Validation of Turbulence Models used in CFD Analysis of Indoor Natural Convection with a Cold Window and a Thermal Manikin

Drafts induced by cold windows deteriorate thermal comfort for occupants during winter. Computational fluid dynamics (CFD) is a practical design tool for evaluating countermeasures against drafts. To balance accuracy and computational cost, the Reynolds-averaged Navier-Stokes (RANS) approach is commonly used. In thermally driven flows, the turbulence model selection and buoyancy treatment significantly affect the accuracy of the simulation results. Previous studies suggest that low Reynolds-number k-ɛ models, which account for turbulence damping under stable stratification, are suitable for natural convection, whereas the SST(Shear Stress Transport) k-ω model often performs well for thermal manikins. In this study, an experiment replicating a winter room with a cold window and a thermal manikin was conducted to obtain validation data for establishing appropriate turbulence model settings in the CFD simulations. CFD simulations reproducing the thermal environment of the experiment were performed using two turbulence models: a low Reynolds-number k-ɛ model and an SST k-ω model. Horizontal profiles of airflow velocity and temperature at 0.5 m above the floor were analyzed. The SST model reproduced the boundary layer thickness along the cooled window more accurately than the Lien low Reynolds-number k-ɛ model. However, it tended to overestimate the airflow velocity near the wall.