Modeling and control of electrochemical batteries

Transcription

1 Toward autonomous photovoltaic building energy management : Modeling and control of electrochemical batteries Dr. Hoang-Anh Dang HaUI Institute of Technology This research is under supervision of Prof. Benoit Delinchant and Prof. Frederic Wurtz in Grenoble Electrical Engineering Laboratory

2 Context Smart Building has to use less energy and can be optimally controlled by occupant Two main strategies of energy management Reduce energy consumption and develop renewable sources Optimize power supply that depends on production, distribution and storage Importance of electrical storage management High energy consumption Their battery could be used for energy management 23/6/215 Hoang-Anh DANG 2

3 PREDIS Smart Building «Monitoring et Habitat Intelligent» (MHI) Local GTC Shed Local CTA Computer room Office room Lobby PREDIS MHI EP RECH Induction motor & variable speed drive Temperature sensor Thematic researches : Metering and characterization Management and design User behavior Wattmeter and controllable switch My research direction in G2Elab 23/6/215 Hoang-Anh DANG 3

4 Summary Context PREDIS MHI Electrical management in PREDIS MHI Battery modelling PREDIS MHI case study application Conclusions and Perspectives 23/6/215 Hoang-Anh DANG 4

5 Electrical management in PREDIS MHI General objectives Autonomous renewable resources Control the energy profile consumption by using electrical storage Case study of PREDIS Computer room Laptop 15 Inverter DC/AC Solar panels Laptop 2 Disposition of laptops G2Elab electrical grid Laptop 1 Power system Electrical distribution system of computer room 23/6/215 Hoang-Anh DANG 5

6 Necessary researches Capitalize models for system management Prevision of photovoltaic production (estimated from the weather forecast) Prevision of PC load (estimated from timetables) Prevision of electric prices (given by electricity distributors) Prevision of storage capacity (calculated from the battery model?) Model electrochemical batteries Develop the storage management algorithm 23/6/215 Hoang-Anh DANG 6

7 Summary Context PREDIS MHI Electrical management in PREDIS MHI Battery modelling PREDIS MHI case study application Conclusions and Perspectives 23/6/215 Hoang-Anh DANG 7

8 Battery modelling problematic should be replaced by V OC R V I B V B Charge capacity of a classic model Charger float voltage of a classic model Charge current of a classic model 23/6/215 Hoang-Anh DANG 8

9 Electrochemical battery modelling Functional specification P sp Parameters, Initial States BATTERY P b SOC SOH Joule losses P discharge_available P charge_available Shepherd s hypothesis discharge mode (I B > ) 23/6/215 Hoang-Anh DANG Polarization zone P sp : Power set point P b : Output battery power SOC : State Of Charge SOH : State Of Health P discharge_available : Available discharge power P charge_available : Available charge power Nominal zone V B V R V B, I B : Voltage and current R I : Internal resistance K, A, B : Voltage factors Q max : Maximal capacity I I Polarization zone B Q K Q Q Q max I B t c A e I B Exponential zone B( Q Q max Q Q max Charge mode (I B < ) dt : Instantaneous charge ) max Q 9

10 Shepherd s model : calculation from the power set point P sp Calculate SOC and SOH P B V B I B I 2 B I B P sp Q SOC 1 % and Q max Resolve SOH SOH Determinate the battery current I B I dt B 2 NC Qmax_ initial 1% N C : Cycle durability Functional constraints A case study of battery simulation (Ni-Cd, 1,2 V, 42 mah, SOC =5%) Puissance (% Pnom) P discharge_available Puissance calculée Puissance de consigne P charge_available Etat de charge (%) SOC calculé Temps (s) x /6/215 Hoang-Anh DANG SOC min = 2% Temps (s) x 1 4 1

11 Model parameters Identified from measurement or/and in catalogue datasheet Estimated from existed tools or/and experimental publications Lead acid Li ion Ni Cd Ni Mh Full charge voltage (V full ) 1,888V nom 1,164V nom 1,1442V nom 1,178V nom Discharge current (I nom ),2Q rat,43478q rat,2q rat,2q rat Internal resistance (R I ) Maximal capacity (Q max ) Extracted capacity at nominal voltage (Q nom ) Vnom Q 1 nom 1,5Q nom,3128q rat,9435q rat,96136q rat,96154q rat Exponential voltage (V exp ) 1,181V nom 1,84V nom 1,671V nom 1,847V nom Extracted capacity at the end of exponential zone (Q exp ),3333Q rat,4913q rat,27955q rat,2q rat Table of typical battery operation points (Battery model/matlab Simulink) (Tremblay & Dessaint, 29) 23/6/215 Hoang-Anh DANG 11

12 Validation Validation in charge mode (DELL PRECISION, Li-ion, 11,1 V, 85 Wh, SOC =5%) P sp = 19,5V 6,7A = 13,65 W 8 7 Puissance mesurée Puissance calculée 6 Puissance de charge (W) Estimated end of charge time Real end of charge time Temps (s) Error = 7% 23/6/215 Hoang-Anh DANG 12

13 Summary Context PREDIS MHI Electrical management in PREDIS MHI Battery modelling PREDIS MHI case study application Conclusions and Perspectives 23/6/215 Hoang-Anh DANG 13

14 Electrical management system 6 Solar power forecast Photovoltaic power (W) Solar panels Time (h) Inverter DC/AC Wattmeter 15 Wattmeter 2 Laptop 15 RF Zigbee protocol USB Xbee communication module Commander server TCP/IP Wattmeter 1 Laptop 2 RF receiver and transmitter RF X1 protocol RF transmitter Power system Electrical price ( /Kwh) Electricity prices forecast Laptop 1 Total power consumption forecast Total power consumption (W) /6/215 Hoang-Anh DANG Time (h) Time (h) 14

15 Electrical management strategy : predictive et real-time control Predictive control : batteries are pre-charged during chosen time intervals Objective : buy the electricity at lowest price moment The charge plan is generated, from charge starting time and charge duration (are calculated by using the battery model) Real-time control : batteries are charged/discharged by ON/OFF controllable switches Objective : maximize de la solar power production and ensure the uninterrupted power consumption Case of SOC SOC min : charge mode is mandatory Case SOC > SOC min : charge mode is sorted Predictive control phase Puissance PV Consommation totale Real-time control phase 7% 4% 15% Reactive point 23/6/215 Hoang-Anh DANG

16 Simulation : Inputs Puissance totale (W) Temps (h) Total power consumption.1 Puissance photovoltaïque (W) Temps (h) Solar power Prix d'électricité ( /KWh) Temps (h) 23/6/215 Hoang-Anh DANG Electricity prices 16

17 Simulation : Results Etat de charge (%) COM Temps (h) Battery states of charge et control states of controllable switches Monthly cost ( ) Ratio Classic case study 6,4 1,67 PREDIS case study,6 1 Electricity cost comparison 23/6/215 Hoang-Anh DANG 17 PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC1 PC11 PC12 PC13 PC14 PC15 Puissance (W) Puissance (W) Puissance de consommation totale Puissance photovoltaïque Temps (h) Comparison between the total power consumption (at electrical outlet) and the solar power production Temps (h) Exchanged power with the power system Prix d'électricité ( /KWh)

18 Real system application : Inputs Puissance (W) Prediction for 22/5/ Temps (minutes) Power consumption forecast 1 Puissance (W) Temps (minutes) Solar power forecast Prix d'électricité ( /MWh) Prix moyen pondéré à la baisse Prix moyen pondéré à la hausse Temps (heures) 23/6/215 Hoang-Anh DANG 18 Electricity prices forecast

19 Real system application: Results SOC (%) Commutation Measurement in 22/5/ PC1 PC2 PC3 PC4 PC5 PC6 PC7 PC8 PC9 PC Temps (minutes) Temps (minutes) Battery states of charge et control states of controllable switches Monthly cost ( ) Ratio Classic case study 5,8 11,6 PREDIS case study,5 1 Electricity cost comparison Temps (minutes) 23/6/215 Hoang-Anh DANG 19 Puissance (W) Puissance (W) Temps (minutes) Real total power consomption Puissance consommée Puissance photovoltaïque Prix réel d'électricité Comparison between the total power consumption (at electrical outlet) and the solar power production Prix d'électricité ( /MWh)

20 Summary Context PREDIS MHI Electrical management in PREDIS MHI Battery modelling PREDIS MHI case study application Conclusions and Perspectives 23/6/215 Hoang-Anh DANG 2

21 Conclusions The battery model developed is simple enough to be implemented only from the typical characteristics of the battery and its nominal variables, It remains sufficiently realistic regarding the evaluation of battery power, which depends on the state of charge, This model has been validated and used to test different strategies for energy management in PREDIS platform in order to maximize the photovoltaic autonomous. 23/6/215 Hoang-Anh DANG 21

22 Perspectives Portable electrical devices management V2H management Application in Vietnam situation? 23/6/215 Hoang-Anh DANG 22

23 Project: Micro Smart Grid Development and Application for Building Energy Management PV panels 15 kwp Load bank Weather station Energy manager AC Grid ALR Modbus Distribution cabinet PLCs Energy controller Transducer elevator Lab HVAC Classroom Battery energy 23/6/215 Hoang-Anh storage DANG station Real Loads USTH building 23

24 Next seminars? PREDIS Building Simulation Energy consumption in one year Température ambiante de la salle informatique 3 Modèle global COMFIE Pléiades Température ( C) PREDIS thermal management Temps (mois) Température ambiante de l'espace bureaux Température ( C) Modèle global COMFIE Pléiades CO 2 regulation temperature regulation Temps (mois) 23/6/215 Hoang-Anh DANG 24

25 Thank you for your attention 23/6/215 Hoang-Anh DANG 25