When predicting the performance of night ventilation with Building Energy Simulation models, the results are highly sensitive to the selection of the convective heat transfer coecient (CHTC). Therefore, a numerical study was conducted, simulating rooms with Fluent 12 for 8 hours of flow time. The study was limited to 2D and results must be considered as qualitative. 14 initial conditions were applied to 4 geometries, including the thermal mass
in floor and ceiling. The ACH was varied from 4 to 11 h-1 and initial K between room air and supply air from 1.25 to 10 K. The flow development showed three flow phases, influencing the CHTC at the ceiling. The implications
for the selection of CHTC in BES are discussed and applied in a sensitivity study in TRNSYS 17. The sensitivity on the thermal comfort was investigated through a parameter variation, using convection correlations from literature. When comparing with the internal CHTC calculation in TRNSYS, the usage of natural convection correlations increased the predicted weighted overheating hours at 4 and 12 ACH respectively with 18 Kh (19 %) and 105 Kh (12 %), whereas forced convection correlations reduced this at 4 and 12 ACH respectively with 30 Kh (11 %) and 210 Kh (29 %).