Building Energy Management Systems (BEMS) optimization, by modeling occupants' behavior towards an agent-based approach

In the last decades, in response to the high impact of buildings on global energy consumption and on the greenhouse gases emission, recent international directives have introduced the standard of "Nearly Zero Energy Building (NZEB)" to be realized from 2021. Despite the increasing attention to the development of strategies and innovative technology solutions for the energy efficiency of building components and HVAC systems, the human dimension, especially regarding the operating modes of the building-HVAC system by occupants, is often neglected. In most cases, this causes a significant discrepancy between the designed and the real total energy use in build-ings. Indeed, monitoring studies for identical dwellings having the same type of installations have shown great variation in energy use. Occupants constitute one of the major source of microclimate alteration in built environment, both as "passive agents" (for sensible and latent energy emissions, and emissions of pollutants), both as "active agents" as result of interaction with the buildings in order to achieve the desired comfort level (by acting on thermostats, by changing the state of opening or closing of windows and /or shading, by activating artificial lighting, etc.) Above all in the buildings characterized by higher levels of the insulation and air tightness, the occupants behavior may have a great influence on the energy con-sumptions and on indoor environment conditions. If the occupants have the possibility to manipulate the set-points temperature, the ventilation rates etc., the performance of the building will be affected by the behavior of the occupants. As consequence, even the most efficient building, may give rise to waste in case of incorrect use by occupants. Nowadays the understanding of occupant behavior results inappropriate, overly simplified, leading to inaccurate expectations of building energy performance. A common approach to model occupant behavior consists of assumptions based on scientists’ thoughts or literature reviews. Typically human actions (operation of lights, blinds, and windows) are modeled based on predefined fixed schedules or predefined rules. In contrast to the deterministic methods, stochastic and above all agent-based models (ABM) are the most powerful and suitable methods for modeling a system as complex as the human behavior. Especially in residential buildings, where the interaction of the occupants on the building-HVAC system is significant and hence the occupant behaviors may affect highly on building performance, the integration of Building Energy Management Sys-tems (BEMS) may provide significant energy savings, going not only to remedy an incorrect or inadequate management by occupants, but also optimizing the activation timing and management methods. Strictly connected with the “resilience” concept, the object of this research is to design “adaptive” Building Energy Management Systems (BEMS), able to maintain energy performance at the desired level despite the diverse operating conditions by occupants, by optimizing building components. In detail, several control logics for BEMS are analyzed in the residential buildings, by optimizing the thermal and visual comfort and by modeling the occupant behav-iors by means of an agent based oriented approach. In this thesis the optimization goals are based on the adaptive thermal comfort according to EN 15251. The thesis is structured in five chapters. In the introduction chapter (chapter 1) the main factors influencing the building performance towards the design of the NZEB are presented. Then, literature review regarding different studies that have analyzed the impact of occupant behaviors and the interaction with building-HVAC system (chapter 2) are reported. The results of a questionnaire survey conducted on occupant behaviors in residential buildings are described in the chapter 3. Large differences in the behavior patterns of occupants are found between dwellings. Indeed, for the oldest buildings, where the thermal discomfort conditions are the highest, the occupants usually turn on active system, by causing more energy waste. Furthermore, it is resulted that while in winter occupants act less on the building components to improve their thermal comfort conditions, (indeed the main actions are wearing heavy clothes and turning on heating system), in summer season the occupants mostly interact with the building components, by changing the window and shading status or by adjusting set-point thermostat. Because the actions on window and blind status are impactful on building perfor-mance, with the aim of reduce the thermal discomfort conditions and hence the vari-ability tied to the occupant behaviors, control logics of natural ventilation and of the solar shading system for passive cooling are designed. Indeed by reducing the ther-mal discomfort conditions, also the actions and the interactions of occupant with building components may be less. In detail, in an Italian dwelling with technological/typological features of sixties buildings, several studies are conducted with the aim to design BEMS for passive cooling that minimize the thermal discomfort situations, by means of Particle Swarm Optimization (PSO) method. The results of these studies are reported in the chapter 4. In the second part of the work, in order to have BEMS adaptable to the actions and preferences of occupants, a further study is conducted (Chapter 5), where occupant behaviors are simulated in more detail, by means of an agent-based approach. In detail, actions like opening/closing windows and shielding and cooling system activation are implemented in the energy software simulation (TRNSYS), using algorithms deduced by field investigations in real buildings. The same control logics of the BEMS (reported in the Chapter 4) are then revalued in this different occupant behavior modeling and the comparison between the models where the occupant behavior is assumed in deterministic way and then though a probabilistic and agent-based approach, allowed to assess the impact of human behavior and the designed BEMS on building performance. This work highlighted how the BEMS may ensure high levels of comfort and energy efficiency, through the dynamic control of some components based on external and internal environmental parameters and on the occupancy conditions. The implementation of different occupant behaviors into energy simulation soft-ware, simulated by means of an ABM method and the coupling of optimization goal for BEMS represent an innovative contribution of the work. A co-simulation architec-ture is created between TRNSYS (for building-HVAC model), TRNFLOW (for building air flow network), MATLAB (for PSO optimization) and DAYSIM (for visive analysis).