Modelling and (co-)simulation of power systems, controls ...

Modelling and (Co-)simulation of power systems, controls and components for analysing complex energy systems Workshop on Software Tools for Power System Modelling and Analysis University College Dublin – 11th October 2013 Stifter Matthias – AIT, Energy Department, Complex Energy Systems

About Co-Simulation

08.10.2013

3

Co-Simulation Motivation 

Multi-domain problems not covered in one tool  Electrical power system  Communication system  Controls / SCADA

    

Different simulation time steps Different concepts: continuous time and discrete event time systems Use existing highly complex and detailed models Integration of real components Scalability and flexibility

 Combine different dedicated simulation tools and domains  Analyse different subsystems and their interactions

08.10.2013

4

Co-Simulation Overview 

Various types of modelling and simulation: number of modelers

closed simulation

distributed simulation

>1

re-unification of the separately modeled subsystems equations

co-simulation

distributed modelling

=1

„classical“ Simulation

model separation for simulation

closed modelling

=1

>1

number of integrators

Source: M. Geimer, T. Krüger, P. Linsel, Co-Simulation, gekoppelte Simulation oder Simulatorkopplung? Simulation 2006.

08.10.2013

5

Co-Simulation (1) 

Coupling on model level

Application 1

Application 2

Model level

Model definition, equations Model code export,

Model level

Integrator level

Model and integrator code export

Integrator level

Source: M. Geimer, T. Krüger, P. Linsel, Co-Simulation, gekoppelte Simulation oder Simulatorkopplung? Simulation 2006.

11.10.2013

6

Co-Simulation (2) 

Coupling on integrator level

Application 1

Application 2

Model level

Equation evaluation (e.g. function call)

Model level

Integrator level

Integrator evaluation (distributed simulation)

Integrator level

Source: M. Geimer, T. Krüger, P. Linsel, Co-Simulation, gekoppelte Simulation oder Simulatorkopplung? Simulation 2006.

11.10.2013

7

Simulation challenge: power system and controls System with controller to fulfilll certain functionalities: e.g., voltage control task consists typically of the following parts:    

Power system model Component models Controller / SCADA

SCADA system

TCP, UPD, IEC 61850, IEC 60870-5-104

Power Grid Simulation Controller transient/ steady state

Power System Analysis

Com Interface

Com Interface

Com Interface

-

TCP, OPC

TCP, OPC

Component Simulation -

10.10.2013

8

Co-Simulation approach In this simulation coupling, the individual simulators run in real time and parallel, exchanging results and set-points.  PowerFactory  Matlab / SimPowerSystem  PowerFactory  4DIAC (distributed control system)  4DIAC  ScadaBR Grid Component Simulation (e.g., Distributed Energy Resources)

Supervisory Control and Monitoring Simulation

Component Simulation

SCADA Emulation

ASN.1 over TCP/IP

Electricity Grid Simulation

10.10.2013

Power System Analysis

ASN.1 over TCP/IP

ASN.1 over TCP/IP

Controller Emulation

(Embedded) Control System Development and Simulation

9

Simulation challenge: dynamic EV charging Need for simulation of :  impact of the electric vehicle energy demand  energy depends on trip length, temperature, etc.  test charging management strategies Continuous  charging power is not static System

Hybrid System

 Hybrid system: discontinuity taking place at discrete events P

 Need for simulation of the EV trip to get the energy needed to recharge

P Charge

a)

t

b)

Recuperation

t

Discharge

 Combine the power system simulation with the simulation of the discharging 10.10.2013

10

Simulation challenge: dynamic EV charging household load profiles

taken from measurement campaign

small scale distribution grid

realistic battery model

medium/low voltage network with consumers

stochastic driving patterns

derived from real data charging control algorithm

distributed charging power regulation

20 May 2013

11

Simulation tools

Modelica

PSAT

 dedicated multi-domain (physics) simulation language  supports event handling  acausal simulation (continuous time-based simulation)

 Power system analysis toolbox  Matlab/Simulink and Octave  continuous time-based simulation

4Diac / Forte

GridLAB-D

 framework for distributed control systems  intended for event based controls in real time  open source IDE and Runtime

 multi-agent based power system simulation  includes various energy-related modules  discrete event-based simulation

Main challenge  coupling discrete event-based simulation with continuous time-based models. 08.10.2013

12

Co-Simulation Interfaces and Mechanisms

10.10.2013

13

Co-Simulation: synchronous, sequentially 

Co-Simulation run sequentially and blocks until the results are available



simulate same time step twice t0

t1

t2

PowerFactory

t3

Data exchange (OPC, TCP/IP)

Co-Simulation (parallel)

step ∆t 1

data exchange

2

simulation time

controller co-sim integration time

power flow send result send ready

Ubus,Pmeas notify IN

receive wait simulate

t0

t1

t2

t3

receive wait

t11

2

data exchange

simulation time

Pset notify OUT

send result send ready

power flow next step

continuous model co-sim integration time



simulate at next time step

10.10.2013

M. Stifter, R. Schwalbe, F. Andrén, and T. Strasser, “Steady-state co-simulation with PowerFactory,” in Modeling and Simulation of Cyber-Physical Energy Systems, Berkeley, California, 2013.

14

Co-Simulation: asynchronous, parallel 

Co-Simulation via external DLL PowerFactory stability (rms)

Co-Simulation (sequential) via external DLL (digexdyn.dll)

step ∆t

t0

t1

t2 t3' t3

input simulation time

∆tInt

event DSL integration time

simulate

DSL model (external) emit event

invoke

store state return

result

exec. event next step

M. Stifter, R. Schwalbe, F. Andrén, and T. Strasser, “Steady-state co-simulation with PowerFactory,” in Modeling and Simulation of Cyber-Physical Energy Systems, Berkeley, California, 2013. .

10.10.2013

15

Co-Simulation: real time synchronisation 



Dynamic behaviour  C-HIL / Regler  Filters Synchronisation with system time / scaled time base

t0 1

t1

t2

t3 real-time

∆tlag 2

∆tcommunication delay parallel co-simulation

10.10.2013

16

Functional Mock-Up Interface for Model Exchange 

FMI: A standarized API for describing models of DAE-based modeling environments (Modelica, Simulink, etc)



Functional Mock-Up Unit  model interface (shared library)  model description (XML file)



Executable according to C API  low-level approach  most fundamental functionalities only  tool/platform independent



Gaining popularity among tool vendors  CATIA, Simulink, OpenModelica, Dymola, JModelica, SimulationX, etc.

08.10.2013

17

Simulation environment for FMI for Model Exchange 

Simulation master algorithm not covered by FMI specification  tool independence



Requirements  initialization  numerical integration  event handling  orchestration



Well suited for simulation tools focusing on continuous time-based modelling



What about plug-ins for discrete event-driven simulation tools?

08.10.2013

18

The FMI++ library  High-level access to FMUs  model initialization, get/set variable by name, etc.  High-level FMU functionalities  integrators, advanced event handling, rollback mechanism, look-ahead predictions, etc.  Open-source C++ library  tested on Linux and Windows (MinGW/GCC and Visual Studio)  available at sourceforge.net  synchronous interaction between discrete event-based environment and a continuous equation-based component 08.10.2013

19

Example Application: Controlled Charging

08.10.2013

20

GridLAB-D as co-simulation master 





Discrete event-based simulator as co-simulation master  GridLAB-D’s core functionalities deployed as master algorithm EV model:  agent based behavior over time (individual driving patterns)  energy demand due to the trip  charging station handling

Traffic Pattern / Electric Vehicles (GridLAB-D) Charge Control

EV

EV

+

-

EV EV +

-

Pset Pcharge

Charging Station Charge Control

EV

+

-

Pset Pcharge

Charging Station

Interface  plugin validated tools and continuous models via standardized interface (FMI & FMI++, etc.)

08.10.2013

21

PSAT (Octave) 

Power system analysis  power flow to determine bus voltages



Network model:  medium voltage network with low voltage networks  load presenting household and charging station



Interface:  Connect to GridLAB-D via Octave API  + thin wrapper to access C arrays instead of data types

08.10.2013

22

4Diac / FORTE 

Distributed control system  IEC 61499 reference model for distributed automation



Model / control  Local voltage control (anxilliary service)  keep voltage limits



Interface:  TCP/IP socket communication (ASN.1 format)  suitable for embedded system environments

08.10.2013

23

OpenModelica 

multi-domain (physics) simulation language  supports event handling  acausal simulation (continuous time-based simulation)



Model  Industry proofed library for a detailed Li-Ion battery model  constant current / constant voltage charger



Interface  plug-in continuous time-based models via FMI++

08.10.2013

24

Co-Sim Power System, EVs, Components and Controls Open source based approach of electric vehicle energy management for voltage control Traffic Pattern / Electric Vehicles (GridLAB-D) Charge Control

EV

EV

-

EV EV +

-

FMI

08.10.2013

+

Charging Station

Pset Pcharge

API Interface



Uactual Pcharge

Charge Control

EV

+

-

Pset Pcharge

Charging Station

Uactual, Pcharge

ASN1 (TCP/IP)

FMI

Battery Model

Battery Model

(OpenModelica)

(OpenModelica)

Distributed Control System (4DIAC)

Power System Analysis (PSAT/Octave)

25

Results 

Voltage during charging process

Simulation time step varies with time due to updates of control algorithm 08.10.2013

26

Results 

Reduced charging power has impact on battery’s SOC

Increased number of updates controller notifies the simulation core more frequently, thus increasing the simulation step resolution.  Investigate interesting dynamic effects more precisely.

08.10.2013

27

Co-Sim Power System and Control 

Electric Vehicle charge management for voltage control

(GridLAB-D)

simulation control

driving behaviour distributions (departure, duration, length) EV EV electric vehicle schedule battery charger

Pset Pcharge

Pcharge,Pset FMI BatteryModel Model Battery battery model (OpenModelica) (OpenModelica) (OpenModelica)

trip data

ChargingPoint Point Charging charging point group Group Group charging Charging Charging management Management Management charging point

Pcharge,V API distribution grid (PowerFactory)

P. Palensky, E. Widl, A. Elsheikh, and M. Stifter, “Modeling intelligent energy systems: Lessons learned from a flexible-demand EV charging management,” Smart Grids, IEEE Transactions on (accepted), 2013.

10.10.2013

28

Co-Sim Power System, Components and Control System  

Detailed battery model based on chemical equations + Energy management Power Flow for every time step control system

SCADA

power system

PV system + inverter

10.10.2013

simulation results

29

Co-Sim Power System with Electrical Energy Storage  

Detailed battery model based on chemical equations + Energy management Power Flow for every time step

F. Andrén, M. Stifter, T. Strasser, and D. Burnier de Castro, “Framework for co-ordinated simulation of power networks and components in smart grids using common communication protocols,” in IECON 2011 - 37th

10.10.2013

30

Co-Sim Power System, Communication and Control 

Co-Simulation with PowerFactory API: Loose Coupling via Message Bus

M. Ralf, K. Friederich, F. Mario, and S. Matthias, “Loose coupling architecture for co-simulation of heterogeneous components – support of controller prototyping for smart grid applications,” in submitted to IECON 2013 - 39th Annual Conference on IEEE Industrial Electronics Society, Vienna, Austria, 2013.

10.10.2013

31

Electric Vehicle Simulation Environment

10.10.2013

32

EV Simulation Environment Results

Power Grid Simulation power demand

CP & EV Simulation trip data

Mobility Simulation traffic data Scenario & Use-Case 10.10.2013

33

EVSim - Architecture Architecture Transportation Simulation

Charging Point Simulation

Charge Management

Power Power System System

Agents Agents // Events Events EV EV EV EV

+

-

Charging Station

-

Charging Station

Pset SOC

EV

+

Electric Vehicle Simulation EVSim

PDG EV Charge Controller

Uactual

Pcharge

Charging Controller

PowerFactory

Specification for the simulation scenario  EV + battery, plug types including efficiency (temperature, losses)  locations and charging points  distributed generation (location based) 10.10.2013

34

EVSim - Functionality 









Configuration  simulation time: start / stop / step-size / loop / real-time  temperature dependency / performance of the battery Output (csv)  SOC, power Interfacing / co-simulation  OPC  OCPP (Open Charge Point Protocol) Behaviour  connection / disconnection handling, authorisation  time to plug / unplug  charging noise Charging algorithm  G2V, V2G, matching with local generation

10.10.2013

35

EVSim – Parameters and Interface Internal variables and parameters Electric Vehicle - battery capacity - battery Type - range ideal - range real - charge Pmax - discharge Pmax - reactive power Qmax,min - charge efficiency - charge temp. perf. - battery performance - time of stay / leave

Pmax, Pmin Pactual Pset Emax SOC, Eactual Ewish status plugtype

Charging Point - group ID - location - plugtype / phases - active power Pact - reactive power Qact - current I1act,I2act,I3act - voltage Ubus - connection Status - time of connection - time of departure

agents

temperature events, profile behavior

Pset Emax SOC, Eactual Ewish Uactual RFID

SLA

SLA

- Total Pactual

Generation - Total Pgeneration

10.10.2013

Pactual

1/3 phase

Charging Group T

Pmax, Pmin

Charge Controller Optimize Echarge=∑[Emax,Ewish] Global or location based contraints: Pactual(t)=[Pmin,Pmax] ∑Pactual = Pgeneration Umin < Uactual < Umax SLA(EV)

Pactual

Pgeneration

36

Validation of charging management 

Real and simulated EVs for charging management validation Electric Vehicle Simulation Environment EVSim EV

Charging Station

EV

Charging Station

EV

Charge Controller Renewable Generation Data Exchange Interface (OPC)

Charging Station

DEMS

ECAR DV

Distributed Energy Management

Charging Management

System

Real World

Charging Station

550 500 450 400 350 300 250 200 150 100 50 0

Tolerance range Pact Pset

60 Level Pset ΔP

Power [kW]

40 20 0 -20

11:30

11:00

10:30

10:00

09:30

09:00

08:30

08:00

07:30

07:00

06:30

06:00

11:30

11:00

10:30

10:00

09:30

09:00

08:30

08:00

07:30

07:00

06:30

-40

06:00

Power [kW]

EV

time [hh:mm]

time [hh:mm]

10.10.2013

Tolerance range, Pset and Pact during simulation and deviation

37

Simulation of EVs and power system 

Impact of unbalanced charging in low voltage networks

222,66

222,66

221,82

221,82

220,97

220,97

220,13

220,13

219,28

219,28

218,44 0,00

288,

576,

863,

1151,

[-]

1439,

DA61368_60240880: Leiter-Neutral Spannung, Betrag L1 in V DA61368_60240880: Leiter-Neutral Spannung, Betrag L2 in V DA61368_60240880: Leiter-Neutral Spannung, Betrag L3 in V

Voltages a the charging point for symmetrical loaded phases (3 times 3.68kW). Note the voltage rise due to PV generation

10.10.2013

218,44 0,00

288,

576,

863,

1151,

[-]

1439,

DA61368_60240880: Leiter-Neutral Spannung, Betrag L1 in V DA61368_60240880: Leiter-Neutral Spannung, Betrag L2 in V DA61368_60240880: Leiter-Neutral Spannung, Betrag L3 in V

Voltages for unsymmetrical loaded phase (red: 11.04kW). Note the rise in the other phase (green) due to neutral point displacement.

38

Temperature dependency 

Region Lungau (Upper Austria) – approx. 6000 EVs

Micro-simulation of region Lungau, generating trip data

10.10.2013

Charging power for opportunity charging on a winter and summer day

39

Simulation of EVs, control and power system 

Local supply - demand match in medium voltage networks Two days simulation in summer

Generation from Wind Generation from PV

800

Uncontrolled 11kW Controlled 11kW; SOC-Level: 50%

700

Ppeak [kW]

Charged Energy [kWh]

DER Energy [kWh]

DER Coverage [%]

Charging Mode

empty EVs

uncontrolled 11kW

15

751

9964

8079

54%

controlled 11kW/SOC50

55

366

6832

8079

89%

controlled 11kW/SOC25

66

324

6229

8079

99%

Controlled 11kW; SOC-Level: 25%

Power [kW]

600 500 400 300 200

Two days simulation in winter

100

20:00

16:00

12:00

08:00

04:00

00:00

20:00

16:00

12:00

08:00

04:00

00:00

0

Ppeak [kW]

Charged Energy [kWh]

DER Energy [kWh]

DER Coverage [%]

Charging Mode

empty Evs

uncontrolled 11kW

135

883

12613

3971

26%

controlled 11kW/SOC50

197

552

7267

3971

50%

controlled 11kW/SOC25

218

353

5051

3971

71%

time [hh:mm]

Uncontrolled and controlled charging of 306 EVs with 11 kW during two sumer days. Note: wind is accumulated on top of PV generation 10.10.2013

40

Summary 

Multi-agent based simulation tool, time-continuous multi-physics simulation, controls and power system simulation, e.g.: GridLAB-D, OpenModelica, PSAT, 4Diac, PowerFactory  Open source software can be used for co-simulation environments



A number of interfaces possibilities exist to couple simulation tools.  Model and integrator based coupling can be realised.



Proprietary interfaces are light-weighted but not very general, not usable for other tools, no simulation time step synchronisation  Functional Mockup Interface for model exchange + co-simulation.

08.10.2013

41