Electric Power Systems Research 110 (2014) 172–179 Contents lists available at ScienceDirect Electric Power Systems Research j o ur nal ho me page: www.elsevier.com/lo cate/epsr Management of electric vehicle battery charging in distribution networks with multi-agent systems I˜ naki Grau Unda a,∗ , Panagiotis Papadopoulos b , Spyros Skarvelis-Kazakos c , Liana M. Cipcigan a , Nick Jenkins a , Eduardo Zabala d a Institute of Energy, Cardiff University, CF24 3AA Cardiff, United Kingdom b EDF Energy R&D UK Centre, SW1W 0AU London, United Kingdom c School of Engineering, University of Greenwich, ME4 4TB Kent, United Kingdom d TECNALIA, Parque Tecnológico de Bizkaia, E-48160 Derio, Vizcaya, Spain a r t i c l e i n f o Article history: Received 29 May 2013 Received in revised form 9 January 2014 Accepted 18 January 2014 Available online 16 February 2014 Keywords: Electric vehicle Multi-agent systems Distribution networks Active demand Smart grids Distributed control a b s t r a c t An agent-based control system that manages the battery charging of electric vehicles in power distribu- tion networks is presented. The electric vehicle battery charging schedules are calculated according to electricity prices and distribution network technical constraints. The design of the multi-agent system is described. The real-time operation of the multi-agent system was demonstrated in a test-bed of a laboratory micro-grid. © 2014 Elsevier B.V. All rights reserved. 1. Introduction A high electric vehicle (EV) uptake is expected in the forth- coming years [1]. EV battery charging will increase the electricity demand and distribution networks will be required to cope with this increase [2]. To aggregate EV resources and enable their bat- tery charging management, the control approaches that have been proposed in the literature are: (i) centralised approach such as [3–5] where the system’s decision making is implemented in a cen- tral coordination unit and (ii) distributed approach such as [6–10] where the system’s decision making is distributed among the con- trol units of the system. As reported in [6,7], one of the main drawbacks of centralised control approach is the computational intensity and data transfer for the battery charging management of large populations of EVs. Distributed control has been identified in [11] as one of the most promising applications of multi-agent system (MAS) in power systems applications, where the main benefits of using MAS are presented as: (i) flexibility, (ii) extensibility, (iii) fault tolerance, (iv) open architectures and (v) distribution. Some practical MAS ∗ Corresponding author. Tel.: +44 029 208 70669. E-mail addresses: grauinaki@yahoo.com, graui@cf.ac.uk (I. Grau Unda). implementations in power systems including distributed control can be found in [12]. The use of MAS for real-time management of EVs is proposed in the literature for load levelling [6,7], for reducing imbalance costs [8,9] and for the provision of frequency regulation services [10]. This paper contributes to the literature by demonstrating a real- time MAS control approach in which EVs are managed according to electricity prices, technical constraints and having the capability to provide active demand services. The management framework developed within the European Union (EU) project Mobile Energy Resources in Grids of Electricity (MERGE) [13] was adopted. The operational framework was further developed at Cardiff University within the EU project Distributed Energy Resources Research Infrastructures (DERRI) [14]. 2. Multi-agent system description The MAS has a hierarchical architecture, presented in Fig. 1, and consists of: • The electric vehicle aggregator, which is responsible for the EV battery charging management. The electric vehicle aggregator comprises three types of agents: EV agent, local area agent and coordinator agent. The EV agent, located in the EV, sends the 0378-7796/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.epsr.2014.01.014