Mumma, Committee Member
  • Bohumil Kasal, Committee Member
  • wind tunnel experiment
  • turbulence modeling
  • CFD
The airflow around buildings and turbulent dispersion of the airborne contaminants has been studied extensively in the last few decades. However, most of the studies restricted their attention to isolated buildings or obstacles placed in relatively simple surroundings with simplified boundary and initial conditions. In addition, more realistic scenarios, such as multiple buildings in urban areas, require more advanced numerical tools. Current engineering practice in computational fluid dynamics involves two equation ¡§k-ƒÕ¡¨ turbulence model, which is typical representative of Reynolds averaged Navier-Stokes (RANS) turbulence modeling approach. Unfortunately, this turbulence modeling approach is not capable of resolving all important features of the engineering flows with complex geometries and boundary flow conditions. Improvements in numerical modeling with k-ƒÕ turbulence model are typically achieved by mesh refinement in the regions with sharp changes in the boundary conditions. However, the increase in the computational cost does not justify very fine mesh refinement in many practical applications. On the other hand, more advanced numerical techniques, such as large eddy simulation (LES) and direct numerical simulations (DNS) are still not entirely applicable tool in engineering design because of high sensitivity to the variations in initial and boundary conditions, and requirement for finer meshes at higher computational cost when compared with Reynolds averaged Navier-Stokes (RANS) turbulence modeling approach. The main objective of the thesis was to provide faster, yet reliable and simple modeling tool for simulation of outdoor airflow around multiple buildings. The intent was to improve the existing zero-equation turbulence models mainly developed and used for modeling of airflows in indoor environments. The zero-equation turbulence model adopts the local mean velocity and distance to the nearest wall surface as the characteristic velocity and length scale to estimate the turbulent viscosity. However, the coefficients in the model equation require assessment for each particular application. For that reason, the wind tunnel modeling of complex airflow around four scaled student dorm buildings at the Penn State University campus has been carried out in the closed loop, low speed wind tunnel at the Department of Aerospace Engineering at the Pennsylvania State University using triple hot wire anemometry to collect the velocity time series. The obtained results on 1:250 scale modeled buildings provided the base for calibration of the coefficients in the zero-equation turbulent model. The derived expression for calibration coefficient implements both the bulk and turbulent Reynolds number as relevant characteristics of approaching wind flow. The calibrated zero-equation turbulence model has been incorporated in commercial CFD software PHOENICS and tested against the wind tunnel results. The simulation results show satisfactory agreement with measured data for longitudinal velocity components. The prediction of vertical velocity component was less accurate and inferior performance of the model was observed for lateral velocity component prediction. Limitations in measurement equipment and negligence of all mean strain rate tensor components including lateral and vertical velocity gradients are two most likely sources of discrepancy between the numerical and wind tunnel experimental results. Further estimate of exponents in the derived expression for calibration coefficient is recommended for different buildings layouts. Experimental study with regular array of cubes placed in the boundary layer wind tunnel provided the data for additional validation of the developed model. The experimental results comprised of streamwise and vertical velocity profiles along the centerline of the modeled building cube array. The developed model for calibration coefficients in the expression for turbulent eddy viscosity in the zero-equation turbulence model demonstrated good agreement in streamwise velocity prediction with experimentally acquired results. Implemented turbulence model showed comparative competitiveness with more advanced one-point turbulence closure models, such as two equation ¡§k-epsilon¡¨ and modified ¡§Kato-Launder¡¨ version of the ¡§k-epsilon¡¨ model, although, the full adjustment of the calibration coefficients in the model was not feasible due to lack of input information that was not measured in the wind tunnel experiment. Part of the thesis effort focused on possible improvements in the description of inlet boundary conditions of incoming wind. For that purpose, an extensive literature review on the atmospheric surface layer flows, their characterization and parameterization has been performed. In addition, the systematic overview of various strategies for synthetic turbulent wind speeds generation has been discussed in detail. As an outcome, the most suitable techniques for implementation in CFD models are described with sufficient level of detail for easy coding. The described techniques should provide substantial foundation for the future implementation of unsteady Reynolds averaged Navier-Stokes equations (URANS) turbulence modeling approach for simulation of outdoor airflow in complex urban settings. The developed methodology for calibration coefficients in the zero-equation turbulence models represents the major contribution of this study. Further improvements of the model are possible with provision of experimental data with high spatial resolution using advanced non-invasive measurement techniques such as Particle Image Velocimetry (PIV) and Laser Doppler Anemometry (LDA). More comprehensive set of experimental data would enable further refinement of the proposed model in terms of adjustment of the exponents in the proposed equation for calibration coefficients. Instead of using single constant value, the exponents should be defined as a function of plan area density, frontal area density and other morphological parameters already established to describe various buildings layouts and boundary layer flows in urban environments. To end this journey through the magical world of turbulence, I will quote several lines from the famous poem that many people have quoted before, but in my case, this time, with much deeper understanding and appreciation: ¡§What we call the beginning is often the end And to make and end is to make a beginning. The end is where we start from¡K ¡KWe shall not cease from exploration And the end of all our exploring Will be to arrive where we started And know the place for the first time.¡¨ Tomas S. Elliot in Little Gidding (1942)