Conference Year

July 2018


predictive maintenance, performance decay, thermal energy storage, scheduling, cost minimization


While various models and methods have been proposed for operating and controlling building heating, ventilation and air conditioning (HVAC) systems, equipment decay (e.g., chiller tube fouling or boiler scaling), which results in lower energy efficiency and higher cost, has received limited attention. Accordingly, in this paper, we present an optimization model (mixed-integer linear programming, MILP) for predictive maintenance of HVAC systems with sufficient thermal energy storage (TES). We simultaneously consider the operation and maintenance schedule because of their close mutual interdependence. In addition, a method to accurately approximate equipment operation is provided for long-term scheduling. The proposed model offers decisions on execution of maintenance tasks based on the simultaneously optimized operation schedule (e.g., on/off status and load of equipment). Two computational experiments illustrate the applicability of the model. First, we show that the proposed model can approximate the hourly equipment operation without significant loss of accuracy. Second, we show that the model can lead to satisfactory cost saving when compared to the “fixed” schedule.