Building performance optimization for operational rule extraction

Abstract

Operational parameters regarding daily and seasonal scheduling of systems' availability and setpoints play an important role in building performance. This paper first reviews the key operational parameters of 14 Canadian office buildings to better understand their range in practice. The results reveal that over 60% of the air handling units (AHUs) are not turned off outside of normal occupied hours, and most of them do not have an economizer cycle program. Subsequently, a mixed-integer genetic algorithm is applied to a building performance simulation model to identify its optimal operational parameters for four different climate zones, nine different occupancy and three different envelope scenarios. The effects of the variables of these scenarios and the optimal operational parameters are examined by employing a decision trees-based rule extraction method.