THE E-VISIØN PROJECT: ELECTRIC-VEHICLE INTEGRATION FOR SMART INNOVATIVE 0-CO 2 NETWORKS

Transcription

1 THE E-VISIØN PROJECT: ELECTRIC-VEHICLE INTEGRATION FOR SMART INNOVATIVE 0-CO 2 NETWORKS G. Celli, S. Mocci, N. Natale, F. Pilo, S. Ruggeri, G. G. Soma DIEE - University of Cagliari Cagliari, Italy susanna.mocci@diee.unica.it Abstract The low-carbon society requires significant changes in the operation and planning of distribution systems. In order to exploit existing low voltage systems, Demand Side Integration is essential for offering services that increase security and quality of supply, improve energy efficiency and reduce the energy cost. The paper describes the three-year project e-visiøn, currently under completion, aimed at the integration of plug-in electric vehicles in Smart Grids and the simultaneous better exploitation of renewable energy sources in order to reach zero-co 2 networks and sustainable mobility. Keywords Distribution networks, Smart Grids, Electric Vehicles, Fast Charging Stations. I. INTRODUCTION The achievement of the international targets in the reduction of CO 2 and greenhouse gas emissions requires, amongst possible actions, a massive exploitation of Renewable Energy Sources (RES), combined with a radical change in both public and private transportation systems. Indeed, the concept of e-mobility (electrical-mobility) has been gaining importance in the international scenario, thanks to its potential ability of ensuring a drastic reduction in CO 2 emissions (the transportation sector alone accounts for 27-30% of the total) and for the natural integration with RES that need load flexibility and storage in order to minimize the harmful unpredictability and not dispatch-ability. This new mobility paradigm is based on the use of Electric Vehicles (EVs), equipped with storage systems capable not only of ensuring adequate autonomy but, in the more futuristic scenarios, also of injecting reactive and active power into the grid while connected to domestic and public recharging systems (the so-called Vehicle-to-Grid concept, V2G). The e-mobility is capable to offer a positive contribution to the operation of wind and photovoltaic generation in the distribution system. But, in order not to increase the burden on distribution systems, the position and size of public charging stations should be carefully planned, by deeply integrating power and urban planning. Furthermore the operation of distribution systems should be changed in order to include the flexibility that comes with the storage capabilities of EV in both passive and V2G applications. These are the main subjects of the project e-visiøn (electric-vehicle integration for smart innovative 0-CO 2 networks, Fig. 1). By following an internationally recognized approach, e-visiøn is devoted to operation, by developing algorithms to include EVs into the smart operation of distribution systems, and to planning, by including operation models into planning strategies [1]. Furthermore, attention has been paid to regulatory and market issues as well as to the identification of roles and functions of new players to make it possible the participation of EV owners. e-visiøn developed control algorithms for the optimal operation of EV fleets with public charging stations and domestic systems in coordination with other electrical energy storage and RES production systems. These algorithms have been integrated into a Distribution Management System (DMS) model. The DMS operates the system to fix voltage and power flows issues. The stationary and mobile storage systems and the coordination with flexible demand improve the quality of DMS actions, reduce operation expenditures and minimize RES curtailment. Finally, the operational models for smart grid have been included in medium-long term expansion planning algorithms since the EV influences the development of the distribution system. As an example, in the paper the problem of optimal allocation of fast charging stations is dealt with by using a multi-objective approach that finds a compromise between EV owners and the distribution system operator (DSO) needs. Fig. 1 The e-visiøn logo. II. THE E-VISIØN PROJECT The e-visiøn project is developed by the Research Unit of Electrical Systems for Energy at the Department of Electrical and Electronic Engineering of the University of Cagliari. The Sardinian Regional Government funds the project, lasting three years. e-visiøn is a research project funded under the framework of the L.R. n. 7/2007. It is split into four Work Packages (WP):

2 WP1 - State of the art, training and methodological approach; WP2 - Models for the integrated operation of V2G and RES; WP3 Distribution planning algorithms for the transition to Smart Grid with RES and V2G; WP4 - Dissemination of the results. A. General Objectives The main objectives of the e-visiøn project are: To assess the impact of EV in the distribution grid, with particular reference to the peculiarities of the system of Sardinia. To study the coordinated use of EVs and active management, or Smart Grid (i.e. control and dispatching of Distributed Generation, DG). To increase the capacity of integration of the DG in distribution networks to maximize production from RES through the coordinated use of EVs and Smart Grid. To develop flexible planning tools that enable the efficient use of energy resources, including EVs, taking into account the capabilities of Smart Grid in network management. To give young researchers key skills for the development of the entrepreneurial system of Sardinia in which RES and e- mobility will take a key role. B. E-mobility Scenarios In the project, two scenarios are taken into account: a short-term scenario and a long-term scenario. 1) Short-term scenario The development of e-mobility in Europe will focus on EVs as city car with still relatively small autonomy (within km), now coming to distribution via the ordinary commercial networks, or hybrid plug-in vehicles rechargeable by the network but with an extended autonomy thanks to the presence of endothermic engines. Today, an EV travels on average 160 km with an energy input into the battery equivalent to about 3 litres of gasoline, thanks to a specific real average consumption of 150 Wh/km (battery of about 25 kwh). It is expected that the autonomy will increase over the next five years, both for technological refinement of the vehicles (general optimization of weight and greater braking energy recovery) and by increasing the capacity of the battery installed (up to kwh). The majority of EVs will make use of "slow" recharge (about 6-8 hours, obtained with a power of 3 kw, even without wire - with new wireless technologies magnetic resonance - in the home environment or non-domestic) or of public "fast" recharge facilities with power in the order of kw and recharge times between 40 and 80 minutes. 2) Long-term scenario The forecasted technological improvement of batteries (about 25 kwh today to at least kwh) will allow autonomy levels of several hundred kilometres, necessary to integrate urban mobility and long distance or inter-regional mobility. The new generation of EVs will require "quick" or "immediate" recharging systems, capable to recharge a high autonomy battery in the same times of a traditional fuel supply. These FCSs must be connected to the MV network. A home charging compatible with charging times of a few minutes is not conceivable. They will make use of smart technologies that combine with the European initiatives for the development of Smart Grids. The availability of technical standards at European level will be crucial to promote a common system of charging for smart cars, commercial vehicles, electric assisted scooters and bicycles. Appropriate planning tools are developed for the identification of best compromise solutions that take into account the benefits offered by Smart Grids even at low voltage level where the impact of EV is critical. The extension of Smart Grid planning and operation to LV systems is a complex task for the lack of information about LV system, the number of connection points involved and the lack of planning programs in DSO headquarters. From MV point of view, the problem of planning and operation of systems hosting significant quantities of EV is accrued by the need of charging stations, fast and ultra-fast. These new high demanding consumption points will allow charging the vehicle in minutes and will therefore need to be localized in strategic points easily accessible by vehicles that need filling but should not impact the network excessively [1]. III. SHORT TERM SCENARIO OPERATION OF DOMESTIC CHARGING STATIONS In the short term the EVs do significantly impact MV and LV networks since the recharge process will be essentially domestic or public but slow. Indeed, EVs absorb and store energy when parked and plugged increasing the load demand, mostly in the evening peak hours, when the owners come back home from work and want to charge the EV as fastest as possible. This behaviour (called dumb charging, Fig. 2) causes thermal overloading of distribution transformers and cables, voltage excursions mostly in the LV part of the distribution network and increases energy losses. Nevertheless, the EVs offer several benefits: increasing the flexibility of the electrical distribution system operation, improving the reliability (in customer side) of the distribution system, providing extra economic benefits to the vehicle owners, and reducing the home electricity purchase cost. Fig. 2 Load demand and EV charging profile in the dumb charging.

3 In this context, it is obvious the need of control systems that allow the coordination of all the resources in the network (DG, active loads, and EVs) under the constraints of the grid security and quality of supply. In the e-visiøn project two control techniques have been studied and developed. A. DMS Control In [2] a centralised DMS for the operation of active distribution networks is developed within the framework of the ATLANTIDE project [3]. The algorithm is capable to find solutions that allow increasing the network hosting capacity and improving the efficiency of power delivery. It relieves power flow congestions at MV or LV level, collects information like energy price, forecasted load demand and weather forecast, improves voltage regulation and allows fast reconfiguration, by sending signals to the Distributed Energy Resources (DERs) local controllers (Fig. 3). The Aggregator implemented aggregates the total demand to sell energy and services in the electricity market, acquires the flexibilities of the consumers involved in the AD program, and also collects a large amount of information about EVs characteristics and EV drivers behavior (Fig. 4). As a consequence, the Aggregator, or Master Agent, coordinates independent Agents for moving demand to the most convenient hours without degrading power quality. The AD Agents and the EV Agents represent the active customers and EVs, respectively. These Agents use a combination of local information (own behaviour, flexibility, EVs characteristics) and global data (energy price) to perform the optimisation following the Nash s theory on games. The MAS does not require huge bidirectional information flows and is better suited than centralised control systems in LV networks. Fig. 4 MAS Decentralised control (MV/LV level). Fig. 3 DMS centralised control. The centralised control in the primary substation (HV/MV) is not particularly suited to include in the operation big numbers of flexible customers and EV charging stations. LV customers are seen by centralised DMS as aggregated loads whose behaviour is taken into account with probabilistic models. Indeed, the centralized control system is not a suitable solution for LV systems because it needs significant computational resources and expensive data communication infrastructures. For these reasons e-visiøn models are mainly devoted to decentralised operational systems coordinated with the centralised ATLANTIDE DMS model. The two research projects have mutual interactions; particularly e-visiøn uses ATLANTIDE network models and DMS. B. Multi-Agent System (MAS) Control In [4]-[5] authors proposed a Multi-Agent System (MAS) for the control of LV networks with active loads and EVs, with a Master-Slave interaction that allows finding a global optimum without a direct control of each resource. The MAS control exploits the model of the Aggregator developed in recent European projects (e.g. ADDRESS) [6]. C. Comparison of Decentralised and Centralised Approach In [7] authors proposed a comparison of the outcomes of the centralised and the distributed control systems, respectively. The MAS for handling AD implementation based on direct control of loads in the LV distribution networks is compared with a centralized DMS. The main distinction is where decisions are taken. The optimization algorithms proposed allow designing valid and effective demand response program, able to analyse and meet the load needs to contribute to the voltage control of the distribution network. The outcome is the difference between the solutions obtained with the two different control systems and a comparison between the system costs in both cases, with particular reference to LV systems. Since the costs in different strategies are comparable, the DMS presents lower total costs than the decentralised MAS. On the other hand, it is important to notice that the MAS direct control considerably reduces the flow of data and information and can be better suited than centralized control systems for LV applications. IV. LONG TERM SCENARIO PLANNING FOR INTEGRATING FAST CHARGING Fast charging refers to DC charging poles with nominal power equal to or higher than 50 kw. Consequently, a FCS will be characterized by high momentary peak power absorptions and it is connected to MV networks. For those

4 reasons, new tools are needed to assess the EV impact on the electric distribution network and to correctly plan the expansion of the power system. The planning tools developed by the authors are capable to include smart network operation and EV provided that models of new load are given. For these reasons, in [8] daily demand profiles of the FCS, modelled by the authors, are used in a new planning tool to find the optimal allocation of public fast charging stations in a given network structure. The tool applies an evolutionary Multi-Objective philosophy, based on the Nondominated Sorting Genetic Algorithm-II, in order to minimize the negative effects of concentrated and intense absorptions of power in the FCSs across the distribution network without penalizing their diffusion in the urban texture. A. Definition of FCS Load Profiles The impact of EV charging on electricity grids is becoming an increasingly important subject of study, but detailed knowledge about the future charging profiles of EVs is missing in Literature. For these reasons, it is crucial to analyse these situations and to model the FCS consumption, in order to correctly plan the expansion of the MV system. In [9] an original Monte Carlo methodology for assessing the power demand of a FCS has been implemented. The methodology allows assessing the average consumption and its variance taking into account several stochastic quantities and the availability of FCS nearby. The analysis of the results allows, for instance, identifying the impact of FCS number and position on queue effect. The maximum values of the demand in each hour of the load profile is bounded to the maximum power available, and some saturation effects start to appear shifting the time of fast recharge also in the night. In Fig. 5 and Fig. 6 this behaviour is shown for the case with 15 FCSs and with 10 FCSs. Obviously, in these cases the average duration of the fast recharge increases due to the waiting time spent on the queue (about 24 minutes with 15 FCSs and more than 36 minutes with only 10 FCSs). The EV charge profiles should be used in the planning and in the EV impact analysis of the future networks. B. Optimal Allocation of FCS with NSGA-II Since the effective FCS power demand is directly influenced by the number of charging facilities available as well as by the traffic conditions, in [8] an integrated probabilistic planning algorithm has been developed for the optimal placement of FCSs. The methodology, based on a Multi-Objective approach, allows considering multiple contrasting goals in order to drive the research towards a configuration not only economically and technically convenient but also robust in respect to the selected objectives. The optimization algorithm is based on a MO technique named NSGA-II. The NSGA-II algorithm sorts a population into different non-dominated levels or fronts (the nondomination rank). Initially, it finds the Pareto optimal set of the current population; then, it excludes temporarily these solutions and searches again the Pareto optimal set among the remaining individuals of the population. This process is repeated until all fronts are identified and associated to all individuals. In order to allot a unique fitness value to each solution, a second attribute is calculated that orders the individuals in each front on the basis of their density along the front (crowding distance). Fig. 5 Daily power demand profile of the whole EV fleet for fast charging with 15 FCSs available. Fig. 6 Daily power demand profile of the whole EV fleet for fast charging with 10 FCSs available. In the optimal allocation, the network topology is assumed fixed, all the branches are known, and the evaluation of the objective functions depends only on size, expected daily demand profile and location of FCSs. For those reasons, each solution has been coded by using a vector, whose size is equal to the number of MV/LV nodes, in which each element contains the information on the presence of a FCS unit. The network considered for the study is the representative urban network of the Italian MV distribution system, developed in the ATLANTIDE project [3]. An in-depth analysis of the solutions, often assisted by a Decision Theory tool, allows the planner to select the most adequate according to his goals or the best compromise over the Pareto set. Looking at the DNO point of view the best configuration (solution C in Fig. 7) is identified by a minimum number of FCSs (three, one for each sub-area) connected to the feeders with the lowest energy demand, conditions that allow minimizing the overall energy losses. Also for FCS owners there is convenience to install a low number of charging facilities in order to limit their investments and maximize their incomes.

5 The best configuration (solution B in Fig. 7) is characterized by three FCSs connected to the sub-areas with medium or high traffic density. In contrast with the previous stakeholders, the EV drivers aspire to maximize the number of FCSs so as to minimize the charging time. A good compromise among contrasting goals can be found in the knee of the Pareto front. Fig. 7 Best FCS configurations for DNO and FCS owners in urban distribution network. V. MAIN RESULTS In the e-visiøn project different techniques, based on original algorithms, to control, operate and integrate DER into active networks have been analysed and developed. Particularly, a distributed control system based on MAS technology for direct control of the active participation of small consumers in the electricity system that support the integration of the EVs in the LV distribution network, reducing their harmful impact on voltage regulation has been developed. Moreover, the definition of a FCS daily load profile has been the first step of a more complex study that aims at identifying the optimal number and position of FCS in a given area, taking into due account of the expected consumption for fast recharge. Since the effective FCS power demand is directly influenced by the number of charging facilities available as well as by the traffic conditions, an integrated probabilistic planning algorithm is proposed for the optimal placement of FCSs. The proposed methodology, based on a Multi-Objective approach, allows considering multiple contrasting goals in order to drive the research towards a configuration not only economically and technically convenient but also robust in respect to the selected objectives. VI. FURTHER WORK The further research is intended to assess the impact of RES and V2G in the development of distribution networks, distinguishing different types of load density in urban areas. The planning studies will evaluate the positive impact of the use of V2G in terms of greater integration of RES with less investment to upgrade the network; the costs for the various parties involved, the need for any incentives and/or feed-in tariffs, the recovery of energy efficiency and reduction in CO 2 emission will also be evaluated. ACKNOWLEDGMENT The project e-visiøn (electric-vehicle integration for smart innovative 0-CO2 networks) is financially supported by the Sardinian Regional Government (L.R. 7/2007). Progress of this project can be found at REFERENCES [1] "Planning and Optimization Methods for Active Distribution Systems", CIGRÈ Technical Brochure 591, 2014 Working Group C [2] F. Pilo, G. Pisano, G. G. Soma, Optimal Coordination of Energy Resources with a Two-stage on-line Active Management, IEEE Trans. on Industrial Electronics, Oct 2011, vol. 58, n. 10. [3] ATLANTIDE Project [Online]. Available: [4] S. Mocci, N. Natale, S. Ruggeri, F. Pilo, Multi-Agent Control System for increasing hosting capacity in Active Distribution Networks with EV, Proceedings IEEE Energycon Conference, Dubrovnik, May [5] S. Mocci, N. Natale, F. Pilo, S. Ruggeri, "Demand Side Integration in LV Smart Grids with Multi-Agent Control System", in Elsevier - EPSR Electric Power System Research Journal, Vol. 125, pp , August [6] ADDRESS Project. [Online]. Available: [7] S. Mocci, N. Natale, F. Pilo, S. Ruggeri, "Decentralized and Centralized Approach in the Active Management of Distribution Networks: a comparison through business cases, in Proc. of CIRED 2015, Lyon, June [8] G. Celli, S. Mocci, F. Pilo, G. G. Soma, "Planning of Fast Charging Station Placement", in Proc. of International Conference on Electricity Distribution CIRED 2014 Workshop, Roma, June [9] G. Celli, F. Lacu, S. Mocci, N. Natale, F. Pilo, G.G. Soma, "Aggregated Electric Vehicles load profiles with Fast Charging Stations", in Proc. of Power System Computation Conference PSCC 2014, Wroclaw, Polonia, August 2014.