Validating computational predictions of natural ventilation in Stanford’s Y2E2 building

Natural ventilation can significantly reduce building energy consumption, but robust design is challenging due to uncertainties in a building’s operating conditions. In a previous study, we used an integral model and a computational fluid dynamics (CFD) model with uncertainty quantification to model night-time ventilation in Stanford’s Y2E2 building. The prediction of indoor air temperature showed a too high cooling rate compared to building measurements; variance based sensitivity analysis showed that the initial thermal mass temperature and internal load had an important influence on results.

The objective of the present study is to perform additional measurements to further investigate the discrepancies between the model predictions and measurements. Thermal mass temperatures are measured, and Bayesian inference is used to estimate the internal load. Extensive measurements of spatial variability in the indoor air temperature are conducted to support more accurate validation of the predictions. Future work will leverage the findings of this study to further improve the models and support robust design of natural ventilation systems.