Martine Chlela, Carlos Rangel, Geza JoosMcGill University

Electric Energy Systems Laboratory

5-8 September 2017

Real-Time Hardware-in-the-Loop Co-Simulation Platform

for Microgrid Analysis

Outline Context & Motivation

Real-Time HIL Co-Simulation Platform

Microgrid Controller Testing

Comprehensive Cyber Security Analysis

Conclusion

2

Context & Motivation Energy challenges in modern power grids include:

Increased integration of intermittent renewable energy

Integration of distributed energy resources (DER)

Balance supply & demand

Two-way flow of power & information

Meeting economic & environmental constraints

To overcome the challenges smart grids should rely on information & communication technologies (ICT) to provide: Real-time sensing & measurement

Advanced control capabilities

Remote maintenance & monitoring

3

Context & Motivation The IEEE P2030 defines 3 interoperability architecture perspectives (IAP) that

together comprise the smart grid: Power System IAP - power elements & their interoperability

Communication Technology IAP - communications elements & networks

Information Technology IAP - information flows, entities & protocols used to exchange that information

A co-simulation platform needs to be developed to enable: Detailed modeling, seamless interfacing & synchronization of the 3 constituting layers to

operate as a single entity

Implementation & testing of advanced control algorithms

Analysis of the interactions between the layers & comprehensive analysis of cyber security (cyber-attack modeling, impact assessment & mitigation strategies)

4

Co-Simulation Platforms Many platforms were developed to interface the electric grids’ power system, information

exchange & communication network layers:

K. Hopkinson, et al. - EPOCHS

W. Li., et al. - VPNET

V. Liberatore, et al. - Powernet

Major drawbacks of these setups include:

Information is exchanged in the inter-synchronization period, leading to an accumulation of errors jeopardizing the fidelity of the simulations, especially the time-critical ones

Lack of scalability – not suitable for the modeling of large scale power networks

Do not operate in real-time

Do not provide a detailed modeling of either the power system or communication network components

5

Developed Real-Time Co-Simulation Platform - Key Features Study of multiple scenarios in near real conditions & without risk

Fast accurate & reliable implementation without introducing artificial delays

Seamless interfacing of the power system, communication network & information layers causing noaccumulation of errors

Hardware-in-the-Loop (HIL) capabilities with sub microsecond time steps & communication protocolimplementation capability

Optimized grid models using advanced decoupling techniques & flexible for a variety of applications

Highly scalable – could be easily extended to model larger grids

Adaptable to technological developments in engineering & configurations

6

Co-Simulation Platform Constituting Layers

DERs Modeling

Renewable DERs (WTG, PV)

ESS

Diesel Generator

Primary power management control loops

Voltage & frequency regulation

Applicable in grid-connected & islanded mode

Measurement Devices WTG ESS DIESEL THERMAL LOAD

EPS25 kV

XS1 XS2

PCC

150 kW320 kW/400 kVA

100 kW/25 kWh

600 kW

±150 kVar

25 kV/ 600 V

3 -167 kVA

25 kV/ 600 V

3 -167 kVA

25 kV/ 600 V

3 -167 kVA

320 kW/400 kVA

PEI PEI PEISG

RTS1- Microgrid Feeder & DERs

Microgrid Controller EMS

RTS2- Microgrid EMS

AnalogI/Os

IEC 61850 GOOSE

publishers/subscribers

GOOSEGOOSE

An

alo

gIn

/A

nal

ogO

ut

Co

ntr

ols

Communication Network

GOOSE GOOSE

IEC 61850 GOOSE publishers/subscribers

Secondary energy management system (EMS)

Receives microgrid measurements & evaluates dispatch points & command to operate loads & DERs

Implemented on a digital controller

Requires a communication network for information exchange

Communication network

Medium for information exchange between microgrid & EMS

Switched network providing 2-way flow of information

Information exchange

IEC 61850 GOOSE messaging protocol

TCP/IP

Analog I/Os

7

Real-Time HIL Co-Simulation Setup

GOOSEGOOSE

GOOSE

OPAL-RT RTS 1Microgrid feeder and DERs

OPAL-RT RTS 2Microgrid Controller EMS

NI cRIO Digital Controller

Host Computer 4 running EMS optimization script & LabView

Host Computer 2 Host Computer 1Host Computer 3 running OPNET

TCP

SITL Node 2 SITL Node 1

TCP

Analog I/O

GOOSE

IP Network

RTS EMS

RTS Microgrid feeder &

DERS

EMS Controller

GOOSEAnalog I/O

Commands & dispatch

Measurements

Publisher for commands

Subscriber for commands

Subscriber for measurements

Publisher for measurements

8

Microgrid Controller EMS Formulation

Optimization objectives & performance metrics

Objectives

1 Reduction of energy cost & liters offuel consumed

2 Reduction of the amount of dieselenergy used

3 Minimization of total load curtailed

4 Increase the capacity of hostingrenewable energy & reduction ofpower & energy violations

Metrics Unit

Total fuel consumed Liters

Net diesel generator energy kWh

Net cost of energy dispatchedby the diesel generator

$

Average cost of using the dieselgenerator

$/kWh

9

Microgrid Controller EMS FormulationCreation of wind speed, wind power & load

demand forecast for 2 days

Set up the size of the wind power, demand response & load

Selection of the moving forward window time in hours

Creation of the reference power data

Initialization of the parameters (SoC of ESS)

Algorithm decisions

Send dispatch results of the current time period to controller

Send dispatch commands through analog channels to emulated network

Receive measurements from the microgrid

Send the information back to the optimization engine (Matlab)

Elapsed time >= 24 hrs?

End program

NO

YES

10

Microgrid Controller EMS Testing

2 4 6 8 10 12 14 16 18 20 22 24-50

0

50

100

150

200

250

300

350

400

Time (Hours)

Po

we

r (k

W)

Pwtg Pload

2 4 6 8 10 12 14 16 18 20 22 24100

150

200

250

300

350

Time (Hours)

Die

se

l P

ow

er

(kW

)

PdieselPdiesel(offline)

2 4 6 8 10 12 14 16 18 20 22 24

-40

-20

0

20

40

60

80

Time (Hours)P

ow

er

ES

S (

kW

)

PessPess(offline)

Lo

ad &

Win

d P

rofi

les

(kW

)

Time (hrs)

Time (hrs)E

SS p

ow

er (

kW

)

Die

sel G

ener

ato

r p

ow

er (

kW

)

Time (hrs)

11

Amount of diesel consumed (L)

Diesel energy consumed (kWh)

Diesel generator operational cost ($)

Diesel energy cost ($/kWh)

Real-Time 1271.9 5191.5 1729.8 0.3332

Offline 1355.7 5750 1815 0.35

Microgrid Controller EMS Testing

2 4 6 8 10 12 14 16 18 20 22 2410

20

30

40

50

60

70

80

90

100

Time (Hours)

SO

C (

kWh

)

SOC (offline) SOC

Time (hrs)

ESS

SO

C (

%)

12

Cyber-Attacks ModelingAttacks compromising data integrity

False Data Injection Attacks (FDI)

PDER_FDI i∆t = PDER i∆t + B (tA)

s. t. i − 1 ∆t < tA ≤ i∆t &

B t = B if t ≥ tA0 elsewhere

Attacks compromising availability

Distributed Denial-of-Service (DDoS) Attacks

𝑃𝐷𝐸𝑅−𝐷𝐷𝑜𝑆 𝑘 𝑖 + 𝑗 ∆𝑡

= 𝑃𝐷𝐸𝑅 𝑘 𝑖 − 1 ∆𝑡

𝑓𝑜𝑟 𝑗 = 0, 1,… ,𝑇𝐴∆𝑡

𝑠. 𝑡. 𝑖 − 1 ∆𝑡 < 𝑡𝐴 ≤ 𝑖∆𝑡 , 𝑘 = 1, . . , 𝑛

13

Cyber-Attacks Implementation

GOOSEGOOSE

GOOSE

OPAL-RT RTS 1Microgrid feeder and DERs

OPAL-RT RTS 2Microgrid Controller EMS

NI cRIO Digital Controller

Host Computer 4 running EMS optimization script & LabView

Host Computer 2 Host Computer 1Host Computer 3 running OPNET

TCP

Linux Host Computer 5 launching cyber-attacks

SITL Node 2 SITL Node 1

TCP

Analog I/O

GOOSE

IP Network

14

Cyber-Attacks Implementation

Wireshark capture for a FDI cyber-attack

15

FDI Attack Impact Quantification Synchronous

machine’s slowdynamics & systemlow inertia cause thesmall disturbances toresult in largefrequency excursions

Transient & steady-state instability

Unnecessary activationof protection schemes

Loss in reliability &cost

100 200 300 400 500 600

-20

0

20

40

60

Time (s)

ES

S p

ow

er

(kW

)

W/O mitigation

W/ mitigation

100 200 300 400 500 600

40

50

60

70

Time (s)

Fre

qu

en

cy (

Hz)

W/O mitigation

W/ mitigation

100 200 300 400 500 600240

260

280

300

Time (s)Die

se

l g

en

era

tor

po

we

r (k

W)

W/O mitigation

W/ mitigation

100 200 300 400 500 600320

340

360

380

400

Time (s)

Lo

ad

po

we

r (k

W)

W/O mitigation

W/ mitigation

16

DDoS Attack Impact Quantification Loss of communicated

commands &measurements

As load & generation mixvary, DERs local controllercannot compensate toensure balance

Lack of coordinationbetween resources

Large excursions &unnecessary activation ofprotection schemes

Loss of reliability &uneconomic operation

100 200 300 400 500 600 700 800 9000

10

20

30

40

50

Time (s)

ES

S p

ow

er

(kW

)

W/O mitigation

W/ mitigation

100 200 300 400 500 600 700 800 900200

250

300

Time (s)Die

se

l g

en

era

tor

po

we

r (k

W)

W/O mitigation

W/ mitigation

100 200 300 400 500 600 700 800 900

45

50

55

60

Time (s)

Fre

qu

en

cy (

Hz)

W/O mitigation

W/ mitigation

100 200 300 400 500 600 700 800 900

330

340

350

360

370

380

Time (s)

Lo

ad

po

we

r (k

W)

W/O mitigation

W/ mitigation

17

Multi-Stage Cyber-Resilient Control Strategy

Main Grid

PCC

Renewable

DERs Energy Storage

System

Controllable

Loads

Diesel

Generator

Microgrid Controller

EMS

18

Comprehensive Cyber Security Analysis

Min/Max Frequency (Hz)

Renewable Energy Shed (kWh)

Load Not Served (kWh)

Average Cost of Energy ($/kWh)

S1 – No Control 36.8 / 60 0 13.6 0.2645

S1 - Control 59.3 / 60 0 0 0.2483

S2 – No Control 13.4 / 70.4 40.4 0 0.3315

S2 - Control 59.9 / 60.5 0 0 0.3195

S1 - No Control 42.1 / 60 0 9.1742 0.2505

S1 - Control 59.3 / 60 0 0 0.2481

S2 - No Control 8.2 / 65.6 47.95 0 0.3962S2 - Control 59.9 / 60.5 0 0 0.3335

FD

ID

Do

S

The multi-stage control infrastructure provides: Enhanced resiliency against cyber-attacks

Higher microgrid ability to host renewable energy & supply critical loads

Transient & steady-state stability

Lower cost of operation

19

Potential Uses of the Platform The proposed co-simulation platform evaluated the performance of a

microgrid controller EMS & analyzed cyber security mainly from a power system perspective

Although the platform models microgrid systems, it has all the building blocks to be easily extended to model large power grids

Amongst many, other applications of the co-simulation platform include: Testing the performance of different communication technologies & protocols

Evaluating the effectiveness of recommended cyber security practices & guidelines applied at the communication network layer to ensure resiliency

Conducting other power system studies (protection, EV charging & energy management algorithms, demand response…)

20

Conclusions Detailed modeling of the real-time HIL co-simulation setup, its

constituting power system, information exchange & communication network layers & their interfacing was presented

An EMS microgrid controller has been formulated & its performance were validated

Cyber security studies have been conducted to model cyber-attacks, assess their impact & propose mitigation solutions

The benefits of the platform have been detailed and the broad range of applications to which its suitable were presented

21

THANK YOUMartine Chlela

martine.chlela@mail.mcgill.ca

Carlos Rangelcarlos.rangel@mail.mcgill.ca

Electric Energy Systems Laboratory, McGill Universityhttp://www.power.ece.mcgill.ca/