Natural ventilation in buildings can create a comfortable and healthy indoor environment,and can save energy used in the mechanical ventilation systems. Two subgrid-scale models of
large eddy simulation (LES), a Smagorinsky subgrid-scale (SS) model and a Filtered dynamic subgrid-scale (FDS) model, have been used to study airflow in buildings with natural ventilation. It was found that, for fully developed turbulence flow with a high Reynolds number, both the SS and FDS models provide good results. However, if the flow has both turbulent and laminar characteristics, or the wall effect is significant, the SS model performs
poory due to its constant model coefficient. The FDS model can still predict this flow correctly
because its model coefficient varies over space and time according to flow types. Furthermore,
for a single-sided ventilation, it is important to obtain instantaneous flow information in order
to correctly predict ventilation rate and air change effectiveness. Reynolds Averaged Navier-
Stokes (RANS) modeling cannot correctly calculate the ventilation rate in this case. r 2001
Elsevier Science Ltd. All rights reserved.