A systematic validation of the computational fluid dynamics code ANSYS CFX for atria geometries is presented. Turbulent natural convection, radiation heat transfer and conjugate heat transfer are essential to the performance of an atrium and are all validated separately. In order to be thorough, the initial validations involve the fundamental simulations for each phenomenon. A simulation of a complete atrium is also presented using the conclusions reached in the previous validations.
The validation of each of the phenomena was successful. Turbulent natural convection simulations yielded two suitable turbulence models, with the preference for k-ω model being decided by a narrow margin. The radiation validations proved that the Discrete Transfer model was an accurate model and the best offered by ANSYS CFX. Conjugate heat transfer showed that ANSYS CFX was capable of capturing the qualitative aspects of the phenomenon. The final atrium simulations showed the expected over prediction of temperature in the atrium, and an under prediction of the stratification. The atrium simulation proved insensitive to façade emissivity. The magnitude of the solar radiation heat flux did change the temperature and velocity field. It was shown that ANSYS CFX was capable of modeling the important phenomena but more accurate boundary conditions are required to obtain the best results possible.