Summer University. Implementing city and citizen friendly electric vehicles May Palma (Mallorca) Dr. Liana Cipcigan

Transcription

1 Summer University Implementing city and citizen friendly electric vehicles May 2014 Palma (Mallorca) Dr. Liana Cipcigan Cardiff University, School of Engineering

2 My Background Member of CIREGS Centre for Integrating Renewable Energy Generation and Supply Member of Electric Vehicle Centre of Excellence, Leader of the Energy Management Theme Invited Expert of Working Group Smart Charging under CEN/CENELEC M468 and M490 Member of WG p , Guide for Transportation Electrification, IEEE Standards Association Member of the Low Carbon Vehicles Steering Group, Welsh Government PI of the projects EPSRC - Smart Management of Electric Vehicles EPSRC - Electric Vehicle Value Chain, Bridging the gaps TSB - Agent-based controllers for electric vehicles and microgenerators TSB - Ebbs and Flows of Energy Systems Member of the projects EPSRC Grid Economics, Planning and Business Models for Smart Electric Mobility ICT-Smart Energy Grids - MAS 2 TERING - Multi-Agent Systems and Secured coupling of Telecom and Energy grids for Next Generation smart grid services IEE - ebridge: empowering e-fleets for business and private purposes in cities IEE - I CVUE : Incentives for Clean Vehicles in Urban Europe COST ARTS Autonomic Road Transport Systems FP7 - MERGE Mobile Energy Resources in Grids of Electricity INTERREG - ENATE European network on electric vehicles and transferring expertise UKERC - Scenarios for the development of smart grids in the UK EADS - Innovation Works - SCADA Cyber Security Lifecycle 2

3 Electrical Infrastructure Smart Grid a Network of Networks MARKET Transportation/Charging Infrastructure Intelligence Infrastructure 3

4 Cardiff University Integrated approach of s integration New products New clients New technologies, new suppliers How will the market respond? How do I sell these? What if something breaks? Range anxiety; cost Who invests? Who makes money? What s in it for us? Do we need more capacity? Smart grids? Intelligent infrastructure / Smart Grids INTEGRATED MODEL 4

5 Why go Electric? Independent drivers for E-mobility Energy: Smart Grids Transport: Sustainable urban mobility Smart Cities Electromobility as part of the Smart Grid As mobile loads, sufficient energy to charge the electric car Control charging Electric car as storage devices Electromobility as integral part of urban mobility CO2 reduction through electric car Complete urban mobility through integration if individual and public transport Infrastructure interdependencies 5

6 Is charging an Electric Vehicle as simple as plugging in? Analysis, Electricity grid How many? uptake scenarios, impact on generation system, impact on distribution networks When will they charge? temporal analysis Where will they connect for charging? spatial analysis Evaluation & Control, smart charging What are the infrastructure challenges of fleet? What are the options for managing the spatial-temporal nature of the load? What is the role of the Aggregator? Intelligent charging? Synergies with Smart Grids? Electricity and Transport systems integration as a link between the electricity and transport systems 6

7 Drivers for uptake Charging infrastructure Plugged-in Places (PiP) program ( ) provided match-funding worth 30 million for the installation of charging points - London, North East and Milton Keynes; then extended to Northern Ireland, Scotland, Greater Manchester, Midlands and East of England Stimulated private investment in the sector (non-pip CPs now represent around 70% of installed points). Half of public PiP charge points are rated at 7kW (~50%), followed by 3kW units (~45%), while fast and rapid CPs (20kW+ AC and 40kW+ AC and DC) only account for a small proportion of the network (~5%, < kW DC). Element Energy, Pathways to high penetration of electric vehicles, Final report for The Committee on Climate Change, December

8 Charging infrastructure 8

9 Drivers for uptake Policy and support measures Element Energy, Pathways to high penetration of electric vehicles, Final report for The Committee on Climate Change, December

10 Electric Vehicles and Power Networks From a power system viewpoint, Electric Vehicles (s) may be seen as: Simple loads, drawing a continuous current from the electricity network. Flexible loads that may allow a management company to interrupt or coordinate their battery charging procedure. Storage devices that may allow a management company to interrupt or coordinate their charging procedure, or even request power injections from their batteries back to the grid. The latter is known as Vehicle to Grid (V2G) concept. Our Research Impact of battery charging on grid demand at a national level. Impact of battery charging on distribution networks. Intelligent control algorithms for coordination of battery charging. 10

11 How are the load shape and peaks going to be affected with battery charging? 11

12 Peak Electricity Demand (GW) Without s (69.9GW) Low Uptake Uncontrolled Dual Tariff Variable Price Mixed Charging High Uptake Uncontrolled Dual Tariff Variable Price Mixed Charging Without s (70.1 GW) Low Uptake Uncontrolled Dual Tariff Variable Price Mixed Charging High Uptake Uncontrolled Dual Tariff Variable Price Mixed Charging Impact on Grid Demand Peaks at National Level Great Britain Spain Peak Electricity Demand (GW) Ref P. Papadopoulos, O. Akizu, L. M. Cipcigan, N. Jenkins, E. Zabala, Electricity Demand with Electric Cars: Comparing GB and Spain, Proc. IMechE Vol. 225 Part A: J. Power and Energy, pp , (2011) 12

13 Electricity Demand (GW) Installed Generation Effective Generation Low Uptake Demand without s Demand without s Low Uptake Effective Generation Installed Generation Electricity Demand (GW) Electricity Demand (GW) GB/ Winter Season Selected Results and Conclusions Spain/ Winter Season Time of Day Time of Day Uncontrolled Mixed SPAIN Charging Dynamic Price GB Dual Tariff Base Case % % 32% Electricity Demand (GW)

14 impact on Generation at National Level ~ 3mil cars of ~42mil vehicle fleet (7% Low market penetration prediction) Isn t enough to make a real impact on energy demand at the national level s impact is expected to be at the local level Impact on LV distribution hotspots depends on clustering 14

15 Impact of battery charging on Distribution Networks Low Voltage Network Medium Voltage Network Parking Area Parking Area Parking Area How are the distribution network equipment and operating parameters going to be affected with battery charging? Key Power Network Electric Vehicle Smart Meter Communications Network charging point T Transformer Distributed Generation 15

16 Impact of battery charging on UK generic distribution network Assumed Uptake Levels in 2030 Low uptake (12.5% of residences) Medium uptake (33.3% of residences) High uptake (70.8% of residences) Source ~ 500 MVA 33/11.5kV UK Generic Distribution Network 384 residences 3072 residences 96 residences 11kV/ 0.433kV Residential charging of batteries will overload distribution networks and modify voltage profile of feeders. The distribution transformer was found to be overloaded for medium and high penetration. The voltage limits would be violated for medium and high penetrations. The 185mm 2 cable was found to be overloaded for most 2030 cases. The results from this research are used for the design of algorithms to allow the efficient management of charging infrastructure Ref Papadopoulos P, Skarvelis-Kazakos S, Grau I, Cipcigan LM, Jenkins N, Electric vehicles' impact on British distribution networks, IET Electrical Systems in Transportation, 2 (3) (2012)

17 Load Forecast The electricity demand will increase due to the recharging of the batteries s impact is expected to be at the local level creating hotspots depending on clustering Smart Management of Electric Vehicles charging Accurate Forecasts: Critical Component for the successful control Feedback Loop necessary to improve the forecast. 17

18 Load Forecast Influential Factors-Attributes Connection time Disconnection Time Seasonality Periodicity Average charging demand Weather Charging Mode Battery capacity Travel patterns Electricity price Big variation Artificial Intelligence methods 18

19 Case study United Kingdom Around 15,000 Charging Events Domestic, Commercial and Public Charging Events 396 owners 200 Charging Stations Geographical Location of each Station Charging Rates and Technical Specifications of most stations Total Energy 1242 MWh 19

20 Distribution of Charging Rates for each Parking 20

21 Average Charging Demand Profile per Week 21

22 Average Daily Charging Demand Train dataset Test Dataset 22

23 Forecast of the Average Daily Charging 23

24 Case Study France Real Charging Events Paris 71 s Public chargers 27/4/2011 until 29/2/ charging events Total Energy kwh 11.10kWh Average Energy per charge event E Xydas, C Marmaras, L M Cipcigan, A S Hassan, N Jenkins, Forecasting Electric Vehicles Charging Demand Using Support Vector Machines, presented in UPEC 2013, Dublin E Xydas, C Marmaras, L M Cipcigan, A S Hassan, N Jenkins, Electric Vehicle Load Forecasting Using Data Mining Methods, 4th Hybrid and Electric Vehicle IET Conference (HC 2013), London 24 24

25 Case Study France 25

26 Case Study The project (USA) The Project The largest deployment of electric vehicle charge infrastructure in history 7937 s were enrolled until now Over 2.3 million charging events 8,278 Residential SE Installed 3,166 Commercial 76 DC Fast Chargers E Xydas, C Marmaras, L M Cipcigan, A S Hassan, N Jenkins, Forecasting Electric Vehicles Charging Demand Using Support Vector Machines, presented in UPEC 2013, Dublin E Xydas, C Marmaras, L M Cipcigan, A S Hassan, N Jenkins, Electric Vehicle Load Forecasting Using Data Mining Methods, 4th Hybrid and Electric Vehicle IET Conference (HC 2013), London 26

27 Energy per charge event Residential Private Non-Residential Public 27

28 Charging Stations usage Residential Private Non-Residential Public 28

29 Next Day Load Forecast ANN SVM MAPE RMSE r It was estimated in the U.K. power system that every 1% increase in the forecasting error costs 10 million pounds per year*. *M. Matijaš, M. Vukićcević, and S. Krajcar, Supplier Short Term Load Forecasting Using Support Vector Regression and Exogenous Input Journal of Electrical Engineering, vol. 62, no. 5, pp , Sep

30 Smart Management of s Charging Virtual Power Plant * s??? Ref * Virtual Power Plant Concept in Electrical Networks. Juan Martí (2007) [FENIX project] 30

31 Electric Vehicle Supplier / Aggregator Basic Functions Individual s have small power capabilities; therefore their participation in the electricity and ancillary services markets will require a new entity: the Aggregator (A). The A will serve as an intermediary between large number of s and market players and/or system operators. Aggregator: Entity which sells electricity to the owners, aggregates and manages their load demand. 31

32 Direct Control Architecture Interaction between the VPP Control Center and the VPP resources, DSO, TSO and market in the direct control approach A. F. Raab, M. Ferdowsi, E. Karfopoulos, I. Grau Unda, S. Skarvelis-Kazakos, P. Papadopoulos, E. Abbasi, L.M. Cipcigan, N. Jenkins, N. Hatziargyriou, and K. Strunz, Virtual Power Plant Control Concepts with Electric Vehicles, ISAP 2011, Crete,

33 Hierarchical Control Architecture Interaction between the VPP control center and the VPP resources, DSO, TSO and market in the hierarchical approach 33

34 Distributed Control Architecture Interaction between the VPP control center and the VPP resources, DSO, TSO and market in the distributed control approach 34

35 Modelling Framework Control Techniques Optimization Reach the optimal solution, with given constraints Machine Learning Data Mining Behavior Analysis Game Theory Cooperative/Noncooperative Games between the s Bidding/Auction Games for the aggregators 35

36 Example Hierarchical Control MAS design aims Satisfy owner preferences according to: Distribution network technical constraints. Electricity prices. Location of agents in a power distribution system 36 P. Papadopoulos PhD thesis

37 PHYSICAL LAYER VIRTUAL LAYER Regional Aggregator agent DSO agent p a n Load Forecasting Low Voltage Network Medium Voltage Network Local Aggregator agent pan Parking Area Power Flow Parking Area Parking Area Operation of the Agent Based Control System 37

38 Resources Used Laboratory of Tecnalia User Interface Charging Point Controller Power Quality Meter User Interface Vehicle Controller -ON Platform Batteries Inverter Avtron K595 Load Bank Avtron Millenium Load Bank GaugeTech DMMS300 Measurement Device Connection Switch SoC Estimator Smart Meter Connection Cable Hardware Resources Software Resources 1. -ON Platform software was used for: State of Charge (SoC) measurement acquisition Set-point application 2. Communication Software for Distributed Energy Resources (CSDER) was used to: Monitor and switch on/off the load banks steps Acquire measurements from the DMMS300 measurement device 38 P. Papadopoulos PhD thesis

39 Experimental Evaluation of Hierarchical Control (Papadopoulos PhD thesis) UK Generic LV Distribution Network Grid Supply ~ Software Layer DSO Regional Aggregator Local Aggregator 31 simulated agents. HV/MV Substation Level 33/ 11.5kV DSO Regional Aggregator Load Banks Controller Medium Voltage Level 15,360 Residential Customers (Lumped) CSDER (IEC 61850) MV/LV Substation Level Low Voltage Level 11/ 0.433kV ,688 Residential Customers and 96 s (Lumped) 24 Residential Customers Key Equivalent 8 s Equivalent Agent 288 Residential Customers and 96 s (Lumped) Local Aggregator Residential Feeder ~ Grid Supply 400V/ 50Hz Physical Layer Key Normal Operation Emergency Operation Monitoring Control Electrical Communication Communication Connections Agent Create an equivalent of the residential feeder P. Papadopoulos PhD thesis DMMS Avtron Millenium Avtron K595 -ON 300 Load Bank Load Bank Platform 39

40 Feeder Loading (kw) Experimental Evaluation of Hierarchical Control (Papadopoulos PhD thesis) Experiment Conditions Purpose Experiment 1 Normal Normal operating operating conditions conditions, Experiment 2 aiming to minimise the load demand in a specific hour Experimental Results To evaluate the MAS operation during normal operating To evaluate conditions the MAS aiming operation to follow when the the electricity demand price reduction signals. would be required in the LV area, during a specific hour Reference Case: No Control Feeder Limit Actual Load Measured Load Load without s Experiment 2: 1: Demand Normal operation reduction 20 06:00 10:00 14:00 18:00 22:00 02:00 06:00 Time of Day P. Papadopoulos PhD thesis 40 40

41 Experimental Evaluation of Hierarchical Control (Papadopoulos PhD thesis) Experimental Results Experiment Conditions Purpose Experiment 3 Normal operating conditions and allowing s to provide power back to the grid To evaluate the MAS operation when power injections would be required from the specific LV area, during a specific hour. Experiment 1: Normal operation Experiment 3: V2G 41 41

42 DSO agent Experimental Evaluation of Centralised Control (Grau PhD thesis) Coordinator agent UK Generic LV Distribution Network agent agent agent Grid Supply 500MVA 33/11.5kV 11/0.433kV Local Area agent agent agent agent Workplace Area Grid Supply 400V/50Hz Experiments conducted: a) Validated operation. b) Technical invalidation. c) Emergency operation. d) Demand reduction service. Laboratory configuration Key: Normal Operation Emergency Operation Monitor agent Communication Software for Distributed Energy Resources DSO agent Load Bank Controller agent Coordinator agent agent... agent Local Area agent agent 42 42

43 Experimental Evaluation of MAS Centralised Control (Grau PhD thesis) Economic Optimisation Tests Executed Demand Reduction Emergency Operation Transformer Loading without s (kw) Transformer Loading Limit (kw) Electricity Prices (p/kwh) Transformer Loading with s µgrid Emulator Output (kw) Transformer Loading with s µgrid Actual Output (kw) -ON Demand 43

44 Conclusions Smart Management of Electric Vehicles Charging Control Options a. Distributed b. Hierarchical c. Centralized Charging options a. Uncontrolled - Prolonged transformer stress - Voltage Drops and Energy Losses b. Smart - Optimal charging profile improved by the Load Forecast c. From DG - When available, is the best option- Minimum impact to the Grid Optimal choice of the Charging Control Model - Location/Demographic Characteristics - Network Technical Characteristics - Cost - Expandability 44

45 New Market model & framework Turnkey (e.g. BetterPlace,Tesla) Investment business models? Charge Station Centric (e.g. Coulomb Tech) Utility Centric (e.g. ESB,RWE) Utility Demand mgmt Power supply Car Manufacturer Utility Demand mgmt Power supply & Billing (opt.) Car Manufacturer Utility Demand mgmt Power supply & Billing Car Manufacturer Electricity Distributor Own battery & charge station Electricity distributor Charge stations Billing Public and private charging stations by multiple operators Subscribe $/ mile Battery rental Pay to charge Pay utility bill Pay utility bill Charge Sell car w/o battery Sell car w/ battery (or lease) Sell car w/ battery (or lease) 45 Geoff Ryder, SAP Labs, LLC, ISM 270, January 27, 2011

46 Why go Electric? Supply Chain new entrants 1 Automotive industry 12 Energy Storage producers 2 Commercial and industrial manufacturers 13 Renewable power producers 3 Retailers 14 Regulators 4 Consumers, residential, commercial and industrial 15 Standard and specification development organisations 5 Electric transportation industry 16 Relevant Government Agencies 6 Electric distribution industry 17 R&D organisations and academia 7 Electricity and financial market traders (including aggregators) 18 Professional societies, user groups, and industry consortia 8 Charging infrastructure developers 19 services integrators 9 ICT infrastructure and service providers 20 Local councils, local government 10 Information technology application developers and integrators 21 Virtual Power Plants, Aggregators 11 State Regulators (Ofgem, UK) 22 Venture Capital, Investors.. 46

47 Conclusions Smart Grids and Integration Electric vehicle charge stations: grid connection points for power and ancillary services delivery Vehicles can respond very fast compared to power plants Interaction with the grid s becomes an active participant in grid operations Potential for energy storage Ancillary services Grid regulation s synergistic with Smart Grid Digital Communications - Information flow between vehicle and utility on some level is critical to maximizing value Information Flow Control Power Flow Control Decision Algorithms 47

48 Conclusions Smart Grids and Integration Electricity as a transportation fuel Complex management of large fleets Integrated analysis of electricity / smart grids / transportation / market Make charging infrastructure convenient for the user strong support to purchase Ensure that vehicle operators have sufficient energy for driving while enabling the delivery of that energy to vehicles in ways that minimize stress upon the grid The charging solution should be able to manage the power flow of network connection point in order to reduce peak load and thus avoid overloading of electricity connection. Pilot projects and experimental work experiences of what works, what doesn t and commonalities for standardization 48

49 Conclusions Smart Grids and Integration Not a revolution but evolution will evolve over many years Created through the incremental deployment and integration of system intelligence and increasing grid observability capabilities Intelligent systems deployed to meet specific customer, utility business and technical regulatory drivers and measurable benefits Each utility has different: Drivers and benefits Starting points Paths Deployment speed 49

50 Thank you! Dr. Liana Cipcigan Contact Details Cardiff University School of Engineering Queen's Buildings The Parade, Cardiff CF24 3AA Research Team: Babis Marmaras (RA), Kroton Xydas (PhD student), Dr. Panos Papadopoulos (graduated working at EDF, Dr. Inaki Gradu-Unda graduated)