Optimal temperature sensor placement in buildings with buoyancy-driven natural ventilation using computational fluid dynamics and uncertainty quantification

Abstract Natural ventilation and cooling can significantly reduce building energy consumption, but inherent variability in the boundary and operating conditions makes robust design and operation a challenging task. Computational models can be leveraged to evaluate the effects of these uncertainties and support robust design; however, the model accuracy should be thoroughly assessed, ideally through validation with carefully designed full-scale measurements. This study presents a novel approach for using computational fluid dynamics (CFD) simulations with uncertainty quantification to optimize temperature sensor placement in buildings with buoyancy-driven natural ventilation, such that the measurements are representative of the volume-averaged temperature, while also supporting analysis of the spatial variability in the temperature field. The approach uses a polynomial chaos expansion method to quantify the effect of the variability in the boundary and initial conditions on the predicted buoyancy-driven flow and indoor temperature field. Results for an operational building that employs night-time cooling are analyzed to identify sensor locations that will support robust observation of the time series of the volume-averaged temperature and the temperature range. The approach is shown to significantly reduce the potential error between the temperature recorded at a random location and the true volume-average temperature, while also providing a representative measure of the temperature range in the building. The results indicate that this approach can also support identifying optimal locations for temperature sensors used for operational control.