Emtp Mahseredjian R&D

EMTP‐RVEMTP‐RVResearch and developmentp

Jean MahseredjianProfessorProfessor

[email protected]École Polytechnique de MontréalÉcole Polytechnique de Montréal

Thursday, April 29

History: R&D projectHistory: R&D project

R h d d l t i ti D l t• Research and development organization: Development Coordination Group (DCG‐EMTP)

• EMTP: Electromagnetic Transients Program, developed since the g g p70s, major versions in 90 and 96

• Completely new software and technology: EMTP‐RV• Large and complex project: total duration 5 years• Large and complex project: total duration 5 years• First commercial release version 1 in 2003• Large scale software with more than 1 million lines of codeg• New computational engine and New graphical user interface

(GUI)C i li d t• Commercialized: www.emtp.com

• DCG Members: Hydro‐Québec, Électricité de France, CRIEPI (Japan), Entergy, American Electric Power, Western Area Power

2

( p ), gy, ,Administration, US Bureau of Reclamation, Hydro‐One, CEATI

Old EMTP software and technology

New Computation MethodsNew EMTP‐RVNew EMTP RV

(Restructured Version)

3

Support and developmentSupport and development• Level 1: Neil MacKenzie, Capilano Computinge e e ac e e, Cap a o Co put g• Level 2: Awa‐Marie Ndiaye, CEATI• Level 3: Jean Mahseredjian, École Polytechnique• Development: Jean Mahseredjian, Chris Dewhurst (Capilano)

– Team at École Polytechnique• Luis Daniel Bellomo research associateLuis Daniel Bellomo, research associate• Many Ph.D. students• Many M. A. Sc. students

• Special developments:• Special developments:– Several funded projects with Hydro‐Québec– Several funded projects with EDFp j

• Major contributors:– Hydro‐Québec– EDF– Developments, funding, funding of research

Courses on EMTP RVCourses on EMTP‐RV

C i 2008• Courses in 2008– Australia (May)Saudi Arabia (June)– Saudi Arabia (June)

– Madison (University of Wisconsin)– Montréal (September)– Montréal (September)– Paris (Supélec, September)– Orléans (Vergnet, éoliennes, September)Orléans (Vergnet, éoliennes, September)

• Courses in 2009– Special course for Hydro‐Québec, MarchSpecial course for Hydro Québec, March– Croatia, April– New Orleans, US, November, ,

Other coursesOther courses

• Courses on transients (not software)– Seoul, South Korea, Sungkyunkwan University, , , g y y,April 2009

– Special long course every year ÉcoleSpecial long course, every year, École Polytechnique de Montréal (web page)

Seoul South Korea Sungkyunkwan University– Seoul, South Korea, Sungkyunkwan University, August 2009

New version 2.2New version 2.2• What is new in 2.2

– Full compatibility with Vista– New documentation system with new navigation features– Various improvements and additions to models. The data handling

features for several models are now simplified to allow easier loading h l l l d dwhen separately calculated data.

– New capability to store complete circuits in libraries. A circuit appearing in a library folder now becomes listed in the library Parts Palette and can be dragged and dropped into a design just likePalette and can be dragged and dropped into a design just like standard parts. This is a very powerful feature that provides easy access to user circuits and allows maintaining more complex models through libraries.g

– Subcircuits are now given the Model or Physical attribute in the Subcircuit Info menu. A model subcircuit is primarily intended to define the operation of the device represented by its parent symbol. A h i l b i it i i il d t t i f th t Thphysical subcircuit is primarily used to contain some of the system. The

devices inside the subcircuit represent actual physical elements of the system. The physical subcircuit may contain Model subcircuits. This distinction allows propagating computed data into Physical subcircuitsdistinction allows propagating computed data into Physical subcircuitsfor visualization purposes.

New version 2 2

S l i i h d i l di d i

New version 2.2

– Several new scripting methods, including: dynamic modification of device symbol using a separately stored symbol drawingstored symbol drawing.

– Several improvements• New ScopeViewNew ScopeView

– Vista compatible– Several improvements

A HVDC d l b h k (f 50 H d 60 H k )• A new HVDC model benchmark (for 50 Hz and 60 Hz networks) originally developed by professor Vijay Sood (University of Ontario Institute of Technology) is now available upon request. This work resulted from a collaboration with Sébastien Dennetière (Électricité de France) and École Polytechnique de Montréal.

Scenario attributeScenario attribute• Allows changing scenarios in one easy step

• Each device is given a Scenario attribute and a Scenario.Script attributep– Built‐in

• Simple user‐defined scenariosSimple user defined scenariosdev=defaultObject()Scenario=dev.getAttribute('Scenario');switch (Scenario){switch (Scenario){

case '1' :dev.setAttribute('Exclude','Ex')

break;break;

case '2' :dev setAttribute('Exclude' '')dev.setAttribute( Exclude , )

break;}

Recently completed R&D projectsRecently completed R&D projects

0 f S h hi• 0‐Hz startup of Synchronous machine– Project EDF R&D, Clamart– Allows using the synchronous machine model without 60 Hz or 50 Hz initialisationS f 0 H– Starts from 0 Hz.

– Allows studying the machine startup and synchronization onto the networksynchronization onto the network

– For pumped storage studiesFor black start st dies– For black‐start studies

• Improved wind generator models

Modeling and Simulation of the Startup of a Pumped Storage Power Plant Unit

• IPST 2009 U K J M h dji S D tiè• IPST‐2009 paper, U. Karaagac, J. Mahseredjian, S. Dennetière

re zdHzed

+-

+-

Freq

uenc

yC

ontro

ller

Ang

leC

ontro

ller

+- Speed & TorqueControl

Cur

rent

Lim

iter

Ir

SM

freq

uenc

y

Grid

angl

e

Δf<

1Hz

Δθ<

45

Activ

ated

whe

n

°

f>

47H

Activ

ate

whe

n

Pos

ition

-en

tro

ller

rter

rolle

r

+

L

G

ridfre

quen

cy

a

Grid

frequ

ency

SM

ang

le

>47

Hz

tivat

edlta

geco

ntro

len

Rot

or P

Cur

reC

ontr

PLL

Inve

rC

ontr + -

+ -

PLL

Ire Grid

ang

le

Grid

volta

ge

f>Act

Vol

wh+

+

+

+ N

etw

ork

+- SM

Exc

itatio

nS

yste

m

+

S

Mvo

ltage

+

+

• Measured and simulated frequenciesMeasured and simulated frequencies

51

50

Hz)

49

quen

cy (H

48

Freq

105 110 115 120 12547

Time (s)

1250A

)

1200urre

nt (A

1200

field

cu

1150

Mac

hine

95 100 105 110 1151100

Time (s)

M

Machine field currents

Time (s)

1.8 x 104V

)

1 6ltage

(V

1.6

-line

vol

1.4

s lin

e-to

-

80 90 100 110 1201.2

i ( )

rms

M hi t i l li t li lt

Time (s)

Machine terminal rms line‐to‐line voltage

5

0W

)

-5

ower

(MW

-10

ctiv

e Po

-15Ac

80 90 100 110 120-20

Time (s)

Active power delivered by the machine

Improved Wind generator modelsImproved Wind generator models

• Generic models– Detailed

– Mean‐value models

M t hi f PSS/E lt f l t i t• Matching of PSS/E results for slow transients

• Initialization scriptsp

• Flicker meters

W k l d b L D B ll d J• Work completed by L. D. Bellomo and J. Mahseredjian (École Polytechnique)

10 generatorsWTG1

+

SW1

WINDLV11.00/_6.6

+ ZnO

O1

12

?

34.5

/0.6

9

ZnO

+

1.00/_6.3

WINDLV2

1 2

230/34.5

YD_1

+

230kVRMSLL /_0

Network 1 2

34.5/0.69

WTG2

+

MAIN_SW

+

+

+

nO3

+ SW2

BUS12

10 generators

.25

5Ohm ZZ

WTG21

2 34.5

/0.6

9

+ZnO

Zn +Zn

OZnO2

WTG3

+ SW3

WINDLV3

0.99/_5.9

10 generatorsWTG3

60

80

0

20

40

60

V)

60

-40

-20

0

(kV

0 0.5 1 1.5 2-80

-60

time (s)

2 5

3

3.5

Voltage

1.5

2

2.5

(pu)

0

0.5

1

Obvervoltage trip signal Crowbar signal

0 0.5 1 1.5 2 2.50

time (s)

Improvements to the load‐flow module (next versions)

P i d l i f i h l i• Presentation and location of worst mismatch locations• Presentation and location of reactive power violations• Presentation of PQ power on transmission lines (on the design symbols)

l l f• Automatic calculation of tap positions– Automatic initialization for tap control signals

l l f h h l• Automatic calculation of asynchronous machine slip from mechanical power or electrical powerTh t l ti• The area control notion

• Attribute scripting for device data based on LF solution

ToolboxesToolboxes

CRINOLINE l t ti tibilit• CRINOLINE: electromagnetic compatibility• EGERIE

– Short‐circuit analysis packageShort circuit analysis package– Automates short‐circuit studies

• Harmonic analysis– Harmonic source models– Analysis tools– Compensator modelsCompensator models

• Parametric studies– Advanced functions, high level scripting– Scenario studies

• LIPS: Lightning impact on power systemsAutomation level for lightning analysis– Automation level for lightning analysis

Other worksOther works

C i f i i d i i t t th bj t• Conversion of remaining device scripts to the object‐oriented version

• Scripts for automatic layout of signals automaticScripts for automatic layout of signals, automatic connections for building entire networks

• Simplified SVC model: controlled inductance (currently p ( yavailable)

• Switching to the Intel compilerb l f– Compatibility of DLLs

• New C/C++ DLL (prebuilt) for direct interfacing through DLL (IREQ)DLL (IREQ)

• New DLL specific to control systems, based on perturbation theoryp y

Modeling of transmission lines and cables

C li i i• Current limitations– The Wideband model may encounter numerical problems

• Can be fixed by user manipulations of the fitting function not• Can be fixed by user manipulations of the fitting function, not simple

• Complex research problem in the literature, many papers• Prominent problem for short cable

• Development of a new fitting method: WVFb f l h• Contribution of an error control technique in time‐

domainM b t t bl d l– More robust, stable model

• Results presented in IEEE papers

( )l= −H exp YZ ( )11e e

−−TΛT ΛH T T( )lH exp YZ ( )e e= =H T T

1

Ns n

modenn

cH e

s pτ−

=≅

+∑ ( )

1 1( )

NM m ij mn s mij

mnm n

cH s e

s pτ− ⋅

= =

⎡ ⎤≅ ⎢ ⎥

+⎢ ⎥⎣ ⎦∑ ∑

1 nn p= ⎣ ⎦

102

104 Magnitude of modes in H(1,1)WB Calculated

4,5,3,2 wb

100

10

gnitu

de 1

, , ,

1 wb

10-4

10-2

Mag

7 wb

6 wb

100 102 104 10610-6

(Hz)

7 6,3,2,4,5

2CEWB FDQ WB Δt 1 microsec

CEWB V10

1.5

2 CEWB_V10FDQ_V10 WB_V10

1

Vol

tage

0 0 01 0 02 0 03 0 04 0 05 0 06 0 07 0 080

0.5

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08t (ms)

2WB CWB FDQ Δt 0.1 μs

CWB_V10FDQ V10

1.5

age

Q_WB_V10

0.5

1

Vol

ta

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.080

t (ms)

Other R&D based on EMTP RVOther R&D based on EMTP‐RV• New hysteretic reactor model, completed M. A. Sc. project

– Better fitting method• Other hysteretic reactor models:

– Preisach based model (University of Toronto), completed– Programming of the old EMTP type 96, started

• Vacuum breaker model, currently available• Fast to superfast computationsFast to superfast computations

– The dynamic phasor approach for slow transients (stability analysis needs)

– Relaxation techniquesq– Automatic adjustment of synchronous machine solutions for slower

transients– Parallel computationsp

• Using the Multi‐Core processors• One simulation to many simulations

– New solution methods for control systemsNew solution methods for control systems– FPGA programming of a sparse‐matrix based solver solver

New solution methods for control systems (research)

I t f d• Improvement of speed• Reduction of Jacobian matrix size (demonstration prototype)prototype)

• Elimination of the matrix based solverE i d i i d 5 10 i• Estimated gains in speed: 5 to 10 times

• Research on a single system of equations: power d t l di b d d land control‐diagram based models

N6

Control system equationsN6

L1 L2 L3 L4 L5

N7

Computation of Jacobian matrix

Iterative solverN7 N7 L4

N6 N6 L5

f ( )1f ( )1

1 k 0

⎡ ⎤⎡ ⎤ ⎡ ⎤⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥

x xx xx Computation of Jacobian matrix

by perturbationL5L5

L4L4

L3

1 k 01 k 0

1 1 -1 01 k 0

⎢ ⎥⎢ ⎥ ⎢ ⎥− ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥−⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥

=

xxxx

=Jx bL2 L2

L1

1 k 01 1 -1 0

1 su

⎢ ⎥⎢ ⎥− ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦⎣ ⎦ ⎢ ⎥⎣ ⎦

xx

27

Other R&D based on EMTP RVOther R&D based on EMTP‐RV

• Database!

• Development of portable data modelingDevelopment of portable data modeling methods

P t bilit t d d CIM V il VHDL?– Portability standards: CIM, Verilog‐VHDL?

– Data

– Portable modeling between applications

• New IEEE Task ForceNew IEEE Task Force

Very large networksVery large networksG2

Radisson_b720

+

+

+

330

MX

+

7063

7062

L706

1

144

BUS2BUS1 O

ZnO

O

_ZnO

C61

C62

C63

O

ZnO

+

L

+

L

+

L

+24

+

144P Q

Load

+ ++ G1

SEND

BUS1

+Zn

O

CXC

63_Z

+Zn

O

CX

C61

_

+ CXC

+ CX

C

+ CX

C

+

+

330

MX

+

+

330

MX

+

+Zn

O

CXC

62_Z

+

330

MX

B18

B19

East

mai

nlas

arce

lle

Nemiscau b780

+193+

B

+Zn

O

+Zn

O

SEND

REC

SV

C

+

+

330

MX

+

+

330

MX

+

V

I

NemiscauCLC

+

330

MX

Nemiscau_b780

+

L708

2

+

L708

1

+

L708

0

EquivalentDetails

+Zn

O

CXC

82_Z

nO +

ZnO

CXC

81_Z

nO

+Zn

O

CXC

80_Z

nO

+

CXC

80

+

CXC

81

+

CXC

82

Details

Hydro Québec NetworkHydro‐Québec Network

• IPST‐2009 paper, L. Gérin‐Lajoie, J. Mahseredjian

• Complete network (L)p ( )– The complete Hydro‐Québec network is organized using a multilevel hierarchical design structured on 6using a multilevel hierarchical design structured on 6 pages in the GUI. There are a total of 30000 physical devices and 28000 signals. The list of physical devices g p yincludes 19000 control devices and coupled 3, 6 or 9‐phase devices are counted once. The signal count adds 8000 power nodes to 20000 control system signals.

• Complete network (L)Th l l li i ( b k d) f i– The top level listing (subnetwork contents are not counted) of main devices is:

– 1100 transmission lines representing the existing 1560 lines and derivationsderivations

– 296 three‐phase transformers representing the existing 1500 three‐phase units connected in Ynyn, DD, Dyn, Ynd, Ynynd, Yndd and ZigZaggrounding banksgrounding banks

– 532 load models representing a total of 36000 MW. All medium and high voltage shunt capacitors and inductors were modeled separately. Some loads were modeled with the transformer and shunt capacitor pat the lower voltage level.

– 7 SVC (Static Var Compensator) models of 300 Mvars and 600 Mvars. The SVCs have been combined on some buses by creating 600 Mvar

d lmodels.– 32 series capacitor MOVs and 303 nonlinear inductances used for high

voltage power transformer saturation representation.99 h hi (SM) ith i t d t l ti– 99 synchronous machines (SM) with associated controls representing more than 49 power stations and four synchronous compensators. All synchronous machine devices are matched to corresponding load‐flow type devices for specifying the PV constraints used for initializingtype devices for specifying the PV constraints used for initializing machine phasors at load‐flow solution convergence. All machines are given a single‐mass model except one nuclear power plant generator modeled using 10 masses.

• Reduced networkReduced network– The reduced network has a total of 24000 physical devices and around 24000 signals. There are 4000 power devices and 2500 power nodes. The listing of top level devices is:170 lines with 75 lines at the 735 kV level 53 at– 170 lines, with 75 lines at the 735 kV level, 53 at 315 kV, 23 at 230 kV and 19 at 120 kV

– 90 three‐phase transformersp– 27 load models, 7 at 315 kV, 6 at 230 kV, 4 at 161 kV, 6 at 120 kV and 4 at 13.8 kV for a total of 33800 MW

d l– 7 SVC models– 39 synchronous machines with AVRs for representing 31 power stations and 3 synchronous compensators31 power stations and 3 synchronous compensators for a total of 35600 MW of generation.

400Substation no.1

400Substation no.4

0

100

200

300

0

100

200

300

0 200 400 600 800 1000 12000

300

400Substation no.2

0 200 400 600 800 1000 12000

60

Substation no.5

0 200 400 600 800 1000 12000

100

200

300

0 200 400 600 800 1000 12000

20

40

0 200 400 600 800 1000 1200

300

400Substation no.3

0 200 400 600 800 1000 1200

100

Substation no.6

0 200 400 600 800 1000 12000

100

200

Frequency (Hz)0 200 400 600 800 1000 1200

0

50

Frequency (Hz)

Frequency response (positive sequence impedance) plots for the complete (blue) and reduced (green) networks. Left column plots show th 735 kV b t ti d i ht l l t h th 315 kV

Frequency (Hz) Frequency (Hz)

three 735 kV substations and right column plots show three 315 kV substations.

1.04Substation no.1 - Bus voltage (pu)

1Substation no.2 - Bus voltage (pu)

1.01

1.02

1.03

0 0.2 0.4 0.6 0.81

2460Line no.2 - Transmitted Power (MW)

0 0.2 0.4 0.6 0.80.99

2280Line no.1 - Transmitted Power (MW)

2430

2440

2450

2220

2240

2260

0 0.2 0.4 0.6 0.82420

2620Power plant no.1 - Power flow (MW)

0 0.2 0.4 0.6 0.82200


2580

2600

5620

5640

0 0.2 0.4 0.6 0.82560

Time (s)0 0.2 0.4 0.6 0.8

5600

Time (s)

Network initialization test without SVCs, L‐Network (blue), R‐Network (green) and PSS/E (red)

1.04Substation no.1 - Bus voltage (pu)

1Substation no.2 - Bus voltage (pu)

1

1.02

0 99

0.995

0 0.2 0.4 0.6 0.81

2440

Line no. 2 - Transmitted Power (MW)0 0.2 0.4 0.6 0.8

0.99

2300Line no. 1 - Transmitted Power (MW)

24202250

0 0.2 0.4 0.6 0.82400

2600

Power plant no.1 - Power flow (MW)

0 0.2 0.4 0.6 0.82200


2540

2560

2580

2600

5600

5650

( )

0 0.2 0.4 0.6 0.82520

Time (s)0 0.2 0.4 0.6 0.8

5550

Time (s)

Network initialization test with SVCs, L‐Network (blue), R‐Network (green) and PSS/E (red)

1.061.08

Substation no.1 - Bus voltage (pu)

11.021.04

Substation no.2 - Bus voltage (pu)

0 981

1.02

1.04

0 920.940.960.98

1

0 5 100.98

0 2 4 6 8 100.92

2300


2500


2100

2200

2300

2300

2400

2500

0 5 102000

0 2 4 6 8 102200


5800


2400

2600

5400

5600

5800

Simulation of a 3 phase fault and loss of a 735 kV transmission line

0 5 102200

Time (s)0 2 4 6 8 10

5200

Time (s)

Simulation of a 3‐phase fault and loss of a 735 kV transmission line, L‐Network (blue), R‐Network (green) and PSS/E (red)

64

66 a) Generator frequencies at James Bay Complex

60

62

64

Hz

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 260

1

2 b) Prospective TOV at LVD7

2

-1

0pu

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-2

1

2 c) TOV at LVD7 with LVD7-Montreal tripping

-1

0

1

pu

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-2

time (s)

James Bay system voltage oscillations due to an extreme disturbance

30000 devices28000 signals

0

2000

4000

6000

80008000

10000

0 2000 4000 6000 8000 10000 1200012000

Solved time‐domain sparse matrix for the L‐Network, 50269 non‐zeros

CPU ti i ( ) f 10 i l ti i t lCPU timings (s) for a 10 s simulation interval

CPU Timers L-Network R-Network

GUI File (design) load 9 4

Data generation 10 3

Load-flow solution 181 (6 iterations) 21 (7 iterations)

Steady-state solution 0.48 0.12

Time-step 100 µs 200 µs 100 µs 200 µs

Time-domain network equations 4710 2548 538 276

Time-domain control equations 846 435 715 389

Time-domain updating 409 210 75 36

Time-domain solution total 596599 min

310352 min

132822 min

70112 min