Thermal Comfort Control in Air-Conditioned Buildings: new data-driven approaches to Neutral Temperature estimation

An essential requirement for thermal comfort is maintaining an indoor air temperature considered satisfactory by the majority of occupants. According to the so-called adaptive theory, such a neutral temperature may change with the outdoor air temperature, a feature that can be exploited to achieve energy savings without reducing thermal comfort. In the literature, the characterization of this dependence relies on a two-step procedure. First, occupants' thermal sensation votes are processed at building level and outdoor temperatures are averaged to obtain a neutral and outdoor temperature pair for each building. The pairs are then used to fit the neutral temperature model. Herein, three approaches for estimating neutral temperature models without need for a preprocessing at building level are proposed and validated on the summer data of the ASHRAE RP-884 database: (i) regression of temperatures considered neutral by the occupants against the outdoor temperature; (ii) direct regression of the ASHRAE votes against indoor and outdoor temperature; (ii) estimation via logistic regression of the indoor temperature that maximizes the percentage of satisfied users. Overall, seven neutral temperature models are successfully worked out. A first finding is that the the adaptive hypothesis can be ascertained also by models formulated at raw data level. When compared to each other, the seven models are in good agreement, especially those within the same approach, thus demonstrating the viability of neutral temperature modeling based on individual raw data.