Development of a model predictive control framework through real-time building energy management system data

Over the past several years, studies have been conducted on the model predictive control (MPC), which has analyzed the amount of savings through model-based predictive control. As its result is dependent on the precision and accuracy of the model, minimizing the errors due to arbitrary events caused by the occupants as well as uncertain data input is important. In this study, to address these errors, real-time building energy simulation was conducted in accordance with which energy consumption could be predicted and a MPC framework was established for control. To use the building energy management system data, co-simulation was implemented based on the Building Controls Virtual Test Bed (BCVTB) as it allows for a real-time simulation based on the external data input. The real-time predicted and measured energy consumption values were compared to the statistical indices such as the hourly Mean Bias Error and Coefficient Variation of the Root-Mean-Squared Error with acceptable values of −0.7% and 19.1%, respectively. As a case study, an enthalpy control algorithm with the real-time monitoring data was successfully implemented producing the damper position of the air handling unit.