Workshop on Optimal Control of Thermal Systems in Buildings using Modelica location:University of Freiburg date:23 - 24 March 2015
The research of optimal control in residential building clusters is approached from different disciplines: building simulations and control engineering. Control engineers focus on the research and development of sophisticated optimal control strategies combined with highlevel simulation tools but less accurate building models for fast prototyping of new control strategies. On the other hand, building simulation experts develop detailed building models which provide realistic and accurate building representations, however often in a simulation environment which is less suited for control. This paper proposes a methodology to extend a detailed neighborhood model in Modelica, which is an objectoriented modelling and simulation language, with a Python control layer in order to bridge the gap between both disciplines. The methodology tries to leverage the advantages of both approaches by enabling the combination of both in an integrated simulation while keeping the development of the building models and control strategies separate. Control algorithms developed in Python can then easily be tested on a detailed neighborhood model in Modelica. As such, the Modelica simulation model is used as an emulator or virtual test bed. These integrated simulations can provide new insights in the behavior of building clusters by using sophisticated control algorithms. The methodology is tested by implementing a central model predictive control distributed by a multiagent control system in Python on a residential neighborhood in Modelica. The results are promising: the test implementation shows improved control performance compared to rulebased control1 using a simple aggregated model for the MPC. Using this methodology, more complex and sophisticated MPC models could easily be implemented.Besides illustrating the methodology for interfacing both disciplines, this case study allows identifying the shortcomings and potential future work related to current simulation tools.