Occupant-centric models for thermal comfort in buildings: Theoretical and experimental analysis of methods for enhancing user comfort in dynamic thermal indoor environments

Abstract

In the last decades, increasing global energy demand, a foreseen reduction of available fossil fuels and increasing evidence for global warming have shown the urgency to rethink the built environment and promote energy transition. Indeed, in most industrialised nations, the building sector accounts for about 40% of the total energy consumption (space heating and cooling, domestic hot water, ventilation, lighting and appliance use). A significant share of this energy is used to thermal control buildings and provide thermally comfortable indoor environments. However, technical building systems are typically designed and operated considering fixed set-point temperatures based on the ‘one-size-fits-all’ principle – which has been questioned in the last fifty years – assuming universal thermal comfort requirements. Furthermore, the indoor environment frequently changes abruptly across buildings or between various parts within a single building. For instance, manually operating thermostats, windows and solar shades can result in considerable and not systematic changes in the indoor environment. Also, automatic controllers exhibit, to a lesser degree, a similar behaviour. Moreover, individual activity modifies the basal metabolic rate over time, and the addition or removal of clothes affects the heat balance of the human body as well. In other words, the steady-state temperature settings are the exception rather than the rule. Building temperature ranges should instead be based on real-time empirical evidence regarding the needs of its occupants, which is obtained through their feedback (usually on a rating scale). This thesis investigates these topics and relies upon an experimental study to explore the human reaction to dynamic thermal environments. The general approach utilised in this thesis encompasses a technical/methodological aspect, namely a newer controlled experimental procedure and a robust and replicable methodology for human feedback acquisition, and a statistical aspect, namely an original statistics-enabled occupant-centric modelling. The technical/methodological aspect refers to how thermal comfort data are collected; that is, the specific approach and experimental set-up utilised. The statistical aspect refers to how thermal comfort data are analysed, namely, the specific statistical technique to be adopted and the needed modelling steps. It was found that the human reaction to dynamic thermal stimuli is asymmetric with respect to heating and cooling processes, and two distinct mechanisms cause discomfort due to overheating and overcooling. Compared to the recommendations regarding temperature cycles, drifts and ramps included in the ASHRAE Standard 55, this result showed that current recommendations underestimate the risk of thermal discomfort during a cooling process while overestimating it during a heating one. Concerning the subjective thermal comfort data analysis, the choice of the statistical method affects the conclusions. While it may seem a trivial consideration, till now, it is common in the thermal comfort field to find studies that use, for example, linear regression on ordinal data following an old approximation used to overcome the lack of statistical tools and computational power (which are not anymore limiting aspects in statistical analysis). Particularly, we showed that applying a linear regression model to ordinal data suggested that there is no difference in means and effect size between genders (female/male). In contrast, an ordinal regression model leaded to the opposite conclusion. This is considered one of the reasons why there is no consensus in the scientific literature on whether gender is an influential factor when assessing the perception of the thermal environment. This result points out that greater attention should be paid to the choice of the statistical method used to analyse subjective data, which should consider the level of measurement used during the data gathering. Furthermore, two different procedures were proposed to facilitate the integration of the occupants and their actual needs into the design and operation of buildings: the former is suitable for a better-informed design phase, where the target is the optimal thermal comfort conditions expressed for an ‘average’ occupant; the other is appropriate for including the human-in-the-control-loop of a building, where satisfying the needs of a specific occupant is the primary goal. Through this thesis work, new knowledge concerning the human reactions to a dynamic thermal environment was created, which can improve the understanding of the extent to which the indoor environmental conditions can vary both naturally and artificially. Designing and implementing thermally comfortable set-point modulations that consider the occupant feedback capabilities would be beneficial to increase perceived thermal comfort and productivity, potentially reduce energy consumption and significantly support the clean energy transition. In addition, several recommendations for future research are presented.