Human-Building Interaction Framework for Personalized Thermal Comfort-Driven Systems in Office Buildings

TLDR

Commercial HVAC systems are centrally controlled by BMS using predefined settings and assumptions, which fails to account for personalized comfort preferences and leads to occupant dissatisfaction. This study aims to integrate occupants into the HVAC control loop by learning their comfort profiles and using those profiles to drive HVAC operation. The framework fuses occupants’ comfort votes and ambient temperature data, applies a fuzzy rule‑based model to compute comfort profiles, and implements a proportional controller that regulates zone temperatures to lie midway between occupants’ preferred temperatures. Validation shows the algorithm accurately captures the nonlinear pattern of the thermal comfort scale, and controller experiments demonstrate that the proportional algorithm keeps zone temperatures within occupants’ preferred ranges.

Abstract

Centrally controlled heating, ventilation, and air conditioning (HVAC) systems in commercial buildings are operated by building management systems (BMS) based on the predefined operational settings and a set of assumptions. Despite the high rate of energy consumption by HVAC systems in commercial buildings, observations showed that a significant portion of the occupants remain dissatisfied with thermal conditions. One of the main reasons is that HVAC systems do not take into account personalized comfort preferences in their operational rules. This study proposes a framework to integrate building occupants in the HVAC control loop, learn their comfort profiles, and control the HVAC system based on occupants' personalized comfort profiles. The framework fuses occupants' comfort perception indices (i.e., comfort votes provided by users and mapped to a numerical value), collected through participatory sensing, and ambient temperature data, collected through a sensor network, and computes occupants' comfort profiles by using a fuzzy rule-based descriptive and predictive model. The performance of the comfort-profiling algorithm was assessed using human subject data and synthetically generated data. For actuation, a BMS controller was proposed and tested in two zones of an office building. The BMS controller uses a proportional controller algorithm that regulates room temperatures to be equidistant from preferred temperatures of all occupants in the same thermal zone. Validation of the framework components demonstrated that the nonlinear underlying pattern of the thermal comfort sensation scale could accurately be recognized. Results of the BMS controller experiments revealed that the proportional controller algorithm is capable of keeping the thermal zones' temperatures in the ranges of preferred temperatures.