CFD simulation of cross-ventilation flow for different isolated building configurations: Validation with wind tunnel measurements and analysis of physical and numerical diffusion effects

Abstract

a b s t r a c tComputational Fluid Dynamics (CFD) has become one of the most important tools for the assessment of natural cross-ventilation of buildings. To ensure the accuracy and reliability of CFD simulations, solution verification and validation studies are needed, as well as detailed sensitivity studies to analyse the impact of computational parameters on the results. In a previous study by the present authors, the impact of a wide range of computational parameters on the cross-ventilation flow in a generic isolated single-zone building was investigated. This paper presents the follow-up study that focuses in more detail on validation with wind tunnel measurements and on the effects of physical and numerical diffusion on the cross-ventilation flow. The CFD simulations are performed with the 3D steady Reynolds-Averaged Navier-Stokes (RANS) approach with the SST k-o model to provide closure. Validation of the coupled outdoor wind flow and indoor airflow simulations is performed based on Particle Image Velocimetry (PIV) measurements for four different building configurations. The analysis of numerical diffusion effects is performed in two parts. First, the effect of physical diffusion is analysed by changing the inlet profiles of turbulent kinetic energy within a realistic range. Second, the effect of numerical diffusion is investigated by changing the grid resolution and by applying both first-order and second-order discretisation schemes. The results of the validation study show a good to a very good agreement for three of the four configurations, while a somewhat less good agreement is obtained for the fourth configuration. The results of the diffusion study show that the effects of physical and numerical diffusion are very similar. Along the centreline between the openings, these effects are most pronounced inside the building, and less pronounced outside the building. The velocity-vector fields however show that increased physical and numerical diffusion decreases the size of the upstream standing vortex and increase the spread of the jet entering the buildings. It is concluded that diffusion is an important transport mechanism in cross-ventilation of buildings, and that special care is needed to select the right amount of physical diffusion and to reduce the numerical diffusion, using highresolution grids and using at least second-order accurate discretisation schemes.

Low-Reynolds number mixing ventilation flows: Impact of physical and numerical diffusion on flow and dispersion

“…Skovgaard and Nielsen 1990;Jones and Whittle 1992;Ramponi and Blocken 2012a). In 1992, Jones and Whittle (1992) stated in their overview paper that sufficient grid resolution should be applied in jet regions, boundary layers and in thermal plumes, and that: "A failure to provide enough mesh in these areas will result in the supply jet or the boundary layer flow being insufficiently resolved (where the momentum is rapidly and artificially diffused) resulting in an unrealistic local (and possibly global) solution.".…”

“…Sørensen and Nielsen 2003;Ramponi and Blocken 2012a) reported that the effect of numerical (artificial) diffusion can be compared to the physical diffusion caused by the turbulence in CFD simulations. Both physical and numerical diffusion smear out the results and a high level of numerical diffusion will negatively impact the numerical accuracy.…”

“…Both physical and numerical diffusion smear out the results and a high level of numerical diffusion will negatively impact the numerical accuracy. Ramponi and Blocken (2012a) showed that the proper choice of the inlet TI was equally important as the choice of a proper grid resolution and higher-order discretization scheme. In the past, some numerical studies focused on the effect of inlet TI on the flow pattern.…”

A model-based data-interpretation framework for improving wind predictions around buildings

“…A second-order discretization scheme is used to interpolate pressure from values at cell centers. The convergence criteria, based on the scaled residuals, are set to 4 1 0 − for all variables. Before simulations were terminated, the predicted values of the variables of interest were constant.…”

“…Wind can also be harnessed for natural ventilation of buildings [2]. Computational Fluid Dynamics (CFD) simulations have been widely used to simulate wind around and through buildings [3,4]. Although guidelines have been established to improve simulation predictions [5], large discrepancies remain when simulation predictions are compared to field measurements.…”