Use of model predictive control for experimental microgrid optimization Alessandra Parisio a,⇑,1 , Evangelos Rikos b , George Tzamalis c , Luigi Glielmo d a ACCESS Linnaeus Center and the Automatic Control Lab, The School of Electrical Engineering, KTH Royal Institute of Technology, Sweden b Department of PVs and DERs Systems, Center for Renewable Energy Sources and Saving (CRES), Pikermi, Athens, Greece c Department of RES and Hydrogen Technologies, Center for Renewable Energy Sources and Saving (CRES), Pikermi, Athens, Greece d Department of Engineering, Università degli Studi del Sannio, Benevento, Italy highlights  A model of a microgrid with storages and controllable loads is developed.  Our storage model guarantees a feasible behavior of the storage unit.  Using our model the microgrid operations optimization problem is tractable.  An MPC controller for minimizing the microgrid running costs is developed.  The proposed method is applied to an experimental microgrid located in Greece. article info Article history: Received 7 June 2013 Received in revised form 2 October 2013 Accepted 10 October 2013 Available online 25 November 2013 Keywords: Model predictive control Microgrids Optimization Mixed Integer Linear Programming abstract In this paper we deal with the problem of efficiently optimizing microgrid operations while satisfying a time-varying request and operation constraints. Microgrids are subsystems of the distribution grid com- prising sufficient generating resources to operate in isolation from the main grid, in a deliberate and con- trolled way. The Model Predictive Control (MPC) approach is applied for achieving economic efficiency in microgrid operation management. The method is thus applied to an experimental microgrid located in Athens, Greece: experimental results show the feasibility and the effectiveness of the proposed approach. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction Microgrids are integrated energy systems comprising loads and a combination of Distributed Energy Resources (DERs), such as Dis- tributed Generators (DGs), which are controllable units, and Renewable Energy Resources (RESs), which are non controllable units. In a smart grid scenario, the microgrid concept is a promising approach, since it is capable of managing and coordinating DGs, storages and loads in a more decentralized way reducing the need for the centralized coordination and management [1]. The key concept that differentiates the microgrid paradigm from a conven- tional power utility is that the power generators are small. They are also located in close proximity to the energy users. A microgrid can be either grid-connected or islanded; when the microgrid operates in parallel with the grid, it can buy and sell power to and from its energy suppliers [2,3]. In this work we solve the problem of operating the microgrid in order to minimize its running costs while meeting the predicted energy demand over a certain period (typically one day). The over- all control strategy has to consider that the system must operate within its technical and physical limits; this means that complex operational constraints have to be fulfilled, such as DGs’ minimum on/off time. The microgrid operational management problem needs to in- clude policies for controllable loads (Demand Side Management, DSM), interaction with the utility grid and storage models, which require both continuous (such as storage charge or discharge rates) and discrete (such as on/off states of DGs) decision variables. Thus, the problem is generally stated as a Mixed Integer Nonlinear Prob- lem (MINLP) (see, for example, [4–6]), which are really hard to solve. The modeling capabilities and the computational advances of Mixed Integer Problem (MIP) algorithms, have led several Independent System Operators (ISOs) and Regional Transmission 0306-2619/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.apenergy.2013.10.027 ⇑ Corresponding author. Tel.: +46 700406605. E-mail address: parisio@kth.se (A. Parisio). 1 This work was supported by the European Commission, through the Distributed Energy Resources Research Infrastructures (DERri) project (EU Project No. 228449) and the Sustainable-Smart Grid Open System for the Aggregated Control, Monitoring and Management of Energy (e-GOTHAM) project. Applied Energy 115 (2014) 37–46 Contents lists available at ScienceDirect Applied Energy journal homepage: www.elsevier.com/locate/apenergy