Modelica, TACO, Model Predictive Control, Optimisation, Control
Model predictive control (MPC) is a promising alternative to rule-based control since it is more suitable to control increasingly complex buildings and thereby realising energy savings and comfort improvement. Practical implementations are however hampered by the complexity of MPC and the expertise required for developing MPC. Therefore, a toolchain for automated control and optimization (TACO) has been developed that automatically translates an object-oriented Modelica model into an efficient MPC code. Since object-oriented models from the Modelica IDEAS library are used, the expertise requirement and development time are reduced significantly. TACO has, however, not yet been applied to a real building and its robustness in real operation still must be demonstrated. The purpose of this paper is to provide a comprehensive overview of the steps that are proposed for implementing an MPC using TACO. We therefore summarise our existing methodology and describe our future extension plans to implement an MPC in the Infrax office building in Brussels by September 2018.