Load flow PPT

93
Power flow calculations Dirk Van Hertem Hakan Ergun Priyanko Guha Thakurta Research group Electa Department of electrical engineering (ESAT) K.U.Leuven, Belgium DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 1 / 33

description

Load flow PPT

Transcript of Load flow PPT

Page 1: Load flow PPT

Power flow calculations

Dirk Van HertemHakan Ergun

Priyanko Guha Thakurta

Research group ElectaDepartment of electrical engineering (ESAT)

K.U.Leuven, Belgium

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 1 / 33

Page 2: Load flow PPT

Introduction

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 2 / 33

Page 3: Load flow PPT

Introduction

Introduction: load or power flow

What are “power flow calculations”

Calculating the power flow (active and reactive) through all thelines in the power system

Calculating the voltages (amplitudes and angles) at every node(substation)

Determination of the static state of a given system

Knowing only:

Grid configuration and parameters (R and X )Power outputs of generator unitsLoads (active and reactive)Some voltages

“Load flow” and “power flow” are synonyms

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 3 / 33

Page 4: Load flow PPT

Introduction

Introduction: load or power flow

Why is load flow important?

Assessing if the power system is:

Within operational limitsSafe (N-1)

Basis for other (e.g. dynamic) calculations

Checking whether future situations are valid

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 3 / 33

Page 5: Load flow PPT

Introduction

Introduction: load or power flow

When is it used?System planning

System operations

State estimation

Dynamic simulations (basis, first calculation)

. . .

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 3 / 33

Page 6: Load flow PPT

Introduction

Introduction: load or power flow

2˚30’ E . Greenwich 2˚40’ 2˚50’ 3˚00’ 3˚10’ 3˚20’ 3˚30’ 3˚40’ 3˚50’ 4˚00’ 4˚10’ 4˚20’ 4˚30’ 4˚40’ 4˚50’ 5˚00’ 5˚10’ 5˚20’ 5˚30’ 5˚40’ 5˚50’ 6˚00’ 6˚10’ 6˚20’ 6˚30’ 6˚40’ 6˚50’

2˚40’ E . Greenwich 2˚50’ 3˚00’ 3˚10’ 3˚20’ 3˚30’ 3˚40’ 3˚50’ 4˚00’ 4˚10’ 4˚20’ 4˚30’ 4˚40’ 4˚50’ 5˚00’ 5˚10’ 5˚20’ 5˚30’ 5˚40’ 5˚50’ 6˚00’ 6˚10’ 6˚20’ 6˚30’ 6˚40’ 6˚50’

49˚30’49˚40’

49˚50’50˚00’

50˚10’50˚20’

50˚30’50˚40’

50˚50’51˚00’

51˚10’51˚20’

51˚30’

7˚00’

51˚3

0’51

˚20’

51˚1

0’51

˚00’

50˚5

0’50

˚40’

50˚3

0’50

˚20’

50˚1

0’50

˚00’

49˚5

0’49

˚40’

49˚3

0’

Alzette

Mos

elle

Our

S˚re

Vierre

Semois

Lesse

Ourthe

Sambre

Warche

Amblève

Gileppe

Vesdre

Meu

se

Maas

Demer

Grote Nete

Ourthe

Vesdre

Meu

se

Dyle

Dyle

Rupel

Dender

Dendre

Sche

lde

Leie

Escaut

IJzer

Meu

se

Sambre

Eau

d'H

eure

Echelle Schaal

0 10 20 30km

1:1000000Situation au

stand op 1-1-2005

Institut Géographique National Nationaal Geografisch Instituut

CENTRALESCENTRALES

c. thermique

c. nucléaire

c. hydraulique

c. de pompage

existante

en projet ferme

existante

en projet ferme

existante

en projet ferme

existante

en projet ferme

bestaande

vast ontworpen

bestaande

vast ontworpen

bestaande

vast ontworpen

bestaande

vast ontworpen

kernenergie c.

thermische c.

hydraulische c.

pomp-c.

POSTES STATIONS

existants

en projet

bestaande

in ontwerp

22

LIGNES AERIENNES BOVENGRONDSE LIJNEN

Tension nominale

380 kV

220 kV

150 kV

70 kV

Nominale spanning

380 kV

220 kV

150 kV

70 kV

en construction ou en projet

CABLES SOUTERRAINS ONDERGRONDSE KABELS

150 kV

70 kV

meervoudige stroomketen

150 kV

70 kV

circuits multiples

2e terne en construction ou en projet 2de draadstel in aanbouw of in ontwerp

in aanbouw of in ontwerp

2

2

2

2

2

2

3

2

2

2

2

2

2

2

2

AVELGEM

MASTAING

RODENHUIZE

ZOMERGEM

MARQUAIN

WARANDE

AVELIN

CHEVALET

VERBRANDEBRUG

MACHELEN

HOENDERVELD

WESPELAAR

KOKSIJDE

BEERST

STADEN

SLIJKENS

BRUGGE

ZEDELGEM

MALDEGEM

OOSTBURG

AALTER

EEKLO

LANGERBRUGGESADA.

RINGVAART

EEKLO NOORD

SIDMAR

NIEUWEVAARTHAM

FLORA

DRONGEN

KENNEDYLAAN

DEINZE

ST-BAAFS-VIJVE

WORTEGEM

RUIEN

OUDENAARDE

MOEN

PITTEM

TIELT

MUIZELAAR

BEVEREN

RUMBEKE

IZEGEM

Oostrozebeke

DESSELGEM

HARELBEKE

HEULE

ZWEVEGEMWEVELGEM

PEKKE

MOUSCRON

DOTTIGNIES

MENENWEST

IEPER

IEPERNOORD

GEERTRUIDENBERG

KREEKRAKBORSSELE

BUGGENHOUT

MALDEREN

LEEST

MALLEBEERSE

POEDERLEE

HEZE

MASSENHOVEN

AMOCO

ESSOCHEM

HERCULES

TESSENDERLO

BERINGEN

MOL

BALEN

DIEST

VIEILLEMONTAGNE

LOMMELOVERPELT

HOUTHALEN

MHO

DODEWAARD

EINDHOVEN

MAASBRACHT

OBERZIER

EISDEN

KNP

ZUTENDAAL

STALEN

GENK-LANGERLO

GODSHEIDE

SIKELWIJGMAAL

WILSELE

WOLUWE-ST-L.ST-L.WOLUWE

GRIMBERGEN

RELEGEM

TERLINDEN

KOBBEGEM

MOLENBEEK

DILBEEKHELIPORTWIERTZ

DHANISQ.DEMETSk.

ZUID/MIDI

IXELLESELSENEFOREST

VORST

LABORELEC

RHODEST-G.ST-G.RODE

BUIZINGEN

BRUEGEL

DROGENBOS

MEKINGEN

EIZERINGEN

SIDAL

PUTTEMECHELEN

KRUISBAAN

HEIST/BERG

ZELE

ST-GILLIS-DENDERMONDE

MEERHOUT

HERDERENBRUSTEM

AVERNAS

TIENEN

CORBAIS

LECHENOI

WATERLOO

BRAINE-L'ALLEUD

BAISY-THY

VIEUXGENAPPE

NIVELLES

OISQUERCQ

CHIEVRES

LIGNETHIEULAIN

WATTINESGAURAIN

Carr.Milieu

ANTOING

THUILLIES

JAMIOLLE

NEUVILLEPLATE-TAILLE

CHOOZ

VIREUX

LUMES

REVIN

MAZURES

LONNY

VESLE

Vierre

HEINSCH

LATOURCAB

St-MARD

AUBANGE

MOULAINE

HERSERANGE

OXYLUX

LANDRES

BERTRANGE

HEISDORF

TRIERROOST

FLEBOUR

VIANDENS.E.O.BAULER

NIEDERSTEDEM

VILLEROUX

MONT-LEZ-HOUFFALIZE

BRUMECOO-TROIS-PONTS

Bévercé

Heid-de-Goreux

Butgenbach

RIMIERETIHANGE

GRAMME

HteSARTEAndenne

SEILLES

Grands-Malades

COGNELEE

CHAMPION

Croix-Chabot

AWIRS

CLERMONT

Ivoz

LEVAL

LATROQUE

SERAING

BRESSOUXJUPILLE

Monsin

CHERTAL

BERNEAULIXHE

Gileppe

Eupen

PT-RECHAIN

MARCOURT

ACHENE

NINOVE

MERCHTEMAALST NOORD

AALST

RIMIERETIHANGE

GRAMME

HteSARTE

Croix-Chabot

AWIRS

CLERMONT

Ivoz-Ramet

LATROQUE

SERAING ROMSEE

BRESSOUX

JUPILLE

CHERTAL

BERNEAU

LIXHE

Ampsin-Neuville

Ans

Vottem

LEVAL

BELLAIRE

Monsin

CHERATTE

MARCHIN

ZANDVLIET

DOEL

BASF

SOLVAYLILLO

BAYER

FINAKETENISSE

EKEREN

7eHAVENDOK

MERKSEMSCHELDE-

LAANKALLO

ZURENBORGBURCHT

ZWIJNDRECHT

WOMMELGEM

OELEGEMBEVEREN

MORTSEL

SCHELLE

-DORP

ST-PAUWELS

HEIMOLENMERCATOR

St-JOB

LIER

LINT

ZANDVLIET

DOEL

BASF

SOLVAY

LILLOBAYER

FINAKETENISSE

EKEREN7eHAVENDOK

MERKSEMSCHELDELAAN

KALLO

ZURENBORG

BURCHT

ZWIJNDRECHT

WOMMELGEM

OELEGEM

BEVEREN

MORTSEL

SCHELLE

-SCH.DORP

ST-PAUWELS

HEIMOLEN

MERCATOR

LIER

LINT

OORDEREN

WAARLOOS

MARCHE-LEZ-ECAUSSINNES

PETROCHIMFELUY

BAUDOUR

AIRLIQ.

TERTRE

QUAREGNON

GHLINOBOURG

HARMIGNIES

V/HAINE GOUYLACROYERE

BASCOUP

TRIVIERES

PERONNES

BINCHE

COURCELLES

GOSSELIES

FLEURUS

AMERCŒUR

DAMPREMY

MONCEAUMONTIGNIES

ST-AMAND

TERGNEE

AUVELAIS

FARCIENNES

JEMEPPE-SOLVAY

Gembloux

MARCHE-LEZ-ECAUSSINNES

PETROCHIM

FELUY

BAUDOURAIR

LIQUIDE

TERTRE

QUAREGNON

GHLIN

OBOURG

HARMIGNIES

VILLE/HAINE

GOUY

LACROYERE

BASCOUP

TRIVIERES

PERONNES

BINCHE

COURCELLES

GOSSELIES

FLEURUS

AMERCŒUR

DAMPREMY

MONCEAUMONTIGNIES

ST-AMAND

TERGNEE

AUVELAIS

FARCIENNES

JEMEPPE-SOLVAY

Gembloux

JEMAPPES

BOELLL

BOELHF

BOELTCC

BOELFOUR

MALFALISE

CARALFOC

CSTMMF.DEFER

BLANCHISSERIE

LAPRAYE

PONT-de-LOUP

BUISSERET

150+

70

220+

150

380+150

150(380)

380+150

150+70

150

+70

220+150

150+70

380+

220(

2x38

0)

380+

220

(2x3

80)

380+

220

150(220)

150 + 70

150 + 70

70+

150

150+

70

380+150

380+150

150+70

150+70

150+

70

150+70

150 +70150 + 70

150+70

150+30

150+70

150+70

150+

36

150+36

380+220

150+70

380 + 150

150+

36

380+

150

150 +36

150

150+70

150+70

150+70

150+70

150+70

150+70

150+70

150+70

6

1

9

1

7

19

9

7

16

12

12

4

2

11

16

152

34

8

17

14

16

16

164

3

13

3

2

2

2

ZAVENTEM

2

BLAUWE TOREN

3HERDERSBRUG

150+70

150+70

2

380+

150

DAMPLEIN

MONTST.MARTIN

Pouplin

DAMPLEIN

HARENHEIDE

220+150(220)

Blandain

Noordschote

Gavere

ST-Denijs Boekel

Zottegem

Geraardsbergen

Deux-Acren

Ronse

KUURNEBekaert

Kortrijk-NMBS

K.Oost

Bas Warneton

Westrozebeke

Bornem Willebroek

Tisselt

Rijkevorsel

Turnhout

Ravels

HerentalsOlen

Herenthout

Geel/Oevel

Nijlen

Hechtel

(I.E.)

Gerdingen

St-Huibr.-Lille

Maaseik

Opglabbeek

Bekaert

Maasmechelen

Bilzen

Paalsteenstr.

Hasselt

(NMBS)

Alken

Halen

Kersbeek

AarschotDorenberg

Kessel-lo

PellenbergLeuven(NMBS)

HeverleeGasthuisberg

Duffel

NMBS

Muizen

Langveld TIP

Baasrode

Lokeren

Lanaken

Tongeren

BorgloonSt-Truiden

LandenNMBS

Jodoigne

Glatigny

Aische-en-Refail

Sauvenière

Court-St-Et.

Ottignies(SNCB)

SNCBBaulers

Ronquières

SNCB

Enghien (SNCB)

Braine-le-C.

Soignies

Hoves

Meslin

Ath(SNCB)

Quevaucamps

Tournai

SNCB

Harchies

Lobbes

Clermont

Renlies

Beaumont

Solre St-Géry

Thy-le-Château

MomigniesLesForges Baileux

FOURMIES

Couvin

Romedenne

Pondrôme

Monceau-en-Ard.

Fays-les-Veneurs

Orgeo

Hatrival(SNCB)

Recogne

NeufchâteauLonglier(SNCB)

Marbehan(SNCB)

Respelt

Villers-s/SemoisChiny Bonnert

SNCB

Arlon

Differd.Arbed

Belv.Arbed

Herbaimont

St-Vith

Cierreux

Trois-Ponts

Bronrome

SpaTuron

Amel

Comblain

Bomal

Soy

Statte(SNCB)

MiécretFlorée

SNCB

Ciney

Sart-Bernard(SNCB)

Dorinne

Yvoir(SNCB)

Warnant

Namur

SNCB

Marche-les-D.

Waret

Gerpinnes

Hanzinelle

Leuze

Fooz

Pepinster

Battice

Stembert

Welkenraedt (SNCB)

Les Plenesses

Henri-Chapelle

Montzen(SNCB)

Saives

Hogne(SNCB)Marche-en-

Famenne

CharneuxOn

Forrières(SNCB)

Dinant

Sommière

Hastière

Herfelingen

Appelterre

Welle EsseneDenderleeuw

(NMBS)

Fooz

Saives

CBR

Visé(SNCB)

Voroux(SNCB)

Hollogne

Alleur

Glain

FN

Montegnée

Tilleur

Ougr.Sclessin

Sart-Tilman

Angleur

Jemeppe

Profondval

Flémalle

Grivegnée

Chénée

Magotteaux

FN

Herstal

EheinHermalle

s/Huy

Abée-ScryPoulseur

Esneux

Rivage(SNCB)

Kalmthout

St-Niklaas

Hamme

Temse

St-Niklaas

Temse

BPCHM

OeverBelliard.

Hovenier.Tabak.

Moons.Berchem(NMBS)

WilrijkMHo

Hoboken

Aartselaar Kontich

Elouges Ciply

Pâturages

SNCB

Bois-de-Villers

St-Servais

Lens

Elouges

CiplyPâturages

Boussu

Maisières

Mons

Distrigaz

Piéton(SNCB)

Fontainel'Evêque Marchienne

Charleroi

Gilly

PosteEstFosses-la-Ville

Moustier

Keumiée

Sombreffe

Marbais(SNCB)

Chassart

Heppigniessud

Liberchies

Jumet

Seneffe

Hannut

70(1

50)

70(150)

70(150)70(220)

70(150)

70(150)

70(150)

70

10(7

0)

12(70)

15(70)

150+70

70(220)

70(220)

15(70)

12(70)

15(70)

70(150)

70(150)

70(1

50)

70(150)

(30)70+30(150)

70(150)

70(150)

70(150)

70(150)

70(150)

70(150)

30(150)

70(150)

70(150)

30(150)

70(1

50)

70(150)

70(150)

150+

70

70(150)

70(150)

70(1

50)

150+70

70(150)

70 (150)

70(150)

70+150

Anthisnes

Garnstock

Soiron

ZONHOVEN

Dowchemical

Ceroux

Stéphanshof

70(150)

Ramet

12(7

0)

36(1

50)

Wierde

Amylum

70(1

50)

REPPEL

Echelle Schaal

0 10 20 30km

1:1000000

1:500000

1:500000

1:500000

LAPRAYEFOUR

LA PRAYEFOUR

BERENDRECHTSLUIS.

BERENDRECHTSLUIS.

ANTWERPENZUID

ANTWERPENZUID

WALGOED

WALGOED

ESCH-SUR-ALZETTEBELVALSCHIFFLANGE

Schif.

Lummen

380+

150

LaLouvière

ROMSEE

VIGY

3

2

2

2

2

3

6

1

9

1

7

19

9

12

2

11

16

152

34

2

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 3 / 33

Page 7: Load flow PPT

Introduction Example

ExampleFile: case6 wh.m, from the book “computational methods for electricpower systems”, M. Crow.

Bus databus type Pd Qd Vm Va Vmax Vmin

1 3 25 10 1.05 0 1.05 1.052 2 15 5 1.05 0 1.05 1.053 2 27.5 11 1 0 1.05 0.954 1 0 0 1 0 1.05 0.955 1 15 9 1 0 1.05 0.956 1 25 15 1 0 1.05 0.95

Generator dataGen Pg Pq Pmax

1 0 0 2002 50 0 150

1#1

4 3

256

#2#3

#4

#5

#6

#7

G

G

Branch dataline from to R X B

1 1 4 0.020 0.185 0.0092 1 6 0.031 0.259 0.0103 2 3 0.006 0.025 0.0004 2 5 0.071 0.320 0.0155 4 6 0.024 0.204 0.0106 3 4 0.075 0.067 0.0007 5 6 0.025 0.150 0.017

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 4 / 33

Page 8: Load flow PPT

Introduction Example

ExampleFile: case6 wh.m, from the book “computational methods for electricpower systems”, M. Crow.

Bus databus type Pd Qd Vm Va Vmax Vmin

1 3 25 10 1.05 0 1.05 1.052 2 15 5 1.05 0 1.05 1.053 2 27.5 11 1 0 1.05 0.954 1 0 0 1 0 1.05 0.955 1 15 9 1 0 1.05 0.956 1 25 15 1 0 1.05 0.95

Generator dataGen Pg Pq Pmax

1 0 0 2002 50 0 150

1#1

4 3

256

#2#3

#4

#5

#6

#7

G

G

Branch dataline from to R X B

1 1 4 0.020 0.185 0.0092 1 6 0.031 0.259 0.0103 2 3 0.006 0.025 0.0004 2 5 0.071 0.320 0.0155 4 6 0.024 0.204 0.0106 3 4 0.075 0.067 0.0007 5 6 0.025 0.150 0.017

How would you solve this simple example by hand?

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 4 / 33

Page 9: Load flow PPT

System representation

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 5 / 33

Page 10: Load flow PPT

System representation

System representation

Most power systems are three phase AC

Normal power flow uses one phase equivalents

⇒ We only focus on this one today

One phase power flow only valid for balanced systems

Systems are usually given in per unit values

Lines can be represented by a π-equivalent

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 6 / 33

Page 11: Load flow PPT

System representation

Per-unit calculations

Normalized representation of the four basic properties: voltage,current, impedance and complex power

Of these, two can be chosen independently

Normally rated phase voltage and one phase rated power aretaken as basis

Upu = UUbase

and Spu = SSbase

Ibase = Sbase

Ubase

Zbase = Ubase

Ibaseor Zbase =

U2base

Sbase

Logical values: for a 11.8 kV , 60 MVA machine,Ubasis = 11.8√

3kV and Sbasis = 60

3 MVA

For a 400 kV line, with 100 MVA: Ubase = 400/√

3,

Sbase = 100/3 ⇒ Rbase =(

400√3

)2

· 3100 = 1600Ω

Why are voltage and complex power chosen as fixed values?

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 7 / 33

Page 12: Load flow PPT

System representation

Per-unit calculations

Normalized representation of the four basic properties: voltage,current, impedance and complex power

Of these, two can be chosen independently

Normally rated phase voltage and one phase rated power aretaken as basis

Upu = UUbase

and Spu = SSbase

Ibase = Sbase

Ubase

Zbase = Ubase

Ibaseor Zbase =

U2base

Sbase

Logical values: for a 11.8 kV , 60 MVA machine,Ubasis = 11.8√

3kV and Sbasis = 60

3 MVA

For a 400 kV line, with 100 MVA: Ubase = 400/√

3,

Sbase = 100/3 ⇒ Rbase =(

400√3

)2

· 3100 = 1600Ω

Why are voltage and complex power chosen as fixed values?

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 7 / 33

Page 13: Load flow PPT

System representation

Example per-unit

Generator example of before

Ubasis = 11.8√3kV and Sbasis = 60

3 MVA

Basis for current: Ibase = Sbase

Ubase= 60·

√3

11.8·3 = 2.9357 kA

Basis for impedance: Zbase =U2

base

Sbase=

(11.8√

3

)2

603

= 2.3207 Ω

Line connecting load: 0, 5 + 1 Ω = 0.21546 + 0.43091 pu

Afterwards, calculate using per-unit instead of original values

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 8 / 33

Page 14: Load flow PPT

System representation

Per-unit and transformers

Z1 Z2

n1 : n2

U2U1

Voltage at both sides of the transformer is different → differentbasisOne of the major advantages of per-unit calculations because ofsimplification

Z ′2 = Z2 ·(

n21

n22

)= Z2 ·

(U2

1

U22

)Zp = Z1 + Z ′2

Zs = Zp ·(

U22

U21

)and Zs(pu) = Zs

Zbase (sec)

The per-unit impedance is the same on both sides of thetransformer⇒ can be replaced by one series impedance!

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 9 / 33

Page 15: Load flow PPT

System representation

Per-unit and transformers

Z1n1 : n2

Z ′2

U2U1

Voltage at both sides of the transformer is different → differentbasisOne of the major advantages of per-unit calculations because ofsimplification

Z ′2 = Z2 ·(

n21

n22

)= Z2 ·

(U2

1

U22

)

Zp = Z1 + Z ′2

Zs = Zp ·(

U22

U21

)and Zs(pu) = Zs

Zbase (sec)

The per-unit impedance is the same on both sides of thetransformer⇒ can be replaced by one series impedance!

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 9 / 33

Page 16: Load flow PPT

System representation

Per-unit and transformers

Zp

n1 : n2

U1 U2

Voltage at both sides of the transformer is different → differentbasisOne of the major advantages of per-unit calculations because ofsimplification

Z ′2 = Z2 ·(

n21

n22

)= Z2 ·

(U2

1

U22

)Zp = Z1 + Z ′2

Zs = Zp ·(

U22

U21

)and Zs(pu) = Zs

Zbase (sec)

The per-unit impedance is the same on both sides of thetransformer⇒ can be replaced by one series impedance!

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 9 / 33

Page 17: Load flow PPT

System representation

Per-unit and transformers

Zs

n1 : n2

U2U1

Voltage at both sides of the transformer is different → differentbasisOne of the major advantages of per-unit calculations because ofsimplification

Z ′2 = Z2 ·(

n21

n22

)= Z2 ·

(U2

1

U22

)Zp = Z1 + Z ′2

Zs = Zp ·(

U22

U21

)and Zs(pu) = Zs

Zbase (sec)

The per-unit impedance is the same on both sides of thetransformer⇒ can be replaced by one series impedance!

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 9 / 33

Page 18: Load flow PPT

System representation

Representation of a transmission line

B2

X

B2

G2

G2

R

π-equivalent

Valid for lines up to 240 km

All values are normally small

Other equivalents exist and are sometimes used in practice

Normally, G can be neglected

With overhead lines, B can be neglected as well, for cables this isnot the case (see chapters on lines and cables)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 10 / 33

Page 19: Load flow PPT

The load flow problem

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 11 / 33

Page 20: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0

# lines (branches) × # nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 21: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0line # 2 -1 0 1 0

# lines (branches) × # nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 22: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0line # 2 -1 0 1 0line # 3 0 -1 1 0

# lines (branches) × # nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 23: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0line # 2 -1 0 1 0line # 3 0 -1 1 0line # 4 0 -1 0 1

# lines (branches) × # nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 24: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0line # 2 -1 0 1 0line # 3 0 -1 1 0line # 4 0 -1 0 1line # 5 0 0 -1 1

# lines (branches) × # nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 25: Load flow PPT

The load flow problem

A bit of algebra: incidence matrixDirected Graph

Incidence matrix (A0)

nodesline # 1 -1 1 0 0line # 2 -1 0 1 0line # 3 0 -1 1 0line # 4 0 -1 0 1line # 5 0 0 -1 1

# lines (branches) × # nodesthe columns are dependent

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 12 / 33

Page 26: Load flow PPT

The load flow problem

Incidence matrix

Incidence matrix is written as A0

Meaning of the incidence matrix

Describes the directed graph

Produces differences

−1 1 0 0−1 0 1 00 −1 1 0−1 0 0 10 −1 0 1

·U1

U2

U3

U4

=

U2 − U1

U3 − U1

U3 − U2

U4 − U1

U4 − U2

(1)

Some symbols

Ii : Current injected atnode i

Iij : Current from node ito node j

Ui : Potential of node i

Eij : Potential difference(voltage) betweennodes i and j

Cij : Conductance of theline between nodes iand j

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 13 / 33

Page 27: Load flow PPT

The load flow problem

Incidence matrix

Incidence matrix is written as A0

Meaning of the incidence matrix

Describes the directed graph

Produces differences

−1 1 0 0−1 0 1 00 −1 1 0−1 0 0 10 −1 0 1

·U1

U2

U3

U4

=

U2 − U1

U3 − U1

U3 − U2

U4 − U1

U4 − U2

(1)

Setting U4 = 0

Resulting matrix is the incidence matrix: A

Some symbols

Ii : Current injected atnode i

Iij : Current from node ito node j

Ui : Potential of node i

Eij : Potential difference(voltage) betweennodes i and j

Cij : Conductance of theline between nodes iand j

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 13 / 33

Page 28: Load flow PPT

The load flow problem

Incidence matrix

Incidence matrix is written as A0

Meaning of the incidence matrix

Describes the directed graph

Produces differences

−1 1 0−1 0 10 −1 1−1 0 00 −1 0

·U1

U2

U3

=

U2 − U1

U3 − U1

U3 − U2

− U1

− U2

(1)

Setting U4 = 0

Resulting matrix is the incidence matrix: A

Some symbols

Ii : Current injected atnode i

Iij : Current from node ito node j

Ui : Potential of node i

Eij : Potential difference(voltage) betweennodes i and j

Cij : Conductance of theline between nodes iand j

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 13 / 33

Page 29: Load flow PPT

The load flow problem

Meaning of the incidence matrix1 Incidence matrix is A0 with one node removed

(grounded, reference)2 Ii is a nodal current injection, Iij is a branch flowI1I2

I3

=

−1 1 0−1 0 10 −1 10 −1 00 0 −1

T

·

I12

I13

I23

I24

I34

3 Ui is a nodal voltage/potential, Eij represents a

potential drop over line ij4 The relation between the voltage difference (e)

and line flows (f): Ohms law (take Cij theconductance of i to j)

5 Link on youtube

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 14 / 33

Page 30: Load flow PPT

The load flow problem

Meaning of the incidence matrix1 Incidence matrix is A0 with one node removed

(grounded, reference)2 Ii is a nodal current injection, Iij is a branch flow3 Ui is a nodal voltage/potential, Eij represents a

potential drop over line ijE12

E13

E23

E24

E34

=

−1 1 0−1 0 10 −1 10 −1 00 0 −1

·U1

U2

U3

4 The relation between the voltage difference (e)

and line flows (f): Ohms law (take Cij theconductance of i to j)

5 Link on youtube

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 14 / 33

Page 31: Load flow PPT

The load flow problem

Meaning of the incidence matrix1 Incidence matrix is A0 with one node removed

(grounded, reference)2 Ii is a nodal current injection, Iij is a branch flow3 Ui is a nodal voltage/potential, Eij represents a

potential drop over line ij4 The relation between the voltage difference (e)

and line flows (f): Ohms law (take Cij theconductance of i to j)I12

I13

I23

I24

I34

=

C12 0 0 0 00 C13 0 0 00 0 C23 0 00 0 0 C24 00 0 0 0 C34

·E12

E13

E23

E24

E34

5 Link on youtube

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 14 / 33

Page 32: Load flow PPT

The load flow problem

Putting it together

AT · C · A · U = I

Ybus · U = I

Ybus is the bus admittance matrix

Representation of the entirenetwork by an admittance matrix, avector of nodal voltages and avector of nodal current injections

Yij = −yij (admittance betweennode i and j)

Yii =∑n

j yij (sum of the rest of therow + yii , the impedance to thereference)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 15 / 33

Page 33: Load flow PPT

The load flow problem

The power system represented

The power system consists of:

Generators: delivering P and Q

Loads: consuming P and Q

Lines or branches: connecting generation and load

Wanted: Power flow of P and Q through these lines

Nodes or busbars: connections points in the power system

Wanted: Voltage amplitude (U) and voltage angle (θ) at eachnode

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 16 / 33

Page 34: Load flow PPT

The load flow problem

The power system represented

The power system consists of:

Generators: delivering P and Q

Loads: consuming P and Q

Lines or branches: connecting generation and load

Wanted: Power flow of P and Q through these lines

Nodes or busbars: connections points in the power system

Wanted: Voltage amplitude (U) and voltage angle (θ) at eachnode

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 16 / 33

Page 35: Load flow PPT

The load flow problem

Mathematical statement of the problem

Uc∠θcUa∠θaUb∠θb

Ia

IbIbc

Iac

Icybc

yabyacc©

Iab

Neutral

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 17 / 33

Page 36: Load flow PPT

The load flow problem

Mathematical statement of the problem

Uc∠θcUa∠θaUb∠θb

Ia

IbIbc

Iac

Icybc

yabyacc©

Iab

Neutral

Voltage of node i to neutral is Ui∠θiAdmittance between i and j is yij

Current from i to j is IijThe injected current at i is Ii

Ia = Iab + IacIa = (Ua − Ub) · yab + (Ua − Uc) · yacIa = Ua · (yab + yac)− Ub · yab − Uc · yac

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 17 / 33

Page 37: Load flow PPT

The load flow problem

Mathematical statement of the problem

Last equation repeated: Ia = Ua · (yab + yac)− Ub · yab − Uc · yacWe take Yaa = yab + yac = yaa +

∑ni 6=a yai

yaa = ya0 = the parallel branches to node a (in this example,yaa = 0)

We take Yai = −yai⇒ as with Ybus , the bus admittance matrix

Which results in:Ia = Yaa · Ua + Yab · Ub + Yac · Uc (2)

Or for the entire system: IaIbIc

=

Yaa Yab Yac

Yba Ybb Ybc

Yca Ycb Ycc

· Ua

Ub

Uc

(3)

or I = Ybus ·U (4)

and Yij = Yji in symmetrical systems

(e.g. not with phase shifting transformers)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 17 / 33

Page 38: Load flow PPT

The load flow problem

Mathematical statement of the problem

Resulting equations for a general system with n nodes

Ii =n∑

j=1

Yij · Uj ∀i ∈ N ≤ n (5)

S∗i = U∗i · Ii (6)

Above equations form the basis of power flow

There are 4 basic quantities for each node in power flowcalculations:

Voltage amplitude |U|Voltage angle θ between the voltage vector and the voltagereferenceActive power injection, withdrawal at a nodeReactive power injection, withdrawal at a node

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 17 / 33

Page 39: Load flow PPT

The load flow problem

Mathematical statement of the problem

Where do shunt elements fit? IaIbIc

=

Yaa Yab Yac

Yba Ybb Ybc

Yca Ycb Ycc

· Ua

Ub

Uc

(7)

Yii =n∑

j=1

yij (8)

=n∑

j=1j 6=i

−Yij + yii (9)

yii is the term to the node that has been “grounded”

In practice: shunt elements

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 17 / 33

Page 40: Load flow PPT

The load flow problem

Types of nodes

Three distinct types of nodes (important)

PV bus: A generating source is connected to the bus; the nodalvoltage is controlled at a certain magnitude U by injectingor absorbing reactive energy. The generated power PG isset at a specified value. θ and QG are computed. Constantvoltage operation is only possible when the generator iswithin its reactive energy generation limits.

PQ bus: P and Q are the control variables. This is the case whenthere is only a load connected to the bus or the generator isoutside its reactive power limits.

Slack (swing) bus: one of the generator busses is chosen to be the slackbus where the nodal voltage magnitude, Uslack , and phaseangle θslack are specified. This bus is needed to provide a“compensation” for the electrical losses that are not knownin advance. The bus forms a reference for the voltage angle.

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 18 / 33

Page 41: Load flow PPT

The load flow problem

Types of nodes

Three distinct types of nodes (important)

PV bus: A generating source is connected to the bus; the nodalvoltage is controlled at a certain magnitude U by injectingor absorbing reactive energy. The generated power PG isset at a specified value. θ and QG are computed. Constantvoltage operation is only possible when the generator iswithin its reactive energy generation limits.

PQ bus: P and Q are the control variables. This is the case whenthere is only a load connected to the bus or the generator isoutside its reactive power limits.

Slack (swing) bus: one of the generator busses is chosen to be the slackbus where the nodal voltage magnitude, Uslack , and phaseangle θslack are specified. This bus is needed to provide a“compensation” for the electrical losses that are not knownin advance. The bus forms a reference for the voltage angle.

What is the mathematical meaning of the slack bus?

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 18 / 33

Page 42: Load flow PPT

The load flow problem

Types of nodes

Three distinct types of nodes (important)

PV bus: A generating source is connected to the bus; the nodalvoltage is controlled at a certain magnitude U by injectingor absorbing reactive energy. The generated power PG isset at a specified value. θ and QG are computed. Constantvoltage operation is only possible when the generator iswithin its reactive energy generation limits.

PQ bus: P and Q are the control variables. This is the case whenthere is only a load connected to the bus or the generator isoutside its reactive power limits.

Slack (swing) bus: one of the generator busses is chosen to be the slackbus where the nodal voltage magnitude, Uslack , and phaseangle θslack are specified. This bus is needed to provide a“compensation” for the electrical losses that are not knownin advance. The bus forms a reference for the voltage angle.

Ii =∑n

j=1 Yij · Uj ∀i ∈ N ≤ n and i 6= nslack

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 18 / 33

Page 43: Load flow PPT

Solving the problem

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 19 / 33

Page 44: Load flow PPT

Solving the problem

The problemKnown data:

Active power injections in my system at generator nodesVoltages at generator nodesActive and reactive withdrawals (load) at PQ nodesSlack node voltage and angleImpedances (Ybus)

Unknowns:

Rest of P (slack), Q (slack and PV), voltage amplitude (PQnodes) and voltage angle (all but slack)

Equations

I = Y · U (10)

S∗ = U∗ · I (11)

S∗ = U∗ · Y · U (12)

P − Q = U∗ · Y · U (13)

Pi − Qi = U∗i ·n∑

j=1

Yij · Uj (14)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 20 / 33

Page 45: Load flow PPT

Solving the problem Gauss-Seidel

Gauss-Seidel

Algorithm

Ii =∑n

j=1 Yij · Uj and S∗i = U∗i · Ii give:

Ui =1

Yii·

S∗iU∗i−

n∑j=1j 6=i

Yij · Uj

(15)

This is solved bus by bus, and solutions of previous calculationsare filled in directly

U(i+1)i =

1

Yii·

S∗i

U∗(i)i

−i∑

j=1j 6=i

Yij · U(i+1)j −

n∑j=i+1j 6=i

Yij · U(i)j

(16)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 21 / 33

Page 46: Load flow PPT

Solving the problem Gauss-Seidel

Gauss-Seidel

1 For busbar 21, calculate I2 =S∗2U∗2

2 Calculate∑n

j=1j 6=2

Y2j · Uj

3 Subtract solution 2 from solution 1 and divide the result by Y22

to obtain a new value for U2

4 For busbar 3, calculate I3 =S∗3U∗3

5 Using the new value of U2 of step 3, calculate∑n

j=1j 6=3

Y3j · Uj

6 Subtract solution 5 from solution 4 and divide the result by Y33

to obtain a new value for U3

7 Repeat for all busses

8 Compare latest set of voltages with previous and check tolerance:U(i+1) −U(i) < ε? If not, go to step 1.

1when 1 is the reference busDVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 21 / 33

Page 47: Load flow PPT

Solving the problem Gauss-Seidel

Gauss-Seidel

Convergence and acceleration

The Gauss-Seidel method converges linearly (slow) with systemsize

Each iteration itself requires limited processing power

Often, the method is corrected with an acceleration factorU

(new)i(acc) = α · U(new)

i − (α− 1) · U(old)i (17)

= α · U(new)i − α · U(old)

i + U(old)i (18)

= U(old)i + α · (U(new)

i − U(old)i ) (19)

1 < α < 2

For large systems, often a value of 1.6 is chosen

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 21 / 33

Page 48: Load flow PPT

Solving the problem Gauss-Seidel

Gauss-Seidel

Gauss-Seidel properties

A starting vector must be chosen

Often, the starting voltages are set to 1∠0 pucalled “Flat start”

If the voltages are calculated in block (and not replaced after onehas calculated the former one), we call the method the Jacobimethod.

The Jacobi method has a slower convergence

The Gauss-Seidel method is not often used anymore

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 21 / 33

Page 49: Load flow PPT

Solving the problem Gauss-Seidel

Gauss-Seidel

Gauss-Seidel properties

A starting vector must be chosen

Often, the starting voltages are set to 1∠0 pucalled “Flat start”

If the voltages are calculated in block (and not replaced after onehas calculated the former one), we call the method the Jacobimethod.

The Jacobi method has a slower convergence

The Gauss-Seidel method is not often used anymore

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 21 / 33

Page 50: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution of equation y = f (x)

Taylor series expansion:

y = f [x(0)]+

df

dx

∣∣∣∣x=x(0)

1!·[x−x(0)]+

d2f

dx2

∣∣∣∣x=x(0)

2!·[x−x(0)]2+. . . (20)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 51: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution of equation y = f (x)

Taylor series expansion:

y = f [x(0)]+

df

dx

∣∣∣∣x=x(0)

1!·[x−x(0)]+

d2f

dx2

∣∣∣∣x=x(0)

2!·[x−x(0)]2+. . . (20)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 52: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution of equation y = f (x)

Taylor series expansion:

y = f [x(0)] +

df

dx

∣∣∣∣x=x(0)

1!· [x − x(0)] (20)

Solving this for x :

x = x(0) +1

df

dx

∣∣∣∣x=x(0)

· [y − f (x(0))] (21)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 53: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 54: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 55: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 56: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

(U , θ)1

f (U , θ)1

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 57: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

(U , θ)2

f (U , θ)2

(U , θ)1

f (U , θ)1

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 58: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution for multivariable nonlinear equations

y1 = f1(x1, x2, . . . , xn)

y2 = f2(x1, x2, . . . , xn)

...

y3 = fn(x1, x2, . . . , xn)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 59: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution for multivariable nonlinear equations

y1

y2

...yn

=

f1(x1(0), x2(0), . . . , xn(0))f2(x1(0), x2(0), . . . , xn(0))

...fn(x1(0), x2(0), . . . , xn(0))

+

∂f1

∂x1

∂f1

∂x2· · · ∂f1

∂xn∂f2

∂x1

∂f2

∂x2· · · ∂f2

∂xn...

.... . .

...∂fn∂x1

∂fn∂x2

· · · ∂fn∂xn

·

x1 − x1(0)x2 − x2(0)

...xn − xn(0)

(22)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 60: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson method: theory

Solution for multivariable nonlinear equations

Summarized, we can write the following:

y = f [x(0)] + J(0) · [x− x(0)] (23)

or solving for x:

x = x(0) + J(0)−1 · [y − f (x(0))] (24)

or in its recursive form:

xi+1 = xi + J−1i · [y − f (xi )] (25)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 22 / 33

Page 61: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Remember

Ii =∑n

j=1 Yij · Uj

S∗i = U∗i · Ii

Equivalents

Si = f (Ui ) is equivalent to y = f (x)

Ui+1 = Ui + Ji · [S− f (Ui )] (26)

S is here the specified complex power at any busbar

f (Ui ) is here the specified complex power at any busbar

∆Si = Ji ·Ui+1 (27)

You normally know the active and reactive power injections ineach node (load and generation)

You want to know the complex voltages at the nodes

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 62: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

S∗i = U∗i∑n

j=1 Yij · Uj and i = 1, . . . , n

Newton-Raphson in rectangular coordinates

Pi = Ui

n∑j=1

Uj · (Gij · cos(θi − θj) + Bij · sin(θi − θj)) (28)

Qi = −Ui

n∑j=1

Uj · (Gij · sin(θi − θj)− Bij · cos(θi − θj)) (29)

Newton-Raphson in polar coordinates

Pi = Ui

n∑j=1

Uj · Yij · cos(θi − θj − φij) (30)

Qi = −Ui

n∑j=1

Uj · Yij · sin(θi − θj − φij) (31)

note: Yij = Gij + · Bij = |Yij |∠φij

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 63: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

∆Pi = Pi, scheduled − Pi, calc ∀PQ and PV (32)

∆Qi = Qi, scheduled − Qi, calc ∀PQ (33)

Pi, scheduled and Qi, scheduled are known from the input data

Pi, calc and Qi, calc are obtained from the calculation inrectangular or polar coordinates

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 64: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Putting it in one equation

Writing the power flow equations (both rectangular and polar) inthe form of equation (25):

[∆P∆Q

](i)

= −

∂P

∂θ

∂P

∂UU

∂Q

∂θ

∂Q

∂UU

(i)

︸ ︷︷ ︸J((U,θ)(i−1))

·[

∆θ∆UU

](i)

(34)

Or written in a simplified form:[∆P∆Q

](i)

= −[

H NM L

](i)

︸ ︷︷ ︸J((U,θ)(i−1))

·[

∆θ∆UU

](i)

(35)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 65: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Equation explained

∆θi = θ(i+1)i − θ(i)

i

∆Ui = U(i+1)i − U

(i)i

Voltages and angles (i + 1) are updated after each iteration andused for the following step

J is the Jacobian, and forms the derivative (tangent, gradient) ofthe power flow equations

∂P

∂UU and

∂Q

∂UU simplify the equations and results in fewer

computations

There are n − 1 equations for ∆P

There are n −#pv − 1 equations for ∆Q

The Jacobian is a square matrix(2 · n −#pv − 2)× (2 · n −#pv − 2)

The Jacobian is a sparse matrix (Special techniques can be usedwhen numerical calculations are performed)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 66: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Equation explained

∆θi = θ(i+1)i − θ(i)

i

∆Ui = U(i+1)i − U

(i)i

Voltages and angles (i + 1) are updated after each iteration andused for the following step

J is the Jacobian, and forms the derivative (tangent, gradient) ofthe power flow equations

∂P

∂UU and

∂Q

∂UU simplify the equations and results in fewer

computations

There are n − 1 equations for ∆P ⇒ Why?

There are n −#pv − 1 equations for ∆Q ⇒ Why?

The Jacobian is a square matrix(2 · n −#pv − 2)× (2 · n −#pv − 2)

The Jacobian is a sparse matrix (Special techniques can be usedwhen numerical calculations are performed)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 67: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Full equations (polar form): off-diagonal∂∆Pi

∂θj= Hij = −Uj · Ui · Yij · sin(θi − θj − φij)

∂∆Pi

∂UjUj = Nij = −Uj · Ui · Yij · cos(θi − θj − φij)

∂∆Qi

∂θj= Mij = Uj · Ui · Yij · cos(θi − θj − φij)

∂∆Qi

∂UjUj = Lij = −Uj · Ui · Yij · sin(θi − θj − φij)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 68: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Full equations (polar form): diagonal∂∆Pi

∂θi= Hii = Ui ·

n∑j=1

UjYij sin(θi − θj − φij) + U2i · Yii · sin(φij)

∂∆Pi

∂UiUi = Nii = −Ui ·

n∑j=1

UjYij cos(θi − θj − φij)− U2i · Yii · cos(φij)

∂∆Qi

∂θi= Mii = −Ui ·

n∑j=1

UjYij cos(θi − θj − φij) + U2i · Yii · cos(φij)

∂∆Qi

∂UiUi = Lii = −Ui ·

n∑j=1

UjYij sin(θi − θj − φij) + U2i · Yii · sin(φij)

Notice the symmetry

Notice that the off-diagonal elements are also in the diagonalelements

M = −N for off-diagonal elements

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 69: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Figure: Sparsity of the Jacobian matrix shown

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 70: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

Numerical aspects

Iterative process until mismatch is below threshold(max(∆P(i); ∆Q(i)) = ε < εlimit)

Quadratic convergence

Major computational effort is calculating the inverse of theJacobian

The Jacobian is sparse, so special techniques can be used (lessstorage)

Ordering schemes can increase speed

Convergence is not guaranteed

A good starting point is needed

Flat start?Previous outcomeDC load flow as starting point

Simplifications exist

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 71: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 72: Load flow PPT

Solving the problem Newton-Raphson power flow

Newton-Raphson and load flow

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 23 / 33

Page 73: Load flow PPT

Solving the problem Simplified Newton-Raphson

Decoupled load flow

In a power system with mostly inductive lines, the power flowequations can be decoupled. (φij ≈ 90)

Active power is related to the angle between nodesReactive power is related to the voltage[

∆P∆Q

](i)

= −[

H NM L

](i)

︸ ︷︷ ︸J((U,θ)(i−1))

·[

∆θ∆UU

](i)

(36)

Advantages and disadvantages

+ Two small inverses instead of one big

+ Faster as only 2 · n3 calculations are needed, and not(2 · n)3 = 8 · n3

- The two subsystems may converge differently

- Convergence rate is slightly reduced

- Not often used nowadays

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 24 / 33

Page 74: Load flow PPT

Solving the problem Simplified Newton-Raphson

Decoupled load flow

In a power system with mostly inductive lines, the power flowequations can be decoupled. (φij ≈ 90)

Active power is related to the angle between nodesReactive power is related to the voltage[

∆P∆Q

](i)

= −[

H 00 L

](i)

︸ ︷︷ ︸J((U,θ)(i−1))

·[

∆θ∆UU

](i)

(37)

Advantages and disadvantages

+ Two small inverses instead of one big

+ Faster as only 2 · n3 calculations are needed, and not(2 · n)3 = 8 · n3

- The two subsystems may converge differently

- Convergence rate is slightly reduced

- Not often used nowadays

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 24 / 33

Page 75: Load flow PPT

Solving the problem Simplified Newton-Raphson

Decoupled load flow

Approximation

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 24 / 33

Page 76: Load flow PPT

Solving the problem Simplified Newton-Raphson

Fast decoupled load flow

In decoupled load flow, a new reduced Jacobian is determinedduring each iteration

Of each new Jacobian, the inverse needs to be calculated

Fast decoupled does not calculate a new Jacobian for eachiteration [

∆P(i)]

= [B′] ·[∆θ(i+1)

](38)[

∆Q(i)

U

]= [B′′] ·

[∆Ui+1

](39)

B′ and B′′ are real, sparse and constant matrices

Only series elements are involved (no shunts)

If the system has high R/X -ratio, large voltage angle deviationsor voltages which seriously differ from 1 pu, convergenceproblems can arise

Slower convergence (more iterations) but each iteration is muchfaster

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 25 / 33

Page 77: Load flow PPT

Solving the problem Simplified Newton-Raphson

Fast decoupled load flow

Approximation

derivative

f (U , θ)0

(U , θ)0

(U , θ)∗

f (U , θ)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 25 / 33

Page 78: Load flow PPT

Solving the problem Simplified Newton-Raphson

DC load flow

If we consider the system to be lossless (Y = B)

And voltages to be around 1.0 pu (∆U = 0)

Voltage angles between busses are small (sin(θi − θj) ≈ (θi − θj)and cos(θi − θj) ≈ 1)

One equation of Newton-Raphson:

∆P = Ui

n∑j=1

Uj · (Gij · cos(θi − θj) + Bij · sin(θi − θj)) (40)

[∆P] = [B′] · [∆δ] (41)

B′ is real

Linear system

One calculation, no iterations

Easy for optimizations

Not correct (approximation)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 26 / 33

Page 79: Load flow PPT

Software

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 27 / 33

Page 80: Load flow PPT

Software

Power flow software

Demo’s + try at home

Free available (open source)

Matpower (matlab based):http://www.pserc.cornell.edu/matpower/

PSAT (matlab based):http://www.power.uwaterloo.ca/~fmilano/psat.htm

InterPSS (Java based): http://www.interpss.org/

Professional software

PSS/EEurostagDigSilentPowerworld (demo athttp://www.powerworld.com/downloads.asp)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 28 / 33

Page 81: Load flow PPT

State estimation

Outline

1 IntroductionExample

2 System representation

3 The load flow problem

4 Solving the problemGauss-SeidelNewton-Raphson power flowSimplified Newton-Raphson

5 Software

6 State estimation

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 29 / 33

Page 82: Load flow PPT

State estimation

State estimation

Known and unknown variables in the real power system

Lines, cables, transformers, location of generation and load ⇒ allknown and constant in time

Voltages, currents, actual generation and load (at that moment),position of power switches, tap-changer settings,. . .⇒ mostlyunknown or variable

Measurements:

P, Q: Generation and load, some linesVoltage: |U| every substation. θ only with PMU (phasormeasurement unit)Tap-changer settingsIncompleteMeasurement errors

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 30 / 33

Page 83: Load flow PPT

State estimation

State estimation

State estimation: what?Monitoring or supplementing data for load flow

Many measurements in the system

Determining measurement errors, estimate and (statistically)analyze

If needed, certain measurements should be rejected

Least Squares approach

Another Youtube video: least squares

Has to be solved iteratively for power systems

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 30 / 33

Page 84: Load flow PPT

State estimation

State estimation

Weighted least-square method

(measurements z , with errors e, h(x) is the true model of the state x)

z =

z1

z2

z3

z4

=

h1(x1, x2, . . . , xn)h2(x1, x2, . . . , xn)h3(x1, x2, . . . , xn)h4(x1, x2, . . . , xn)

+

e1

e2

e3

e4

= h(x) + e (42)

With errors having a zero average, and each independent we get acovariance matrix R:

R =

σ2

1 0 · · · 00 σ2

2 · · · 0...

.... . .

...0 0 · · · σ2

n

(43)

R is the inverse of what we could call the weighting matrixR = inv(W)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 30 / 33

Page 85: Load flow PPT

State estimation

State estimation

Solving the state estimation

The expected values are:

x =

x1

x2

. . .xn

=

HT ·W ·H︸ ︷︷ ︸G

−1

·HT ·W · z = G−1 ·HT ·W · z

(44)

x = G−1 ·HT ·W · (H · x + e) (45)

x = G−1 · (HT ·W ·H)︸ ︷︷ ︸G

·x + G−1 ·HT ·W · e (46)

z = H · x (47)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 30 / 33

Page 86: Load flow PPT

State estimation

State estimation: simple example

Figure: Example network

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 87: Load flow PPT

State estimation

State estimation: simple example

We want to know x1 and x2, which are voltages U1 and U2

Two amperemeters measuring z1 = 9.01 A and z2 = 3.02 A

U1 = 16.0233 V and U2 = 8.0367 V

Two voltmeters measuring z3 = 6.98 V and z4 = 5.01 V

U1 = 15.93 V and U2 = 8.05 V

The system equations can be written as:

z1

z2

z3

z4︸︷︷︸measurements

====

58 · x1 − 1

8 · x2

− 18 · x1 + 5

8 · x238 · x1 + 1

8 · x218 · x1 + 3

8 · x2︸ ︷︷ ︸true values from model

++++

e1

e2

e3

e4︸︷︷︸errors

(48)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 88: Load flow PPT

State estimation

State estimation: simple example

We want to know x1 and x2, which are voltages U1 and U2

Two amperemeters measuring z1 = 9.01 A and z2 = 3.02 A

U1 = 16.0233 V and U2 = 8.0367 V ⇒ Conflict

Two voltmeters measuring z3 = 6.98 V and z4 = 5.01 V

U1 = 15.93 V and U2 = 8.05 V ⇒ Conflict

The system equations can be written as:

z1

z2

z3

z4︸︷︷︸measurements

====

58 · x1 − 1

8 · x2

− 18 · x1 + 5

8 · x238 · x1 + 1

8 · x218 · x1 + 3

8 · x2︸ ︷︷ ︸true values from model

++++

e1

e2

e3

e4︸︷︷︸errors

(48)

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 89: Load flow PPT

State estimation

State estimation: simple example

Calculating the expected values x

We take the following weighting matrix (1/sigma):W = diag([100, 100, 50, 50])

The most probable values for U1 and U2 are 16.00719 and8.02614 resp.

The expected error will be: e =

0.008770.00456−0.02596−0.00070

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 90: Load flow PPT

State estimation

State estimation: simple example

Measurement 4 changes

z4 = 4.4 instead of z4 = 5.01

The best estimate for the voltages: U1 = 15.86807 andU2 = 7.75860

In that case, the expected error will be: e =

0.062280.154380.05964−0.49298

When the expected error is too high, measurements can/shouldbe disregarded

Statistical test are needed to determine when errors are “high”

The weight matrix also has a serious influence on the results

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 91: Load flow PPT

State estimation

State estimation: simple example

Measurement 4 changes

z4 = 4.4 instead of z4 = 5.01

The best estimate for the voltages: U1 = 15.86807 andU2 = 7.75860

In that case, the expected error will be: e =

0.062280.154380.05964−0.49298

When the expected error is too high, measurements can/shouldbe disregarded

Statistical test are needed to determine when errors are “high”

The weight matrix also has a serious influence on the results

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 31 / 33

Page 92: Load flow PPT

State estimation

State estimation for power flow calculations

State estimator calculates voltage magnitudes and relative phaseangles of the system buses

Redundancy in input data

With errors on all measurement data

Non-DC circuit ⇒ non-linear equations: h = h(x)

Iterative solutions (as in the Newton-Raphson method) areneeded

The principle is the same

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 32 / 33

Page 93: Load flow PPT

State estimation

References

Power System Analysis; Grainger, John J. and Stevenson,William D., Jr.

Computational Mehods for Electric Power Systems; Crow,Mariesa

Power System Load Flow Analysis; Powell, Lynn

DVH, HE, PGT (KUL/ESAT/ELECTA) Power flow calculations September 19, 2011 33 / 33