Yoshihide Tominaga

Velocity-pressure field of cross ventilation with open windows analyzed by wind tunnel and numerical simulation

Abstract In order to investigate the mechanism of cross ventilation with open windows, velocity and pressure fields of airflows in and around building models are analyzed in detail by means of wind tunnel tests and numerical simulations. Large eddy simulation (LES) is used for 3D turbulent flow analysis. The results of LES agree very well with those of the wind tunnel tests, and thus the accuracy of the numerical method used here is well validated. By means of LES, the spatial distributions of mean static pressure, turbulence energy, turbulence energy dissipation rate, etc. are examined with sufficient accuracy. The energy dissipating process (total pressure loss) along a cross flow through a building model is examined in relation to the conventional method for predicting the airflow rate of wind-induced ventilation, which uses static pressure drops and the discharge coefficients α of openings. However, in cross ventilation with large openings, the dynamic pressure which has a significantly large value in a room, cannot be neglected. Therefore we cannot predict the airflow rates with the conventional method based on static pressure drops. The airflow through large openings still preserves much of its mean kinetic energy when it remains inside the room and this is reflected in decreased values of the total pressure loss coefficients.

NUMERICAL SIMULATION OF TURBULENT DIFFUSION IN CITIES

Examples of numerical simulations on turbulent diffusion fields in cities are presented including comparison with wind tunnel experiments. The first example is the computation of the flow and diffusion fields around a cooling tower by k-e turbulence model. A composite grid technique, which connects two grids supported by a fortified solution algorithm, is adopted in this computation. The second example is the computations of flow over a heated urban area by Large Eddy Simulation (LES) and k—e model. In the k-e model computations, two different modellings for buoyancy effect are adopted; first is the model proposed by Rodi (1984), and second is the one proposed by Viollet (1987). The last example is the LES computations of turbulent diffusion of buoyant (lighter-than-air) and heavy gases discharged in the wake region behind a building model. Results given from two types of Smagorinsky models, i.e., the standard one and its modified version proposed by Mason et al. (1989, 1990), are compared here. In the concluding remarks, the relative abilities of various turbulence models are discussed.