Evaluation of LES and RANS CFD modelling of multiple steady states in natural ventilation

This paper reports research carried out with the aim of evaluating and comparing the performance of Large Eddy Simulation (LES) and Unsteady Reynolds-Averaged Navier-Stokes (URANS) modelling for predicting the multiple steady states observed in experiments on a buoyancy-driven naturally ventilated enclosure. The sub-grid scales of the flow have been modelled using a Van Driest damped Smagorinsky sub-grid scale model in the case of LES and an RNG k-ε turbulence model has been used for URANS. A novel mesh design strategy was introduced to design the LES mesh to identify an optimum 'well-resolved' mesh, assuming that the flow investigated is free-shear dominated. It was found that the URANS solution eventually settled down into a permanent steady state, displaying no evidence of continuing instabilities or periodic unsteadiness. Both URANS and LES solutions captured the existence of three steady states as observed in experimental studies. However, LES was more accurate in predicting the temperatures inside the enclosure compared to URANS. In the URANS solutions, it was observed that for smaller lower opening areas the average indoor temperature had noticeable discrepancies when compared with experimental results. Unlike URANS, LES correctly predicted different steady state temperatures for different opening areas and the time to reach steady state agreed closely with theoretical predictions.