Summer School Smart Energy Systems 2013 Energy Informatics

Transcription

1 Summer School Smart Energy Systems 2013 Energy Informatics Hartmut Schmeck (KIT) Institute for Applied Informatics and Formal Description Methods (AIFB) Institute for Applied Computer Science (IAI) FZI Research Center for Information Technology KIT University of the State of Baden-Württemberg and National Research Center of the Helmholtz Association Brief intro: Hartmut Schmeck Chair of Applied Informatics at the Institute AIFB ( University mission of KIT) Director at Institute for Applied Informatics (IAI) ( Helmholtz mission of KIT) Director at Forschungszentrum Informatik (FZI) ( Innovation mission of KIT) Research Area: Efficient algorithms Bio-inspired optimization (genetic algorithms, ant colony optimization, ) Organic Computing Smart energy systems ( Energy Informatics ) Major projects (small selection): MeRegio(Mobil) using ICT to improve the energy system, integrate EVs and PHEVs into the energy system ( izeus, CROME, ) Organic Traffic Control self-organizing, adaptive traffic light control Observation and Control of Collaborative Systems fundamental research on self-organizing systems, generic observer-controller architectures, collaborative learning KIT Focus: COMMputation (= COMMunication + COMputation) 2

2 Major Projects 3 Moving towards Minimum Emission Regions Research Question / Scenario Energy Technology Smart Metering Hybrid Generation Demand Side Management Distribution Grid Management Energy Markets Decentralized Trading Price incentives at the power plug Premium Services System Optimization Objectives Optimize power generation & usage from producers to end consumers Intelligent combination of new generator technology, DSM and ICT Price signals for efficient energy allocation Combined Heat and Power MEREGIO-Certificate: Best practice + information dissemination Partners 4 ICT Real-time measurement Safety & Security System Control & Billing Non Repudiable Transactions Pilot Region with ~ 1000 Participants (Freiamt + Göppingen) (5 chairs at KIT: Energy Economics, Informatics, Telematics, Management, Law)

3 ICT for Electromobility Research Question / Scenario Objectives [source: EnBW AG] Intelligent & efficient integration of electric vehicles into the grid Technology assessment & feasibility under real life conditions Seamless integration into MEREGIO pilot region Center of competence at KIT (demo and research lab) Partners Methodology Computer Simulations Field trial with about 100 PEV Living Lab 5 (11 chairs at KIT: Electrical Engineering (2), Energy Economics, Informatics (5), Telematics, Management, Law) Further projects izeus intelligent Zero Emission Urban System (BMWi ICT 4 EM II) CROME Cross-border Mobility with Electric Vehicles French-German electric mobility Interoperability of infrastructures Driving patterns and user response to service infrastructure Leading Edge Cluster : Electric mobility south-west: road to global market Integration of vehicle technology, energy technology, information and communication technology as enablers of electric mobility 4 innovation areas: vehicle, energy, information & communication, production Projects within I&C: intermodal electric mobility, e-fleet management, smart grid integration, green navigation Helmholtz Energy Alliance Technologies for Future Energy Grids HeGrid Hybrid Energy Grid Management (EIT ICT Labs activity) 6

4 Motivation and Need for Energy Informatics Institute for Applied Informatics and Formal Description Methods (AIFB) Institute for Applied Computer Science (IAI) FZI Research Center for Information Technology KIT University of the State of Baden-Württemberg and National Research Center of the Helmholtz Association European Energy Targets: Strategic Energy Technology Plan : March 2007: EU targets to be met by 2020: 20% reduction of EU greenhouse gas emissions (relative to 1990) 20% share of renewables of overall EU energy consumption 20% increase in energy efficiency. (relative to 1990) More ambitious targets of Germany: Fall 2010: 30% renewables by 2020, 50% by 2030, 80% (??) by 2050 Spring 2011: Energiewende For a sailor, wenden means tacking: There won t be just one tack! Highly accelerated replacement of nuclear power with renewables (by 2022) 8

5 If a tack is not done properly 9 Problems: Fluctuations in Demand and Supply Small Scale Short Term Variations Mismatch Dead Calm Variations at different time scales, only partially predictable How to deal with fluctuations? demand and supply management How to compensate for a dead calm?? 10

6 German power generation PV Wind others Fossile Nuclear 11 Problems: Power Generation on , : PV 12 GW, wind 2 GW (peak), nuclear 10,3 GW (steady) 22.1.: PV. 1,8 GW, wind 22 GW (peak), nuclear 5,7 GW (steady) (source: 12

7 Problems with Power Generation : actual and planned on : actual: PV 23 GW, wind 6,4 GW (peak), , planned: PV. 17,3 GW, wind 5,9 GW (peak), How do you deal with these deviations between planned and actual production? (source: ) 13 Study on Energy Situation 2050 (Meteorological Base Year 2007) Power in GW Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Source: Fraunhofer IWES 14

8 Problems due to decentralization: bottlenecks in the low voltage distribution grid Local voltage increase due to PV power infeed Local voltage decrease due to EV charging These visualizations are a result of E-Energy project MeRegio. 15 Impact of PV power input on voltage in the low power grid voltage PV power Problem: all PV panels of one segment are in sync! 16

9 Energy Management: Balancing Demand and Supply Traditional: Demand cannot be controlled. Electricity cannot be stored. Principle: Supply follows demand (Spinning reserve: Primary, secondary, ) Future: Supply only partially controllable and decentralized Potential reversal of power flow New Principle: Demand has to follow supply! Requires more flexible demand Supply HV MV LV Demand Supply HV MV LV Demand Supply 17 Integrated Future Energy System Transmission grid Kurz und längerfristige Fluktuationen Flaute Ungleich gewicht Spannungserhöhung durch PV Distribution grid Collection grid Spannungsabfall durch E-Auto Energy flow (electricity) Information flow (Energy information network with distributed system intelligence) 18

10 Physical power grid SLP SLP SLP Privatkunden SLP RLM Gewerbekunden RLM Extra High Voltage Level High Voltage Level MediumVoltage Level LowVoltage Level Industriekunden Energy provider Transmission System Operator Distribution System Operator Metrering System Operator Energy Consumer / provider 19 Communication in Current Energy Market MSB pro vider VNB (DSO ) MSB Private customer ÜNB (TSO ) pro vider LM VNB (DSO ) MSB Business customer Control power plant RLM Industry customer Energy provider Transmission System Operator Distribution System operator Metering point operator Energy consumer and provider 20

11 Communication in Tomorrow s Power Grid Regelkraftwerk Energieerzeuger Und verbraucher 21 Numerous possibilities for vulnerabilities... d Regelkraftwerk 22

12 Where should system intelligence be located? What do we have to communicate? Power generators Power flow Communication PV PV-panels Substation (transformer) (20kV / 0,4kV) DSL stove WaMa IM E-car CB CHP EEX or other markets BGM Power provider (utility) 23 Integrated Hybrid Grids power gas heat Power storage Power transmission grid Power storage power Wind management Power distribution grid Power plants Wind Power Plant Steam Power Plant Gas &Steam PP H2 / methanation H2 / methanation Gas transmission grid Gas distribution grid gas management biogas Integrated energy management Gas buffer Bio-PP Gas storage Heat distribution grid heat management PV E-Mobility H2- Mobility G T G T CHP boiler Bio-CHP 24 Heat pump Thermal storage

13 Integrated Energy Management Systems Balancing demand and supply within each grid Energy conversion in between gas, power, and heat real conversion of power to gas e.g. by electrolytic methods (H 2 ) and methanation in order to consume overflow of power supply from wind power plants virtual conversion of power to gas in bivalent systems e.g. by switching between gas boiler and electric boiler Interoperability of energy management systems for power, gas, and thermal grids ( standardized interfaces? project HeGrid) Integrated energy information grid with distributed system intelligence in order to increase the efficiency, flexibility, and stability of the combined grids. 25 Tomorrow s Energy Management Challenges Discover and exploit degrees of freedom and leeway for demand (and supply) management. Need for autonomic/organic energy management without reducing personal comfort or industrial productivity Develop new ways of storing (electric) energy Batteries Power to gas to power Virtual storage Strong need for intelligent demand and supply management to increase the reliability of power supply in spite of fluctuating, decentralized and uncontrollable generation of power from renewable sources. Strong need for load flexibility and load shifting 26

14 Requirements on information about the grid Information on the current state of relevant components Energy consumers Energy producers Storage ( Prosumer ) Grid components (Substations, cables, ) Which information? Voltage (+ current? + frequency?) Power Phase shifts (cos ) Current degrees of freedom wrt Amount of consumption / production Time period for consumption / production For whom? Next level, needing this information for decisions on scheduling components or, more general, on managing the energy system. 27 Requirements on communication Exchange information on the current state with those who need this information for performing their duties: Next level energy manager (distribution system operator, energy supplier (trader), demand side manager, ) Partner in the energy system who is involved in cooperation. Communicate derived data: demand/supply forecast values, price and control signals Day-ahead, intra-day, price signals Control signals for the anticipated use of components in the energy system 28

15 Typical questions Which and how much information is needed? On each consumer/producer or aggregated over many? On which time scale? Real-time? Every second, minute, x minutes, hours? How to communicate? Using the power grid (Powerline? Digital current?) Using a data grid (fibre? DSL? phone? Public vs. private data grid? ) Wireless (WLAN, GSM, GPRS, UMTS, LTE, zigbee, )? Where is the information processed? Decentralized (within a house? an EV? a grid segment?) Centralized (in a balancing group? at the power supplier? At the distribution system operator?) Which system architecture is needed? Central, decentral, hierarchical, Tree-like, mesh-like topology? Parallel, multi-core, distributed, cloud, real-time? 29 Typical questions Which are the most appropriate concepts for the control of demand and supply of electrical (and thermal) energy? Market mechanisms Exact or approximate planning and optimisation Strictly local versus strictly global versus hierarchical, Trade-offs What are the objectives? Balancing demand and supply Generation of system services (reactive power, balancing energy,..) Reduction of energy consumption, efficiency vs. flexibility What about data protection and privacy?? Anonymisation / Pseudonymisation Traceability Which problems arise wrt security and safety? Access protection, detecting and preventing attacks Dependability / manipulation of data, Robustness issues (wrt missing data, erroneous data, ) Stability 30

16 Some essential tasks for ICT for energy (or for Energy Informatics ) Communication of essential information between the relevant entities in the grid Efficient and dependable information processing for various tasks Prediction of power demand and supply Modeling and simulation of the energy sytem Planning and optimisation of grid operations Support of market operations (trading agents, ) Demand and supply management Control of batteries of various types of EVs (charging, power-feedback) Virtualisation of components (virtual power plants, virtual storage, ) Support of system services (reactive power, demand response, ) Support of emergency actions (failures, power outages, ) Authentication, roaming, accounting, billing, Various services for integration of electric mobility into the grid Various issues wrt security and safety 31 Necessary competences in Informatics Algorithms and data structures, optimisation methods Information processing in distributed and parallel systems Communication protocols Software architectures Service-oriented computing (Cloud etc.) Concepts for self-organising systems (autonomic/organic computing, multi agent systems, architectures and methods) Validation / verification of software + hardware Security concepts, cryptographic protocols, access protection Dependability, fault tolerance, trustworthiness, robustness, stability 32

17 The evolution of the grid: the Internet of Energy The electricity grid: from a rather unflexible consumption-oriented supply chain with passive consumers Generation Transfer Consumption to a new dynamic interaction architecture with active prosumers Generation Transfer Consumption Storage Needs the real-time crosslinking of all grid components and participants combined with an intelligent self-organisation! 33 What do we mean by the Internet of Energy? There is a global addressing scheme (IP-based) for uniquely addressing all the components of the energy grid. There is a common protocol (TCP/IP and related) for communicating information between the components of the energy grid There are standard services for operating the energy grid. Generation Transfer Consumption Storage 34

18 Remarks The challenges of the upcoming and inevitable transformation of the energy system needs a transdisciplinary approach Informatics ( new discipline Energy Informatics ) Power Engineering Control Engineering Communication Engineering Law Without significant contributions from Informatics (or ICT ) this transformation will not be feasible. E-Energy animation: see for a nice and comprehensive animation of the German view of a future smart grid Animation by acatech (in German): 35 Electric Mobility and Smart Homes Institute for Applied Informatics and Formal Description Methods (AIFB) Institute for Applied Computer Science (IAI) FZI Research Center for Information Technology KIT University of the State of Baden-Württemberg and National Research Center of the Helmholtz Association

19 German National Development Plan for Electric Mobility Phase 2 Market development Phase 3 Volume market Goal for 2020: 1 Mio. E-Vs in DE DE is lead market for E-Mobility 2030: 6 Mio EVs 30 % renewable energy Phase 1 Market-/ Technologiepreparation Development of battery technology and competence centers in DE Provisioning of an interoperable and large-scale charging infrastructure Series production of Battery electric vehicles (BEV) and Plug-In electric vehicles (PHEV) Development of business models 37 Relevant Properties of Electric Vehicles Germany, 2008 (mobility survey): Average daily car usage < 1 h, 94% of trips < 50 km Average net capacity of currently available EVs: 20 kwh At 1 Million BEVs (German objective for 2020): available storage capacity of ~ 20 GWh At charging/discharging power of 3.7 kw: ~ 3.7 GW potential power Consequently: high demand for power, potentially also high supply (if power feedback is possible) Average time for charging: Single phase 3.7 kw: 5 to 7 hours. Three phase 10 kw: ~ 2 hours (but high risk of grid overload!) Potential of high flexibility for load shifting, but also potential of high peak load! Using intelligent control leads to high potential for stabilizing the grid. 38

20 Integration Strategies: Load Balancing Potential Power P in kw Power P in kw :00 original grid load curve Solar power 06:00 12:00 Time 18:00 EV <-> Grid Exchange Charging/Infeed Uncontrolled EV energy charging :00 Power P in kw :00 Controlled EV charging 06:00 12:00 Time Flexibility of battery charging could allow to generate a range of different demand profiles 3 4 Power P in kw resulting load curve 18:00 00: :00 06:00 12:00 Time 18:00 00: :00 06:00 12:00 Time 18:00 00:00 39 Energy Smart Home Living Lab at KIT Energy Mgmt. System 40

21 KIT Energy Smart Home Lab Energy Management Panel (EMP) Visualisation of energy usage EMP Intelligent appliances EMP CHP Discover user preferences EMP Observes and controls electric/thermal consumers & providers EMS A/C EMP Charging station Smart Meter Solar inverter 41 KIT- Smart Home Scenario Decentralized power plants Smart meter Energy provider Personal charging station Inhouse touchscreens data Personal computer / smart phones Car driver Intelligent and classic household appliances Sensors 42

22 KIT- Smart Home Scenario Intelligent appliances Communicate with central control and with each other. Know (and communicate) their current state. May respond to control. Electric car Connected to the home as a mobile storage Bidirectional utilization (charging/discharging) Large consumer/supplier Decentralized power generation (PV/CHP) PCM elements in ceiling (cooling) (Phase-change material) Simulation component ( 4-Quadrant amplifier ) Home Human Device Reduced but effective interaction between human, home management, and devices Discover and exploit degrees of freedom for energy control 43 MeRegioMobil - Electric Vehicles Opel Meriva MeRegioMobil (3) A-Class E-CELL (Daimler) bidirectional power connection ~ 40 Smart ed uni-directional power connection 44

23 Current KIT Research Demonstrates how Electric Cars Could Support the Electricity System. research on vehicle to home and vehicle to grid Mercedes-Benz E-CELL incl. feed-in function KIT Smart Home Charge protocol: ISO15118 enhanced for feed-in control 45 Charge Stations (by EnBW and KIT) Opel Meriva BEV incl. feed-in function Stakeholders of ISO15118 are Customers, OEMs and the Unbundled Energy Utilities. Energy Trader Car manufacturers (OEMs) Private Customers, Energy suppliers possibly with Balance responsible parties decentralized generation Generation Energy Storage feed-in consumption (Transmission) System operator (TSO) Distribution system operator (DSO) Business customers Public charging infrastructure Picture source Ladestation: EnBW 46

24 An Important Success Factor is the Availability of Supported, Practical, International Standards. ISO describes how communication of battery electric cars and charge stations should work Current participation in ISO joint working group OEMs Energy utilities Automotive suppliers Universities/Research Source: Tim Schlüsener (Daimler AG), Impulsvortrag SmartCharging, Lebenswelt Elektromobilität, Mannheim Car Manufacturers are Focused on Battery Protection and Rentable Value Added Services. Battery protection No external direct control SoC and SoH information not accessible to third parties Charge session incl. battery care User experience Support maximum range Value added services Preconditioning of the vehicle cabin is a comfort and range extending feature Chevrolet OnStar/MyVolt MyFord, Nissan Leaf smart/mercedes E-CELL Routing services Energy supplier, All-inclusive packages Smart Phone Apps with preconditioning control Sources: (left pic.): Chevrolet OnStar Android App (right pic): Hawk Asgeirsson (DTE Energy Co.), Plug-in Electric Vehicle Activities and AMI, EPRI IWC Infrastructure Working Council. Tempe, AZ, USA

25 Electric Cars Will Support Optional Departure Time Settings by the Driver (in car, by phone, www). Customer portal MY VOLT of GM / Chevrolet (since 2010) Immediate Charge starts upon plug-in. Delayed Departure Time Charge your vehicle based on departure time. Delayed Rate and Departure Time Charge your vehicle based on departure time and utility rates. Pic.Source: 49 External Control Signals Will Be Sent by a DSM Common term Demand Side Manager Focusing on demand Prosumers not considered (PV, CHP) Enhanced DSM, also managing decentralized generation (Decentralized) Demand and Supply Manager: DDSM Energy Supplier DDSM Function Charge station operator Business Energy Management Home Energy Management Public charge spots Business and semipublic charge spots Private charge spots 50

26 Charge Sessions Consist of Start, Charging, End and Optional Feed-In. Power P Optional feed-in Power consumption Departure and start trip Time t Plug in vehicle End of charge 51 V2G-Enabled Vehicles Could Provide Negative and Positive Balance to the DDSM. Power P Negative balance energy Positive balance energy Time t Which parameters are necessary to fit the requirements of all stakeholders? 52

27 Charge Station and Grid Set the Technical Constraints Technical maximum possible power Load balancing of stations / overload protection Power P Maximum power charge station 1 Data flow Overload Maximum power V2G charge station Time t 53 The Vehicle Knows a Set of Technical and User Requirements Affecting the Charge Session. Vehicle driver defines a minimum energy amount (e.g. to reach next hospital) Minimum power EV needs at least energy to run the charge system Pre-conditioning during the charge session (save energy, extend range) Departure time is known, Charge 100% SoC Power P Filling minimum energy amount Maximum power charger Minimum power of charge system Cabin Preconditioning Departure Time Time t Maximum feed-in power inverter 54

28 The Charge Schedule Is Used to Abstract all EV-Internal Boundary Conditions. Power P 2 Data flow EV starts charging. Maximum power Preferred power EV Energy amount Minimum power Time t Maximum power V2G Energy amount V2G 55 The DDSM Sums up all EVs and Computes Necessary Load Shifting Actions. Power P Purchaised power resp. scheduled power excess demand Short-time load prediction supply surplus Time t 56

29 Necessary Load Shifting Will Be Realized by EVs resp. by all Controllable Devices. Power P Maximum power Preferred power EV Preferred power DDSM Minimum power Time t 57 The DDSM Sends a Particular Preferred Load to Every Selected EV (incl. Feed-in, If Necessary). Power P 3 Data flow Preferred power DDSM Time t 58

30 A New Charge Schedule Will Be Sent by the EV Which Depends on the Preferred Power by DDSM. Power P 4 Data flow Maximum power New, preferred power Energy amount Minimum power Time t Maximum power V2G Energy amount V2G 59 Resulting Optimized Battery Charging for EVs Constrained by power grid, charging station, EV, and user (driver) towards a standardized protocol ISO power [kw] EVSEMaxPower EVMaxPower Grey area = tech. flexibility PMax EVMinPower Shifitng charging / discharging times 60 t 0 EVSEMinPowe r Time [h] t eocevminpowerdischarge ScheduledPower PPreferred PMaxDischarge EVMaxPowerDischarge EVSEMaxPowerDischarge

31 Remarks on EV Charging Batteries of electric vehicles will provide significant load flexibility. The load management of BEVs requires a sophisticated protocol for negotiating the load profile wrt the requirements of all stakeholders. Several other approaches to load management have been suggested, based on specific scenarios At private charging spots: supporting local V2G applications At semi-public charging spots: supporting demand response applications For fleet management: probably the first major users of BEVs will be in commercial traffic. Inductive charging will provide interesting alternative to cable-based charging. 61 KIT- Smart Home Scenario Intelligent appliances Communicate with central control and with each other. Know (and communicate) their current state. May respond to control. Electric car Connected to the home as a mobile storage Bidirectional utilization (charging/discharging) Large consumer/supplier Decentralized power generation (PV/CHP) PCM elements in ceiling (cooling) (Phase-change material) Simulation component ( 4-Quadrant amplifier ) Home Human Device Reduced but effective interaction between human, home management, and devices Discover and exploit degrees of freedom for energy control 62

32 Hardware in the Loop Simulation sensor data grid simulation (power factory) 4Q-controller 63 Focus of our Approach Demand-side load management Appliances are re-scheduled based on external 24h-signal Management of the load- and feed-back-process of the electrical vehicle 64

33 Classification by Degree of Freedom Degree of Freedom Capability of re-scheduling the appliances work-item Classification Appliances with a poor degree of freedom Re-scheduling not possible Re-schedulable appliances Bidirectionally on the timeline (e.g. deep freezer, electric heating) Backward on the timeline (e.g. hot-water boiler) Forward on the timeline (e.g. washing machine and dish washer) 65 Classification of Household Appliances appliance deepfreezer el. heating/ air cond. warm water boiler dishwasher washing machine dryer stove heating plate small appliances multimedia lighting small appliances permanent service timed service predictable un-predictable controllable observable 66

34 24-Hour Signal rating :00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Source (pic): EnBW MeRegio Time dependent price signal Communicated by the energy provider (e.g. multicast) 67 Typical Schedule deep freezer breadmaker toaster stove dishwasher stove toaster washing machine dishwasher 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 A typical schedule is predicted by the Energy Management System based on a large amount of measurement data 68

35 Intelligent Demand Management Original schedule Optimized schedule 4500 Stove usage 4000 cannot 3500 stove be moved stove washing machine 3000 breadmaker dishwasher 2500 toaster dishwasher 2000 deep toaster 1500 freezer Bread machine 0 starts earlier :00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Dishwasher 4000 postponed Washing machine 3500 and dish washer 3000 postponed Battery may1000 be recharged at 500 night 0 therefore: power 00:00 02:00 04:00 06:00 08:00 10:00 12:00 infeed 14:00from 16:00car 18:00 20:00 22:00 battery! 69 Discovering Degrees of Freedom: Energy Management Panel Transparent information on energy consumption Discover and specify degrees of freedom for energy consumption 70

36 Energy Management Panel Energy Consumption and PV-System View 71 Energy Management Panel Device Dashboard Kitchen 72

37 Energy Management Panel Example Tumble Dryer Kitchen Programmed state Degree of freedom Forced start Automatic start: 18:00h 73 Energy Management Panel Energy Supply and High Power Consumers Phase change material in the ceiling Source (PCM pic): 22-insulated-shipping/ 74

38 Energy Management Panel Energy Flows and Electric Vehicle 75 Energy Management System EMS Observe + Predict Learn + Control Observe + Control Communicate 76

39 Origin of Organic Computing Workshops of the GI-/ITG-Sections on Computer Engineering in 2002 Information technology is moving towards the ubiquitous networked computer. Complex ubiquitous systems need new concepts for organization and user interfaces to remain manageable and controllable. Future computer systems have to be designed with respect to human needs. Future computer systems have to be robust, adaptive, and flexible. Future computer systems have to be self-organized but trustworthy. Systems having these properties show life-like behavior Hence, we call them Organic Computer Systems. Based on range of other initiatives: ubiquitous, autonomic, 77 Meta-data? Model-predictive? Controlled-self-organized? Model-based? Organic Computing You-name-it Feedback control? Self-managing? It is not the question, whether adaptive and self-organising systems will emerge, but how they will be designed and controlled. Policies? Autonomic? Feedforward control? Self-adaptive? Self-organized? Emergent control? 78

40 German Framework for Research on OC : Christian Müller-Schloer, Hartmut Schmeck, Theo Ungerer DFG priority program 1183 Organic Computing ( ) 3 phases of two years each, around 18 projects, ~2 Mio per year Nature (1) OC-Principles (2) OC-Technology (3) Technical Applications Emergence Self-x OC-toolbox Dissip. Struct. Ants Swarms Brain. Autonomy Self-x. Emergence Awareness Cooperation Competition Basic technologies Observer/Controller Guarding Helper Threads Embedded Learning Complete system arch. Car Office Telecom Factory Home Health&Care Energy (Generic) Concepts for Control of Self-organising Systems IBM s MAPE cycle for autonomic computing Monitor Plan Analyze Execute Knowledge (called autonomic element ) System under observation and control (SuOC) A set of interacting elements/agents. Does not depend on the existence of observer/controller. Distributed and/or central observer/controller-architecture Driven by external goals Multilevel organization sensors input K productive system selects observation model observation model observer observes reports organic system SuOC controller controls actuators system status goals output 80

41 Generic O/C-Architecture J.Branke, M.Mnif, C. Müller-Schloer, U. Richter, H. Schmeck 2006 model selection model of observation select select select select emergence detector 1 pre-processor system data emergence detector 2 filter... analyze monitor time-space-pattern data analyzer individual data monitor aggregator statistics... cluster prediction log file predictor predict store observer situation parameters simulation model mapping map F i C i A i adapt direct goal / objectives system status adaptation module situation action t evaluation parameters evaluate history history fitness action selector controller raw data action A i system under observation and control (SuOC) 81 Realisation of OC systems 1. Central: One observer/controller for the whole system. 2. Distributed: An observer/controller on each system component. 3. Multi-level: An observer/controller on each system element as well as one for the whole system. adaptive top-down self-organising bottom-up emergent control controlled selforganising bottom-up / top-down observer controller O C SuOC O C SuOC O C observer controller O C SuOC SuOC O C SuOC O C SuOC O C SuOC SuOC O C SuOC O C SuOC O C SuOC O C SuOC O C SuOC SuOC O C SuOC O C SuOC O C SuOC 82

42 observation model select select select select Different view on O/C-architecture model selection emergence detector 1 emergence detector 2... pre-processor data analyzer time-space-pattern monitor or system dataindividual data raw data aggregator statistics... predictor cluster prediction log file observer situation parameters simulation mapping action A i system under observation and control (SuOC) Off-line learning loop On-line learning loop C i F i A i action selector objectives adaptation action situation t evaluationation parameters history history fitness user controller Action generation Layer observer 2 2 Layer 2 situation, quality of control Action selection and evaluation Layer observer 1 1 detector information rule generator EA simulator new rule Controller 1 Layer 1 LCS controller 1 rule set parameters traffic controller light Internal system controller control Layer detectors 0 Layer 0 detectors system component traffic component output 83 Organic Computing has been applied to broad range of technical applications, in particular to Traffic control (self-adaptive traffic lights, self-organizing progressive signal systems, routing recommendations, ) Production systems (adaptive production lines, fault tolerance, reconfiguration) Smart camera systems (supervisory applications) Network control Systems on chip Energy management and control 84

43 Architecture of Energy Management System Combination with EEBus (Kellendonk) 85 EEBus and O/C-Architecture 86

44 Improving self-supply by load management Ratio of self-supply with power from PV and -CHP (typical profile of a 5-person household, without stationary batteries) optimized Non-optimized weeks of a year Source: F. Allerding 87 Improving self-consumption by load management Ratio of self-consumption of power from PV and -CHP (typical profile of a 5-person household, without stationary batteries) optimized optimiert nicht Non-optimized optimiert weeks of a year Source: F. Allerding 88

45 Distributed Optimisation (F. Allerding) 18:00 Goals and external signals Start: 16:00 Subproblem WaMa partial solution WaMa 17:00 2h Subproblem DishWa Global optimisation partial solution DishWa Start: 15:30 Start: 15:30 & 17:00 Subproblem CHP partial solution CHP 89 Distributed optimisation: process view TP TPSP TP TPSP Opt Evolutionary Algorithm A SP SP 90

46 Evolutionary Algorithm Selection Generate Iintial population Evaluation Recombination Stopping criteria? Choose Best solution Mutation Coded value subproblem a Coded value subproblem b Individual 91 Constructing an abstract subproblem Bit vector of length k Evaluation function - Input: k bit - Output: resulting load profile Stores the current solution vector (e.g. representing a strarting time oa a device) Determines the resulting load profile of the household component based on the current value of the bit vector. Back transformation - Input: k bit - Output: concrete solution (e.g. starting time) Generates a concrete instruction for the household component from the current value of the bit vector. (e.g. the starting time) 92

47 Subproblem in more detail power rj : Release Time of component j pj: duration of task of component j dj: Deadline for component j t*: Shift of starting time tstart time rj tstart dj-pj dj Bit vector of length k Degree of freedom tdof Dividing tdof into timeslots and mapping to k bit => Shifting the start by t* Example: Evaluation function Back transformation function => rj + t* = tstart Subproblem for component j 93 rj = 14:00h Value of bit vector is number of minutes k = 6 bit (= degree of freedom max. 63 min) bit vector = ( t*= 19) => starting time for component j : 14:19h Generating the individuals Subproblem WaMa 10 bit Subproblem DishWa 8 bit Subproblem CHP 43 bit 10 bit 8 bit 43 bit Individual 94

48 Optimisation in more detail Initial/current population Subproblem WaMa Recombination/ Mutation Selection Evaluati on Subproblem DishWa Subproblem CHP 95 Optimisation in more detail P Subproblem WaMa Subproblem DishWa Subproblem CHP P P t t t P t Fitness function Evaluation of the individual 96

49 Stopping the optimisation Initial/current population Best individual of last generation Recombination/ Mutation Evaluation Selection Stopping criteria 97 Back transformation Subproblem WaMa Subproblem DishWa Subproblem CHP Decode solution Decode solution Decode solution at local O/C-unit Real starting times 98

50 FZI House of Living Labs (HoLL) Space > m² for lab-, office- and presentation purposes Extension and integration of existing FZI Living Labs AAL - Ambient Assisted Living Automotive MobileIT / SatNav Service Robotics New FZI Living Labs smartenergy smarthome smartautomation smartmobility Innovations at the interfaces of the research topics Implementation and Evaluation of real-life -scenarios 99 FZI House of Living Labs (HoLL) LL smart Automation LL smart Energy LL smart Home LL smart Mobility Presentation Office 100

51 HoLL - smartenergy Utility, DSO Grid s state User Energy Management (EM) LL smarthome LL smart Mobility LL smart Automation LL Service- Robotic LL smart Energy 101 smartenergy Infrastructure Decentralized energy supply Photovoltaic (15 kw p ) Stationary electrical storage (30 kwh) CHP-unit (5,5 kw el, 12 kw th ) Adsorption chiller (9 kw th ) Condensing boiler (100 kw th ) Mobile energy storage via bidirectional connected EV Thermal energy storage (decoupling of supply and demand related to heating and cooling) Intelligent consumption Load shifting of electrical and thermal appliances in the smarthome Respecting user constraints Sensors and actuators for automatic EM of electrical and thermal consumers in the office (building systems for heating, cooling and electrical consumers) smartautomation (monitoring and load shifting of diverse processes) smartmobility Integrated Energy Management System across the whole HoLL by flexible combination of energy supply, consumption, and storages Development of homogenous interfaces for the system integration 102

52 Communication electric utility grid operator user EMS (for office and residential buildings) PLC EIB/KNX Zigbee HabiTEQ REST (XML, HTTP) flexible power consumption system services A/C smart plugs smart meter electric and thermal energy supply smart and conventional household appliances electrical and thermal energy storages 103 Controlled self-organizing energy management Shift imbalances to a later point Handle device restrictions decentralised elite group Organic / Autonomic energy management Significant reduction of the need for balancing energy Primary control by Elite Group Secondary control by pool power internet time consumption production 104

53 Summary ICT 4 EV & Smart Home Power generation from renewable sources needs ICT for new approaches to energy management. Electric vehicles will generate significant capacity for power storage leading to additional demand and supply of power. Integration of EVs into smart home environments allows for intelligent balancing of power demand and supply and for new power system services. An Internet of Energy will have to cope with similar safety and security problems as the Internet of Data. 105 Summary on Energy Informatics Tomorrow s energy system crucially depends on appropriate information at the right time, at the right place, with appropriate content. Future energy management and grid control systems need an adequate integration of control and information processing. An appropriately designed energy information and control grid will be one of the backbones of future integrated energy systems Stability and security of are major challenges for future highly decentralised energy grids Those who are experts in both - information and communication technology and - control and power engineering will have a brilliant future!! 106

54 Contact Address Prof.Dr. Hartmut Schmeck KIT Campus South Institute AIFB Karlsruhe Germany Phone: Fax: