D7.2a Designed distributed MPC based thermal energy control system

The present deliverable reports the conceived control framework for the supervision of the Smart Thermal Energy grid. In particular, a Distributed Model Predictive Control based framework has been designed by adopting a price-driven decomposition technique. In section 1, detailed requirements and objectives to be addressed are reported, by considering also recent developments from the Smart/Micro-grid community. In section 2, the designed distributed architecture is presented, including also the interactions between single company internal energy grid control and cluster level industrial park utility grid control as well as with the energy-aware scheduling system. Section 3 reports the identified Distributed Model Predictive control approach to be developed, by evaluating and comparing the features, advantages and limitations of alternative techniques with reference to the characteristics of the problem at hand. In particular, decentralized, non-cooperative, cooperative and centralized approached to the design of DMPC are compared. Then, Section 4 describes the implemented ADMM-based method for upper layer prices update, including dual decomposition and augmented Lagrangian calculation, minimization steps as well as the considered techniques for stopping iterative execution and improve convergence for the class of non-convex Mixed Integer Programming problem to be solved for Smart energy grid control. The equations constituting the single unit HMPC are reported within section 5. In fact, the resulting MPC module falls into the class of Hybrid MPC systems since it includes both continuous dynamics (as in charge and discharge of storage) and discrete events (start-up and shut-down of the generators and controllable loads). Finally, section 6 describes an easy to use Human Machine Interface that has been conceived to properly support agile configuration of optimization model parameters for the Smart energy grid generation units.