This action might not be possible to undo. Are you sure you want to continue?
Stability analysis of transmission systems with
high penetration of distributed generation
PROEFSCHRIFT
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magniﬁcus prof.dr.ir. J.T. Fokkema,
voorzitter van het College voor Promoties, in het openbaar te verdedigen op
donderdag 21 december 2006 om 10:00 uur
door
Muhamad REZA
elektrotechnisch ingenieur
geboren te Bandung, Indonesi¨e.
Dit proefschrift is goedgekeurd door de promotoren:
Prof.ir. W. L. Kling
Prof.ir. L. van der Sluis
Samenstelling promotiecommissie:
Rector Magniﬁcus, voorzitter
Prof.ir. W. L. Kling, Technische Universiteit Delft, promotor
Prof.ir. L. van der Sluis, Technische Universiteit Delft, promotor
Prof.dr. J. A. Ferreira Technische Universiteit Delft
Prof.dr.ir. J. H. Blom Technische Universiteit Eindhoven
Prof.ir. M. Antal Technische Universiteit Eindhoven (emeritus)
Prof.dr.ir R. Belmans Katholieke Universiteit Leuven, Belgi¨e
Prof.dr. M. J. O’Malley University College Dublin, Ierland
This research has been performed within the framework of the research program
’Intelligent Power Systems’ that is supported ﬁnancially by SenterNovem, an
agency of the Dutch Ministry of Economic Aﬀairs.
Stability analysis of transmission systems with high penetration of distributed
generation.
Dissertation at Delft University of Technology.
Copyright c 2006 by M. Reza.
ISBN: 9162870394
Cover: A miniature of transmission lines tower in Madurodam, The Hague,
photographed and modiﬁed by Muhamad Reza, and used for this thesis with
permission from Madurodam B.V.
To Bapak, Ibu, Novi and Rifqi.
Summary
Stability analysis of transmission systems with
high penetration of distributed generation
Nowadays, interest in generating electricity using decentralized generators of
relatively small scale is increasing. Such generation is known as ’distributed
generation’ (DG). Many of the prime movers of such DG technologies are based
on renewable energy sources resulting in an environmentallyfriendly power gen
eration.
It is wellknown that the implementation of DG inﬂuences the technical
aspects of the distribution grids. The impact of a small amount of DG connected
to the grid on the power system transient stability has not been treated so often.
When the penetration level of DG increases, its impact is no longer restricted
to the distribution network but begins to inﬂuence the whole system.
This work deals with the impact of implementing DG on the transmission
system transient stability, with the emphasis on a potential transition from a
’vertical power system’ to a ’horizontal power system’ (Chapter 1).
For this purpose, it is important to examine characteristics of DG that in
ﬂuence the dynamic stability behavior of a transmission system (Chapter 2).
Therefore, DG units are classiﬁed based on the primary (both conventional and
renewable) energy sources. To a large extent, the type of primary energy source
determines the output power characteristics of DG and the type of grid connec
tion applied. It also determines the utilization of power electronic interfaces.
Based on the DG classiﬁcation, basic models of DG technology to be used in the
transient stability simulation of a large power system can be derived, presented
in Chapter 3.
A problem in power systems is maintaining synchronous operation of all
(centralized) synchronous machines. The stability problem associated is called
rotor angle stability. In this work, the impact of the DG implementation on this
is investigated. Therefore, in Chapter 3, the phenomena of the rotor dynamics
of synchronous machines, that determine the rotor angle stability of a power
system, are explained by means of the swing equation, the powerangle curve
and the equal area criterion concepts. Indicators for assessing the stability
performance of a power system derived from these concepts are the maximum
ii Summary
rotor speed deviation and oscillation duration of the (centralized) synchronous
machines. For simulation purposes, the widely known 39bus New England
dynamic test system is used with minor adjustments. The basic setup of the
test system is described, while details are listed in Appendices B and C. Several
software packages suitable for the transient dynamic simulation of DG in a large
system are highlighted. The representations of diﬀerent DG technologies within
the software packages are discussed, with the emphasis on the representation of
power electronic interfaced (converter connected) DG units.
The investigation of the impact of a high DG (penetration) level on power
system transient stability, is presented in Chapter 4. The impact of increasing
DG penetration levels, DG gridconnectionstrength, diﬀerent DG technologies,
and DG protection schemes of converterconnected DG are simulated and dis
cussed. It is found that DG inﬂuences the system transient stability diﬀerently
depending on the factors above. However, there is no signiﬁcant stability prob
lem observed up to about 30% DG penetration level regardless the technology.
This is logical when all centralized generators remain in the system – as well
as their active and their reactive power control and the inertia of their rotating
masses – along with the increasing DG levels. Furthermore, implementing DG
is a natural way of ‘limiting’ the power ﬂows on the transmission lines. It im
proves the transient stability of a transmission system, since large power ﬂows
may have a detrimental eﬀect on the damping of the oscillations: the heavier the
lines are loaded, the weaker the coupling between generators and loads becomes,
and the larger the oscillations of the centralized generators may be, especially
with long lines.
In Chapter 5, the investigation is focused on the impact of DG levels on
the system transient stability when the increasing DG level is followed by a
reduction of centralized generators in service resulting in a ’vertical to hori
zontal’ transformation of the power system. The emphasis is on the use of
converter connected DG units to supply active power. Therefore, diﬀerent from
the preceding Chapter 4, the increasing DG level implies a reduction in rotat
ing masses (inertia) and reactive power control ability in the system. In some
cases, power system transient instabilities with very high DG levels (more than
50%) are found. Several solutions are proposed for these problems by means
of rescheduling centralized generators and optimizing the power ﬂow. In some
cases, a minimum number of centralized generators has to remain in the system
to avoid system instabilities.
The results discussed in Chapters 4 and 5 are merely based on a deterministic
approach, where the parameters of the test system are set to the typical values.
Beside the deterministic approach, a stochastic analysis can also be used to
study the transient stability of the power systems, as presented in Chapter 6.
The study is focused on the impact of the stochastic behavior of DG. The results
show that including the stochastic behavior of DG leads to a more complete and
detailed view of the system performance.
Finally, Chapter 7 investigates the situation when the power system is pushed
towards a scenario, where DG penetration reaches a level that covers the total
load of the original power system (100% DG level). The DG units are im
Summary iii
plemented via powerelectronic converters within so called “active distribution
systems” (ADS) connected to the transmission system also via powerelectronic
interfaces. The power system is still connected to a source that provides a con
stant 50 Hz voltage that is meant to give a (system) frequency reference for the
generators but generates no power at steady state. Therefore, any power imbal
ance in the system must be compensated by generators in the ADS. However,
due to the powerelectronic interfaces, the output power of the DG units (and
the ADS) are decoupled from the grid frequency. Therefore in this chapter, the
voltage is used to detect and maintain the power balance. For this purpose,
speciﬁc control concepts are developed. The simulation results show that by
applying such control systems, the power balance in the power system can be
maintained by the ADS.
The research performed in this work indicates that from the transmission
system stability point of view, if higher DG penetration levels are coming up,
suﬃcient inertia and voltage support must be installed. Furthermore, one should
be aware of the fact that the system behaves stochastically, especially with DG.
To a certain extent regional balancing of power can be performed by local voltage
control.
iv Summary
Samenvatting in het
Nederlands
Stabiliteitsanalyse van een transmissienet met een
hoge penetratiegraad van decentrale opwekking
Tegenwoordig neemt de interesse in het produceren van elektriciteit met relatief
kleine decentrale eenheden toe. Deze manier van elektriciteitsopwekking wordt
ook wel ‘distributed generation’ (DG) genoemd. Veel aandrijfsystemen van DG
technologie¨en zijn gebaseerd op hernieuwbare energiebronnen wat resulteert in
een milieuvriendelijke energieopwekking.
Het is bekend dat de toepassing van DG de technische aspecten van de dis
tributienetten be¨ınvloedt. De invloed van kleine hoeveelheden DG aangesloten
op het net, op de transi¨ente stabiliteit van het transportnet is nog niet vaak
onderzocht. Wanneer het penetratieniveau van DG stijgt, is het eﬀect niet meer
beperkt tot het distributienet, maar wordt het gehele systeem be¨ınvloed.
Dit proefschrift onderzoekt de eﬀecten van de toepassing van DG op de
transi¨ente stabiliteit van het transmissienet, met de nadruk op een potenti¨ele
overgang van een ’verticaal gericht systeem’ naar een ’horizontaal gericht sys
teem’ (Hoofdstuk 1)
Met dit doel is het belangrijk om de kenmerken van DG te onderzoeken,
die het dynamische stabiliteitsgedrag van een transmissiesysteem be¨ınvloeden
(Hoofdstuk 2). Daarom worden DG eenheden geclassiﬁceerd op basis van de
primaire energiebronnen (zowel conventionele als hernieuwbare). Het soort pri
maire energiebron bepaalt in grote mate de kenmerken van de energieoutput
van DG en het type netaansluiting dat wordt toegepast. Ook bepaalt het of
vermogenelektronische interfaces worden gebruikt. Gebaseerd op de DG classi
ﬁcatie kunnen basismodellen afgeleid worden voor de DG technologie¨en om te
gebruiken in de transi¨ente stabiliteitssimulatie van een transmissienet, wat in
Hoofdstuk 3 behandeld wordt.
Een probleem in elektriciteitsvoorzieningsystemen is het handhaven van syn
chrone werking van alle (centrale) synchrone machines. Het stabiliteitsprob
leem dat daarmee samenhangt heet de rotorhoek stabiliteit. In dit proefschrift
worden de eﬀecten van de toepassing van DG daarop onderzocht. Daartoe
vi Samenvatting
worden in Hoofdstuk 3, de fenomenen van de rotordynamica van een synchro
nische machine verklaard, die de rotorhoek stabiliteit van een transmissienet
bepalen, middels de bewegingsvergelijking, de vermogen versus hoek curve en
het gelijke oppervlakte criterium concept. De indicatoren voor het beoordelen
van het stabiliteitsgedrag van een transmissienet afgeleid uit deze concepten
zijn de maximumafwijking van de rotorsnelheid en de duur van de slingering
van de synchrone machines. Voor simulatiedoeleinden wordt het bekende 39
knooppunten New England dynamisch testsysteem gebruikt, met een aantal aan
passingen daarin. De basisopzet van het testsysteem wordt in dit hoofdstuk
beschreven, terwijl de details worden gegeven in Bijlagen B en C. Verschei
dene softwarepakketten, geschikt voor transi¨ente dynamische simulatie van DG
in een groot transmissienet, worden behandeld. De representatie van de ver
scheidene DG technologie¨en in de softwarepakketten wordt besproken, waarbij
de nadruk ligt op de weergave van DG eenheden met vermogenelektronische
interfaces (converters).
Het onderzoek van de invloed van een hoog (penetratie) niveau van DG op de
transi¨ente stabiliteit van het transmissienet wordt in Hoofdstuk 4 uiteengezet.
De invloed van de toename van het DG niveau, de sterkte van de DG netkop
peling, verschillende DG technologie¨en, en DG beveiligingsschema’s van via ver
mogenselektronica gekoppelde DG, zijn gesimuleerd en bediscussieerd. Er wordt
aangetoond dat DG de transi¨ente stabiliteit van het transmissienet verschil
lend be¨ınvloedt afhankelijk van boven vermelde factoren. Echter, er is geen
signiﬁcant stabiliteitsprobleem gevonden tot de DG een 30% niveau bereikt, on
afhankelijk van de DG technologie. Deze resultaten zijn logisch vanwege de nog
steeds aanwezige centrale generatoren in het net, met de bijbehorende regeling
van actief vermogen en blindvermogen en de inertie van de roterende massa,
bij dit niveau van DG. Daarnaast is toepassing van DG een natuurlijke manier
om de vermogenstransporten in de transmissielijnen te beperken. Dit verbetert
de transi¨ente stabiliteit van het transmissienet, omdat grote vermogenstrans
porten een nadelige invloed hebben op de demping van de rotorslingeringen:
hoe zwaarder de lijnen zijn belast des te zwakker de koppeling tussen de gener
atoren en de belastingen wordt en des te groter de slingeringen van de centrale
generatoren kunnen worden, vooral bij lange lijnen.
In Hoofdstuk 5, ligt de nadruk van het onderzoek op de invloed van het toen
emende DG niveau op de transi¨ente stabiliteit van het net als dit samen gaat met
de vermindering van in bedrijf zijnde centrale generatoren, leidend tot een ‘ver
tikale naar horizontale’ transformatie van het elektriciteitsvoorzieningsysteem.
De nadruk ligt op het gebruik van de vermogenelektronisch gekoppelde DG
als actief vermogen leverancier. Daardoor, anders dan Hoofdstuk 4, impliceert
het toenemende DG niveau een vermindering van roterende massa (inertie) en
regelmogelijkheden van blindvermorgen in het net. In sommige gevallen wordt
instabiliteit van het transmissienet bij een hoog DG niveau (meer dan 50%)
gevonden. Verschillende oplossingen voor dit probleem worden voorgesteld mid
dels verandering van de inzet van de centrale generatoren en optimalisering van
de vermogenstransporten. In sommige gevallen moet een minimaal aantal cen
trale generatoren in bedrijf gehouden worden om instabiliteit van het net te
Samenvatting vii
vermijden.
De resultaten van de Hoofdstukken 4 en 5 zijn gebaseerd op een determin
istische benadering, waarbij de parameters van het testsysteem ingesteld zijn
op hun typische waarden. Behalve een deterministische benadering kan ook een
stochastische analyse worden gebruikt om de transi¨ente stabiliteit te bestuderen,
zoals beschreven in Hoofdstuk 6. De nadruk van de studie ligt op de invloed van
het stochastische gedrag van DG. De resultaten tonen aan dat het meenemen
van het stochastische gedrag van DG leidt tot een vollediger en gedetailleerder
overzicht van het functioneren van het systeem.
Hoofdstuk 7 tot slot behandelt de situatie waarbij het transmissienet aan
een extreem scenario onderwerpen wordt, waar DG de totale belasting van het
net dekt (100% DG niveau). DG eenheden zijn via vermogenselektronische
interfaces opgenomen in zogenaamde “actieve distributie systemen” (ADS) en
aangenomen is dat deze ook met vermogenselektronische interfaces zijn gekop
peld met het transmissienet. Het systeem is verondersteld nog steeds verbonden
te zijn met een bron met constante 50 Hz frequentie bedoeld als een frequentie
referentie voor de generatoren, maar wekt geen vermogen op in de stationaire
toestand. Daarom moet iedere onbalans in vermogen door opwekking in de
ADS gecompenseerd worden. Echter, vanwege de vermogenselektronische inter
faces is het uitgangsvermogen van de DG eenheden (en de ADS) ontkoppelt van
de netfrequentie. Daarom wordt in dit hoofdstuk de spanning gebruikt om de
vermogensbalans te detecteren en te handhaven. Voor dit doel zijn speciﬁeke
regelconcepten ontwikkeld. De simulatieresultaten tonen aan dat door toepass
ing van deze regeltechnieken, de handhaving van de vermogenbalans gerealiseerd
kan worden via de ADS.
Het onderzoek beschreven in dit proefschrift toont aan dat vanuit oogpunt
van stabiliteit van het transmissienet voldoende inertie en spanningsonderste
uning aanwezig moet zijn als hogere DG penetratieniveaus aan de orde zijn.
Verder moet men zich bewust zijn van het feit dat het systeem zich stochastisch
gedraagt, zeker met DG. In zekere mate kan regionale balanshandhaving worden
uitgevoerd met lokale spanningsregeling.
viii Samenvatting
Contents
Summary in English i
Samenvatting in het Nederlands v
Contents ix
1 Introduction 1
1.1 ’Vertical’ Power Systems . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Distributed Generation Concept . . . . . . . . . . . . . . . . . . 3
1.3 ’Horizontal’ Power Systems . . . . . . . . . . . . . . . . . . . . . 4
1.4 Dynamics of Power Systems . . . . . . . . . . . . . . . . . . . . . 4
1.5 Research Framework . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.6 Objectives and Limitations . . . . . . . . . . . . . . . . . . . . . 8
1.7 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Distributed Generation 11
2.1 Stateoftheart DG Technology . . . . . . . . . . . . . . . . . . . 11
2.1.1 Conventional FossilFuel Based Generators . . . . . . . . 12
2.1.2 Microturbines . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.3 Combined Heat and Power (CHP) Plants . . . . . . . . . 12
2.1.4 Small HydroPower Plants . . . . . . . . . . . . . . . . . . 13
2.1.5 Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.6 Photovoltaics . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.7 Fuel Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.8 Geothermal Power Plants . . . . . . . . . . . . . . . . . . 14
2.1.9 Biomass Power Plants . . . . . . . . . . . . . . . . . . . . 14
2.1.10 Tidal Power Plants . . . . . . . . . . . . . . . . . . . . . . 14
2.1.11 Wave Power Plants . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Output Power Characteristics . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Controllable DG . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Noncontrollable DG . . . . . . . . . . . . . . . . . . . . . 16
2.3 Energy Storage Systems . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Hydrogen Fuel Cells . . . . . . . . . . . . . . . . . . . . . 19
x Contents
2.3.3 Redox Flow Batteries . . . . . . . . . . . . . . . . . . . . 20
2.3.4 Flywheel Systems . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.5 Ultracapacitors . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.6 Superconducting Magnetic Energy Storage (SMES) Systems 20
2.3.7 PumpedHydroelectric Plants . . . . . . . . . . . . . . . . 20
2.3.8 CompressedAir Systems . . . . . . . . . . . . . . . . . . . 21
2.4 DG GridConnection Characteristics . . . . . . . . . . . . . . . . 22
2.4.1 Direct GridConnected DG . . . . . . . . . . . . . . . . . 22
2.4.2 Indirect GridConnected DG . . . . . . . . . . . . . . . . 23
2.4.3 Connecting Energy Storage to the Grid . . . . . . . . . . 26
2.5 DG Prospects: ConverterConnected DG . . . . . . . . . . . . . . 26
2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3 Stability of Systems with DG 29
3.1 Classiﬁcation of Power System Stability . . . . . . . . . . . . . . 29
3.1.1 Rotor Angle Stability . . . . . . . . . . . . . . . . . . . . 30
3.1.2 Voltage Stability . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.3 Frequency Stability . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Rotor Dynamics of Synchronous Machines . . . . . . . . . . . . . 31
3.2.1 Swing Equation . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.2 PowerAngle Equation . . . . . . . . . . . . . . . . . . . . 33
3.2.3 Equal Area Criterion . . . . . . . . . . . . . . . . . . . . . 34
3.3 System Stability Indicators . . . . . . . . . . . . . . . . . . . . . 36
3.3.1 Maximum Rotor Speed Deviation . . . . . . . . . . . . . . 37
3.3.2 Oscillation Duration . . . . . . . . . . . . . . . . . . . . . 37
3.4 DG and Large System Dynamic Simulation . . . . . . . . . . . . 38
3.4.1 Modeling DG Technologies . . . . . . . . . . . . . . . . . 39
3.4.2 Power System Dynamics Software Packages . . . . . . . . 41
3.5 Simulation Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.5.1 The IEEE 39bus New England Test System . . . . . . . 43
3.5.2 DG Technology . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5.3 Incorporation of DG in Distribution Networks . . . . . . . 46
3.5.4 Behavior of Centralized Power Plants . . . . . . . . . . . 47
3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4 Impact of DG on Power System Transient Stability 49
4.1 DG Impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1.1 Simulation Scenarios . . . . . . . . . . . . . . . . . . . . . 49
4.1.2 Transient Stability Simulation . . . . . . . . . . . . . . . . 50
4.1.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 51
4.1.4 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 DG GridConnection Strength Impacts . . . . . . . . . . . . . . . 56
4.2.1 Distribution Network and DG Layout . . . . . . . . . . . 56
4.2.2 Simulation Scenarios . . . . . . . . . . . . . . . . . . . . . 57
4.2.3 Transient Stability Simulation . . . . . . . . . . . . . . . . 60
4.2.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 60
Contents xi
4.2.5 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.3 DG Penetration Level and Technology Impacts . . . . . . . . . . 63
4.3.1 Simulation Scenario . . . . . . . . . . . . . . . . . . . . . 63
4.3.2 Transient Stability Simulation . . . . . . . . . . . . . . . . 64
4.3.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 64
4.3.4 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.4 Protection of PowerElectronics Impacts . . . . . . . . . . . . . . 69
4.4.1 Simulation Scenarios . . . . . . . . . . . . . . . . . . . . . 69
4.4.2 Transient Stability Simulation . . . . . . . . . . . . . . . . 71
4.4.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 71
4.4.4 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5 ’Vertical to Horizontal’ Transformation of Power Systems 75
5.1 Simulation Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2 Simulation Results Case I . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Rescheduling Generation Case I . . . . . . . . . . . . . . . . . . . 83
5.4 Simulation Results Case II . . . . . . . . . . . . . . . . . . . . . . 84
5.5 DG with RideThrough Capability . . . . . . . . . . . . . . . . . 88
5.6 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6 Stochastic Approach to Transient Stability of Power Systems
with DG 91
6.1 Stochastic Load Flow . . . . . . . . . . . . . . . . . . . . . . . . 92
6.2 Stochastic Transient Stability Analysis . . . . . . . . . . . . . . . 92
6.2.1 Simulation Scenario . . . . . . . . . . . . . . . . . . . . . 92
6.2.2 Monte Carlo simulation (MCS) Samples . . . . . . . . . . 94
6.2.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 95
6.3 Stochastic Transient Stability Study with Increasing DG . . . . . 99
6.3.1 Simulation Scenario . . . . . . . . . . . . . . . . . . . . . 99
6.3.2 MCS Samples . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.3.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 100
6.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
7 Maintaining Power Balance with Active Distribution Systems 105
7.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2 Power Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3 Model of Power System with ADS . . . . . . . . . . . . . . . . . 107
7.3.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.3.2 Model of ADS . . . . . . . . . . . . . . . . . . . . . . . . 107
7.3.3 Generator Models . . . . . . . . . . . . . . . . . . . . . . 108
7.4 Basic Controller Model . . . . . . . . . . . . . . . . . . . . . . . . 110
7.5 ADS Control Systems . . . . . . . . . . . . . . . . . . . . . . . . 113
7.5.1 Standalone master controller . . . . . . . . . . . . . . . . 114
7.5.2 Decentralizedcontroller with single reference . . . . . . . 116
xii Contents
7.5.3 Decentralized controller with hysteresis . . . . . . . . . . 118
7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
8 Conclusions 123
8.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
8.2 Stochastic Stability Studies . . . . . . . . . . . . . . . . . . . . . 125
8.3 Remarks and Future Works . . . . . . . . . . . . . . . . . . . . . 125
8.3.1 ‘Inertia’ Contribution . . . . . . . . . . . . . . . . . . . . 125
8.3.2 Reactive Power Control . . . . . . . . . . . . . . . . . . . 126
A List of Symbols and Abbreviations 127
B Test System Data 131
C Generator, Governor and Excitation Systems Data 135
D Power Flow Computation 139
D.1 Power Flow Problem . . . . . . . . . . . . . . . . . . . . . . . . . 139
D.2 NewtonRhapson power ﬂow solution . . . . . . . . . . . . . . . . 141
Bibliography 143
Scientiﬁc Contributions 151
Acknowledgment 155
Biography 157
Chapter 1
Introduction
A classical power system is characterized by a relatively small number of large,
socalled centralized power plants for meeting the electric energy demand. These
power plants are built based on the demand estimate for a certain period of
time. Some constraints, however, limit the expansion and use of such large
power plants and may induce a shift towards a more extensive use of small,
decentralized power generators.
It is wellknown that the implementation of small, decentralized power gen
erators brings both positive and negative consequences to the existing power
system. The technical consequences must be considered carefully in order to
maintain the present reliability level of the power system. Some of these nega
tive consequences will be focused on in this work. Several remedies to eliminate
them or limit their impact will be suggested and discussed.
1.1 ’Vertical’ Power Systems
A power system is designed to supply electrical power to the consumers. Until
now, mainly large, centralized generators have been utilized for the power gener
ation. Synchronous generators are typical electromechanical energy transducers
for large power plants, often close to cooling water, energy resources or supply
routes and connected to the transmission system. If hydropower is available it
may be used as input too.
So, a classical power system consists of three technical stages, namely: gen
eration, transmission, and distribution. The generation system converts me
chanical power that results from the conversion of primary energy sources, such
as nuclear, hydro power, coal, gas, etc., into electrical. The transmission system
transports the electrical power over a long distance to the load centers. The dis
tribution system distributes the electrical power to the consumers/loads. This
classical power system can be best illustrated by considering the diﬀerent voltage
levels (Figure 1.1).
2 1.1 ’Vertical’ Power Systems
Generation
Plant
Generation
Substation
Transmission
System
Transmission
Substation
Subtransmission
System
Distribution
Substation
Distribution System
Customer
H
V
/
E
H
V
N
e
t
w
o
r
k
(
1
1
0
k
V
t
o
4
0
0
k
V
)
M
V
N
e
t
w
o
r
k
(
1
0
t
o
3
0
k
V
)
L
V
N
e
t
w
o
r
k
(
2
3
0
/
4
0
0
V
)
"Electrical power
flow"
"Electrical power
flow"
"Electrical power
flow"
D
i
r
e
c
t
i
o
n
o
f
a
c
t
i
v
e
p
o
w
e
r
f
l
o
w
Figure 1.1: VerticallyOperated Power System (modiﬁed from [8])
Introduction 3
After being generated at the power plant (typically at a voltage of 10 kV
to 30 kV), the power is transformed to a higher voltage level in the generation
substation. The High Voltage (HV) or Extra High Voltage (EHV) (110 kV
to 400 kV) transmission systems transports the electrical power further to the
(sub)transmission substations, where the power is transformed to Medium Volt
age (MV) level (typically 10 to 30 kV), and where it enters the primary distribu
tion systems. Finally, in the distribution substations, the power is transformed
to Low Voltage (LV) level and is distributed to the consumers [78]. Therefore,
electrical energy generally ﬂows from the higher to the lower voltage levels in
the network. Based on the diﬀerent voltage levels, this type of power system
can then be viewed as a ’verticallyoperated’ power system, which in this work
will be referred to as a ’vertical’ power system.
The expansion and the construction of large power plants are limited by
both rational (e.g. economical, environmental and geographical) considerations
and irrational constraints (e.g. social and political issues) [56].
1.2 Distributed Generation Concept
Nowadays, an increasing amount of electrical power is generated by decentral
ized power generators of relatively small scale (i.e. smaller than 50100 MW).
This way of electrical power generation is referred to as ’Distributed Generation’
(DG) because it is spread out over the system. These small power generators are
usually located in the vicinity of the electrical loads, and are mostly connected
to distribution networks (i.e. at MV or LVnetworks) [12], [32].
In contrast to the conventional power plants, the development and the imple
mentation of DG units are encouraged mostly by environmental forces. This has
stimulated research, promotion, development and increased use of new, renew
able, clean and environmentally friendly forms of energy [5], [27]. Renewable
energy sources like wind, biomass, sun, tidal, wave and geothermal energy
are used. Most of these renewable energy sources can be converted to electric
power, in units in a range of hundreds kWs to some MWs, by (relatively) small
generators that are connected to the distribution networks, close to the load
centers.
Some types of distributed generation (DG) are based on conventional fossil
energy sources, but are often because of their relatively low carbon emission
classiﬁed as environmentallyfriendly types of power generation. Within this
class are the microturbine generator supplied by natural gas, and the Combined
Heat and Power (CHP) generation, which is practically a parallel conversion of
fuel into electrical and thermal energy (more carbon emission will be produced
if the electrical and thermal energy are generated separately).
The rise of DG is supported by the advancements in supporting technologies
like power electronic converters and controllers.
Currently there are many DG technologies available. An overview of these
DG technologies is given in Chapter 2.
4 1.3 ’Horizontal’ Power Systems
1.3 ’Horizontal’ Power Systems
Due to the possible largescale implementation of DG units in the classical
’vertical’ power system, a transition towards a more ’horizontal’ power system
may take place. In addition to the power injected in the EHV and HV system by
the large power plants, DG units supply the system via the MV or LV networks.
Therefore, the power can ﬂow both ’vertically’, i.e. from the higher to the lower
voltage levels, as well as ’horizontally’ from one MV or LV network to another
or from a generator to a load within the same MV or LV network leading to a
new term: the horizontal power system or horizontallyoperated power system.
The implementation of DG turns the passive distribution network into an
active one. In this active distribution network some costumers not only consume
electricity, but they also generate and if generation surpasses their demand,
supply the network. Figure 1.2 shows an example of both an active and a
passive distribution network. In the active network the power ﬂow is no longer
in one direction (downwards), as is the case in the passive network, but may be
bidirectional (down and upwards).
In this way, it is possible that power is transferred from one distribution
network to another. When we reﬂect further on this issue, we could even imagine
that on certain moments in time the electrical power generated by the DG within
the distribution networks may become suﬃcient to fulﬁll the total demand of the
system. In this particular case, the remaining large (centralized) power plants
may be shut down.
Figure 1.3 illustrates a ’VerticaltoHorizontal’ transformation of a power
system. In the ’ﬁrst’ transformation step, a large amount of DG is implemented
in the power system and all centralized generators remain but generate less
(Figure 1.3: graph (a) changes to graph (b)). In the ’second’ transformation
step, the amount of DG in the system increases in such a way that a number of
centralized generators (power plants) are shut down for eﬃciency reasons (Fig
ure 1.3: graph (b) changes to graph (c)). Finally, in the ’third’ transformation
step, when power generated by DG within the active distribution networks is
suﬃcient to match the total demand, all (remaining) centralized generators are
out of service (Figure 1.3: graph (c) to graph (d)).
It is hard to imagine how a system under graph d could operate but theo
retically these are the steps.
1.4 Dynamics of Power Systems
The power system is a dynamic system. Even under normal operation condi
tions, loads are connected and disconnected frequently and demand for both
active and reactive power changes continuously. Besides, the power system is
subject to disturbances caused by malfunctioning or failing equipment. The
ability of a power system to remain in a state of operating equilibrium under
normal operating conditions and to regain an acceptable state of equilibrium af
ter being subjected to a disturbance is deﬁned as the power system stability [34].
Introduction 5
Distribution
Network
Transmission Network
Industrial customers who
consume electricity
Domestic customers and
small business who
consume electricity
(a) Passive Network
(b) Active Network
Transmission Network
Distribution
Network
Industrial customers with
DG (colored in gray) also
generate some electricity
which flows back into the
network
Domestic customers and
small business with
domestic DG (colored in
gray) can also generate
electricity which flows
back into the network
Distributed generator
e.g. wind turbine
Thin line indicates flow from the network
Thicker line indicates flow from, and to, the network
Thin line indicates flow from the network
Figure 1.2: Conventional (passive) distribution network (top  a) and an active
distribution networks with DG (bottom  b) (modiﬁed from [14])
6 1.4 Dynamics of Power Systems
~ ~
Centralized generators
Loads
Transmission
network
~
Loads and Distributed Generators
Transmission
network
Centralized generators
~
~ ~
Loads and Distributed Generators
Transmission
network
Centralized generator
~
~ ~ ~ ~
Loads and Distributed Generators
Transmission
network
~
~
~ ~ ~ ~
(a) (b)
(c) (d)
1
2
3
~ ~
Figure 1.3: ’VerticaltoHorizontal’ transformation of the power system
More than 100 years of experience have lead to the present power system with
many subsystems and dynamic elements, a lot of them equipped with control
systems to stabilize the operation of the overall system. For instance, power
plants are equipped with prime mover controllers to regulate the speed and
thus the active power output, and excitation controllers to regulate the voltage
and thus the reactive power output. When we consider the interconnected power
system as a single system, a necessary condition for a stable operation is that
more or less all the large synchronous generators remain in synchronism. Then
it is up to the dynamics of the system whether stability can be maintained.
Instability can also occur if a power system cannot maintain the voltage levels
within a required range [34]. The implementation of DG inﬂuences both steady
state and dynamic performance of a power system. In this work, however, the
focus is on the dynamics.
Introduction 7
1.5 Research Framework
The research presented in this work has been performed within the framework of
the ’Intelligent Power Systems’ project. The project is part of the IOPEMVT
program (Innovation Oriented research Program  ElectroMagnetic Power Tech
nology), ﬁnancially supported by SenterNovem, an agency of the Dutch Ministry
of Economical Aﬀairs. The ’Intelligent Power Systems’ project is initiated by
the Electrical Power Systems and Electrical Power Electronics Groups of the
Delft University of Technology and the Electrical Power Systems and Control
Systems Groups of the Eindhoven University of Technology. In total 10 Ph.D.
students are involved and work closely together. The research focuses on the
eﬀects of the structural changes in generation and demand taking place, like for
instance the largescale introduction of distributed (renewable) generators [59].
The project consists of four parts (illustrated in Figure 1.4).
The ﬁrst part (research part 1), inherently stable transmission system, in
vestigates the inﬂuence of uncontrolled decentralized generation on stability
and dynamic behavior of the transmission network. As a consequence of the
transition in the generation, less centralized plants will be connected to the
transmission network as more generation takes place in the distribution net
works, whereas the remainder is possibly generated further away in neighboring
systems. Solutions investigated include the control of centralized and decentral
ized power, the application of power electronic interfaces and monitoring of the
system stability.
The second part (research part 2), manageable distribution networks, focuses
on the distribution network, which becomes ’active’. Technologies and strategies
have to be developed that can operate the distribution network in diﬀerent
modes and support the operation and robustness of the network. The project
investigates how the power electronic interfaces of decentralized generators or
between network parts can be used to support the grid. Also the stability of the
distribution network and the eﬀect of the stochastic behavior of decentralized
generators on the voltage level are investigated.
In the third part (research part 3), selfcontrolling autonomous networks,
autonomous networks are considered. When the amount of power generated
in a part of the distribution network is suﬃcient to supply a local demand,
the network can be operated autonomously but as a matter of fact remains
connected to the rest of the grid for security reasons. The project investigates
the control functions needed to operate the autonomous networks in an optimal
and secure way.
The interaction between the grid and the connected appliances has a large
inﬂuence on the power quality. The fourth part (research part 4), optimal
power quality, of the project analyzes all aspects of power quality. The goal is
to provide elements for the discussion between polluter and grid operator who
has to take measures to comply with the standards and grid codes. Setting up
a power quality test lab is an integral part of the project.
The research described in this thesis is within research part 1: inherently
stable transmission systems.
8 1.6 Objectives and Limitations
Inherently
stable
transmission
system
Manageable
distribution
networks
Selfcontrolling
autonomous
networks
Optimal power
quality
1
2 3 4
Figure 1.4: Research items within the ’Intelligent Power Systems’ research
project
1.6 Objectives and Limitations
In this work, the following two objectives are set:
• Investigate the impact of a high DG penetration level on the stability of
a power system.
• Investigate the stability of a power system that undergoes a ’verticalto
horizontal’ transformation.
This research is unique as it combines the investigation of an increasing DG
penetration level and a power system that transforms from a vertical into a
horizontal one. The focus is only on the dynamic impacts, i.e. the transient
stability. Steady state and economic impacts are beyond the scope of this work.
For this purpose, power system simulation software package PSS/E is used,
where models of the power (test) system and DG are included. Simulation sce
narios of power systems with DG are deﬁned later. Based on the simulation
results, special emphasis is on the behavior of the centralized generators in ser
vice.
Introduction 9
1.7 Outline of the Thesis
The thesis is organized as follows:
• In Chapter 2, an overview of the current DG technologies is given. The
emphasis is on the classiﬁcation of the diﬀerent DG technologies according
to their potential impact on the power system stability.
• An overview of power system stability is presented in Chapter 3. In this
chapter, the term ”inherently stable transmission system” is deﬁned. Fur
thermore the research approach and the simulation setup, used and pre
sented throughout the work, are discussed, including test system, software,
system stability indicators, basic associated controls of system elements
and parameters used.
• In Chapter 4, the impact of DG implementation on the power system
transient stability is discussed. Diﬀerent scenarios of a power system with
a high DG penetration level are developed. The impact of diﬀerent DG
technologies, fault durations and locations, DG penetration level, DG grid
connectionoption, and protection schemes of power electronic interfaced
DG units are investigated.
• In Chapter 5, the scenarios for the ’verticaltohorizontal’ transformation
of power systems are further elaborated. The transient stability impact
of this transformation is analyzed and solutions for reducing the negative
eﬀects are discussed.
• In Chapter 6, a stochastic approach to the transient stability of system
with DG within the framework of “verticaltohorizontal” transformation
is studied.
• In Chapter 7, a power system reaching 100% DG implementation is stud
ied. DG units are implemented within Active Distribution Systems. Con
trol methods to maintain the power balance in such a power system with
Active Distribution Systems are suggested.
• The conclusions and recommendations for future work are given in Chap
ter 8.
10 1.7 Outline of the Thesis
Chapter 2
Distributed Generation
The impact of distributed generation (DG) on the dynamic stability of power
systems is studied. Therefore, it is important to examine characteristics of DG
that inﬂuence this behavior. As mentioned in Chapter 1, DG units can be
based on various (both conventional and alternative) primary energy sources.
The type of primary energy source and the conversion process determine, to
a large extent, the output power characteristics of DG and the type of grid
connection applied.
Based on the output power characteristics, DG can be classiﬁed as dispatch
able or nondispatchable as is described in Section 2.2. The output power of
nondispatchable units, especially the ones driven by renewable energy sources,
can show high outputpower ﬂuctuations. Energy storage systems, as described
in Section 2.3, can be applied to smooth this intermittent eﬀect.
In Section 2.4 the way DG is connected to the network (grid) is reviewed.
There are two options, depending on both the type of primary energy source and
prime mover: a direct and indirect grid connection. A direct grid connection
is made by using the common/classical synchronous and induction generators,
whereas an indirect grid connection is made by means of powerelectronic con
verters. Section 2.5 elaborates on this issue. Concluding remarks are made in
Section 2.6.
2.1 Stateoftheart DG Technology
Many deﬁnitions of distributed generation (DG) exist. CIGRE Working Group
37.23, for example, has deﬁned distributed generation (DG) as electrical gener
ation that is not centrally planned, not centrally dispatched
∗
, and connected to
the distribution network [12]. A DG unit usually produces electric power well
below 100 MW [32], [33]. Other literature however advocates a boarder and
∗
not centrally dispatched: it cannot be controlled from a system control center; it does not
implicate that the unit cannot be controlled locally
12 2.1 Stateoftheart DG Technology
more straightforward deﬁnition of DG: a DG source is an electric power gener
ation source connected directly to the distribution network or on the customer
side of the meter [1], [52]. Thus, the way that a generator is implemented in
a power system determines its classiﬁcation as DG, and not the type of pri
mary energy source used. However, many generator units that are driven by
renewable sources of energy inherently possess the characteristics of DG.
2.1.1 Conventional FossilFuel Based Generators
Within the category DG, the term ’Conventional FossilFuel Based Generator’
is used to describe small fossilfueled power plants within a range of kWs up
to 100 MW [7], [12], [33]. The reciprocating engines and gas turbines are the
most common in this category.
Reciprocating engines are characterized by low capital cost, possible thermal
and electrical cogeneration, and good modularity and ﬂexibility. Furthermore
they are reliable [7], [12]. Reciprocating engines, however, have drawbacks.
The use of diesel or gasoline gives high emission levels [30]. The emission can be
reduced to some extent by using natural gas as energy source. A large number of
moving parts leads to high noise levels pollution and increases the maintenance
cost [7].
Gas turbines are commonly used in industry [12]. In oil industry for example,
the associated gas from the oilﬁeld is frequently used to generate electricity. The
use of natural gas results in lower emission when compared to reciprocating
engines. As DG, gas turbines are mostly encouraged by the development of
microturbines, highlighted in the next subsection.
2.1.2 Microturbines
A ’micro’ gas turbine (microturbine) produces electric power in the range of 25
500 kW. An electrical generator is integrated within the microturbine, that op
erates at a high speed (50,000 to 120,000 RPM). The electric power is produced
with a frequency (in the order) of thousands of Hz [2]. Therefore, a power
electronic converter is used to interface the generator and the grid. Within the
powerelectronic interface, the highfrequency electrical power is converted to
DC before it is inverted back to the lowfrequency AC of the grid.
Most microturbines use natural gas. As a consequence, microturbines are
typically characterized by low emission levels. The use of renewable energy
sources such as ethanol is also possible [15].
2.1.3 Combined Heat and Power (CHP) Plants
Combined Heat and Power (CHP), also known as cogeneration, is the simulta
neous production of electrical power and useful heat [33]. Reciprocating engines,
gas turbines, and microturbines can be used in CHP schemes. CHP generation
on a large scale is usually based on fossil fuel. In general, CHP is heat driven
Distributed Generation 13
and electricity is the byproduct. With this simultaneous process, the overall
eﬃciency of a CHP plant can be around 85% [12], [33].
2.1.4 Small HydroPower Plants
A hydropower plant generates electricity from the motion of a mass of water,
where a power house is installed. This water movement can be obtained, for
example, from a runof river or a river with a small impoundment [66]. A small
hydropower plant produces electric power up to 10 MW. Hydro power plant
technology has reached maturity. A small hydropower plant has less impact on
the environment and ecosystem, when compared to a large hydropower plant,
and is easy to build within a short construction schedule [66]. Once built, its
maintenance cost is minimal [62].
2.1.5 Wind Turbines
A wind turbine generates electricity by extracting kinetic energy from the wind
passing through its blades. Wind energy is one of the most promising energy
sources to be used for renewable electricity generation [51]. Apart from us
ing a small wind turbine as DG for generating emissionfree power, however,
the increasing interest for implementing wind turbines is mostly driven by the
availability of wind energy for generating power at large scale of MWs or even
GWs [77].
2.1.6 Photovoltaics
Photovoltaic (PV) power generation systems convert sunlight directly into elec
tricity [51]. A PV cell consists of two or more semiconductor layers of speciﬁc
physical properties. These layers are arranged in such a way that when the
PV cell is exposed to sunlight, the photons cause the electrons to move in one
direction (crossing the junctions of the layers) and a direct current (DC) is
generated.
Currently, PV energy cost is still high. However, the capital cost of PV
modules has declined in the past decades. PV implementation is encouraged
by the inﬁnite availability of sun energy, long life cycle and simple maintenance
(since there are no moving parts), high modularity and mobility, and short
design, installation and startup time of a new plant [51].
2.1.7 Fuel Cells
DC power can also be generated by an electrochemical process. An example
is the socalled fuel cell. It consists of a positive electrode (anode) and a neg
ative electrode (cathode). To generate electricity, fuel (usually hydrogen) and
an oxidant must be supplied to the anode and the cathode, respectively. Elec
trochemical reactions create ion ﬂows, that generate electricity. One fuel cell
14 2.1 Stateoftheart DG Technology
only produces a small amount of electricity, and larger amounts can be obtained
from a stack of fuel cells [7], [13].
Fuel cells are modular, portable and produce low noise pollution, because
there are no moving parts [39]. These characteristics make fuel cells suitable as
DG in, for examples, remote areas. In the future, electrical networks (both AC
and DC) can be combined with a gas and hydrogen infrastructure. This new
structure may further increase the implementation of fuel cells as DG [19].
2.1.8 Geothermal Power Plants
Geothermal power plants convert the energy contained in hot rock into elec
tricity by using water to absorb the heat from the rock and transport it to the
surface of the earth. The heat from geothermal reservoirs provides the force that
rotates the turbine generators and produces the electricity. The used geothermal
water is then returned (injected back) into the reservoir to be reheated. This
cycle will maintain the pressure of the reservoir and sustain the reservoir [57].
A geothermal power plant is relatively sustainable.A ﬁeld may remain pro
ductive over a period of tens of years. It produces no pollutant and no unwanted
product, if any, can be disposed underground [68], [80].
2.1.9 Biomass Power Plants
The term “biomass” describes all organic matter that is produced by photo
synthesis. It includes all water and landbased vegetation and trees, municipal
biosolids (sewage), animal wastes (manures), forestry and agricultural residues,
and certain types of industrial wastes [26], [72]. Biomass is considered a substi
tute for fossil fuels. Practically, biomass is converted to thermal energy, liquid,
solid or gaseous fuels and other chemical products through a variety of conver
sion processes [72]. The latter forms are then converted into electricity. The
biomass products, for example, can be used as fuel to generate electricity. The
gaseous fuels can be applied in fuel cell systems.
In general, biomass is abundantly available and can be considered as a re
newable.
2.1.10 Tidal Power Plants
Tidal energy is derived from the gravitational forces of attraction that operate
between the earth and the moon, and the earth and the sun. Energy is extracted
either directly by harnessing the kinetic energy of currents due the tides or by
using a basin to capture potential energy from the diﬀerence in height of a
rising and falling mass of water. To generate electricity, tidal ﬂow is extracted
by means of propellers with large diameters. In the latter technique, a huge
dam, called a ’barrage’ is built across a river estuary. When the tide goes in
and out, the water ﬂows through tunnels in the dam. The ebb and ﬂow of the
tides can be used to turn a turbine. When the tides comes into the shore, they
Distributed Generation 15
can be trapped in reservoirs behind dams. Later, when the tide drops, the water
dam can be used like in a regular operation of a hydroelectric power plant [16].
Tidal power is a renewable energy source. Tidal power plants produce no
pollutant. They also cause no fundamental change of the natural rhythm of the
tidal cycle and no inundation of the adjacent area. These factors encourage the
implementation of tidal power plants [6].
However, building a tidal power plant has to be planned carefully considering
the potential ecological impacts, especially during the construction [22].
2.1.11 Wave Power Plants
Waves are generated at the surface of oceans by wind eﬀects which in turn result
from the diﬀerential heating of the earth’s surface. Wave energy is complemen
tary to tidal power, it uses the essentially upanddown motion of the sea surface
(wave power), instead of using the energy of the sea rushing backwards and for
wards (tidal power). A wave power plant extracts wave energy and converts it
into electricity [79].
The wave power plant is promoted as electricity generation available in abun
dance throughout the world; it is clean and nonpolluting, renewable, and suited
to electrify remote communities, especially as DG [17]. However, just like a tidal
plant, the erection of a wave power plant should be planned carefully, so that
the ecological impacts are minimized.
2.2 Output Power Characteristics
One of the main characteristics of a power system is that the supply and the
demand must be kept in balance at any time. In a steadystate operation of a
traditional power system, the use of synchronous generators (within the power
plants) enables the power output of each plant (and each generating unit within
the plant) to be dispatched for any speciﬁed load condition [25]. Dispatching
a power plant (and a generator unit) is a function of the availability of the
primary energy sources that drives the prime mover and the ﬂexibility of the
conversion process (ramping up and down). By deﬁnition DG units are not
centrally dispatched
†
, but several DG technologies enable the DG unit to be
controlled locally. The DG operator can determine an exact power output of the
DG units by controlling the primary energy sources (or fuels) that are supplied
to the DG units. Other DG technologies are based on renewable energy sources
where the operator cannot dispatch the DG units because the behavior of the
primary energy sources cannot be controlled. In case of wind turbines and
photovoltaic panels, for example, no extra primary energy can be supplied to
the generator units in order to produce more electricity. Normally, most of the
renewable energy based electrical generation is operated in such a way that the
electricity production is maximized.
†
the nondispatchable characteristics of DG units are important when a stochastic approach
is considered. Such approach is applied in this thesis in Chapter 6
16 2.2 Output Power Characteristics
Table 2.1: Controllable and Noncontrollable Classiﬁcation of DG
DG technology Controllable Noncontrollable
Conventional FossilFuel Based Generators
√
Microturbines
√
Combined Heat and Power (CHP) Plants
√
Small hydropower plants
√
Wind turbines
√
Photovoltaics
√
Fuel cells
√
Geothermal power plants
√
Biomass power plants
√
Tidal power plants
√
Wave power plants
√
In short, DG technologies can be classiﬁed into two categories:
• Controllable DG.
• NonControllable DG.
As a summary, Table 2.1 lists the various DG technologies and their classi
ﬁcation as controllable and/or noncontrollable generation.
2.2.1 Controllable DG
Controllable DG is characterized by its ability to control the fuel supply to the
generator. As a result, the output power can be determined, and dispatched.
Among the DG technologies that can be classiﬁed as controllable DG are
conventional fossilfuel based generators, microturbines, fuel cells, geothermal
power plants, and power plants driven by biomass. Except for the fuel cells,
these technologies utilize conventional rotating electrical machines for power
conversion (synchronous or induction generators), driven by prime movers based
on reciprocating or combustion turbine technologies. By controlling the fuel that
is supplied to the prime mover, the torque of the prime mover can be adjusted.
A note should be made with regard to geothermal power plants. The geother
mal primary energy source is not as ﬂexible as fossil fuels for dispatching the
generator units [68].
2.2.2 Noncontrollable DG
Noncontrollable DG represents DG technologies where the DG operator cannot
determine the power output of the DG units. Among the DG technologies that
can be classiﬁed as noncontrollable DG are small hydro power plants, wind
turbines, photovoltaics, tidal and wave power plants and CHP plants.
Distributed Generation 17
Small hydropower plants
Because of the nonavailability of large power impounding (dam), the power
output of a hydro turbine (the prime mover in the small hydropower plant) is
practically driven by a directcaptured water ﬂow. A simple expression of the
power output for a small hydroplant is [33]
P = QHηρg, (2.1)
with P the output power [W], Q the ﬂow rate [m
3
s
−1
], H the eﬀective head [m],
η the overall eﬃciency, ρ the density of water [kgm
−3
], and g the gravitional
constant [ms
−2
].
For small hydropower plants, H, η, ρ, and g in (2.1) are deterministic and
constant. Without signiﬁcant storage capacity, a small hydropower plant may
experience a very large variation in available water ﬂow (Q), and output power
(P) [33]. Thus, a small hydropower unit is nondispatchable.
Wind turbines
The power generated by a wind turbine (provided that the upstream wind ve
locity, v, is between minimal and maximal values, e.g. 4 < v < 25 [ms
−1
]) can
be expressed as [33], [51]
P =
1
2
C
p
ρv
3
A, with C
p
=
(1 +
v
o
v
)[1 −(
v
o
v
)
2
]
2
. (2.2)
In (2.2), P denotes the output power [W], C
p
the power coeﬃcient, v
o
the
downstream wind velocity at the exit of the rotor blades [ms
−1
], ρ the air density
[kgm
−3
], and A the swept area of the rotor blades [m
2
].
In practice, ρ, A, and to some extent C
p
, are deterministic and constant
values. Thus, the power produced by a wind turbine is mainly characterized by
the wind velocity. The wind velocity itself has a stochastic nature; any wind
speed can occur at any time [61]. Moreover, when the upstream wind velocity
(v) is either below minimal or above maximal operating values of the wind
plant, e.g. v < 4 or v > 25 [ms
−1
], the output power equals zero. As a result, a
stochastic output power results, especially when a single wind turbine or plant
is regarded [50].
Photovoltaics
The power generated by a PV module is given in (2.3) [51] as
P = η ×(E
ed
×A
PV total
+E
es
×Area
PV withsun
), (2.3)
where
A
PV withsun
= (
−→
S ×
−→
P ) ×A
PV total
(2.4)
18 2.2 Output Power Characteristics
and
−→
S = [S
x
S
y
S
z
], 
−→
S  = 1, (2.5)
−→
P = [P
x
P
y
P
z
], 
−→
P  = 1, (2.6)
−→
S
x
= cos(θ) ×cos(α
sun
), (2.7)
−→
S
y
= cos(θ) ×sin(α
sun
), (2.8)
−→
S
z
= sin(θ), (2.9)
−→
P
x
= cos(β) ×cos(α
panel
), (2.10)
−→
P
y
= cos(β) ×sin(α
panel
), (2.11)
−→
P
z
= sin(β). (2.12)
In (2.3) to (2.12) P denotes the power extracted from the sunlight [W], η the
eﬃciency of the solar panel, E
ed
and E
es
the diﬀuse and the directhorizontal
irradiance [Wm
−2
],
−→
S and
−→
P the solar and panel orientation, θ and α
sun
the
altitude and azimuth angle of the sun [rad], and β and α
panel
the altitude and
the azimuth angle of the panel [rad].
In practice, A
PV total
, η, β and α
panel
are deterministic and constant. There
fore, the generated electricity is characterized by E
ed
, E
es
, θ and α
sun
. The
altitude and the azimuth angle of the sun (θ and α
sun
) have daily and seasonal
patterns, whereas the characteristics of E
ed
and E
es
are intermittent. Weather
changes and cloud movement, for example, strongly inﬂuence the values of E
ed
and E
es
, and the generated electricity. The power generation of PV is non
controllable [81].
Tidal power plants
The power output of a turbine operating in ﬂowing water is [74]
P =
1
2
ρAC
p
v
3
. (2.13)
In (2.13), P denotes the output power [W], ρ the density of the ﬂuid [kgm
−3
],
A the area of the ﬂow covered by the device [m
2
], C
p
the power coeﬃcient of
the device (the percentage of power that the turbine can extract from the water
ﬂowing through the turbine), and v the velocity of the water [ms
−1
].
For a tidal power plant, ρ, A, and C
p
in (2.13) are deterministic and constant.
Therefore, the output power P depends on the velocity of the water v. Thus,
the tide, which is predicable but variable in nature [9], is the only factor that
aﬀects the generating activity of a tidal power plant. This makes the tidal power
generation noncontrollable.
Wave power plants
The power production of a wave power plant can be assessed using [45], namely
P
abs
= αA
w
H
1.5
s
. (2.14)
Distributed Generation 19
In (2.14) P
abs
denotes the average absorbed power, A
w
the ﬂoat water plain
area [m
2
], and H
s
the signiﬁcant wave height [m].α is a coeﬃcient that equals
0.166 [kgm
−1.5
s
−3
] under ideal conditions [45].
For a wave power plant, α and A
w
are deterministic. The output power
depends practically on the wave height (H
s
), neither constant nor controllable.
Hence, a wave power plant is noncontrollable.
2.3 Energy Storage Systems
A largescale implementation of renewable energy sources for power generation
could be expected in future. With the noncontrollable nature of these gener
ators, especially when renewable energy sources such as wind, wave and sun
are exploited, the output power is diﬃcult to predict. Moreover, high power
ﬂuctuations from these generators can be expected. In this case, energy storage
systems/devices may be needed to cover the resulting imbalances between the
power generation and the consumptions [64], [67]. In the following sections,
energy storage systems, and their technical characteristics from a power system
point of view, are highlighted.
2.3.1 Batteries
Batteries store energy in electrochemical form. There are two basic types of
batteries. The socalled primary battery converts chemical energy into electrical
energy in a nonreversible process and is discarded after discharge. The sec
ondary battery works in a reversible reaction. It converts chemical energy into
electrical in discharge, and vice versa in charge mode [51].
Batteries are the most widelyused devices for electrical energy storage in
a variety of applications. Batteries can store rather large amounts of energy
in a relatively small volume. They are modular, so that output power in the
order of MWs is accessible. They are also quiet during operation, which makes
them suited for implementation near load centers where also DG can be imple
mented [36], [64].
2.3.2 Hydrogen Fuel Cells
Hydrogen fuel cells store energy in electrochemical form. Hydrogen is produced
by electrolysis of water using the oﬀpeak electricity such as coming from wind
turbines, photovoltaics, hydro or even nuclear power plants. This hydrogen can
be used to operate fuel cells when there is a high demand for electricity. In
this way, the hydrogen fuel cell energy storage can be ’charged’ and ’discharged’
reversibly.
Hydrogen fuel cells are environmentally friendly, if the hydrogen is produced
by electrolysis of water, and no fossilbased fuel is used [64].
20 2.3 Energy Storage Systems
2.3.3 Redox Flow Batteries
Redox ﬂow batteries store energy in electrochemical form. They are classiﬁed
in between secondary batteries and hydrogen fuel cells and they have the char
acteristic of secondary batteries as they can be charged and discharged [54].
Flow batteries have a number of advantages. For example, the power output
can easily be varied by increasing the size of the membranes, and the storage
capacity can be raised by increasing the size of the tanks of the electrolytes [54].
2.3.4 Flywheel Systems
Flywheels store energy mechanically as kinetic energy [64]. A ﬂywheel system
consists of a ﬂywheel, a motor/generator, and powerelectronic converter. The
ﬂywheel speeds up as it accumulates energy and slows down as energy is re
leased [29]. A ﬂywheel system has a high energystorage density. It also has an
unlimited number of charging and discharging cycles.
2.3.5 Ultracapacitors
Ultracapacitors store energy in the form of electrostatic energy (static charge).
Similar to a regular capacitor, the electric energy is stored by means of charge
separation [10]. However, compared to ordinary electrolyte capacitors, ultraca
pacitor’s capacitance can be more than tens times higher. Its equivalent internal
resistance is more than tens times lower than that of a battery, allowing more
than tens times higher discharging/charging currents [23]. This makes the ul
tracapacitor suited for shortterm, highpower applications. An ultracapacitor
has an unlimited number of charge and discharge cycles at high rates [10], [23].
2.3.6 Superconducting Magnetic Energy Storage (SMES)
Systems
Superconducting Magnetic Energy Storage (SMES) systems store energy in the
magnetic ﬁeld created by the ﬂow of direct current in a superconducting coil.
The stored energy can be released by discharging the coil.
The use of SMES is encouraged by several factors [40]. Firstly, its eﬃciency
is very high (up to 95%) as no conversion of energy to other forms is involved.
Secondly, it can very rapidly dump or absorb power from the grid as the only
limitations are the control loop and the switching time of the solidstate com
ponents connecting the coil to the grid.
2.3.7 PumpedHydroelectric Plants
Pumpedhydroelectric plants store energy in the form of potential energy. They
use oﬀpeak power to pump water uphill to an elevated reservoir. When electric
ity is needed, the water is released to ﬂow to a lower reservoir, and its potential
energy is used to drive turbines [64].
Distributed Generation 21
A pumpedhydroelectric energy system enables largescale energy storage
with a high capacity and power rating. It has an unlimited charging and dis
charging cycle and long life duration. The implementation of such an energy
storage system is limited by the requirement of a signiﬁcant land area with
suitable topography for the upper and lower plants.
2.3.8 CompressedAir Systems
Compressedair energy storage systems store energy in the form of potential
energy by compressing air within an air reservoir, using a compressor powered
by oﬀpeak electric energy. When electricity is needed, the air is withdrawn,
heated via combustion, and run through expansion turbines to drive an elec
tric generator. During discharge, the plant’s generator produces power using
a conventional natural gas combustor and the compressed air. When charg
ing, the generator operates in reverse  as a motor (powered by the oﬀpeak
electric energ)  to provide mechanical energy to the air compressors [64], [65].
Such a plant uses about onethird of the premium fuel of a conventional simple
cycle combustion turbine and produces onethird of the pollutants per kWh
generated. This technology is considered as a hybrid storage and generation
plant because it uses fuel and electricity in its storage cycle [64]. Among the
positive aspects of a compressedair energy storage system, are its availability
for largescale storage (high capacity and power rating), the unlimited charge
and discharge cycles, and the long life duration [65], however it needs speciﬁc
geography.
Figure 2.1 shows an indication of the working areas of these energy storage
systems.
Flow Batteries
Hydrogen Fuel Cells
Batteries
Flywheels
Ultracapacitors
Pumped
Hydroelectric
Plants &
Compressed
Air Systems
C
a
p
a
c
i
t
y
[
W
h
]
Power [W]
10
9
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
3
10
4
10
5
10
6
10
7
10
8
10
9
SMES
Figure 2.1: Indication of the working areas of diﬀerent energy storage systems [4]
22 2.4 DG GridConnection Characteristics
2.4 DG GridConnection Characteristics
Public electrical power systems operate as AC (Alternating Current) systems
standardized at either 50 Hz or 60 Hz. Power is mostly generated by means of
synchronous generators. The synchronous generators driven by their turbines
are responsible for maintaining the frequency in the system. The frequency
of the system voltage is directly proportional to the speed of the synchronous
generators, i.e.,
f =
p
60
n
syn
, (2.15)
where n
syn
denotes the synchronous speed of the generator in revolutions per
minute [rpm], p the number of pole pairs of the magnetic ﬁeld circuit, and f
the frequency of the generated voltage [Hz].
Due to the variety of primary energy sources/prime movers, DG can generate
electricity by means of either rotating electrical machines or static electrical
generators. When the primary energy is converted into mechanical energy that
is used to drive electric rotating machines (synchronous or induction machines),
AC power is generated. If this AC power is generated at the system frequency
or close to it, the generator can be directly coupled to the grid. However, if
the frequency deviates from the system frequency, power electronic interface
must be used. This may occur if the primary energy sources are intermittent
in nature (e.g. wind, tidal, wave) and if it is economically better to adapt the
speed of the generator accordingly. DG may also generate AC power by means
of a fastrotating prime mover (e.g. a microturbine). In this way, AC power
is generated at a constant, but higher frequency than that of the grid. Also in
this case, an interface is required. When DC power is generated (i.e. f = 0),
as with solar panels and fuel cells, an interface that converts DC to AC (at the
system frequency) is a must.
In this way, the DG connection to the power grid can be classiﬁed into two
categories:
• Direct gridconnected DG.
• Indirect gridconnected DG.
2.4.1 Direct GridConnected DG
Figure 2.2 shows a schematic diagram of a DG unit connected directly to the AC
grid. The prime mover operates at a constant speed, and drives the generator.
In general, this generation (or conversion) can be done by means of either a
synchronous or an induction generator.
DG units equipped with a synchronous generator
For a synchronous generator, (2.15) is applicable. By controlling the prime
mover, so that it operates at a constant speed, the generator can produce power
at the grid frequency. This is the case in steam plants, gas turbines, combined
cycle plants and cogeneration plants. The diﬀerence is in the energy source
that drives the prime mover.
Distributed Generation 23
Prime
mover
Rotary
electrical
generator
Constant
speed
Primary
energy
source
Grid
frequency
AC
Figure 2.2: Schematic diagram of direct gridconnected DG
DG units equipped with an induction generator
When the prime mover does not operate at a constant speed, an induction
generator may be used. In this case, n is no longer constant and
n = (1 −s)n
syn
, (2.16)
where s represents the slip.
Induction generators are usually applied in small hydropower plants and
older design or small wind turbines. In this case the speed of the induction
generator may vary with the turning force (moment, or torque) applied to it.
In practice however, the diﬀerence between the rotational speed at peak power
and at idle is very small, about 1 per cent [69]. Usually, a gearbox is used
(Figure 2.3) to connect the lowspeed driving shaft to the highspeed generator
shaft (1200 to 1500 rpm).
Prime
mover
Gearbox
Low
speed
Primary
energy
source
Rotary
electrical
generator
Grid
frequency
AC
High
speed
Figure 2.3: Schematic diagram of grid connected DG via gearbox
2.4.2 Indirect GridConnected DG
A power system operates at a constant system/grid frequency. Several DG types
generate electricity as DC (e.g. solar panels and fuel cells), highfrequency AC
(e.g. microturbines) or AC with variable frequency (e.g. certain types of wind
turbines). Therefore, an interface is necessary to connect these devices to the
grid. As such, a DG unit is connected to the grid in an indirect way.
24 2.4 DG GridConnection Characteristics
For indirect gridconnected DG, we basically distinguish two situations:
• DG generating DC.
• DG generating either highfrequency AC or AC with a variable frequency.
• Induction generator with power electronic converter in the rotor.
DG generating DC
A DG unit with DC output is primarily characterized by static electric genera
tion, i.e. no rotating parts are involved. Examples of these kind of DG units are
fuel cells and solar panels. Figure 2.4 shows a simpliﬁed layout of such a plant.
The primary energy sources are converted into electricity without of a rotating
electrical machine. The DC output may be ﬂuctuating and is smoothed by a
capacitor before converted to AC at the grid frequency. In addition, a ﬁlter can
be implemented at the outputstage of the inverter to clean the AC voltage.
Static
electrical
generator
DC link
capacitor
DC to AC
converter
Grid
frequency
AC DC DC
Primary
energy
source
Powerelectronic interface
Figure 2.4: Interfaceconnected DG with DC output
DG generating high/variablefrequency AC
Some DG units, such as microturbines, wind turbines and tidal power genera
tors, use rotating electrical machines for electricity generation but are connected
to the grid via powerelectronic interfaces. There are two situations in which
powerelectronic converters are needed to interface the rotating electrical ma
chine to the grid:
• When the rotating electrical machine generates a highfrequency AC (far
beyond the grid frequency).
• When the primary energy sources cause the prime mover to drive the
rotating electrical generator at a variable speed, leading to a variable
frequency AC.
This is illustrated in Figure 2.5. The highfrequency AC, or AC with variable
frequency, is rectiﬁed into DC. A capacitor is used to smooth the DC, before it
is converted into gridfrequency AC. A ﬁlter can be implemented to clean the
resulting AC voltage.
Distributed Generation 25
AC to DC
converter
Prime
mover
Rotary
electrical
generator
Variable or
Very high
speed
Primary
energy
source
Variable or
High
frequency AC
DC link
capacitor
DC to AC
converter
Grid
frequency
AC DC DC
Powerelectronic interface
Figure 2.5: Interfaceconnected DG with AC output
Induction generator with power electronic converter in the rotor
The stator windings of a variable speed induction generator can be connected
directly to the grid with the rotor windings connected to (bidirectional) power
electronic interface (Figure 2.6). The mechanical and electrical rotor frequencies
are controllable over a certain range and the electrical stator and rotor frequency
can be matched, independently of the mechanical rotor speed [28], [75].
Table 2.2 summarizes the grid connection classiﬁcations of DG.
Prime
mover
Rotary
electrical
generator
Variable
speed
Primary
energy
source
Grid
frequency
AC
DC DC
Gearbox
AC to DC
converter
DC link
capacitor
DC to AC
converter
Powerelectronic interface
Figure 2.6: Induction generator with power electronic converter in the rotor
26 2.5 DG Prospects: ConverterConnected DG
Table 2.2: Direct and Indirect gridconnected DG
Rotating Static Direct Indirect
DG Technology Electrical Generator Generator Grid Grid
(AC output) (DC output) Connected Connected
Conventional Generators
√ √
Microturbines
√ √
CHP Plants
√ √
Small HydroPower Plants
√ √
Wind turbines
√ √ √
Photovoltaics
√ √
Fuel cells
√ √
Geothermal Power Plants
√ √
Biomass Power Plants
√ √ √ √
Tidal Power Plants
√ √ √
Wave Power Plants
√ √ √
2.4.3 Connecting Energy Storage to the Grid
Energy storage can be connected to the grid by means of powerelectronic in
terfaces too. When the energy storage device is charged/discharged by DC
(e.g. batteries, hydrogen fuel cells, ﬂow batteries, ultracapacitors, and SMES
systems), powerelectronic inverters are necessary to convert the stored energy
to gridfrequency AC (and vice versa). When a highspeed ﬂywheel is used,
a powerelectronic interface is necessary to convert the highfrequency AC to
gridfrequency AC.
When the energy is stored by means of pumpedhydroelectric plants and
compressedair systems, rotating electrical generators are used to extract the
stored energy and convert it into electricity (gridfrequency AC). In this scheme,
the generator can be connected directly to the grid.
2.5 DG Prospects: ConverterConnected DG
A largescale implementation of DG units may be foreseen, with emphasis on
renewables. Therefore, a large amount of generation will be ’hidden’ behind
powerelectronic interfaces. This will certainly exert a major inﬂuence on the
power system and its operation. Powerelectronic converters can be used to
maximize the energy yield, for active and reactive power control, and for power
quality improvement [60]. The role of powerelectronic interfaces will become
even more important if hybrid networks that combine AC and DC, as well as
ﬂexible energy storages are implemented [19].
When powerelectronic converters are used to interface DG with the grid,
the impact of the protection scheme should be considered, as it may inﬂuence
the power system dynamic behavior [58], [60]. There are two possible pro
tection systems for DG units that are connected to the power system via a
Distributed Generation 27
powerelectronic interface. The ﬁrst type of protection systems automatically
disconnects the DG from the power system when the voltage of the system
drops below a certain level, and reconnects the DG as soon as the voltage re
covers. The other keeps the DG connected to the power network during a fault
(ridethrough capability). In this thesis, powerelectronic interfaced DG with a
protection scheme of the ﬁrst type is considered. However, as comparison, the
impact of powerelectronic interfaced DG with a protection scheme of the latter
type is simulated in Sections 4.4 and 5.5.
Currently, most technical standards require a protection scheme of the ﬁrst
kind [3], [24]. This is mostly motivated by the following objectives:
• To preserve selective protection with simple overcurrent relays as usually
applied in the radiallyoperated MV and LV networks.
• To prevent the DG units from remaining connected to the grid during the
reclosure dead time; they could keep the network energized and negate the
selfextinction of arcing faults and in case of permanent faults the presence
of DG still connected to the grid can be dangerous for utility personnel
during repairs.
2.6 Summary
Many deﬁnitions of DG exist. From the wellknown CIGRE Working Group
37.23 [12], DG is basically deﬁned as electrical generation which is not centrally
planned and dispatched, and is mostly connected to the distribution network.
A broader and more straightforward deﬁnition of DG is advocated by [1] and
[52], which includes all electrical generation sources connected directly to the
distribution network or on the customer side of the meter.
An overview of the stateoftheart of DG technology is given. Although
its classiﬁcation as DG depends on the way a generator is incorporated into a
power system, in many literature, however, DG is distinguished based on the
primary energy sources [7], [52]. The reason is that many generator units driven
by renewable sources of energy inherently possess DG characteristics. In this
chapter, the existing DG technologies are brieﬂy highlighted according to the
primary energy sources.
Another way of examining DG is by looking at the interactions between the
generator and the grid. In this way, there are two classiﬁcations: (1) based
on the output power characteristic, and (2) based on the way DG is connected
to the network. Although these two aspects are wellknown, these classiﬁca
tions are seldom explicitly presented in literature. In this chapter, however,
this distinction is made, as it gives basic ideas and illustrations for supporting
modeling of DG used in the simulation in the later chapters. The classiﬁcation
of DG based on the output power characteristics gives basic ideas on stochastic
approach done in Chapter 6. The classiﬁcation of DG based on the way a DG
is connected to the network provides a basic idea on modeling of DG for power
system transient simulation used in Chapters 35.
28 2.6 Summary
Literature survey on DG characteristics reveals that the output power of non
controllable DG units, especially the ones driven by renewable energy sources,
can show high outputpower ﬂuctuations. Energy storage systems can be applied
to smooth these intermittent eﬀects.
This chapter ends with an overview of powerelectronic interfaced DG. The
basic elements of protection schemes, taken from literature, of such DG units
are highlighted.
Chapter 3
Stability of Systems with
DG
A requirement of power system operation is to balance the electricity supply
and the demand at any time, including grid losses. A properly designed and
operated power system should be able to maintain this balance both under
normal conditions (steady state) as well as after disturbances (dynamic).
A power system is always dynamic. Even under normal operating condi
tions, both active and reactive demands continuously change. As a system with
interconnected machines and components may cover a wide geographical area, a
power system is often subjected to disturbances. For a reliable service, it is key
that the system remains in operation and is able to return to a stable state. The
ability of a power system to return to a state of operating equilibrium under
normal conditions and to regain a new equilibrium after being subjected to a
disturbance is deﬁned as power system stability [34].
The stateoftheart DG technologies presented in Chapter 2 leads to the
classiﬁcation of DG based on the output power behavior and grid connection
characteristics. This chapter emphasizes on the DG impact on the transmission
system stability. An introduction on the phenomena of power system stability
is the start of this chapter. The connection of this chapter with Chapter 2
is mainly found in Section 3.4, where the modeling of DG technologies and
software used for large transmission system transient simulations are presented.
3.1 Classiﬁcation of Power System Stability
All measured (or calculated) physical quantities, such as the magnitude and
phase angle of the voltage at each bus and the active/reactive power ﬂowing
in each line, describe the operating condition of a power system. If they are
constant in time, the system is in steady state. When this steadystate condition
is subjected to a sudden change or a sequence of changes in the system quantities,
the system undergoes a disturbance from its steady state [25].
30 3.1 Classiﬁcation of Power System Stability
Depending on their origin and magnitude, there exist small and large dis
turbances. For small disturbances, a change from a steadystate condition can
be analyzed by using the linearized system’s dynamic and algebraic equations.
Small variations in load and generation are examples of such small disturbances.
Disturbances like transmission system faults, large load changes, loss of gener
ating units, and line switching are examples of large disturbances [25]. In these
cases, the linearized dynamic and algebraic equations are no longer valid.
A system is said to be steadystate stable, if it is able to return to essen
tially the same steadystate condition of operation after being subjected to a
small disturbance. In many cases, especially under large disturbances, a system
reaches a new acceptable steadystate condition is diﬀerent from the original
steadystate condition. The system is called transiently stable.
One can distinguish three types of instability mechanisms depending on
which parameters are most aﬀected by the disturbance: rotor angle, voltage
and frequency.
Furthermore, power system stability can be further classiﬁed based on the
typical ranges of the time period of response actions as shortterm (seconds), or
longterm stabilities (minutes).
The various types of power system stability can be classiﬁed correspondingly
as the diagram in Figure 3.1 [34].
Power system
stability
Rotor angle
stability
Frequency
stability
Voltage
stability
Small signal
stability
Transient
stability
Large
disturbance
voltage stability
Small
disturbance
voltage stability
Short Term
Long Term Short Term
Long Term Short Term
Figure 3.1: Classiﬁcation of power system stability [34]
3.1.1 Rotor Angle Stability
Rotor angle stability concerns the ability of interconnected synchronous ma
chines in a power system to remain in synchronism under normal operating
Stability of Systems with DG 31
conditions and after being subjected to a disturbance. Instability that may re
sult occurs in the form of increasing angular swings of some generators, which
leads to the loss of synchronism [34].
When the disturbance is relatively small, stability (which is then called small
signal stability) can be analyzed using a linearized set of system equations. In
case of transient stability, however, the resulting system response involves large
excursions of generator rotor angles, and is inﬂuenced by the nonlinear power
angle relationship [11], [34]. The latter is mainly focused on in this work.
3.1.2 Voltage Stability
Voltage stability is deﬁned as the ability of a power system to maintain steady
acceptable voltages at all buses in the system under normal operating conditions
and after being subjected to a disturbance. The instability that may result
occurs in a progressive and uncontrollable drop in voltage [34]. Voltage stability
can be classiﬁed into small disturbance voltage stability when it is concerned
with a system’s ability to control voltages following small perturbations such as
small changes in loads, or large disturbance voltage stability when it is concerned
with a system’s ability to control voltages following large disturbances such as
transmission system faults [11], [34]. Analyzing voltage stability is not in the
scope of this thesis.
3.1.3 Frequency Stability
Frequency stability is the ability of a power system to maintain the frequency
within an acceptable range, following a system upset resulting in a signiﬁcant im
balance between generation and load [34]. The instability may result in switch
ing oﬀ of generators, overloading of lines and most probable splitting up the
system in subsystems. Frequency stability could concern any disturbance and
therefore it is not classiﬁed into the small nor the large disturbances. Diﬀerent
from rotor angle stability, frequency stability is determined by the overall re
sponse of the system (or each island, in case of the system split into islands) as
evidenced by the mean frequency, rather than the relative motion of generators.
Frequency is only touched upon in Chapter 7, where keeping power balance in
power system with Active Distribution Systems will be discussed.
3.2 Rotor Dynamics of Synchronous Machines
As long as a power system can rely on synchronous machines for generation of
electrical power, a necessary condition for satisfactory system operation is that
all synchronous machines remain in synchronism [34].
3.2.1 Swing Equation
The governing equation for rotor motion of a synchronous machine is based
on the elementary principle in dynamics which states that the net accelerating
32 3.2 Rotor Dynamics of Synchronous Machines
torque, T
a
, is the product of the total moment of inertia of the rotor, J, and its
angular acceleration [25]
J
d
2
θ
m
dt
2
= T
a
= T
m
−T
e
, (3.1)
with θ
m
the angular displacement of the rotor with respect to a stationary
axis. In (3.1), T
m
and T
e
are the driving torque of the prime mover and the
net electromagnetic torque, respectively. The machine is said to be working in
synchronous speed (or in synchronism) if T
a
= 0, i.e. T
m
= T
e
.
For a given speed of synchronously rotating reference axis, ω
sm
, the angular
displacement can be rewritten as
θ
m
= ω
sm
· t +δ
m
, (3.2)
where δ
m
as the angular displacement of the rotor from the synchronously ro
tating reference axis. By introducing the angular velocity of the rotor from the
synchronously rotating reference axis ω
m
=
dθ
m
dt
for a convenient notation and
twice diﬀerentiating (3.2) with respect to time, combining it with (3.1) and re
calling that power (P) equals torque (T) times angular velocity (ω) one arrives
at the equation
M
d
2
δ
m
dt
2
= P
a
= P
m
−P
e
, (3.3)
where M = Jω
m
is the inertia constant of the machine, and P
a
, P
e
and P
m
are the accelerating, the electrical and the mechanical power, respectively. The
above equation (3.3) can be further normalized in terms of unit inertia constant
H, deﬁned as the kinetic energy at rated speed divided by the rated apparent
power of the generator S
mach
as [25]
H =
1
2
Jω
2
sm
S
mach
(3.4)
or
H =
1
2
Mω
sm
S
mach
, (3.5)
yielding
2H
ω
sm
d
2
δ
m
dt
2
= P
a
= P
m
−P
e
. (3.6)
Moreover, by noting that both δ
m
and ω
sm
are expressed as mechanical speed,
provided both ω
s
(the synchronous speed of the rotor) and δ (the angular dis
placement of the rotor from the synchronously rotating reference axis) have
consistent units, (3.6) can be written as [25]
2H
ω
s
d
2
δ
dt
2
= P
a
= P
m
−P
e
. (3.7)
Equation (3.7) is called the swing equation of the machine. It is the funda
mental equation in the stability study, governing the rotational dynamics of a
Stability of Systems with DG 33
synchronous machine. The graph of its solution is known as the swing curve.
The dynamic behavior of a synchronous machine can be described by means of
this equation and its graph, given initial conditions for speed and angle. An
inspection of the swing curves of all machines of the system can be used to show
whether a particular machine remains in synchronism after being subjected to
a disturbance [25].
3.2.2 PowerAngle Equation
When using the swing equation (3.7) H and ω
s
are known parameters of the
synchronous machine. Therefore, the behavior of the angular positions of the
rotor of synchronous machines (δ) are dictated only by P
m
and P
e
for a given
machine and rated system frequency. Due to the characteristic of the prime
mover and the related controllers, P
m
can be assumed constant during a tran
sient disturbance. This assumption is based on the fact that, while the electrical
network reacts almost instantaneously under disturbances, the turbine has some
delay before its control mechanism causes it to react. Therefore, only P
e
is es
sential for the solution of the swing equation. This is valid for time frames
smaller than 10 seconds (short term stability).
The machine will always operate at steadystate synchronous speed if P
e
=
P
m
. If P
e
= P
m
, e.g. in a response to a disturbance, the rotor deviates from
synchronous speed.
The behavior of P
e
is better explained using a generic twomachine system
as shown in Figure 3.2. This system consists of a synchronous generator(1),
connected to a large external system (2), modeled as a machine with very large
inertia, a so called “inﬁnite bus”. In this system, a simple model consisting
of a constant voltage behind a transient reactance is used to represent both
machines.
Transmission Network
E
1
E
2
1 2
1 2
Figure 3.2: Schematic diagram of the twomachine system for stability stud
ies. Transient reactances associated with the transient internal voltages of both
machines E
1
and E
2
are included in the transmission network [25]
34 3.2 Rotor Dynamics of Synchronous Machines
The admittance matrix
Y
bus
=
¸
Y
11
∠θ
11
Y
12
∠θ
12
Y
21
∠θ
21
Y
22
∠θ
22
(3.8)
represents the transmission network between bus 1 and 2, including the transient
reactances of the two equivalent machines. The electric power output of the
feeding generator P
e
is determined by
P
e
= P
c
+P
max
sin(δ −γ), (3.9)
where
P
c
= E
1
2
Re(Y
11
), P
max
= E
1
E
2
Y
12
, (3.10)
and δ = δ
1
− δ
2
, with δ
1
and δ
2
the angular displacement of the rotor from
the synchronously rotating reference axis associated with the transient internal
voltages E
1
and E
2
. Furthermore, γ = θ
12
− π/2, with θ
12
= arg (Y
12
). If the
network resistance is neglected, such that all the elements of Y
bus
are purely
imaginary (susceptances), then P
c
= 0 and γ = 0, and (3.9) reduces to
P
e
= P
max
sin δ, (3.11)
where P
max
= E
1
E
2
/X
12
with X
12
the transfer reactance between E
1
and E
2
,
corresponding to Y
12
.
3.2.3 Equal Area Criterion
Combining (3.7) and (3.11), yields
d
2
δ
dt
2
=
ω
s
P
m
2H
−
ω
s
P
max
2H
sin δ, (3.12)
or
d
2
δ
dt
2
=
ω
s
P
m
2H
−
ω
s
E
′
1
E
′
2
2HX
12
sin δ. (3.13)
The nonlinearity is clear even for this simple example.
Formal solutions of such an equation cannot be explicitly found. In this case,
one could rely on numerical methods to obtain the solution [25]. For illustration
purposes a twomachine system is commonly used in many text books. In this
case, the examination of the system stability can be done with a direct approach
without solving the swing equation. It is assumed that a temporary threephase
fault occurs at bus1 (Figure 3.3). The fault is cleared after a certain period of
time, without disconnecting any transmission equipment. The plotted power
angle curve of most critical stable situation is shown in Figure 3.4.
The original (steadystate) operation is characterized by the (prefault) rotor
angle δ
0
, located at the crossing of the horizontal line P = P
m
with a curve
drawn and sketching the original P
e0
curve. For simpliﬁcation purposes, a three
phase to ground fault through no intermediate fault impedance is assumed. The
Stability of Systems with DG 35
*
Figure 3.3: Example of twomachine system for stability studies. A fault is
applied at bus1
P
P
m
P cr max
= 
0
P
eD
= 0
P
e0
= P
eP
A
acc
A
dec
P
max
= P
max
sin
Figure 3.4: Plotted power angle curves showing the critical clearing angle δ
cr
.
P
e0
, P
eD
and P
eP
represent the airgap power as well as the terminal power
before (prefault), during and after the fault (postfault), respectively. The
accelerating area A
acc
and the decelerating area A
dec
are equal
eﬀective transmission system is unaltered except when the fault is present. The
short circuit is eﬀective at bus1 so that the electrical power output from the
generator, P
eD
, is zero until the fault is cleared. So, the duringfault curve P
eD
equals zero and the postfault curve P
eP
equals the original P
e0
curve. The
postfault equilibrium point is determined by the intersection of P
m
with P
eP
(equals P
e0
) and this provides δ
P
. Respectively, the unstable equilibrium point
(UEP) is equal δ
max
= π −δ
0
. If the angle δ reaches a value larger than UEP,
there is insuﬃcient decelerating energy. This leads to instability [82].
The Equal Area Criterion (EAC) simply recognizes the fact that the faulted
system is still capable of recovering stability as long as the inequality A
acc
<
A
dec
is satisﬁed. The borderline case corresponds to
A
dec
= A
acc
. (3.14)
36 3.3 System Stability Indicators
Considering the applied fault and the powerangle curve (Figure 3.4), the accel
erating A
acc
and the decelerating area A
dec
can be written as
A
acc
=
δ
cr
δ
0
P
m
dδ, A
dec
=
δ
max
δ
cr
(P
max
sin δ −P
m
)dδ. (3.15)
By solving (3.15) in combination with (3.14), the system reaches the critical
clearing angle (CCA, denoted by δ
cr
) at
δ
cr
= cos
−1
((π −2δ
0
) sin δ
0
−cos δ
0
). (3.16)
Integrating (3.7) twice, provided that P
a
= P
m
(P
e
, i.e. P
eD
, is zero during
the disturbance), at the instant of the critical fault clearing, the increase in
rotor speed and the angle separation between the generator and the inﬁnite bus
become
δ(t)
t=t
cr
=
ω
s
P
m
4H
t
2
cr
+δ
0
. (3.17)
The corresponding critical clearing time (CCT) is obtained as
t
cr
=
4H(δ
cr
−δ
0
)
ω
s
P
m
. (3.18)
3.3 System Stability Indicators
To assess the performance of power systems indicators are needed. As it has
been derived from the Equal Area Criterion concept, the Critical Clearing Angle
(CCA) of (3.16) and the Critical Clearing Time (CCT) of (3.18) indicate a
borderline situation where the faulted system is capable of recovering stability
as long the angle of the synchronous machine is less than the CCA or the fault
clearing time is shorter than the CCT. On the other hand, when both CCA and
CCT are surpassed, the rotor of the machines speeds up, and the rotor angle
increases without limit. Therefore, considering this clear analytical/explicit
relationships between the CCA/CCT and power system stability, the CCA and
CCT are often used as power system stability indicators. When a fault is applied
in a power system, the diﬀerence between the actual clearing angle and actual
clearing time and their respective critical counterparts (CCT and CCA) deﬁnes
the “stability margin” of the system (assuming the system is stable and there
is a positive margin).
However, when simulations are to be done on large power systems, as well as
on a more general case than a simple power system (e.g. twomachine system or
one machine inﬁnite bus system) the CCA and CCT cannot be explicitly found
without computer simulation [25]. One of the practical methods for determining
CCT (and CCA) of a power system is the timedomain numerical integration
method (the stepbystep time domain simulation) [82], where the determination
of CCT (and CCA) requires several time domain simulation runs.
Stability of Systems with DG 37
In this work more practical transient system indicators are used, as proposed
ﬁrstly in [71], namely:
• Maximal rotor speed deviation.
• Oscillation duration.
3.3.1 Maximum Rotor Speed Deviation
The maximum rotor speed deviation is deﬁned as the maximum rotor speed
value attained during the transient phenomenon [71].
This indicator is proposed to assess the rotoranglestability performance of
(centralized) synchronous generators that drive a transmission system with lim
ited inertia. It suggests that the more/faster the rotor speed (of the synchronous
generators) deviates from the rated value when a disturbance occurs, the more
instable the system becomes. Thus when two cases are compared, as a fault is
simulated, at a certain clearing time a higher maximum rotor speed deviation
(the faster the rotor accelerates) suggests a lower stability margin.
A remark should be made here. In this work, a test system with limited
inertia is used (Section 3.5.1). The system is considered stable if after a fault
(disturbance) all (or a limited number of) centralized generator units in the test
system remain in synchronism.
3.3.2 Oscillation Duration
The oscillation duration is deﬁned as the time interval between the start of the
fault and the instant after which the rotor speed stays within a bandwidth of
10
−4
pu (on the basis of the rated rotor speed) during a time interval longer
than 2.5 s [71].
In the absence of damping, the rotor would continue to oscillate around the
rotor angle of the operating point (δ
0
in Figure 3.14). As long as this is the case,
there is a risk of instability. In the analysis within this work, when scenarios
are compared, an identical disturbance will be applied in each scenario. The
oscillation of the rotor speed that deviates from the rated speed is recorded.
The indicator of oscillation duration implies that the longer the rotor speed
deviates from the rated value when a disturbance occurs, the more instable the
system is. In other words, the shorter the rotor speed deviates from rated, when
a disturbance occurs, the more stable the system is.
Above all, eﬃciency in computing time is obtained by using these indicators
rather than CCT (and CCA).
Figure 3.5 shows both indicators of maximum rotor speed deviation and the
oscillation duration [71]. To quantify the indicators, (3.19)(3.21) are used:
maximum rotor speed deviation = ω
r,max
−ω
r,nom
, (3.19)
with ω
r,max
and ω
r,nom
denoting the maximum and the rated rotor speed of a
centralized generator [rpm],
oscillation duration = t
osc
−t
f
, (3.20)
38 3.4 DG and Large System Dynamic Simulation
0 5 10 15
−3
−2
−1
0
1
2
3
4
x 10
−3
Time [s]
R
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
max. rotor speed deviation
(ω
r,max
− ω
r,nom
/ω
r,nom
)
t
f
t
osc
t
osc
+ 2.5
oscillation
duration
ω
r
(t+n∆ t) − ω
r
(t)≤ 10
−4
;n=1,...,2.5/∆ t
Figure 3.5: Transient stability indicators: maximum rotor speed deviation and
oscillation duration [71]
with t
f
the time [s] when the fault is applied, and
t
osc
= min{t : ω
r
(t +n∆t) −ω
r
(t) ≤ 10
−4
; n = 1, · · · ,
2.5
∆t
}, (3.21)
with ω
r
(t) the rotor speed at a time t and ∆t the simulation step (in this research
∆t = 10
−3
s is used).
3.4 DG and Large System Dynamic Simulation
In the earliest stages, DG was implemented in power systems on a small scale.
In this case, DG can be treated as negative load in power system dynamics
studies. As DG implementation increases, the penetration level of DG may
become important. In this case, the (original) conventional centralized power
plants (e.g. thermal, nuclear or hydro) remain in operation and still cover a
signiﬁcant part of the load and the voltage and frequency control are still within
the responsibility of the large/centralized synchronous generators. However,
when DG penetration becomes very high, it begins to inﬂuence the overall power
system behavior. Therefore, the dynamic behavior of the existing system due to
this high degree of DG needs to be understood/studied, especially its interaction
with the existing generating equipment, to which the power system stability
problem is mostly related.
Stability of Systems with DG 39
3.4.1 Modeling DG Technologies
As elaborated in Chapter 2, DG can be connected to the grid directly and/or
indirectly via a power electronic converter.
There is a major diﬀerence in the response to voltage dips (that occur during
a disturbance) of electrical machines at the one hand and power electronic con
verters at the other. The response of electrical machines is determined by the
fundamental electromechanical laws and physical construction of the machine.
Machines therefore have an inherent behavior during voltage disturbances. The
power electronic converters response, not only depends on the physical con
struction (e.g. power electronic components cannot withstand large currents),
but also for the largest part on the controller and its parameters. This makes
it more diﬃcult to provide a general model that can be used for simulation in
power system studies [42].
Electrical Rotating Machines
The direct gridconnected DGs are based on directly coupled synchronous or
induction generators. These technologies can be modeled using models of a con
ventional synchronous or induction generator, readily available in power system
dynamics simulations software [53]. Therefore, the available models within the
software can be used and they do not pose special problems in power system
dynamics.
The challenge however lies on how to incorporate the DG in the distribution
system commonly represented only by the load at the main connection between
the transmission system and the distribution feeder in power system simulations.
Section 3.5.3 deals further with this issue.
Converter Connected DG: Voltage Source Converter
When the indirect gridconnected DGs are considered, the output of these DG
technologies are driven by the output of the power electronic interface basically
representing a voltage source converter, the most used converter nowadays. The
example of a circuit topology of such converter is taken from [42] as shown in
Figure 3.6, where a ﬁrstorder ﬁlter (L
f
and R
f
) between the converter and the
grid is used to reduce the harmonics injected by the switching of the converters.
The operating principle of the grid connected voltage source converter can
be summarized as follows [42], [70]:
• A setpoint for the active power to be supplied to the grid is determined by
the electrical conversion either from a static electrical generator (Chap
ter 2) or prime mover and ACDCconverter (Chapter 2).
• A setpoint for the reactive power is derived from the actual value of the
terminal voltage (U
grid
) if a terminal voltage controller is present.
• The actual values of terminal voltage and currents are measured and are
inputs to the current controller.
40 3.4 DG and Large System Dynamic Simulation
• The current controller generates control signals for the semiconductor
switches in such a way that the current that ﬂows, injects the desired
amounts of active and reactive power in the grid.
C U
dc
I
dc
+

U
grid
I
a
I
b
I
c
U
an
U
bn
U
cn
L
f
R
f
Figure 3.6: Threephase full bridge Voltage Source Converter with 6 IGBT
switches [42]
Converter Connected DG: Constant PQSources
When transient studies are performed on large systems, it is usual to model the
DGs as PQsources. Only the active power P and reactive power Q (or active
and reactive current) supplied by the converter are of interest then. This is espe
cially advantageous in power system dynamics simulations, because it facilitates
the interaction with the load ﬂow module, that uses active and reactive power
as an input, as it is required by power system dynamics simulation packages.
Moreover, the grid representation in the simulation packages and the typical
time step used, do not allow detailed modeling of power electronics. The use of
this PQsource model for DG is justiﬁed in [70]. The steady state values and
the time required to reach a new set point for this PQsource model is similar
to both the voltage source converter and the controlled current source converter
with hardly any overshoot occurring [70]. Applying this model to power system
dynamic simulation can done basically by following Algorithm 1:
Algorithm 1 Calculate I
DG
Require: P
nom
; Q
nom
P
DG
⇐ P
nom
Q
DG
⇐ Q
nom
−−→
I
DG
⇐
(P
DG
+jQ
DG
)
∗
−−−→
U
DG
P
nom
and Q
nom
denote the nominal/setpoint of the active and reactive power
output of the converter connected DG unit. P
DG
and Q
DG
denote the ‘real’
Stability of Systems with DG 41
active and reactive power output of the converter connected DG unit.
−−→
I
DG
denotes the current generated by the DG unit and
−−→
U
DG
the terminal voltage
where DG unit is connected (as complex phasors).
Due to the diﬀerent protection schemes of converter connected DG, addi
tional equations should be applied to Algorithm 1.
When the protection system automatically disconnects DG from the network
when the voltage of the system drops below a certain level U
min
(e.g. 0.85 pu)
and reconnects the DG as soon as the voltage recovers, Algorithm 2 is applied.
Algorithm 2 Calculate I
DG
Require: P
nom
; Q
nom
; U
min
if U
DG
≥ U
min
then
P
DG
⇐ P
nom
Q
DG
⇐ Q
nom
−−→
I
DG
⇐
(P
DG
+jQ
DG
)
∗
−−−→
U
DG
else
P
DG
⇐ 0
Q
DG
⇐ 0
−−→
I
DG
⇐ 0
When the protection system keeps the DG connected to the power network
during a fault, Algorithm 3 is applied. Note that a current limiter is commonly
applied in the latter case, due to the limitations (“by construction”) of the
converter. Two ‘minimal’ voltage levels are commonly applied here. When the
voltage of the system drops below the ﬁrst voltage level U
min
1
(e.g. 0.85 pu),
the output current of the DG is limited to the maximum value I
max
(in practice,
100% up to 120% of the rated value). Further, when the voltage drops below
the second voltage level U
min
2
(e.g. 0.15 pu), the DG is disconnected from the
network.
Most DG based on renewable energy generation is intended to maximize the
active power output. Therefore it is common to set a PQsource to represent the
converter connected DG for these DG units as a source that delivers constant
active power P and limited amount of reactive power Q [71]. On the other hand,
the current limiter and the operating voltage applied in the DG gives a limitation
in the total power S
nom
=
P
2
nom
+Q
2
nom
. Therefore, for a given operating
active power P setting, the Q must be within a certain area, as constrained by
the total apparent power.
3.4.2 Power System Dynamics Software Packages
For the investigation of DG impact on power system transient stability, the
above DG models are integrated in the various simulation software packages,
each with their own modeling assumptions and limitations, as discussed below.
42 3.4 DG and Large System Dynamic Simulation
Algorithm 3 Calculate I
DG
Require: P
nom
; Q
nom
; U
min
1
; U
min
2
; I
max
if U
DG
≥ U
min
1
then
P
DG
⇐ P
nom
Q
DG
⇐ Q
nom
−−→
I
DG
⇐
(P
DG
+jQ
DG
)
∗
−−−→
U
DG
else if U
min
1
≥ U
DG
≥ U
min
2
then
I
DG
⇐ I
max
P
DG
⇐ Re(
−−→
U
DG
−−→
I
DG
∗)
Q
DG
⇐ Im(
−−→
U
DG
−−→
I
DG
∗)
else {U
DG
< U
min
2
}
P
DG
⇐ 0
Q
DG
⇐ 0
−−→
I
DG
⇐ 0
There is a lot of power system dynamic simulation software available on
the market today. Within this work, three power system software packages are
utilized: PSS/E, MATLAB Power Systems Blockset and RTDS (which is not
really software, but a “hardware digital simulator”). The capabilities of these
three tools complement one another.
Basically, power system dynamics simulation software can be used when the
phenomena of interest have a frequency of about 1 to 10 Hz. PSS/E falls within
the software packages that oﬀer dynamic simulation capability. In this type
of software, only the fundamental harmonic component is simulated, whereas
higher harmonics are neglected. This approach enables the representation of the
network by a constant admittance matrix, as in load ﬂow calculations. Further,
it reduces the number of diﬀerential equations, as no diﬀerential equations are
associated with the network and fewer with generating equipment and as it
enables the use of a larger simulation time step [34]. As a result of this approach,
power system dynamics simulation software alternately executes a load ﬂow
and a timedomain calculation, via integration of the diﬀerential equations that
model dynamic system devices. From the load ﬂow calculation, node voltages
and branch ﬂows result. During the timedomain calculation, the response of
the dynamic device models to changes in their terminal voltage, current and/or
frequency are determined. Therefore the use of this software is preferred when
many simulation scenarios of DG implementation need to be run in a large
power system like the simulations in this work in Chapters 4 and 5. Here, the
converter connected DG is modeled as a constant PQSource.
When the frequencies of interest are higher, the packages that contain more
detailed and higher order equipment models should be used. MATLAB Power
Systems Blockset and the Real Time Digital Simulation (RTDS) fall within this
category. In these packages for examples, the transmission network is repre
sented in threephase timedomain and a time step down to 1 µs is possible.
Therefore the modeling of DG units in details as Voltage Source Converter or
Stability of Systems with DG 43
Current Source Converter is only possible when such software is used. However
this small step will have an impact on the execution speed, especially when sim
ulation is done on large power systems. In the (rotor angle) transient stability
studies performed in Chapters 46, the phenomena of interest have a frequency
of between 1 to 10 Hz. Therefore, this level of detail is not needed, and power
system dynamics software package PSS/E is used.
Particularly in Chapter 7 RTDS is used. The RTDS models of the voltage
and current source converters that represent DG in that chapter are derived
from the model earlier developed in MATLAB Power Systems Blockset [42].
3.5 Simulation Setup
This work deals with issues of DG aﬀecting the bulk transmission system when
a high DG penetration level is seen.
For this purpose, a model of the transmission system is needed. This model
should be detailed enough to assess transient stability. At this level there are
two possible data sets to use. The ﬁrst possibility is using a real national
transmission system, for example, the high voltage 380kV Dutch network. The
second possibility is using an existing test system normally used for dynamic
stability. Here the second option is taken.
3.5.1 The IEEE 39bus New England Test System
The IEEE 39bus New England is a widely known test system used for dynamic
simulations (Figure 3.7 and Table 3.1). As such, this system has a beneﬁt
when compared to other systems, such as the 380kV Dutch High Voltage Grid,
as it has been used extensively and described in literature, and the results
from the simulations presented throughout this work can (to some extent) be
compared with other work. The system is relatively small (39 bus, 10 centralized
generators (CG), 46 transmission lines), but comparable to the 380kV Dutch
Network (28 bus, 19 CG (≥ 250 MW), 27 transmission lines (380 kV and 220
kV)) [73].
Table 3.1: Characteristics of the New England Test System
System Characteristic Value
# of buses 39
# of generators 10
# of loads 19
# of transmission lines 46
Total generation 6140.7 MW / 1264.3 Mvar
Total generation 6097.1 MW / 1408.7 Mvar
44 3.5 Simulation Setup
10
8
1
2
3
5 4
7
6
9
37
25
30
2
1
39
9
8
7
5
4
3
18
17
27
26 28
29
38
24
16
21
35
22
23
36
19
33
20
34
32
31
10
11
13
12 6
14
15
Figure 3.7: Singleline diagram of the 39bus New England test system [49]
In this research, the basic parameters of the IEEE 39bus New England
dynamic test system are taken from [49], and listed in Appendix B. Note that
some minor adjustments have been made. To represent a system with limited
inertia, no bus is modeled as an inﬁnite bus (every generator in this test system
is set to have its inertia). Each CG is modeled as a twoaxis (dq) model of
synchronous machine [53] and is equipped with a simpliﬁed excitation system
model and a steam turbine governor model. The details of these models and
the representative values for the parameters are taken from [53] and [34], and
attached in Appendix B. Each load is equally divided in constant impedance,
constant power and constant current.
3.5.2 DG Technology
Two basic technologies can be used for connecting DG to the grid (Section 3.4.1):
(i) electrical rotating machine connected directly to power grid, or (ii) converter
connected generator indirectly connected to the grid. In this work, some options
elaborated from both basic models are used to investigate the general impact of
the DG on the system stability, namely:
• Squirrel cage induction generator (ASM), simulated by means of a third
order induction generator model [53].
Stability of Systems with DG 45
• Synchronous generator without grid voltage and frequency control (SM),
as discussed in [71].
• Synchronous generator with grid voltage and frequency control (SMC) [71].
• Power electronic converter (PE), modeled as a constant PQsource [71].
• Power electronic converter (modeled as a constant PQsource), with grid
voltage and frequency control (PEC) [71].
In case of ASM and SM , the electrical rotating machine is modeled based on
the existing models available in the power system dynamic simulation software
(Appendix C). Furthermore, PE is modeled by a source of constant active
power and reactive power (PQSource). Since a standard model is not available
for representing power electronics in the version of PSS/E version used [53], a
socalled userwritten model of a power electronic converter has been developed.
This model is based on the algorithms 2 and 3 in Section 3.4.1. Controllers are
applied when DG with grid voltage and frequency control is simulated (SMC
and PEC). When SMC is considered, each DG is equipped with the simpliﬁed
excitation system and a steam turbine governor model, as listed in Appendix C.
When PEC is simulated, the active and reactive power controllers as depicted in
Figure 3.8 are incorporated in the model. The challenge lies in the possibilities
that many types and sizes of DG are incorporated in the system so that the
generators as well and the controller parameters can be optimized. In this
case, it is very complex to optimize the controllers parameters for the DG unit.
Therefore simpliﬁcations are done as only one generic model is used to simulate
one DG technology. Furthermore, once the types and the parameters of the
controllers are set, they will neither be changed nor optimized throughout the
simulations. This issue is also discussed in next section.
U
t,ref
+

Q
ref
K
p
+
Q
Q
+
Q
max
Q
min
U
t
T
i
s
1
ref
+

P
ref
K
p
+
P
P
+
P
max
P
min
Figure 3.8: Reactive (above) and active (below) power controllers of a power
electronic converter with grid voltage and frequency control in the test system
46 3.5 Simulation Setup
3.5.3 Incorporation of DG in Distribution Networks
In test systems used for transient stability studies, the highvoltage transmis
sion system is normally modeled in detail, while the distribution system is rep
resented only by the load at the main connections between the transmission
system and the distribution systems. As described in Section 1.3, the struc
ture of distribution systems is likely to change due to the implementation of
DG. This is because, in an active distribution network, some generation and
energy storage systems are located within the distribution system. Therefore,
the most thorough approach to study the impact of such DG implementation on
the transient stability is to model the distribution networks in detail along with
all load types, distributed generators and energy storage. In many situations,
however, it is not practical to obtain and to apply these details in the lower level
distribution networks on a test system originally used to study the transmission
level. Therefore, a simpliﬁcation must be made using a distribution network
model representing the aggregated load and generation in the system.
In this work a simple approach is used by representing the incorporation
of DG in a distribution network as an equivalent load and generator earlier
proposed in [18] and [71]. Only one DG technology (and not a mixture of
several DG technologies) is assumed to be implemented at one load bus for one
simulation case. After that, comparison of the impacts of one DG technology
on the transmission system transient stability to another are investigated. In
this way, the general model of connecting DG at a particular load bus as shown
in Figure 3.9 can be used [18], [71].
Load Bus DG Bus
DG
jX
TDG
R
LDG
+ jX
LDG
P
Load
+ jQ
Load
Figure 3.9: Model of connecting DG at a particular load bus
In this model the impact of diﬀerent DG topology can be approached by
changing the parameters of the impedance connecting DG (X
T,DG
, R
L,DG
,
X
L,DG
) to the transmission system. X
T,DG
and X
L,DG
represent the reactance
of the transformer and the line between the DG and the transmission network
respectively, and R
L,DG
represents the resistance of the line. For example, to
compare the impacts of DG implemented far away from or close to the connec
tion feeder, the impedance parameters can be set by increasing the parameters
in the ﬁrst case and reducing them for the latter. Also, when a distribution
Stability of Systems with DG 47
network with a singlepoint concentrated DG is compared to a distribution net
work with scattered DG (that suggests a parallel connection), the impedance
can be set diﬀerently (e.g. the line parameters R
L,DG
, X
L,DG
). These issues
are treated in Section 4.2.
3.5.4 Behavior of Centralized Power Plants
Synchronous generators within centralized power plants that deliver the elec
trical energy within a power system are equipped with controllers (governing
systems  governors, excitation systems  exciters) and protection schemes.
The governing systems for example provide a means of controlling power and
frequency, a function commonly referred to as loadfrequency control or auto
matic generation control. The excitation systems provide direct current to the
synchronous machine ﬁeld winding, in order to control the ﬁeld current. The
protective function ensures that the physical (mechanical and electrical) capa
bility limit of the synchronous machine, excitation system and other equipment
are not exceeded [34].
The characteristics of these control systems of course have an impact on
the overall performance of the power system. Basically the eﬀectiveness of
the control systems in enhancing power system stability can be estimated and
further optimized. For example, by adjusting the parameters of these controllers
the behavior of the power system transient stability can be inﬂuenced [34].
When DG is implemented on a large scale it can be expected that the tran
sient performance of the power system changes. It can be expected too that
changing, for instance, the parameters of the controllers could counteract the
inﬂuence on the performance of the system due to this DG implementation. This
suggests that these parameters can be optimized for having the most optimal
system transient stability performance. However, as the focus of this work is to
investigate stability of the power system due to DG implementation (‘vertical to
horizontal’ transformation of power system), we do not include the optimization
of these controllers.
3.6 Summary
This chapter gives an overview of the power system stability phenomena. This
chapter begins with classiﬁcation of power system stability, which is taken from
existing literature [35]. Then, review of existing knowledge is presented to pro
vide insights into the mechanism, which leads to transient instability in power
systems. Highlights are put on the rotor angle stability (of synchronous ma
chines), which is the main focus of this research, by means of the swing equa
tion, powerangle equation and equal area criterion. Attention is drawn on some
of the existing system stability indicators (Critical Clearing Time (CCT) and
Critical Clearing Angle (CCA)), and on more practical transient system indi
cators such as maximal rotor speed deviation and oscillation duration proposed
48 3.6 Summary
by [71]. The latter is used to assess power system stability performance on large
power systems in this work.
Models and assumptions of DG implementation on (large) power system
transient simulation are discussed, leading to the use of existing models of DG
as electrical rotating machines and constant PQsource for converter connected
DG. The basic algorithm for simulation of converter connected DG as constant
PQsource is ﬁrst introduced in [71]. In this thesis, diﬀerent protection schemes
of converter connected DG are added to the basic algorithm. Some power system
dynamics software packages are highlighted, and the motivations that lead to
the use of power system dynamics software package PSS/E in this research are
presented.
This chapter also covers the development of the simulation setup used in
this work: IEEE 39bus New England test system [49]. Five DG connection
technology options are taken into account based on the two basic alternatives:
DG directly or indirectly connected to the grid. The options are: (1) squirrel
cage induction generator, simulated by means of a third order induction gener
ator model, (2) synchronous generator without or (3) with grid voltage and
frequency control models derived from electricalrotatingmachinebased DG
technology (an existing library in PSS/E [53]) and (4) power electronic con
verter [71] modeled as a constant PQsource without or (5) with grid voltage
and frequency control, for which, userwritten model with diﬀerent protection
schemes are developed.
The representation and aggregation of DG in the distribution network [18]
is discussed in this chapter. Finally, the chapter ends with highlights on the
behavior and assumptions of the centralized power plants that are potentially
impacted due the DG implementation.
Chapter 4
Impact of DG on Power
System Transient Stability
This chapter investigates the impact of DG implementation on power system
transient stability. Diﬀerent scenarios of a power system with a high DG pen
etration level are developed. The impact of diﬀerent DG penetration levels,
fault durations and locations, DG gridconnectionstrengths, DG technologies
and protection schemes of powerelectronic interfaced DG units are investigated
and discussed.
The 39bus New England Test System forms the starting point of the in
vestigations. Throughout the chapter, DG technologies and DG topology are
based on the simulation setup deﬁned in Section 3.5.
4.1 DG Impacts
4.1.1 Simulation Scenarios
In this section, the impact of the DG penetration level on the power system
stability is investigated. For this purpose, two simulation scenarios are deﬁned.
In the ﬁrst scenario, Scenario I, the DG is implemented to cover the increment
of the load, so that the DG penetration level rises along with the increasing load.
In the second, Scenario II, DG supplies (part of) the existing (constant) load.
As a result, the total power output from the centralized generators is reduced.
The DG penetration level is deﬁned as [18]
%DG
penetrationlevel
=
P
DG
P
DG
+P
CG
×100, (4.1)
with P
DG
and P
CG
the total active power generated by DG and CG respectively.
50 4.1 DG Impacts
The details of the scenarios are as follows:
Scenario I:
• The DG penetration level is raised to keep track with the increasing real
and reactive power consumption of all loads. The increment of the real
power consumption is covered by an equal amount of power produced by
DG connected to each load bus via a j0.05 pu impedance (representing
X
T
+X
L
in Figure 3.9 with R
L
neglected) on the 100 MVA system base.
The DG penetration level increases in steps of 3.33 % up to 33.33%, cor
responding to a 50% increment of the load i.e. the load increase within
25 years at a load growth of about 1.8%. Thus eleven subscenarios are
obtained with DG penetration levels of 0.0, 3.33, 6.67, 10.0, 13.33, 16.67,
20.0, 23.33, 26.67, 30.0, and 33.33%.
• The active power generated by the CG is kept constant, except for the
active power generated by the generator that acts as the swing bus (gen
erator nr. 2, see Figure 3.7) for covering the losses. The increasing reactive
power consumption is provided by centralized generators.
Scenario II:
• The DG (penetration) level is raised by decreasing the CG active power
output in steps of 3.33 % down to a reduction of 33.33%, and the im
plementation of DG at every load bus to cover this decrement of active
power. In this way, again eleven scenarios are obtained.
• The load remains constant in this scenario.
To assess the results of Scenarios I and II, Base Case I and Base Case II are
additionally deﬁned as:
Base case I:
• The load is increased in steps similar to that in Scenario I. However, the
increasing load is supplied by raising the active power output of the CG.
Thus in this case the centralized generators cover both active and reactive
power increment.
Base case II:
• No increase in load similar to Scenario II and no DG penetration.
4.1.2 Transient Stability Simulation
The transient stability of the test system is investigated by applying a permanent
threephase fault to all possible branches cleared by tripping the faulty line after
150 ms. Every line, that can be missed according the (N−1) adequacy standard,
is subjected to a fault. In this way, 35 possible locations for faulty branches are
simulated. Details are shown in Table 4.1.
To assess the transmission system stability, two transient stability indicators
are examined:
• Maximum rotor speed deviation of large centralized generators.
• Oscillation duration.
Details of the indicators can be found in Section 3.3.
Impact of DG on Power System Transient Stability 51
Table 4.1: Branch Number and Corresponding Buses
Faulty Corresponding Faulty Corresponding Faulty Corresponding
branch buses branch buses branch buses
(nr.) (nr.) (nr.)
1 12 13 89 25 1727
2 23 14 1011 26 2122
3 225 15 1013 27 2223
4 34 16 1112 28 2324
5 318 17 1213 29 2526
6 45 18 1314 30 2627
7 414 19 1415 31 2628
8 56 20 1516 32 2629
9 58 21 1617 33 2829
10 67 22 1621 34 139
11 611 23 1624 35 939
12 78 24 1718
4.1.3 Simulation Results
Scenario I
Figure 4.1 and Figure 4.2 show the simulation results displaying the transient
stability indicators when the DG penetration level is increased according to
Scenario I; the results of Base case I serve as a reference. The xaxis of each
graph represents the number of the faulty branch (Table 4.1). The yaxis rep
resents the DG penetration level [%] and the corresponding load increase [%].
Note that in Base case I the yaxis represents the load increase [%] only since
there is no DG implemented (DG
level
= 0%). The zaxis represents the value of
the stability indicator used (the maximum rotor speed deviation in Figure 4.1,
and the oscillation duration in Figure 4.2). The bottom graphs of Figure 4.1
and 4.2, labeled CG, indicate the simulation results of Base case I. The titles
ASM (squirrelcage induction generator), SM (synchronous generator without
grid voltage and frequency control), or PE (powerelectronic converter without
grid voltage and frequency control) above the graphs indicate the type of DG
technology simulated (Section 3.5.2).
When the increasing load within the test system is covered only by increasing
the CG active power output (Base case I), the stability indicators are generally
increasing (i.e. a reduced stability in the system). When DG is implemented to
cover the increased load within the system (Scenario I), in general the indicators
do not increase (see graphs ’ASM’, ’SM’). Some exceptions can be observed when
powerelectronic interfaced DG (without grid voltage and frequency control) is
implemented (see graph ’PE’), but those can be prevented when the DG is
equipped with grid voltage and frequency control. In general, the indicators
show an improved stability (i.e. decreasing indicators) when the DG is equipped
with grid voltage and frequency control. These results are not visualized here.
52 4.1 DG Impacts
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
0.005
0.01
Faulty branch [nr.]
ASM
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
0.005
0.01
Faulty branch [nr.]
SM
DG level (load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
0.005
0.01
Faulty branch [nr.]
PE
DG level (load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
0 (20)
0 (50)
0
0.005
0.01
Faulty branch [nr.]
CG
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
Figure 4.1: Maximum rotor speed deviation when the DG penetration level is
simulated according to Scenario I (graphs ’ASM’, ’SM’ and ’PE’) and the Base
case I (graph ’CG’), and a fault is simulated in all deﬁned branches (Table 4.1)
Impact of DG on Power System Transient Stability 53
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
10
20
Faulty branch [nr.]
ASM
DG level (Load increase) [%]
o
s
c
.
[
s
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
10
20
Faulty branch [nr.]
SM
DG level (Load increase) [%]
o
s
c
.
[
s
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
10
20
Faulty branch [nr.]
PE
DG level (Load increase) [%]
o
s
c
.
[
s
]
1 5
10
15
20
25
30
35
0 (0)
0 (20)
0 (50)
0
10
20
Faulty branch [nr.]
CG
DG level (Load increase) [%]
o
s
c
.
[
s
]
Figure 4.2: Oscillation duration when the DG penetration level is simulated
according to Scenario I (graphs ’ASM’, ’SM’ and ’PE’) and the Base case I
(graph ’CG’), and a fault is simulated in all deﬁned branches (Table 4.1)
54 4.1 DG Impacts
Scenario II
Figure 4.3 displays the maximum rotor speed deviation when the DG pene
tration level is increased according to Scenario II. The xaxis of each graph
represents the number of the faulty branch. The yaxis represents the DG pene
tration level [%] and the corresponding load increase [%]. The zaxis represents
the value of the maximum rotor speed deviation in per unit. The bottom graph
of Figure 4.3 (labeled ’CG’) indicates the Base case II. In general the maximum
rotor speed deviation decreases. The oscillation duration decreases along with
the increasing DG level too (not visualized here). Both indicators show even
better results when the DG is equipped with grid voltage and frequency control.
4.1.4 Remarks
In the simulation Scenarios I and II, no signiﬁcant (transient) stability problems
were found when the DG penetration level is increased up to 33.33% regardless of
the DG technology used. Note that in these scenarios all centralized generators
(the 10 CGs in the test system) remain in the system when the DG penetration
level is increased.
The results can be explained as follows. In Scenarios I and II, all centralized
generators remain in the system. Therefore, all active and reactive power con
trols as well as the inertia of the centralized generators are unchanged. However,
the active power ﬂows in the transmission lines are lower when DG is imple
mented as the active power generated by the DG is consumed directly by the
load at the same feeder. Thus, the active power ﬂows on the transmission lines
are more or less constant when the DG covers the increasing load (Scenario I).
The active power ﬂows in the transmission lines even decrease when the DG
covers (part of) the existing load (Scenario II). It is known that large power
ﬂows have a detrimental eﬀect on the damping of oscillations [31]: the heavier
the lines are loaded, the weaker the connections between the generators and
the loads and the larger the oscillations of the centralized generators. Thus,
when all CG units remain in the system, and the DG is implemented close to
the loads, the DG implementation is a kind of ‘loadreduction’ that reduces the
power ﬂows in the transmission network and improves its transient stability.
We can compare the active power ﬂows in each of the simulated branches
according to simulation Scenarios I, II and also the Base cases I and II, with
the system indicators. It can be observed that the ’surface’ of the branch power
ﬂows are comparable with that of the system indicators of the scenarios.
Figures 4.44.6 show some examples of these comparisons.
Impact of DG on Power System Transient Stability 55
1 5
10
15
20
25
30
35
0 (0)
16.7 (0)
33.3 (0)
0
0.005
0.01
Faulty branch [nr.]
ASM
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (0)
33.3 (0)
0
0.005
0.01
Faulty branch [nr.]
SM
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (0)
33.3 (0)
0
0.005
0.01
Faulty branch [nr.]
PE
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
1 5
10
15
20
25
30
35
0 (0)
0 (0)
0 (0)
0
0.005
0.01
Faulty branch [nr.]
CG
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
Figure 4.3: Maximum rotor speed deviation when the DG penetration level is
simulated according to Scenario II and the Base case II (graph ‘CG’), and a
fault is simulated in all deﬁned branches (Table 4.1)
56 4.2 DG GridConnection Strength Impacts
1 5
10
15
20
25
30
35
0 (0)
0 (20)
0 (50)
0
500
1000
Branch [nr.]
Base case I
DG level (Load increase) [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
1 5
10
15
20
25
30
35
0 (0)
0 (20)
0 (50)
0
0.005
0.01
Faulty branch [nr.]
CG
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
Figure 4.4: Active power ﬂows in the simulated branches in the test system
(top) and maximum rotor speed deviation (bottom) in Base case I
4.2 DG GridConnection Strength Impacts
In this section, the DG gridconnection strength is varied by changing the pa
rameters of the impedance connecting the DG units to the system, as suggested
in Section 3.5.3.
4.2.1 Distribution Network and DG Layout
The distribution system carries the power to the individual customers, in a cer
tain geographical area. Both the geographical situation and the distribution of
customers can vary from one area to another [21]. Moreover, there are many
ways to connect DG to the distribution network. From the transmission system
point of view, both elements may result in a DG implementation that can have a
weaker or a stronger grid connection even when in both situations the DG pene
tration levels are equal. For example, DG may be implemented in a distribution
network with a relatively low impedance, e.g. distribution networks in towns or
city centers, or a distribution network with a relatively high impedance, e.g. the
distribution network in a rural area with long laterals. DG units may also be
spread throughout a distribution network, such as solar panels mounted on the
Impact of DG on Power System Transient Stability 57
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
500
1000
Branch [nr.]
Scenario I
DG level (Load increase) [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (20)
33.3 (50)
0
0.005
0.01
Faulty branch [nr.]
SM
DG level (load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
Figure 4.5: Active power ﬂows in the simulated branches in the test system
(top) and maximum rotor speed deviation (bottom) in simulation Scenario I
(’SM’ refers to the DG technology implemented)
roofs of houses, or concentrated in a few locations, e.g. a wind park connected
to a distribution network at one substation only. The latter case may result in
a higher impedance between DG and the transmission network.
4.2.2 Simulation Scenarios
The incorporation of DG in a distribution network is represented as an equiv
alent load and generator (Figure 3.9). The impedance between the aggregated
DG (implemented at a particular load bus) and the transmission network is set
according to
Z
DG
i
= jX
T,DG
i
+R
L,DG
i
+jX
L,DG
i
, (4.2)
where Z
DG
i
is the impedance between DG
i
(the aggregate DG connected to
load busi) and the transmission network.
The parameters in (4.2) are varied considering the following points:
• When DG is implemented in laterals of diﬀerent length the value of jX
L,DGi
is varied.
58 4.2 DG GridConnection Strength Impacts
1 5
10
15
20
25
30
35
0 (0)
0 (0)
0 (0)
0
500
1000
Branch [nr.]
Scenario II
DG level [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
1 5
10
15
20
25
30
35
0 (0)
16.7 (0)
33.3 (0)
0
0.005
0.01
Faulty branch [nr.]
SM
DG level (Load increase) [%]
m
a
x
.
d
e
v
.
[
p
u
]
Figure 4.6: Active power ﬂows in the simulated branches in the test system
(top) and the maximum rotor speed deviation (bottom) in simulation Scenario
II (’SM’ refers to the DG technology implemented)
• When DG is spread out over many substations instead of concentrated in
one, the value of jX
L,DGi
is varied related to the “size of DG” implemented
in a particular distribution network:
jX
L,DG
i
=
j0.05pu
par(P
DG
i
)
, (4.3)
where par(P
DG
i
) is an integer whose value is proportional to the ”size of
DG
i
” (the active power generated by the aggregate DG at a particular
load busi).
• When DG is implemented in a distribution network with a high resistance,
the value of R
L,DGi
is taken into account too.
To investigate the transmission system stability in relation to the DG grid
connection strength, various Scenarios (IIIA, IIIB, IVA, and IVB) are de
ﬁned (Table 4.2).
Impact of DG on Power System Transient Stability 59
Table 4.2: Simulation Scenarios
Scenario Keywords Impedance Resistance Reactance DG
(pu) (pu) (pu) penetration
(Z
DG
i
) (R
L,DG
i
) (jX
T,DG
i
level
+jX
L,DG
i
) (%)
IIIA Laterals {0.035 · · · 0.08} 0 j0.01 20
diﬀerent +{j0.025 · · · j0.07}
length,
resistance
neglected
IIIB Laterals {0.035 · · · 0.08}
1
3
j0.01 j0.01 + jX
L,DG
i
20
diﬀerent +X
L,DG
i

length,
resistance
included
IVA Constant 0.06 0 j0.01 + j0.05 {3.33 · · · 33.3}
length,
concentrated
IVB Constant Agrees to 0 j0.01 +
j0.05
par(P
DG
i
)
; {3.33 · · · 33.3}
length, equation (4.2)
spread par(P
DG
i
) =
out the integer
over of
P
DG
i
100.0
laterals
Note that X
L,DG
i
in Scenario IIIB is obtained by substituting the values of Z
DG
i
,
R
L,DG
i
and X
T,DG
i
in this table that correspond to Scenario IIIB into equation (4.2).
In this way, the values of X
L,DG
i
and R
L,DG
i
increase along with the increasing value
of Z
DG
i
Scenario IIIA and Scenario IIIB are deﬁned to investigate the impact of
the DG gridconnection strength on the transmission system stability, when
DG is implemented in laterals of diﬀerent length. In Scenario IIIA the value of
R
L,DG
i
is neglected, while in Scenario IIIB it is taken into account. In these
scenarios, a 20% DG penetration level is assumed (4.1).
Furthermore, ten subscenarios are derived from Scenario IIIA by adjusting
the value of jX
L,DG
i
(jX
T,DG
i
is kept constant at j0.01 pu) in such a way that
the value of Z
TN,DG
i
 in (4.2) raises from 0.035 pu to 0.08 pu, in steps of 0.005
pu. Accordingly, ten subscenarios are derived from Scenario IIIB by adjusting
the values of jX
L,DG
i
and R
L,DG
i
(jX
T,DG
i
is kept constant at j0.01 pu) in
such a way that the value Z
TN,DG
i
 (in (4.2)) increases from 0.035 pu to 0.08
pu, in steps of 0.005 pu while the value of R
L,DG
i
equals
1
3
of (X
L,DG
i
+X
T,DG
i
).
60 4.2 DG GridConnection Strength Impacts
Scenario IVA and Scenario IVB are deﬁned to investigate the impact of the
DG grid connection strength on the system stability, when DG is implemented
either concentrated in one substation (Scenario IVA) or spread out over many
substations (Scenario IVB). In Scenario IVA it is assumed that when DG is
implemented at load busi (DG
i
), it is implemented in one particular point in
the distribution network regardless the ”size of DG”, so that the parameters
of (4.2) are constant: jX
T,DG
i
= j0.01 pu and jX
L,DG
i
= j0.05 pu (R
L,DG
i
is neglected). In Scenario IVB, it is assumed that DG is implemented in a
distribution network with radial laterals; every 100 MW of aggregate DG is
implemented in a diﬀerent lateral. The parameters in (4.2) are jX
T,DG
i
= j0.01
pu, R
L,DG
i
is neglected and jX
L,DG
i
is set according to (4.3) with par(P
DG
i
) =
P
DG
i
100.0
(MW)
(MW)
rounded to the next larger integer. Thus, it is assumed that the
larger the size of the aggregate DG is, the more spread out the locations of the
DG units are, and the lower the impedance between DG and the transmission
network becomes. Ten subscenarios are derived from both Scenario IVA and
Scenario IVB by varying the DG penetration level in the test system from
3.33% to 33.33% in steps of 3.33% (4.1).
4.2.3 Transient Stability Simulation
The transient stability of the test system is investigated by applying a perma
nent threephase fault, cleared after 150 ms, to all possible branches taken into
account in the N −1 adequacy standard (Table 4.1).
To assess the transmission system stability, the maximum rotor speed devi
ation and the oscillation duration are investigated according to Section 3.3.
4.2.4 Simulation Results
Scenario III
With the DG penetration level ﬁxed at 20%, increasing the impedance of the
connection of the DG to the transmission system generally results in a slightly
less stable transmission system, regardless of the DG technology implemented.
Figure 4.7 shows an example of the slightly increasing value for the transient
stability indicators along with the increasing impedance connecting DG accord
ing to Scenario IIIA, when the DG is implemented as synchronous machines
without grid voltage and frequency control (SM). In Figure 4.7, the xaxis rep
resents the number of the branch where a fault is applied. The yaxis represents
the ten subscenarios with the increasing impedance value of the DG connection
as deﬁned in Scenario IIIA, and the zaxis represents the stability indicator.
For a better overview, the zaxis represents ’relative’ values of the indicators
used: the diﬀerence between the resulting maximum rotor speed deviation and
the base case value and the diﬀerence between the resulting oscillation duration
and the base case value. The base case corresponds to subscenario nr. 1.
Impact of DG on Power System Transient Stability 61
1
5
10
15
20
25
30
35
1
5
10
−2
0
2
x 10
−4
Faulty branch [nr.]
SM
DG grid connection
scenario [nr.]
r
e
l
a
t
i
v
e
m
a
x
.
d
e
v
.
[
p
u
]
1
5
10
15
20
25
30
35
1
5
10
−1
0
1
Faulty branch [nr.]
DG grid connection
scenario [nr.]
r
e
l
a
t
i
v
e
o
s
c
.
[
s
]
Figure 4.7: Relative value in system indicators (actual value minus basecase
value (subscenario nr. 1)): maximum rotor speed deviations (upper graph);
oscillation durations (lower graph) when Scenario IIIA is applied and DG is
implemented as synchronous machines without grid voltage and frequency con
trol (SM)
When the resistive part of the impedance between DG and the transmission
system is taken into account (Scenario IIIB), similar results are obtained. Sce
nario IIIB gives slightly higher values of the stability indicators (i.e. a reduced
system stability) than Scenario IIIA. This can be seen from the positive val
ues that result from the substraction of values of Scenario IIIA from those of
Scenario IIIB. Figure 4.8 shows an example of those diﬀerences, when DG is
implemented as synchronous machines (SM).
Scenario IV
DG that is concentrated in one lateral (according to Scenario IVA) results
generally in higher system indicators than when the DG is spread out of several
laterals (according to Scenario IVB). By comparing the ’relative’ values of
the stability indicators (i.e. the diﬀerence between the maximum rotor speed
deviations of Scenarios IVA and IVB and the diﬀerence between the oscillation
durations of Scenarios IVA and IVB) generally positive values are obtained.
Figure 4.9 shows an example of those diﬀerences, in case DG is implemented as
synchronous machines (SM).
62 4.2 DG GridConnection Strength Impacts
4.2.5 Remarks
In this section a general tendency is found that the values of the stability indi
cators slightly increase when the impedance in between the DG and the trans
mission system is raised. A higher impedance leads to a less stable system [63],
[69].
Taking into account the resistance of the distribution system in which the
DG is implemented, results in slightly higher stability indicators than when the
resistance is neglected. This observation can be explained as follows. When we
take the resistance into account, higher active power losses occur that must be
supplied by the swing generator (one of the CGs) and gives somewhat larger
power ﬂows in the transmission lines.
1
5
10
15
20
25
30
35
1
5
10
−5
0
5
x 10
−4
Faulty branch (nr.)
SM
DG grid connection
scenario (nr)
r
e
l
a
t
i
v
e
m
a
x
.
d
e
v
.
[
p
u
]
1
5
10
15
20
25
30
35
1
5
10
−1
0
1
Faulty branch (nr.)
DG grid connection
scenario (nr)
r
e
l
a
t
i
v
e
o
s
c
.
d
u
r
.
[
p
u
]
Figure 4.8: Relative value in system indicators (Scenario IIIB minus Scenario
IIIA which means including resistance): maximum rotor speed deviations (up
per graph); oscillation durations (lower graph) when DG is implemented as
synchronous machines without grid voltage and frequency control (SM)
Impact of DG on Power System Transient Stability 63
1
5
10
15
20
25
30
35
3
20
33
−2
−1
0
1
2
x 10
−4
Faulty branch (nr.)
SM
DG level [%]
r
e
l
a
t
i
v
e
m
a
x
.
d
e
v
.
[
p
u
]
1
5
10
15
20
25
30
35
3
20
33
−2
0
2
Faulty branch (nr.)
DG level [%]
r
e
l
a
t
i
v
e
o
s
c
.
[
s
]
Figure 4.9: Relative value in system indicators (Scenario IVA minus Scenario
IVB which means more spreading of DG): maximum rotor speed deviations
(upper graph); oscillation durations (lower graph) when DG is implemented as
synchronous machines without grid voltage and frequency control (SM)
4.3 DG Penetration Level and Technology Im
pacts
Although the results of Sections 4.1 and 4.2 show that some impact of raising
DG penetration levels on the system stability can be generalized as regardless of
the DG technologies implemented, some small diﬀerences were observed. This
section investigates the impacts of raising DG penetration levels on the system
transient stability with respect to the DG technology.
4.3.1 Simulation Scenario
The test system, the DG technologies and the DG topology are set according
to the simulation setup described in Section 3.5, whereas DG penetration levels
are varied in the similar way as in Section 4.1 with a minor diﬀerence. In
Section 4.1.1 DG penetration level is raised in steps of 3.33% and the load
increment is adjusted accordingly. In this section, the load is increased in steps
of 5%, and the DG penetration level is adjusted accordingly.
64 4.3 DG Penetration Level and Technology Impacts
4.3.2 Transient Stability Simulation
The transient stability of the test system is investigated by applying a permanent
fault to a transmission line. The transmission line between buses 15 and 16
(Figure 3.7), carrying 315 MW and 150 Mvar in the prefault scenario with
no DG implemented, is chosen arbitrarily. The fault is cleared by tripping the
faulty line after a certain fault duration. Three fault durations are simulated:
100, 150 and 200 ms. The eleven penetration level scenarios are extended with
three subscenarios based on the three fault durations. Furthermore, ﬁve DG
technologies as described in Section 3.5.2 are simulated together with the eleven
penetration levels and the three fault durations. To assess the system stability in
these scenarios, the maximum rotor speed deviation and the oscillation duration
are used again.
4.3.3 Simulation Results
At a ﬁrst glance, the simulation results show that the implementation of DG in
the test system (along with the increase in the loads) aﬀects the transient sta
bility diﬀerently depending on DG penetration level, DG technology, and fault
duration. To investigate the results systematically, each of the ﬁve main DG
technologies simulated in this section  the squirrel cage induction generator, the
uncontrolled or controlled synchronous generator, and the DG coupled through
an uncontrolled or controlled powerelectronic interface  are commented on
separately.
Induction Generator
Figure 4.10 shows the worst values of maximum rotor speed deviation (top) and
oscillation duration (bottom) for large centralized generators, when DG with
induction generator technology is implemented. The results are shown for three
faulttime durations: 100, 150, and 200 ms.
When a 100 ms fault duration is applied, the maximum rotor speed deviation
decreases when the penetration level increases from 0% up to 29%. However,
increasing the penetration further, up to 33%, leads to growing maximum rotor
speed deviations. When 150 and 200 ms fault clearing durations are applied, the
maximum rotor speed deviations decrease when the penetration level rises from
0% up to 23% and they increase when the penetration goes further up to 33%.
Overall, the relative changes of the maximum rotor speed deviations (either
increasing or decreasing) are within 11% (at 100 ms fault clearing duration)
and 14% (at 150 and 200 ms fault clearing durations).
Furthermore, when 100 and 150 ms fault clearing durations are applied,
the oscillation durations tend to decrease. When the duration is 200 ms the
oscillation duration tends to increase with increasing DG penetration level.
When the maximum rotor speed deviation and the oscillation duration of
each large generator are displayed separately (Figure 4.11), a better insight is
obtained. For the three fault durations applied, it can be observed that the
maximum rotor speed deviation of each large generator tends to go up or down
Impact of DG on Power System Transient Stability 65
ASM
0.003
0.005
0.007
0.009
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
4.50
5.50
6.50
7.50
8.50
9.50
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
Load increase [%]
(DG penetration level [%])
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
Figure 4.10: Maximum rotor speed deviation (top) and oscillation duration
(bottom) using induction generator (ASM) DG technology, and fault clearing
durations of 100 (), 150 (), and 200 ms (△)
consistently. The curves of the maximum rotor speed deviation in Figure 4.10
(ﬁrst descending, then ascending) are thus made up by two generators: one with
an increasing, and the other with a decreasing maximum rotor speed deviation
due to the growing DG penetration level. A similar tendency is found for the
oscillation duration.
These results may be explained as follows [47], [71]. The eﬀect of induc
tion generators on the power system stability depends on their distance to the
synchronous generators:
• When they are located near the synchronous generators and the latter
speed up during a fault, the stator frequency of the induction generators
increases. This leads to a decrease in the slip frequency and thus in ex
tra generated power, which in turn slows down the speeding up of the
synchronous generators.
• When they are at a larger distance and more weakly coupled to the syn
chronous generator, its speeding up during the fault results in an increasing
reactive power demand. This leads to a lower terminal voltage at the far
away synchronous generator and thus in a decrease of the synchronizing
torque and a faster increase in rotor speed.
66 4.3 DG Penetration Level and Technology Impacts
ASM
2.5
3.5
4.5
5.5
6.5
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
Load increase [%]
(DG penetration level [%])
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
0.0025
0.0035
0.0045
0.0055
0.0065
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
Figure 4.11: Maximum rotor speed deviation (top) and oscillation duration
(bottom) for each large centralized generator when induction generator (ASM)
DG technology is implemented and a 150 ms fault clearing duration is applied.
Generator nr. 1 (G1) = dash; G2 = dash△; G3 = solid×; G4 = solid; G5
= solid♦; G6 = solid; G7 = solid△; G8 = dash×; G9 = dash; G10 =
dash♦
As an example, the synchronous generator 1 at bus 39 (Figure 3.7) is con
sidered. From Figure 4.11, it is clear that generator 1 shows an increasing
maximum rotor speed deviation with growing DG penetration levels. Since at
buses 1, 2 and 9 no DG is implemented  i.e. generator 1 is not located close
to DG  this result is what could be expected: a faster increase in rotor speed
with growing DG penetration levels.
Synchronous Generator
The worst values of the transient stability indicators for the large centralized
generators, in case that DG is implemented using uncontrolled or controlled
synchronous generator technologies are displayed in Figure 4.12. The maxi
mum rotor speed deviation consistently decreases when DG penetration level is
increased. However, the oscillation duration is diﬀerent when these two tech
nologies are applied. While the oscillation duration tends to increase in case
of uncontrolled synchronous DG, it tends to decrease in case of controlled syn
chronous DG.
Impact of DG on Power System Transient Stability 67
SM
0.003
0.005
0.007
0.009
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
4.50
5.50
6.50
7.50
8.50
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
Load increase [%]
(DG penetration level [%])
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
Figure 4.12: Maximum rotor speed deviation (top) and oscillation duration
(bottom) when DG using uncontrolled, SM, (solid lines) and controlled, SMC,
(dashed lines) synchronous generator technologies are implemented and fault
clearing durations of 100 (), 150 (), and 200 ms (△) are applied
The results can be explained as follows. Both uncontrolled and controlled
distributed synchronous generators are equipped with an excitation winding on
the rotor, keeping the generators excited during the fault. When a fault occurs,
the distributed generators supply the fault current and the voltage drop during
the fault is not as severe as is the case without DG. Thus during the fault, a
higher DG penetration level results in a higher terminal voltage and less over
speed.
PowerElectronic Converter
When DG is coupled through uncontrolled or controlled powerelectronic con
verters, the maximum rotor speed deviation of the synchronous generators de
creases consistently with increasing DG penetration levels (Figure 4.13). The
oscillation duration shows diﬀerent results. Along with increasing DG pene
tration levels, the oscillation duration of the synchronous generators tends to
increase when DG is coupled through uncontrolled powerelectronic convert
ers. However, it tends to decrease slightly if DG is coupled through controlled
powerelectronic converters.
68 4.3 DG Penetration Level and Technology Impacts
PE / PEC
0.0025
0.0045
0.0065
0.0085
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
4.00
5.00
6.00
7.00
8.00
9.00
10.00
0
(0)
5
(5)
10
(9)
15
(13)
20
(17)
25
(20)
30
(23)
35
(26)
40
(29)
45
(31)
50
(33)
Load increase [%]
(DG penetration level [%])
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
Figure 4.13: Maximum rotor speed deviation (top) and oscillation duration (bot
tom) when DG is coupled through a powerelectronic interface without (solid
lines) or with voltage and frequency control (dashed lines), and fault clearing
durations of 100 (), 150 (), and 200 ms (△) are applied
The simulation results can be explained as follows. The protection sys
tems applied, disconnect the distributed generators connected through a power
electronic interface as soon as the voltage level drops below 0.85 pu, e.g. during
a fault. The protection system reconnects DG as soon as a recovery action is
taken and when the voltage level increases again and passes 0.85 pu. Thus,
during a fault, the distributed generators are lost and the rotor acceleration of
the large synchronous generators is reduced. As the same protection schemes
are applied for both controlled and uncontrolled powerelectronic converters,
the maximum rotor speed deviations are the same.
The diﬀerent results found for the oscillation durations are caused by the
frequency controller actions of the controlled powerelectronic converter. When
the distributed generators are gridconnected by means of a controlled power
electronic interface, the frequency controller damps the rotor oscillations of the
centralized generators after the protection system reconnects the DG.
Impact of DG on Power System Transient Stability 69
4.3.4 Remarks
The computations in this section show that the impact of DG on the system
stability depends on both DG technology and penetration level. The maxi
mum rotor speed deviation for most of the synchronous centralized generators
decreases with increasing DG penetration levels for the main DG technologies,
either controlled or uncontrolled. However, when induction generator based DG
is implemented, this is only true for generators located in the vicinity of the DG.
The results for the oscillation duration are ambiguous, although it appears that
the oscillation duration tends to decrease with increasing DG penetration levels
when a DG technology equipped with controllers (Section 3.4.1) is applied. This
suggests that the performance of the system can be improved by enhancing the
control capabilities of the DG technology.
4.4 Protection of PowerElectronics Impacts
In this section, the impact of implementing powerelectronic interfaced DG on
the system stability is further investigated, with the focus on the two protection
systems: DG that disconnects from the network automatically when the voltage
drops below a certain level, and reconnects as soon as the voltage recovers and
DG that remains connected to the power network during a fault (ridethrough
capability [42]).
4.4.1 Simulation Scenarios
The test system and the DG topology are again set according to the simulation
setup described in Section 3.5.
DG connected to the grid via a powerelectronic converter without grid volt
age and frequency control (PE) (Sections 3.4.1) is simulated.
The DG penetration levels in the test system are varied in the same way as
in Section 4.1 (Scenario I, eleven (sub)scenarios). The DG connected to each
load bus produces an amount of active power equal to the increased real power
consumption of the load at that particular bus.
Two protection schemes for DG are applied:
• DG is disconnected from the power network when the voltage level of the
system drops below 0.85 pu and reconnected as soon as the voltage level
passes 0.85 pu (Protection I).
• DG remains connected to the network during a fault. However, due to
limitations of the components of a powerelectronic interface, it is assumed
that the current through the powerelectronic interface is limited to a
maximum value of 1.2 pu. Thus, in case the voltage level drops during
the fault, the power supplied by the DG drops accordingly, whereas the
current is constant and limited to its maximum value (Protection II).
70 4.4 Protection of PowerElectronics Impacts
Figure 4.14 shows an example of the diﬀerent active power output values of
converterconnected DG (PE) in case of Protection I (top) and Protection II
(middle). The fault occurs between the buses 15 and 16, and both protection
schemes, Protection I and Protection II, are applied. The bottom graph shows
the voltage of the corresponding bus obtained from the case without voltage and
frequency control and Protection I. The case of Protection II results in a slightly
diﬀerent value of the corresponding bus voltage. Yet it is quite comparable to the
case of Protection I. The xaxis represents the time and the yaxis represents
the DG unit. The zaxis, represents the value of the active power output of
the DG unit (top and middle) and the value of the corresponding bus voltage
(bottom).
PE
1
1.15
1.5
1
19
0
0.5
1
1.5
time [s]
DG [nr.]
A
c
t
i
v
e
P
o
w
e
r
[
p
u
.
]
1
1.15
1.5
1
19
0
0.5
1
1.5
time [s]
DG [nr.]
A
c
t
i
v
e
P
o
w
e
r
[
p
u
.
]
1
1.15
1.5
1
19
0
0.5
1
1.5
time [s]
DG [nr.]
B
u
s
V
o
l
t
a
g
e
[
p
u
.
]
Figure 4.14: DG active power outputs when implemented via a powerelectronic
interface without grid voltage and frequency control (PE) in the case of Protec
tion I (top) or Protection II (middle) at a DG penetration level of 33.33% and
the corresponding bus voltages of Protection I case (bottom)
Impact of DG on Power System Transient Stability 71
4.4.2 Transient Stability Simulation
The transient stability of the test system is investigated by applying a permanent
threephase fault, cleared after 150 ms, to all possible branches. The maximum
rotor speed deviation and the oscillation duration are again applied.
4.4.3 Simulation Results
To investigate the impact of both protection schemes, the simulation results of
both schemes are compared. The values of the indicators in Protection scheme
II are generally higher than that of Protection scheme I, especially when DG
is implemented without grid voltage and frequency control (PE) (Figure 4.15).
This can be explained as follows.
1
5
10
15
20
25
30
35
0
20
33
−5
0
5
x 10
−3
Faulty branch [nr.]
PE
DG level [%]
r
e
l
.
m
a
x
.
d
e
v
.
[
p
u
]
1
5
10
15
20
25
30
35
0
20
33
−5
0
5
Faulty branch [nr.]
DG level [%]
r
e
l
.
o
s
c
.
d
u
r
.
[
s
]
Figure 4.15: Relative values in maximum rotor speed deviations (top) and rel
ative oscillation durations (bottom), i.e. the results of Protection scheme II
minus Protection scheme I. DG is connected via a powerelectronic interface
without grid voltage and frequency control (PE).
In case DG remains connected (Protection II) to the system, DG covers part
of the power consumption. As a result, the CG electrical power output, P
e
is
lower (accelerating power, P
a
, is larger then) than when the DG disconnects
(Protection I) from the system. In other words, when DG disconnects from the
system (Protection I), the part of the power consumption that in Protection
72 4.5 Conclusions
scheme II would be covered by the keptconnected DG is covered by CG. This
results in a higher electrical power output, P
e
, and less accelerating power, P
a
.
4.4.4 Remarks
This section shows that when the powerelectronic interfaced DG remains con
nected to the network during the fault, higher generator speeds result (i.e. re
duced transient stability) than in case DG disconnects during the fault. How
ever, from the power supply point of view, it may be important to keep (parts
of) the DG connected to the system during a fault in order to supply power
to the less involved areas. In this case, a rule could be applied that prescribes
when the DG disconnects from the system or when it remains in the system
during a fault. As an example, a rule based on the voltage level can be used for
this purpose.
4.5 Conclusions
It is wellknown that the implementation of DG inﬂuences the technical aspects
of the distribution grids. However, the impact of a small amount of DG con
nected to the grid on the power system transient stability has not been treated
so often. When the penetration level of DG increases, its impact is no longer
restricted to the distribution network, but begins to inﬂuence the whole sys
tem [76].
In this chapter therefore the DG penetration level on the power system tran
sient stability is analyzed. Several factors are analyzed such as load scenario, DG
grid connection strength, DG technology, and the protection scheme of power
electronic interfaced DG. When load is increased and DG is implemented in the
distribution grid, the active power generated by the DG is consumed directly
by the load at the same feeder and the active power ﬂows on the transmission
lines are more or less constant instead of increasing active power ﬂows when
CG needs to cover the increased load. In contrast, the active power ﬂows in the
transmission lines decrease when DG covers also part of the existing load. Less
power ﬂows should intuitively result in a more stable system, and in this work,
as the system stability indicators show less maximum rotor speed deviation and
shorter oscillation duration. Although a mathematical proof is not established,
[31] supports this correlation, and results in this chapter support this as well.
It is also found that the values of the stability indicators show a tendency to in
crease slightly when the impedance between DG and the transmission system is
raised. Furthermore, taking into account the resistance of the distribution sys
tem in which DG is implemented results in slightly higher stability indicators
than when the resistance is neglected.
It is found that the maximum rotor speed deviation for most of the syn
chronous centralized generators decreases with increasing DG penetration levels
for all DG connection technologies, either controlled or uncontrolled. However,
when induction generator based DG is implemented, this is only true for gener
Impact of DG on Power System Transient Stability 73
ators located in the vicinity of the DG. The results for the oscillation duration
are ambiguous, although it appears that the oscillation duration tends to de
crease with increasing DG penetration levels when a DG connection technology
equipped with controllers is applied. This suggests that the performance of
the system can be improved by enhancing the control capabilities of the DG
technology.
It is also found that when the powerelectronic interfaced DG remains con
nected to the network during and after a fault, higher generator speeds appear
than in case DG disconnects during the fault (i.e. reduced transient stability).
However, from the power supply point of view, it is important to keep (parts
of) the DG connected (ridethroughcapability) to the system during a fault in
order not to loose too much power in the system. In that case, a rule could be
applied that prescribes when the DG disconnects from the system or when it
remains in the system during a fault, for example based on the voltage level to
which it is connected [20].
In general, there appear no signiﬁcant stability problems up to the 30% DG
penetration level examined. Higher DG penetration levels have not been tested
in this chapter. Note that in this chapter, all centralized generators remain in
the system  as well as their active and reactive power controls  along with the
increasing DG penetration levels.
74 4.5 Conclusions
Chapter 5
’Vertical to Horizontal’
Transformation of Power
Systems
A largescale implementation of DG may be expected in future. Eventually,
the high amount of DG in a power system may cause a number of centralized
generators (power plants) to be shut down for eﬃciency (or environmental)
reasons. This results in a gradual transition from the current ’vertical’ into a
future ’horizontal’ power system (Section 1.3). Figure 5.1 illustrates the concept
of a ’vertical’ and a ’horizontal’ power system (redeﬁned from Figure 1.3).
Loads and Distributed Generators
Transmission
network
Centralized generators
~
~ ~
~ ~
Loads and Distributed Generators
Transmission
network
Centralized generator
~
~ ~ ~ ~
(a) (b)
Figure 5.1: Illustration of the ’vertical’ (a) and the ’horizontal’ power system
(b)
In Chapter 4, all centralized generators and their corresponding inertias
and control functions remain in the system when the DG level is increased.
However, a ’verticaltohorizontal’ transformation of power systems results in
76 5.1 Simulation Setup
less centralized generators in service and, correspondingly, reductions of their
control functions alongside with the increasing DG level. Moreover, the use
of largescale power electronicinterfaced DG units implies a reduction in the
rotating masses (inertia) in the system. In this chapter, the impact of such
’verticaltohorizontal’ transformation of power systems on transient stability is
investigated.
Simpliﬁcations are assumed in shutting down the centralized generators.
The main idea is to give a consistent sequence of shutting them down. A simple
economic dispatch program is used for the remaining units in service.
The results of this chapter show that based on power system transient sta
bility constraints, a limit of DG penetration level in a power system can be
reached.
Some remedies to improve the stability of the system that goes through the
‘verticaltohorizontal’ transition are also discussed (Sections 5.3 through 5.5).
5.1 Simulation Setup
Test system, load modeling and DG topology for simulations in this chapter are
set according to the simulation setup deﬁned in Section 3.5.
DG is implemented as powerelectronic interfaced DG without grid voltage
and frequency control (modeled as a constant PQsource; Section 3.5.2).
The transformation from a vertical to a horizontal power system is done in
the following way:
• Both the active and reactive power of all loads are kept constant.
• The penetration level of DG is raised by increasing the fraction of the
total load in the test system served by DG.
• DG is connected to every load bus via a j0.05 p.u. impedance at the
100 MVA system base. The fraction of the total load served by DG is
distributed among the modeled DG units proportional to the active power
consumed by the load at a particular bus.
• The remaining power is divided among the (dispatchable) centralized gen
erator (CG) units by considering the economic operation of the power
system, where some CG units, if necessary, are taken out of service.
The DG penetration level in the system is deﬁned as
%DG
level
=
P
DG
P
Load,Total
×100, (5.1)
a modiﬁed version of the deﬁnition of DG penetration level used Chapter 4.
P
Load,Total
is the total amount of active load within the test system and P
DG
is the total amount of active power generated by DG.
Minimum and maximum loading limits of the CG units are assumed. All
centralized generators are initially dispatchable and have a piecewise linear cost
curve [53]
f
i
= a
i
P
CGi
+b
i
, (5.2)
’Vertical to Horizontal’ Transformation of Power Systems 77
with f
i
the cost of CG unit i to generate an amount of active power P
CGi
.
In (5.2), a
i
(incremental cost) and b
i
(startup cost) are constants corresponding
to the CG unit i. To diﬀerentiate the economic eﬃciency of each CG, and thus
give a consistent sequence of taking out of service centralized generators, the
constants a
i
and b
i
are deﬁned as
a
i
= αc
i
, (5.3)
b
i
= βc
i
, (5.4)
with α and β constant, and c
i
deﬁned as
c
i
= {c
1
+γ(i −1)i = 2, . . . , 10, γ ∈ R
+
}. (5.5)
In this way, P
CGj
is higher than the P
CG(j−1)
(e.g, in the test system CG nr.
10 is more expensive than CG nr. 9, CG nr. 9 is more expensive than CG
nr. 8, and so forth). An optimal power ﬂow (OPF) program, with objective
function to minimize the fuel cost is then run each time the DG penetration
level is raised. As the main constraint, the voltage limits with 5% margin are
set in this OPF program. The most ineﬃcient CG unit whose power output
falls below its minimum loading limit, is taken out of service, and a ﬁxed shunt
device is implemented at the location of the shut down CG to compensate the
former’s reactive power generation. The optimal power ﬂow program is then re
run. In this way, the ’verticaltohorizontal’ transformation of the power system,
is simulated. The ﬂowchart of Figure 5.2 illustrates this process. In this study,
the value used for increasing the DG level in each simulation is 5%, and a DG
level of 90% is set as maximum. A 90% maximum DG
level
is chosen so that
in all scenarios at least one centralized generator remains in operation when
the optimal power ﬂow is performed. This generator is necessary to provide
a reference frequency for all power electronic converters of the DG units. The
stability of these last centralized generators that remain connected determines
the stability of the system as a whole.
The transient stability of the test system is investigated by applying a per
manent fault to the (arbitrarily chosen) transmission line between buses 15 and
16 . A single line diagram of the test system is shown in Figure 3.7, and detailed
in Appendix B. It is assumed that the fault is cleared by tripping the faulty
line after 100 ms.
The protection system applied disconnects the powerelectronic interfaced
DG units as soon as the voltage level drops below 0.85 pu, e.g. during a fault,
and it reconnects the DG as soon as the voltage level increases again above
(or equal to) the 0.85 pu level (Section 3.4.1) after the fault is cleared. In
Section 5.5, DG with “ridethrough” capability is applied. In that case, the
rated current of the DG converter is limited to 1.2 pu of the power electronic
connected DG.
To assess the transmission system stability, two transient stability indicators,
maximum rotor speed deviation and oscillation duration (Section 3.3), are used
to observe the rotor angle stability and to quantify the severity of the rotor
78 5.2 Simulation Results Case I
Start
Take basecase with 0 % DG
penetration level (DG
level
)
Run Optimal Power Flow Program
Any CG to shut
down?
Shut down CG
Implement shunt device at the location
of shut down CG
Increase DG
level
by 5%
Define case with DG penetration
level = DG
level
Stop
No
Yes
Run Optimal Power Flow Program
DG
level
= 90% ? No
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
Yes
Figure 5.2: Flowchart depicting ’verticaltohorizontal’ transformation process
speed oscillations of the remaining centralized generators. The transient sta
bility performance of the cases within the verticaltohorizontal transformation
scenario is compared based on these indicators.
5.2 Simulation Results Case I
The approach depicted in Figure 5.2 results in a number of simulations for a
power system that goes trough the ’verticaltohorizontal’ transformation. As a
result of performing the optimal power ﬂow, CG units are taken out of service
as the DG penetration level increases. The result of this process is illustrated in
Figure 5.3. In this research, each shutdown CG unit is replaced by a ﬁxed (at
’Vertical to Horizontal’ Transformation of Power Systems 79
one value) reactive shunt device (cheaper than the more sophisticated variable
switched shunt devices or FACTS devices). With this approach, the power ﬂow
program keeps 4 centralized generators in operation. Note that due to the small
values, in Figure 5.3 the output power generated by CG nr. 2 and CG nr. 4 are
not clearly visible. This is called Case I.
0
1000
2000
3000
4000
5000
6000
7000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
CG10
CG9
CG8
CG7
CG6
CG5
CG4
CG3
CG2
CG1
Figure 5.3: Dispatched CGs as a function of the DG penetration level (Case I)
Figure 5.4 shows both stability indicators as a function of the DG level, when
a permanent fault to the transmission line between buses 15 and 16 is applied.
More detailed values are given in Table 5.1.
It can be observed from Figure 5.4 and Table 5.1, that up to 45% DG level the
indicators do not change much. The indicators signiﬁcantly increase when the
DG level rises beyond this value. At 60%75% DG penetration levels the applied
fault causes instability. When DG levels are increased even further (80%90%),
the applied fault does no longer cause instability, although the indicator values
are high compared to the values resulting from 5%45% DG levels.
This result can be explained by considering the fact that two competing
mechanisms play a role in inﬂuencing the transient stability behavior of the
system in this ’verticaltohorizontal’ transformation:
• The total inertia (i.e. stored kinetic energy) within the rotating masses of
the machines in the system decreases, so stability decreases.
• The branch ﬂows, in particular in the faulty branch, decrease so stability
increases.
80 5.2 Simulation Results Case I
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
0.02
0.04
0.06
0.08
DG Level [%]
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
10
20
30
DG Level [%]
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
i
n
s
t
a
b
i
l
i
t
y
i
n
s
t
a
b
i
l
i
t
y
Figure 5.4: Transient stability indicators: maximum rotor speed deviation (top)
and oscillation duration (bottom) as function of DG penetration level. Note that
at DG levels of 60% through 75% the simulation shows that the applied fault
causes instability in Case I
Table 5.1: Stability indicators as a Function of the DG Level (Case I)
DG Level Max. Rotor Speed Oscillation
Scenario(%) Deviation (pu) Duration (s)
0 0.0044 5.05
5 0.0045 5.13
10 0.0043 5.12
15 0.0042 5.96
20 0.0045 6.23
25 0.0043 6.31
30 0.0039 5.16
35 0.0033 5.25
40 0.0034 5.32
45 0.0035 5.54
50 0.0080 11.12
55 0.0532 14.24
60 unstable unstable
65 unstable unstable
70 unstable unstable
75 unstable unstable
80 0.0674 24.57
85 0.0638 24.22
90 0.0811 26.11
’Vertical to Horizontal’ Transformation of Power Systems 81
In this study, we assume that all centralized generators are of the same type
and technology. The inertia constants of all centralized generators in per unit
thus are equal. Using the inertia constant H (3.4) and using the dispatched CG
values found from the optimal power ﬂow simulation as shown in Figure 5.2,
we obtain the total stored kinetic energy of the system at synchronous speed as
shown in Figure 5.5. With H the same for all generators, the plot in Figure 5.5 is
equivalent to the plot of the sum of machine ratings for the generators connected.
The circles (solid line) and their data labels mark the stored kinetic energy
(in per unit) at the speed of the moment immediately following fault clearing
(
¸
n
CG
i=1
H
i
(ω
pf
i
)
2
; pf stands for postfault).
The total stored kinetic energy of the system decreases consistently as the
DG level increases up to 55% and remains constant thereafter because the min
imum of four CGs are kept in the system.
0
5000
10000
15000
20000
25000
30000
35000
40000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
DG Level [%]
S
t
o
r
e
d
K
i
n
e
t
i
c
E
n
e
r
g
y
[
M
J
]
CG10
CG9
CG8
CG7
CG6
CG5
CG4
CG3
CG2
CG1
40.2
36.2 36.2 36.1
32.2 32.1
28.1
28.1
24.1 24.1
20.1
16.1
16.0
16.0
16.0
15.9 15.9
15.9 15.9
Figure 5.5: Total stored kinetic energy in the system at synchronous speed in
Case I. The circles (solid line) and their data labels mark the stored kinetic
energy at the speed of the moment immediately following fault clearing
The sum of active power ﬂows in all grid lines, as well as the power ﬂow in
the branch between buses 1516, are shown in Figure 5.6.
The load ﬂow in the test system, particularly the active power ﬂow in the
branch between buses 15 and 16 where the fault is applied, behaves diﬀerently.
This ﬂow initially decreases when the DG level is raised from 0% to 45%. Later,
it increases when the DG level is continuously raised up to 70% and decreases
again when the DG level increases further to 90% (Figure 5.6). This is due
to the combination of implementing DG and taking out of service of CG units
during the ’verticaltohorizontal’ transformation of the power system. It can be
concluded that in the ﬁrst place, the decreasing total inertia of the system (total
82 5.2 Simulation Results Case I
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
100
200
300
400
500
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
1
5
−
1
6
[
M
W
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
5000
10000
DG Level [%]
S
u
m
o
f
A
l
l
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
s
[
M
W
]
Figure 5.6: Active power ﬂow in the branch between buses 1516, where the
fault is applied in the simulation (top) and the sum of all active power ﬂows in
the grid lines of the test system (bottom) in Case I
stored kinetic energy in the system) that occurs when the DG level increases
from 0% to 45% is ’compensated’ by the decreasing active power ﬂow, so that
the stability performance of the system with a DG level up to 45% does not
vary much. However, when the DG level increases further, high(er) transient
stability indicators for the cases of 50% and 55% DG level are observed, after
which instability occurs for 60%75% DG levels. With 80%90% DG levels, the
system is again stabilized, although the transient indicators for these cases are
quite high (Figure 5.4). The reason is that the total inertia of the system is
not changed when the DG levels are increased from 55% to 90% (Figure 5.5),
while the active power ﬂow in the branch where the fault is applied, experiences
its maximum at the 70% DG level (Figure 5.6). Note that in Figure 5.6 the
sum of all ﬂows in the system does not experience a monotonic decrease as the
DG penetration level goes up, because the objective function of the OPF is to
minimize the fuel cost rather than to minimize the branch ﬂows. Therefore, it
seems that in the case of 55% and 80%90% DG levels, the small active power
ﬂowing in the branch where the fault is applied ‘compensates’ for the low total
inertia of the system, so that the applied fault does not cause instability (unlike
in the case of 60%75% DG levels with the high active power ﬂowing in the
faulty branch and the low total system inertia).
’Vertical to Horizontal’ Transformation of Power Systems 83
5.3 Rescheduling Generation Case I
Figures 5.4 through 5.6 show that a combination of the smaller total system in
ertia and the increased power ﬂow for 60%75% DG levels (compared to 0%55%
and 80%90% DG levels) results in system instability. Therefore, a logical rem
edy to eliminate the instability is optimizing the load ﬂow in case of reduction
of the total system inertia.
0
1000
2000
3000
4000
5000
6000
7000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
CG10
CG9
CG8
CG7
CG6
CG5
CG4
CG3
CG2
CG1
Figure 5.7: Dispatched CGs as a function of the DG penetration level, Case I
after CG outputs are rescheduled
In this section, the outputs of the remaining centralized generators (CGs) in
the system are rescheduled to alleviate instability problems. For this purpose,
after the approach depicted in Figure 5.2 is performed in which some CG units
are taken out of service, an OPF program is run, with the objective function set
to minimize the losses, so that the sum of all branch ﬂows and the ﬂow in branch
1516 are more or less minimized. Figure 5.7 shows the dispatched CG units as
a function of the DG levels after the CG outputs are rescheduled. Figure 5.8
shows the corresponding sum of all branch ﬂows and the ﬂow in branch 1516.
As may be expected, after the load ﬂow is optimized, the instabilities for cases
with 60%70% DG levels are eliminated (Figure 5.9). It is not obvious that in
all cases rescheduling of generation is eﬀective enough, but here it is.
84 5.4 Simulation Results Case II
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
100
200
300
400
500
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
1
5
−
1
6
[
M
W
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
5000
10000
DG Level [%]
S
u
m
o
f
A
l
l
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
s
[
M
W
]
Figure 5.8: Active power ﬂow in the branch between buses 1516 where the
fault is applied in the simulation (top) and the sum of all active power ﬂows
in the grid lines of the test system (bottom). The triangles (dashed line) mark
the results of rescheduling CG outputs and the circles (dotted line) mark the
original Case I
5.4 Simulation Results Case II
In Section 5.2, if a CG unit is taken out of service, a ﬁxed switched shunt
device is used to replace it and to provide reactive power (Figure 5.2, step No.
7). When the gradual transition is continued, once a CG unit is taken out of
service, the topology and reactive power balance of the test system are changed.
Since this ﬁxed switched shunt device has no ﬂexibility of supporting the voltage
in the system, unlike the replaced CG units, the power ﬂow program forces the 4
generators to keep running, in order to fulﬁll the reactive power demand changes
within the system and to hold the voltage in between speciﬁed margins.
To avoid this constraint, in this section, instead of using a ﬁxed switched
shunt device, a variable switched shunt device is used in the process No. 7 of
the ﬂowchart depicted in Figure 5.2. In this way, the reactive power supplied
by the shunt device is adjusted at every step of the gradual transition in such a
way that the voltage at the bus is kept between 0.95–1.05 pu. As a result, when
the DG level increases up to 90%, the load ﬂow program can manage up to only
2 generators running (8 CG units are taken out of service). This is called Case
II. Figure 5.10 shows the resultant dispatched CG. The corresponding inertias
of the CG units are shown in Figure 5.11.
Applying the 100 ms permanent fault clearance on the line between bus 15
16 results in the indicators shown in Figure 5.12. In Case II, rescheduling the
output power of CG units only improves the indicator at a DG level of 60%.
’Vertical to Horizontal’ Transformation of Power Systems 85
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
0.02
0.04
0.06
0.08
DG Level [%]
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
10
20
30
DG Level [%]
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
Figure 5.9: Transient stability indicators: maximum rotor speed deviation (top)
and oscillation duration (bottom) as a function of the DG penetration level. The
triangles (dashed line) mark the results of rescheduling CG outputs and the
circles (dotted line) the original case. Note that by rescheduling CG outputs,
the applied fault does not cause instability at DG levels of 60% through 75%
0
1000
2000
3000
4000
5000
6000
7000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
CG10
CG9
CG8
CG7
CG6
CG5
CG4
CG3
CG2
CG1
Figure 5.10: Dispatched CGs as a function of the DG penetration level when
variable, instead of ﬁxed, shunt devices are used (Case II instead of Case I)
86 5.4 Simulation Results Case II
0
5000
10000
15000
20000
25000
30000
35000
40000
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
DG Level [%]
S
t
o
r
e
d
K
i
n
e
t
i
c
E
n
e
r
g
y
[
M
J
]
CG10
CG9
CG8
CG7
CG6
CG5
CG4
CG3
CG2
CG1
40.2
36.2 36.2 36.1
32.2 32.1
28.1
28.1
24.1 24.1
20.1
16.1
16.0
12.0
12.0
12.0
8.0 8.0 8.0
Figure 5.11: Total stored kinetic energy in the system at synchronous speed, as
a function of the DG penetration level in Case II. The circles (solid line) and
their data labels mark the stored kinetic energy (in per unit) at the speed of
the moment immediately following fault clearing
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
0.02
0.04
0.06
DG Level [%]
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
5
10
15
20
DG Level [%]
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
instability limit
(rescheduling generator)
instability limit
(original case)
instability limit
(rescheduling generator)
instability limit
(original case)
Figure 5.12: Transient stability indicators: maximum rotor speed deviation
(top) and oscillation duration (bottom) as a function of the DG penetration
level. The circles (dotted line) mark Case II and the triangles (dashed line) the
results of rescheduling CG output in Case II. Note that by rescheduling CG
outputs, the applied fault does not cause instability at DG levels of 60%
’Vertical to Horizontal’ Transformation of Power Systems 87
As shown in Figure 5.13, the load ﬂow in the case of 60% DG level with
rescheduling CG units is much improved compared to the original one of 60%
DG level. At higher DG level rescheduling CG does not help, although the load
ﬂow still decreases. This result is logical when the inertia decreases. This leads
to the conclusion that there is a limit to the DG penetration level in power
systems for the test system used.
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
100
200
300
400
500
DG Level [%]
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
1
5
−
1
6
[
M
W
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
5000
10000
DG Level [%]
S
u
m
o
f
A
l
l
A
c
t
i
v
e
P
o
w
e
r
F
l
o
w
s
[
M
W
]
Figure 5.13: Active power ﬂow in the faulted branch (top) and sum of all active
power ﬂows in the grid lines of the test system (bottom) in Case II. The triangles
(dashed line) mark the results of rescheduling CG outputs and the circles (dotted
line) the original Case II
This is interesting as so far the DG penetration level is mostly straightfor
wardly used without a limit. This, however, does not explicitly take into account
how many CG units still are in operation to provide the inertia or reactive power
service to the system. Therefore, this result suggests another way of deﬁning the
limit of DG penetration level. This new approach must be based on the stored
kinetic energy provided by the system. Like the example in this section, the DG
level is limited when the system has a total inertia of 30% or less compared to
the original one. But, the availability of reactive power support can inﬂuence
this limit.
The limit obtained in this section is sensitive to the network/test system and
verticaltohorizontal scenario. However, the merit of this section is in showing
that, in order to determine the DG limit for a given system, analysis has to be
done on the network. Similar remedial actions might be needed as discussed
here. This requires the speciﬁc sequence of shutting down the CG units, and
identifying if certain unstable situations are correctable by rescheduling the
remaining generation. Of course all possible faults and branches in the system
have to be taken into account in such an analysis.
88 5.5 DG with RideThrough Capability
5.5 DG with RideThrough Capability
In this section, DG with ridethrough capability is applied to the simulation
setup of Section 5.4. The rated current of the DG converter is limited to 1.2 pu
of the power electronic connected DG.
As loads are equally divided in constant impedance, constant power and
constant current, it is shown that the use of ridethrough capable DG results
in an improved voltage support for the system, and hence in system stable
operation even up to 80% penetration levels. Figure 5.14 shows the results
of system indicators when ridethrough DG is used. Figure 5.15 shows the
voltage level at a certain bus in the system for 50% and 55% DG respectively.
This result again proves that the limit of DG penetration level is inﬂuenced by a
number of factors such as system inertia, reactive power support, and protection
mechanism of the powerelectronic interfaced DG.
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
0
0.02
0.04
0.06
DG Level [%]
M
a
x
.
r
o
t
o
r
s
p
e
e
d
d
e
v
i
a
t
i
o
n
[
p
u
]
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
5
10
15
20
DG Level [%]
O
s
c
i
l
l
a
t
i
o
n
d
u
r
a
t
i
o
n
[
s
]
stability limit
(DG with
ride−through
capability)
stability limit
(original case II)
stability limit
(original case II)
stability limit
(DG with
ride−through
capability)
Figure 5.14: Transient stability indicators: maximum rotor speed deviation
(top) and oscillation duration (bottom) as a function of the DG penetration
level. The circles (dotted line) mark the original Case II and the triangles
(dashed line) the results of applying DG with ridethrough capability in this
case. Note that by applying DG with ridethrough capability, the applied fault
does not cause instability at DG levels of 60% through 80%
The results shown in Figure 5.14 are consistent with those of Chapter 4.
When the penetration level of DG is low, disconnecting DG from the system
during fault acts like introducing a resistive load to the system, damping the
synchronous machine oscillations (higher indicators values at DG level up to
50% when DG is kept connected). However, in the higher level, when DG is
supporting the voltage, it improves system stability.
’Vertical to Horizontal’ Transformation of Power Systems 89
0 1 2 3 4 5
0
0.5
1
1.5
Time [s]
V
o
l
t
a
g
e
[
p
u
]
0 1 2 3 4 5
0
0.5
1
1.5
Time [s]
V
o
l
t
a
g
e
[
p
u
]
Figure 5.15: Node voltage of bus 16 at 50 % DG level (upper graph) and 55% DG
level (lower graph). The solid lines correspond to the original Case II (i.e. DG is
disconnected during the fault), the dashed lines to the case where ridethrough
capability DG is used
5.6 Remarks
In this section, it is assumed that all power electronic interfaced DG units
require a reference frequency for their proper operation, i.e. the power electronic
converters need a frequency reference taken from the grid. Therefore, within
this chapter, the ’verticaltohorizontal’ transformation is limited until at least
two centralized generators are still running (Section 5.4). The remaining CG
units albeit small in percentage (e.g., only 2 units) are crucial for the operation
of the DG units and thus the whole system to give the frequency reference to
the powerelectronic interfaced DG. When these machines are unstable, they
cannot be simply disconnected from the system. Moreover, when these CG
units provide additional reactive support, it is even more important not to lose
these units.
5.7 Conclusion
Higher penetration levels of DG leads to taking out of service of centralized
generators. In this way, an increase in DG penetration level is followed by a
decrease in inertia and reactive power support ability in the system. Power
system inertia and the respectively stored kinetic energy of the total genera
tion system is an important component inﬂuencing power system stability [43].
Therefore, intuitively when the increasing penetration level of converter con
nected DG that adds no inertia to the system is in parallel with the decreasing
90 5.7 Conclusion
inertia due to taking out of service of centralized generators, the system will
be more vulnerable to disturbances [48]. This chapter examines the impact of
such a change in supply structure on the system transient stability. Studies
on the impact on the power system with DG and limited inertia have been
presented in literatures [38], [44]. Yet, the contribution of this work is the
more fundamental approach of transforming a power system from the classical
’verticallyoperated power system’ supported mainly by several large centralized
generators into a ’horizontallyoperated power system’, having a large number
of small to mediumsized distributed generators, that is where the simultaneous
eﬀects of increasing DG and decreasing system inertia occurs.
It is then found that a limit to DG penetration level in a given power system
can be reached. This limit is based on the total kinetic energy remaining in the
system and the reactive power support scheme adopted. This ﬁnding is inter
esting since DG penetration level is classically deﬁned as a share of active power
output generated by DG units compared to the total active power generated by
all generators in the system, or as a share of the active power output generated
by DG units compared to the total load. The transient stability analysis re
sults that the ability of centralized generators to supply supporting services like
system inertia and reactive power, is of most importance. Therefore a limit in
DG penetration level cannot solely be set based on the proportion of the active
power production of DG.
Reactive power support depends signiﬁcantly on the network topology and
the optimal power ﬂow strategy. Therefore, the limit obtained is sensitive to
the network analyzed and the verticaltohorizontal scenario. Therefore, to de
termine the DG penetration limit for a speciﬁc case, detailed analysis has to
be done on the network. Furthermore, both the sequence of shutting down
CG units and the strategy for substituting reactive power support have to be
speciﬁed.
A straightforward remedy for instability problems that may occur could be
advocated by assigning certain “mustrun” CG units that provide inertia to the
system and, in addition, the reactive power support required. Another approach
as suggested in literature could be for instance to equip the converter connected
generators with control system such that generators contribute in stabilizing the
system [37], [43].
If higher DG penetration levels are wanted, suﬃcient inertia and voltage
support must be installed. Doing so, in theory one could consider aiming for
100% DG. This is the subject of the Chapter 7.
Chapter 6
Stochastic Approach to
Transient Stability of Power
Systems with DG
In Chapters 4 and 5 the transient stability of the power systems is investigated
based on a deterministic approach. In this approach, power system parameters
are set at the rated values. It is found that the transient stability performance
of power systems depends on the (prefault) load ﬂow in the system.
It is well known, however, that some parameters of a power system do not
behave in a deterministic way. For example, demand ﬂuctuates due to the be
havior of customers, that is by deﬁnition stochastic. The implementation of
renewable–energy–source–based DG also implies a more stochastic approach on
the electricity generation side (Chapter 2). This stochastic behavior impacts,
among other things, the load ﬂow within the system. Therefore, one may think
of analysis methods that incorporate the stochastic nature to investigate the
transient stability of the power systems. In fact, this approach is more compre
hensive than the deterministic one.
In this chapter, we extend the transient stability analysis to and focus more
on the stochastic approach. As the stochastic behavior of a power system is, by
nature, extremely complex, we restrict ourselves to the study of the stochastic
behavior of the DG output (Chapter 2). Throughout the analysis we assume
that every DG unit is a customerowned synchronous generators; i.e. the owners
decide whether the units operate or not. Even though the DG units are now in
essence “deterministic” they still characterize stochastic generation. The results
from this approach show that the inclusion of the stochastic nature of DG leads
to a more complete and detailed view of the system transient stability.
We begin our discussion with the impact of stochastic behavior of DG on
the load ﬂow in Section 6.1. The impact on the system transient stability is
discussed in Sections 6.2 and 6.3.
92 6.1 Stochastic Load Flow
6.1 Stochastic Load Flow
In the classical “verticallyoperated” power system there is only a “small” num
ber of large centralized generators “dispatchable”, i.e. controllable to meet the
demand. In the “horizontallyoperated” power system, the DG units in the
“active” distribution networks, (called Active Distribution Systems (ADS)), are
basically “nondispatchable”. This nondispatchable behavior results from the
fact that certain DG units generate power from primary energy sources with in
herently stochastic characteristics, such as wind and solar energy (Chapter 2).
Even when DG units are in essence “deterministic”, if they are customerowned
units, the owners can decide whether the units operate or not. In both cases
DG possesses a stochastic generation characteristic, and the modeling discussed
in the previous chapters needs be adjusted to account for this.
The power ﬂow solution of a system with stochastic DG is computed by
including the stochastic behavior of the parameters of the speciﬁed active and
reactive power of each bus representing ADS in a power system. With DG
implemented in every load bus, each load bus (i.e. distribution network) contains
both consumption and generation (modeled as negative consumption) in the
steadystate simulation. Therefore, seen from the transmission level, each active
distribution system (ADS) can be represented by an aggregated load in parallel
with aggregated generation
P
ADS(i)
=
N
L
i
¸
j=1
P
L(i,j)
−
N
DG
i
¸
k=1
P
DG(i,k)
, i = 1, . . . , N
ADS
, (6.1)
with N
ADS
the number of ADS, N
L
i
the number of loads and N
DG
the number
of DG in ADS i. P
ADS(i)
is incorporated in the load ﬂow algorithm (i.e., as
the corresponding P
i,sch
in the Newton Rhapson algorithm described in Ap
pendix D, by considering that both the rigthhand side terms of equation (6.1)
contain stochastic elements and a stochastic load ﬂow results.
6.2 Stochastic Transient Stability Analysis
A method is proposed to investigate the impact of a largescale DG imple
mentation on the power system transient stability in which both the stochastic
behavior of the DG units and the loads are taken into account. The test system
and the simulation setup are set according to Section 3.5.
6.2.1 Simulation Scenario
A 50% DG level is considered, whereas the total load of the test system is kept
constant. The fraction of the total load served by the DG is distributed among
the DG units, proportional to the active power consumed by the load at that
particular bus. The remaining power generation is divided among dispatchable
CG units by considering the economic operation of the power system, similar
Stochastic Approach to Transient Stability of Power Systems with DG 93
to the approach done in Chapter 5. Minimum and maximum loading limits,
as well as the cost models of the CG units are chosen in such a way that the
highernumbered CG units are more expensive than the lowernumbered ones
(e.g. power production with CG
3
is more expensive than with CG
2
). An optimal
power ﬂow program, whose objective function is to minimize the fuel cost, is run
with a DG penetration level increased to 50%. The most ineﬃcient CG units
whose power output falls below their minimum loading limits are shut down
and switched shunt devices are implemented at the location of the shut down
CG units to compensate for the former reactive power generation. In this way,
5 CG units are shut down (i.e CG nr. 6 to 10; see Appendix B).
The DG units are modeled as synchronous generators without grid voltage
and frequency control (Chapter 3). A DG unit is connected to every load bus via
a j0.05 pu impedance on the 100 MVA system base, and represents the aggregate
active power generation of all DG units in an ADS [18]. Customerowned DG
units that are installed in the ADS supply only active power (1 MW rated active
power each) and no reactive power. The owners decide whether the units are
running or not: i.e. the DG units are randomly connected to and disconnected
from the network. Therefore, the aggregated DG unit is stochastically calculated
as a binomial distribution where each DG unit within the DES is connected to
the system with a probability p
(DG=on)
. We set that p
(DG=on)
×
¸
N
DG
i
k=1
P
DG(i,k)
is equal to the load value at bus i. Thus, the maximum DG capacity installed in
one load bus (
¸
N
DG
i
k=1
P
DG(i,k)
) can be 1/p
(DG=on)
times the rated value. The
loads are following a normal distribution, where the mean values equal the rated
values (Appendix B), and the standard deviations equal σ. Several scenarios
are developed in which both parameters p
(DG=on)
and σ are varied according
to Table 6.1.
Table 6.1: Scenario cases
Case Loads’ standard deviation Probability that DG unit is turned on
(nr.) (σ) (p
(DG=on)
)
1 0.01 0.8
2 0.03 0.8
3 0.05 0.8
4 0.03 0.5
5 0.03 0.7
6 0.03 0.9
A Monte Carlo simulation (MCS) is used to generate samples of aggregate
DG output power and load at each load bus. 10,000 samples are generated in
each scenario. These 10,000 samples are considered to be suﬃcient since the
MCS converges after 10,000 samples.
94 6.2 Stochastic Transient Stability Analysis
6.2.2 Monte Carlo simulation (MCS) Samples
Figure 6.1 shows the MCS generated samples of the active power consumption of
each load. The loads are normally distributed functions, where the mean values
equal the rated values and the standard deviations σ vary according to scenario
nr. 1, 2 and 3 in Table 6.1. The reactive load samples are generated based on
the MCS samples of the active load data, by keeping the active and reactive
power ratio (P
i
/Q
i
) at each load bus (bus i) constant. From Figure 6.1, one
can observe larger spread around the rated load values in case of an increasing
standard deviation (σ).
0 200 400 600 800 1000 1200 1400
0
200
400
600
800
1000
N
o
.
o
f
S
a
m
p
l
e
s
STD dev. = 0.01
3 4
7
8
1
2
1
5
1
6
1
8
2
0
2
1
2
3
2
4 2
5
2
6
2
7
2
8
2
9
3
1
3
9
0 200 400 600 800 1000 1200 1400
0
200
400
600
800
1000
N
o
.
o
f
S
a
m
p
l
e
s
STD dev. = 0.03
3 4
7
8
1
2
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
1
3
9
0 200 400 600 800 1000 1200 1400
0
200
400
600
800
1000
Active (Consumed) Power [MW]
N
o
.
o
f
S
a
m
p
l
e
s
STD dev. = 0.05
3 4 7
8
1
2
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
1
3
9
Figure 6.1: MCS generated samples of the active power consumption of each
load bus (Numbers in the graphs indicate the number of the load bus)
Figure 6.2 shows the MCS generated samples of the DG power generation at
each load bus (except for bus nr. 12 and 31, not displayed due to small values).
Figure 6.2 indicates that the spread of the MCS samples of the DG power
generation decreases along with a rising probability p
(DG=on)
. The reason is
because we set the average power output of the DG units as the rated. Therefore,
if each DG unit has a lower probability to be connected to the grid, more DG
units will be implemented in order to generate the average power equal to the
rated value.
For each MCS sample, the transient stability of the test system is investi
gated by applying a fault to the transmission line between the buses 15 and 16
of the test system. After 150 ms, the fault is cleared by tripping the faulty line.
None of the CG or DG is disconnected during the fault.
Stochastic Approach to Transient Stability of Power Systems with DG 95
0 100 200 300 400 500 600
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
p
DG=on
= 0.5
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
0 100 200 300 400 500 600
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
p
DG=on
= 0.7
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7 2
8
2
9 3
9
0 100 200 300 400 500 600
0
1000
2000
3000
4000
Active (Generated) Power [MW]
N
o
.
o
f
S
a
m
p
l
e
s
p
DG=on
= 0.9
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9 3
9
Figure 6.2: MCS generated samples of the DG power generation at each load
bus (Numbers in the graphs indicate the number of the load bus)
To assess the transmission system stability, two indicators are applied to
quantify the rotor speed oscillations of the CG in the system: maximum rotor
speed deviation and oscillation duration (Chapter 3).
Algorithm 4 shows the stochastic transient analysis of power systems with
DG in an ADS.
6.2.3 Simulation Results
We run Algorithm 4 for all cases given in Table 6.1. The results are shown in
Figures 6.3 and 6.4.
For Cases No. 1–3 the computations show that the spread of the total
active power ﬂow in the lines becomes larger with increasing spread in the load
(Figure 6.3 (left)). An immediate consequence of this growth is an increase of
the spread of probability distributions of the maximum rotor speed deviation
(Figure 6.3 (right)). It is obvious from these results that large power ﬂows have
a detrimental eﬀect on the damping of oscillations. This is in accordance with
what already explained in Chapters 4 and 5.
In Figure 6.4 (left), the spread of the stored kinetic energy is shown for
Cases 4–6. Down from Case 6 to 4, the spread of the stored kinetic energy in
the system rises as the probability of the DG units to be connected to the grid
decreases and only the rotating masses of the connected DG units contribute
to the total stored kinetic energy in the system. Due to this eﬀect, the spread
96 6.2 Stochastic Transient Stability Analysis
Algorithm 4 Stochastic transient stability analysis
1: Set test system parameters, with i = 1, · · · , N the number of buses
2: Set DG
level
, the mean (rated value) µ
i
and the standard deviation σ
i
3: Set the number of generators connected in each busi N
DG,i
and the prob
ability of DG being connected p
(DG=on)
4: Set N
MCS
:= 1000
5: for k = 1, . . . , until convergent do
6: for j = 1 to N
MCS
do
7: Generate normal random numbers for P
i
and Q
i
with µ
i
and σ
i
8: Generate random numbers for P
DG,i
from the binomial distribution
with parameters N
DG,i
and p
(DG=on)
9: Simulate a permanent fault 150 ms (t
fault
)
10: Calculate maximum rotor speed deviation and oscillation duration
11: if MCS convergent with N
MCS
samples then
12: Calculate probability distribution function (pdf) of maximum rotor
speed deviation and oscillation duration
13: quit
14: end if
15: end for
16: N
MCS
:= N
MCS
+ 1000
17: end for
8000 8500 9000 9500 10000 10500 11000
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
Case 1
5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
0
1000
2000
3000
4000
Case 1
8000 8500 9000 9500 10000 10500 11000
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
Case 2
5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
0
1000
2000
3000
4000
Case 2
8000 8500 9000 9500 10000 10500 11000
0
1000
2000
3000
4000
Active Power Flow [MW]
N
o
.
o
f
S
a
m
p
l
e
s
Case 3
5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
0
1000
2000
3000
4000
Max. Rotor Speed Deviation [x 10 pu]
Case 3
−3
Figure 6.3: Total active power ﬂows (left) and maximum rotor speed deviations
(right) in the system in Case 1 (upper), Case 2 (middle) and Case 3 (lower)
Stochastic Approach to Transient Stability of Power Systems with DG 97
3.25 3.3 3.35 3.4 3.45
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
Case 4
5.5 6 6.5
0
1000
2000
3000
4000
Case 4
3.25 3.3 3.35 3.4 3.45
0
1000
2000
3000
4000
N
o
.
o
f
S
a
m
p
l
e
s
Case 5
5.5 6 6.5
0
1000
2000
3000
4000
Case 5
3.25 3.3 3.35 3.4 3.45
0
1000
2000
3000
4000
Energy (Kinetic) [x 10 MJ]
N
o
.
o
f
S
a
m
p
l
e
s
Case 6
4
5.5 6 6.5
0
1000
2000
3000
4000
Max. Rotor Speed Deviation [x 10 pu]
Case 6
−3
Figure 6.4: Total stored kinetic energy (left) and maximum rotor speed devi
ations (right) in the system in Case 4 (upper), Case 5 (middle) and Case 6
(lower)
of the probability distributions of the maximum rotor speed deviation increases
slightly (Figure 6.4 (right)). Hence, the lower the stored kinetic energy in the
system (i.e. inertia) becomes, the more vulnerable the system is in terms of
transient stability (e.g. due to disturbance) [25], [33] (Chapters 4 and 5).
The stochastic behavior of consumption and generation within an ADS re
sults in stochastic power ﬂows in the lines. In addition, the stochastic behavior
of connecting/disconnecting DG units results in a stochastic amount of stored
kinetic energy in the system. Accordingly, the transient stability analysis does
not result in a single value of the stability indicators, but in a probabilistic distri
bution of the stability indicators. In other words, in stead of a blackandwhite
statement on the system stability, a more nuanced result is obtained.
Table 6.2 gives a comparison of the indicators obtained from the determin
istic approach and the ones (average values) obtained from the stochastic ap
proach. Especially the average values of the oscillation duration are signiﬁcantly
higher than the ones obtained from the deterministic approach, which gives an
indication that the system tends to be more unstable when the stochastic be
havior of the ADS is taken into account.
98 6.2 Stochastic Transient Stability Analysis
Table 6.2: Maximum rotor speed deviations (max. dev.) and oscillation du
rations (osc. dur.) that result from both deterministic and stochastic stability
analysis
Case nr. Deterministic approach Stochastic approach
(average value)
Max. dev. [pu] Osc. dur. [s] Max. dev. [pu] Osc. dur. [s]
1 0.0057 7.16 0.0057 7.16
2 0.0057 7.16 0.0058 7.25
3 0.0057 7.16 0.0058 7.48
4 0.0057 7.16 0.0058 7.30
5 0.0057 7.16 0.0058 7.26
6 0.0057 7.16 0.0058 7.23
An advantage of the stochastic transient analysis is that it does not result in
a single value of the stability indicators, but in a probabilistic distribution of the
stability indicators. Although the average values of the maximum rotor speed
deviations are almost equal to the deterministic ones (Table 6.2), the proba
bility that the value of the indicators obtained from the stochastic approach is
larger than the ones obtained from the deterministic approach can be quite high
(Table 6.3). This means that it gives a more complete picture of the transient
stability of a system with DG in ADS.
Table 6.3: Probability that the values of the stability indicators that result
from the stochastic approach are larger than the ones from the deterministic
approach
Case nr. Probability [%]
Maximum rotor speed deviation [pu] Oscillation duration [s]
1 38.0 1.3
2 55.6 70.8
3 68.4 92.4
4 58.0 82.3
5 55.2 85.0
6 51.2 70.8
Stochastic Approach to Transient Stability of Power Systems with DG 99
6.3 Stochastic Transient Stability Study with In
creasing DG
The stochastic approach is used in this section to investigate the transient stabil
ity of power system that goes through a ‘verticaltohorizontal’ transformation
(Chapter 5).
6.3.1 Simulation Scenario
A ‘verticaltohorizontal’ transformation is obtained by gradually increasing the
DG penetration level, i.e. by increasing the fraction of the total load in the
test system served by DG in steps of 10% up to 50%. The remaining power is
divided among the dispatchable CG units by considering the economic operation
of the power system, similar to the approach in Chapter 5. As in Section 6.2,
a scheme of minimum and maximum loading limits and the cost models of the
CG units are taken in such a way that the highernumbered CG units are the
more expensive ones. An optimal power ﬂow program, with objective function
to minimize the fuel cost, is run with a DG penetration level from 10% to 50%.
The most ineﬃcient CG units whose power output falls below their minimum
loading limits are shut down. Switched shunt devices are implemented at the
location of the shut down CG units to compensate for the reactive power lost.
6.3.2 MCS Samples
The total (aggregated) customerowned DG is stochastically calculated as a
binomial distribution where each DG unit within the distribution network is
connected to the system with a probability (p
on
) equal to 0.8. The penetration
level of DG is deﬁned as: (
(p
on
×P
DG,installed
)
P
load,nominal
) × 100%. Thus, the maximum
capacity of DG installed at one load bus can be 125% of the rated load value
(i.e. when 100% DG is supposed). The DG units supply only active power
(1 MW rated active power each) and no reactive power. The loads are following
a normal distribution, where the average is the rated load (Appendix B), and the
standard deviations equal 2.5%. 5,000 Samples are generated for the aggregate
DG output power and the load at each load bus, in every scenario of increasing
DG penetration level. These 5,000 samples are considered to be suﬃcient since
the MCS converges after 5,000 samples.
Figure 6.5 shows the contours of the histograms of the MCS generated sam
ples representing the DG power generation at each load bus, (except at bus nr.
12 and 31, not shown due to small values), at 10 to 50% DG penetration level.
The transient stability of the test system is investigated by applying a per
manent fault to the transmission line between buses 15 and 16 for each MCS
sample. The fault is cleared by tripping the faulty line after 150 ms. To assess
the transmission system stability, two indicators are applied to quantify the ro
tor speed oscillations of the CG in the system, maximum rotor speed deviation
and oscillation duration (Chapter 3).
100 6.3 Stochastic Transient Stability Study with Increasing DG
0 100 200 300 400 500 600
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
10% DG Level
3 4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
3 4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
0 100 200 300 400 500 600
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
20% DG Level
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
0 100 200 300 400 500 600
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
30% DG Level
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5 2
6
2
7
2
8
2
9
3
9
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5 2
6
2
7
2
8
2
9
3
9
0 100 200 300 400 500 600
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
40% DG Level 3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7
2
8
2
9
3
9
0 100 200 300 400 500 600
0
1000
2000
Active (Generated) Power [MW]
N
o
.
o
f
S
a
m
p
l
e
s
50% DG Level
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7 2
8
2
9
3
9
3
4
7
8
1
5
1
6
1
8
2
0
2
1
2
3
2
4
2
5
2
6
2
7 2
8
2
9
3
9
Figure 6.5: MCS generated samples representing the DG power generation at
each load bus at a 10%, 20%, 30%, 40% and 50% DG penetration level
6.3.3 Simulation Results
Figure 6.6 shows a number of scenarios of the ‘verticaltohorizontal’ power
system transformation. It can be observed from Figure 6.6 that there are ﬁve
CG units operating within all scenarios (DG penetration level from 0% to 50%):
CG nr. 1, 2, 3, 4, and 5 (i.e. the lowest 5 bars of ﬁg. 6.6). To compare the
transient stability performance in the diﬀerent scenarios, the transient stability
indicators are applied on these ﬁve CG units.
The values of both stability indicators, i.e. the average value in case of the
stochastic approach and the deterministic value in the deterministic approach,
are shown in Table 6.4.
Stochastic Approach to Transient Stability of Power Systems with DG 101
0 10 20 30 40 50
6000
5000
4000
3000
2000
1000
7000
0
CG−9
CG−10
CG−8
CG−7
CG−6
CG−5
CG−1
CG−2
CG−3
CG−4
DG Penetration Level [%]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
Figure 6.6: Dispatched CGs as a function of the DG penetration level
Table 6.4: Maximum rotor speed deviations and oscillation durations result
ing from the deterministic and the stochastic (average) approaches of all MCS
simulations at the ﬁve DG penetration levels
Deterministic approach Stochastic approach
DG Max. Oscillation Average of the max. Average of the
level rotor speed duration rotor speed oscillation
Scenario deviation deviation duration
(%) [pu] [s] [pu] [s]
10 0.0053 6.31 0.0053 6.32
20 0.0055 7.09 0.0055 7.09
30 0.0055 7.36 0.0055 7.07
40 0.0055 6.31 0.0055 6.86
50 0.0063 6.88 0.0063 7.20
From Table 6.4, it can be concluded that the stochastic approach and the
deterministic approach result in the same average maximum rotor speed devi
ation. The oscillation duration is, however, diﬀerent for both approaches. The
latter fact gives an indication that the results obtained from the determinis
tic approach may give either too pessimistic (e.g. at the 30% DG penetration
level) or too optimistic (e.g. at the 40% and 50% DG penetration level) results
compared to the results obtained from the stochastic approach.
102 6.3 Stochastic Transient Stability Study with Increasing DG
P
r
o
b
a
b
i
l
i
t
y
P
r
1 2
x
%DG Level (i+1) %DG Level (i)
x x
Indicator values
Figure 6.7: Probability distributions of stability indicators overlap
Furthermore, with the stochastic approach proposed in this section, the
probability distributions of the stability indicators are obtained and contain
additional information. First of all, the blackandwhite statement that the sys
tem becomes more unstable when e.g. the DG penetration level increases from
10% to 20% can be more nuanced. Equation (6.2) and Figure 6.7 illustrate the
use of this approach. In Figure 6.8, for example, is shown that the probability
distributions of the maximum rotor speed deviations overlap in some ranges.
P
r
(g
1
(x) > g
2
(x)) =
x
2
x
1
g
1
(x)
x
x
1
g
2
(y) dy
dx. (6.2)
Discretizing and applying (6.2) to the results obtained from the stochastic ap
proach, leads to Table 6.5.
Table 6.5: Probability that the indicators resulting from the scenario with lower
DG level surpass the indicators resulting from the scenario with the onestep
higher DG level
%DG level P
r
(g
1
(x) > g
2
(x))
g
1
(x) g
2
(x) x = indicator x = indicator
of max. rotor speed deviation of oscillation duration
10% 20% 0.0043 0
20% 30% 0.1406 0.3984
30% 40% 0.3211 0.6942
40% 50% 0 0.2413
Stochastic Approach to Transient Stability of Power Systems with DG 103
5 5.5 6 6.5 7
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
10% DG Level
5 5.5 6 6.5 7
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
20% DG Level
5 5.5 6 6.5 7
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
30% DG Level
5 5.5 6 6.5 7
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
40% DG Level
5 5.5 6 6.5 7
0
1000
2000
N
o
.
o
f
S
a
m
p
l
e
s
50% DG Level
Max. Rotor Speed Deviation [x 10 pu]
−3
Figure 6.8: Histograms of the maximum rotor speed deviations (in 10
−3
pu).
Note that the xaxis and yaxis of all graphs are adjusted to be uniform
6.4 Conclusions
Results from the preceding chapters suggest that the transient stability perfor
mance of power systems depends on the (prefault) load ﬂow within the system.
Moreover, the implementation of DG, based on renewable energy sources, im
plies more stochastic behavior of the electricity generation. Therefore, stochastic
approach should be considered to study the transient stability of power systems.
This chapter deals with examining the consequences of a stochastic approach
on the system transient stability. The study is mainly focused on the stochastic
behavior of the DG output, that is simulated using Monte Carlo Simulation
method, within the framework of the “verticaltohorizontal” transformation.
104 6.4 Conclusions
It is shown that although many simpliﬁcations are applied for the stochas
tic parameters, the inclusion of the stochastic behavior of DG leads to a more
complete and detailed view of the steady state situations. When the transient
response of the system is simulated, the average values of the stability indica
tors obtained from the stochastic approach may be the same as the indicators
obtained from the deterministic approach. Yet, in the stochastic approach, also
the probability distributions for the indicators are estimated. When compared
to the single result of the deterministic approach, it is shown that the determin
istic approach may give either undervalued or overvalued results.
The probability distributions of the stability indicators of two diﬀerent DG
penetration levels may overlap. Therefore, the deterministic statement that a
system becomes more stable (or unstable) when the DG penetration level is
increased can be more nuanced as that the system becomes more stable (or
unstable) to certain degrees.
The merit of stochastic stability studies is evident even in the case of a rel
atively simple example studied here. Therefore, a stochastic approach becomes
a necessity when, e.g. DG units depend on highly intermittent energy sources
(e.g. wind and solar energy), or when the dependence between the generation
of DG units and the loads becomes important.
Chapter 7
Maintaining Power Balance
with Active Distribution
Systems
This chapter investigates the situation when the power system is pushed towards
a scenario, where DG penetration reaches a level that covers the total load of
the original power system (100% DG implementation in the system).
The assumption is that all DG units are implemented via powerelectronic
converters within active distribution systems (ADS). The ADS are connected to
the transmission system also via powerelectronic interfaces. The power system
is still connected to one source that provides a constant 50 Hz voltage. This can
be, for example, a connection (tie line) to a strong external system. This source
is meant only to gives a (system) frequency reference for the other generators
and generates no power at steady state. Therefore, any power imbalance in the
power system must be compensated by generators in the ADS.
Due to the powerelectronic interfaces, the output power of all generators
within the ADS are decoupled from the system frequency. Therefore, voltage
deviations are proposed to detect power imbalances in the system. Remedies to
eliminate the negative consequences of using the voltage deviations to detect the
power imbalances are discussed and appropriate control systems are suggested.
Figure 7.1 shows an illustration of the power system with ADS.
7.1 Background
The implementation of the distributed generation (DG) turns the current passive
distribution network into an active one (Chapter 1). This active distribution
network does not only consume, but it also generates power and supplies it to
the transmission system [41], [55]. In this way, power can be transferred from
one distribution network to another. When we reﬂect further on this issue to the
106 7.2 Power Balance
3
3
3
Figure 7.1: Illustration of the power system with ADS
extreme, we could imagine that at a certain moment in time the electrical power
generated by the DG within the distribution networks may become suﬃcient to
cover the total demand in the transmission system.
Most of the DG implementations based on renewable energy sources or
environmentallyfriendly technologies are connected to the grid via power elec
tronic converters (Chapter 2). Moreover, when the distribution networks evolve
to become more hybrid, including ACDC, and with ﬂexible storage systems,
it is envisaged that connecting these distribution networks to the transmission
system via power electronic interfaces will be an interesting option [19].
7.2 Power Balance
One of the primary tasks of power system operation is to balance the electricity
supply from generators and loads including the losses in the network at any
time. However, a disturbance may occur in the system that causes a power
imbalance. It is important that the power system can restore the power bal
ance and return to a stable state after this disturbance, called power system
stability [34] (Chapter 3).
In this chapter, all generators within the ADS are ‘hidden’ behind power
electronic interfaces. The output power of the generators are decoupled from
the grid frequency. On the other hand, the source that provides the constant 50
Hz voltage is meant only to give a frequency reference for the other generators
(in the ADS) and not meant to generate power at steady state. Therefore, any
Maintaining Power Balance with Active Distribution Systems 107
power imbalance in the power system must be compensated by generators in the
ADS. However, as a result of the decoupling between the output power and the
grid frequency in the ADS, the power imbalance cannot be detected by the ADS
in the classical way, as an altered system frequency. Therefore in this chapter,
the use of voltages to detect a power imbalance in the transmission system with
active distribution systems is proposed.
Traditionally, however, two problems may arise from using the voltages to
detect a power imbalance in a power system. Firstly, the voltage (magnitude) at
a bus is aﬀected by the ﬂow of the reactive power. Secondly, the voltages at the
buses throughout a power system are not the same (in contrast to the system
frequency). Therefore in the following sections, remedies to eliminate these
problems are presented. Assumptions are applied on the model of the power
system with ADS to eliminate the ﬁrst problem as it is not essential. To deal
with the latter problem, several control systems are proposed and simulated.
7.3 Model of Power System with ADS
7.3.1 Assumptions
To decouple the changes of the voltages with the changes of the reactive power,
some assumptions are applied on the model of the power system as the following:
• The distribution networks are equipped with reactive power sources. The
reactive power source is suﬃcient to fulﬁll the reactive power needs within
the distribution networks.
• The (active) distribution networks are connected to the transmission sys
tem via power electronic interfaces. The power electronic interface is as
sumed to permit only active power to ﬂow (bidirectional).
• The reactances of the transmission lines are compensated, in such a way
that they behave like resistive lines.
7.3.2 Model of ADS
This chapter focuses on functioning of the ADS to maintain the power balance
in the systems. The ADS are considered as follows:
• The (distributed) generators are connected via power electronic converters
and generate only active power. The generators are initially set to balance
active power demand.
• The loads demand active and reactive power. They are modeled as con
stant impedance and constant power. Electrical motors and the corre
sponding inertias are hidden behind power electronic interfaces from the
grid; therefore, electrical motors, if any, are assumed to be included in the
constant power model of the loads.
• The reactive power is supplied by dedicated reactive power sources. The
reactive power sources are preset to balance the reactive power demand.
108 7.3 Model of Power System with ADS
• A power imbalance is simulated by changing the active power demand of
the load. A reactive power imbalance is not simulated. Therefore, only
active power ﬂows between the ADS and the transmission system. Fur
thermore, to balance the reactive power within the ADS, reactive power
sources are applied and modeled as shunt devices.
Thus in the simulation, each ADS is modeled as shown in Figure 7.2.
s/
Transmission
System
DG
Load
3
ADS
Figure 7.2: Model of ADS used in the simulation of maintaining the power
balance in the system
7.3.3 Generator Models
In the power system with ADS, generators perform the following functions:
• One central generator operates as a reference. It serves as voltage and
frequency reference for the other generators in the system and generates
no power at steady state.
• The distributed generators operate as active power sources. They only
supply active power.
• One (or more) central or decentralized generator serves as a ’slack’ gener
ator that either supplies or absorbs any deﬁcit or surplus of active power
in the system. This generator is assumed to have no current limiter and
to be equipped with suﬃcient (energy) storage.
Three converter connected generator models are used:
• A constant voltage source is used to represent the generator providing
voltage and frequency reference. Figure 7.3 shows the representation of
this generator, where U
s
is the constant (reference) voltage with a ﬁxed
frequency, and Z
s
is a source impedance.
• A constant current source that generates a current in phase with the ter
minal voltages U
t
of the generator (a Phase Locked Loop, PLL, is used for
this purpose) is used to represent the distributed generator that supplies
active power. Figure 7.4 shows the representation of this generator.
• A controlled current source is used to represent the generator serving as
the ’slack’ generator (Figure 7.5).
Maintaining Power Balance with Active Distribution Systems 109
U
s
Z
s
Figure 7.3: Constant voltage source model
Ì
s
U
t
Z
s
Figure 7.4: Constant current source model
Ì
s
Controller
U
t
Figure 7.5: Controlled current source model
A remark should be made that power electronic interfaces driving the out
put of converterconnected (distributed) generators basically represent a voltage
source converter, as mostly used nowadays [70]. Yet, the use of constant cur
rent sources to represent converterconnected generators in this simulation is
supported by the following assumptions (Chapter 3):
• A converter is actually a voltage source converter, but it behaves like a
current source, so that current source models can substitute voltage source
converters when simulations are made in large systems [42].
110 7.4 Basic Controller Model
• It is usual to model the sources as PQsources in studies of large sys
tems [69]. A constant current source generating a current in phase with
the (terminal) voltage represents a PQsource generating constant active
power (P) and zero reactive power (Q = 0), as long as the terminal voltage
of the generator is constant.
• In practice, a converter is equipped with a current limiter. When the
terminal voltage drops, the converter supplies less active power. Thus,
the use of a constant current source corresponds to a converter whose
current is limited to the rated value (in practice, 100% up to 120% of the
rated value).
• When the terminal voltage raises temporarily, the use of the constant
current source can be justiﬁed by assuming that the generator is equipped
with energy storage so that the generator can supply extra active power
for a short period of time.
7.4 Basic Controller Model
The function of maintaining the power balance is supposed to be done by the
controllerblock of the ‘slack’ generator(s). The ‘slack’ generator(s) represented
by the controlled current source(s) (Figure 7.5) must supply or absorb any
deﬁcit or surplus of active power in the system. The basic functionality of the
controllerblock is highlighted hereunder.
U
t
Controller
U
S
Ì
g
Ì
c
P
load
Z
Ì
load
Ì
g Ì
s
Ì
c
1
Figure 7.6: Basic controller idea applied at a singlebus test system
In Figure 7.6, three types of converterconnected generators  a constant
voltage source providing a reference voltage (U
s
), a constant current source
generating a constant current (I
g
) and a controlled current source supplying a
changeable current (I
c
)  and an active load consuming active power (P
load
) are
implemented.
In steady state, the current of the voltage source (master generator providing
a reference voltage and frequency) should be zero (I
s
= 0), so that there is no
voltage drop across the impedance Z. Hence, the terminal voltage (U
t
) of bus1
is equal to the reference voltage U
s
.
Maintaining Power Balance with Active Distribution Systems 111
Any power imbalance should be eliminated by controlling the current I
c
generated by the controlled current source. This controlled current source rep
resents the ’slack’ generator. When, for example, the active power consumption
of the load P
load
increases, the current ﬂow to the load (I
load
) rises. Both gen
erators modeled as current sources do not react (yet) and the voltage source
starts to supply active power in order to balance the power. The voltage source,
representing a strong source acting as the frequency reference, takes care of the
primary action. This source, however, is meant to provide the frequency refer
ence only: the source should not generate active power at steady state. There
fore, as I
s
increases and causes a voltage drop across Z so that U
t
decreases, a
secondary action is taken. The controller of the controlled current source should
detect this voltage drop. As a result, the controlled current source supplies more
active power, i.e. it injects more current I
c
, until the power balance is restored.
The controller used in the basic model is a proportionalintegral controller
(PI controller), a common feedback loop component in industrial control ap
plications. The controller compares a measured value from a process with a
reference setpoint. The diﬀerence or “error” signal is processed to calculate a
new value for a manipulated process input, in which the new value brings the
process measured value back to its desired setpoint. Unlike simpler control al
gorithms, the PI controller can adjust process inputs based on history and rate
of change of the error signal, which gives a more accurate and stable control. It
can be shown mathematically that a PI loop produces accurate stable control in
cases where other control algorithms would either exhibit a steadystate error
or cause the process to oscillate. The mathematical form of PI controller can
be shown as [46]
U(t) = K
P
E(t) +K
I
E(t)dt, (7.1)
where U(t) is output of the controller, E(t) is error signal (diﬀerence between
desired and real output), K
P
is proportional and K
I
integral term.
To verify the basic controller model, a load jump is simulated to cause a
power imbalance in the system (Figure 7.6). The system voltage is set at 10
kV. The load demands 60 MW of active power (P
load
) initially, equally divided in
constant impedance and constant power. The constant current source supplies
initially all demand. The currents generated by the constant voltage source
(I
s
) and the controlled current source (I
g
) are zero. A load jump is applied by
increasing the load modeled as the constant impedance by 30 MW.
Figures 7.7, 7.8 and 7.9 show transients of voltages, currents and active
power when the load jump occurs. Note that the P
‘constantactivepower
′
generator
,
P
‘slack
′
generator
, and P
‘master
′
generator
represent the active power supplied by the
constant current source, controlled current source and constant voltage source
of the system in Figure 7.6 respectively. Simulations are performed on a Real
Time Digital Simulator (RTDS) with a time step of 50 µs.
112 7.4 Basic Controller Model
0 2 4 6 8 10
0
2
4
6
8
10
12
V
o
l
t
a
g
e
(
k
V
)
Time (s)
U
s
U
t
( bus−1)
Figure 7.7: Transients of the voltages when a 30 MW load jump is applied at
bus1 of the system shown in Figure 7.6
0 2 4 6 8 10
0
1
2
3
4
5
6
7
8
9
10
C
u
r
r
e
n
t
(
k
A
)
Time (s)
I
g
I
load
I
c
I
s
Figure 7.8: Transients of the currents when a 30 MW load jump is applied at
bus1 of the system shown in Figure 7.6
Maintaining Power Balance with Active Distribution Systems 113
0 2 4 6 8 10
−10
0
10
20
30
40
50
60
70
80
90
100
A
c
t
i
v
e
P
o
w
e
r
(
M
W
)
Time (s)
P
load
P
‘constant active power’ generator
P
‘slack’ generator
P
‘master’ generator
Figure 7.9: Transients of the active power when a 30 MW load jump is applied
at bus1 of the system shown in Figure 7.6
We can see from these ﬁgures that, following the load jump, the power
balance is restored, and the ’extra’ power demand is ﬁnally supplied at the
steady state by the ’slack’ generator (increasing current I
c
and active power
P
‘slack
′
generator
in Figures 7.8 and 7.9). The voltage level of the bus also restores
to the steadystate value that lies within a +/5% margin of the rated voltage.
7.5 ADS Control Systems
Practically, a power system consists of more than one bus. In this case, the sys
tem is decoupled and no longer linear, due to the interdependency of the buses.
In addition, the voltages at the buses throughout the system are not the same
(Section 7.2). Consequently, applying the basic control model of Section 7.4 to
each bus gives potential diﬃculties, since the basic control model is linear.
In this section, three existing control system methods are proposed [46]:
• Standalone master controller.
• Decentralizedcontroller with single reference.
• Decentralizedcontroller with hysteresis.
To verify these control systems, a simple test system that consists of 3 buses
is deﬁned (Figure 7.10). Tables 7.1 and 7.2 show respectively the component
parameters used and the load ﬂow settings and computed results in the 3bus
test system. Note that G
ref
refers to the reference (‘master’) generator. G
j
refers to the constantpower generator at busj, Load
j
to the load at busj
114 7.5 ADS Control Systems
1 2
T
L12
Z
G
2
G
1
G
ref
Load
1
T
L13
T
L23
Load
3
G
3
3
C
2
Load
2
C
3
C
1
T
1
T
2
T
3
Figure 7.10: A simple 3bus test system
and C
1
to the reactive power source at busj. Also note that the transmission
lines introduce quite some capacitance in the system; each transmission line
generates around 16 Mvar. T
j
indicates the transformer at busj and T
Ljk
the
transmission line between busj and busk. In Table 7.1, S
base
denotes the
complex power base of the test system. U
HV
and U
MV
denote the system high
and medium voltage levels. R, X
L
and B denote the resistance, reactance and
susceptance of the transmission lines. X
T
denotes the transformer reactance
and Z the impedance between the reference ’master’ generator and bus1.
7.5.1 Standalone master controller
The most simple way to overcome the nonlinearity problem is to use a stand
alone master controller. In this approach only one basic controller is applied
and connected to one of the buses of the system. In this case, the controller
only regulates the voltage at one bus and lets the system come to balance using
its own connectivity.
Figure 7.11 shows the implementation of the standalone master controller
in the test system. The ‘slack’ generator represented by the controlled current
source, is implemented at bus1. In this controller scheme, only one ‘slack’
generator is implemented. The generator should handle the active power im
balance occuring in the system. Note that G
Cj
refers to the controlledpower
‘slack’ generator (at busj): in this case G
C1
at bus1.
Maintaining Power Balance with Active Distribution Systems 115
Table 7.1: Component parameters used in the 3bus test system
Description Parameter Value Unit
System base S
base
100 MVA
System voltage U
HV
100 kV
U
MV
10 kV
Transmission lines: R 0.0064 pu
(T
L12
, T
L13
, T
L23
) X
L
0.0322 pu
B
2
0.0306 pu
Transformer X
T
0.1 pu
(T
1
, T
2
, T
3
)
Impedance Z 1.0 pu
Table 7.2: Load ﬂow settings and computed results in the 3bus test system
Description Parameters Setting Computed Setting Computed
(MW) (MW) (Mvar) (Mvar)
Generation Gref 0.0 0.5 0.0 6.5
G
1
60.0 59.9 0.0 2.8
G
2
60.0 60.4 0.0 2.8
G
3
60.0 60.4 0.0 2.8
Load Load
1
60.0 59.0 30.0 31.8
Load
2
60.0 59.7 30.0 32.2
Load
3
60.0 59.7 30.0 32.2
Shunt device C
1
0.0 0.0 15.0 15.0
C
2
0.0 0.0 15.0 15.3
C
3
0.0 0.0 15.0 15.3
A load jump is applied by increasing the constant power load at bus2,
the load being modeled as a constant power of 30 MW. Figure 7.12 shows the
transients of the active power where all the 30 MW power (of the load jump) is
supplied by the ‘slack’ generator. The power balance is restored and all system
parameters are back to stable, steady state values.
When the standalone master controller scheme is implemented, the problem
of ‘diﬀerent signal’ of voltages for the control input is eliminated by using only
one controller at bus1. In this way, the controller uses only one voltage as its
reference signal. However, the challenge is that one generator should compensate
for the power balance of the whole system. In the following sections, other
proposals concentrate on dividing the ‘slack’ generator task.
116 7.5 ADS Control Systems
1 2
Controller1
T
L12
Z
G
2
G
1
G
ref
Load
1
T
L13
T
L23
Load
3
G
3
3
C
2
Load
2
C
3
C
1
T
1
T
2
T
3
U
1
G
C1
Figure 7.11: Implementation of the standalone master controller in the test
system
7.5.2 Decentralizedcontroller with single reference
In practice, it is impossible to use one central generator to compensate for the
whole system. Because of that, in this example three decentralized generators
with their own controllers are applied. However, if these three controllers are
applied ’as is’ in the test system, even though the voltage can be regulated, the
power balance of the generators is not achieved. It is already expected that
the system is decoupled and the voltage at each bus is not exactly the same.
Therefore, decentralizedcontrollers with single reference is proposed here. In
this approach, the ‘slack’ generator at each bus has its own controller, but the
reference signal is common to all of them and can be taken from any voltage
point.
Figure 7.13 shows the implementation of the decentralizedcontroller with
single reference in the test system. Each of the ‘slack’ generators, represented by
the controlled current source is implemented at buses 1, 2 and 3. The generators
altogether should take care of the active power imbalance that occurs in the
system. One control signal is used by all generators, i.e. the voltage at bus1
(U
1
). Note that G
Cj
refers to the controlledpower ‘slack’ generator at busj.
Maintaining Power Balance with Active Distribution Systems 117
0 2 4 6 8 10
−10
0
10
20
30
40
50
60
70
80
90
100
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
Time [s]
G
ref
G
C1
Load
1
, Load
3
Load
2
G
1
, G
2
, G
3
Figure 7.12: Transients of the active power when a 30 MW load jump is applied
at bus2 of the system shown in Figure 7.11
A load jump is applied by increasing the constant power load at bus2,
the load being modeled as constant power, of 30 MW. Figure 7.14 shows the
transients of the active power. The 30 MW power of the load jump is supplied
by the three ‘slack’ generators at buses 1, 2 and 3. Each generator supplies the
required power into balance, i.e. 10 MW. Also, all system parameters are back
to stable, steady state values.
In the same way as in the ﬁrst approach, by implementing the decentralized
controller with a single reference, the problem of ‘diﬀerent signal’ of voltages
for the control input is eliminated by using only one controller at bus1. The
diﬀerence is that this control signal is used for all three controllers. With this
approach the control generator is no longer centralized. However there is one
aspect that should be considered. This approach needs a communication link
between each controller to transfer the reference data signal. It might happen
that not all system has this luxury.
In the next section, the third method that can be applied for the system
without communication link between each controller is described.
118 7.5 ADS Control Systems
1 2
Controller1
T
L12
Z
G
2
G
1
G
ref
Load
1
T
L13
T
L23
Load
3
G
3
3
C
2
Load
2
C
3
C
1
T
1
T
2
T
3
U
1
G
C1
G
C3
Controller3
U
1
G
C2
Controller2
U
1
Figure 7.13: Implementation of the decentralizedcontroller with single reference
in the test system
7.5.3 Decentralized controller with hysteresis
The third approach proposed uses three decentralized controllers, one applied
at each bus. As discussed above, the linear controller cannot work perfectly
in the nonlinear decoupled system. The symptom that happens by applying
the basic controller over the test system, is that in steadystate the controller
oscillates as the system is decoupled and the reference signal is not exactly the
same.
To prevent this symptom, hysteresis is applied in the controller input: the
controller stops regulating the system whenever the value of the voltage lies
within the hysteresis boundary. The use of hysteresis is allowed as long as the
width of it is less or equal than the tolerance of the voltage, being 1% of the
rated value.
Figure 7.15 shows the implementation of the decentralized controller with
hysteresis. The ‘slack’ generators represented by the controlled current sources
are implemented at buses 1, 2 and 3. One ‘slack’ generator is implemented at
each bus. In this control scheme, each ‘slack’ generator uses the voltage where
the generator is implemented. The generators altogether should take care of the
active power imbalance occurring in the system. Note that G
Cj
refers to the
‘slack’ generator at busj. The generator G
Cj
is also linked to the control signal
the voltage of busj (U
j
).
Maintaining Power Balance with Active Distribution Systems 119
0 2 4 6 8 10
−10
0
10
20
30
40
50
60
70
80
90
Time [s]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
Load
2
G
1
, G
2
, G
3
Load
1
, Load
3
G
ref
G
C1
, G
C2
, G
C3
Figure 7.14: Transients of the active power when a 30 MW load jump is applied
at bus2 of the system shown in Figure 7.13
1 2
Controller1
T
L12
Z
G
2
G
1
G
ref
Load
1
T
L13
T
L23
Load
3
G
3
3
C
2
Load
2
C
3
C
1
T
1
T
2
T
3
U
1
G
C1
G
C3
Controller3
U
3
G
C2
Controller2
U
2
Figure 7.15: Implementation of the robust controller in the test system
120 7.6 Conclusions
A load jump is applied by increasing the load at bus2, the load being mod
eled as constant power of 30 MW. Figure 7.16 shows the transients of the active
power. The 30 MW power of the load jump is supplied by the three ‘slack’
generators at buses 1, 2 and 3, each generator supplying 10 MW. All system
parameters are restored to stable, steady state values.
0 2 4 6 8 10
−10
0
10
20
30
40
50
60
70
80
90
Time [s]
A
c
t
i
v
e
P
o
w
e
r
[
M
W
]
Load
2
G
1
, G
2
, G
3
Load
1
, Load
3
G
ref
G
C1
, G
C2
, G
C3
Figure 7.16: Transients of the active power when a 30 MW load jump is applied
at bus2 of the system shown in Figure 7.15
By applying this approach, the power balance of the system can be achieved.
The eﬀect of hysteresis is that there is a steady state error on the voltage, but
while the steady state value is smaller than the tolerance of the voltage value,
it can be accepted. Thus, the voltage regulation is maintained. It might be not
the optimal solution, but the system still operates within the speciﬁcations. The
main advantage of this approach is that the requirement of the communication
link between controllers is eliminated.
7.6 Conclusions
This chapter presents models and control systems for maintaining the power
balance in the power system with active distribution networks (ADS), where
the DG units covering the total demand of the original power system are im
plemented within the ADS via powerelectronic converters. Moreover, the ADS
are connected to the transmission system via powerelectronic interfaces.
Maintaining Power Balance with Active Distribution Systems 121
Due to the powerelectronic interfaces, the output power of the DG units and
the ADS are decoupled from the grid frequency. As a result, the ADS cannot
detect power imbalance as an altered system frequency such as in classical power
systems [44], [43]. Therefore, in this chapter the idea is developed to use the
voltage to detect and maintain the power balance.
Some assumptions are applied to the power system model to deal with the
problem that the voltage magnitude at a bus is aﬀected by the reactive power
ﬂow. Moreover, control systems are developed to deal with the problem that
the voltages at the buses throughout the power system are not the same.
For this purpose, three possible control system concepts are adjusted and
applied: standalone master controller, decentralizedcontroller with single ref
erence, and decentralized controller with hysteresis. The simulation results show
that by applying each of these three control systems, the power balance in the
power system can be maintained by the ADS. In the ﬁrst control system, the
power imbalance is taken care of by one ADS unit while in the latter two, the
power imbalance is taken care of by several ADS units. The simulation results
show that when a load jump is applied in the power system to simulate the
power imbalance, the power balance is restored by means of the ADS with each
of the three control systems developed, and all system parameters return to
stable, steady state values.
Each control approach proposed here has its own advantages and drawbacks.
The standalone master controller is the most simple one, easy to implement
and has a good performance. The main disadvantage is that it uses a single
generator to supply the required power. The second approach, the decentralized
controller with single reference can overcome the single generator problem, at
the price of requiring a communication link between controllers to transfer the
reference signal data. The most interesting approach to be implemented is
the third one, the decentralized controller with hysteresis, since this approach
uses a split generator for each bus and in addition the three controllers are
totally independent. Thus, the need of communication between controllers is
eliminated.
It should be noted that the primary action is taken care of by a strong source
which acts as frequency reference.
122 7.6 Conclusions
Chapter 8
Conclusions
8.1 Overview
Currently, concerns on environmental issues lead to an increasing implementa
tion of environmentallyfriendly generation in electrical power systems. Many of
these generation technologies tend to be connected to the distribution systems,
in which they are referred to as distributed/decentralized/embedded generation
(DG). In literature, the technical impact of DG on the distribution systems have
been studied often. Due to the small DG penetration level, the impact of DG
on the transmission system has not been treated so often. However, when the
penetration level of DG in a power system becomes higher, it can be expected
that the impact is no longer restricted to the distribution systems, but inﬂuences
the system as a whole.
In this work, we have investigated and discussed the impact of DG on power
systems with a focus on the transmission system. Power system studies are
usually based on computer simulations. This simulationbased approach is used
in this work as well. For this purpose, we have introduced the concept of a
‘verticaltohorizontal’ transformation of power systems. This transformation is
the result of an increasing DG penetration level in the system. To study the
DG impact on power systems, simulations have been performed on scenarios
developed within the ‘verticaltohorizontal’ transformation context.
DG technology and especially the way DG is connected to the grid determine
DG characteristics inﬂuencing a power system. Dispatchable/nondispatchable
DG or direct/indirect gridconnected DG are included in these characteristics,
and become the basis for the DG models used in the system simulation.
We have focused on the impacts of DG on the system transient stability, and
more speciﬁcally on the rotor angle stability. By means of the swing equation,
power angle equation, and equal area criterion, the transient behavior of the
rotational dynamics of synchronous machines is studied. To assess the system
transient stability, after a disturbance, we have used two indicators: maximum
rotor speed deviation and oscillation duration, rather than the classical sys
124 8.1 Overview
tem stability indicators: critical clearing angle and the corresponding critical
clearing time. Due to the large number of scenarios to be simulated in combi
nation with a relatively large power system, the use of the classical indicators
is impractical.
Simulation have been performed on the wellknown 39bus New England
transmission test system. This test system has several advantages: not only
does it represents a relatively large power system, the system parameters are
easily accessible and it has been widely used in literature.
Distribution systems, with DG incorporated, are represented in a simpli
ﬁed way according to models found in literature. The power system simulation
software package PSS/E v.25.4 has been used to simulate the test system. DG
based on classical rotating machines is modelled by means of the available mod
els in the software package libraries. Userwritten models have been developed
for DG based on powerelectronic converters.
We have simulated the incorporation of DG in the test system with diﬀerent
increasing load scenarios, DG gridconnection strengths, DG technologies, and
protection schemes of powerelectronic interfaced DG. As expected, these factors
inﬂuence the power system transient stability in a diﬀerent way. There appears
to be no signiﬁcant stability problem up to about 30% DG penetration level.
This 30% DG penetration is obtained when DG is implemented to cover a 50%
load increase. Therefore, we consider this 30% DG penetration level to be a
high DG penetration level in the system. In general, when DG, regardless the
technology, is implemented to cover a load increase in the system, the transient
stability is better than when only the centralized generation (CG) covers the
increasing load. This makes sense as all centralized generators remain in the
system  including their active and their reactive power controls and the energy
stored in their rotating masses, along with the increasing DG penetration levels.
In this way, implementing DG is a natural way of ’limiting’ the power ﬂows in
the transmission lines. Therefore it inherently improves transient stability of
the transmission system, since large power ﬂows have a detrimental eﬀect on
the damping of the oscillations.
To simulate the ‘verticaltohorizontal’ power system transformation, scenar
ios have been developed in which the DG penetration level is raised while the
load is constant. This leads to an overcapacity in the system, and an economic
dispatch is used in these simulations so that the most ineﬃcient CG unit(s)
is(are) taken out of service (eﬃciency considerations). DG has been imple
mented as powerelectronic interfaced DG so that it does not contribute to the
inertia (stored kinetic energy) of the system. In this way, the total inertia of the
system reduces as CG units are taken out of service, and the system becomes
more vulnerable in the face of a disturbance. However, the fact that the DG
generates power close to the loads causes a reduction in the total real power
ﬂowing in the system, and through the faulty branch. This appears to coun
teract and compensate the tendency of the system to become more unstable.
Therefore, by optimizing the load ﬂow in the system (e.g. by rescheduling the
output of the CG units), the transient stability of the system that goes through
the ‘verticaltohorizontal’ transformation can even be improved.
Conclusions 125
The simulation results within this ‘vertical to horizontal’ power system trans
formation framework also suggest that a DG penetration limit exists and this
limit is inﬂuenced by the total kinetic energy remaining in the system, and
the reactive power support scheme. Therefore, a straightforward remedy for
instability problems that may occur due to the DG implementation could be
advocated by assigning certain CG units as synchronous condensers, that pro
vide inertia to the system and, in addition, the needed reactive power support.
The simulation of a power system reaching 100% DG implementation has
been done where assumptions are taken that all DG units are implemented via
powerelectronic converters within active distribution systems (ADS). Moreover,
the ADS are connected to the transmission system via powerelectronic inter
faces too. It has been modeled that the power system is connected to one source
providing a constant 50 Hz voltage, that can be, for instance, a connection (tie
line) to an external system. This source is meant only to gives a (system) fre
quency reference for the other generators and supplies no power at steady state.
Since the the output power of all generators within the ADS are decoupled from
the system frequency due to the powerelectronic interfaces, the voltages have
been used so that the ADS can detect and maintain the power balance in the
power system.
8.2 Stochastic Stability Studies
The merit of stochastic stability studies is evident even in the case of a relatively
simple example as studied in this thesis. The stochastic approach becomes
a necessity when, for example, DG units are based on (highly) intermittent
renewable energy sources e.g. wind and solar energy, or when the dependence
between the renewable DG units (power generation) and the loads need to be
taken into account.
8.3 Remarks and Future Works
8.3.1 ‘Inertia’ Contribution
When a ‘verticaltohorizontal’ transformation in the power system takes place,
the inertia in the system decreases. DG units could provide/emulate ‘inertia’ to
the system in order to conﬁne the impact such as unwanted frequency deviations
for a largescale DG integration. For that purpose, some converterconnected
DG technologies could be equipped with controllers enabling DG to contribute
to the primary frequency control, for example by enabling the extraction of
kinetic energy stored in the turbine blades of variablespeed wind turbines or
by combining diﬀerent types/technologies of DG units.
126 8.3 Remarks and Future Works
8.3.2 Reactive Power Control
The possibilities for reactive power control within the system are also reduced
during the transformation. As DG is set to maximize the active power genera
tion, providing reactive power to the system may become a problem. However,
various solutions are available for providing reactive power to the system (e.g.
capacitor banks, synchronous condensers, FACTS devices, etc).
When the passive distribution systems transform into active ones, i.e. both
generation and consumption, the active distribution systems should take over
some functions of the shutdown CG units to provide inertia and/or reactive
power to the system.
Appendix A
List of Symbols and
Abbreviations
Latin symbols
Upper case
%DG
level
DG penetration level
A Area [m
2
]
B (line) susceptance [S, pu]
C
p
power coeﬃcient, the power coeﬃcient of the device (turbine)
E energy irradiance [Wm
−2
] (Chapter 2), or transient
internal voltages of machine [V, pu] (Chapter 5)
E(t) error signal, diﬀerence between desired and real output;
G conductance [S]
H eﬀective head of small hydropower plants [m],
or signiﬁcant wave height [m] (Chapter 2),
or per unit inertia constant [s] (Chapters 36)
I current [A, pu]
J moment of inertia [kgm
2
]
K constant term (in control blocks)
L inductance [H, pu]
M inertia constant [J/mech rad]
N number of units
P active power [W, pu], output power [W, pu]
P
r
Probability distribution (Chapter 6)
Q ﬂow rate [m
3
s
−1
] (Chapter 2), or reactive power [Var, pu]
R (line) resistance [Ω, pu]
128 List of Symbols and Abbreviations
Upper case, continued
S apparent power [VA]
T torque [Nm]
U voltage [V,pu]
U(t) controller output signal (Chapter 7)
X reactance [Ω, pu]
Y admittance [S, pu]
Z impedance [Ω, pu]
Lower case
a
i
constant corresponding to CG unit generation cost i
b
i
constant corresponding to CG unit generation cost i
c
i
constant corresponding to CG unit generation cost i
f frequency [Hz]
f
i
energy generation cost of CG unit i (in cost unit)
g gravitional constant [ms
−2
]
j complex identity (
√
−1)
n generator rotational speed [rpm]
p number of magnetic pole pairs of the ﬁeld circuit of the generator
(Chapter 3), or probability value (Chapter 6)
par(P
DG
i
) integer proportional to the ”size of DG
i
”,
the active power generated by the aggregate DG
at a particular load busi (Chapter 4)
s slip of induction machine
t time [s]
v velocity [ms
−1
]
Greek symbols
α azimuth angle [rad] (Chapter 2), or constant
corresponding to CG unit generation cost (Chapter 5)
β altitude angle [rad] (Chapter 2), or constant
corresponding to CG unit generation (Chapter 5)
δ angular rotor displacement [rad]
∆ simulation step [s], usually for time t
η eﬃciency
φ phase angle [rad]
ρ density of ﬂuid [kgm
−3
]
σ standard deviation
θ angle [rad]
ϕ current phase angle [rad]
ω angular velocity [rads
−1
]
List of Symbols and Abbreviations 129
List of Abbreviations
AC Alternating Current
ADS Active Distribution Systems
ASM squirrel cage induction generator
CCA Critical Clearing Angle
CCT Critical Clearing Time
CG Centralized Generator
CHP Combined Heat and Power
DC Direct Current
DG Distributed Generation
DN Distribution Network
EAC Equal Area Criterion
EHV ExtraHigh Voltage
FACTS Flexible AC Transmission System
G Generator (Chapter 7)
HV High Voltage
IEEE Institute of Electrical and Electronics Engineers
IGBT Insulated Gate Bipolar Transistor
LV Low Voltage
Max. dev. Maximum rotor speed deviation
MCS Monte Carlo Simulation
MV Medium Voltage
OPF Optimal Power Flow
Osc. dur. Oscillation duration
PE power electronic interfaced DG
without grid voltage and frequency control
PEC power electronic interfaced DG 
with grid voltage and frequency control
PI ProportionalIntegral controller
PLL Phase Locked Loop
PQsource constant active and reactive power generator
PSS/E Power System Simulator for Engineering
PV Photovoltaic
PWM Pulse Width Modulation
RMS RootMeanSquare
RTDS Real Time Digital Simulator
Sin Wave Gen SineWave Generator
SM synchronous generator 
without grid voltage and frequency control
SMC synchronous generator 
with grid voltage and frequency control
130 List of Symbols and Abbreviations
List of Abbreviations, continued
SMES Superconducting Magnetic Energy Storage
T
L
Transmission Line (Chapter 7)
T Transformer (Chapter 7)
TN Transmission Network
UEP Unstable Equilibrium Point
VSC Voltage Source Converter
Appendix B
Test System Data
10
8
1
2
3
5 4
7
6
9
37
25
30
2
1
39
9
8
7
5
4
3
18
17
27
26 28
29
38
24
16
21
35
22
23
36
19
33
20
34
32
31
10
11
13
12 6
14
15
Figure B.1: Single line diagram of the 39bus New England test system [49]
132 Test System Data
Table B.1: Bus Data of the New England 39 Bus Test System [49]
Bus Volts Load Load Gen Gen
(nr.) (pu) (MW) (MVAr) (MW) (MVAr)
1  0.0 0.0  
2  0.0 0.0  
3  322.0 2.4  
4  500.0 184.0  
5  0.0 0.0  
6  0.0 0.0  
7  233.8 84.0  
8  522.0 176.0  
9  0.0 0.0  
10  0.0 0.0  
11  0.0 0.0  
12  7.5 88.0  
13  0.0 0.0  
14  0.0 0.0  
15  320.0 153.0  
16  329.0 32.3  
17  0.0 0.0  
18  158.0 30.0  
19  0.0 0.0  
20  628.0 103.0  
21  274.0 115.0  
22  0.0 0.0  
23  247.5 84.6  
24  308.6 92.2  
25  224.0 47.2  
26  139.0 17.0  
27  281.0 75.5  
28  206.0 27.6  
29  283.5 26.9  
30 1.0475 0.0 0.0 250 
31 0.982 9.2 4.6  
32 0.9831 0.0 0.0 650 
33 0.9972 0.0 0.0 632 
34 1.0123 0.0 0.0 508 
35 1.0493 0.0 0.0 650 
36 1.0635 0.0 0.0 560 
37 1.0278 0.0 0.0 540 
38 1.0265 0.0 0.0 830 
39 1.03 1104.0 250.0 1000 
Test System Data 133
Table B.2: Line Data of the New England 39 Bus Test System [49]
Line Data Transformer Tap
Bus Bus Resistance Reactance Susceptance Magnitude Angle
1 2 0.0035 0.0411 0.6987 0 0
1 39 0.0010 0.0250 0.7500 0 0
2 3 0.0013 0.0151 0.2572 0 0
2 25 0.0070 0.0086 0.1460 0 0
3 4 0.0013 0.0213 0.2214 0 0
3 18 0.0011 0.0133 0.2138 0 0
4 5 0.0008 0.0128 0.1342 0 0
4 14 0.0008 0.0129 0.1382 0 0
5 6 0.0002 0.0026 0.0434 0 0
5 8 0.0008 0.0112 0.1476 0 0
6 7 0.0006 0.0092 0.1130 0 0
6 11 0.0007 0.0082 0.1389 0 0
7 8 0.0004 0.0046 0.0780 0 0
8 9 0.0023 0.0363 0.3804 0 0
9 39 0.0010 0.0250 1.2000 0 0
10 11 0.0004 0.0043 0.0729 0 0
10 13 0.0004 0.0043 0.0729 0 0
13 14 0.0009 0.0101 0.1723 0 0
14 15 0.0018 0.0217 0.3660 0 0
15 16 0.0009 0.0094 0.1710 0 0
16 17 0.0007 0.0089 0.1342 0 0
16 19 0.0016 0.0195 0.3040 0 0
16 21 0.0008 0.0135 0.2548 0 0
16 24 0.0003 0.0059 0.0680 0 0
17 18 0.0007 0.0082 0.1319 0 0
17 27 0.0013 0.0173 0.3216 0 0
21 22 0.0008 0.0140 0.2565 0 0
22 23 0.0006 0.0096 0.1846 0 0
23 24 0.0022 0.0350 0.3610 0 0
25 26 0.0032 0.0323 0.5130 0 0
26 27 0.0014 0.0147 0.2396 0 0
26 28 0.0043 0.0474 0.7802 0 0
26 29 0.0057 0.0625 1.0290 0 0
28 29 0.0014 0.0151 0.2490 0 0
12 11 0.0016 0.0435 0.0000 1.006 0
12 13 0.0016 0.0435 0.0000 1.006 0
6 31 0.0000 0.0250 0.0000 1.07 0
10 32 0.0000 0.0200 0.0000 1.07 0
19 33 0.0007 0.0142 0.0000 1.07 0
20 34 0.0009 0.0180 0.0000 1.009 0
22 35 0.0000 0.0143 0.0000 1.025 0
23 36 0.0005 0.0272 0.0000 1 0
25 37 0.0006 0.0232 0.0000 1.025 0
2 30 0.0000 0.0181 0.0000 1.025 0
29 38 0.0008 0.0156 0.0000 1.025 0
19 20 0.0007 0.0138 0.0000 1.06 0
134 Test System Data
Appendix C
Generator, Governor and
Excitation Systems Data
Table C.1: Detailed model unit data for a cylindrical synchronous generator (as
CG or DG) [53]
Parameter T
do
T
do
T
qo
T
qo
H D X
d
Value 5.0 0.05 1.0 0.04 4 0 1.75
Parameter X
q
X
d
X
q
X
d
X
l
S(1.0) S(1.2)
Value 1.65 0.30 0.75 0.20 0.175 0.2 0.4
Table C.2: Detailed model unit data for induction genrator (as DG) [53]
Parameter T T H X X X X
l
Value 0.98 0 3.5 3.1 0.18 0 0.1
Parameter E
1
S(E
1
) E
2
S(E
2
) 0. SY N −POW∗
Value 1 0 1.2 0 0 0
*SYNPOW, Mechanical Power At Synchronous Speed (> 0). Used only to start machine,
otherwise ignored.
136 Generator, Governor and Excitation Systems Data
Ψ″
d
/ Ψ″/

Ψ″
q
L
q
L
l
– ( )
/ Ψ″/ ( L
d
L
l
) –

/ Ψ″/
∆L
a d
i
f d
Σ Σ Σ
Σ
L″
d
L
l
–
L′
d
L
l
–

L′
d
L″
d
–
L′
d
L
l
–

L
d
′ L
l
–
1
T
do
″s

L′
d
L″
d
–
L′
d
L
l
– ( )
2

1
T
do
′ s

Σ
L
d
L′
d
–
+
–
+
–
–
+
+
+
+
+
+
Ψ″
d
i
d
daxis
Σ
L″
q
L
l
–
L′
q
L
l
–

L′
q
L″
q
–
L′
q
L
l
–

L′
q
L
l
–
1
T″
qo
s

L′
q
L″
q
–
L′
q
L
l
– ( )
2

1
T′
qo
s

Σ
L
q
L′
q
–
+
–
+
+
+
–
+
–
+
Ψ – ″
q
i
q
qaxis
E
f d
L
ad
i
f d
+
+
Ψ″
Σ Σ
Figure C.1: Synchronous generator model block diagram [53]
Table C.3: Detailed model unit governing system data for synchronous ma
chine [53]
Parameter T
A
/T
B
T
B
K T
E
E
MIN
E
MAX
Value 0.1 10 300 0.05 0 5
Generator, Governor and Excitation Systems Data 137
X
X
E
R
″
E″

E
I
″
E″

X X′ –
X X′ –
X′  X″
X′  X
l
( )
2

X′  X″
X′  X
l
( )
2

E
I
″
E
R
″
E
KR
X
T″pθ
E
KI
X
T″pθ
X″ X
l
X′  X
l

X′  X″
X′  X
l

E
KI
X″ X
l
X′  X
l

X′  X″
X′  X
l

E
KR
1
T″S

X′  X
l
X′  X
l
1
T′ S

1
T′ S

I
R
I
I
+ 
+

+
+
+
+
+
+



+
+ +
+
+
+
+
+





E″
E′
R
1
T″S

Figure C.2: Induction machine model block diagram [53]
138 Generator, Governor and Excitation Systems Data
VAR(L)
Reference
1
R

Σ
D
t
1
1 + T
1
s 1 + T
3
s
V
MAX
V
MIN
1 + T
2
s
∆ω
SPEED
+
Σ

+
PMECH
Figure C.3: Governor model block diagram [53]
K
1 + T
E
s
V
REF

+
E
MIN
E
MAX
E
C
V
S
EFD
p.u.
+
1 + T
A
s
1 + T
B
s
Σ
Figure C.4: Governor model block diagram [53]
Table C.4: Detailed model unit excitation system data for synchronous ma
chine [53]
Parameter R T
1
V
MAX
V
MIN
T
2
T
3
D
t
Value 0.05 0.05 0.91 0 2.1 7 0
Appendix D
Power Flow Computation
D.1 Power Flow Problem
One of the most common studies in power systems is a load ﬂow (or power ﬂow)
calculation. This computation provides insight in the state of a power system
for a speciﬁc steadystate situation, as it computes, among others, the voltage
(magnitude and phase angle) at each bus and the power ﬂow (real and reactive)
in each line [25].
In mathematical terms, solving a load ﬂow problem is nothing more than
solving a system of nonlinear algebraic equations. Solving the load ﬂow problem
starts from obtaining the singleline diagram data of power systems [25], where
the transmission lines – represented by the equivalent circuits and the numerical
values for the series impedance Z and the total linecharging admittance Y –
are used to determine all elements of the N × N bus admittance matrix of a
system (with N buses), with the typical element Y
ij
of this matrix given by
Y
ij
= Y
ij
∠θ
ij
,
= Y
ij
 cos θ
ij
+jY
ij
 sin θ
ij
, (D.1)
= Gij +
ˆ
jB
ij
.
Y
ij
is the admittance element between bus i and j, a complex identity repre
sented in the polar coordinate by the magnitude Y
ij
 and the angle θ
ij
and in
the cartesian coordinate by G
ij
and B
ij
, where G
ij
, B
ij
are the conductance
and susceptance of the element Y
ij
. The voltage U at a typical bus i of the
system is
U
i
= U
i
∠δ
i
,
= U
i
 cos(δ
i
) +j sin(δ
i
), (D.2)
140 Power Flow Computation
while the current injected into the network at bus i in terms of the elements Y
in
of Y
bus
is
I
i
= Y
i1
U
1
+Y
i2
U
2
+· · · +Y
iN
U
N
,
=
N
¸
n=1
Y
in
U
n
. (D.3)
According to this representation, the load ﬂow equations can be written as
P
i
= U
i

2
G
ii
+
N
¸
n=1,n=i
U
i
U
n
Y
in
 cos(θ
in
+δ
n
−δ
i
), (D.4)
Q
i
= −U
i

2
B
ii
+
N
¸
n=1,n=i
U
i
U
n
Y
in
 sin(θ
in
+δ
n
−δ
i
), (D.5)
where P
i
, Q
i
are respectively the active and reactive power injections at node
i. In the above equations, G
ii
, B
ii
are the conductance and susceptance of the
element Y
in
of the admittance matrix, respectively, and U
i
 and δ
i
the voltage
magnitude and angle at node i, respectively.
Using (D.4) and (D.5), the net real P
i
and reactive Q
i
power entering the
network at typical bus i can be computed. Denoting P
gi
as the scheduled power
generated at bus i and P
di
as the scheduled power demand of the load at bus i,
the net scheduled power being injected in the network at bus i, P
i,sch
, can be
deﬁned as
P
i,sch
= P
gi
−P
di
. (D.6)
Now, let P
i,calc
be the calculated value of P
i
, and
∆P
i
= P
i,sch
−P
i,calc
,
= (P
gi
−P
di
) −P
i,calc
, (D.7)
∆Q
i
= Q
i,sch
−Q
i,calc
,
= (Q
gi
−Q
di
) −Q
i,calc
. (D.8)
In reality, the calculated values do not always coincide with the scheduled ones.
In this case, ∆P
i
= 0 or ∆Q
i
= 0, and it is said that a mismatch occurs.
Suppose that mismatch does not occur. In this case ∆P
i
= ∆Q
i
= 0, and the
powerbalance equations (g
′
i
and g
′′
i
) can be written as
g
′
i
= P
i
−P
i,sch
,
= P
i
−(P
gi
−P
di
), (D.9)
= 0,
g
′′
i
= Q
i
−Q
i,sch
,
= Q
i
−(Q
gi
−Q
di
), (D.10)
= 0.
Power Flow Computation 141
Notice that the subscript i indicates that each bus of a power network has the
above two equations. To this end, we can formulate the power ﬂow problem as
follows:
Find U
i
, δ
i
, P
i
and Q
i
, i = 1, 2, . . . from (D.4) and (D.5) such that (D.10) and
(D.11) are satisﬁed.
Each bus i in a power system may be associated with four potentially un
known quantities P
i
, Q
i
, δ
i
and U
i
. Thus, the above stated problem is overde
termined and has an inﬁnite number of solutions. In order to make the solu
tions uniquely determined, two of the four unknowns have to be speciﬁed. The
remaining are computed. There exist three scenarios to choose the speciﬁed
variables, related to three types of buses:
1. P
i
and Q
i
speciﬁed, Load bus.
2. P
i
and U
i
 speciﬁed, Voltagecontrolled bus.
3. δ
i
and U
i
 speciﬁed, Slack bus.
In the power ﬂow study, the unscheduled busvoltage magnitudes and angles
are called state variables or dependent variables since their values, describing
the state of the system, depend on the quantities speciﬁed at all buses. The
powerﬂow problem is to determine values for all state variables by solving an
equal number of powerﬂow equations based on the input data speciﬁcations. As
the functions P
i
and Q
i
are nonlinear with respect to δ
i
and U
i
, a powerﬂow
calculation should employ iterative techniques.
D.2 NewtonRhapson power ﬂow solution
One method to solve a nonlinear system of equations is the NewtonRhapson
method. This method is widely explained in standard numerical analysis books,
and a basic tool within the PSS/E software [53]. Adaptation of this method to
the powerﬂow problem is described in the following algorithm 5, [25]:
Remark D.2.1 In Line 7, a linear system of type Ax = b has to be solved for
x. This in general requires inversion of A. In our case, J := A and so on. As
J is a densely populated matrix, this inversion may become very time consuming
for a large number of buses. In Line 8,  · 
∞
= max
i
{ ·
i
}, the inﬁnity norm.
So, the NewtonRhapson process is terminated if the maximum absolute value
of the corrections is lower than ǫ.
Remark D.2.2 In line 4, note that ∆P
1
and ∆Q
1
of the slack bus are undeﬁned
when P
1
and Q
1
are not scheduled. Since the slack bus serves as reference for
the angles of all other bus voltages, all terms involving δ
1
and ∆U
1
 are omitted
from the equations because those corrections are both zero at the slack bus.
142 Power Flow Computation
Algorithm 5 NewtonRhapson for powerﬂow problem
1: Guess the initial unknown voltages: U
i

(0)
, δ
(0)
i
2: Set tolerance ǫ
3: for j = 1, 2, ... do
4: Compute P
(j)
i,calc
, Q
(j)
i,calc
(from (D.4) and (D.5))
5: Compute the power mismatches (D.7), (D.8)
6: Compute the Jacobian J
(j)
J
(j)
=
∂P
2
∂δ
2
· · ·
∂P
2
∂δ
N
U
2

∂P
2
∂U
2

· · · U
2

∂P
2
∂U
N

.
.
. J
11
.
.
.
.
.
. J
12
.
.
.
∂P
N
∂δ
2
· · ·
∂P
N
∂δ
N
U
2

∂P
N
∂U
2

· · · U
N

∂P
2
∂U
N

∂Q
2
∂δ
2
· · ·
∂Q
2
∂δ
N
U
2

∂Q
2
∂U
2

· · · U
2

∂Q
2
∂U
N

.
.
. J
21
.
.
.
.
.
. J
22
.
.
.
∂Q
N
∂δ
2
· · ·
∂Q
N
∂δ
N
U
2

∂Q
N
∂U
2

· · · U
N

∂Q
2
∂U
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
(j)
(D.11)
7: Compute the correction from the following system:
J
(j)
∆δ
2
.
.
.
∆δ
N
∆U
2

.
.
.
∆U
N

¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
(j)
=
∆P
2
.
.
.
∆P
N
∆Q
2
.
.
.
∆Q
N
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
(j)
(D.12)
8: if ∆δ
(j)
i
∞
, ∆U
i

(j)
i
∞
< ǫ convergence is reached. quit
9: Add the correction to the previous value:
δ
(j+1)
i
= δ
(j)
i
+ ∆δ
(j)
i
(D.13)
U
i

(j+1)
= U
i

(j)
+ ∆U
(j)
i
 (D.14)
Bibliography
[1] T. Ackermann, G. Andersson, and L. Soder, Distributed generation:
a deﬁnition, Electric Power Systems Research, 57 (2001), pp. 195–204.
[2] A. AlHinai and A. Feliachi, Dynamic model of a microturbine used as
distributed generator, in 34
th
Southeastern Symposium on System Theory,
1819 March 2002, pp. 209–213.
[3] D. Audring and P. Balzer, Operating stationary fuel cells on power
systems and microgrids, in 2003 IEEE Bologna PowerTech Conference,
Bologna, Italy, 2326 June 2003, 6 pages.
[4] H. Barten and R. J. F. van Gerwen, Flexibele elektriciteitsopslag,
Energietechniek, 79(12) (2001), pp. 601–605.
[5] M. Begovic, A. Pregeli, A. Rohatgi, and C. Honsberg, Green
power: Status and perspectives, Proceedings of the IEEE, 89(12) (2001),
pp. 1734–1743.
[6] L. B. Bernshtein, Tidal power development  a realistic, justiﬁable and
topical problem of today, IEEE Transactions on Energy Conversion, 10 (3)
(1995), pp. 591–599.
[7] A. M. Borbely and J. F. Kreider, Distributed Generation the Power
Paradigm for the New Millennium, CRC Press LLC, 2001.
[8] R. E. Brown, Electric Power Distribution Reliability, Marcel Dekker Inc.,
New York, 2002.
[9] A. G. Bryans, B. Fox, P. A. Crossley, and M. O’Malley, Impact of
tidal generation on power system operation in ireland, IEEE Transactions
on Power Systems, 20(4) (2005), pp. 2034–2040.
[10] A. Burke, Ultracapacitors: Why, how and where is the technology, Journal
of Power Sources, 91 (1) (2000), pp. 37–50.
[11] Cigre Technical Brochure TF 38.01.10, Modeling new forms of gen
eration and storage, November 2000.
144 Bibliography
[12] Cigre Working Group 37.23, Impact of increasing contribution of dis
persed generation on the power system, Cigre Technical Brochure no. 137,
February 1999.
[13] B. Cook, Introduction to fuel cells and hydrogen technology, IEE Engi
neering Science and Education Journal, 11(6) (2002), pp. 205–216.
[14] A. Cooke, personal communication on august 1, OFGEM, (2003).
[15] M. W. Davis, Microturbines  an economic and reliability evaluation for
commercial, residential, and remote load applications, IEEE Transactions
on Power Systems, 14(4) (1999), pp. 1556–1562.
[16] D. N. Deokar, N. S. Lingayat, S. A. Khaparde, and S. P.
Sukhatme, Modelling tidal power plant at saphale, in 1998 IEEE Region
10 International Conference on Global Connectivity in Energy, TENCON
’98, vol. 2, 1719 December 1008, pp. 544–547.
[17] R. Dettmer, Wave energy gets seaworthy, IEE Review, 48(5) (2002),
pp. 14–19.
[18] M. K. Donnelly, J. E. Dagle, D. J. Trudnowski, and G. J.
Rogers, Impacts of the distributed utility on transmission system stability,
IEEE Transactions on Power Systems, 11(2) (1996), pp. 741–746.
[19] J. H. R. Enslin, Interconnection of distributed power to the distribution
network, in IEEE Young Researcher Symposium: Intelligent Energy Con
versions, Delft, the Netherlands, 1819 March 2004.
[20] E.On Netz GmbH, Bayreuth, Germany, E.On Netz, Grid Codem,
August 2003.
[21] L. M. Faulkenberry and W. Coffer, Electrical Power Distribution
and Transmission, PrenticeHall, New Jersey, 1996.
[22] J. P. Frau, Tidel energy: Promising projects la rance, a successful
industrialscale experiment, IEEE Transactions on Energy Conversion, 8(3)
(1993), pp. 552–558.
[23] W. Gao, Performance comparison of a fuel cellbattery hybrid powertrain
and a fuel cell ultracapacitor hybrid powertrain, in 2004 IEEE Power Elec
tronics in Transportation, Detroit, Michigan, USA, 2122 October 2004,
pp. 143–150.
[24] F. M. Gatta, F. iliceto, S. Lauria, and P. Masato, Modelling and
computer simulation of dispersed generation in distribution networks. mea
sures to prevent disconnection during system disturbances, in 2003 IEEE
Bologna PowerTech Conference, Bologna, Italy, 2326 June 2003, 10 pages.
[25] J. J. Grainger and W. D. Stevenson, Power System Analysis,
McGrawHill Inc., 1968.
Bibliography 145
[26] G. Grassi and A. V. Bridgwater, European research in biomass energy,
in 25th Intersociety Energy Conversion Engineering Conference IECEC90,
1217 August 1990, pp. 126–130.
[27] S. W. Hadley and W. Short, Electricity sector analysis in the clean
energy futures study, Energy Policy, 29(14) (2001), pp. 1285–1298.
[28] A. D. Hansen, F. Iov, P. Sorensen, and F. Blaabjerg, Overall con
trol strategy of variable speed doublyfed induction generator wind turbine,
in Nordic Wind Power Conference, Gotheborg, Sweden, 12 March 2004, 7
pages.
[29] R. Hebner, J. Beno, and A. Walls, Flywheel batteries come around
again, IEEE Spectrum, 39(4) (2002), pp. 46–51.
[30] J. W. Herzog, Current and nearterm emission control strategies for
diesel powered generator sets, in 24
th
Annual International Telecommu
nication Energy Conference INTELEC, 2002, pp. 394–399.
[31] N. Janssens, Impact of power ﬂows on interarea oscillations and mitiga
tion by means of svcs or qpss, in 14
th
Power Systems Computation Con
ference, 2428 June 2002, 7 pages.
[32] N. Jenkins, Impact of dispersed generation on power systems, ELECTRA,
199 (1999), pp. 6–13.
[33] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac,
Embedded Generation, no. 31 in Power Energy Series, IEE, 2000.
[34] P. Kundur, Power System Stability and Control, McGrawHill Inc., New
York, ﬁrst ed., 1994.
[35] P. Kundur, J. Paserba, and S. Vitet, Overview on deﬁnition and
classiﬁcation of power system stability, in CIGRE/IEEE PES International
Symposium Quality and Security of Electric Power Delivery Systems, 2003,
pp. 1–4.
[36] W. R. Lachs and D. Sutanto, Battery storage plant within large load
centres, IEEE Transaction on Power Systems, 7 (1992), pp. 762–767.
[37] G. Lalor, A. Mullane, and M. O’Malley, Frequency control and
wind turbine technologies, IEEE Transaction on Power Systems, 20 (2005),
pp. 1905–1913.
[38] G. Lalor, J. Ritchie, S. Rourke, D. Flynn, and M. J. O’Malley,
Dynamic frequency control with increasing wind generation, in 2004 IEEE
Power Engineering Society General Meeting, Denver, Colorado, USA,
vol. 2, 610 June 2004, pp. 1715–1720.
[39] M. A. Laughton, Fuel cells, IEE Engineering Science and Education
Journal, 11 (2002), pp. 7–16.
146 Bibliography
[40] C. A. Luongo, Superconducting storage systems: An overview, IEEE
Transactions on Magnetics, 32 (1996), pp. 2214–2223.
[41] A. P. S. Meliopoulos, Challenges in simulation and design of µ grids,
in 2002 IEEE Power Engineering Society Winter Meeting, vol. 1, 2002,
pp. 309–314.
[42] J. Morren, S. W. H. de Haan, and J. A. Ferreira, Model reduction
and control of electronic interfaces of voltage dip proof dg units, in 2004
IEEE Power Engineering Society General Meeting, vol. 2, 610 June 2004,
pp. 2168–2173.
[43] A. Mullane, G. Bryans, and M. O’Malley, Kinetic energy and fre
quency response comparison for renewable generation systems, in Interna
tional Conference on Future Power Systems 2005, Amsterdam, the Nether
lands, 1618 November 2005, 6 pages.
[44] A. Mullane and M. O’Malley, The inertial response of induction
machinebased wind turbines, IEEE Transactions on Power Systems, 20
(2005), pp. 1496–1503.
[45] K. Nielsen, Wave energy plants  the way forward, in IEE Colloqium
Wave Power: An Engineering and Commercial Perspective (Digest No:
1997/098), 13 March 1997, pp. 7/1–7/5.
[46] K. Ogata, Modern control engineering, PrenticeHall, 4
th
ed., 2002.
[47] I. T. F. on Load Representation for Dynamics Performance,
Load representation for dynamic analysis, IEEE Transactions on Power
Systems, 8(2) (1993), pp. 472–483.
[48] J. O’Sullivan, M. Power, M. Flynn, and M. O’Malley, Modelling
of frequency control in an island system, in 1999 IEEE Power Engineering
Society Winter Meeting, New York, New York, USA, vol. 1, 31 January4
February 1999, pp. 574–579.
[49] M. A. Pai, Energy Function Analysis for Power System Stability, Kluwer
Academic Publishers, Boston, 1989.
[50] G. Papaefthymiou, A. Tsanakas, M. Reza, P. H. Schavemaker,
and L. van der Sluis, Stochastic modelling and analysis of horizontally
operated power systems with a high wind energy penetration, in 2005 IEEE
St. Petersburg PowerTech Conference, St. Petersburg, Russia, 2730 June
2005, 7 pages.
[51] M. R. Patel, Wind and Solar Power Systems, CRC Press LLC, 1999.
[52] G. Pepermans, J. Driesen, D. Haeseldonckx, R. Belmans, and
W. D’Haeseleer, Distributed generation: deﬁnition, beneﬁts and issues,
Energy Policy, 33 (2005), pp. 787–798.
Bibliography 147
[53] Power Technologies, Inc., Schenectady, US, PSS/E 25.4 Online
Documentation, December 1997.
[54] A. Price, Technologies for energy storagepresent and future: Flow bat
teries, in 2000 IEEE Power Engineering Society Summer Meeting, vol. 3,
1620 July 2000, pp. 1541–1545.
[55] F. Provoost, A. Ishchenko, A. Jokic, J. M. A. Myrzik, and W. L.
Kling, Self controlling autonomous operating power networks, in 18
th
In
ternational Conference on Electricity Distribution, CIRED 2005, Turin,
Italy, 69 June 2005, 5 pages.
[56] M. Rabinowitz, Power systems of the future (part 4), IEEE Power Engi
neering Review, 20 (2000), pp. 4–6.
[57] S. Rahman, Green power: What is it and where can we ﬁnd it?, IEEE
Power and Energy Magazine, 1 (2003), pp. 30–37.
[58] M. Reza, J. Morren, P. H. Schavemaker, W. L. Kling, and
L. van der Sluis, Power electronic interfaced DG units: Impact of control
strategy on power system transient stability, in 2005 IEE 3
rd
Reliability of
Transmission and Distribution Networks, London, United Kingdom, 1517
February 2005, 4 pages.
[59] M. Reza, P. H. Schavemaker, W. L. Kling, and L. van der Sluis,
A research program on intelligent power systems: Self controlling and self
adapting power systems equipped to deal with the structural changes in the
generation and the way of consumption, in 17
th
International Conference on
Electricity Distribution, CIRED 2003, Barcelona, Spain, 1215 May 2003,
6 pages.
[60] M. Reza, P. H. Schavemaker, J. G. Slootweg, W. L. Kling, and
L. van der Sluis, Impacts of distributed generation penetration levels on
power systems transient stability, in IEEE Young Researcher Symposium:
Intelligent Energy Conversions, Delft, the Netherlands, 1819 March 2004,
6 pages.
[61] M. Reza, A. M. van Voorden, P. H. Schavemaker, G. C. Paap,
and L. van der Sluis, Implementation of renewable electrical energy gen
eration in an urban distribution network: Impacts of energy storage and de
mand growth, in 17
th
International Conference on Electricity Distribution,
CIRED 2003, Barcelona, Spain, 1215 May 2003, 6 pages.
[62] P. Robinson, Microhydro in Papua New Guineathe experiences of an
electrical engineer, IEE Power Engineering Journal, (1988), pp. 273–280.
[63] G. Rogers, Demystifying power system oscillations, IEEE Computer Ap
plications in Power, 9 (1996), pp. 30–35.
148 Bibliography
[64] R. B. Schainker, Executive overview: Energy storage options for a sus
tainable energy future, in 2004 IEEE Power Engineering Society General
Meeting, vol. 2, 610 June 2004, pp. 2309–2314.
[65] S. M. Schoenung and C. Burns, Utility energy storage applications
studies, IEEE Transactions on Energy Conversion, 11 (1996), pp. 658–665.
[66] F. Schwartz, R. Pegallapati, and M. Shahidehpour, Small hydro
as green power, in 2005 IEEE Power Engineering Society General Meeting,
1216 June 2005, pp. 1883–1890.
[67] T. Sels, C. Dragu, T. V. Craenenbroeck, and R. Belmans,
Overview of new energy storage systems for an improved power quality and
load managing on distribution level, in 16
th
International Conference and
Exhibition on Electricity Distribution (IEE Conf. Publ No. 482), CIRED
2001, Amsterdam, the Netherlands, vol. 4, 1821 June 2001, 5 pages.
[68] S. Sheth and M. Shahidehpour, Geothermal energy in power systems,
in 2004 IEEE Power Engineering Society General Meeting, Denver, Col
orado, USA, 610 June 2004, pp. 1972–1977.
[69] J. G. Slootweg, Wind Power: Modelling and Impact on Power Systems
Dynamics, PhD Thesis, Delft University of Technology, 2003.
[70] J. G. Slootweg, S. W. H. de Haan, H. Polinder, and W. L. Kling,
Modeling new generation and storage technologies in power system dynam
ics simulations, in 2002 IEEE Power Engineering Society Summer Meeting,
Chicago, Illinois, USA, vol. 2, 2125 July 2002, pp. 868–873.
[71] J. G. Slootweg and W. L. Kling, Impacts of distributed generation on
power system transient stability, in 2002 IEEE Power Engineering Society
Summer Meeting, Chicago, Illinois, USA, vol. 2, 2125 July 2002, pp. 503–
508.
[72] N. Sriram and M. Shahidehpour, Renewable biomass energy, in 2005
IEEE Power Engineering Society General Meeting, 1216 June 2005,
pp. 1910–1915.
[73] Tennet, Arnhem, the Netherlands, Technische gegevens van compo
nenten van het door TenneT beheerde 380 kV en 220 kVnet situatie per
1 november 2003, November 2003.
[74] A. M. Tuckey, D. J. Patterson, and J. Swenson, A kinetic energy
tidal generator in the Northern Territory Results, in 23
rd
International
Conference on Industrial Electronics, Control and Instrumentation IECON
97, vol. 2, 914 November 1997, pp. 937–942.
[75] T. V. Van and R. Belmans, Distributed generation overview: cur
rent status and challenges, International Review of Electrical Engineering
(IREE), 1 (2006), pp. 178–189.
Bibliography 149
[76] T. V. Van, J. Driesen, and R. Belmans, Interconnection of distributed
generators and their inﬂuences on power system, International Energy
Journal, 6 (2005), pp. 127–140.
[77] G. A. M. van Kuik, Are wind turbines growing too fast?, in 2001 European
Wind Energy Conference and Exhibition, 26 July 2001, pp. 69–72.
[78] B. M. Weedy, Electric Power Systems, McGrawHill Inc., ﬁrst ed., 1994.
[79] T. J. T. Whittaker, The development of shoreline wavepower in the
uk, in International Conference on Renewable Energy  Clean Power 2001,
1719 November 1993, pp. 117–120.
[80] K. H. Williamson, R. P. Gunderson, G. M. Hamblin, D. L.
Gallup, and K. Kitz, Geothermal power technology, Proceedings of the
IEEE, 89 (2001), pp. 1783–1792.
[81] A. Woyte, T. V. Van, R. Belmans, and J. Nijs, Voltage ﬂuctuations
on distribution level introduced by photovoltaic systems, IEEE Transactions
on energy conversion, 21 (2006), pp. 202–209.
[82] Y. Xue, T. van Cutsem, and M. RibbensPavella, A simple direct
method for fast transient stability assessment of large power systems, IEEE
Transactions on Power Systems, 3 (1988), pp. 400–412.
150 Bibliography
Scientiﬁc Contributions
Publications
1. M. Reza, A. O. Dominguez, P. H. Schavemaker, W. L. Kling. Maintaining
the power balance in an ’Empty Network’. European Transactions on
Electrical Power, Vol. 16, No. 5, SeptemberOctober 2006, pp. 479493.
2. M. Reza, D. Sudarmadi, F. A. Viawan, W. L. Kling, L. van der Sluis. Dy
namic Stability of Power Systems with Power Electronic Interfaced DG. In
Proceedings of IEEE Power Systems Conference and Exposition, Atlanta,
Georgia, USA, 29 October1 November 2006, 6 pages.
3. M. Reza, G. Papaefthymiou, W. L. Kling. Investigating Transient Stability
Impacts of a ’VerticaltoHorizontal’ Transformation of Power Systems.In
Proceedings of IEEE Young Researcher Symposium, Gent, Belgium, 2728
April 2006, 6 pages.
4. M. Reza, A. O. Dominguez, P. H. Schavemaker, W. L. Kling. Maintain
ing the power balance in an ’Empty Network’. In Proceedings of Interna
tional Conference on Future Power Systems (FPS) 2005, Amsterdam, The
Netherlands, 1618 November 2005, 8 pages.
5. M. Reza, C. P. Rodriguez, P. H. Schavemaker, W. L. Kling. A Transient
Stability Studies of a ’VerticaltoHorizontal’ Transformation of Power
Systems on a Real Time Digital Simulator. In Proceedings of Interna
tional Conference on Future Power Systems (FPS) 2005, Amsterdam, The
Netherlands, 1618 November 2005, 6 pages.
6. M. Reza, M. Gibescu, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Transient Stability Impacts of a ’VerticaltoHorizontal’ Transformation
of Power Systems. In Proceedings of IEEE PowerTech Conference, St
Petersburg, Russia, 2327 June 2005, 6 pages.
7. M. Reza, G. Papaefthymiou, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Stochastic Transient Stability Analysis of Power Systems with Distributed
Energy Systems. In Proceedings of CIGRE Symposium, Athens, Greece,
1516 April 2005, 8 pages.
152 Scientiﬁc Contributions
8. M. Reza, J. Morren, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Power Electronic Interfaced DG Units: Impact of Control Strategy on
Power System Transient Stability. In Proceedings of 3
rd
Reliability of
Transmission and Distribution Networks (RTDN) 2005, London, United
Kingdom, 1517 February 2005, 4 pages.
9. M. Reza, G. Papaefthymiou, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Investigating Network Constraints (Congestions) of HorizontallyOperated
Power Systems with Stochastic Distributed Generation. In Proceedings of
3
rd
Reliability of Transmission and Distribution Networks (RTDN) 2005,
London, United Kingdom, 1517 February 2005, 4 pages.
10. M. Reza, P. H. Schavemaker, J. G. Slootweg, W. L. Kling, L. van der Sluis.
Impact of Distributed Generation Grid Connection Strength on Power
System Transient Stability. In Proceedings of the 4
th
IASTED Interna
tional Conference on Power and Energy Systems (EuroPES) 2004, Rhodes,
Greece, 2830 June 2004, 6 pages.
11. M. Reza, J. G. Slootweg, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Impacts of Distributed Generation Penetration Levels on Power Systems
Transient Stability. In Proceedings of IEEE Power Engineering Society,
General Meeting, Denver, Colorado, USA, 611 June 2004, 6 pages.
12. M. Reza, J. Morren, P. H. Schavemaker, J. G. Slootweg, W. L. Kling,
L. van der Sluis. Impacts of Converter Connected Distributed Generation
on Power System Transient Stability. In Proceedings of IEEE Young Re
searcher Symposium: Intelligent Energy Conversions, Delft, The Nether
lands, 1819 March 2004, 6 pages.
13. M. Reza, G. Papaefthymiou, P. H. Schavemaker, L. van der Sluis. Poten
tial for Transmission Lines Losses Reduction in Electrical Power System
Operation with Distributed Generation. In Proceedings of 8
th
Indonesian
Student Scientiﬁc Meeting 2003 Conference, Delft, The Netherlands, 910
October 2003, 4 pages.
14. M. Reza, J. G. Slootweg, P. H. Schavemaker, W. L. Kling, L. van der Sluis.
Investigating Impacts of Distributed Generation on Transmission System
Stability. In Proceedings of IEEE PowerTech Conference, Bologna, Italy,
2326 June 2003, 7 pages.
15. M. Reza, P. H. Schavemaker, W. L. Kling, L. van der Sluis. A Research
Program on Intelligent Power Systems: Self Controlling and Self Adapt
ing Power Systems Equipped to Deal with the Structural Changes in the
Generation and the Way of Consumption. In Proceedings of CIRED 2003
Conference, Barcelona, Spain, 1215 May 2003, 6 pages.
16. M. Reza, A. M. van Voorden, P. H. Schavemaker, G. C. Paap, L. van der Sluis.
Implementation of Renewable Electrical Energy Generation in an Urban
Distribution Network: Impacts of Energy Storage and Demand Growth.
Scientiﬁc Contributions 153
In Proceedings of CIRED 2003 Conference, Barcelona, Spain, 1215 May
2003, 6 pages.
17. M. Reza, W. L. Kling. Solving Network Constraints (Congestions) by
Investigating Impact of Phase Shifters on the Flows. In Proceedings of
the 2
nd
IASTED International Conference Power and Energy Systems
(EuroPES) 2002, Crete, Grece, 2528 June 2002, pp. 189193.
18. D. Sudarmadi, M. Reza, G. C. Paap, L. van der Sluis. DC Interconnection
between Java and Sumatera, in Indonesia. In Proceedings of IEEE Power
Systems Conference and Exposition, Atlanta, Georgia, USA, 29 October1
November 2006, 6 pages.
19. A. Mostavan, E. Joelianto, M. Reza, L. van der Sluis. Discrete Event
Controller Design of a Solar PV Water Pumping System using Signal
Interpreted Petri Net. In Proceedings of the 6
th
Asian Control Conference,
Bali, Indonesia, 1821 July 2006, 6 pages.
20. E. Joelianto, A. Mostavan, M. Reza, L. van der Sluis. A Discrete Event
Controller for PVWind Hybrid Energy System using Signal Interpreted
Petri Net. In Proceedings of the 6
th
Asian Control Conference, Bali,
Indonesia, 1821 July 2006, 8 pages.
21. F. A. Viawan, M. Reza, The Impact of Distributed Generation on Voltage
Dip and Overcurrent Protection Coordination. In Proceedings of Interna
tional Conference on Future Power Systems (FPS) 2005, Amsterdam, The
Netherlands, 1618 November 2005, 6 pages.
22. G. Papaefthymiou, A. Tsanakas, M. Reza, P. H. Schavemaker, L. van der Sluis.
Stochastic Modelling and Analysis of Horizontally Operated Power Sys
tems with High Wind Energy Penetration. In Proceedings of IEEE Pow
erTech Conference, St Petersburg, Russia, 2327 June 2005, 7 pages.
23. G. Papaefthymiou, A. Tsanakas, M. Reza, P. H. Schavemaker, L. van der Sluis.
Reliability Assessment of HV/MV Transformer Links for Distributed Power
Systems Planning. In Proceedings of 3
rd
Reliability of Transmission and
Distribution Networks (RTDN) 2005, London, United Kingdom, 1517
February 2005, 6 pages.
Talks
1. Impacts of Distributed Generation and Distribution Systems on Power
System Transient Stability, ABB Corporate Research, V¨aster˚as, Sweden,
May 18, 2006.
2. ‘Empty Network’. Panel Discussion on “Reliable Power  Europes Strength”,
Brussels, Belgium, May 10, 2006 (poster).
154 Scientiﬁc Contributions
3. Inherently Stable Transmission Systems on a VerticaltoHorizontal Trans
formation of Power Systems with Distributed Generations, Lunch Lecture
KEMA, Arnhem, The Netherlands, February 16, 2006.
4. Developing Indonesian Electricity with Distributed Generation. Interna
tional Conference on Indonesia Toward 2020, Hamburg, Germany, March
45, 2005.
5. An Example of Utilizing a Stochastic Approach to Elaborate the Investiga
tion of the Impacts of Distributed Generation on Power Systems. The 3
rd
Indonesian Applied Mathematics Society in The Netherlands (IAMSN)
Seminar on Applied Mathematics, Delft, The Netherlands, October 21,
2004.
6. Potential of RenewableEnergyBased Decentralized Generation in Power
Network for Carbon Dioxide Emission Reduction. The 9
th
Indonesian
Student Scientiﬁc Meeting, Aachen, Germany, October 79, 2004.
7. Potential for Distributed Generation in Indonesian Electrical Power Sys
tems. The 8
th
Indonesian Student Scientiﬁc Meeting, Delft, The Nether
lands, October 910, 2003.
Acknowledgment
This thesis describes my work that was performed during my appointment at
the Electrical Power System (EPS) Research Group, Faculty of Electrical En
gineering, Mathematics and Computer Science, Delft University of Technology
(TU Delft), the Netherlands.
Upon completion, I am indebted to a number of people for their direct or
indirect contributions.
I would like to specially thank prof.ir. W. L. Kling (Wil) and prof.ir. L.
van der Sluis (Lou) for being the promotor and the copromotor of my PhD
research, and especially for giving me a great opportunity to do research in
the EPS Research Group. My special thanks also go to my daily supervisor
dr.ir. P. H. Schavemaker (Pieter) and dr.ir. M. Gibescu (Madeleine) for fruitful
discussions and encouragement during the research.
Many encouraging discussions were done during the appointment with highly
experienced people in the ﬁeld. In this respect, my appreciations should go to
prof.dr.ir. J. A. Ferreira (TU Delft’s Electrical Power Processing), prof.dr.ir.
J. H. Blom (Technische Universiteit Eindhoven), prof.ir. M. Antal (Technis
che Universiteit Eindhoven, emeritus), prof.dr.ir R. Belmans (Katholieke Uni
versiteit Leuven, Belgium) and prof.dr. M. J. O’Malley (University College
Dublin, Ireland) of the promotion committee for their time to carefully read the
manuscript.
This research has been performed within the framework of the research pro
gram ’intelligent power systems’ of the IOPEMVT (Innovatiegerichte Onder
zoeksprogramma’s  ElektroMagnetische VermogensTechniek) Program, ﬁnan
cially supported by SenterNovem. SenterNovem is an agency of the Dutch min
istry of Economic Aﬀairs. The ’Intelligent Power Systems’ research program is
conducted by the Electrical Power Systems and the Electrical Power Processing
Research Groups of the Delft University of Technology (TU Delft) and the Elec
trical Power Systems and Control System Research Groups of the Eindhoven
University of Technology (TU Eindhoven), the Netherlands. For this, I would
like to thank again prof.ir. M. Antal, the chairman of the IOPEMVT program
committee, ir. G. W. Boltje (SenterNovem), the program coordinator, and the
leaders and supervisors of the project, prof. Kling, prof. van der Sluis, and
dr. Schavemaker mentioned earlier and furthermore ir. S. W. H. de Haan (TU
Delft’s Electrical Power Processing), dr.ir. J. M. A. Myrzik (TU Eindhoven’s
Electrical Power Systems), prof.dr.ir P. van den Bosch (TU Eindoven’s Control
156 Acknowledgment
Systems), and dr.ir. A. A. H. Damen (TU Eindhoven’s Control Systems), for
the scientiﬁc discussions and the technical feedback from the ‘real world’. It was
so often that these discussions paved the way to better understand many aspects
that were and are still beyond my knowledge. My thanks are also to George
Papaefthymiou, Johan Morren, Frans Provoost, Anton Ishchenko, Andrej Jo
kic, Roald de Graaﬀ, Jody Verboomen, Cai Rong, and Sjef Cobben, the Ph.D.
students, with whom I discussed scientiﬁc subjects and technical experiences we
faced in the IOPEMVT project.
Special thanks are also once more to George Papaefthymiou, my oﬃcemate,
with whom I shared interests in many subjects from the very technical to the
philosophical and spiritual ones, and Bob Paap, Marjan Popov, Tirza Drizi,
Boukje Ypma, Ezra van Lanen, Arjan van Voorden, Johan Vijftigschild, Jan
Heydeman, Barbara Slagter, Bart Ummels, Ralph Hendriks, Didik Sudarmadi,
and I Made Ro Sakya, the EPSmates, all in all, with whom I experienced
a highly stimulating and pleasant working atmosphere during my stay at the
EPS group. The initial part of this thesis is based to some extent on dr.ir. J.
G. Slootweg’s thesis, who was kind enough to share his knowledge and experi
ence with me. The students Alejandro Dominguez and Cristovo Rodriguez are
acknowledged for their contributions to this research.
Many thanks also go to those who support me with their hospitality and
friendship during my stay in the Netherlands. I found it not easy to mention
all their names, since the names would make a couple of pages! Nevertheless,
I would like to acknowledge Yogi Erlangga, Dedy Wicaksono, Uly Nasution,
Diah Chaerani and Anita Pharmatrisanti with whom I shared experiences as
Indonesian Ph.D. students in Delft, and discussed our hopes and dreams in the
future, and the younger Averrouz Mostavan, Zulﬁkar Dharmawan, Datuk Ary
Samsura, Boy Fadhilah, Ikshan Rashad and Yusuf Maury who sincerely helped
me with practical things from repairing bikes to reinstalling crashed computers,
and also becoming nice friends for chatting during our interaction in Delft. I
would like also to acknowledge Sheikh Sharief and the Suﬁ brothers and sisters
in Rotterdam, with whom I shared colorful and enjoyable experiences on life
and spirituality.
I wish to deeply thank my parents, my sister Mia Miranti and my brother
Muhammad Lukman for their sincere support and prayers so that I could make
my way up to where I am now. I am also grateful to my grand parents and my
parentsinlaw for their constant support and encouragement.
Special thanks should go to my wife Novi Ineke Cempaka Wangi for her
endless support and encouragement throughout the diﬃculties I was facing in
this work, as well as to my son Muhammad Rifqi Aulia Yahya, who can erase
my problems just by smiling and laughing.
Most of all, I ultimately praise and thank Allah the Almighty, the Creator
who creates all these nice people and the pleasant opportunities.
Muhamad Reza
Delft, Summer 2006
Biography
Muhamad Reza was born on November 4, 1974, in Bandung, the capital of
the province of West Java, Indonesia. He ﬁnished the secondary school educa
tion at SMA Negeri 3 Bandung, Indonesia in 1993. In 1997, he obtained a B.Sc.
degree in Electrical Engineering from Bandung Institute of Technology (ITB)
with honor (cum laude). From 1998 until 2000 he attended Delft University
of Technology (TU Delft), the Netherlands, from where he received an M.Sc.
degree in Electrical Engineering with honor (cum laude). Since February 2002,
he joined Electrical Power System Research Groups, chaired by Professor Lou
van der Sluis, at the Faculty of Electrical Engineering, Mathematics and Com
puter Science as a PhD student within the framework of the research program
’Intelligent Power Systems’ supported ﬁnancially by SenterNovem, an agency of
the Dutch Ministry of Economic Aﬀairs.
From 19941997 he was awarded fellowship by ToyotaAstra Foundation,
Indonesia for his B.Sc. study. In 1997 he was awarded Ganesha Prize as the
best ITB student Year 1997 (highest in the university), and in 1998 he received a
grant from Bandoengsche Technische Hogeschool Fonds (BTHF) where he used
it for working on a research activity at the Electrical Power System Research
Group (EPS), TU Delft, for three months. From 19982000 he was awarded a
fellowship within the Highly Talented Indonesian Student (Thalis) Program for
his M.Sc. study. In 2000 he was awarded the Best Grade Average of the Master
of Science International Program 19982000, at TU Delft.
His work experience include student internships in Schlumberger Wireline,
Manila, the Philippines (1996), and in PT Tesla Daya Elektrika, Bandung, In
donesia (1997). In 19971998 and 20002002 he was with the Electrical Power
System and Distribution Laboratory, Department of Electrical Engineering,
ITB, where he involved in some projects with PLN, Indonesia (2001), YPF
Maxus, Indonesia (2001), and Pertamina, Indonesia (2001). He is currently
working as a Research & Development scientist in power systems at ABB Cor
porate Research, V¨aster˚as, Sweden.
Muhamad Reza is married to Novi Ineke Cempaka Wangi and has a son,
Muhammad Rifqi Aulia Yahya.
Dit proefschrift is goedgekeurd door de promotoren: Prof.ir. W. L. Kling Prof.ir. L. van der Sluis Samenstelling promotiecommissie: Rector Magniﬁcus, Prof.ir. W. L. Kling, Prof.ir. L. van der Sluis, Prof.dr. J. A. Ferreira Prof.dr.ir. J. H. Blom Prof.ir. M. Antal Prof.dr.ir R. Belmans Prof.dr. M. J. O’Malley voorzitter Technische Universiteit Delft, promotor Technische Universiteit Delft, promotor Technische Universiteit Delft Technische Universiteit Eindhoven Technische Universiteit Eindhoven (emeritus) e Katholieke Universiteit Leuven, Belgi¨ University College Dublin, Ierland
This research has been performed within the framework of the research program ’Intelligent Power Systems’ that is supported ﬁnancially by SenterNovem, an agency of the Dutch Ministry of Economic Aﬀairs.
Stability analysis of transmission systems with high penetration of distributed generation. Dissertation at Delft University of Technology. Copyright c 2006 by M. Reza. ISBN: 9162870394 Cover: A miniature of transmission lines tower in Madurodam, The Hague, photographed and modiﬁed by Muhamad Reza, and used for this thesis with permission from Madurodam B.V.
To Bapak, Ibu, Novi and Rifqi.
.
It is wellknown that the implementation of DG inﬂuences the technical aspects of the distribution grids. the phenomena of the rotor dynamics of synchronous machines. In this work.Summary Stability analysis of transmission systems with high penetration of distributed generation Nowadays. Many of the prime movers of such DG technologies are based on renewable energy sources resulting in an environmentallyfriendly power generation. This work deals with the impact of implementing DG on the transmission system transient stability. Indicators for assessing the stability performance of a power system derived from these concepts are the maximum . Therefore. the type of primary energy source determines the output power characteristics of DG and the type of grid connection applied. in Chapter 3. It also determines the utilization of power electronic interfaces. the impact of the DG implementation on this is investigated. the powerangle curve and the equal area criterion concepts. it is important to examine characteristics of DG that inﬂuence the dynamic stability behavior of a transmission system (Chapter 2). Based on the DG classiﬁcation. Such generation is known as ’distributed generation’ (DG). with the emphasis on a potential transition from a ’vertical power system’ to a ’horizontal power system’ (Chapter 1). its impact is no longer restricted to the distribution network but begins to inﬂuence the whole system. DG units are classiﬁed based on the primary (both conventional and renewable) energy sources. presented in Chapter 3. A problem in power systems is maintaining synchronous operation of all (centralized) synchronous machines. interest in generating electricity using decentralized generators of relatively small scale is increasing. The impact of a small amount of DG connected to the grid on the power system transient stability has not been treated so often. basic models of DG technology to be used in the transient stability simulation of a large power system can be derived. are explained by means of the swing equation. For this purpose. that determine the rotor angle stability of a power system. Therefore. When the penetration level of DG increases. The stability problem associated is called rotor angle stability. To a large extent.
It is found that DG inﬂuences the system transient stability diﬀerently depending on the factors above. Furthermore. and DG protection schemes of converterconnected DG are simulated and discussed. However. The DG units are im . since large power ﬂows may have a detrimental eﬀect on the damping of the oscillations: the heavier the lines are loaded. with the emphasis on the representation of power electronic interfaced (converter connected) DG units. diﬀerent from the preceding Chapter 4. is presented in Chapter 4. implementing DG is a natural way of ‘limiting’ the power ﬂows on the transmission lines. The impact of increasing DG penetration levels. power system transient instabilities with very high DG levels (more than 50%) are found. where DG penetration reaches a level that covers the total load of the original power system (100% DG level). especially with long lines. The study is focused on the impact of the stochastic behavior of DG. and the larger the oscillations of the centralized generators may be. DG gridconnectionstrength. Beside the deterministic approach. Therefore. while details are listed in Appendices B and C. the widely known 39bus New England dynamic test system is used with minor adjustments. In Chapter 5. the investigation is focused on the impact of DG levels on the system transient stability when the increasing DG level is followed by a reduction of centralized generators in service resulting in a ’vertical to horizontal’ transformation of the power system. The results discussed in Chapters 4 and 5 are merely based on a deterministic approach. Chapter 7 investigates the situation when the power system is pushed towards a scenario. For simulation purposes. the increasing DG level implies a reduction in rotating masses (inertia) and reactive power control ability in the system. Several solutions are proposed for these problems by means of rescheduling centralized generators and optimizing the power ﬂow. Finally. the weaker the coupling between generators and loads becomes. a stochastic analysis can also be used to study the transient stability of the power systems. diﬀerent DG technologies. This is logical when all centralized generators remain in the system – as well as their active and their reactive power control and the inertia of their rotating masses – along with the increasing DG levels. The representations of diﬀerent DG technologies within the software packages are discussed. as presented in Chapter 6.ii Summary rotor speed deviation and oscillation duration of the (centralized) synchronous machines. The investigation of the impact of a high DG (penetration) level on power system transient stability. The emphasis is on the use of converter connected DG units to supply active power. The results show that including the stochastic behavior of DG leads to a more complete and detailed view of the system performance. there is no signiﬁcant stability problem observed up to about 30% DG penetration level regardless the technology. In some cases. The basic setup of the test system is described. In some cases. Several software packages suitable for the transient dynamic simulation of DG in a large system are highlighted. a minimum number of centralized generators has to remain in the system to avoid system instabilities. where the parameters of the test system are set to the typical values. It improves the transient stability of a transmission system.
the output power of the DG units (and the ADS) are decoupled from the grid frequency. Therefore. suﬃcient inertia and voltage support must be installed. The simulation results show that by applying such control systems. Furthermore. due to the powerelectronic interfaces. The power system is still connected to a source that provides a constant 50 Hz voltage that is meant to give a (system) frequency reference for the generators but generates no power at steady state. To a certain extent regional balancing of power can be performed by local voltage control. any power imbalance in the system must be compensated by generators in the ADS. However.Summary iii plemented via powerelectronic converters within so called “active distribution systems” (ADS) connected to the transmission system also via powerelectronic interfaces. For this purpose. one should be aware of the fact that the system behaves stochastically. especially with DG. Therefore in this chapter. the voltage is used to detect and maintain the power balance. speciﬁc control concepts are developed. the power balance in the power system can be maintained by the ADS. . The research performed in this work indicates that from the transmission system stability point of view. if higher DG penetration levels are coming up.
iv Summary .
In dit proefschrift worden de eﬀecten van de toepassing van DG daarop onderzocht. Het stabiliteitsprobleem dat daarmee samenhangt heet de rotorhoek stabiliteit. wat in e Hoofdstuk 3 behandeld wordt. op de transi¨nte stabiliteit van het transportnet is nog niet vaak e onderzocht. Het soort primaire energiebron bepaalt in grote mate de kenmerken van de energieoutput van DG en het type netaansluiting dat wordt toegepast. Veel aandrijfsystemen van DG technologie¨n zijn gebaseerd op hernieuwbare energiebronnen wat resulteert in e een milieuvriendelijke energieopwekking. Wanneer het penetratieniveau van DG stijgt. Ook bepaalt het of vermogenelektronische interfaces worden gebruikt. Een probleem in elektriciteitsvoorzieningsystemen is het handhaven van synchrone werking van alle (centrale) synchrone machines. is het eﬀect niet meer beperkt tot het distributienet. Dit proefschrift onderzoekt de eﬀecten van de toepassing van DG op de transi¨nte stabiliteit van het transmissienet. Daartoe . Deze manier van elektriciteitsopwekking wordt ook wel ‘distributed generation’ (DG) genoemd. die het dynamische stabiliteitsgedrag van een transmissiesysteem be¨ ınvloeden (Hoofdstuk 2). met de nadruk op een potenti¨le e e overgang van een ’verticaal gericht systeem’ naar een ’horizontaal gericht systeem’ (Hoofdstuk 1) Met dit doel is het belangrijk om de kenmerken van DG te onderzoeken. Daarom worden DG eenheden geclassiﬁceerd op basis van de primaire energiebronnen (zowel conventionele als hernieuwbare). De invloed van kleine hoeveelheden DG aangesloten op het net. maar wordt het gehele systeem be¨ ınvloed. Het is bekend dat de toepassing van DG de technische aspecten van de distributienetten be¨ ınvloedt. Gebaseerd op de DG classiﬁcatie kunnen basismodellen afgeleid worden voor de DG technologie¨n om te e gebruiken in de transi¨nte stabiliteitssimulatie van een transmissienet.Samenvatting in het Nederlands Stabiliteitsanalyse van een transmissienet met een hoge penetratiegraad van decentrale opwekking Tegenwoordig neemt de interesse in het produceren van elektriciteit met relatief kleine decentrale eenheden toe.
ligt de nadruk van het onderzoek op de invloed van het toenemende DG niveau op de transi¨nte stabiliteit van het net als dit samen gaat met e de vermindering van in bedrijf zijnde centrale generatoren. In Hoofdstuk 5. de fenomenen van de rotordynamica van een synchronische machine verklaard. e De invloed van de toename van het DG niveau. vooral bij lange lijnen. Echter. Er wordt aangetoond dat DG de transi¨nte stabiliteit van het transmissienet verschile lend be¨ ınvloedt afhankelijk van boven vermelde factoren. De representatie van de verscheidene DG technologie¨n in de softwarepakketten wordt besproken. anders dan Hoofdstuk 4. er is geen signiﬁcant stabiliteitsprobleem gevonden tot de DG een 30% niveau bereikt. zijn gesimuleerd en bediscussieerd. de sterkte van de DG netkoppeling.vi Samenvatting worden in Hoofdstuk 3. leidend tot een ‘vertikale naar horizontale’ transformatie van het elektriciteitsvoorzieningsysteem. verschillende DG technologie¨n. onafhankelijk van de DG technologie. Verscheidene softwarepakketten. Het onderzoek van de invloed van een hoog (penetratie) niveau van DG op de transi¨nte stabiliteit van het transmissienet wordt in Hoofdstuk 4 uiteengezet. die de rotorhoek stabiliteit van een transmissienet bepalen. Deze resultaten zijn logisch vanwege de nog steeds aanwezige centrale generatoren in het net. waarbij e de nadruk ligt op de weergave van DG eenheden met vermogenelektronische interfaces (converters). Verschillende oplossingen voor dit probleem worden voorgesteld middels verandering van de inzet van de centrale generatoren en optimalisering van de vermogenstransporten. De indicatoren voor het beoordelen van het stabiliteitsgedrag van een transmissienet afgeleid uit deze concepten zijn de maximumafwijking van de rotorsnelheid en de duur van de slingering van de synchrone machines. De basisopzet van het testsysteem wordt in dit hoofdstuk beschreven. Daarnaast is toepassing van DG een natuurlijke manier om de vermogenstransporten in de transmissielijnen te beperken. omdat grote vermogenstranse porten een nadelige invloed hebben op de demping van de rotorslingeringen: hoe zwaarder de lijnen zijn belast des te zwakker de koppeling tussen de generatoren en de belastingen wordt en des te groter de slingeringen van de centrale generatoren kunnen worden. In sommige gevallen wordt instabiliteit van het transmissienet bij een hoog DG niveau (meer dan 50%) gevonden. bij dit niveau van DG. met de bijbehorende regeling van actief vermogen en blindvermogen en de inertie van de roterende massa. geschikt voor transi¨nte dynamische simulatie van DG e in een groot transmissienet. De nadruk ligt op het gebruik van de vermogenelektronisch gekoppelde DG als actief vermogen leverancier. terwijl de details worden gegeven in Bijlagen B en C. Dit verbetert de transi¨nte stabiliteit van het transmissienet. de vermogen versus hoek curve en het gelijke oppervlakte criterium concept. Voor simulatiedoeleinden wordt het bekende 39knooppunten New England dynamisch testsysteem gebruikt. middels de bewegingsvergelijking. en DG beveiligingsschema’s van via vere mogenselektronica gekoppelde DG. In sommige gevallen moet een minimaal aantal centrale generatoren in bedrijf gehouden worden om instabiliteit van het net te . impliceert het toenemende DG niveau een vermindering van roterende massa (inertie) en regelmogelijkheden van blindvermorgen in het net. Daardoor. met een aantal aanpassingen daarin. worden behandeld.
De resultaten tonen aan dat het meenemen van het stochastische gedrag van DG leidt tot een vollediger en gedetailleerder overzicht van het functioneren van het systeem. De resultaten van de Hoofdstukken 4 en 5 zijn gebaseerd op een deterministische benadering. DG eenheden zijn via vermogenselektronische interfaces opgenomen in zogenaamde “actieve distributie systemen” (ADS) en aangenomen is dat deze ook met vermogenselektronische interfaces zijn gekoppeld met het transmissienet. Hoofdstuk 7 tot slot behandelt de situatie waarbij het transmissienet aan een extreem scenario onderwerpen wordt. waarbij de parameters van het testsysteem ingesteld zijn op hun typische waarden.Samenvatting vii vermijden. Behalve een deterministische benadering kan ook een stochastische analyse worden gebruikt om de transi¨nte stabiliteit te bestuderen. Daarom moet iedere onbalans in vermogen door opwekking in de ADS gecompenseerd worden. zeker met DG. De nadruk van de studie ligt op de invloed van het stochastische gedrag van DG. maar wekt geen vermogen op in de stationaire toestand. Voor dit doel zijn speciﬁeke regelconcepten ontwikkeld. Echter. waar DG de totale belasting van het net dekt (100% DG niveau). . Het onderzoek beschreven in dit proefschrift toont aan dat vanuit oogpunt van stabiliteit van het transmissienet voldoende inertie en spanningsondersteuning aanwezig moet zijn als hogere DG penetratieniveaus aan de orde zijn. In zekere mate kan regionale balanshandhaving worden uitgevoerd met lokale spanningsregeling. De simulatieresultaten tonen aan dat door toepassing van deze regeltechnieken. de handhaving van de vermogenbalans gerealiseerd kan worden via de ADS. Het systeem is verondersteld nog steeds verbonden te zijn met een bron met constante 50 Hz frequentie bedoeld als een frequentie referentie voor de generatoren. e zoals beschreven in Hoofdstuk 6. Verder moet men zich bewust zijn van het feit dat het systeem zich stochastisch gedraagt. Daarom wordt in dit hoofdstuk de spanning gebruikt om de vermogensbalans te detecteren en te handhaven. vanwege de vermogenselektronische interfaces is het uitgangsvermogen van de DG eenheden (en de ADS) ontkoppelt van de netfrequentie.
viii Samenvatting .
. . . . . . . . . . . .1 ’Vertical’ Power Systems . . . . . . . . . . . . . . . . . . . . . . . . . .1 Stateoftheart DG Technology . . . . . . . . .1. 2. . . . . . . . . . . 2. . . .3 ’Horizontal’ Power Systems . . 2. . . . . . . . . . . . . 2. . . . . . . . . . .2 Output Power Characteristics . . . .1. . . . .8 Geothermal Power Plants . . . .1 Conventional FossilFuel Based Generators 2. . . 1. . . . . . . .1. . . . . . . . . . . . . . . . . . 2. .2. .1 Batteries . . . . . . . . . . . . . . . . .4 Dynamics of Power Systems . . . . . . . . 1. .9 Biomass Power Plants . . . . . . . . . . . . . .3 Energy Storage Systems . . . . . . . . . . . . . . . . .1. . . . . . . . . . 1. . . . . . . . . .4 Small HydroPower Plants . . . . . . . . . . . . . . . . . . . . 1. . . . .1. . . . .1 Controllable DG . . . . . . .2 Distributed Generation Concept 1. . . . . . . . i v ix 1 1 3 4 4 7 8 9 11 11 12 12 12 13 13 13 13 14 14 14 15 15 16 16 19 19 19 . . . . 2. . 2. . . . . . .1. . . . . . . . . . . . . . . . . . . . . . . 2. .2 Hydrogen Fuel Cells . . . . . . . . . . . . . .6 Photovoltaics . . . . . . . 2. . .3 Combined Heat and Power (CHP) Plants . .1. . . . .3.11 Wave Power Plants . . . . . . . 2. . . . . . . . . . . . . . . . . . . . .1.10 Tidal Power Plants . . . . . . . . . . . .Contents Summary in English Samenvatting in het Nederlands Contents 1 Introduction 1. . . . . . . 2. . . . . .1. .2. 1. . . . . . . 2. . . . 2. . . . . . . . . . 2. . .7 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Objectives and Limitations . . . . . . .5 Research Framework . . . . .7 Fuel Cells . . . . 2 Distributed Generation 2. . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . . . . 2. . . . . . 2. . . . . . . . . . . . . . . . . . . . . .2 Microturbines . . . . . . . . . . . . . . . . . . . . . .2 Noncontrollable DG . .1. . . . .1. . . . . . . . . . . . . . . .5 Wind Turbines . . .
. . . . . . . . 2. . . . . . . . DG Prospects: ConverterConnected DG . . . .1. . 3. . . . . . . . . . . . .x Contents 2. . . . . . . . . . . . . . . . . . . . . . .3 Connecting Energy Storage to the Grid . 3. . . . . . . . . . . .3. .4. . . . . . .4. . . .1 DG Impacts . . . . . . . . .4 Behavior of Centralized Power Plants . . . . . . . . . . . 2.3 Redox Flow Batteries . . . . . . .2. . . . . . . .5. . 3. . . . . . . . . . . . . . . . . .8 CompressedAir Systems . . . . . . . . . . . . . . . . . . . . . . . . . 2. .2. . . . . . . . . 4 Impact of DG on Power System Transient Stability 4.2. . . . . . .2 PowerAngle Equation . . . . . . . . . . . . . . . . . .1 Swing Equation . . . . . . . . . . . .1 Distribution Network and DG Layout . 3. 3. . . . .4 Flywheel Systems . . . . . . . . . . . .1 Simulation Scenarios . 4. . . . 3. . . . . . .6 Summary . . . . . . . . . . .7 PumpedHydroelectric Plants . . .1 The IEEE 39bus New England Test System . . .1. . 4. . . . . . . . . . . .3 Incorporation of DG in Distribution Networks . . . . . . . . . . . .1. . . 4. . . .2 Rotor Dynamics of Synchronous Machines .6 3 Stability of Systems with DG 3. . . . . . . . . . 4. . . . . . .1 Maximum Rotor Speed Deviation . . . . . . . .2. . . . . . .6 Superconducting Magnetic Energy Storage (SMES) Systems 2. . . . . . 3. . . . . . . . . . . . . . 4. 2. . . . . . . . . . . . . . . 4. . . . . . . . . . . . . . . . . . . . . . . . .1. . . . .5 Ultracapacitors . . . . . 20 20 20 20 20 21 22 22 23 26 26 27 29 29 30 31 31 31 31 33 34 36 37 37 38 39 41 43 43 44 46 47 47 49 49 49 50 51 54 56 56 57 60 60 2. . . .3. . . . . . . .2 Simulation Scenarios . . . . .1 Rotor Angle Stability .4. . .1. .4 DG and Large System Dynamic Simulation . . . . .3. . .4 Remarks . . . . . .4 2. . . . . .2. 3. .4. . . . . . . . . . 3. . . .2 Power System Dynamics Software Packages . . . . . . . . . . . .4. . . . . . . . . .1 Classiﬁcation of Power System Stability . .1. . . . . . . . .3 Transient Stability Simulation . . . . . . . . . . . . . . . . . . .1 Modeling DG Technologies . . . . . . . . . .4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . .3.2 Oscillation Duration . . . .3 Equal Area Criterion .5.2 Transient Stability Simulation . .1 Direct GridConnected DG . . . . . . 3. . . . .2 Voltage Stability . . . 3. . . . . . . .3 System Stability Indicators . . . . . . . . . . . . . . 3. 3. 3. . . . . . . 3. . . .3 Frequency Stability . . .2. . . . . . . . . . .5 Simulation Setup . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . .2 DG GridConnection Strength Impacts . . . . . . . . 3. . . . . . . . Summary . . . . .2 Indirect GridConnected DG . . .5. . . . . . . 2. . . . . . . . . . . . . . . 3. 3. . . 3. . . . . . . . . . . . . . . . . . . . .3 Simulation Results . . .5 2. . . . .5. . . . . . . . 4. . . . . . . . . . . . . .3. . . . . 4. . .1. .2. . 2. . . . . . . . . . . . . . . . . . 2. . . . . . .3. . . . . . . . . . . . . . 4. . . . . . . . DG GridConnection Characteristics . . . . . .3. . . . .2 DG Technology . . . . . . . . . . . .
. . . . . . . . . . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 5 ’Vertical to Horizontal’ Transformation of Power Systems 5. .4. . . . . . . . . . . . . . . . .3. . . . .Contents 4. . . . . . . . . . . . 95 6. . 92 6. . . . . . . . . . . . . . . . . . . . . . . 5. . . . . .3 Simulation Results . . . . . . . . . . . . .3 Stochastic Transient Stability Study with Increasing DG . . . . . . . . . . . . 4. . . . . . . . . . . . . . . . . . . . . . .3. 4. . . .5. . . . . . . . . . . . 4. . . .3. . .2 Transient Stability Simulation . . . . . . . .4 Remarks . . . . . . . . . . . . . . . . . . .2 Transient Stability Simulation . . . . . . . . . . . 103 7 Maintaining Power Balance with Active Distribution Systems 105 7. . . . . 108 7. . . . . . 113 7. . . . . . . . . . . 99 6. . . . . . . . . . . . .4 Simulation Results Case II . . 107 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . .1 Background . . . . . . . . . . . . . . . .1 Simulation Scenario . . . .4. . . .3 Rescheduling Generation Case I . . . . . .5 DG with RideThrough Capability . . . Conclusions . . . 94 6.2 Decentralizedcontroller with single reference . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . . . . . . . . .7 Conclusion . . . . . . . .3 Simulation Results .3 Model of Power System with ADS . . . . . . .1 Stochastic Load Flow . . . .3 Simulation Results . . . 4. . . . . . . . . . . . .2. . . Protection of PowerElectronics Impacts . . 5. . .2 Simulation Results Case I . . . . 4. 100 6. . 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 4. . . . . .2 Stochastic Transient Stability Analysis . . . . . . . . .2.3. . .3. . . . . . . 107 7. . .1 Standalone master controller . . . . . . . . . . . . . . .1 Simulation Setup . . . . . . . . . . .2 Model of ADS . . . . . . . . . . 116 .5 ADS Control Systems . . . . . . . . . . . . . . . 5.3 Generator Models . . . . . . . . . . . . . 92 6. . . . . . . . . . . . . . . . . . . . . . . xi 62 63 63 64 64 69 69 69 71 71 72 72 75 76 78 83 84 88 89 89 4. . . . . . . . . . . . . . .5. . . . . . . . . . . . . . . . . .1 Simulation Scenario . . . . . . . . 114 7. . .3 4. . . .1 Assumptions . . . . . . . 5. . . . . . . . . . . . . . .2 MCS Samples .4. . . . DG Penetration Level and Technology Impacts 4. . . . . . . . . . . . . . . . .3. . . . . . . 5. . . . . . . . . . . . .3. 107 7. . . 106 7. 99 6. . . . . . . . . . . . . . . . . . . .4 Basic Controller Model . .4 Conclusions . .1 Simulation Scenario . . 110 7. . 92 6. .3. .2 Power Balance .1 Simulation Scenarios . . 4. . . . . . . . . 105 7. . . . . .4 Remarks . . . . . . . . .6 Remarks . . . . . . . . . . .2 Monte Carlo simulation (MCS) Samples . . . . . . . . . . . . . . . . . . . . . . .3 Simulation Results . . . . 99 6. . . . . . . . . . . . . . 6 Stochastic Approach to Transient Stability of Power Systems with DG 91 6. .2. .
.1 Power Flow Problem .3 Decentralized controller with hysteresis .xii Contents 7. . . . . . . . .2 Stochastic Stability Studies . . . 8. . . . .2 Reactive Power Control A List of Symbols and Abbreviations B Test System Data C Generator. . . . . . . . . . . .6 8 Conclusions 8. . . . . . .2 NewtonRhapson power ﬂow solution . . . . . . . . . . 123 123 125 125 125 126 127 131 135 7. . . .3 Remarks and Future Works . . . 118 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . .1 ‘Inertia’ Contribution . . . . . . . . . . 120 . 8. . . . . . . . . . . . . . . . . . . . . . 8. . . . . . .3. . . . . . . . Governor and Excitation Systems Data D Power Flow Computation 139 D. . . . . . . . . .1 Overview . . 8. . . . . . . . . . . . .5. . . . . . . . . . . . . . . . . . . . . . . . . . 141 Bibliography Scientiﬁc Contributions Acknowledgment Biography 143 151 155 157 . 139 D.
Some constraints. The distribution system distributes the electrical power to the consumers/loads.1). The generation system converts mechanical power that results from the conversion of primary energy sources. Some of these negative consequences will be focused on in this work. namely: generation. a classical power system consists of three technical stages. These power plants are built based on the demand estimate for a certain period of time. decentralized power generators. centralized generators have been utilized for the power generation.Chapter 1 Introduction A classical power system is characterized by a relatively small number of large. etc. limit the expansion and use of such large power plants and may induce a shift towards a more extensive use of small. . If hydropower is available it may be used as input too. coal. transmission. mainly large. and distribution. Several remedies to eliminate them or limit their impact will be suggested and discussed. gas. So. The technical consequences must be considered carefully in order to maintain the present reliability level of the power system. Until now.. This classical power system can be best illustrated by considering the diﬀerent voltage levels (Figure 1. 1. The transmission system transports the electrical power over a long distance to the load centers. hydro power. It is wellknown that the implementation of small. energy resources or supply routes and connected to the transmission system. however. socalled centralized power plants for meeting the electric energy demand. such as nuclear. Synchronous generators are typical electromechanical energy transducers for large power plants. often close to cooling water.1 ’Vertical’ Power Systems A power system is designed to supply electrical power to the consumers. into electrical. decentralized power generators brings both positive and negative consequences to the existing power system.
1 ’Vertical’ Power Systems *HQHUDWLRQ 3ODQW *HQHUDWLRQ 6XEVWDWLRQ +9(+9 1HWZRUN N9 WR N9.2 1.
³(OHFWULFDO SRZHU IORZ´ 7UDQVPLVVLRQ 6XEVWDWLRQ 7UDQVPLVVLRQ 6\VWHP 'LUHFWLRQ RI DFWLYH SRZHU IORZ 09 1HWZRUN WR N9.
³(OHFWULFDO SRZHU IORZ´ 'LVWULEXWLRQ 6XEVWDWLRQ 6XEWUDQVPLVVLRQ 6\VWHP ³(OHFWULFDO SRZHU IORZ´ /9 1HWZRUN 9.
&XVWRPHU 'LVWULEXWLRQ 6\VWHP Figure 1.1: VerticallyOperated Power System (modiﬁed from [8]) .
and geothermal energy are used. Therefore. development and increased use of new. economical. and where it enters the primary distribution systems. but are often because of their relatively low carbon emission classiﬁed as environmentallyfriendly types of power generation. sun. and the Combined Heat and Power (CHP) generation. An overview of these DG technologies is given in Chapter 2. These small power generators are usually located in the vicinity of the electrical loads.e. promotion. environmental and geographical) considerations and irrational constraints (e. the power is transformed to Low Voltage (LV) level and is distributed to the consumers [78]. wave. 3 1. Within this class are the microturbine generator supplied by natural gas.g. Some types of distributed generation (DG) are based on conventional fossil energy sources. an increasing amount of electrical power is generated by decentralized power generators of relatively small scale (i. which is practically a parallel conversion of fuel into electrical and thermal energy (more carbon emission will be produced if the electrical and thermal energy are generated separately). The expansion and the construction of large power plants are limited by both rational (e. clean and environmentally friendly forms of energy [5]. This way of electrical power generation is referred to as ’Distributed Generation’ (DG) because it is spread out over the system. at MV. [27].or LVnetworks) [12]. [32]. biomass. tidal. In contrast to the conventional power plants.g. the development and the implementation of DG units are encouraged mostly by environmental forces. Most of these renewable energy sources can be converted to electric power. social and political issues) [56]. in units in a range of hundreds kWs to some MWs. Renewable energy sources like wind.e. Finally. which in this work will be referred to as a ’vertical’ power system. where the power is transformed to Medium Voltage (MV) level (typically 10 to 30 kV). the power is transformed to a higher voltage level in the generation substation. in the distribution substations. and are mostly connected to distribution networks (i.2 Distributed Generation Concept Nowadays. renewable. Based on the diﬀerent voltage levels. this type of power system can then be viewed as a ’verticallyoperated’ power system. The rise of DG is supported by the advancements in supporting technologies like power electronic converters and controllers.Introduction After being generated at the power plant (typically at a voltage of 10 kV to 30 kV). close to the load centers. . Currently there are many DG technologies available. This has stimulated research. electrical energy generally ﬂows from the higher to the lower voltage levels in the network. by (relatively) small generators that are connected to the distribution networks. The High Voltage (HV) or Extra High Voltage (EHV) (110 kV to 400 kV) transmission systems transports the electrical power further to the (sub)transmission substations. smaller than 50100 MW).
3: graph (a) changes to graph (b)). loads are connected and disconnected frequently and demand for both active and reactive power changes continuously. In this way. from the higher to the lower voltage levels.4 1. . we could even imagine that on certain moments in time the electrical power generated by the DG within the distribution networks may become suﬃcient to fulﬁll the total demand of the system. In addition to the power injected in the EHV and HV system by the large power plants. Figure 1. a large amount of DG is implemented in the power system and all centralized generators remain but generate less (Figure 1.and upwards). as is the case in the passive network. In the active network the power ﬂow is no longer in one direction (downwards). but may be bidirectional (down. as well as ’horizontally’ from one MV or LV network to another or from a generator to a load within the same MV or LV network leading to a new term: the horizontal power system or horizontallyoperated power system. it is possible that power is transferred from one distribution network to another. Finally. Besides. the amount of DG in the system increases in such a way that a number of centralized generators (power plants) are shut down for eﬃciency reasons (Figure 1. the power system is subject to disturbances caused by malfunctioning or failing equipment. In the ’second’ transformation step. The ability of a power system to remain in a state of operating equilibrium under normal operating conditions and to regain an acceptable state of equilibrium after being subjected to a disturbance is deﬁned as the power system stability [34]. but they also generate and if generation surpasses their demand.3 ’Horizontal’ Power Systems Due to the possible largescale implementation of DG units in the classical ’vertical’ power system.3: graph (c) to graph (d)).3: graph (b) changes to graph (c)). The implementation of DG turns the passive distribution network into an active one.2 shows an example of both an active and a passive distribution network. supply the network. i. DG units supply the system via the MV or LV networks. a transition towards a more ’horizontal’ power system may take place. When we reﬂect further on this issue. It is hard to imagine how a system under graph d could operate but theoretically these are the steps. In the ’ﬁrst’ transformation step. Figure 1.3 ’Horizontal’ Power Systems 1.4 Dynamics of Power Systems The power system is a dynamic system. the remaining large (centralized) power plants may be shut down.e. In this active distribution network some costumers not only consume electricity.3 illustrates a ’VerticaltoHorizontal’ transformation of a power system. the power can ﬂow both ’vertically’. Even under normal operation conditions. In this particular case. all (remaining) centralized generators are out of service (Figure 1. in the ’third’ transformation step. when power generated by DG within the active distribution networks is suﬃcient to match the total demand. 1. Therefore.
2: Conventional (passive) distribution network (top . the network (b) Active Network Figure 1.a) and an active distribution networks with DG (bottom . and to.b) (modiﬁed from [14]) .g.Introduction 5 Transmission Network Industrial customers who consume electricity Distribution Network Domestic customers and small business who consume electricity Thin line indicates flow from the network (a) Passive Network Transmission Network Industrial customers with DG (colored in gray) also generate some electricity which flows back into the network Distribution Network Distributed generator e. wind turbine Domestic customers and small business with domestic DG (colored in gray) can also generate electricity which flows back into the network Thin line indicates flow from the network Thicker line indicates flow from.
6 1.4 Dynamics of Power Systems &HQWUDOL]HG JHQHUDWRUV a a a &HQWUDOL]HG JHQHUDWRUV a a a 7UDQVPLVVLRQ QHWZRUN 7UDQVPLVVLRQ QHWZRUN a a /RDGV D.
/RDGV DQG 'LVWULEXWHG *HQHUDWRUV E.
&HQWUDOL]HG JHQHUDWRU a 7UDQVPLVVLRQ QHWZRUN a 7UDQVPLVVLRQ QHWZRUN a a a a a a a a a /RDGV DQG 'LVWULEXWHG *HQHUDWRUV F.
/RDGV DQG 'LVWULEXWHG *HQHUDWRUV G.
Instability can also occur if a power system cannot maintain the voltage levels within a required range [34]. Then it is up to the dynamics of the system whether stability can be maintained. Figure 1. the focus is on the dynamics. a lot of them equipped with control systems to stabilize the operation of the overall system. and excitation controllers to regulate the voltage and thus the reactive power output.3: ’VerticaltoHorizontal’ transformation of the power system More than 100 years of experience have lead to the present power system with many subsystems and dynamic elements. When we consider the interconnected power system as a single system. In this work. The implementation of DG inﬂuences both steady state and dynamic performance of a power system. For instance. however. a necessary condition for a stable operation is that more or less all the large synchronous generators remain in synchronism. . power plants are equipped with prime mover controllers to regulate the speed and thus the active power output.
which becomes ’active’. The ’Intelligent Power Systems’ project is initiated by the Electrical Power Systems and Electrical Power Electronics Groups of the Delft University of Technology and the Electrical Power Systems and Control Systems Groups of the Eindhoven University of Technology. selfcontrolling autonomous networks. the network can be operated autonomously but as a matter of fact remains connected to the rest of the grid for security reasons. The research described in this thesis is within research part 1: inherently stable transmission systems. ﬁnancially supported by SenterNovem. inherently stable transmission system. The research focuses on the eﬀects of the structural changes in generation and demand taking place. whereas the remainder is possibly generated further away in neighboring systems. The project investigates how the power electronic interfaces of decentralized generators or between network parts can be used to support the grid.Introduction 7 1. of the project analyzes all aspects of power quality. focuses on the distribution network. an agency of the Dutch Ministry of Economical Aﬀairs.5 Research Framework The research presented in this work has been performed within the framework of the ’Intelligent Power Systems’ project. The project consists of four parts (illustrated in Figure 1.4). The interaction between the grid and the connected appliances has a large inﬂuence on the power quality. When the amount of power generated in a part of the distribution network is suﬃcient to supply a local demand. In total 10 Ph. like for instance the largescale introduction of distributed (renewable) generators [59]. In the third part (research part 3). Technologies and strategies have to be developed that can operate the distribution network in diﬀerent modes and support the operation and robustness of the network.D. investigates the inﬂuence of uncontrolled decentralized generation on stability and dynamic behavior of the transmission network. Setting up a power quality test lab is an integral part of the project. manageable distribution networks. The second part (research part 2). Solutions investigated include the control of centralized and decentralized power. . autonomous networks are considered. less centralized plants will be connected to the transmission network as more generation takes place in the distribution networks. As a consequence of the transition in the generation. students are involved and work closely together. Also the stability of the distribution network and the eﬀect of the stochastic behavior of decentralized generators on the voltage level are investigated. The goal is to provide elements for the discussion between polluter and grid operator who has to take measures to comply with the standards and grid codes. The ﬁrst part (research part 1). the application of power electronic interfaces and monitoring of the system stability. The project is part of the IOPEMVT program (Innovation Oriented research Program . The fourth part (research part 4).ElectroMagnetic Power Technology). The project investigates the control functions needed to operate the autonomous networks in an optimal and secure way. optimal power quality.
Based on the simulation results. Simulation scenarios of power systems with DG are deﬁned later. .8 1. power system simulation software package PSS/E is used. This research is unique as it combines the investigation of an increasing DG penetration level and a power system that transforms from a vertical into a horizontal one.e. Steady state and economic impacts are beyond the scope of this work. the following two objectives are set: • Investigate the impact of a high DG penetration level on the stability of a power system. The focus is only on the dynamic impacts. special emphasis is on the behavior of the centralized generators in service. the transient stability.6 Objectives and Limitations In this work. • Investigate the stability of a power system that undergoes a ’verticaltohorizontal’ transformation.6 Objectives and Limitations 1 Inherently stable transmission system 4 Optimal power quality 2 Manageable distribution networks 3 Selfcontrolling autonomous networks Figure 1. i. where models of the power (test) system and DG are included. For this purpose.4: Research items within the ’Intelligent Power Systems’ research project 1.
7 Outline of the Thesis The thesis is organized as follows: • In Chapter 2.Introduction 9 1. . are discussed. a stochastic approach to the transient stability of system with DG within the framework of “verticaltohorizontal” transformation is studied. • In Chapter 6. Diﬀerent scenarios of a power system with a high DG penetration level are developed. an overview of the current DG technologies is given. system stability indicators. the term ”inherently stable transmission system” is deﬁned. • An overview of power system stability is presented in Chapter 3. In this chapter. software. DG units are implemented within Active Distribution Systems. The impact of diﬀerent DG technologies. fault durations and locations. The transient stability impact of this transformation is analyzed and solutions for reducing the negative eﬀects are discussed. basic associated controls of system elements and parameters used. used and presented throughout the work. • In Chapter 4. DG gridconnectionoption. the scenarios for the ’verticaltohorizontal’ transformation of power systems are further elaborated. Furthermore the research approach and the simulation setup. • The conclusions and recommendations for future work are given in Chapter 8. • In Chapter 5. the impact of DG implementation on the power system transient stability is discussed. • In Chapter 7. The emphasis is on the classiﬁcation of the diﬀerent DG technologies according to their potential impact on the power system stability. including test system. a power system reaching 100% DG implementation is studied. DG penetration level. Control methods to maintain the power balance in such a power system with Active Distribution Systems are suggested. and protection schemes of power electronic interfaced DG units are investigated.
10 1.7 Outline of the Thesis .
Energy storage systems.6. 2. it is important to examine characteristics of DG that inﬂuence this behavior. Concluding remarks are made in Section 2. The type of primary energy source and the conversion process determine. A direct grid connection is made by using the common/classical synchronous and induction generators.5 elaborates on this issue. In Section 2. depending on both the type of primary energy source and prime mover: a direct and indirect grid connection. can be applied to smooth this intermittent eﬀect.2. DG can be classiﬁed as dispatchable or nondispatchable as is described in Section 2. A DG unit usually produces electric power well below 100 MW [32]. for example. can show high outputpower ﬂuctuations. and connected to the distribution network [12]. whereas an indirect grid connection is made by means of powerelectronic converters.1 Stateoftheart DG Technology Many deﬁnitions of distributed generation (DG) exist. as described in Section 2.4 the way DG is connected to the network (grid) is reviewed. Based on the output power characteristics. There are two options.Chapter 2 Distributed Generation The impact of distributed generation (DG) on the dynamic stability of power systems is studied. The output power of nondispatchable units. it does not implicate that the unit cannot be controlled locally . especially the ones driven by renewable energy sources. Other literature however advocates a boarder and ∗ not centrally dispatched: it cannot be controlled from a system control center. As mentioned in Chapter 1.3. CIGRE Working Group 37.23. not centrally dispatched∗ . [33]. to a large extent. the output power characteristics of DG and the type of grid connection applied. DG units can be based on various (both conventional and alternative) primary energy sources. Section 2. has deﬁned distributed generation (DG) as electrical generation that is not centrally planned. Therefore.
1.000 to 120. [12]. The use of diesel or gasoline gives high emission levels [30].1 Stateoftheart DG Technology more straightforward deﬁnition of DG: a DG source is an electric power generation source connected directly to the distribution network or on the customer side of the meter [1]. and good modularity and ﬂexibility.2 Microturbines A ’micro’ gas turbine (microturbine) produces electric power in the range of 25500 kW.1. CHP generation on a large scale is usually based on fossil fuel. Reciprocating engines are characterized by low capital cost. As DG.12 2. gas turbines. [52]. Reciprocating engines. 2. have drawbacks. Therefore. In general. highlighted in the next subsection. gas turbines are mostly encouraged by the development of microturbines. The use of renewable energy sources such as ethanol is also possible [15].3 Combined Heat and Power (CHP) Plants Combined Heat and Power (CHP). a powerelectronic converter is used to interface the generator and the grid. Gas turbines are commonly used in industry [12]. The emission can be reduced to some extent by using natural gas as energy source. 2. Reciprocating engines. The use of natural gas results in lower emission when compared to reciprocating engines. Most microturbines use natural gas. An electrical generator is integrated within the microturbine. The electric power is produced with a frequency (in the order) of thousands of Hz [2]. microturbines are typically characterized by low emission levels. and not the type of primary energy source used. However. however. also known as cogeneration. the highfrequency electrical power is converted to DC before it is inverted back to the lowfrequency AC of the grid. [12]. is the simultaneous production of electrical power and useful heat [33].1 Conventional FossilFuel Based Generators Within the category DG. and microturbines can be used in CHP schemes. The reciprocating engines and gas turbines are the most common in this category. CHP is heat driven .000 RPM). 2. possible thermal and electrical cogeneration. the term ’Conventional FossilFuel Based Generator’ is used to describe small fossilfueled power plants within a range of kWs up to 100 MW [7]. A large number of moving parts leads to high noise levels pollution and increases the maintenance cost [7]. the associated gas from the oilﬁeld is frequently used to generate electricity. Thus. many generator units that are driven by renewable sources of energy inherently possess the characteristics of DG. As a consequence. Furthermore they are reliable [7]. that operates at a high speed (50. Within the powerelectronic interface. the way that a generator is implemented in a power system determines its classiﬁcation as DG.1. [33]. In oil industry for example.
This water movement can be obtained. fuel (usually hydrogen) and an oxidant must be supplied to the anode and the cathode. where a power house is installed. high modularity and mobility. A small hydropower plant has less impact on the environment and ecosystem. and short design. from a runof river or a river with a small impoundment [66]. Currently.7 Fuel Cells DC power can also be generated by an electrochemical process. A PV cell consists of two or more semiconductor layers of speciﬁc physical properties. To generate electricity. when compared to a large hydropower plant. With this simultaneous process. for example. [33]. the overall eﬃciency of a CHP plant can be around 85% [12].1. Once built.1. 2.4 Small HydroPower Plants A hydropower plant generates electricity from the motion of a mass of water. 2. its maintenance cost is minimal [62]. and is easy to build within a short construction schedule [66]. the photons cause the electrons to move in one direction (crossing the junctions of the layers) and a direct current (DC) is generated.6 Photovoltaics Photovoltaic (PV) power generation systems convert sunlight directly into electricity [51]. Wind energy is one of the most promising energy sources to be used for renewable electricity generation [51]. A small hydropower plant produces electric power up to 10 MW. installation and startup time of a new plant [51]. These layers are arranged in such a way that when the PV cell is exposed to sunlight.Distributed Generation 13 and electricity is the byproduct. respectively. Hydro power plant technology has reached maturity. that generate electricity.5 Wind Turbines A wind turbine generates electricity by extracting kinetic energy from the wind passing through its blades. however. 2. the capital cost of PV modules has declined in the past decades. One fuel cell . the increasing interest for implementing wind turbines is mostly driven by the availability of wind energy for generating power at large scale of MWs or even GWs [77]. long life cycle and simple maintenance (since there are no moving parts). Electrochemical reactions create ion ﬂows. It consists of a positive electrode (anode) and a negative electrode (cathode). An example is the socalled fuel cell.1. Apart from using a small wind turbine as DG for generating emissionfree power. However. PV implementation is encouraged by the inﬁnite availability of sun energy. PV energy cost is still high. 2.1.
because there are no moving parts [39]. liquid. When the tides comes into the shore. 2. biomass is converted to thermal energy. This cycle will maintain the pressure of the reservoir and sustain the reservoir [57]. In the future. Practically. The heat from geothermal reservoirs provides the force that rotates the turbine generators and produces the electricity. can be disposed underground [68]. It produces no pollutant and no unwanted product. Biomass is considered a substitute for fossil fuels. animal wastes (manures). a huge dam. In general. forestry and agricultural residues. and the earth and the sun. The biomass products. The gaseous fuels can be applied in fuel cell systems. These characteristics make fuel cells suitable as DG in. called a ’barrage’ is built across a river estuary.and landbased vegetation and trees.1 Stateoftheart DG Technology only produces a small amount of electricity. The used geothermal water is then returned (injected back) into the reservoir to be reheated.1. Energy is extracted either directly by harnessing the kinetic energy of currents due the tides or by using a basin to capture potential energy from the diﬀerence in height of a rising and falling mass of water. the water ﬂows through tunnels in the dam.14 2. The ebb and ﬂow of the tides can be used to turn a turbine. if any.10 Tidal Power Plants Tidal energy is derived from the gravitational forces of attraction that operate between the earth and the moon. portable and produce low noise pollution. and larger amounts can be obtained from a stack of fuel cells [7]. In the latter technique. can be used as fuel to generate electricity. municipal biosolids (sewage). A geothermal power plant is relatively sustainable. When the tide goes in and out. This new structure may further increase the implementation of fuel cells as DG [19]. 2. electrical networks (both AC and DC) can be combined with a gas and hydrogen infrastructure. [13].8 Geothermal Power Plants Geothermal power plants convert the energy contained in hot rock into electricity by using water to absorb the heat from the rock and transport it to the surface of the earth. tidal ﬂow is extracted by means of propellers with large diameters. remote areas. for example. The latter forms are then converted into electricity. solid or gaseous fuels and other chemical products through a variety of conversion processes [72]. [72].1. It includes all water. they .A ﬁeld may remain productive over a period of tens of years. [80]. 2. biomass is abundantly available and can be considered as a renewable. To generate electricity. Fuel cells are modular. for examples. and certain types of industrial wastes [26].9 Biomass Power Plants The term “biomass” describes all organic matter that is produced by photosynthesis.1.
2. the use of synchronous generators (within the power plants) enables the power output of each plant (and each generating unit within the plant) to be dispatched for any speciﬁed load condition [25]. no extra primary energy can be supplied to the generator units in order to produce more electricity. renewable. when the tide drops.11 Wave Power Plants Waves are generated at the surface of oceans by wind eﬀects which in turn result from the diﬀerential heating of the earth’s surface. By deﬁnition DG units are not centrally dispatched† .Distributed Generation 15 can be trapped in reservoirs behind dams. The DG operator can determine an exact power output of the DG units by controlling the primary energy sources (or fuels) that are supplied to the DG units. building a tidal power plant has to be planned carefully considering the potential ecological impacts. especially during the construction [22]. but several DG technologies enable the DG unit to be controlled locally. especially as DG [17]. the water dam can be used like in a regular operation of a hydroelectric power plant [16]. for example. Later.1. The wave power plant is promoted as electricity generation available in abundance throughout the world. Dispatching a power plant (and a generator unit) is a function of the availability of the primary energy sources that drives the prime mover and the ﬂexibility of the conversion process (ramping up and down). Other DG technologies are based on renewable energy sources where the operator cannot dispatch the DG units because the behavior of the primary energy sources cannot be controlled. just like a tidal plant. A wave power plant extracts wave energy and converts it into electricity [79]. it uses the essentially upanddown motion of the sea surface (wave power). and suited to electrify remote communities. They also cause no fundamental change of the natural rhythm of the tidal cycle and no inundation of the adjacent area. In a steadystate operation of a traditional power system. † the nondispatchable characteristics of DG units are important when a stochastic approach is considered. most of the renewable energy based electrical generation is operated in such a way that the electricity production is maximized. Tidal power plants produce no pollutant. Wave energy is complementary to tidal power. 2. the erection of a wave power plant should be planned carefully. However. Such approach is applied in this thesis in Chapter 6 . instead of using the energy of the sea rushing backwards and forwards (tidal power).2 Output Power Characteristics One of the main characteristics of a power system is that the supply and the demand must be kept in balance at any time. However. Tidal power is a renewable energy source. These factors encourage the implementation of tidal power plants [6]. it is clean and nonpolluting. so that the ecological impacts are minimized. Normally. In case of wind turbines and photovoltaic panels.
A note should be made with regard to geothermal power plants.and wave power plants and CHP plants. microturbines.16 2. Except for the fuel cells. As a summary. Among the DG technologies that can be classiﬁed as noncontrollable DG are small hydro power plants. As a result.2. and power plants driven by biomass. 2.1 lists the various DG technologies and their classiﬁcation as controllable and/or noncontrollable generation. .1 Controllable DG Controllable DG is characterized by its ability to control the fuel supply to the generator. By controlling the fuel that is supplied to the prime mover. and dispatched. these technologies utilize conventional rotating electrical machines for power conversion (synchronous or induction generators). the output power can be determined. driven by prime movers based on reciprocating or combustion turbine technologies.2 Noncontrollable DG Noncontrollable DG represents DG technologies where the DG operator cannot determine the power output of the DG units. fuel cells. geothermal power plants.2. the torque of the prime mover can be adjusted. Table 2. wind turbines. DG technologies can be classiﬁed into two categories: • Controllable DG.2 Output Power Characteristics Table 2. Among the DG technologies that can be classiﬁed as controllable DG are conventional fossilfuel based generators. • NonControllable DG. The geothermal primary energy source is not as ﬂexible as fossil fuels for dispatching the generator units [68].1: Controllable and Noncontrollable Classiﬁcation of DG DG technology Conventional FossilFuel Based Generators Microturbines Combined Heat and Power (CHP) Plants Small hydropower plants Wind turbines Photovoltaics Fuel cells Geothermal power plants Biomass power plants Tidal power plants Wave power plants Controllable √ √ Noncontrollable √ √ √ √ √ √ √ √ √ In short. 2. tidal. photovoltaics.
vo the downstream wind velocity at the exit of the rotor blades [ms−1 ]. a stochastic output power results. ρ. a small hydropower unit is nondispatchable. In practice.3) [51] as P = η × (Eed × AP V total + Ees × AreaP V withsun ). Q the ﬂow rate [m3 s−1 ]. the power produced by a wind turbine is mainly characterized by the wind velocity. is between minimal and maximal values. P denotes the output power [W]. η. where − − → → AP V withsun = ( S × P ) × AP V total (2. H the eﬀective head [m]. are deterministic and constant values. ρ.1) with P the output power [W]. η the overall eﬃciency. e.g. [51] P = (1 + 1 Cp ρv 3 A. (2. Without signiﬁcant storage capacity. Thus. especially when a single wind turbine or plant is regarded [50].2). and A the swept area of the rotor blades [m2 ]. with Cp = 2 vo v )[1 2 − ( vo )2 ] v . Thus. and g the gravitional constant [ms−2 ]. (2. H. and g in (2. A simple expression of the power output for a small hydroplant is [33] P = QHηρg. the power output of a hydro turbine (the prime mover in the small hydropower plant) is practically driven by a directcaptured water ﬂow. The wind velocity itself has a stochastic nature.2) In (2. Photovoltaics The power generated by a PV module is given in (2. As a result.4) . e. v < 4 or v > 25 [ms−1 ]. A. the output power equals zero. 4 < v < 25 [ms−1 ]) can be expressed as [33]. v. when the upstream wind velocity (v) is either below minimal or above maximal operating values of the wind plant. Cp the power coeﬃcient. and to some extent Cp . a small hydropower plant may experience a very large variation in available water ﬂow (Q). ρ the air density [kgm−3 ].Distributed Generation Small hydropower plants 17 Because of the nonavailability of large power impounding (dam).g. Wind turbines The power generated by a wind turbine (provided that the upstream wind velocity. ρ the density of water [kgm−3 ].1) are deterministic and constant.3) (2. any wind speed can occur at any time [61]. and output power (P ) [33]. For small hydropower plants. Moreover.
whereas the characteristics of Eed and Ees are intermittent.3) to (2.12) In (2. which is predicable but variable in nature [9]. P denotes the output power [W]. strongly inﬂuence the values of Eed and Ees . = cos(θ) × cos(αsun ). Therefore. β and αpanel are deterministic and constant.13). Weather changes and cloud movement. Thus.11) (2. In practice.8) (2. the generated electricity is characterized by Eed .  S  = 1. = cos(β) × sin(αpanel ).and the azimuth angle of the sun (θ and αsun ) have daily and seasonal patterns.9) (2.10) (2. A the area of the ﬂow covered by the device [m2 ].14) .6) (2.18 and − → S − → P − → Sx − → Sy − → Sz − → Px − → Py − → Pz 2. and the generated electricity. namely 1. the tide. The power generation of PV is noncontrollable [81]. Eed and Ees the diﬀuse.13) In (2. S and P the solar. = sin(θ). = cos(θ) × sin(αsun ). Wave power plants The power production of a wave power plant can be assessed using [45]. η. η the eﬃciency of the solar panel. AP V total .and panel orientation. Ees .12) P denotes the power extracted from the sunlight [W]. and v the velocity of the water [ms−1 ].and the azimuth angle of the panel [rad]. is the only factor that aﬀects the generating activity of a tidal power plant. − → = [Px Py Pz ]. For a tidal power plant.5 Pabs = αAw Hs . 2 (2. ρ.and the directhorizontal − → − → irradiance [Wm−2 ].7) (2. for example. Tidal power plants The power output of a turbine operating in ﬂowing water is [74] P = 1 ρACp v 3 . This makes the tidal power generation noncontrollable. the output power P depends on the velocity of the water v. (2. The altitude.5) (2. Therefore.2 Output Power Characteristics − → = [Sx Sy Sz ]. (2. A.  P  = 1. and β and αpanel the altitude. and Cp in (2. θ and αsun the altitude.and azimuth angle of the sun [rad]. = sin(β). = cos(β) × cos(αpanel ). ρ the density of the ﬂuid [kgm−3 ].13) are deterministic and constant. Cp the power coeﬃcient of the device (the percentage of power that the turbine can extract from the water ﬂowing through the turbine). θ and αsun .
and no fossilbased fuel is used [64]. wave and sun are exploited. are highlighted. which makes them suited for implementation near load centers where also DG can be implemented [36]. For a wave power plant. the output power is diﬃcult to predict. Batteries are the most widelyused devices for electrical energy storage in a variety of applications.3 Energy Storage Systems A largescale implementation of renewable energy sources for power generation could be expected in future. Batteries can store rather large amounts of energy in a relatively small volume.14) Pabs denotes the average absorbed power. They are also quiet during operation. the hydrogen fuel cell energy storage can be ’charged’ and ’discharged’ reversibly.3. Hydrogen fuel cells are environmentally friendly. In this way.3. 2. The secondary battery works in a reversible reaction. . This hydrogen can be used to operate fuel cells when there is a high demand for electricity. In this case.1 Batteries Batteries store energy in electrochemical form.166 [kgm−1. [67].2 Hydrogen Fuel Cells Hydrogen fuel cells store energy in electrochemical form. Hydrogen is produced by electrolysis of water using the oﬀpeak electricity such as coming from wind turbines. [64]. photovoltaics. α and Aw are deterministic. It converts chemical energy into electrical in discharge. and their technical characteristics from a power system point of view.α is a coeﬃcient that equals 0.Distributed Generation 19 In (2. The socalled primary battery converts chemical energy into electrical energy in a nonreversible process and is discarded after discharge. neither constant nor controllable. In the following sections. The output power depends practically on the wave height (Hs ). so that output power in the order of MWs is accessible. 2. high power ﬂuctuations from these generators can be expected.5 s−3 ] under ideal conditions [45]. They are modular. Hence. if the hydrogen is produced by electrolysis of water. especially when renewable energy sources such as wind. Aw the ﬂoat water plain area [m2 ]. and vice versa in charge mode [51]. energy storage systems. energy storage systems/devices may be needed to cover the resulting imbalances between the power generation and the consumptions [64]. Moreover. a wave power plant is noncontrollable. There are two basic types of batteries. With the noncontrollable nature of these generators. hydro or even nuclear power plants. and Hs the signiﬁcant wave height [m]. 2.
7 PumpedHydroelectric Plants Pumpedhydroelectric plants store energy in the form of potential energy. Flow batteries have a number of advantages. and powerelectronic converter. and the storage capacity can be raised by increasing the size of the tanks of the electrolytes [54].6 Superconducting Magnetic Energy Storage (SMES) Systems Superconducting Magnetic Energy Storage (SMES) systems store energy in the magnetic ﬁeld created by the ﬂow of direct current in a superconducting coil.3 Energy Storage Systems 2. However.3. 2.20 2.3. [23].3. 2.3. The use of SMES is encouraged by several factors [40]. compared to ordinary electrolyte capacitors. A ﬂywheel system consists of a ﬂywheel. An ultracapacitor has an unlimited number of charge and discharge cycles at high rates [10]. a motor/generator.5 Ultracapacitors Ultracapacitors store energy in the form of electrostatic energy (static charge). They use oﬀpeak power to pump water uphill to an elevated reservoir.4 Flywheel Systems Flywheels store energy mechanically as kinetic energy [64]. 2. They are classiﬁed in between secondary batteries and hydrogen fuel cells and they have the characteristic of secondary batteries as they can be charged and discharged [54]. The stored energy can be released by discharging the coil. Similar to a regular capacitor. 2. the power output can easily be varied by increasing the size of the membranes. Secondly.3. it can very rapidly dump or absorb power from the grid as the only limitations are the control loop and the switching time of the solidstate components connecting the coil to the grid. its eﬃciency is very high (up to 95%) as no conversion of energy to other forms is involved. For example. Firstly. the water is released to ﬂow to a lower reservoir. It also has an unlimited number of charging and discharging cycles. . Its equivalent internal resistance is more than tens times lower than that of a battery. and its potential energy is used to drive turbines [64]. allowing more than tens times higher discharging/charging currents [23]. ultracapacitor’s capacitance can be more than tens times higher. The ﬂywheel speeds up as it accumulates energy and slows down as energy is released [29]. the electric energy is stored by means of charge separation [10]. A ﬂywheel system has a high energystorage density.3 Redox Flow Batteries Redox ﬂow batteries store energy in electrochemical form. When electricity is needed. This makes the ultracapacitor suited for shortterm. highpower applications.
the plant’s generator produces power using a conventional natural gas combustor and the compressed air. [65].as a motor (powered by the oﬀpeak electric energ) . the generator operates in reverse .8 CompressedAir Systems Compressedair energy storage systems store energy in the form of potential energy by compressing air within an air reservoir.to provide mechanical energy to the air compressors [64]. are its availability for largescale storage (high capacity and power rating). When electricity is needed. and the long life duration [65]. )ORZ %DWWHULHV 3XPSHG +\GURHOHFWULF 3ODQWV &RPSUHVVHG $LU 6\VWHPV &DSDFLW\ >:K@ +\GURJHQ )XHO &HOOV %DWWHULHV 60(6 )O\ZKHHOV 8OWUDFDSDFLWRUV 3RZHU >:@ Figure 2. and run through expansion turbines to drive an electric generator. the unlimited charge and discharge cycles. When charging. using a compressor powered by oﬀpeak electric energy.3. Among the positive aspects of a compressedair energy storage system. It has an unlimited charging and discharging cycle and long life duration.1 shows an indication of the working areas of these energy storage systems. Figure 2. the air is withdrawn. however it needs speciﬁc geography. Such a plant uses about onethird of the premium fuel of a conventional simplecycle combustion turbine and produces onethird of the pollutants per kWh generated. This technology is considered as a hybrid storage and generation plant because it uses fuel and electricity in its storage cycle [64]. heated via combustion.Distributed Generation 21 A pumpedhydroelectric energy system enables largescale energy storage with a high capacity and power rating.1: Indication of the working areas of diﬀerent energy storage systems [4] . During discharge. The implementation of such an energy storage system is limited by the requirement of a signiﬁcant land area with suitable topography for the upper and lower plants. 2.
p the number of pole pairs of the magnetic ﬁeld circuit.4. power electronic interface must be used.1 Direct GridConnected DG Figure 2.22 2. the generator can produce power at the grid frequency.15) is applicable. In this way. p nsyn . as with solar panels and fuel cells. but higher frequency than that of the grid. (2. so that it operates at a constant speed. gas turbines. By controlling the prime mover. This may occur if the primary energy sources are intermittent in nature (e.e. and f the frequency of the generated voltage [Hz]. DG may also generate AC power by means of a fastrotating prime mover (e. In this way. The diﬀerence is in the energy source that drives the prime mover. In general. If this AC power is generated at the system frequency or close to it. The frequency of the system voltage is directly proportional to the speed of the synchronous generators.g. an interface that converts DC to AC (at the system frequency) is a must. wind.15) f= 60 where nsyn denotes the synchronous speed of the generator in revolutions per minute [rpm]. this generation (or conversion) can be done by means of either a synchronous or an induction generator. the generator can be directly coupled to the grid. the DG connection to the power grid can be classiﬁed into two categories: • Direct gridconnected DG. combined cycle plants and cogeneration plants. and drives the generator. 2. When the primary energy is converted into mechanical energy that is used to drive electric rotating machines (synchronous or induction machines).g. When DC power is generated (i.. wave) and if it is economically better to adapt the speed of the generator accordingly. • Indirect gridconnected DG. a microturbine). However. AC power is generated at a constant. The synchronous generators driven by their turbines are responsible for maintaining the frequency in the system. tidal. (2. Also in this case. This is the case in steam plants. f = 0).e. Due to the variety of primary energy sources/prime movers.2 shows a schematic diagram of a DG unit connected directly to the AC grid. Power is mostly generated by means of synchronous generators. . if the frequency deviates from the system frequency. an interface is required. DG units equipped with a synchronous generator For a synchronous generator. The prime mover operates at a constant speed. i.4 DG GridConnection Characteristics 2. DG can generate electricity by means of either rotating electrical machines or static electrical generators. AC power is generated.4 DG GridConnection Characteristics Public electrical power systems operate as AC (Alternating Current) systems standardized at either 50 Hz or 60 Hz.
or torque) applied to it. certain types of wind turbines). highfrequency AC (e. solar panels and fuel cells). (2. about 1 per cent [69]. . an induction generator may be used. the diﬀerence between the rotational speed at peak power and at idle is very small. Usually.g. As such. microturbines) or AC with variable frequency (e.16) where s represents the slip.3) to connect the lowspeed driving shaft to the highspeed generator shaft (1200 to 1500 rpm).4. In this case. In practice however.Distributed Generation 23 3ULPDU\ HQHUJ\ VRXUFH 3ULPH PRYHU &RQVWDQW VSHHG 5RWDU\ HOHFWULFDO JHQHUDWRU *ULG IUHTXHQF\ $& Figure 2. Therefore. a DG unit is connected to the grid in an indirect way.g. n is no longer constant and n = (1 − s)nsyn .g. 3ULPDU\ HQHUJ\ VRXUFH 3ULPH PRYHU *ULG IUHTXHQF\ $& 5RWDU\ HOHFWULFDO JHQHUDWRU /RZ VSHHG +LJK VSHHG *HDUER[ Figure 2. Induction generators are usually applied in small hydropower plants and older design or small wind turbines. an interface is necessary to connect these devices to the grid.2: Schematic diagram of direct gridconnected DG DG units equipped with an induction generator When the prime mover does not operate at a constant speed. a gearbox is used (Figure 2. Several DG types generate electricity as DC (e. In this case the speed of the induction generator may vary with the turning force (moment.3: Schematic diagram of grid connected DG via gearbox 2.2 Indirect GridConnected DG A power system operates at a constant system/grid frequency.
Examples of these kind of DG units are fuel cells and solar panels. • Induction generator with power electronic converter in the rotor.4 shows a simpliﬁed layout of such a plant. use rotating electrical machines for electricity generation but are connected to the grid via powerelectronic interfaces.24 2.4: Interfaceconnected DG with DC output DG generating high/variablefrequency AC Some DG units. or AC with variable frequency. Figure 2. • When the primary energy sources cause the prime mover to drive the rotating electrical generator at a variable speed. The primary energy sources are converted into electricity without of a rotating electrical machine. This is illustrated in Figure 2. • DG generating either highfrequency AC or AC with a variable frequency. we basically distinguish two situations: • DG generating DC. such as microturbines. There are two situations in which powerelectronic converters are needed to interface the rotating electrical machine to the grid: • When the rotating electrical machine generates a highfrequency AC (far beyond the grid frequency). no rotating parts are involved. DG generating DC A DG unit with DC output is primarily characterized by static electric generation. a ﬁlter can be implemented at the outputstage of the inverter to clean the AC voltage. leading to a variablefrequency AC. wind turbines and tidal power generators.e. is rectiﬁed into DC. i. 3ULPDU\ HQHUJ\ VRXUFH 6WDWLF HOHFWULFDO JHQHUDWRU *ULG IUHTXHQF\ $& '& WR $& FRQYHUWHU '& '& OLQN FDSDFLWRU '& 3RZHUHOHFWURQLF LQWHUIDFH Figure 2. before it is converted into gridfrequency AC. The DC output may be ﬂuctuating and is smoothed by a capacitor before converted to AC at the grid frequency. . A capacitor is used to smooth the DC. A ﬁlter can be implemented to clean the resulting AC voltage.4 DG GridConnection Characteristics For indirect gridconnected DG. In addition.5. The highfrequency AC.
independently of the mechanical rotor speed [28]. [75].6). Table 2. The mechanical and electrical rotor frequencies are controllable over a certain range and the electrical stator and rotor frequency can be matched.6: Induction generator with power electronic converter in the rotor .Distributed Generation 3ULPDU\ HQHUJ\ VRXUFH 3ULPH PRYHU 9DULDEOH RU 9HU\ KLJK VSHHG 9DULDEOH RU +LJK IUHTXHQF\ $& 5RWDU\ HOHFWULFDO JHQHUDWRU $& WR '& FRQYHUWHU *ULG IUHTXHQF\ $& '& WR $& FRQYHUWHU 25 '& '& OLQN FDSDFLWRU '& 3RZHUHOHFWURQLF LQWHUIDFH Figure 2.5: Interfaceconnected DG with AC output Induction generator with power electronic converter in the rotor The stator windings of a variable speed induction generator can be connected directly to the grid with the rotor windings connected to (bidirectional) power electronic interface (Figure 2.2 summarizes the grid connection classiﬁcations of DG. Primary energy source Prime mover Gridfrequency AC Rotary electrical generator DC AC to DC converter DC link capacitor DC DC to AC converter Variable speed Gearbox Powerelectronic interface Figure 2.
26 2. a large amount of generation will be ’hidden’ behind powerelectronic interfaces. a powerelectronic interface is necessary to convert the highfrequency AC to gridfrequency AC.g.5 DG Prospects: ConverterConnected DG Table 2. hydrogen fuel cells. When a highspeed ﬂywheel is used. as well as ﬂexible energy storages are implemented [19]. batteries. Therefore. When the energy is stored by means of pumpedhydroelectric plants and compressedair systems. for active and reactive power control. 2. powerelectronic inverters are necessary to convert the stored energy to gridfrequency AC (and vice versa). Powerelectronic converters can be used to maximize the energy yield.3 Connecting Energy Storage to the Grid Energy storage can be connected to the grid by means of powerelectronic interfaces too.4. When the energy storage device is charged/discharged by DC (e. and SMES systems). The role of powerelectronic interfaces will become even more important if hybrid networks that combine AC and DC. and for powerquality improvement [60]. When powerelectronic converters are used to interface DG with the grid. This will certainly exert a major inﬂuence on the power system and its operation.2: Direct and Indirect gridconnected DG Rotating Electrical Generator (AC output) √ √ √ √ √ Static Generator (DC output) Direct GridConnected √ √ √ √ √ √ √ √ √ √ √ √ √ √ √ Indirect GridConnected √ DG Technology Conventional Generators Microturbines CHP Plants Small HydroPower Plants Wind turbines Photovoltaics Fuel cells Geothermal Power Plants Biomass Power Plants Tidal Power Plants Wave Power Plants √ √ √ √ √ √ 2. [60].5 DG Prospects: ConverterConnected DG A largescale implementation of DG units may be foreseen. ﬂow batteries. the generator can be connected directly to the grid. There are two possible protection systems for DG units that are connected to the power system via a . with emphasis on renewables. rotating electrical generators are used to extract the stored energy and convert it into electricity (gridfrequency AC). as it may inﬂuence the power system dynamic behavior [58]. the impact of the protection scheme should be considered. ultracapacitors. In this scheme.
In this chapter.Distributed Generation 27 powerelectronic interface. In this chapter. An overview of the stateoftheart of DG technology is given. Although its classiﬁcation as DG depends on the way a generator is incorporated into a power system. [24]. A broader and more straightforward deﬁnition of DG is advocated by [1] and [52]. and reconnects the DG as soon as the voltage recovers. This is mostly motivated by the following objectives: • To preserve selective protection with simple overcurrent relays as usually applied in the radiallyoperated MV and LV networks. however. From the wellknown CIGRE Working Group 37. However. powerelectronic interfaced DG with a protection scheme of the ﬁrst type is considered. this distinction is made. Although these two aspects are wellknown. . Another way of examining DG is by looking at the interactions between the generator and the grid.23 [12]. the impact of powerelectronic interfaced DG with a protection scheme of the latter type is simulated in Sections 4.6 Summary Many deﬁnitions of DG exist. which includes all electrical generation sources connected directly to the distribution network or on the customer side of the meter. in many literature. and is mostly connected to the distribution network. In this way. as it gives basic ideas and illustrations for supporting modeling of DG used in the simulation in the later chapters. however. The reason is that many generator units driven by renewable sources of energy inherently possess DG characteristics. In this thesis. The other keeps the DG connected to the power network during a fault (ridethrough capability). The classiﬁcation of DG based on the output power characteristics gives basic ideas on stochastic approach done in Chapter 6. these classiﬁcations are seldom explicitly presented in literature. DG is distinguished based on the primary energy sources [7]. DG is basically deﬁned as electrical generation which is not centrally planned and dispatched. • To prevent the DG units from remaining connected to the grid during the reclosure dead time. the existing DG technologies are brieﬂy highlighted according to the primary energy sources. 2. most technical standards require a protection scheme of the ﬁrst kind [3]. they could keep the network energized and negate the selfextinction of arcing faults and in case of permanent faults the presence of DG still connected to the grid can be dangerous for utility personnel during repairs. The classiﬁcation of DG based on the way a DG is connected to the network provides a basic idea on modeling of DG for power system transient simulation used in Chapters 35. as comparison. Currently. The ﬁrst type of protection systems automatically disconnects the DG from the power system when the voltage of the system drops below a certain level. there are two classiﬁcations: (1) based on the output power characteristic. and (2) based on the way DG is connected to the network.5.4 and 5. [52].
Energy storage systems can be applied to smooth these intermittent eﬀects.28 2. .6 Summary Literature survey on DG characteristics reveals that the output power of noncontrollable DG units. especially the ones driven by renewable energy sources. of such DG units are highlighted. can show high outputpower ﬂuctuations. taken from literature. This chapter ends with an overview of powerelectronic interfaced DG. The basic elements of protection schemes.
such as the magnitude and phase angle of the voltage at each bus and the active/reactive power ﬂowing in each line.Chapter 3 Stability of Systems with DG A requirement of power system operation is to balance the electricity supply and the demand at any time. the system is in steady state. both active and reactive demands continuously change. When this steadystate condition is subjected to a sudden change or a sequence of changes in the system quantities. describe the operating condition of a power system. where the modeling of DG technologies and software used for large transmission system transient simulations are presented. including grid losses. The connection of this chapter with Chapter 2 is mainly found in Section 3. it is key that the system remains in operation and is able to return to a stable state. An introduction on the phenomena of power system stability is the start of this chapter. The stateoftheart DG technologies presented in Chapter 2 leads to the classiﬁcation of DG based on the output power behavior and grid connection characteristics. If they are constant in time. The ability of a power system to return to a state of operating equilibrium under normal conditions and to regain a new equilibrium after being subjected to a disturbance is deﬁned as power system stability [34]. 3.1 Classiﬁcation of Power System Stability All measured (or calculated) physical quantities.4. A properly designed and operated power system should be able to maintain this balance both under normal conditions (steady state) as well as after disturbances (dynamic). Even under normal operating conditions. This chapter emphasizes on the DG impact on the transmission system stability. For a reliable service. a power system is often subjected to disturbances. the system undergoes a disturbance from its steady state [25]. A power system is always dynamic. . As a system with interconnected machines and components may cover a wide geographical area.
The various types of power system stability can be classiﬁed correspondingly as the diagram in Figure 3. large load changes. voltage and frequency.30 3.1 [34]. A system is said to be steadystate stable. a system reaches a new acceptable steadystate condition is diﬀerent from the original steadystate condition. and line switching are examples of large disturbances [25]. For small disturbances.1: Classiﬁcation of power system stability [34] 3. One can distinguish three types of instability mechanisms depending on which parameters are most aﬀected by the disturbance: rotor angle. The system is called transiently stable. there exist small and large disturbances. loss of generating units. a change from a steadystate condition can be analyzed by using the linearized system’s dynamic and algebraic equations.1. In many cases. Small variations in load and generation are examples of such small disturbances. In these cases. power system stability can be further classiﬁed based on the typical ranges of the time period of response actions as shortterm (seconds).1 Rotor Angle Stability Rotor angle stability concerns the ability of interconnected synchronous machines in a power system to remain in synchronism under normal operating . or longterm stabilities (minutes). Furthermore. 3RZHU V\VWHP VWDELOLW\ 5RWRU DQJOH VWDELOLW\ )UHTXHQF\ VWDELOLW\ 9ROWDJH VWDELOLW\ 6PDOO VLJQDO VWDELOLW\ 7UDQVLHQW VWDELOLW\ /DUJH GLVWXUEDQFH YROWDJH VWDELOLW\ 6PDOO GLVWXUEDQFH YROWDJH VWDELOLW\ 6KRUW 7HUP 6KRUW 7HUP /RQJ 7HUP 6KRUW 7HUP /RQJ 7HUP Figure 3. the linearized dynamic and algebraic equations are no longer valid. Disturbances like transmission system faults.1 Classiﬁcation of Power System Stability Depending on their origin and magnitude. especially under large disturbances. if it is able to return to essentially the same steadystate condition of operation after being subjected to a small disturbance.
however.1. in case of the system split into islands) as evidenced by the mean frequency. Diﬀerent from rotor angle stability. The instability may result in switching oﬀ of generators. [34]. Frequency is only touched upon in Chapter 7. Instability that may result occurs in the form of increasing angular swings of some generators. or large disturbance voltage stability when it is concerned with a system’s ability to control voltages following large disturbances such as transmission system faults [11]. 3.3 Frequency Stability Frequency stability is the ability of a power system to maintain the frequency within an acceptable range. the resulting system response involves large excursions of generator rotor angles. 3. rather than the relative motion of generators. 3.2 Voltage Stability Voltage stability is deﬁned as the ability of a power system to maintain steady acceptable voltages at all buses in the system under normal operating conditions and after being subjected to a disturbance. which leads to the loss of synchronism [34]. stability (which is then called small signal stability) can be analyzed using a linearized set of system equations. overloading of lines and most probable splitting up the system in subsystems. Analyzing voltage stability is not in the scope of this thesis. The instability that may result occurs in a progressive and uncontrollable drop in voltage [34]. and is inﬂuenced by the nonlinear powerangle relationship [11]. [34]. 3. following a system upset resulting in a signiﬁcant imbalance between generation and load [34].1 Swing Equation The governing equation for rotor motion of a synchronous machine is based on the elementary principle in dynamics which states that the net accelerating . When the disturbance is relatively small.2 Rotor Dynamics of Synchronous Machines As long as a power system can rely on synchronous machines for generation of electrical power.2. a necessary condition for satisfactory system operation is that all synchronous machines remain in synchronism [34]. frequency stability is determined by the overall response of the system (or each island. where keeping power balance in power system with Active Distribution Systems will be discussed.Stability of Systems with DG 31 conditions and after being subjected to a disturbance. Voltage stability can be classiﬁed into small disturbance voltage stability when it is concerned with a system’s ability to control voltages following small perturbations such as small changes in loads.1. The latter is mainly focused on in this work. Frequency stability could concern any disturbance and therefore it is not classiﬁed into the small nor the large disturbances. In case of transient stability.
e. 2 Smach (3.3) can be further normalized in terms of unit inertia constant H. In (3.7) Equation (3. (3. governing the rotational dynamics of a . ωs dt2 (3.5) 2H d2 δm = Pa = Pm − Pe . (3.4) 1 M ωsm . The above equation (3. and its angular acceleration [25] J d2 θm = Ta = Tm − Te . combining it with (3.1) with θm the angular displacement of the rotor with respect to a stationary axis. Pe and Pm are the accelerating. deﬁned as the kinetic energy at rated speed divided by the rated apparent power of the generator Smach as [25] H= or H= yielding 2 1 Jωsm 2 Smach (3.1). dt2 (3.1) and recalling that power (P ) equals torque (T ) times angular velocity (ω) one arrives at the equation d2 δm = Pa = Pm − Pe .6) ωsm dt2 Moreover. the electrical and the mechanical power.7) is called the swing equation of the machine. By introducing the angular velocity of the rotor from the synchronously rotating reference axis ωm = dθm for a convenient notation and dt twice diﬀerentiating (3. provided both ωs (the synchronous speed of the rotor) and δ (the angular displacement of the rotor from the synchronously rotating reference axis) have consistent units. (3. Ta .32 3. For a given speed of synchronously rotating reference axis.2) where δm as the angular displacement of the rotor from the synchronously rotating reference axis. Tm and Te are the driving torque of the prime mover and the net electromagnetic torque. the angular displacement can be rewritten as θm = ωsm · t + δm . Tm = Te .2 Rotor Dynamics of Synchronous Machines torque. It is the fundamental equation in the stability study. J.3) M dt2 where M = Jωm is the inertia constant of the machine. by noting that both δm and ωsm are expressed as mechanical speed. The machine is said to be working in synchronous speed (or in synchronism) if Ta = 0. respectively. (3. respectively. i. ωsm .2) with respect to time. is the product of the total moment of inertia of the rotor. and Pa .6) can be written as [25] 2H d2 δ = Pa = Pm − Pe .
in a response to a disturbance. This system consists of a synchronous generator(1).2: Schematic diagram of the twomachine system for stability studies. while the electrical network reacts almost instantaneously under disturbances. In this system. the turbine has some delay before its control mechanism causes it to react. the rotor deviates from synchronous speed. ( ( 7UDQVPLVVLRQ 1HWZRUN Figure 3. connected to a large external system (2). Due to the characteristic of the prime mover and the related controllers. e. The machine will always operate at steadystate synchronous speed if Pe = Pm . The graph of its solution is known as the swing curve.7) H and ωs are known parameters of the synchronous machine. This assumption is based on the fact that. a so called “inﬁnite bus”. An inspection of the swing curves of all machines of the system can be used to show whether a particular machine remains in synchronism after being subjected to a disturbance [25]. This is valid for time frames smaller than 10 seconds (short term stability).g. modeled as a machine with very large inertia.2 PowerAngle Equation When using the swing equation (3. Therefore.Stability of Systems with DG 33 synchronous machine.2.2. given initial conditions for speed and angle. The dynamic behavior of a synchronous machine can be described by means of this equation and its graph. The behavior of Pe is better explained using a generic twomachine system as shown in Figure 3. Pm can be assumed constant during a transient disturbance. If Pe = Pm . a simple model consisting of a constant voltage behind a transient reactance is used to represent both machines. only Pe is essential for the solution of the swing equation. Transient reactances associated with the transient internal voltages of both machines E1 and E2 are included in the transmission network [25] . Therefore. the behavior of the angular positions of the rotor of synchronous machines (δ) are dictated only by Pm and Pe for a given machine and rated system frequency. 3.
The plotted powerangle curve of most critical stable situation is shown in Figure 3.34 The admittance matrix Ybus = 3. located at the crossing of the horizontal line P = Pm with a curve drawn and sketching the original Pe0 curve.8) represents the transmission network between bus 1 and 2. corresponding to Y12 . 3. Furthermore.3). then Pc = 0 and γ = 0. The fault is cleared after a certain period of time. a three phase to ground fault through no intermediate fault impedance is assumed. Pmax = E1 E2 Y12 . For simpliﬁcation purposes. and (3. For illustration purposes a twomachine system is commonly used in many text books. without disconnecting any transmission equipment.13) 2 dt 2H 2HX12 The nonlinearity is clear even for this simple example.10) = Pc + Pmax sin(δ − γ). The electric power output of the feeding generator Pe is determined by Pe where Pc = E1 2 Re(Y11 ).9) and δ = δ1 − δ2 .2.4.12) ′ ′ ωs Pm ωs E1 E2 d2 δ = − sin δ.9) reduces to Pe = Pmax sin δ. including the transient reactances of the two equivalent machines. one could rely on numerical methods to obtain the solution [25]. The original (steadystate) operation is characterized by the (prefault) rotor angle δ0 . (3.11). (3. with δ1 and δ2 the angular displacement of the rotor from the synchronously rotating reference axis associated with the transient internal voltages E1 and E2 . yields (3. such that all the elements of Ybus are purely imaginary (susceptances). In this case.11) where Pmax = E1 E2 /X12 with X12 the transfer reactance between E1 and E2 .2 Rotor Dynamics of Synchronous Machines Y11 ∠θ11 Y21 ∠θ21 Y12 ∠θ12 Y22 ∠θ22 (3. (3. Formal solutions of such an equation cannot be explicitly found.7) and (3. The or . (3. It is assumed that a temporary threephase fault occurs at bus1 (Figure 3. If the network resistance is neglected. In this case.3 Equal Area Criterion d2 δ ωs Pm ωs Pmax = − sin δ. 2 dt 2H 2H Combining (3. γ = θ12 − π/2. the examination of the system stability can be done with a direct approach without solving the swing equation. with θ12 = arg (Y12 ).
Pe0 . the unstable equilibrium point (UEP) is equal δmax = π − δ0 . If the angle δ reaches a value larger than UEP. So. Respectively. PeD . the duringfault curve PeD equals zero and the postfault curve PeP equals the original Pe0 curve. PeD and PeP represent the airgap power as well as the terminal power before (prefault). The accelerating area Aacc and the decelerating area Adec are equal eﬀective transmission system is unaltered except when the fault is present. (3. The Equal Area Criterion (EAC) simply recognizes the fact that the faulted system is still capable of recovering stability as long as the inequality Aacc < Adec is satisﬁed. there is insuﬃcient decelerating energy. This leads to instability [82].3: Example of twomachine system for stability studies. during and after the fault (postfault). The short circuit is eﬀective at bus1 so that the electrical power output from the generator. The postfault equilibrium point is determined by the intersection of Pm with PeP (equals Pe0 ) and this provides δP .4: Plotted power angle curves showing the critical clearing angle δcr .Stability of Systems with DG 35 Figure 3. A fault is applied at bus1 3 3PD[ $GHF 3P $DFF 3H ' 3H 3H3 3PD[ VLQ 3 FU PD[ * Figure 3. respectively.14) . The borderline case corresponds to Adec = Aacc . is zero until the fault is cleared.
g. is zero during the disturbance). the increase in rotor speed and the angle separation between the generator and the inﬁnite bus become ωs Pm 2 δ(t)t=tcr = t + δ0 . On the other hand. and the rotor angle increases without limit. the Critical Clearing Angle (CCA) of (3. the CCA and CCT are often used as power system stability indicators.18) indicate a borderline situation where the faulted system is capable of recovering stability as long the angle of the synchronous machine is less than the CCA or the fault clearing time is shorter than the CCT. twomachine system or one machine inﬁnite bus system) the CCA and CCT cannot be explicitly found without computer simulation [25]. as well as on a more general case than a simple power system (e.36 3. ωs Pm (3. As it has been derived from the Equal Area Criterion concept. when both CCA and CCT are surpassed. when simulations are to be done on large power systems.14). denoted by δcr ) at δcr = cos−1 ((π − 2δ0 ) sin δ0 − cos δ0 ).3 System Stability Indicators Considering the applied fault and the powerangle curve (Figure 3.e.7) twice. at the instant of the critical fault clearing. Therefore.3 System Stability Indicators To assess the performance of power systems indicators are needed. the accelerating Aacc and the decelerating area Adec can be written as δcr δmax Aacc = δ0 Pm dδ.17) 4H cr The corresponding critical clearing time (CCT) is obtained as tcr = 4H(δcr − δ0 ) .15) in combination with (3.16) and the Critical Clearing Time (CCT) of (3. the diﬀerence between the actual clearing angle and actual clearing time and their respective critical counterparts (CCT and CCA) deﬁnes the “stability margin” of the system (assuming the system is stable and there is a positive margin). where the determination of CCT (and CCA) requires several time domain simulation runs.15) By solving (3. Adec = δcr (Pmax sin δ − Pm )dδ. i. the system reaches the critical clearing angle (CCA. (3.4). PeD . .16) Integrating (3.18) 3. (3. (3. One of the practical methods for determining CCT (and CCA) of a power system is the timedomain numerical integration method (the stepbystep time domain simulation) [82]. provided that Pa = Pm (Pe . the rotor of the machines speeds up. However. considering this clear analytical/explicit relationships between the CCA/CCT and power system stability. When a fault is applied in a power system.
the more stable the system is.5. As long as this is the case. • Oscillation duration.nom denoting the maximum and the rated rotor speed of a centralized generator [rpm]. the more instable the system is.5 shows both indicators of maximum rotor speed deviation and the oscillation duration [71]. when scenarios are compared.3. the rotor would continue to oscillate around the rotor angle of the operating point (δ0 in Figure 3. an identical disturbance will be applied in each scenario. when a disturbance occurs.nom . To quantify the indicators.5 s [71].max and ωr. 3. as proposed ﬁrstly in [71]. In the absence of damping. This indicator is proposed to assess the rotoranglestability performance of (centralized) synchronous generators that drive a transmission system with limited inertia.Stability of Systems with DG 37 In this work more practical transient system indicators are used.19) with ωr. the shorter the rotor speed deviates from rated.1 Maximum Rotor Speed Deviation The maximum rotor speed deviation is deﬁned as the maximum rotor speed value attained during the transient phenomenon [71].21) are used: maximum rotor speed deviation = ωr. In other words.3. In this work. (3.max − ωr. namely: • Maximal rotor speed deviation. The indicator of oscillation duration implies that the longer the rotor speed deviates from the rated value when a disturbance occurs.1). at a certain clearing time a higher maximum rotor speed deviation (the faster the rotor accelerates) suggests a lower stability margin. (3.20) . eﬃciency in computing time is obtained by using these indicators rather than CCT (and CCA).19)(3.14). A remark should be made here. as a fault is simulated. Above all.2 Oscillation Duration The oscillation duration is deﬁned as the time interval between the start of the fault and the instant after which the rotor speed stays within a bandwidth of 10−4 pu (on the basis of the rated rotor speed) during a time interval longer than 2. The system is considered stable if after a fault (disturbance) all (or a limited number of) centralized generator units in the test system remain in synchronism. 3. the more instable the system becomes. oscillation duration = tosc − tf . a test system with limited inertia is used (Section 3. In the analysis within this work. there is a risk of instability. The oscillation of the rotor speed that deviates from the rated speed is recorded. It suggests that the more/faster the rotor speed (of the synchronous generators) deviates from the rated value when a disturbance occurs. (3. Figure 3. Thus when two cases are compared.
thermal. the (original) conventional centralized power plants (e.nom) ωr(t+n∆ t) − ωr(t)≤ 10−4. it begins to inﬂuence the overall power system behavior.5 10 Time [s] 15 −3 0 Figure 3. ∆t (3.4 DG and Large System Dynamic Simulation 4 3 Rotor speed deviation [pu] 2 1 0 −1 −2 max. .n=1. n = 1. when DG penetration becomes very high. In this case. DG was implemented in power systems on a small scale. In this case..5/∆ t oscillation duration tf tosc 5 tosc + 2. the dynamic behavior of the existing system due to this high degree of DG needs to be understood/studied. the penetration level of DG may become important.4 DG and Large System Dynamic Simulation In the earliest stages. 2. rotor speed deviation (ωr.2. 3. nuclear or hydro) remain in operation and still cover a signiﬁcant part of the load and the voltage and frequency control are still within the responsibility of the large/centralized synchronous generators. · · · . DG can be treated as negative load in power system dynamics studies. However.5 }. especially its interaction with the existing generating equipment.. Therefore.5: Transient stability indicators: maximum rotor speed deviation and oscillation duration [71] with tf the time [s] when the fault is applied.38 x 10 −3 3..nom/ωr. and tosc = min{t : ωr (t + n∆t) − ωr (t) ≤ 10−4 .. As DG implementation increases. to which the power system stability problem is mostly related.g.max − ωr.21) with ωr (t) the rotor speed at a time t and ∆t the simulation step (in this research ∆t = 10−3 s is used).
Therefore. not only depends on the physical construction (e.g. the available models within the software can be used and they do not pose special problems in power system dynamics. DG can be connected to the grid directly and/or indirectly via a power electronic converter. Machines therefore have an inherent behavior during voltage disturbances. but also for the largest part on the controller and its parameters. Converter Connected DG: Voltage Source Converter When the indirect gridconnected DGs are considered. Section 3. The response of electrical machines is determined by the fundamental electromechanical laws and physical construction of the machine. where a ﬁrstorder ﬁlter (Lf and Rf ) between the converter and the grid is used to reduce the harmonics injected by the switching of the converters. The challenge however lies on how to incorporate the DG in the distribution system commonly represented only by the load at the main connection between the transmission system and the distribution feeder in power system simulations.1 Modeling DG Technologies As elaborated in Chapter 2. • A setpoint for the reactive power is derived from the actual value of the terminal voltage (Ugrid ) if a terminal voltage controller is present. the output of these DG technologies are driven by the output of the power electronic interface basically representing a voltage source converter. There is a major diﬀerence in the response to voltage dips (that occur during a disturbance) of electrical machines at the one hand and power electronic converters at the other. the most used converter nowadays. This makes it more diﬃcult to provide a general model that can be used for simulation in power system studies [42]. The power electronic converters response.5. .3 deals further with this issue.4. Electrical Rotating Machines The direct gridconnected DGs are based on directly coupled synchronous or induction generators.Stability of Systems with DG 39 3.6. • The actual values of terminal voltage and currents are measured and are inputs to the current controller. [70]: • A setpoint for the active power to be supplied to the grid is determined by the electrical conversion either from a static electrical generator (Chapter 2) or prime mover and ACDCconverter (Chapter 2). power electronic components cannot withstand large currents). The operating principle of the grid connected voltage source converter can be summarized as follows [42]. These technologies can be modeled using models of a conventional synchronous or induction generator. The example of a circuit topology of such converter is taken from [42] as shown in Figure 3. readily available in power system dynamics simulations software [53].
because it facilitates the interaction with the load ﬂow module. PDG and QDG denote the ‘real’ .4 DG and Large System Dynamic Simulation • The current controller generates control signals for the semiconductor switches in such a way that the current that ﬂows. the grid representation in the simulation packages and the typical time step used. Applying this model to power system dynamic simulation can done basically by following Algorithm 1: Algorithm 1 Calculate IDG Require: Pnom .40 3. + Idc Lf Rf Ia Ib Ic Ugrid Udc C  Uan Ubn Ucn Figure 3. Qnom PDG ⇐ Pnom QDG ⇐ Qnom ∗ −→ − +jQ IDG ⇐ (PDG− − DG ) −→ UDG Pnom and Qnom denote the nominal/setpoint of the active and reactive power output of the converter connected DG unit.6: Threephase full bridge Voltage Source Converter with 6 IGBT switches [42] Converter Connected DG: Constant PQSources When transient studies are performed on large systems. do not allow detailed modeling of power electronics. injects the desired amounts of active and reactive power in the grid. as it is required by power system dynamics simulation packages. that uses active and reactive power as an input. Moreover. The use of this PQsource model for DG is justiﬁed in [70]. Only the active power P and reactive power Q (or active and reactive current) supplied by the converter are of interest then. The steady state values and the time required to reach a new set point for this PQsource model is similar to both the voltage source converter and the controlled current source converter with hardly any overshoot occurring [70]. it is usual to model the DGs as PQsources. This is especially advantageous in power system dynamics simulations.
each with their own modeling assumptions and limitations. 0.85 pu). On the other hand. additional equations should be applied to Algorithm 1.15 pu).g. as discussed below. Two ‘minimal’ voltage levels are commonly applied here. Algorithm 3 is applied. the current limiter and the operating voltage applied in the DG gives a limitation 2 2 in the total power Snom = Pnom + Qnom . when the voltage drops below the second voltage level Umin2 (e. IDG −→ − denotes the current generated by the DG unit and UDG the terminal voltage where DG unit is connected (as complex phasors). Further. the above DG models are integrated in the various simulation software packages.2 Power System Dynamics Software Packages For the investigation of DG impact on power system transient stability. 0.g. the DG is disconnected from the network. . Qnom . Due to the diﬀerent protection schemes of converter connected DG. 100% up to 120% of the rated value). Therefore it is common to set a PQsource to represent the converter connected DG for these DG units as a source that delivers constant active power P and limited amount of reactive power Q [71]. When the protection system automatically disconnects DG from the network when the voltage of the system drops below a certain level Umin (e. 0. for a given operating active power P setting. Therefore. due to the limitations (“by construction”) of the converter. Most DG based on renewable energy generation is intended to maximize the active power output.g. Note that a current limiter is commonly applied in the latter case. the output current of the DG is limited to the maximum value Imax (in practice. the Q must be within a certain area.85 pu) and reconnects the DG as soon as the voltage recovers. When the voltage of the system drops below the ﬁrst voltage level Umin1 (e. as constrained by the total apparent power. Umin if UDG ≥ Umin then PDG ⇐ Pnom QDG ⇐ Qnom ∗ −→ − +jQ IDG ⇐ (PDG− − DG ) −→ UDG else PDG ⇐ 0 QDG ⇐ 0 −→ − IDG ⇐ 0 When the protection system keeps the DG connected to the power network during a fault. 3.4. Algorithm 2 Calculate IDG Require: Pnom .Stability of Systems with DG 41 −→ − active and reactive power output of the converter connected DG unit. Algorithm 2 is applied.
Basically. current and/or frequency are determined. PSS/E falls within the software packages that oﬀer dynamic simulation capability. In these packages for examples. node voltages and branch ﬂows result.42 3. Imax if UDG ≥ Umin1 then PDG ⇐ Pnom QDG ⇐ Qnom ∗ −→ − +jQ IDG ⇐ (PDG− − DG ) −→ UDG else if Umin1 ≥ UDG ≥ Umin2 then IDG ⇐ Imax − →−→ − − PDG ⇐ Re(UDG IDG ∗) − →−→ − − QDG ⇐ Im(UDG IDG ∗) else {UDG < Umin2 } PDG ⇐ 0 QDG ⇐ 0 −→ − IDG ⇐ 0 There is a lot of power system dynamic simulation software available on the market today. Therefore the use of this software is preferred when many simulation scenarios of DG implementation need to be run in a large power system like the simulations in this work in Chapters 4 and 5. Therefore the modeling of DG units in details as Voltage Source Converter or . Further. MATLAB Power Systems Blockset and RTDS (which is not really software. the packages that contain more detailed and higher order equipment models should be used. From the load ﬂow calculation. Qnom . as no diﬀerential equations are associated with the network and fewer with generating equipment and as it enables the use of a larger simulation time step [34]. power system dynamics simulation software can be used when the phenomena of interest have a frequency of about 1 to 10 Hz. it reduces the number of diﬀerential equations. When the frequencies of interest are higher. only the fundamental harmonic component is simulated.4 DG and Large System Dynamic Simulation Algorithm 3 Calculate IDG Require: Pnom . The capabilities of these three tools complement one another. whereas higher harmonics are neglected. power system dynamics simulation software alternately executes a load ﬂow and a timedomain calculation. This approach enables the representation of the network by a constant admittance matrix. via integration of the diﬀerential equations that model dynamic system devices. Umin2 . In this type of software. Here. During the timedomain calculation. the converter connected DG is modeled as a constant PQSource. Within this work. but a “hardware digital simulator”). as in load ﬂow calculations. MATLAB Power Systems Blockset and the Real Time Digital Simulation (RTDS) fall within this category. As a result of this approach. Umin1 . the transmission network is represented in threephase timedomain and a time step down to 1 µs is possible. three power system software packages are utilized: PSS/E. the response of the dynamic device models to changes in their terminal voltage.
such as the 380kV Dutch High Voltage Grid. Table 3. 46 transmission lines).7 Mvar . The system is relatively small (39 bus.1). The RTDS models of the voltage and current source converters that represent DG in that chapter are derived from the model earlier developed in MATLAB Power Systems Blockset [42]. as it has been used extensively and described in literature.3 Mvar 6097. 3. In the (rotor angle) transient stability studies performed in Chapters 46. 10 centralized generators (CG). Therefore. As such. and the results from the simulations presented throughout this work can (to some extent) be compared with other work. this system has a beneﬁt when compared to other systems. 27 transmission lines (380 kV and 220 kV)) [73]. Particularly in Chapter 7 RTDS is used. the high voltage 380kV Dutch network. This model should be detailed enough to assess transient stability. this level of detail is not needed. Here the second option is taken. especially when simulation is done on large power systems.Stability of Systems with DG 43 Current Source Converter is only possible when such software is used.1: Characteristics of the New England Test System System Characteristic # of buses # of generators # of loads # of transmission lines Total generation Total generation Value 39 10 19 46 6140. a model of the transmission system is needed.5 Simulation Setup This work deals with issues of DG aﬀecting the bulk transmission system when a high DG penetration level is seen. For this purpose. 19 CG (≥ 250 MW). and power system dynamics software package PSS/E is used. 3. The second possibility is using an existing test system normally used for dynamic stability.7 MW / 1264.5. for example. the phenomena of interest have a frequency of between 1 to 10 Hz. The ﬁrst possibility is using a real national transmission system.7 and Table 3.1 The IEEE 39bus New England Test System The IEEE 39bus New England is a widely known test system used for dynamic simulations (Figure 3. However this small step will have an impact on the execution speed. At this level there are two possible data sets to use. but comparable to the 380kV Dutch Network (28 bus.1 MW / 1408.
In this work. simulated by means of a third order induction generator model [53]. the basic parameters of the IEEE 39bus New England dynamic test system are taken from [49]. namely: • Squirrel cage induction generator (ASM).2 DG Technology Two basic technologies can be used for connecting DG to the grid (Section 3. and listed in Appendix B. To represent a system with limited inertia. or (ii) converter connected generator indirectly connected to the grid. and attached in Appendix B. Each load is equally divided in constant impedance. some options elaborated from both basic models are used to investigate the general impact of the DG on the system stability. no bus is modeled as an inﬁnite bus (every generator in this test system is set to have its inertia).1): (i) electrical rotating machine connected directly to power grid. Note that some minor adjustments have been made.44 3. 3. constant power and constant current.4.5 Simulation Setup Figure 3.5. . Each CG is modeled as a twoaxis (dq) model of synchronous machine [53] and is equipped with a simpliﬁed excitation system model and a steam turbine governor model. The details of these models and the representative values for the parameters are taken from [53] and [34].7: Singleline diagram of the 39bus New England test system [49] In this research.
Stability of Systems with DG
45
• Synchronous generator without grid voltage and frequency control (SM), as discussed in [71]. • Synchronous generator with grid voltage and frequency control (SMC) [71]. • Power electronic converter (PE), modeled as a constant PQsource [71]. • Power electronic converter (modeled as a constant PQsource), with grid voltage and frequency control (PEC) [71]. In case of ASM and SM , the electrical rotating machine is modeled based on the existing models available in the power system dynamic simulation software (Appendix C). Furthermore, PE is modeled by a source of constant active power and reactive power (PQSource). Since a standard model is not available for representing power electronics in the version of PSS/E version used [53], a socalled userwritten model of a power electronic converter has been developed. This model is based on the algorithms 2 and 3 in Section 3.4.1. Controllers are applied when DG with grid voltage and frequency control is simulated (SMC and PEC). When SMC is considered, each DG is equipped with the simpliﬁed excitation system and a steam turbine governor model, as listed in Appendix C. When PEC is simulated, the active and reactive power controllers as depicted in Figure 3.8 are incorporated in the model. The challenge lies in the possibilities that many types and sizes of DG are incorporated in the system so that the generators as well and the controller parameters can be optimized. In this case, it is very complex to optimize the controllers parameters for the DG unit. Therefore simpliﬁcations are done as only one generic model is used to simulate one DG technology. Furthermore, once the types and the parameters of the controllers are set, they will neither be changed nor optimized throughout the simulations. This issue is also discussed in next section.
4UHI 8WUHI 8W
3UHI
UHI
.S
7LV
4
4PD[ 4
4PLQ
.S
3
3PD[ 3
3PLQ
Figure 3.8: Reactive (above) and active (below) power controllers of a power electronic converter with grid voltage and frequency control in the test system
46
3.5 Simulation Setup
3.5.3
Incorporation of DG in Distribution Networks
In test systems used for transient stability studies, the highvoltage transmission system is normally modeled in detail, while the distribution system is represented only by the load at the main connections between the transmission system and the distribution systems. As described in Section 1.3, the structure of distribution systems is likely to change due to the implementation of DG. This is because, in an active distribution network, some generation and energy storage systems are located within the distribution system. Therefore, the most thorough approach to study the impact of such DG implementation on the transient stability is to model the distribution networks in detail along with all load types, distributed generators and energy storage. In many situations, however, it is not practical to obtain and to apply these details in the lower level distribution networks on a test system originally used to study the transmission level. Therefore, a simpliﬁcation must be made using a distribution network model representing the aggregated load and generation in the system. In this work a simple approach is used by representing the incorporation of DG in a distribution network as an equivalent load and generator earlier proposed in [18] and [71]. Only one DG technology (and not a mixture of several DG technologies) is assumed to be implemented at one load bus for one simulation case. After that, comparison of the impacts of one DG technology on the transmission system transient stability to another are investigated. In this way, the general model of connecting DG at a particular load bus as shown in Figure 3.9 can be used [18], [71].
/RDG %XV M;7'* 5/'* M;/'* '* '* %XV
3/RDG M4/RDG
Figure 3.9: Model of connecting DG at a particular load bus In this model the impact of diﬀerent DG topology can be approached by changing the parameters of the impedance connecting DG (XT,DG , RL,DG , XL,DG ) to the transmission system. XT,DG and XL,DG represent the reactance of the transformer and the line between the DG and the transmission network respectively, and RL,DG represents the resistance of the line. For example, to compare the impacts of DG implemented far away from or close to the connection feeder, the impedance parameters can be set by increasing the parameters in the ﬁrst case and reducing them for the latter. Also, when a distribution
Stability of Systems with DG
47
network with a singlepoint concentrated DG is compared to a distribution network with scattered DG (that suggests a parallel connection), the impedance can be set diﬀerently (e.g. the line parameters RL,DG , XL,DG ). These issues are treated in Section 4.2.
3.5.4
Behavior of Centralized Power Plants
Synchronous generators within centralized power plants that deliver the electrical energy within a power system are equipped with controllers (governing systems  governors, excitation systems  exciters) and protection schemes. The governing systems for example provide a means of controlling power and frequency, a function commonly referred to as loadfrequency control or automatic generation control. The excitation systems provide direct current to the synchronous machine ﬁeld winding, in order to control the ﬁeld current. The protective function ensures that the physical (mechanical and electrical) capability limit of the synchronous machine, excitation system and other equipment are not exceeded [34]. The characteristics of these control systems of course have an impact on the overall performance of the power system. Basically the eﬀectiveness of the control systems in enhancing power system stability can be estimated and further optimized. For example, by adjusting the parameters of these controllers the behavior of the power system transient stability can be inﬂuenced [34]. When DG is implemented on a large scale it can be expected that the transient performance of the power system changes. It can be expected too that changing, for instance, the parameters of the controllers could counteract the inﬂuence on the performance of the system due to this DG implementation. This suggests that these parameters can be optimized for having the most optimal system transient stability performance. However, as the focus of this work is to investigate stability of the power system due to DG implementation (‘vertical to horizontal’ transformation of power system), we do not include the optimization of these controllers.
3.6
Summary
This chapter gives an overview of the power system stability phenomena. This chapter begins with classiﬁcation of power system stability, which is taken from existing literature [35]. Then, review of existing knowledge is presented to provide insights into the mechanism, which leads to transient instability in power systems. Highlights are put on the rotor angle stability (of synchronous machines), which is the main focus of this research, by means of the swing equation, powerangle equation and equal area criterion. Attention is drawn on some of the existing system stability indicators (Critical Clearing Time (CCT) and Critical Clearing Angle (CCA)), and on more practical transient system indicators such as maximal rotor speed deviation and oscillation duration proposed
diﬀerent protection schemes of converter connected DG are added to the basic algorithm. The basic algorithm for simulation of converter connected DG as constant PQsource is ﬁrst introduced in [71]. for which.48 3. userwritten model with diﬀerent protection schemes are developed. Models and assumptions of DG implementation on (large) power system transient simulation are discussed.6 Summary by [71]. The options are: (1) squirrel cage induction generator. the chapter ends with highlights on the behavior and assumptions of the centralized power plants that are potentially impacted due the DG implementation. The representation and aggregation of DG in the distribution network [18] is discussed in this chapter. leading to the use of existing models of DG as electrical rotating machines and constant PQsource for converter connected DG. Finally. In this thesis. The latter is used to assess power system stability performance on large power systems in this work. This chapter also covers the development of the simulation setup used in this work: IEEE 39bus New England test system [49]. simulated by means of a third order induction generator model. and the motivations that lead to the use of power system dynamics software package PSS/E in this research are presented. . Five DG connection technology options are taken into account based on the two basic alternatives: DG directly or indirectly connected to the grid. (2) synchronous generator without or (3) with grid voltage and frequency control models derived from electricalrotatingmachinebased DG technology (an existing library in PSS/E [53]) and (4) power electronic converter [71] modeled as a constant PQsource without or (5) with grid voltage and frequency control. Some power system dynamics software packages are highlighted.
Chapter 4
Impact of DG on Power System Transient Stability
This chapter investigates the impact of DG implementation on power system transient stability. Diﬀerent scenarios of a power system with a high DG penetration level are developed. The impact of diﬀerent DG penetration levels, fault durations and locations, DG gridconnectionstrengths, DG technologies and protection schemes of powerelectronic interfaced DG units are investigated and discussed. The 39bus New England Test System forms the starting point of the investigations. Throughout the chapter, DG technologies and DG topology are based on the simulation setup deﬁned in Section 3.5.
4.1
4.1.1
DG Impacts
Simulation Scenarios
In this section, the impact of the DG penetration level on the power system stability is investigated. For this purpose, two simulation scenarios are deﬁned. In the ﬁrst scenario, Scenario I, the DG is implemented to cover the increment of the load, so that the DG penetration level rises along with the increasing load. In the second, Scenario II, DG supplies (part of) the existing (constant) load. As a result, the total power output from the centralized generators is reduced. The DG penetration level is deﬁned as [18] %DGpenetrationlevel = PDG × 100, PDG + PCG (4.1)
with PDG and PCG the total active power generated by DG and CG respectively.
50
4.1 DG Impacts The details of the scenarios are as follows: Scenario I: • The DG penetration level is raised to keep track with the increasing real and reactive power consumption of all loads. The increment of the real power consumption is covered by an equal amount of power produced by DG connected to each load bus via a j0.05 pu impedance (representing XT + XL in Figure 3.9 with RL neglected) on the 100 MVA system base. The DG penetration level increases in steps of 3.33 % up to 33.33%, corresponding to a 50% increment of the load i.e. the load increase within 25 years at a load growth of about 1.8%. Thus eleven subscenarios are obtained with DG penetration levels of 0.0, 3.33, 6.67, 10.0, 13.33, 16.67, 20.0, 23.33, 26.67, 30.0, and 33.33%. • The active power generated by the CG is kept constant, except for the active power generated by the generator that acts as the swing bus (generator nr. 2, see Figure 3.7) for covering the losses. The increasing reactive power consumption is provided by centralized generators. Scenario II: • The DG (penetration) level is raised by decreasing the CG active power output in steps of 3.33 % down to a reduction of 33.33%, and the implementation of DG at every load bus to cover this decrement of active power. In this way, again eleven scenarios are obtained. • The load remains constant in this scenario.
To assess the results of Scenarios I and II, Base Case I and Base Case II are additionally deﬁned as: Base case I: • The load is increased in steps similar to that in Scenario I. However, the increasing load is supplied by raising the active power output of the CG. Thus in this case the centralized generators cover both active and reactive power increment. Base case II: • No increase in load similar to Scenario II and no DG penetration.
4.1.2
Transient Stability Simulation
The transient stability of the test system is investigated by applying a permanent threephase fault to all possible branches cleared by tripping the faulty line after 150 ms. Every line, that can be missed according the (N −1) adequacy standard, is subjected to a fault. In this way, 35 possible locations for faulty branches are simulated. Details are shown in Table 4.1. To assess the transmission system stability, two transient stability indicators are examined: • Maximum rotor speed deviation of large centralized generators. • Oscillation duration. Details of the indicators can be found in Section 3.3.
Impact of DG on Power System Transient Stability
51
Table 4.1: Branch Number and Corresponding Buses
Faulty branch (nr.) 1 2 3 4 5 6 7 8 9 10 11 12 Corresponding buses 12 23 225 34 318 45 414 56 58 67 611 78 Faulty branch (nr.) 13 14 15 16 17 18 19 20 21 22 23 24 Corresponding buses 89 1011 1013 1112 1213 1314 1415 1516 1617 1621 1624 1718 Faulty branch (nr.) 25 26 27 28 29 30 31 32 33 34 35 Corresponding buses 1727 2122 2223 2324 2526 2627 2628 2629 2829 139 939
4.1.3
Simulation Results
Scenario I Figure 4.1 and Figure 4.2 show the simulation results displaying the transient stability indicators when the DG penetration level is increased according to Scenario I; the results of Base case I serve as a reference. The xaxis of each graph represents the number of the faulty branch (Table 4.1). The yaxis represents the DG penetration level [%] and the corresponding load increase [%]. Note that in Base case I the yaxis represents the load increase [%] only since there is no DG implemented (DGlevel = 0%). The zaxis represents the value of the stability indicator used (the maximum rotor speed deviation in Figure 4.1, and the oscillation duration in Figure 4.2). The bottom graphs of Figure 4.1 and 4.2, labeled CG, indicate the simulation results of Base case I. The titles ASM (squirrelcage induction generator), SM (synchronous generator without grid voltage and frequency control), or PE (powerelectronic converter without grid voltage and frequency control) above the graphs indicate the type of DG technology simulated (Section 3.5.2). When the increasing load within the test system is covered only by increasing the CG active power output (Base case I), the stability indicators are generally increasing (i.e. a reduced stability in the system). When DG is implemented to cover the increased load within the system (Scenario I), in general the indicators do not increase (see graphs ’ASM’, ’SM’). Some exceptions can be observed when powerelectronic interfaced DG (without grid voltage and frequency control) is implemented (see graph ’PE’), but those can be prevented when the DG is equipped with grid voltage and frequency control. In general, the indicators show an improved stability (i.e. decreasing indicators) when the DG is equipped with grid voltage and frequency control. These results are not visualized here.
7 (20) 0 (0) DG level (load increase) [%] CG max.01 0. dev.01 0.3 (50) 16. dev.005 0 0 (50) 0 (20) 0 (0) 1 5 10 15 20 25 4.005 0 33.7 (20) 0 (0) DG level (load increase) [%] PE max.1 DG Impacts 1 5 10 15 20 25 30 35 Faulty branch [nr.7 (20) 0 (0) DG level (Load increase) [%] SM max.] 30 35 DG level (Load increase) [%] Faulty branch [nr.005 0 33. dev. [pu] 0. and a fault is simulated in all deﬁned branches (Table 4.1) .3 (50) 16. [pu] 0.1: Maximum rotor speed deviation when the DG penetration level is simulated according to Scenario I (graphs ’ASM’. [pu] 0. [pu] 0.01 0.] Figure 4.] 1 5 10 15 20 25 30 35 Faulty branch [nr.] 1 5 10 15 20 25 30 35 Faulty branch [nr. ’SM’ and ’PE’) and the Base case I (graph ’CG’).005 0 33.52 ASM max.3 (50) 16. dev.01 0.
7 (20) 0 (0) DG level (Load increase) [%] CG 20 10 0 0 (50) 0 (20) 0 (0) 1 5 10 15 20 25 30 35 35 35 35 53 osc. [s] DG level (Load increase) [%] Faulty branch [nr. [s] 1 5 10 15 20 25 30 Faulty branch [nr.] osc.] Figure 4.] osc.3 (50) 16.1) .3 (50) 16.Impact of DG on Power System Transient Stability ASM 20 10 0 33. [s] 1 5 10 15 20 25 30 Faulty branch [nr.2: Oscillation duration when the DG penetration level is simulated according to Scenario I (graphs ’ASM’.7 (20) 0 (0) DG level (Load increase) [%] SM 20 10 0 33.7 (20) 0 (0) DG level (Load increase) [%] PE 20 10 0 33. [s] 1 5 10 15 20 25 30 Faulty branch [nr.] osc.3 (50) 16. and a fault is simulated in all deﬁned branches (Table 4. ’SM’ and ’PE’) and the Base case I (graph ’CG’).
all active and reactive power controls as well as the inertia of the centralized generators are unchanged. Thus. The active power ﬂows in the transmission lines even decrease when the DG covers (part of) the existing load (Scenario II). The zaxis represents the value of the maximum rotor speed deviation in per unit. Therefore.6 show some examples of these comparisons. the active power ﬂows in the transmission lines are lower when DG is implemented as the active power generated by the DG is consumed directly by the load at the same feeder. We can compare the active power ﬂows in each of the simulated branches according to simulation Scenarios I. . The xaxis of each graph represents the number of the faulty branch. It can be observed that the ’surface’ of the branch power ﬂows are comparable with that of the system indicators of the scenarios. Note that in these scenarios all centralized generators (the 10 CGs in the test system) remain in the system when the DG penetration level is increased.1.1 DG Impacts Figure 4. Thus.3 (labeled ’CG’) indicates the Base case II. The results can be explained as follows.3 displays the maximum rotor speed deviation when the DG penetration level is increased according to Scenario II. and the DG is implemented close to the loads. the active power ﬂows on the transmission lines are more or less constant when the DG covers the increasing load (Scenario I). when all CG units remain in the system. Both indicators show even better results when the DG is equipped with grid voltage and frequency control. 4. In general the maximum rotor speed deviation decreases. It is known that large power ﬂows have a detrimental eﬀect on the damping of oscillations [31]: the heavier the lines are loaded. The yaxis represents the DG penetration level [%] and the corresponding load increase [%].4 Remarks In the simulation Scenarios I and II.33% regardless of the DG technology used.54 Scenario II 4. In Scenarios I and II. The bottom graph of Figure 4. the weaker the connections between the generators and the loads and the larger the oscillations of the centralized generators. all centralized generators remain in the system.44. II and also the Base cases I and II. However. Figures 4. the DG implementation is a kind of ‘loadreduction’ that reduces the power ﬂows in the transmission network and improves its transient stability. The oscillation duration decreases along with the increasing DG level too (not visualized here). no signiﬁcant (transient) stability problems were found when the DG penetration level is increased up to 33. with the system indicators.
01 0.3 (0) 16.005 0 33. [pu] 0.01 0. dev.3: Maximum rotor speed deviation when the DG penetration level is simulated according to Scenario II and the Base case II (graph ‘CG’). dev.7 (0) 0 (0) 1 5 10 15 20 25 30 35 55 DG level (Load increase) [%] SM max.] 15 20 25 30 35 DG level (Load increase) [%] CG max. dev.005 0 33. [pu] 0. dev. [pu] 0.7 (0) 0 (0) 1 5 10 Faulty branch [nr.005 0 33.] 15 20 25 30 35 DG level (Load increase) [%] PE max.01 0.] Figure 4.1) .01 0.005 0 0 (0) 0 (0) 0 (0) 1 5 10 Faulty branch [nr.Impact of DG on Power System Transient Stability ASM max.] 15 20 25 30 35 DG level (Load increase) [%] Faulty branch [nr. and a fault is simulated in all deﬁned branches (Table 4.3 (0) 16.7 (0) 0 (0) 1 5 10 Faulty branch [nr. [pu] 0.3 (0) 16.
the DG gridconnection strength is varied by changing the parameters of the impedance connecting the DG units to the system. the distribution network in a rural area with long laterals.1 Distribution Network and DG Layout The distribution system carries the power to the individual customers.2 DG GridConnection Strength Impacts Base case I Active Power [MW] 1000 500 0 0 (50) 0 (20) 1 5 10 15 20 25 30 35 0 (0) DG level (Load increase) [%] Branch [nr. DG may be implemented in a distribution network with a relatively low impedance. From the transmission system point of view.005 0 0 (50) 0 (20) 0 (0) 1 5 10 15 20 25 30 35 DG level (Load increase) [%] Faulty branch [nr. [pu] 0.g. both elements may result in a DG implementation that can have a weaker or a stronger grid connection even when in both situations the DG penetration levels are equal. Moreover. 4.56 4. e.] CG max.] Figure 4. as suggested in Section 3.5. Both the geographical situation and the distribution of customers can vary from one area to another [21]. e. For example. distribution networks in towns or city centers. DG units may also be spread throughout a distribution network. in a certain geographical area.2 DG GridConnection Strength Impacts In this section.g. there are many ways to connect DG to the distribution network.2. such as solar panels mounted on the .01 0.3.4: Active power ﬂows in the simulated branches in the test system (top) and maximum rotor speed deviation (bottom) in Base case I 4. or a distribution network with a relatively high impedance. dev.
e. 4. The impedance between the aggregated DG (implemented at a particular load bus) and the transmission network is set according to ZDGi = jXT. .] Figure 4.5: Active power ﬂows in the simulated branches in the test system (top) and maximum rotor speed deviation (bottom) in simulation Scenario I (’SM’ refers to the DG technology implemented) roofs of houses.2) where ZDGi is the impedance between DGi (the aggregate DG connected to load busi) and the transmission network. The latter case may result in a higher impedance between DG and the transmission network. a wind park connected to a distribution network at one substation only.DGi + RL. [pu] 0.7 (20) 0 (0) DG level (Load increase) [%] 1 5 10 15 20 25 30 Branch [nr.7 (20) 0 (0) DG level (load increase) [%] 35 1 5 10 15 20 25 30 Faulty branch [nr. (4.3 (50) 16.] SM max.DGi .2 Simulation Scenarios The incorporation of DG in a distribution network is represented as an equivalent load and generator (Figure 3.9).g. dev.DGi is varied.005 0 33.01 0.DGi + jXL.2.Impact of DG on Power System Transient Stability Scenario I Active Power [MW] 1000 500 0 35 57 33.2) are varied considering the following points: • When DG is implemented in laterals of diﬀerent length the value of jXL. or concentrated in a few locations. The parameters in (4.3 (50) 16.
DGi is taken into account too.2). and IVB) are deﬁned (Table 4. IVA. . par(PDGi ) (4.3) where par(PDGi ) is an integer whose value is proportional to the ”size of DGi ” (the active power generated by the aggregate DG at a particular load busi). To investigate the transmission system stability in relation to the DG grid connection strength.05pu . IIIB.005 0 33.01 0. • When DG is implemented in a distribution network with a high resistance.DGi = j0.DGi is varied related to the “size of DG” implemented in a particular distribution network: jXL. various Scenarios (IIIA.3 (0) 16. the value of jXL. the value of RL.] SM max.2 DG GridConnection Strength Impacts Scenario II Active Power [MW] 1000 500 0 0 (0) 0 (0) DG level [%] 0 (0) 1 5 10 15 20 25 30 35 Branch [nr. dev. [pu] 0.7 (0) 0 (0) 1 5 10 15 20 25 30 35 DG level (Load increase) [%] Faulty branch [nr.] Figure 4.6: Active power ﬂows in the simulated branches in the test system (top) and the maximum rotor speed deviation (bottom) in simulation Scenario II (’SM’ refers to the DG technology implemented) • When DG is spread out over many substations instead of concentrated in one.58 4.
in steps of 0. resistance included Constant length.DGi is neglected.2: Simulation Scenarios Scenario Keywords Impedance (pu) (ZDGi ) {0.08} 1 j0.01 + j0.2) 0 j0. Accordingly.DGi in Scenario IIIB is obtained by substituting the values of ZDGi .DGi +jXL.DGi  in (4.DGi in this table that correspond to Scenario IIIB into equation (4.08} Resistance (pu) (RL.1).005 pu.DGi is kept constant at j0.DGi is kept constant at j0.01 pu) in such a way that the value ZT N.08 pu.035 · · · 0.Impact of DG on Power System Transient Stability 59 Table 4.07} DG penetration level (%) 20 IIIA Laterals diﬀerent length.005 pu while the value of RL.01 pu) in such a way that the value of ZT N.DGi ) 0 Reactance (pu) (jXT .DGi 20 IVA 0.2)) increases from 0.01 3 +XL. 3 .DGi and RL.3} IVB Agrees to equation (4.05 . par(PDGi ) {3. the values of XL.33 · · · 33.33 · · · 33. concentrated Constant length. when DG is implemented in laterals of diﬀerent length.035 · · · 0.01 + j0. while in Scenario IIIB it is taken into account.DGi (jXT.2). ten subscenarios are derived from Scenario IIIB by adjusting the values of jXL. ten subscenarios are derived from Scenario IIIA by adjusting the value of jXL. In this way.0 Note that XL. in steps of 0.01 + jXL. spread out over laterals IIIB {0.DGi  (in (4. RL.DGi and RL.01 +{j0.035 pu to 0.DGi +XT. In these scenarios.DGi ).DGi increase along with the increasing value of ZDGi Scenario IIIA and Scenario IIIB are deﬁned to investigate the impact of the DG gridconnection strength on the transmission system stability. a 20% DG penetration level is assumed (4.DGi (jXT.035 pu to 0.08 pu.DGi equals 1 of (XL. Furthermore.DGi ) j0.DGi  j0. resistance neglected Laterals diﬀerent length.3} par(PDGi ) = the integer PDGi of 100.025 · · · j0.06 0 j0.05 {3. In Scenario IIIA the value of RL.DGi and XT .2) raises from 0.
4. In Scenario IVB.DGi is neglected).01 pu and jXL.7 shows an example of the slightly increasing value for the transient stability indicators along with the increasing impedance connecting DG according to Scenario IIIA. the maximum rotor speed deviation and the oscillation duration are investigated according to Section 3. The parameters in (4. the zaxis represents ’relative’ values of the indicators used: the diﬀerence between the resulting maximum rotor speed deviation and the base case value and the diﬀerence between the resulting oscillation duration and the base case value. In Figure 4. when DG is implemented either concentrated in one substation (Scenario IVA) or spread out over many substations (Scenario IVB). so that the parameters of (4.3 Transient Stability Simulation The transient stability of the test system is investigated by applying a permanent threephase fault. To assess the transmission system stability. and the lower the impedance between DG and the transmission network becomes. it is implemented in one particular point in the distribution network regardless the ”size of DG”. Thus. For a better overview. regardless of the DG technology implemented.3. the xaxis represents the number of the branch where a fault is applied.60 4. increasing the impedance of the connection of the DG to the transmission system generally results in a slightly less stable transmission system.DGi = j0.1). the more spread out the locations of the DG units are. Ten subscenarios are derived from both Scenario IVA and Scenario IVB by varying the DG penetration level in the test system from 3. when the DG is implemented as synchronous machines without grid voltage and frequency control (SM).33% to 33.4 Simulation Results Scenario III With the DG penetration level ﬁxed at 20%.1). every 100 MW of aggregate DG is implemented in a diﬀerent lateral.DGi = j0. and the zaxis represents the stability indicator. RL.2. The base case corresponds to subscenario nr.05 pu (RL. In Scenario IVA it is assumed that when DG is implemented at load busi (DGi ).3) with par(PDGi ) = PDGi 100. The yaxis represents the ten subscenarios with the increasing impedance value of the DG connection as deﬁned in Scenario IIIA. it is assumed that DG is implemented in a distribution network with radial laterals.2 DG GridConnection Strength Impacts Scenario IVA and Scenario IVB are deﬁned to investigate the impact of the DG grid connection strength on the system stability. 4.2) are constant: jXT. to all possible branches taken into account in the N − 1 adequacy standard (Table 4.DGi is neglected and jXL.33% (4.0 (MW) rounded to the next larger integer.33% in steps of 3.7. Figure 4. . 1.01 pu.DGi is set according to (4.DGi = j0.2.2) are jXT. it is assumed that the (MW) larger the size of the aggregate DG is. cleared after 150 ms.
when DG is implemented as synchronous machines (SM).] 1 5 10 15 20 25 Faulty branch [nr.e. . [pu] 2 0 −2 10 SM 5 1 DG grid connection scenario [nr.] 30 35 30 35 Figure 4. dev.e.] 1 5 10 15 20 25 Faulty branch [nr. Scenario IV DG that is concentrated in one lateral (according to Scenario IVA) results generally in higher system indicators than when the DG is spread out of several laterals (according to Scenario IVB). [s] 1 0 −1 10 5 1 DG grid connection scenario [nr. Figure 4.8 shows an example of those diﬀerences. similar results are obtained. in case DG is implemented as synchronous machines (SM). Figure 4. By comparing the ’relative’ values of the stability indicators (i. a reduced system stability) than Scenario IIIA. Scenario IIIB gives slightly higher values of the stability indicators (i.] relative osc. oscillation durations (lower graph) when Scenario IIIA is applied and DG is implemented as synchronous machines without grid voltage and frequency control (SM) When the resistive part of the impedance between DG and the transmission system is taken into account (Scenario IIIB). the diﬀerence between the maximum rotor speed deviations of Scenarios IVA and IVB and the diﬀerence between the oscillation durations of Scenarios IVA and IVB) generally positive values are obtained. This can be seen from the positive values that result from the substraction of values of Scenario IIIA from those of Scenario IIIB.Impact of DG on Power System Transient Stability −4 61 x 10 relative max.9 shows an example of those diﬀerences. 1)): maximum rotor speed deviations (upper graph).7: Relative value in system indicators (actual value minus basecase value (subscenario nr.
x 10 relative max. [pu] 5 0 −5 10 −4 SM 5 1 DG grid connection scenario (nr) 1 5 10 15 20 25 Faulty branch (nr. Taking into account the resistance of the distribution system in which the DG is implemented.2.62 4. dur.8: Relative value in system indicators (Scenario IIIB minus Scenario IIIA which means including resistance): maximum rotor speed deviations (upper graph).2 DG GridConnection Strength Impacts 4. higher active power losses occur that must be supplied by the swing generator (one of the CGs) and gives somewhat larger power ﬂows in the transmission lines. oscillation durations (lower graph) when DG is implemented as synchronous machines without grid voltage and frequency control (SM) .) 30 35 relative osc. This observation can be explained as follows. [pu] 1 0 −1 10 5 1 5 10 15 20 25 30 35 DG grid connection scenario (nr) 1 Faulty branch (nr. results in slightly higher stability indicators than when the resistance is neglected.) Figure 4.5 Remarks In this section a general tendency is found that the values of the stability indicators slightly increase when the impedance in between the DG and the transmission system is raised. dev. [69]. When we take the resistance into account. A higher impedance leads to a less stable system [63].
4. the DG technologies and the DG topology are set according to the simulation setup described in Section 3.1 DG penetration level is raised in steps of 3.3.33% and the load increment is adjusted accordingly. [s] 2 0 −2 33 20 DG level [%] 3 1 5 10 15 20 25 30 Faulty branch (nr. .) 35 relative osc.Impact of DG on Power System Transient Stability −4 63 x 10 relative max.) 35 Figure 4.3 DG Penetration Level and Technology Impacts Although the results of Sections 4. [pu] 2 1 0 −1 −2 33 SM 20 DG level [%] 3 1 5 10 15 20 25 30 Faulty branch (nr. some small diﬀerences were observed. oscillation durations (lower graph) when DG is implemented as synchronous machines without grid voltage and frequency control (SM) 4.5. whereas DG penetration levels are varied in the similar way as in Section 4. and the DG penetration level is adjusted accordingly. dev.1 Simulation Scenario The test system. In Section 4.9: Relative value in system indicators (Scenario IVA minus Scenario IVB which means more spreading of DG): maximum rotor speed deviations (upper graph).1 and 4.1. the load is increased in steps of 5%.1 with a minor diﬀerence. In this section. This section investigates the impacts of raising DG penetration levels on the system transient stability with respect to the DG technology.2 show that some impact of raising DG penetration levels on the system stability can be generalized as regardless of the DG technologies implemented.
Furthermore. each of the ﬁve main DG technologies simulated in this section . when DG with induction generator technology is implemented.64 4. Induction Generator Figure 4. increasing the penetration further. and the DG coupled through an uncontrolled or controlled powerelectronic interface . the maximum rotor speed deviation decreases when the penetration level increases from 0% up to 29%. the maximum rotor speed deviations decrease when the penetration level rises from 0% up to 23% and they increase when the penetration goes further up to 33%. carrying 315 MW and 150 Mvar in the prefault scenario with no DG implemented.3. Furthermore.5.7). the oscillation durations tend to decrease. is chosen arbitrarily.3 Simulation Results At a ﬁrst glance. When 150 and 200 ms fault clearing durations are applied. For the three fault durations applied.3 DG Penetration Level and Technology Impacts 4. However. When the maximum rotor speed deviation and the oscillation duration of each large generator are displayed separately (Figure 4. the maximum rotor speed deviation and the oscillation duration are used again. a better insight is obtained.10 shows the worst values of maximum rotor speed deviation (top) and oscillation duration (bottom) for large centralized generators.are commented on separately.2 are simulated together with the eleven penetration levels and the three fault durations. When the duration is 200 ms the oscillation duration tends to increase with increasing DG penetration level. and 200 ms. it can be observed that the maximum rotor speed deviation of each large generator tends to go up or down . The results are shown for three faulttime durations: 100.2 Transient Stability Simulation The transient stability of the test system is investigated by applying a permanent fault to a transmission line. when 100 and 150 ms fault clearing durations are applied. To investigate the results systematically. Three fault durations are simulated: 100.the squirrel cage induction generator.3. the relative changes of the maximum rotor speed deviations (either increasing or decreasing) are within 11% (at 100 ms fault clearing duration) and 14% (at 150 and 200 ms fault clearing durations). and fault duration. 4. leads to growing maximum rotor speed deviations. When a 100 ms fault duration is applied. ﬁve DG technologies as described in Section 3. DG technology. To assess the system stability in these scenarios. the uncontrolled or controlled synchronous generator. The transmission line between buses 15 and 16 (Figure 3. The fault is cleared by tripping the faulty line after a certain fault duration. the simulation results show that the implementation of DG in the test system (along with the increase in the loads) aﬀects the transient stability diﬀerently depending on DG penetration level. up to 33%. Overall. 150 and 200 ms. The eleven penetration level scenarios are extended with three subscenarios based on the three fault durations.11). 150.
Impact of DG on Power System Transient Stability ASM 65 Max. then ascending) are thus made up by two generators: one with an increasing.007 0.50 0 (0) 5 (5) 10 (9) 15 (13) 20 (17) 25 (20) 30 (23) 35 (26) 40 (29) 45 (31) 50 (33) Load increase [%] (DG penetration level [%]) Figure 4. The eﬀect of induction generators on the power system stability depends on their distance to the synchronous generators: • When they are located near the synchronous generators and the latter speed up during a fault. and 200 ms (△) consistently. • When they are at a larger distance and more weakly coupled to the synchronous generator. 150 ( ). which in turn slows down the speeding up of the synchronous generators. A similar tendency is found for the oscillation duration. These results may be explained as follows [47]. .005 0. and fault clearing durations of 100 ( ).50 5.50 6. the stator frequency of the induction generators increases.003 0 (0) 9. This leads to a decrease in the slip frequency and thus in extra generated power.50 5 (5) 10 (9) 15 20 25 30 35 40 45 50 (13) (17) (20) (23) (26) (29) (31) (33) Oscillation duration [s] 8. [71].50 7.009 0. its speeding up during the fault results in an increasing reactive power demand. and the other with a decreasing maximum rotor speed deviation due to the growing DG penetration level. The curves of the maximum rotor speed deviation in Figure 4. This leads to a lower terminal voltage at the far away synchronous generator and thus in a decrease of the synchronizing torque and a faster increase in rotor speed.10 (ﬁrst descending. rotor speed deviation [pu] 0.10: Maximum rotor speed deviation (top) and oscillation duration (bottom) using induction generator (ASM) DG technology.50 4.
. . it is clear that generator 1 shows an increasing maximum rotor speed deviation with growing DG penetration levels.0045 0. While the oscillation duration tends to increase in case of uncontrolled synchronous DG. However. G2 = dash△.5 3..66 4. G5 = solid♦. G4 = solid.11.. generator 1 is not located close to DG . Synchronous Generator The worst values of the transient stability indicators for the large centralized generators. G9 = dash.0035 0.7) is considered. G10 = dash♦ As an example. From Figure 4. G7 = solid△. 1 (G1) = dash. rotor speed deviation [pu] 0.0065 0.e.0025 0 (0) 5 (5) 10 (9) 15 20 25 30 35 40 45 50 (13) (17) (20) (23) (26) (29) (31) (33) Oscillation duration [s] 6.0055 0..12. the synchronous generator 1 at bus 39 (Figure 3.5 5. in case that DG is implemented using uncontrolled or controlled synchronous generator technologies are displayed in Figure 4. G8 = dash×. Generator nr. Since at buses 1.5 2.this result is what could be expected: a faster increase in rotor speed with growing DG penetration levels.5 0 (0) 5 (5) 10 (9) 15 (13) 20 25 30 35 (17) (20) (23) (26) Load increase [%] (DG penetration level [%]) 40 (29) 45 (31) 50 (33) Figure 4. 2 and 9 no DG is implemented . G3 = solid×.3 DG Penetration Level and Technology Impacts ASM Max.11: Maximum rotor speed deviation (top) and oscillation duration (bottom) for each large centralized generator when induction generator (ASM) DG technology is implemented and a 150 ms fault clearing duration is applied.5 4. the oscillation duration is diﬀerent when these two technologies are applied. it tends to decrease in case of controlled synchronous DG. The maximum rotor speed deviation consistently decreases when DG penetration level is increased. G6 = solid.i.
rotor speed deviation [pu] 0. (dashed lines) synchronous generator technologies are implemented and fault clearing durations of 100 ( ). and 200 ms (△) are applied The results can be explained as follows. PowerElectronic Converter When DG is coupled through uncontrolled or controlled powerelectronic converters. SMC. Along with increasing DG penetration levels. (solid lines) and controlled. Both uncontrolled and controlled distributed synchronous generators are equipped with an excitation winding on the rotor.Impact of DG on Power System Transient Stability SM 67 Max.005 0. Thus during the fault.003 0 (0) 8. the maximum rotor speed deviation of the synchronous generators decreases consistently with increasing DG penetration levels (Figure 4. the distributed generators supply the fault current and the voltage drop during the fault is not as severe as is the case without DG. The oscillation duration shows diﬀerent results. However. When a fault occurs.13). . a higher DG penetration level results in a higher terminal voltage and less overspeed.50 0 (0) 5 (5) 10 (9) 15 20 25 30 35 (13) (17) (20) (23) (26) Load increase [%] (DG penetration level [%]) 40 (29) 45 (31) 50 (33) Figure 4. the oscillation duration of the synchronous generators tends to increase when DG is coupled through uncontrolled powerelectronic converters.50 4.50 5.007 0.50 6.009 0.12: Maximum rotor speed deviation (top) and oscillation duration (bottom) when DG using uncontrolled. it tends to decrease slightly if DG is coupled through controlled powerelectronic converters. SM. keeping the generators excited during the fault. 150 ( ).50 5 (5) 10 (9) 15 20 25 30 35 40 45 50 (13) (17) (20) (23) (26) (29) (31) (33) Oscillation duration [s] 7.
The diﬀerent results found for the oscillation durations are caused by the frequency controller actions of the controlled powerelectronic converter. rotor speed deviation [pu] 0. As the same protection schemes are applied for both controlled and uncontrolled powerelectronic converters.00 Oscillation duration [s] 9. e.00 5.13: Maximum rotor speed deviation (top) and oscillation duration (bottom) when DG is coupled through a powerelectronic interface without (solid lines) or with voltage and frequency control (dashed lines).3 DG Penetration Level and Technology Impacts PE / PEC Max.85 pu. during a fault. . the distributed generators are lost and the rotor acceleration of the large synchronous generators is reduced.0085 0.00 0 (0) 5 (5) 10 (9) 15 20 25 30 35 40 45 50 (13) (17) (20) (23) (26) (29) (31) (33) Load increase [%] (DG penetration level [%]) Figure 4. the maximum rotor speed deviations are the same.00 6.0025 0 (0) 5 (5) 10 (9) 15 20 25 30 35 40 45 50 (13) (17) (20) (23) (26) (29) (31) (33) 10.g. The protection systems applied. 150 ( ). The protection system reconnects DG as soon as a recovery action is taken and when the voltage level increases again and passes 0. and fault clearing durations of 100 ( ).0045 0.00 7. and 200 ms (△) are applied The simulation results can be explained as follows.68 4. When the distributed generators are gridconnected by means of a controlled powerelectronic interface.00 4.85 pu.00 8. the frequency controller damps the rotor oscillations of the centralized generators after the protection system reconnects the DG. Thus. during a fault.0065 0. disconnect the distributed generators connected through a powerelectronic interface as soon as the voltage level drops below 0.
• DG remains connected to the network during a fault.85 pu (Protection I).4.3. the power supplied by the DG drops accordingly.4. The DG connected to each load bus produces an amount of active power equal to the increased real power consumption of the load at that particular bus.4 Remarks The computations in this section show that the impact of DG on the system stability depends on both DG technology and penetration level. although it appears that the oscillation duration tends to decrease with increasing DG penetration levels when a DG technology equipped with controllers (Section 3. due to limitations of the components of a powerelectronic interface. this is only true for generators located in the vicinity of the DG.5.1 (Scenario I. DG connected to the grid via a powerelectronic converter without grid voltage and frequency control (PE) (Sections 3. and reconnects as soon as the voltage recovers and DG that remains connected to the power network during a fault (ridethroughcapability [42]). it is assumed that the current through the powerelectronic interface is limited to a maximum value of 1.Impact of DG on Power System Transient Stability 69 4. in case the voltage level drops during the fault. whereas the current is constant and limited to its maximum value (Protection II). eleven (sub)scenarios). The results for the oscillation duration are ambiguous. when induction generator based DG is implemented.1) is simulated. The DG penetration levels in the test system are varied in the same way as in Section 4. . However.2 pu. This suggests that the performance of the system can be improved by enhancing the control capabilities of the DG technology. However.1 Simulation Scenarios The test system and the DG topology are again set according to the simulation setup described in Section 3.85 pu and reconnected as soon as the voltage level passes 0. The maximum rotor speed deviation for most of the synchronous centralized generators decreases with increasing DG penetration levels for the main DG technologies. the impact of implementing powerelectronic interfaced DG on the system stability is further investigated.4. 4. Two protection schemes for DG are applied: • DG is disconnected from the power network when the voltage level of the system drops below 0. with the focus on the two protection systems: DG that disconnects from the network automatically when the voltage drops below a certain level. either controlled or uncontrolled.4 Protection of PowerElectronics Impacts In this section. Thus. 4.1) is applied.
Protection I and Protection II.5 0 19 1 DG [nr. The case of Protection II results in a slightly diﬀerent value of the corresponding bus voltage.15 time [s] 1 1.] 1 1.33% and the corresponding bus voltages of Protection I case (bottom) .70 4.15 time [s] 1. Yet it is quite comparable to the case of Protection I. and both protection schemes. PE Active Power [pu.5 1 0. The fault occurs between the buses 15 and 16. are applied.] 1.5 0 19 1 DG [nr.5 1.] 1.] 1.] Active Power [pu.15 time [s] 1 1.4 Protection of PowerElectronics Impacts Figure 4.5 0 19 1 DG [nr.14 shows an example of the diﬀerent active power output values of converterconnected DG (PE) in case of Protection I (top) and Protection II (middle).5 1.5 1 0.] Bus Voltage [pu.14: DG active power outputs when implemented via a powerelectronic interface without grid voltage and frequency control (PE) in the case of Protection I (top) or Protection II (middle) at a DG penetration level of 33. The bottom graph shows the voltage of the corresponding bus obtained from the case without voltage and frequency control and Protection I. The xaxis represents the time and the yaxis represents the DG unit. The zaxis.5 Figure 4. represents the value of the active power output of the DG unit (top and middle) and the value of the corresponding bus voltage (bottom).5 1 0.
DG is connected via a powerelectronic interface without grid voltage and frequency control (PE). Pe is lower (accelerating power.4.15). i.] rel. In other words. is larger then) than when the DG disconnects (Protection I) from the system. DG covers part of the power consumption. [s] 5 0 −5 33 20 0 1 5 10 15 20 25 30 35 DG level [%] Faulty branch [nr. x 10 rel. [pu] 5 0 −5 33 20 DG level [%] 0 1 5 10 15 20 25 30 35 −3 PE Faulty branch [nr. when DG disconnects from the system (Protection I).Impact of DG on Power System Transient Stability 71 4.] Figure 4. the CG electrical power output.15: Relative values in maximum rotor speed deviations (top) and relative oscillation durations (bottom). cleared after 150 ms. dur. In case DG remains connected (Protection II) to the system.e.4. to all possible branches.2 Transient Stability Simulation The transient stability of the test system is investigated by applying a permanent threephase fault. Pa . especially when DG is implemented without grid voltage and frequency control (PE) (Figure 4. The maximum rotor speed deviation and the oscillation duration are again applied. osc. dev.3 Simulation Results To investigate the impact of both protection schemes. The values of the indicators in Protection scheme II are generally higher than that of Protection scheme I. As a result. This can be explained as follows. the part of the power consumption that in Protection . 4. the simulation results of both schemes are compared. max. the results of Protection scheme II minus Protection scheme I.
higher generator speeds result (i. and the protection scheme of powerelectronic interfaced DG. It is also found that the values of the stability indicators show a tendency to increase slightly when the impedance between DG and the transmission system is raised.4. In this chapter therefore the DG penetration level on the power system transient stability is analyzed. As an example. it may be important to keep (parts of) the DG connected to the system during a fault in order to supply power to the less involved areas. [31] supports this correlation. as the system stability indicators show less maximum rotor speed deviation and shorter oscillation duration. Several factors are analyzed such as load scenario. 4. a rule based on the voltage level can be used for this purpose. When load is increased and DG is implemented in the distribution grid. taking into account the resistance of the distribution system in which DG is implemented results in slightly higher stability indicators than when the resistance is neglected. the active power ﬂows in the transmission lines decrease when DG covers also part of the existing load. either controlled or uncontrolled. Less power ﬂows should intuitively result in a more stable system. When the penetration level of DG increases. DG grid connection strength. In this case. 4.72 4. the impact of a small amount of DG connected to the grid on the power system transient stability has not been treated so often.5 Conclusions scheme II would be covered by the keptconnected DG is covered by CG. when induction generator based DG is implemented. reduced transient stability) than in case DG disconnects during the fault. from the power supply point of view.4 Remarks This section shows that when the powerelectronic interfaced DG remains connected to the network during the fault. However. In contrast. Pe . but begins to inﬂuence the whole system [76].5 Conclusions It is wellknown that the implementation of DG inﬂuences the technical aspects of the distribution grids. a rule could be applied that prescribes when the DG disconnects from the system or when it remains in the system during a fault. This results in a higher electrical power output. However. its impact is no longer restricted to the distribution network. and in this work. and less accelerating power. the active power generated by the DG is consumed directly by the load at the same feeder and the active power ﬂows on the transmission lines are more or less constant instead of increasing active power ﬂows when CG needs to cover the increased load.e. Furthermore. this is only true for gener . Although a mathematical proof is not established. Pa . However. It is found that the maximum rotor speed deviation for most of the synchronous centralized generators decreases with increasing DG penetration levels for all DG connection technologies. DG technology. and results in this chapter support this as well.
In general. Note that in this chapter.along with the increasing DG penetration levels. In that case. although it appears that the oscillation duration tends to decrease with increasing DG penetration levels when a DG connection technology equipped with controllers is applied.Impact of DG on Power System Transient Stability 73 ators located in the vicinity of the DG. there appear no signiﬁcant stability problems up to the 30% DG penetration level examined. This suggests that the performance of the system can be improved by enhancing the control capabilities of the DG technology. However. It is also found that when the powerelectronic interfaced DG remains connected to the network during and after a fault. . it is important to keep (parts of) the DG connected (ridethroughcapability) to the system during a fault in order not to loose too much power in the system.as well as their active and reactive power controls . a rule could be applied that prescribes when the DG disconnects from the system or when it remains in the system during a fault. Higher DG penetration levels have not been tested in this chapter. from the power supply point of view. all centralized generators remain in the system . for example based on the voltage level to which it is connected [20]. The results for the oscillation duration are ambiguous. higher generator speeds appear than in case DG disconnects during the fault (i. reduced transient stability).e.
5 Conclusions .74 4.
Figure 5. &HQWUDOL]HG JHQHUDWRUV a a a &HQWUDOL]HG JHQHUDWRU a 7UDQVPLVVLRQ QHWZRUN 7UDQVPLVVLRQ QHWZRUN a a a a a a /RDGV DQG 'LVWULEXWHG *HQHUDWRUV D. This results in a gradual transition from the current ’vertical’ into a future ’horizontal’ power system (Section 1.1 illustrates the concept of a ’vertical’ and a ’horizontal’ power system (redeﬁned from Figure 1.3). Eventually.3).Chapter 5 ’Vertical to Horizontal’ Transformation of Power Systems A largescale implementation of DG may be expected in future. the high amount of DG in a power system may cause a number of centralized generators (power plants) to be shut down for eﬃciency (or environmental) reasons.
/RDGV DQG 'LVWULEXWHG *HQHUDWRUV E.
all centralized generators and their corresponding inertias and control functions remain in the system when the DG level is increased. a ’verticaltohorizontal’ transformation of power systems results in . However. Figure 5.1: Illustration of the ’vertical’ (a) and the ’horizontal’ power system (b) In Chapter 4.
1) a modiﬁed version of the deﬁnition of DG penetration level used Chapter 4. All centralized generators are initially dispatchable and have a piecewise linear cost curve [53] fi = ai PCGi + bi . if necessary.1 Simulation Setup less centralized generators in service and. are taken out of service. the impact of such ’verticaltohorizontal’ transformation of power systems on transient stability is investigated.u. reductions of their control functions alongside with the increasing DG level. The transformation from a vertical to a horizontal power system is done in the following way: • Both the active and reactive power of all loads are kept constant.2).5.T otal (5. load modeling and DG topology for simulations in this chapter are set according to the simulation setup deﬁned in Section 3. In this chapter. correspondingly. The main idea is to give a consistent sequence of shutting them down. Simpliﬁcations are assumed in shutting down the centralized generators.2) . • The penetration level of DG is raised by increasing the fraction of the total load in the test system served by DG. (5. • DG is connected to every load bus via a j0. A simple economic dispatch program is used for the remaining units in service.3 through 5.5).5.76 5. Section 3. The results of this chapter show that based on power system transient stability constraints. impedance at the 100 MVA system base. where some CG units. PLoad.05 p.1 Simulation Setup Test system. Moreover. Minimum and maximum loading limits of the CG units are assumed. The fraction of the total load served by DG is distributed among the modeled DG units proportional to the active power consumed by the load at a particular bus. • The remaining power is divided among the (dispatchable) centralized generator (CG) units by considering the economic operation of the power system. 5.T otal is the total amount of active load within the test system and PDG is the total amount of active power generated by DG. a limit of DG penetration level in a power system can be reached. The DG penetration level in the system is deﬁned as %DGlevel = PDG × 100. the use of largescale power electronicinterfaced DG units implies a reduction in the rotating masses (inertia) in the system. DG is implemented as powerelectronic interfaced DG without grid voltage and frequency control (modeled as a constant PQsource. Some remedies to improve the stability of the system that goes through the ‘verticaltohorizontal’ transition are also discussed (Sections 5. PLoad.
A single line diagram of the test system is shown in Figure 3. . A 90% maximum DGlevel is chosen so that in all scenarios at least one centralized generator remains in operation when the optimal power ﬂow is performed. are used to observe the rotor angle stability and to quantify the severity of the rotor . (5.2 pu of the power electronic connected DG. 9. and thus give a consistent sequence of taking out of service centralized generators. (5. .2 illustrates this process. The ﬂowchart of Figure 5. ai (incremental cost) and bi (startup cost) are constants corresponding to the CG unit i. and a ﬁxed shunt device is implemented at the location of the shut down CG to compensate the former’s reactive power generation. with objective function to minimize the fuel cost is then run each time the DG penetration level is raised.4.g. is simulated. during a fault. . the voltage limits with 5% margin are set in this OPF program.g. PCGj is higher than the PCG(j−1) (e. In (5. The protection system applied disconnects the powerelectronic interfaced DG units as soon as the voltage level drops below 0.’Vertical to Horizontal’ Transformation of Power Systems 77 with fi the cost of CG unit i to generate an amount of active power PCGi . two transient stability indicators. the constants ai and bi are deﬁned as ai bi = = αci . As the main constraint. and it reconnects the DG as soon as the voltage level increases again above (or equal to) the 0. and so forth). 10.5) In this way.7. . The most ineﬃcient CG unit whose power output falls below its minimum loading limit. The transient stability of the test system is investigated by applying a permanent fault to the (arbitrarily chosen) transmission line between buses 15 and 16 . and a DG level of 90% is set as maximum. To diﬀerentiate the economic eﬃciency of each CG. In that case. in the test system CG nr.85 pu level (Section 3. This generator is necessary to provide a reference frequency for all power electronic converters of the DG units. 10 is more expensive than CG nr. the ’verticaltohorizontal’ transformation of the power system.3) (5. and ci deﬁned as ci = {c1 + γ(i − 1)i = 2. The stability of these last centralized generators that remain connected determines the stability of the system as a whole. maximum rotor speed deviation and oscillation duration (Section 3. CG nr.1) after the fault is cleared. In this way. the rated current of the DG converter is limited to 1. DG with “ridethrough” capability is applied.3). e. and detailed in Appendix B. the value used for increasing the DG level in each simulation is 5%. In this study.5.4) with α and β constant.85 pu.2). 9 is more expensive than CG nr. An optimal power ﬂow (OPF) program. In Section 5. is taken out of service. It is assumed that the fault is cleared by tripping the faulty line after 100 ms. 8. βci . γ ∈ R+ }. The optimal power ﬂow program is then rerun. To assess the transmission system stability.
2 results in a number of simulations for a power system that goes trough the ’verticaltohorizontal’ transformation. CG units are taken out of service as the DG penetration level increases. The result of this process is illustrated in Figure 5. 5.78 5. In this research. As a result of performing the optimal power ﬂow. each shutdown CG unit is replaced by a ﬁxed (at .3. The transient stability performance of the cases within the verticaltohorizontal transformation scenario is compared based on these indicators.2 Simulation Results Case I (1) Start (2) Take basecase with 0 % DG penetration level (DGlevel) (3) (4) Increase DGlevel by 5% Run Optimal Power Flow Program Yes (6) Shut down CG Any CG to shut down? (5) (7) Implement shunt device at the location of shut down CG Run Optimal Power Flow Program No (8) No Define case with DG penetration level = DGlevel DGlevel = 90% ? (9) (10) Yes Stop (11) Figure 5.2 Simulation Results Case I The approach depicted in Figure 5.2: Flowchart depicting ’verticaltohorizontal’ transformation process speed oscillations of the remaining centralized generators.
1. This is called Case I. so stability decreases. 4 are not clearly visible.3 the output power generated by CG nr. although the indicator values are high compared to the values resulting from 5%45% DG levels. When DG levels are increased even further (80%90%). that up to 45% DG level the indicators do not change much. • The branch ﬂows.3: Dispatched CGs as a function of the DG penetration level (Case I) Figure 5.4 shows both stability indicators as a function of the DG level. The indicators signiﬁcantly increase when the DG level rises beyond this value.’Vertical to Horizontal’ Transformation of Power Systems 79 one value) reactive shunt device (cheaper than the more sophisticated variable switched shunt devices or FACTS devices). With this approach.1.e. in Figure 5. This result can be explained by considering the fact that two competing mechanisms play a role in inﬂuencing the transient stability behavior of the system in this ’verticaltohorizontal’ transformation: • The total inertia (i. 2 and CG nr.4 and Table 5. . stored kinetic energy) within the rotating masses of the machines in the system decreases. At 60%75% DG penetration levels the applied fault causes instability. Note that due to the small values. It can be observed from Figure 5. 7000 6000 Active Power [MW] CG10 CG9 CG8 CG7 CG6 CG5 CG4 CG3 5000 4000 3000 2000 1000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] CG2 CG1 Figure 5. in particular in the faulty branch. decrease so stability increases. the power ﬂow program keeps 4 centralized generators in operation. More detailed values are given in Table 5. the applied fault does no longer cause instability. when a permanent fault to the transmission line between buses 15 and 16 is applied.
0044 0.54 11.16 5.0080 0.57 24.0532 unstable unstable unstable unstable 0.23 6.06 0.05 5.80 Oscillation duration [s] Max.96 6.0034 0.32 5. Note that at DG levels of 60% through 75% the simulation shows that the applied fault causes instability in Case I Table 5.0043 0.4: Transient stability indicators: maximum rotor speed deviation (top) and oscillation duration (bottom) as function of DG penetration level.12 5.2 Simulation Results Case I 0.0638 0.25 5.0042 0. Rotor Speed Deviation (pu) 0.1: Stability indicators as a Function of the DG Level (Case I) DG Level Scenario(%) 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Max.08 0.0035 0.0045 0.04 0.0033 0.12 14.0045 0.11 .0043 0.31 5.0039 0. rotor speed deviation [pu] 5.24 unstable unstable unstable unstable 24.0674 0.0811 Oscillation Duration (s) 5.02 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] 30 20 10 instability instability 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5.22 26.13 5.
we obtain the total stored kinetic energy of the system at synchronous speed as shown in Figure 5.2 36.9 CG10 CG9 CG8 CG7 CG6 CG5 CG4 CG3 CG2 35000 30000 25000 20000 15000 10000 5000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] CG1 Figure 5. we assume that all centralized generators are of the same type and technology.1 16.1 32.1 28.9 15.9 15.0 16.4) and using the dispatched CG values found from the optimal power ﬂow simulation as shown in Figure 5.2 32.1 28.2.1 16.1 24. it increases when the DG level is continuously raised up to 70% and decreases again when the DG level increases further to 90% (Figure 5. particularly the active power ﬂow in the branch between buses 15 and 16 where the fault is applied. The circles (solid line) and their data labels mark the stored kinetic energy (in per unit) at the speed of the moment immediately following fault clearing pf ( nCG Hi (ωi )2 . behaves diﬀerently. pf stands for postfault).6). the plot in Figure 5. The circles (solid line) and their data labels mark the stored kinetic energy at the speed of the moment immediately following fault clearing The sum of active power ﬂows in all grid lines.2 40000 Stored Kinetic Energy [MJ] 36. The inertia constants of all centralized generators in per unit thus are equal.5 is equivalent to the plot of the sum of machine ratings for the generators connected. are shown in Figure 5. the decreasing total inertia of the system (total . The load ﬂow in the test system.2 36. It can be concluded that in the ﬁrst place.0 16.6.1 20. i=1 The total stored kinetic energy of the system decreases consistently as the DG level increases up to 55% and remains constant thereafter because the minimum of four CGs are kept in the system. Using the inertia constant H (3.5. as well as the power ﬂow in the branch between buses 1516. With H the same for all generators. Later.’Vertical to Horizontal’ Transformation of Power Systems 81 In this study. This is due to the combination of implementing DG and taking out of service of CG units during the ’verticaltohorizontal’ transformation of the power system. 40.0 15.1 24.5: Total stored kinetic energy in the system at synchronous speed in Case I. This ﬂow initially decreases when the DG level is raised from 0% to 45%.9 15.
With 80%90% DG levels. it seems that in the case of 55% and 80%90% DG levels.6 the sum of all ﬂows in the system does not experience a monotonic decrease as the DG penetration level goes up.5). after which instability occurs for 60%75% DG levels. The reason is that the total inertia of the system is not changed when the DG levels are increased from 55% to 90% (Figure 5. so that the stability performance of the system with a DG level up to 45% does not vary much. Note that in Figure 5. although the transient indicators for these cases are quite high (Figure 5.6).6: Active power ﬂow in the branch between buses 1516. the system is again stabilized.2 Simulation Results Case I Sum of All Active Power Flows [MW] 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] 10000 5000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5. high(er) transient stability indicators for the cases of 50% and 55% DG level are observed. so that the applied fault does not cause instability (unlike in the case of 60%75% DG levels with the high active power ﬂowing in the faulty branch and the low total system inertia). when the DG level increases further. experiences its maximum at the 70% DG level (Figure 5.4). because the objective function of the OPF is to minimize the fuel cost rather than to minimize the branch ﬂows. the small active power ﬂowing in the branch where the fault is applied ‘compensates’ for the low total inertia of the system. .82 500 Active Power Flow 15−16 [MW] 400 300 200 100 5. Therefore. where the fault is applied in the simulation (top) and the sum of all active power ﬂows in the grid lines of the test system (bottom) in Case I stored kinetic energy in the system) that occurs when the DG level increases from 0% to 45% is ’compensated’ by the decreasing active power ﬂow. However. while the active power ﬂow in the branch where the fault is applied.
It is not obvious that in all cases rescheduling of generation is eﬀective enough. so that the sum of all branch ﬂows and the ﬂow in branch 1516 are more or less minimized. For this purpose.3 Rescheduling Generation Case I Figures 5. Figure 5. with the objective function set to minimize the losses. Therefore.’Vertical to Horizontal’ Transformation of Power Systems 83 5. after the load ﬂow is optimized.8 shows the corresponding sum of all branch ﬂows and the ﬂow in branch 1516. Case I after CG outputs are rescheduled In this section.6 show that a combination of the smaller total system inertia and the increased power ﬂow for 60%75% DG levels (compared to 0%55% and 80%90% DG levels) results in system instability. As may be expected. after the approach depicted in Figure 5. but here it is.4 through 5.2 is performed in which some CG units are taken out of service. the instabilities for cases with 60%70% DG levels are eliminated (Figure 5. 7000 6000 Active Power [MW] 5000 4000 3000 2000 1000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] CG10 CG9 CG8 CG7 CG6 CG5 CG4 CG3 CG2 CG1 Figure 5. Figure 5.7 shows the dispatched CG units as a function of the DG levels after the CG outputs are rescheduled. a logical remedy to eliminate the instability is optimizing the load ﬂow in case of reduction of the total system inertia. an OPF program is run. .7: Dispatched CGs as a function of the DG penetration level.9). the outputs of the remaining centralized generators (CGs) in the system are rescheduled to alleviate instability problems.
once a CG unit is taken out of service. in order to fulﬁll the reactive power demand changes within the system and to hold the voltage in between speciﬁed margins.4 Simulation Results Case II In Section 5.4 Simulation Results Case II 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Sum of All Active Power Flows [MW] 10000 5000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5. the reactive power supplied by the shunt device is adjusted at every step of the gradual transition in such a way that the voltage at the bus is kept between 0. This is called Case II. the topology and reactive power balance of the test system are changed. step No. Figure 5.95–1.12. In Case II. when the DG level increases up to 90%. As a result.2.8: Active power ﬂow in the branch between buses 1516 where the fault is applied in the simulation (top) and the sum of all active power ﬂows in the grid lines of the test system (bottom). When the gradual transition is continued.2. the load ﬂow program can manage up to only 2 generators running (8 CG units are taken out of service). The corresponding inertias of the CG units are shown in Figure 5. To avoid this constraint. In this way. Applying the 100 ms permanent fault clearance on the line between bus 1516 results in the indicators shown in Figure 5.84 500 Active Power Flow 15−16 [MW] 400 300 200 100 5.10 shows the resultant dispatched CG.2. if a CG unit is taken out of service. 7). 7 of the ﬂowchart depicted in Figure 5. unlike the replaced CG units. rescheduling the output power of CG units only improves the indicator at a DG level of 60%. . a variable switched shunt device is used in the process No. Since this ﬁxed switched shunt device has no ﬂexibility of supporting the voltage in the system. instead of using a ﬁxed switched shunt device. in this section.11.05 pu. the power ﬂow program forces the 4 generators to keep running. The triangles (dashed line) mark the results of rescheduling CG outputs and the circles (dotted line) mark the original Case I 5. a ﬁxed switched shunt device is used to replace it and to provide reactive power (Figure 5.
02 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] 30 20 10 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5.10: Dispatched CGs as a function of the DG penetration level when variable. rotor speed deviation [pu] 85 0. The triangles (dashed line) mark the results of rescheduling CG outputs and the circles (dotted line) the original case. Note that by rescheduling CG outputs. instead of ﬁxed.9: Transient stability indicators: maximum rotor speed deviation (top) and oscillation duration (bottom) as a function of the DG penetration level.’Vertical to Horizontal’ Transformation of Power Systems Max.06 0.08 0. the applied fault does not cause instability at DG levels of 60% through 75% 7000 6000 Active Power [MW] 5000 4000 3000 2000 1000 CG3 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] CG2 CG1 CG10 CG9 CG8 CG7 CG6 CG5 CG4 Figure 5.04 0. shunt devices are used (Case II instead of Case I) Oscillation duration [s] .
86 5.1 24. Note that by rescheduling CG outputs.1 28.2 40000 36.1 32. The circles (solid line) and their data labels mark the stored kinetic energy (in per unit) at the speed of the moment immediately following fault clearing Oscillation duration [s] Max. The circles (dotted line) mark Case II and the triangles (dashed line) the results of rescheduling CG output in Case II.11: Total stored kinetic energy in the system at synchronous speed.0 12. rotor speed deviation [pu] 0.2 32.2 36.12: Transient stability indicators: maximum rotor speed deviation (top) and oscillation duration (bottom) as a function of the DG penetration level.0 8.1 28.0 12. the applied fault does not cause instability at DG levels of 60% .4 Simulation Results Case II 40.2 36. as a function of the DG penetration level in Case II.1 16.06 0.02 0 0 20 15 10 5 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] instability limit (original case) instability limit (rescheduling generator) instability limit (original case) instability limit (rescheduling generator) 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5.0 CG10 CG9 CG8 CG7 CG6 CG5 CG4 CG3 CG2 Stored Kinetic Energy [MJ] 35000 30000 25000 20000 15000 10000 5000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] CG1 Figure 5.1 24.0 8.0 8.0 12.1 20.1 16.04 0.
’Vertical to Horizontal’ Transformation of Power Systems 87 As shown in Figure 5. although the load ﬂow still decreases. This result is logical when the inertia decreases. . 500 400 300 200 100 0 0 10000 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Sum of All Active Power Flows [MW] Active Power Flow 15−16 [MW] 5000 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5. the merit of this section is in showing that.13. and identifying if certain unstable situations are correctable by rescheduling the remaining generation. At higher DG level rescheduling CG does not help. the DG level is limited when the system has a total inertia of 30% or less compared to the original one. the load ﬂow in the case of 60% DG level with rescheduling CG units is much improved compared to the original one of 60% DG level. does not explicitly take into account how many CG units still are in operation to provide the inertia or reactive power service to the system. analysis has to be done on the network. This leads to the conclusion that there is a limit to the DG penetration level in power systems for the test system used. Similar remedial actions might be needed as discussed here. This new approach must be based on the stored kinetic energy provided by the system. this result suggests another way of deﬁning the limit of DG penetration level. This requires the speciﬁc sequence of shutting down the CG units. But. Therefore. however. the availability of reactive power support can inﬂuence this limit.13: Active power ﬂow in the faulted branch (top) and sum of all active power ﬂows in the grid lines of the test system (bottom) in Case II. This. Of course all possible faults and branches in the system have to be taken into account in such an analysis. However. The triangles (dashed line) mark the results of rescheduling CG outputs and the circles (dotted line) the original Case II This is interesting as so far the DG penetration level is mostly straightforwardly used without a limit. The limit obtained in this section is sensitive to the network/test system and verticaltohorizontal scenario. Like the example in this section. in order to determine the DG limit for a given system.
14 shows the results of system indicators when ridethrough DG is used. when DG is supporting the voltage.4. in the higher level. This result again proves that the limit of DG penetration level is inﬂuenced by a number of factors such as system inertia.5 DG with RideThrough Capability In this section.02 0 0 20 15 10 5 0 stability limit (original case II) stability limit (DG with ride−through capability) 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] stability limit (original case II) stability limit (DG with ride−through capability) Oscillation duration [s] 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 DG Level [%] Figure 5. it is shown that the use of ridethrough capable DG results in an improved voltage support for the system. Note that by applying DG with ridethrough capability.14 are consistent with those of Chapter 4. disconnecting DG from the system during fault acts like introducing a resistive load to the system. The circles (dotted line) mark the original Case II and the triangles (dashed line) the results of applying DG with ridethrough capability in this case.5 DG with RideThrough Capability 5.04 0. rotor speed deviation [pu] 0. and protection mechanism of the powerelectronic interfaced DG. However. Figure 5.2 pu of the power electronic connected DG. constant power and constant current. Max.06 0. As loads are equally divided in constant impedance.15 shows the voltage level at a certain bus in the system for 50% and 55% DG respectively. damping the synchronous machine oscillations (higher indicators values at DG level up to 50% when DG is kept connected). reactive power support. DG with ridethrough capability is applied to the simulation setup of Section 5. and hence in system stable operation even up to 80% penetration levels. The rated current of the DG converter is limited to 1. . Figure 5.88 5. the applied fault does not cause instability at DG levels of 60% through 80% The results shown in Figure 5.14: Transient stability indicators: maximum rotor speed deviation (top) and oscillation duration (bottom) as a function of the DG penetration level. When the penetration level of DG is low. it improves system stability.
Therefore.e. within this chapter. the power electronic converters need a frequency reference taken from the grid.5 0 0 1. it is even more important not to lose these units.6 Remarks In this section.’Vertical to Horizontal’ Transformation of Power Systems 1. Moreover. When these machines are unstable. the ’verticaltohorizontal’ transformation is limited until at least two centralized generators are still running (Section 5. The remaining CG units albeit small in percentage (e. The solid lines correspond to the original Case II (i. intuitively when the increasing penetration level of converter connected DG that adds no inertia to the system is in parallel with the decreasing . Therefore.g.. they cannot be simply disconnected from the system. DG is disconnected during the fault). the dashed lines to the case where ridethrough capability DG is used 5.e. i. In this way.4). only 2 units) are crucial for the operation of the DG units and thus the whole system to give the frequency reference to the powerelectronic interfaced DG. Power system inertia and the respectively stored kinetic energy of the total generation system is an important component inﬂuencing power system stability [43].15: Node voltage of bus 16 at 50 % DG level (upper graph) and 55% DG level (lower graph).5 Voltage [pu] 1 0. it is assumed that all power electronic interfaced DG units require a reference frequency for their proper operation. 5. an increase in DG penetration level is followed by a decrease in inertia and reactive power support ability in the system.7 Conclusion Higher penetration levels of DG leads to taking out of service of centralized generators.5 Voltage [pu] 1 0.5 0 0 89 1 2 Time [s] 3 4 5 1 2 Time [s] 3 4 5 Figure 5. when these CG units provide additional reactive support.
the reactive power support required. Therefore a limit in DG penetration level cannot solely be set based on the proportion of the active power production of DG. Reactive power support depends signiﬁcantly on the network topology and the optimal power ﬂow strategy. Therefore. The transient stability analysis results that the ability of centralized generators to supply supporting services like system inertia and reactive power. having a large number of small to mediumsized distributed generators. the system will be more vulnerable to disturbances [48]. or as a share of the active power output generated by DG units compared to the total load. .90 5. Studies on the impact on the power system with DG and limited inertia have been presented in literatures [38]. [44].7 Conclusion inertia due to taking out of service of centralized generators. to determine the DG penetration limit for a speciﬁc case. that is where the simultaneous eﬀects of increasing DG and decreasing system inertia occurs. If higher DG penetration levels are wanted. This is the subject of the Chapter 7. [43]. in addition. It is then found that a limit to DG penetration level in a given power system can be reached. Another approach as suggested in literature could be for instance to equip the converter connected generators with control system such that generators contribute in stabilizing the system [37]. detailed analysis has to be done on the network. A straightforward remedy for instability problems that may occur could be advocated by assigning certain “mustrun” CG units that provide inertia to the system and. is of most importance. This limit is based on the total kinetic energy remaining in the system and the reactive power support scheme adopted. both the sequence of shutting down CG units and the strategy for substituting reactive power support have to be speciﬁed. in theory one could consider aiming for 100% DG. the limit obtained is sensitive to the network analyzed and the verticaltohorizontal scenario. Furthermore. This chapter examines the impact of such a change in supply structure on the system transient stability. Doing so. the contribution of this work is the more fundamental approach of transforming a power system from the classical ’verticallyoperated power system’ supported mainly by several large centralized generators into a ’horizontallyoperated power system’. Therefore. Yet. suﬃcient inertia and voltage support must be installed. This ﬁnding is interesting since DG penetration level is classically deﬁned as a share of active power output generated by DG units compared to the total active power generated by all generators in the system.
We begin our discussion with the impact of stochastic behavior of DG on the load ﬂow in Section 6.Chapter 6 Stochastic Approach to Transient Stability of Power Systems with DG In Chapters 4 and 5 the transient stability of the power systems is investigated based on a deterministic approach. by nature. however. one may think of analysis methods that incorporate the stochastic nature to investigate the transient stability of the power systems. Throughout the analysis we assume that every DG unit is a customerowned synchronous generators. we extend the transient stability analysis to and focus more on the stochastic approach. Therefore. In this approach. demand ﬂuctuates due to the behavior of customers. The impact on the system transient stability is discussed in Sections 6.e. In fact. .3. that some parameters of a power system do not behave in a deterministic way. It is found that the transient stability performance of power systems depends on the (prefault) load ﬂow in the system. Even though the DG units are now in essence “deterministic” they still characterize stochastic generation. we restrict ourselves to the study of the stochastic behavior of the DG output (Chapter 2).2 and 6. i. As the stochastic behavior of a power system is. among other things.1. For example. the owners decide whether the units operate or not. this approach is more comprehensive than the deterministic one. the load ﬂow within the system. The implementation of renewable–energy–source–based DG also implies a more stochastic approach on the electricity generation side (Chapter 2). It is well known. extremely complex. that is by deﬁnition stochastic. This stochastic behavior impacts. power system parameters are set at the rated values. In this chapter. The results from this approach show that the inclusion of the stochastic nature of DG leads to a more complete and detailed view of the system transient stability.
2. .e. Even when DG units are in essence “deterministic”.92 6.. each active distribution system (ADS) can be represented by an aggregated load in parallel with aggregated generation NLi NDGi PADS(i) = j=1 PL(i. With DG implemented in every load bus. The fraction of the total load served by the DG is distributed among the DG units.5. if they are customerowned units. . (called Active Distribution Systems (ADS)). NADS . In the “horizontallyoperated” power system. This nondispatchable behavior results from the fact that certain DG units generate power from primary energy sources with inherently stochastic characteristics.1 Stochastic Load Flow In the classical “verticallyoperated” power system there is only a “small” number of large centralized generators “dispatchable”. distribution network) contains both consumption and generation (modeled as negative consumption) in the steadystate simulation. The remaining power generation is divided among dispatchable CG units by considering the economic operation of the power system.k) . . proportional to the active power consumed by the load at that particular bus. The test system and the simulation setup are set according to Section 3.1 Simulation Scenario A 50% DG level is considered. the DG units in the “active” distribution networks. Therefore. PADS(i) is incorporated in the load ﬂow algorithm (i. controllable to meet the demand. are basically “nondispatchable”. each load bus (i.j) − PDG(i. i = 1.e. such as wind and solar energy (Chapter 2). as the corresponding Pi. . The power ﬂow solution of a system with stochastic DG is computed by including the stochastic behavior of the parameters of the speciﬁed active and reactive power of each bus representing ADS in a power system.1 Stochastic Load Flow 6. NLi the number of loads and NDG the number of DG in ADS i. whereas the total load of the test system is kept constant. 6. 6.2 Stochastic Transient Stability Analysis A method is proposed to investigate the impact of a largescale DG implementation on the power system transient stability in which both the stochastic behavior of the DG units and the loads are taken into account.1) contain stochastic elements and a stochastic load ﬂow results. In both cases DG possesses a stochastic generation characteristic. k=1 (6. by considering that both the rigthhand side terms of equation (6.1) with NADS the number of ADS. i.e.sch in the Newton Rhapson algorithm described in Appendix D. and the modeling discussed in the previous chapters needs be adjusted to account for this. the owners can decide whether the units operate or not. similar . seen from the transmission level.
We set that p(DG=on) × k=1 i PDG(i. . Several scenarios are developed in which both parameters p(DG=on) and σ are varied according to Table 6.05 pu impedance on the 100 MVA system base. see Appendix B).7 0. Therefore.01 0. The DG units are modeled as synchronous generators without grid voltage and frequency control (Chapter 3). is run with a DG penetration level increased to 50%.1.03 0. The owners decide whether the units are running or not: i. 5 CG units are shut down (i.Stochastic Approach to Transient Stability of Power Systems with DG 93 to the approach done in Chapter 5. Customerowned DG units that are installed in the ADS supply only active power (1 MW rated active power each) and no reactive power. the maximum DG capacity installed in NDG one load bus ( k=1 i PDG(i.k) is equal to the load value at bus i. the aggregated DG unit is stochastically calculated as a binomial distribution where each DG unit within the DES is connected to NDG the system with a probability p(DG=on) .e CG nr. Minimum and maximum loading limits. The loads are following a normal distribution.9 A Monte Carlo simulation (MCS) is used to generate samples of aggregate DG output power and load at each load bus.000 samples are considered to be suﬃcient since the MCS converges after 10. A DG unit is connected to every load bus via a j0.000 samples are generated in each scenario. whose objective function is to minimize the fuel cost.03 0. These 10. Table 6.k) ) can be 1/p(DG=on) times the rated value. Thus. the DG units are randomly connected to and disconnected from the network.e.03 0. power production with CG3 is more expensive than with CG2 ).03 Probability that DG unit is turned on (p(DG=on) ) 0.000 samples. where the mean values equal the rated values (Appendix B).8 0. The most ineﬃcient CG units whose power output falls below their minimum loading limits are shut down and switched shunt devices are implemented at the location of the shut down CG units to compensate for the former reactive power generation. and the standard deviations equal σ. An optimal power ﬂow program.8 0.8 0. as well as the cost models of the CG units are chosen in such a way that the highernumbered CG units are more expensive than the lowernumbered ones (e. 6 to 10.1: Scenario cases Case (nr. In this way. and represents the aggregate active power generation of all DG units in an ADS [18].05 0.g. 10.5 0.) 1 2 3 4 5 6 Loads’ standard deviation (σ) 0.
12 and 31. by keeping the active and reactive power ratio (Pi /Qi ) at each load bus (bus i) constant. if each DG unit has a lower probability to be connected to the grid.05 31 600 400 200 4 8 20 0 12 0 200 400 600 800 1000 1200 Active (Consumed) Power [MW] 39 1400 Figure 6. Therefore. = 0.2 shows the MCS generated samples of the DG power generation at each load bus (except for bus nr. = 0. of Samples 31 800 600 400 200 26 1 28 8 2 7 25 29 272 321 31 4 15 6 STD dev. more DG units will be implemented in order to generate the average power equal to the rated value. None of the CG or DG is disconnected during the fault. The reason is because we set the average power output of the DG units as the rated.94 6.01 31 800 600 400 200 4 8 12 20 0 1000 0 200 400 600 800 1000 39 1200 1400 No. For each MCS sample.2. one can observe larger spread around the rated load values in case of an increasing standard deviation (σ).2 Monte Carlo simulation (MCS) Samples Figure 6. . where the mean values equal the rated values and the standard deviations σ vary according to scenario nr. After 150 ms. 1. of Samples 26 1 288 7 25 2 29 7 3 21 2 2 316 4 15 800 STD dev.1 shows the MCS generated samples of the active power consumption of each load. 2 and 3 in Table 6. the transient stability of the test system is investigated by applying a fault to the transmission line between the buses 15 and 16 of the test system. No. The loads are normally distributed functions.1. Figure 6. not displayed due to small values).1: MCS generated samples of the active power consumption of each load bus (Numbers in the graphs indicate the number of the load bus) Figure 6. of Samples 26 1 2 28 7 85 2 29 27 2 321 3 1 4 15 6 1000 STD dev.03 12 4 8 20 0 1000 0 200 400 600 800 1000 39 1200 1400 No.2 indicates that the spread of the MCS samples of the DG power generation decreases along with a rising probability p(DG=on) .1. = 0. the fault is cleared by tripping the faulty line. From Figure 6. The reactive load samples are generated based on the MCS samples of the active load data.2 Stochastic Transient Stability Analysis 6.
Due to this eﬀect. It is obvious from these results that large power ﬂows have a detrimental eﬀect on the damping of oscillations.1.9 No.Stochastic Approach to Transient Stability of Power Systems with DG 4000 3000 2000 1000 0 0 100 200 300 400 500 600 95 No.4.2: MCS generated samples of the DG power generation at each load bus (Numbers in the graphs indicate the number of the load bus) To assess the transmission system stability.3 (left)). The results are shown in Figures 6. of Samples 3000 2000 1000 0 0 1826 25 28 7 2 29 32 2721 15 4 1 3 6 4000 8 4 100 200 300 400 500 Active (Generated) Power [MW] 20 39 600 Figure 6. Algorithm 4 shows the stochastic transient analysis of power systems with DG in an ADS. of Samples 26 18 25 287 2 29 3 27 21 15 24 1 3 6 pDG=on = 0. 6. of Samples 4000 1826 28 23 25 7 29 27 21 15 24 1 3 6 3000 2000 1000 0 0 pDG=on = 0. the spread of the stored kinetic energy is shown for Cases 4–6. Down from Case 6 to 4.3 and 6. the spread 39 . 1–3 the computations show that the spread of the total active power ﬂow in the lines becomes larger with increasing spread in the load (Figure 6. This is in accordance with what already explained in Chapters 4 and 5. For Cases No. two indicators are applied to quantify the rotor speed oscillations of the CG in the system: maximum rotor speed deviation and oscillation duration (Chapter 3).3 (right)). An immediate consequence of this growth is an increase of the spread of probability distributions of the maximum rotor speed deviation (Figure 6.7 8 4 100 200 300 20 400 500 39 600 pDG=on = 0. In Figure 6.2.4 (left).3 Simulation Results We run Algorithm 4 for all cases given in Table 6. the spread of the stored kinetic energy in the system rises as the probability of the DG units to be connected to the grid decreases and only the rotating masses of the connected DG units contribute to the total stored kinetic energy in the system.5 8 4 20 No.
2 6.i and the probability of DG being connected p(DG=on) 4: Set NM CS := 1000 5: for k = 1.3: Total active power ﬂows (left) and maximum rotor speed deviations (right) in the system in Case 1 (upper).4 4000 Case 2 3000 2000 1000 8500 9000 9500 10000 10500 11000 0 5. of Samples No.2 6. Case 2 (middle) and Case 3 (lower) .4 6. until convergent do 6: for j = 1 to NM CS do 7: Generate normal random numbers for Pi and Qi with µi and σi 8: Generate random numbers for PDG.8 −3 Max.8 6 6.8 6 6.6 5.4 6.6 6.6 6. Rotor Speed Deviation [x 10 pu] 7 Case 3 5. N the number of buses 2: Set DGlevel .6 5. · · · . of Samples 4000 3000 2000 1000 0 8000 4000 3000 2000 1000 0 8000 Figure 6.8 7 Case 1 No.4 6. .i and p(DG=on) 9: Simulate a permanent fault 150 ms (tf ault ) 10: Calculate maximum rotor speed deviation and oscillation duration 11: if MCS convergent with NM CS samples then 12: Calculate probability distribution function (pdf) of maximum rotor speed deviation and oscillation duration 13: quit end if 14: 15: end for 16: NM CS := NM CS + 1000 17: end for No.8 6 6.4 4000 Case 3 3000 2000 1000 8500 9000 9500 10000 10500 11000 Active Power Flow [MW] 0 5.8 7 Case 2 5.6 6.96 6.4 5.2 6.6 5. . . .2 Stochastic Transient Stability Analysis Algorithm 4 Stochastic transient stability analysis 1: Set test system parameters.i from the binomial distribution with parameters NDG. the mean (rated value) µi and the standard deviation σi 3: Set the number of generators connected in each busi NDG. of Samples 4000 3000 2000 1000 0 8000 8500 9000 9500 10000 10500 11000 Case 1 4000 3000 2000 1000 0 5. with i = 1.
2 gives a comparison of the indicators obtained from the deterministic approach and the ones (average values) obtained from the stochastic approach. the more vulnerable the system is in terms of transient stability (e. due to disturbance) [25].45 0 5.4: Total stored kinetic energy (left) and maximum rotor speed deviations (right) in the system in Case 4 (upper).5 4000 Case 5 3000 2000 1000 3.5 4000 Case 6 3000 2000 1000 3. Accordingly.5 Case 5 6 6. [33] (Chapters 4 and 5). Table 6.35 3. .5 Case 6 6 6. the transient stability analysis does not result in a single value of the stability indicators. in stead of a blackandwhite statement on the system stability. the lower the stored kinetic energy in the system (i.4 Energy (Kinetic) [x 10 4 MJ] 3.4 3. of Samples 4000 3000 2000 1000 0 3.3 3. a more nuanced result is obtained. Case 5 (middle) and Case 6 (lower) of the probability distributions of the maximum rotor speed deviation increases slightly (Figure 6. which gives an indication that the system tends to be more unstable when the stochastic behavior of the ADS is taken into account. but in a probabilistic distribution of the stability indicators.5 6 Max.25 No.Stochastic Approach to Transient Stability of Power Systems with DG 4000 3000 2000 1000 0 3.3 3. of Samples 97 No. the stochastic behavior of connecting/disconnecting DG units results in a stochastic amount of stored kinetic energy in the system. In addition.4 3. Rotor Speed Deviation [x 10 −3 pu] 6. inertia) becomes.45 0 5.35 3. The stochastic behavior of consumption and generation within an ADS results in stochastic power ﬂows in the lines.e.35 3. In other words.4 (right)).3 3.5 Case 4 4000 3000 2000 1000 0 3. Hence.25 Figure 6. of Samples 4000 Case 4 3000 2000 1000 3.45 0 5.25 No.g. Especially the average values of the oscillation duration are signiﬁcantly higher than the ones obtained from the deterministic approach.
0 51.16 7.0057 0.0057 0.3).4 58.0 82.3 55. [pu] 0. Deterministic approach Max.3: Probability that the values of the stability indicators that result from the stochastic approach are larger than the ones from the deterministic approach Case nr.) that result from both deterministic and stochastic stability analysis Case nr. dev. dev. dur. [pu] Osc.0 1. Although the average values of the maximum rotor speed deviations are almost equal to the deterministic ones (Table 6.16 7.2).98 6.) and oscillation durations (osc.16 Stochastic approach (average value) Max.8 .0057 0. dur.0057 0.16 7.0058 7.25 0. This means that it gives a more complete picture of the transient stability of a system with DG in ADS.8 68.4 92.0057 0.26 0. [s] 7. but in a probabilistic distribution of the stability indicators.30 0.2 85.2 Stochastic Transient Stability Analysis Table 6. [s] 0.0057 7. dur.0058 7.0058 7.23 1 2 3 4 5 6 An advantage of the stochastic transient analysis is that it does not result in a single value of the stability indicators.2 70.0058 7.3 55.0058 7.16 0.16 7.0057 Osc. dev.16 7. 1 2 3 4 5 6 Probability [%] Maximum rotor speed deviation [pu] Oscillation duration [s] 38.6 70.48 0. the probability that the value of the indicators obtained from the stochastic approach is larger than the ones obtained from the deterministic approach can be quite high (Table 6.2: Maximum rotor speed deviations (max. Table 6.
These 5. The remaining power is divided among the dispatchable CG units by considering the economic operation of the power system. is run with a DG penetration level from 10% to 50%.nominal ) × 100%. . similar to the approach in Chapter 5.2 MCS Samples The total (aggregated) customerowned DG is stochastically calculated as a binomial distribution where each DG unit within the distribution network is connected to the system with a probability (pon ) equal to 0.e. the maximum P capacity of DG installed at one load bus can be 125% of the rated load value (i. maximum rotor speed deviation and oscillation duration (Chapter 3).3. i. by increasing the fraction of the total load in the test system served by DG in steps of 10% up to 50%.2. The loads are following a normal distribution. The penetration (p ×PDG. Switched shunt devices are implemented at the location of the shut down CG units to compensate for the reactive power lost.Stochastic Approach to Transient Stability of Power Systems with DG 99 6.1 Simulation Scenario A ‘verticaltohorizontal’ transformation is obtained by gradually increasing the DG penetration level. when 100% DG is supposed). two indicators are applied to quantify the rotor speed oscillations of the CG in the system. and the standard deviations equal 2. in every scenario of increasing DG penetration level. 12 and 31. not shown due to small values). The most ineﬃcient CG units whose power output falls below their minimum loading limits are shut down.8.3 Stochastic Transient Stability Study with Increasing DG The stochastic approach is used in this section to investigate the transient stability of power system that goes through a ‘verticaltohorizontal’ transformation (Chapter 5).installed ) level of DG is deﬁned as: ( on load. a scheme of minimum and maximum loading limits and the cost models of the CG units are taken in such a way that the highernumbered CG units are the more expensive ones. As in Section 6.5%. at 10 to 50% DG penetration level. The DG units supply only active power (1 MW rated active power each) and no reactive power. 6. To assess the transmission system stability.000 samples are considered to be suﬃcient since the MCS converges after 5. 6. An optimal power ﬂow program.e. where the average is the rated load (Appendix B). 5. (except at bus nr. with objective function to minimize the fuel cost.000 Samples are generated for the aggregate DG output power and the load at each load bus. The fault is cleared by tripping the faulty line after 150 ms.000 samples.3.5 shows the contours of the histograms of the MCS generated samples representing the DG power generation at each load bus. The transient stability of the test system is investigated by applying a permanent fault to the transmission line between buses 15 and 16 for each MCS sample. Figure 6. Thus.
4. 6.6 shows a number of scenarios of the ‘verticaltohorizontal’ power system transformation.6).3 Stochastic Transient Stability Study with Increasing DG No. of Samples 261 8 2528 7 23 2 2 2 159 2471 13 6 4 30% DG Level 0 0 2000 100 20 1000 8 200 300 39 400 500 600 No. The values of both stability indicators. 8 . of Samples 2 186 2 2 8 235 7 29 27 21 15 124 3 6 50% DG Level 4 20 1000 0 0 100 200 300 400 500 39 600 Active (Generated) Power [MW] Figure 6. To compare the transient stability performance in the diﬀerent scenarios. of Samples 2 186 28 25 7 23 227 21 9 15 24 3 16 8 4 40% DG Level 0 0 2000 100 200 300 400 39 500 1000 20 600 No. 20%.3 Simulation Results Figure 6. i. It can be observed from Figure 6.5: MCS generated samples representing the DG power generation at each load bus at a 10%. 30%. of Samples 18 2 2 25 2876 129 24 1 23 5 32 7 6 1 4 8 20 10% DG Level 1000 0 0 2000 100 39 200 300 400 500 600 No. the transient stability indicators are applied on these ﬁve CG units. the lowest 5 bars of ﬁg.e.3.6 that there are ﬁve CG units operating within all scenarios (DG penetration level from 0% to 50%): CG nr. and 5 (i. 3.100 2000 6. the average value in case of the stochastic approach and the deterministic value in the deterministic approach. of Samples 1826 25 2 7 28 32 27 9 221 15 13 46 20% DG Level 8 4 20 1000 0 0 2000 100 200 39 300 400 500 600 No.e. are shown in Table 6. 2. 1. 40% and 50% DG penetration level 6.4.
diﬀerent for both approaches.86 0. at the 40% and 50% DG penetration level) results compared to the results obtained from the stochastic approach. The latter fact gives an indication that the results obtained from the deterministic approach may give either too pessimistic (e.31 0. Oscillation rotor speed duration deviation [pu] [s] 0.0055 6.0055 7.0055 7.09 0.88 Stochastic approach Average of the max.09 0.Stochastic Approach to Transient Stability of Power Systems with DG 7000 6000 Active Power [MW] 5000 4000 3000 2000 1000 0 0 10 20 30 40 DG Penetration Level [%] 50 CG−10 CG−9 CG−8 CG−7 CG−6 CG−5 CG−4 CG−3 CG−2 CG−1 101 Figure 6.0053 6.07 0.g.0053 6. however.4: Maximum rotor speed deviations and oscillation durations resulting from the deterministic and the stochastic (average) approaches of all MCS simulations at the ﬁve DG penetration levels Deterministic approach Max.31 0.0063 7.32 0.0055 7.4.0055 6.36 0.0055 7. it can be concluded that the stochastic approach and the deterministic approach result in the same average maximum rotor speed deviation.6: Dispatched CGs as a function of the DG penetration level Table 6. . The oscillation duration is.g. at the 30% DG penetration level) or too optimistic (e.20 DG level Scenario (%) 10 20 30 40 50 From Table 6. Average of the rotor speed oscillation deviation duration [pu] [s] 0.0063 6.
with the stochastic approach proposed in this section.8. is shown that the probability distributions of the maximum rotor speed deviations overlap in some ranges.102 Pr 6.0043 0 0. rotor speed deviation of oscillation duration 0.7 illustrate the use of this approach.3984 0.5. In Figure 6.2413 . First of all. for example.3211 0. Table 6. leads to Table 6. Equation (6.6942 0 0.g.3 Stochastic Transient Stability Study with Increasing DG %DG Level (i) %DG Level (i+1) Probability x1 Indicator values x2 x Figure 6. the blackandwhite statement that the system becomes more unstable when e. the DG penetration level increases from 10% to 20% can be more nuanced. (6. x2 x Pr (g1 (x) > g2 (x)) = x1 g1 (x) x1 g2 (y) dy dx.1406 0.2) Discretizing and applying (6.2) and Figure 6. the probability distributions of the stability indicators are obtained and contain additional information.7: Probability distributions of stability indicators overlap Furthermore.2) to the results obtained from the stochastic approach.5: Probability that the indicators resulting from the scenario with lower DG level surpass the indicators resulting from the scenario with the onestephigher DG level %DG level g1 (x) g2 (x) 10% 20% 30% 40% 20% 30% 40% 50% Pr (g1 (x) > g2 (x)) x = indicator x = indicator of max.
Stochastic Approach to Transient Stability of Power Systems with DG
No. of Samples 2000 1000 0 2000 1000 0 2000 1000 0 2000 1000 0 2000 1000 0 5 5.5 6 6.5 Max. Rotor Speed Deviation [x 10−3 pu] 7 5 5.5 6 6.5 50% DG Level 7 5 5.5 6 6.5 40% DG Level 7 5 5.5 6 6.5 30% DG Level 7 5 5.5 6 6.5 20% DG Level 7
103
10% DG Level
Figure 6.8: Histograms of the maximum rotor speed deviations (in 10−3 pu). Note that the xaxis and yaxis of all graphs are adjusted to be uniform
6.4
Conclusions
Results from the preceding chapters suggest that the transient stability performance of power systems depends on the (prefault) load ﬂow within the system. Moreover, the implementation of DG, based on renewable energy sources, implies more stochastic behavior of the electricity generation. Therefore, stochastic approach should be considered to study the transient stability of power systems. This chapter deals with examining the consequences of a stochastic approach on the system transient stability. The study is mainly focused on the stochastic behavior of the DG output, that is simulated using Monte Carlo Simulation method, within the framework of the “verticaltohorizontal” transformation.
No. of Samples
No. of Samples
No. of Samples
No. of Samples
104
6.4 Conclusions
It is shown that although many simpliﬁcations are applied for the stochastic parameters, the inclusion of the stochastic behavior of DG leads to a more complete and detailed view of the steady state situations. When the transient response of the system is simulated, the average values of the stability indicators obtained from the stochastic approach may be the same as the indicators obtained from the deterministic approach. Yet, in the stochastic approach, also the probability distributions for the indicators are estimated. When compared to the single result of the deterministic approach, it is shown that the deterministic approach may give either undervalued or overvalued results. The probability distributions of the stability indicators of two diﬀerent DG penetration levels may overlap. Therefore, the deterministic statement that a system becomes more stable (or unstable) when the DG penetration level is increased can be more nuanced as that the system becomes more stable (or unstable) to certain degrees. The merit of stochastic stability studies is evident even in the case of a relatively simple example studied here. Therefore, a stochastic approach becomes a necessity when, e.g. DG units depend on highly intermittent energy sources (e.g. wind and solar energy), or when the dependence between the generation of DG units and the loads becomes important.
Chapter 7
Maintaining Power Balance with Active Distribution Systems
This chapter investigates the situation when the power system is pushed towards a scenario, where DG penetration reaches a level that covers the total load of the original power system (100% DG implementation in the system). The assumption is that all DG units are implemented via powerelectronic converters within active distribution systems (ADS). The ADS are connected to the transmission system also via powerelectronic interfaces. The power system is still connected to one source that provides a constant 50 Hz voltage. This can be, for example, a connection (tie line) to a strong external system. This source is meant only to gives a (system) frequency reference for the other generators and generates no power at steady state. Therefore, any power imbalance in the power system must be compensated by generators in the ADS. Due to the powerelectronic interfaces, the output power of all generators within the ADS are decoupled from the system frequency. Therefore, voltage deviations are proposed to detect power imbalances in the system. Remedies to eliminate the negative consequences of using the voltage deviations to detect the power imbalances are discussed and appropriate control systems are suggested. Figure 7.1 shows an illustration of the power system with ADS.
7.1
Background
The implementation of the distributed generation (DG) turns the current passive distribution network into an active one (Chapter 1). This active distribution network does not only consume, but it also generates power and supplies it to the transmission system [41], [55]. In this way, power can be transferred from one distribution network to another. When we reﬂect further on this issue to the
and with ﬂexible storage systems. However. On the other hand. any .2 Power Balance One of the primary tasks of power system operation is to balance the electricity supply from generators and loads including the losses in the network at any time. including ACDC. we could imagine that at a certain moment in time the electrical power generated by the DG within the distribution networks may become suﬃcient to cover the total demand in the transmission system.2 Power Balance 3 3 3 Figure 7. called power system stability [34] (Chapter 3). all generators within the ADS are ‘hidden’ behind powerelectronic interfaces. the source that provides the constant 50 Hz voltage is meant only to give a frequency reference for the other generators (in the ADS) and not meant to generate power at steady state. when the distribution networks evolve to become more hybrid. a disturbance may occur in the system that causes a power imbalance. Therefore. it is envisaged that connecting these distribution networks to the transmission system via power electronic interfaces will be an interesting option [19]. The output power of the generators are decoupled from the grid frequency. 7. Moreover. It is important that the power system can restore the power balance and return to a stable state after this disturbance.1: Illustration of the power system with ADS extreme. In this chapter.106 7. Most of the DG implementations based on renewable energy sources or environmentallyfriendly technologies are connected to the grid via power electronic converters (Chapter 2).
1 Model of Power System with ADS Assumptions To decouple the changes of the voltages with the changes of the reactive power. Secondly. To deal with the latter problem.Maintaining Power Balance with Active Distribution Systems 107 power imbalance in the power system must be compensated by generators in the ADS. as a result of the decoupling between the output power and the grid frequency in the ADS. if any. Traditionally. as an altered system frequency.3.3 7. However. remedies to eliminate these problems are presented. The generators are initially set to balance active power demand. the power imbalance cannot be detected by the ADS in the classical way. some assumptions are applied on the model of the power system as the following: • The distribution networks are equipped with reactive power sources. in such a way that they behave like resistive lines. • The reactances of the transmission lines are compensated. the voltages at the buses throughout a power system are not the same (in contrast to the system frequency). therefore. . The reactive power sources are preset to balance the reactive power demand. • The loads demand active and reactive power. Firstly. The reactive power source is suﬃcient to fulﬁll the reactive power needs within the distribution networks. • The (active) distribution networks are connected to the transmission system via power electronic interfaces. The ADS are considered as follows: • The (distributed) generators are connected via power electronic converters and generate only active power. two problems may arise from using the voltages to detect a power imbalance in a power system. 7. electrical motors. however. 7. Assumptions are applied on the model of the power system with ADS to eliminate the ﬁrst problem as it is not essential.3. Electrical motors and the corresponding inertias are hidden behind power electronic interfaces from the grid. Therefore in this chapter. are assumed to be included in the constant power model of the loads. the voltage (magnitude) at a bus is aﬀected by the ﬂow of the reactive power. They are modeled as constant impedance and constant power. several control systems are proposed and simulated. the use of voltages to detect a power imbalance in the transmission system with active distribution systems is proposed. • The reactive power is supplied by dedicated reactive power sources. Therefore in the following sections. The power electronic interface is assumed to permit only active power to ﬂow (bidirectional).2 Model of ADS This chapter focuses on functioning of the ADS to maintain the power balance in the systems.
2.108 7. generators perform the following functions: • One central generator operates as a reference. They only supply active power.3 shows the representation of this generator. • The distributed generators operate as active power sources. where Us is the constant (reference) voltage with a ﬁxed frequency. only active power ﬂows between the ADS and the transmission system. • A constant current source that generates a current in phase with the terminal voltages Ut of the generator (a Phase Locked Loop. each ADS is modeled as shown in Figure 7.3 Model of Power System with ADS • A power imbalance is simulated by changing the active power demand of the load.2: Model of ADS used in the simulation of maintaining the power balance in the system 7. to balance the reactive power within the ADS. Figure 7. $'6 3 '* 7UDQVPLVVLRQ 6\VWHP V /RDG Figure 7. • One (or more) central or decentralized generator serves as a ’slack’ generator that either supplies or absorbs any deﬁcit or surplus of active power in the system.3. is used for this purpose) is used to represent the distributed generator that supplies active power. reactive power sources are applied and modeled as shunt devices. Therefore. . This generator is assumed to have no current limiter and to be equipped with suﬃcient (energy) storage. • A controlled current source is used to represent the generator serving as the ’slack’ generator (Figure 7. Figure 7. Three converter connected generator models are used: • A constant voltage source is used to represent the generator providing voltage and frequency reference.3 Generator Models In the power system with ADS. Thus in the simulation. PLL.4 shows the representation of this generator. A reactive power imbalance is not simulated. and Zs is a source impedance.5). Furthermore. It serves as voltage and frequency reference for the other generators in the system and generates no power at steady state.
4: Constant current source model 8W &RQWUROOHU .5: Controlled current source model A remark should be made that power electronic interfaces driving the output of converterconnected (distributed) generators basically represent a voltage source converter. but it behaves like a current source. Yet.3: Constant voltage source model 8W . . so that current source models can substitute voltage source converters when simulations are made in large systems [42].V =V Figure 7. as mostly used nowadays [70].V Figure 7. the use of constant current sources to represent converterconnected generators in this simulation is supported by the following assumptions (Chapter 3): • A converter is actually a voltage source converter.Maintaining Power Balance with Active Distribution Systems =V 109 8V Figure 7.
Thus.110 7.J . .4 Basic Controller Model The function of maintaining the power balance is supposed to be done by the controllerblock of the ‘slack’ generator(s). a constant current source generating a constant current (Ig ) and a controlled current source supplying a changeable current (Ic ) . • When the terminal voltage raises temporarily.and an active load consuming active power (Pload ) are implemented.F 3ORDG Figure 7.J &RQWUROOHU . three types of converterconnected generators . In steady state.6. 7. the current of the voltage source (master generator providing a reference voltage and frequency) should be zero (Is = 0). • In practice. The ‘slack’ generator(s) represented by the controlled current source(s) (Figure 7.ORDG . the use of the constant current source can be justiﬁed by assuming that the generator is equipped with energy storage so that the generator can supply extra active power for a short period of time. a converter is equipped with a current limiter. Hence.4 Basic Controller Model • It is usual to model the sources as P Qsources in studies of large systems [69]. When the terminal voltage drops.V 86 . 100% up to 120% of the rated value).F . 8W = . the converter supplies less active power.6: Basic controller idea applied at a singlebus test system In Figure 7. the use of a constant current source corresponds to a converter whose current is limited to the rated value (in practice. The basic functionality of the controllerblock is highlighted hereunder. so that there is no voltage drop across the impedance Z. as long as the terminal voltage of the generator is constant.a constant voltage source providing a reference voltage (Us ). the terminal voltage (Ut ) of bus1 is equal to the reference voltage Us . A constant current source generating a current in phase with the (terminal) voltage represents a P Qsource generating constant active power (P ) and zero reactive power (Q = 0).5) must supply or absorb any deﬁcit or surplus of active power in the system.
7. a load jump is simulated to cause a power imbalance in the system (Figure 7. Unlike simpler control algorithms.8 and 7. The controller compares a measured value from a process with a reference setpoint. equally divided in constant impedance and constant power. (7. Figures 7. in which the new value brings the process measured value back to its desired setpoint. until the power balance is restored.9 show transients of voltages.1) where U (t) is output of the controller. KP is proportional and KI integral term. The currents generated by the constant voltage source (Is ) and the controlled current source (Ig ) are zero. . for example. To verify the basic controller model. The diﬀerence or “error” signal is processed to calculate a new value for a manipulated process input. Simulations are performed on a Real Time Digital Simulator (RTDS) with a time step of 50 µs.6).6 respectively. Therefore. the current ﬂow to the load (Iload ) rises. the controlled current source supplies more active power. E(t) is error signal (diﬀerence between desired and real output). When. The load demands 60 MW of active power (Pload ) initially. takes care of the primary action. As a result. This controlled current source represents the ’slack’ generator. as Is increases and causes a voltage drop across Z so that Ut decreases. the PI controller can adjust process inputs based on history and rate of change of the error signal.7. representing a strong source acting as the frequency reference. The mathematical form of PI controller can be shown as [46] U (t) = KP E(t) + KI E(t)dt. is meant to provide the frequency reference only: the source should not generate active power at steady state. It can be shown mathematically that a PI loop produces accurate stable control in cases where other control algorithms would either exhibit a steadystate error or cause the process to oscillate. The controller used in the basic model is a proportionalintegral controller (PI controller). currents and active power when the load jump occurs. This source. The system voltage is set at 10 kV. a secondary action is taken. P‘slack′ generator . controlled current source and constant voltage source of the system in Figure 7. The voltage source. the active power consumption of the load Pload increases.e. Note that the P‘constantactivepower′ generator .Maintaining Power Balance with Active Distribution Systems 111 Any power imbalance should be eliminated by controlling the current Ic generated by the controlled current source. The controller of the controlled current source should detect this voltage drop. A load jump is applied by increasing the load modeled as the constant impedance by 30 MW. The constant current source supplies initially all demand. which gives a more accurate and stable control. however. and P‘master′ generator represent the active power supplied by the constant current source. it injects more current Ic . i. Both generators modeled as current sources do not react (yet) and the voltage source starts to supply active power in order to balance the power. a common feedback loop component in industrial control applications.
7: Transients of the voltages when a 30 MW load jump is applied at bus1 of the system shown in Figure 7.6 10 9 8 Iload 7 Current (kA) 6 5 4 3 2 1 0 0 2 4 Time (s) Is Ig Ic 6 8 10 Figure 7.112 12 7.4 Basic Controller Model U 10 s 8 Voltage (kV) Ut ( bus−1) 6 4 2 0 0 2 4 6 8 10 Time (s) Figure 7.8: Transients of the currents when a 30 MW load jump is applied at bus1 of the system shown in Figure 7.6 .
Loadj to the load at busj . the voltages at the buses throughout the system are not the same (Section 7. Gj refers to the constantpower generator at busj.Maintaining Power Balance with Active Distribution Systems 100 90 80 70 113 Pload Active Power (MW) 60 50 40 30 20 P ‘constant active power’ generator P ‘slack’ generator P 10 0 −10 0 2 4 ‘master’ generator 6 8 10 Time (s) Figure 7. and the ’extra’ power demand is ﬁnally supplied at the steady state by the ’slack’ generator (increasing current Ic and active power P‘slack′ generator in Figures 7.6 We can see from these ﬁgures that. the system is decoupled and no longer linear.5 ADS Control Systems Practically. In this case. Consequently.9). The voltage level of the bus also restores to the steadystate value that lies within a +/5% margin of the rated voltage. 7.4 to each bus gives potential diﬃculties.10).2 show respectively the component parameters used and the load ﬂow settings and computed results in the 3bus test system.1 and 7. In this section.9: Transients of the active power when a 30 MW load jump is applied at bus1 of the system shown in Figure 7. since the basic control model is linear. • Decentralizedcontroller with single reference. following the load jump. the power balance is restored. a power system consists of more than one bus. To verify these control systems. • Decentralizedcontroller with hysteresis. applying the basic control model of Section 7. three existing control system methods are proposed [46]: • Standalone master controller.8 and 7. Note that Gref refers to the reference (‘master’) generator. due to the interdependency of the buses.2). Tables 7. a simple test system that consists of 3 buses is deﬁned (Figure 7. In addition.
only one ‘slack’ generator is implemented. Also note that the transmission lines introduce quite some capacitance in the system. In this controller scheme. In this case. UHV and UM V denote the system highand medium voltage levels. each transmission line generates around 16 Mvar. The generator should handle the active power imbalance occuring in the system. XT denotes the transformer reactance and Z the impedance between the reference ’master’ generator and bus1. . R. XL and B denote the resistance. 7. In this approach only one basic controller is applied and connected to one of the buses of the system.1 Standalone master controller The most simple way to overcome the nonlinearity problem is to use a standalone master controller.11 shows the implementation of the standalone master controller in the test system.5. Figure 7.5 ADS Control Systems /RDG 7 7/ 7 /RDG = *UHI * & 7/ 7/ & * 7 & * /RDG Figure 7. In Table 7. Tj indicates the transformer at busj and TLjk the transmission line between busj and busk. Note that GCj refers to the controlledpower ‘slack’ generator (at busj): in this case GC1 at bus1. The ‘slack’ generator represented by the controlled current source.114 7. is implemented at bus1. Sbase denotes the complex power base of the test system.10: A simple 3bus test system and C1 to the reactive power source at busj. reactance and susceptance of the transmission lines. the controller only regulates the voltage at one bus and lets the system come to balance using its own connectivity.1.
3 15.2 15.0 30. When the standalone master controller scheme is implemented.0 60.9 60. steady state values.0 0. TL13 .0 30.Maintaining Power Balance with Active Distribution Systems 115 Table 7.0 15. the challenge is that one generator should compensate for the power balance of the whole system. In this way.1 1.7 59.8 2. Figure 7. other proposals concentrate on dividing the ‘slack’ generator task. the problem of ‘diﬀerent signal’ of voltages for the control input is eliminated by using only one controller at bus1.8 32.0064 0.0 30. The power balance is restored and all system parameters are back to stable.12 shows the transients of the active power where all the 30 MW power (of the load jump) is supplied by the ‘slack’ generator.1: Component parameters used in the 3bus test system Description System base System voltage Transmission lines: (TL12 .0 Computed (Mvar) 6.0 0. T3 ) Impedance Parameter Sbase UHV UM V R XL B 2 XT Value 100 100 10 0.8 2.8 31.0 60. the controller uses only one voltage as its reference signal. the load being modeled as a constant power of 30 MW.5 2.2: Load ﬂow settings and computed results in the 3bus test system Description Generation Parameters Gref G1 G2 G3 Load1 Load2 Load3 C1 C2 C3 Setting (MW) 0.0 Setting (Mvar) 0.0 60.0 15. TL23 ) Transformer (T1 .0 0.0 0.0 60. However.0 Unit MVA kV kV pu pu pu pu pu Z Table 7.7 0.0 59.4 59.0 0. In the following sections.0306 0. T2 .0 0.0 Computed (MW) 0.2 32.0 0.0 15.0 0.0 15.0 60.5 59.0322 0.3 Load Shunt device A load jump is applied by increasing the constant power load at bus2.4 60.0 60. .
in this example three decentralized generators with their own controllers are applied. Because of that. Therefore. . Each of the ‘slack’ generators. Note that GCj refers to the controlledpower ‘slack’ generator at busj.11: Implementation of the standalone master controller in the test system 7.e. In this approach. the power balance of the generators is not achieved. Figure 7. i. the voltage at bus1 (U1 ).13 shows the implementation of the decentralizedcontroller with single reference in the test system.116 7. represented by the controlled current source is implemented at buses 1. if these three controllers are applied ’as is’ in the test system. It is already expected that the system is decoupled and the voltage at each bus is not exactly the same.2 Decentralizedcontroller with single reference In practice. One control signal is used by all generators. decentralizedcontrollers with single reference is proposed here.5 ADS Control Systems /RDG 7 7/ 7 /RDG = *UHI 8 * &RQWUROOHU *& & 7/ 7/ & * 7 & * /RDG Figure 7. the ‘slack’ generator at each bus has its own controller. 2 and 3. The generators altogether should take care of the active power imbalance that occurs in the system. it is impossible to use one central generator to compensate for the whole system.5. However. but the reference signal is common to all of them and can be taken from any voltage point. even though the voltage can be regulated.
However there is one aspect that should be considered. It might happen that not all system has this luxury.12: Transients of the active power when a 30 MW load jump is applied at bus2 of the system shown in Figure 7.Maintaining Power Balance with Active Distribution Systems 100 90 80 70 Active Power [MW] 60 50 40 30 20 10 0 −10 0 2 4 Time [s] 6 8 10 G ref 117 Load 2 G .e. steady state values. 2 and 3. Load 1 3 G C1 Figure 7. . The diﬀerence is that this control signal is used for all three controllers. by implementing the decentralizedcontroller with a single reference. Each generator supplies the required power into balance.G .11 A load jump is applied by increasing the constant power load at bus2. With this approach the control generator is no longer centralized. 10 MW.14 shows the transients of the active power. In the next section. The 30 MW power of the load jump is supplied by the three ‘slack’ generators at buses 1. the problem of ‘diﬀerent signal’ of voltages for the control input is eliminated by using only one controller at bus1. Also. of 30 MW. all system parameters are back to stable. In the same way as in the ﬁrst approach. i. Figure 7. the load being modeled as constant power. the third method that can be applied for the system without communication link between each controller is described.G 1 2 3 Load . This approach needs a communication link between each controller to transfer the reference data signal.
Figure 7. each ‘slack’ generator uses the voltage where the generator is implemented. The generators altogether should take care of the active power imbalance occurring in the system. One ‘slack’ generator is implemented at each bus.5 ADS Control Systems /RDG 7 7/ 7 /RDG = *UHI 8 * 8 &RQWUROOHU *& & 7/ 7/ & *& &RQWUROOHU * 7 & * 8 &RQWUROOHU *& /RDG Figure 7. The symptom that happens by applying the basic controller over the test system. The generator GCj is also linked to the control signal the voltage of busj (Uj ). To prevent this symptom. hysteresis is applied in the controller input: the controller stops regulating the system whenever the value of the voltage lies within the hysteresis boundary. As discussed above.13: Implementation of the decentralizedcontroller with single reference in the test system 7. 2 and 3.118 7. Note that GCj refers to the ‘slack’ generator at busj. being 1% of the rated value. In this control scheme. The ‘slack’ generators represented by the controlled current sources are implemented at buses 1. . The use of hysteresis is allowed as long as the width of it is less or equal than the tolerance of the voltage. the linear controller cannot work perfectly in the nonlinear decoupled system.5.3 Decentralized controller with hysteresis The third approach proposed uses three decentralized controllers. is that in steadystate the controller oscillates as the system is decoupled and the reference signal is not exactly the same.15 shows the implementation of the decentralized controller with hysteresis. one applied at each bus.
G Active Power [MW] 60 50 40 30 20 10 0 −10 0 2 G .G .G .14: Transients of the active power when a 30 MW load jump is applied at bus2 of the system shown in Figure 7.Maintaining Power Balance with Active Distribution Systems 90 80 70 G .13 /RDG 7 7/ 7 /RDG = *UHI 8 * 8 & *& 7/ 7/ & *& &RQWUROOHU &RQWUROOHU * 7 & * 8 &RQWUROOHU *& /RDG Figure 7.G C1 C2 1 2 3 119 Load 2 Load .15: Implementation of the robust controller in the test system . Load 1 3 Gref C3 4 Time [s] 6 8 10 Figure 7.
the ADS are connected to the transmission system via powerelectronic interfaces.120 7. steady state values.6 Conclusions This chapter presents models and control systems for maintaining the power balance in the power system with active distribution networks (ADS). Load3 G ref C3 4 Time [s] 6 8 10 Figure 7. Moreover. 2 and 3. .G . The main advantage of this approach is that the requirement of the communication link between controllers is eliminated.G Active Power [MW] 60 50 40 30 20 10 0 −10 0 2 G . Figure 7.6 Conclusions A load jump is applied by increasing the load at bus2. 7. but the system still operates within the speciﬁcations. The eﬀect of hysteresis is that there is a steady state error on the voltage.16 shows the transients of the active power.15 By applying this approach. All system parameters are restored to stable. where the DG units covering the total demand of the original power system are implemented within the ADS via powerelectronic converters. each generator supplying 10 MW. It might be not the optimal solution. Thus.16: Transients of the active power when a 30 MW load jump is applied at bus2 of the system shown in Figure 7. it can be accepted. 90 80 70 G .G . the load being modeled as constant power of 30 MW. The 30 MW power of the load jump is supplied by the three ‘slack’ generators at buses 1. the voltage regulation is maintained. the power balance of the system can be achieved. but while the steady state value is smaller than the tolerance of the voltage value.G C1 C2 1 2 3 Load2 Load1.
and all system parameters return to stable. . easy to implement and has a good performance. steady state values. [43]. As a result. the decentralized controller with hysteresis. Each control approach proposed here has its own advantages and drawbacks. the need of communication between controllers is eliminated. at the price of requiring a communication link between controllers to transfer the reference signal data. the decentralizedcontroller with single reference can overcome the single generator problem. Therefore. the ADS cannot detect power imbalance as an altered system frequency such as in classical power systems [44]. the power balance in the power system can be maintained by the ADS. the output power of the DG units and the ADS are decoupled from the grid frequency. the power balance is restored by means of the ADS with each of the three control systems developed. since this approach uses a split generator for each bus and in addition the three controllers are totally independent. The simulation results show that by applying each of these three control systems. the power imbalance is taken care of by several ADS units. For this purpose. The standalone master controller is the most simple one. decentralizedcontroller with single reference. Some assumptions are applied to the power system model to deal with the problem that the voltage magnitude at a bus is aﬀected by the reactive power ﬂow. in this chapter the idea is developed to use the voltage to detect and maintain the power balance.Maintaining Power Balance with Active Distribution Systems 121 Due to the powerelectronic interfaces. control systems are developed to deal with the problem that the voltages at the buses throughout the power system are not the same. and decentralized controller with hysteresis. Thus. The second approach. The most interesting approach to be implemented is the third one. three possible control system concepts are adjusted and applied: standalone master controller. Moreover. the power imbalance is taken care of by one ADS unit while in the latter two. The simulation results show that when a load jump is applied in the power system to simulate the power imbalance. It should be noted that the primary action is taken care of by a strong source which acts as frequency reference. The main disadvantage is that it uses a single generator to supply the required power. In the ﬁrst control system.
122 7.6 Conclusions .
Dispatchable/nondispatchable DG or direct/indirect gridconnected DG are included in these characteristics. and equal area criterion.1 Overview Currently. and become the basis for the DG models used in the system simulation. after a disturbance. For this purpose. Power system studies are usually based on computer simulations. and more speciﬁcally on the rotor angle stability. in which they are referred to as distributed/decentralized/embedded generation (DG). Due to the small DG penetration level. This transformation is the result of an increasing DG penetration level in the system. the technical impact of DG on the distribution systems have been studied often. To assess the system transient stability. DG technology and especially the way DG is connected to the grid determine DG characteristics inﬂuencing a power system. we have introduced the concept of a ‘verticaltohorizontal’ transformation of power systems. concerns on environmental issues lead to an increasing implementation of environmentallyfriendly generation in electrical power systems. However. This simulationbased approach is used in this work as well. To study the DG impact on power systems. simulations have been performed on scenarios developed within the ‘verticaltohorizontal’ transformation context. In this work. By means of the swing equation. power angle equation. it can be expected that the impact is no longer restricted to the distribution systems.Chapter 8 Conclusions 8. the impact of DG on the transmission system has not been treated so often. we have used two indicators: maximum rotor speed deviation and oscillation duration. but inﬂuences the system as a whole. when the penetration level of DG in a power system becomes higher. the transient behavior of the rotational dynamics of synchronous machines is studied. we have investigated and discussed the impact of DG on power systems with a focus on the transmission system. rather than the classical sys . In literature. Many of these generation technologies tend to be connected to the distribution systems. We have focused on the impacts of DG on the system transient stability.
4 has been used to simulate the test system. and the system becomes more vulnerable in the face of a disturbance. To simulate the ‘verticaltohorizontal’ power system transformation.1 Overview tem stability indicators: critical clearing angle and the corresponding critical clearing time. the total inertia of the system reduces as CG units are taken out of service. Therefore. along with the increasing DG penetration levels. scenarios have been developed in which the DG penetration level is raised while the load is constant.25. This makes sense as all centralized generators remain in the system . the fact that the DG generates power close to the loads causes a reduction in the total real power ﬂowing in the system.g. This leads to an overcapacity in the system. In this way. Therefore it inherently improves transient stability of the transmission system. The power system simulation software package PSS/E v. since large power ﬂows have a detrimental eﬀect on the damping of the oscillations.124 8. we consider this 30% DG penetration level to be a high DG penetration level in the system. when DG. This appears to counteract and compensate the tendency of the system to become more unstable. implementing DG is a natural way of ’limiting’ the power ﬂows in the transmission lines. and through the faulty branch. However. these factors inﬂuence the power system transient stability in a diﬀerent way. are represented in a simpliﬁed way according to models found in literature. and protection schemes of powerelectronic interfaced DG. DG has been implemented as powerelectronic interfaced DG so that it does not contribute to the inertia (stored kinetic energy) of the system.including their active and their reactive power controls and the energy stored in their rotating masses. As expected. the transient stability is better than when only the centralized generation (CG) covers the increasing load. by optimizing the load ﬂow in the system (e. In this way. by rescheduling the output of the CG units). DG gridconnection strengths. In general. the system parameters are easily accessible and it has been widely used in literature. is implemented to cover a load increase in the system. DG based on classical rotating machines is modelled by means of the available models in the software package libraries. Simulation have been performed on the wellknown 39bus New England transmission test system. There appears to be no signiﬁcant stability problem up to about 30% DG penetration level. Therefore. This 30% DG penetration is obtained when DG is implemented to cover a 50% load increase. regardless the technology. . the transient stability of the system that goes through the ‘verticaltohorizontal’ transformation can even be improved. DG technologies. This test system has several advantages: not only does it represents a relatively large power system. Due to the large number of scenarios to be simulated in combination with a relatively large power system. Distribution systems. and an economic dispatch is used in these simulations so that the most ineﬃcient CG unit(s) is(are) taken out of service (eﬃciency considerations). We have simulated the incorporation of DG in the test system with diﬀerent increasing load scenarios. the use of the classical indicators is impractical. with DG incorporated. Userwritten models have been developed for DG based on powerelectronic converters.
Moreover. This source is meant only to gives a (system) frequency reference for the other generators and supplies no power at steady state. for instance. a straightforward remedy for instability problems that may occur due to the DG implementation could be advocated by assigning certain CG units as synchronous condensers.Conclusions 125 The simulation results within this ‘vertical to horizontal’ power system transformation framework also suggest that a DG penetration limit exists and this limit is inﬂuenced by the total kinetic energy remaining in the system.2 Stochastic Stability Studies The merit of stochastic stability studies is evident even in the case of a relatively simple example as studied in this thesis. in addition. For that purpose.g. DG units could provide/emulate ‘inertia’ to the system in order to conﬁne the impact such as unwanted frequency deviations for a largescale DG integration. that can be. 8. a connection (tie line) to an external system. wind and solar energy.1 Remarks and Future Works ‘Inertia’ Contribution When a ‘verticaltohorizontal’ transformation in the power system takes place. the needed reactive power support. the ADS are connected to the transmission system via powerelectronic interfaces too.3. Since the the output power of all generators within the ADS are decoupled from the system frequency due to the powerelectronic interfaces. The stochastic approach becomes a necessity when. some converterconnected DG technologies could be equipped with controllers enabling DG to contribute to the primary frequency control. . DG units are based on (highly) intermittent renewable energy sources e.3 8. The simulation of a power system reaching 100% DG implementation has been done where assumptions are taken that all DG units are implemented via powerelectronic converters within active distribution systems (ADS). and the reactive power support scheme. for example by enabling the extraction of kinetic energy stored in the turbine blades of variablespeed wind turbines or by combining diﬀerent types/technologies of DG units. for example. Therefore. the inertia in the system decreases. or when the dependence between the renewable DG units (power generation) and the loads need to be taken into account. the voltages have been used so that the ADS can detect and maintain the power balance in the power system. that provide inertia to the system and. It has been modeled that the power system is connected to one source providing a constant 50 Hz voltage. 8.
etc).3 Remarks and Future Works 8. FACTS devices. i. synchronous condensers. When the passive distribution systems transform into active ones. However.e. .3.g. providing reactive power to the system may become a problem. both generation and consumption.126 8.2 Reactive Power Control The possibilities for reactive power control within the system are also reduced during the transformation. capacitor banks. As DG is set to maximize the active power generation. the active distribution systems should take over some functions of the shutdown CG units to provide inertia and/or reactive power to the system. various solutions are available for providing reactive power to the system (e.
pu] inertia constant [J/mech rad] number of units active power [W. pu]. or signiﬁcant wave height [m] (Chapter 2). pu] (Chapter 5) error signal. pu] (line) resistance [Ω. diﬀerence between desired and real output.Appendix A List of Symbols and Abbreviations Latin symbols Upper case %DGlevel A B Cp E E(t) G H DG penetration level Area [m2 ] (line) susceptance [S. the power coeﬃcient of the device (turbine) energy irradiance [Wm−2 ] (Chapter 2). or per unit inertia constant [s] (Chapters 36) current [A. pu] power coeﬃcient. pu] I J K L M N P Pr Q R . pu] Probability distribution (Chapter 6) ﬂow rate [m3 s−1 ] (Chapter 2). output power [W. pu] moment of inertia [kgm2 ] constant term (in control blocks) inductance [H. or transient internal voltages of machine [V. or reactive power [Var. conductance [S] eﬀective head of small hydropower plants [m].
pu] impedance [Ω. pu] admittance [S.128 List of Symbols and Abbreviations Upper case. or constant corresponding to CG unit generation (Chapter 5) angular rotor displacement [rad] simulation step [s]. or probability value (Chapter 6) integer proportional to the ”size of DGi ”. pu] Lower case ai bi ci f fi g j n p par(PDGi ) constant corresponding to CG unit generation cost i constant corresponding to CG unit generation cost i constant corresponding to CG unit generation cost i frequency [Hz] energy generation cost of CG unit i (in cost unit) gravitional constant [ms−2 ] √ complex identity ( −1) generator rotational speed [rpm] number of magnetic pole pairs of the ﬁeld circuit of the generator (Chapter 3). or constant corresponding to CG unit generation cost (Chapter 5) altitude angle [rad] (Chapter 2). the active power generated by the aggregate DG at a particular load busi (Chapter 4) slip of induction machine time [s] velocity [ms−1 ] s t v Greek symbols α β δ ∆ η φ ρ σ θ ϕ ω azimuth angle [rad] (Chapter 2). continued S T U U (t) X Y Z apparent power [VA] torque [Nm] voltage [V. usually for time t eﬃciency phase angle [rad] density of ﬂuid [kgm−3 ] standard deviation angle [rad] current phase angle [rad] angular velocity [rads−1 ] .pu] controller output signal (Chapter 7) reactance [Ω.
dev. MCS MV OPF Osc. PE PEC PI PLL PQsource PSS/E PV PWM RMS RTDS Sin Wave Gen SM SMC Alternating Current Active Distribution Systems squirrel cage induction generator Critical Clearing Angle Critical Clearing Time Centralized Generator Combined Heat and Power Direct Current Distributed Generation Distribution Network Equal Area Criterion ExtraHigh Voltage Flexible AC Transmission System Generator (Chapter 7) High Voltage Institute of Electrical and Electronics Engineers Insulated Gate Bipolar Transistor Low Voltage Maximum rotor speed deviation Monte Carlo Simulation Medium Voltage Optimal Power Flow Oscillation duration power electronic interfaced DG without grid voltage and frequency control power electronic interfaced DG with grid voltage and frequency control ProportionalIntegral controller Phase Locked Loop constant active and reactive power generator Power System Simulator for Engineering Photovoltaic Pulse Width Modulation RootMeanSquare Real Time Digital Simulator SineWave Generator synchronous generator without grid voltage and frequency control synchronous generator with grid voltage and frequency control . dur.List of Symbols and Abbreviations 129 List of Abbreviations AC ADS ASM CCA CCT CG CHP DC DG DN EAC EHV FACTS G HV IEEE IGBT LV Max.
continued SMES TL T TN UEP VSC Superconducting Magnetic Energy Storage Transmission Line (Chapter 7) Transformer (Chapter 7) Transmission Network Unstable Equilibrium Point Voltage Source Converter .130 List of Symbols and Abbreviations List of Abbreviations.
Appendix B
Test System Data
Figure B.1: Single line diagram of the 39bus New England test system [49]
132
Test System Data
Table B.1: Bus Data of the New England 39 Bus Test System [49] Bus (nr.) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Volts (pu) 1.0475 0.982 0.9831 0.9972 1.0123 1.0493 1.0635 1.0278 1.0265 1.03 Load (MW) 0.0 0.0 322.0 500.0 0.0 0.0 233.8 522.0 0.0 0.0 0.0 7.5 0.0 0.0 320.0 329.0 0.0 158.0 0.0 628.0 274.0 0.0 247.5 308.6 224.0 139.0 281.0 206.0 283.5 0.0 9.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1104.0 Load (MVAr) 0.0 0.0 2.4 184.0 0.0 0.0 84.0 176.0 0.0 0.0 0.0 88.0 0.0 0.0 153.0 32.3 0.0 30.0 0.0 103.0 115.0 0.0 84.6 92.2 47.2 17.0 75.5 27.6 26.9 0.0 4.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 250.0 Gen (MW) 250 650 632 508 650 560 540 830 1000 Gen (MVAr) 
Test System Data
133
Table B.2: Line Data of the New England 39 Bus Test System [49] Line Bus 1 1 2 2 3 3 4 4 5 5 6 6 7 8 9 10 10 13 14 15 16 16 16 16 17 17 21 22 23 25 26 26 26 28 12 12 6 10 19 20 22 23 25 2 29 19 Data Bus 2 39 3 25 4 18 5 14 6 8 7 11 8 9 39 11 13 14 15 16 17 19 21 24 18 27 22 23 24 26 27 28 29 29 11 13 31 32 33 34 35 36 37 30 38 20 Transformer Magnitude 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1.006 1.006 1.07 1.07 1.07 1.009 1.025 1 1.025 1.025 1.025 1.06 Tap Angle 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Resistance 0.0035 0.0010 0.0013 0.0070 0.0013 0.0011 0.0008 0.0008 0.0002 0.0008 0.0006 0.0007 0.0004 0.0023 0.0010 0.0004 0.0004 0.0009 0.0018 0.0009 0.0007 0.0016 0.0008 0.0003 0.0007 0.0013 0.0008 0.0006 0.0022 0.0032 0.0014 0.0043 0.0057 0.0014 0.0016 0.0016 0.0000 0.0000 0.0007 0.0009 0.0000 0.0005 0.0006 0.0000 0.0008 0.0007
Reactance 0.0411 0.0250 0.0151 0.0086 0.0213 0.0133 0.0128 0.0129 0.0026 0.0112 0.0092 0.0082 0.0046 0.0363 0.0250 0.0043 0.0043 0.0101 0.0217 0.0094 0.0089 0.0195 0.0135 0.0059 0.0082 0.0173 0.0140 0.0096 0.0350 0.0323 0.0147 0.0474 0.0625 0.0151 0.0435 0.0435 0.0250 0.0200 0.0142 0.0180 0.0143 0.0272 0.0232 0.0181 0.0156 0.0138
Susceptance 0.6987 0.7500 0.2572 0.1460 0.2214 0.2138 0.1342 0.1382 0.0434 0.1476 0.1130 0.1389 0.0780 0.3804 1.2000 0.0729 0.0729 0.1723 0.3660 0.1710 0.1342 0.3040 0.2548 0.0680 0.1319 0.3216 0.2565 0.1846 0.3610 0.5130 0.2396 0.7802 1.0290 0.2490 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
134 Test System Data .
1 *SYNPOW.2 Xd 1. 0 X 0 SY N − P OW ∗ 0 Xl 0.5 E2 1.04 Xd 0. Used only to start machine.0) 0.175 D 0 S(1.1: Detailed model unit data for a cylindrical synchronous generator (as CG or DG) [53] Parameter Value Parameter Value Tdo 5.2 X 3. .0 Xq 1.4 Table C.75 S(1.30 Tqo 1. otherwise ignored.20 H 4 Xl 0.2: Detailed model unit data for induction genrator (as DG) [53] Parameter Value Parameter Value T 0.05 Xd 0.Appendix C Generator.75 Tqo 0. Governor and Excitation Systems Data Table C.0 Xq 0.18 0.2) 0. Mechanical Power At Synchronous Speed (> 0).65 Tdo 0.98 E1 1 T 0 S(E1 ) 0 H 3.1 S(E2 ) 0 X 0.
Governor and Excitation Systems Data L ″d – Ll L ′d – Ll + Ef d 1 Td o ′ s – + – – Σ 1 Td o ″ s L ′d – L ″d L ′d – Ll + + Σ Ψ ″d Σ L ′d – L ″d 2 ( L ′d – Ll ) + La d if d Σ + Ψ ″d /Ψ″/ + Ld – L ′d Σ + Ld ′ – Ll daxis + id ∆ La d if d /Ψ″/ Ψ ″q ( Lq – Ll ) / Ψ ″ / ( Ld – Ll ) + Σ – + Lq – L ′q Σ – L ′q – L ″q 2 ( L ′q – Ll ) + + Σ – + Ψ″ iq L ′q – Ll qaxis 1 T ′q o s 1 T ″q o s L ′q – L ″q L ′q – Ll L ″q – Ll L ′q – Ll + Σ + –Ψ ″q Figure C.1: Synchronous generator model block diagram [53] Table C.3: Detailed model unit governing system data for synchronous machine [53] Parameter Value TA /TB 0.05 EM IN 0 EM AX 5 .1 TB 10 K 300 TE 0.136 Generator.
Xl X EK I + + ER ″ 1 T″S T″pθ Figure C. Governor and Excitation Systems Data 137 EK R T″pθ X X ″ .Xl 1 T′S + X′X″ 2 ( X ′ .Xl EK R X′X″ X ′ .Xl X ′ .Xl ) + + 1 T″S EK I X′X″ X′X l + + EI ″ X ′ .Xl X ′ .Xl + X E ″ R E″  + X – X′ + IR E″ X EI ″ E″ +  + X – X′  + II X′X″ 2 ( X ′ .Xl X ″ .Xl ) 1 T′S + E ′R + +  X ′ .2: Induction machine model block diagram [53] .Generator.
91 VM IN 0 T2 2.4: Detailed model unit excitation system data for synchronous machine [53] Parameter Value R 0.1 T3 7 Dt 0 .3: Governor model block diagram [53] VREF + EC p.138 Generator. EMAX  Σ + 1 + TAs 1 + TBs K 1 + TEs EFD VS EMIN Figure C.u.4: Governor model block diagram [53] Table C.05 VM AX 0.05 T1 0. Governor and Excitation Systems Data VMAX Reference + VAR(L) Σ  1 R 1 1 + T1s 1 + T2s 1 + T3s + Σ  PMECH VMIN ∆ω SPEED Dt Figure C.
In mathematical terms.1 Power Flow Problem One of the most common studies in power systems is a load ﬂow (or power ﬂow) calculation. the voltage (magnitude and phase angle) at each bus and the power ﬂow (real and reactive) in each line [25]. jB (D. This computation provides insight in the state of a power system for a speciﬁc steadystate situation. a complex identity represented in the polar coordinate by the magnitude Yij  and the angle θij and in the cartesian coordinate by Gij and Bij . = Gij + ˆ ij . The voltage U at a typical bus i of the system is Ui = Ui ∠δi . Bij are the conductance and susceptance of the element Yij . = Ui  cos(δi ) + j sin(δi ).1) Yij is the admittance element between bus i and j. solving a load ﬂow problem is nothing more than solving a system of nonlinear algebraic equations. as it computes. where the transmission lines – represented by the equivalent circuits and the numerical values for the series impedance Z and the total linecharging admittance Y – are used to determine all elements of the N × N bus admittance matrix of a system (with N buses). with the typical element Yij of this matrix given by Yij = Yij ∠θij .2) . = Yij  cos θij + jYij  sin θij .Appendix D Power Flow Computation D. Solving the load ﬂow problem starts from obtaining the singleline diagram data of power systems [25]. (D. where Gij . among others.
Qi are respectively the active and reactive power injections at node i.5).sch − Qi.7) ∆Qi (D.n=i N Ui Un Yin  cos(θin + δn − δi ). = 0.4) Qi = −Ui 2 Bii + (D. Using (D. 0.calc be the calculated value of Pi .8) In reality. = (Qgi − Qdi ) − Qi. (D. In this case ∆Pi = ∆Qi = 0. = Qi.calc . and ∆Pi = Pi.n=i where Pi .5) n=1. Denoting Pgi as the scheduled power generated at bus i and Pdi as the scheduled power demand of the load at bus i. and Ui  and δi the voltage magnitude and angle at node i.sch = Pgi − Pdi . (D.calc . let Pi. Gii .4) and (D. In the above equations. the net real Pi and reactive Qi power entering the network at typical bus i can be computed.3) According to this representation. (D. Pi.9) ′′ gi = Qi − Qi.10) .sch .sch − Pi. ∆Pi = 0 or ∆Qi = 0.6) Now. and the ′ ′′ powerbalance equations (gi and gi ) can be written as ′ gi = = = Pi − Pi. Ui Un Yin  sin(θin + δn − δi ).sch . In this case. and it is said that a mismatch occurs. the net scheduled power being injected in the network at bus i. Bii are the conductance and susceptance of the element Yin of the admittance matrix. respectively. respectively.140 Power Flow Computation while the current injected into the network at bus i in terms of the elements Yin of Ybus is Ii = = n=1 Yi1 U1 + Yi2 U2 + · · · + YiN UN . Pi − (Pgi − Pdi ). N Yin Un . can be deﬁned as Pi. (D. the load ﬂow equations can be written as N Pi = Ui 2 Gii + n=1.calc . = (Pgi − Pdi ) − Pi.sch . (D. Suppose that mismatch does not occur. = Qi − (Qgi − Qdi ). the calculated values do not always coincide with the scheduled ones.calc . (D.
Pi and Qi . [25]: Remark D. the above stated problem is overdetermined and has an inﬁnite number of solutions. Pi and Qi speciﬁed.1 In Line 7.11) are satisﬁed.10) and (D.2 In line 4. Qi .Power Flow Computation 141 Notice that the subscript i indicates that each bus of a power network has the above two equations.  ·  ∞ = maxi { ·i }. In Line 8. The powerﬂow problem is to determine values for all state variables by solving an equal number of powerﬂow equations based on the input data speciﬁcations. depend on the quantities speciﬁed at all buses. we can formulate the power ﬂow problem as follows: Find Ui . and a basic tool within the PSS/E software [53]. Thus.2. J := A and so on. this inversion may become very time consuming for a large number of buses. There exist three scenarios to choose the speciﬁed variables. .5) such that (D. The remaining are computed. Each bus i in a power system may be associated with four potentially unknown quantities Pi .2 NewtonRhapson power ﬂow solution One method to solve a nonlinear system of equations is the NewtonRhapson method. So. related to three types of buses: 1. Since the slack bus serves as reference for the angles of all other bus voltages. This in general requires inversion of A. As the functions Pi and Qi are nonlinear with respect to δi and Ui . 2. . 2. . Voltagecontrolled bus. Load bus. note that ∆P1 and ∆Q1 of the slack bus are undeﬁned when P1 and Q1 are not scheduled. i = 1. δi and Ui . δi and Ui  speciﬁed. from (D. the inﬁnity norm. describing the state of the system. . Adaptation of this method to the powerﬂow problem is described in the following algorithm 5. Remark D. 3. δi . Pi and Ui  speciﬁed. the unscheduled busvoltage magnitudes and angles are called state variables or dependent variables since their values. all terms involving δ1 and ∆U1  are omitted from the equations because those corrections are both zero at the slack bus. a linear system of type Ax = b has to be solved for x. the NewtonRhapson process is terminated if the maximum absolute value of the corrections is lower than ǫ. As J is a densely populated matrix. In the power ﬂow study. In order to make the solutions uniquely determined.4) and (D. Slack bus. In our case. two of the four unknowns have to be speciﬁed. This method is widely explained in standard numerical analysis books. D. To this end. a powerﬂow calculation should employ iterative techniques.2.
. . . .12) = δi (j) + ∆δi (j) (j) (D.calc . ∂Q2 UN  ∂UN  (j) (D.8) Compute the Jacobian J(j) ∂P2 ∂δ2 (0) J(j) 7: Compute the correction from the following system: ∆δ2 .4) and (D. ∆QN = . .. ∂QN ∂δ2 ∂QN ∂δN ∂P U2  ∂U2  2 . . . δi Set tolerance ǫ for j = 1. ∂PN ∂δN ∂Q2 ∂δN . .14) Ui (j+1) = Ui (j) + ∆Ui  . . . do (j) (j) Compute Pi. . . . . ∂P2 UN  ∂UN  ∂Q2 U2  ∂UN  . . . . ∂Q U2  ∂UN 2 ··· J12 ··· ··· J22 ··· ∂P2 U2  ∂UN  . ··· J11 ··· ··· J21 ··· ∂P2 ∂δN .calc (from (D. . Qi. (j) (j) ∆P2 . ∂PN ∂δ2 ∂Q2 ∂δ2 . ∆UN  (j) (D. (D. . . 2.. . quit Add the correction to the previous value: δi (j+1) (j) ∆δN J(j) ∆U2  . .7). ∆PN = ∆Q2 .13) (D. .142 Power Flow Computation Algorithm 5 NewtonRhapson for powerﬂow problem 1: 2: 3: 4: 5: 6: Guess the initial unknown voltages: Ui (0) . ∆Ui i ∞ < ǫ convergence is reached. .5)) Compute the power mismatches (D.11) 8: 9: if ∆δi ∞ . . ∂P U2  ∂UN 2 ∂Q2 U2  ∂U2  .
Crossley. pp. Feliachi.Bibliography [1] T. A. Electric Power Systems Research. Barten and R. [7] A. E. 591–599. CRC Press LLC. B. Green power: Status and perspectives. [4] H. and C. Kreider. pp. Journal of Power Sources. Distributed Generation the Power Paradigm for the New Millennium. Dynamic model of a microturbine used as distributed generator. [8] R. O’Malley. M. Borbely and J.. 79(12) (2001). [3] D. 195–204. Proceedings of the IEEE. [6] L. 2034–2040. [11] Cigre Technical Brochure TF 38. 91 (1) (2000). Ultracapacitors: Why. Burke. B. Audring and P. Rohatgi. [9] A. G. 89(12) (2001). Honsberg. Electric Power Distribution Reliability. Soder. Modeling new forms of generation and storage. A. and M. in 34th Southeastern Symposium on System Theory. 2001. [2] A. Impact of tidal generation on power system operation in ireland. [10] A. Balzer. [5] M. Tidal power development . 6 pages. Flexibele elektriciteitsopslag. van Gerwen. 1819 March 2002. AlHinai and A. New York. Brown. 57 (2001). Pregeli.a realistic. A. Bernshtein. Andersson. Energietechniek. 1734–1743. F. Bologna. Fox.10. IEEE Transactions on Energy Conversion. J. 2326 June 2003. Begovic. pp. pp. Operating stationary fuel cells on power systems and microgrids. 20(4) (2005). F. Distributed generation: a deﬁnition. and L. pp. IEEE Transactions on Power Systems. justiﬁable and topical problem of today. Italy. Marcel Dekker Inc. 209–213. 37–50. in 2003 IEEE Bologna PowerTech Conference. 2002. Bryans.01. pp. November 2000. Ackermann. P. 10 (3) (1995). . how and where is the technology. 601–605. pp. G.
Coffer. Dagle. J. 14–19. Impacts of the distributed utility on transmission system stability. [14] A. K. 2. 11(6) (2002).On Netz. A. Faulkenberry and W. Frau. pp. Modelling tidal power plant at saphale. Performance comparison of a fuel cellbattery hybrid powertrain and a fuel cell ultracapacitor hybrid powertrain. and G. OFGEM. Microturbines . Cooke. vol. Electrical Power Distribution and Transmission. E. Detroit. 741–746. New Jersey. Introduction to fuel cells and hydrogen technology. McGrawHill Inc. [13] B. IEEE Transactions on Energy Conversion. personal communication on august 1. (2003). 2122 October 2004. S. Trudnowski. N. [21] L. IEE Review. a successful industrialscale experiment. Gatta. [15] M. 10 pages. Cigre Technical Brochure no. Masato. Lauria. Grainger and W. 48(5) (2002). 1719 December 1008. in 1998 IEEE Region 10 International Conference on Global Connectivity in Energy. Enslin. J. 1556–1562. pp. P. Dettmer. E.23. IEEE Transactions on Power Systems. August 2003. Germany. Italy. M. [24] F. pp. IEEE Transactions on Power Systems. J. S. [22] J. pp. Khaparde. 544–547.On Netz GmbH. Michigan. 11(2) (1996). PrenticeHall. and P. and S. pp. Cook. Stevenson. measures to prevent disconnection during system disturbances. IEE Engineering Science and Education Journal. 8(3) (1993). 552–558. . D. TENCON ’98. Rogers. M. Tidel energy: Promising projects la rance. February 1999. in IEEE Young Researcher Symposium: Intelligent Energy Conversions. 1996. [25] J. 1968. J. Power System Analysis. 143–150. USA. S. Donnelly. P. W. 2326 June 2003. and remote load applications. pp. F. [19] J. pp. H. Davis.144 Bibliography [12] Cigre Working Group 37. 205–216. 137. Sukhatme. the Netherlands. R. Lingayat. [18] M. N. [17] R. D. Modelling and computer simulation of dispersed generation in distribution networks. [20] E. Grid Codem. Gao. [23] W. 14(4) (1999). Delft. in 2003 IEEE Bologna PowerTech Conference. Wave energy gets seaworthy. residential. Bologna. Deokar. Interconnection of distributed power to the distribution network. in 2004 IEEE Power Electronics in Transportation. iliceto.. [16] D. Impact of increasing contribution of dispersed generation on the power system.an economic and reliability evaluation for commercial. Bayreuth. 1819 March 2004.
IEEE Spectrum. 394–399. [35] P. 610 June 2004. Vitet. Hansen. D.. [36] W. P.Bibliography 145 [26] G. Denver. Walls. 7 pages. Electricity sector analysis in the clean energy futures study. IEEE Transaction on Power Systems. Laughton. P. pp. Energy Policy. IEEE Transaction on Power Systems. Gotheborg. pp. Paserba. Sweden. Hadley and W. Kundur. 1905–1913. vol. Overview on deﬁnition and classiﬁcation of power system stability. Dynamic frequency control with increasing wind generation. Ritchie. Hebner. S. and M. J. Impact of dispersed generation on power systems. pp. 199 (1999). pp. and G. in 24th Annual International Telecommunication Energy Conference INTELEC. Iov. Allan. and F. 6–13. IEE. [27] S. McGrawHill Inc. [28] A. and M. IEE Engineering Science and Education Journal. 46–51. 1217 August 1990. Flynn. Frequency control and wind turbine technologies. D. Crossley. [31] N. Short. [33] N. [29] R. 12 March 2004. 762–767. Rourke. Overall control strategy of variable speed doublyfed induction generator wind turbine. ﬁrst ed. Lalor. Bridgwater. O’Malley. 7 (1992). 7–16. no. F. J. W. Kundur. Jenkins. Colorado. . Current and nearterm emission control strategies for diesel powered generator sets. 2003. Janssens. R. Sorensen. Blaabjerg. in 25th Intersociety Energy Conversion Engineering Conference IECEC90. ELECTRA. 2. Mullane. New York. in 14th Power Systems Computation Conference. [32] N. A. Kirschen. [37] G. O’Malley. 126–130. J. 20 (2005). 31 in Power Energy Series. Embedded Generation. USA. Strbac. Grassi and A. pp. Fuel cells. D. 11 (2002). 2000. R. in 2004 IEEE Power Engineering Society General Meeting. Sutanto. Beno. 2428 June 2002. 1994. Jenkins. 39(4) (2002). A. pp. [39] M. [34] P. pp.. Lachs and D. pp. Battery storage plant within large load centres. Herzog. 2002. J. [38] G. Impact of power ﬂows on interarea oscillations and mitigation by means of svcs or qpss. V. pp. 7 pages. in CIGRE/IEEE PES International Symposium Quality and Security of Electric Power Delivery Systems. 29(14) (2001). [30] J. pp. 1285–1298. and A. Flywheel batteries come around again. W. and S. 1715–1720. in Nordic Wind Power Conference. 1–4. European research in biomass energy. Lalor. Power System Stability and Control.
33 (2005). PrenticeHall. A. The inertial response of inductionmachinebased wind turbines. Wave energy plants . A. Papaefthymiou. in International Conference on Future Power Systems 2005. Haeseldonckx. 13 March 1997. Schavemaker. W. pp. Luongo. 2730 June 2005. D. 7/1–7/5. D’Haeseleer. pp. Bryans. 1.. 2. 1618 November 2005. M. [46] K. G. vol. vol. Driesen. Reza. 7 pages. 1999. St. Russia. 309–314. pp. 2002. Wind and Solar Power Systems. O’Malley. on Load Representation for Dynamics Performance. and M. R. R. pp. pp. Amsterdam. Stochastic modelling and analysis of horizontallyoperated power systems with a high wind energy penetration. Flynn. Distributed generation: deﬁnition. 8(2) (1993). Energy Policy. 2214–2223. [52] G. in 2002 IEEE Power Engineering Society Winter Meeting. M. [49] M. O’Sullivan. in IEE Colloqium Wave Power: An Engineering and Commercial Perspective (Digest No: 1997/098). 472–483. Pai. New York. in 2004 IEEE Power Engineering Society General Meeting. Patel. [51] M. and W. Kluwer Academic Publishers. J. [41] A. O’Malley. Power. 2168–2173. and M. de Haan. S. H. Modern control engineering. 6 pages. in 2005 IEEE St. T. Morren. Petersburg. pp. Nielsen. 31 January4 February 1999. P. Mullane. CRC Press LLC. Belmans. 787–798. 574–579. Energy Function Analysis for Power System Stability. Meliopoulos. pp. 1496–1503. Tsanakas. Modelling of frequency control in an island system. M. the Netherlands. beneﬁts and issues. 2002. Petersburg PowerTech Conference.the way forward. O’Malley. Ferreira. [48] J. in 1999 IEEE Power Engineering Society Winter Meeting. P. [43] A. 1989. Pepermans. Load representation for dynamic analysis. IEEE Transactions on Magnetics. and L. A. Kinetic energy and frequency response comparison for renewable generation systems. [42] J. [44] A. and J. . 1. pp. 4th ed. IEEE Transactions on Power Systems.146 Bibliography [40] C. [47] I. Ogata. New York. A. [50] G. Challenges in simulation and design of µ grids. 20 (2005). IEEE Transactions on Power Systems. F. van der Sluis. S. Mullane and M. Boston. Superconducting storage systems: An overview. USA. 32 (1996). vol. [45] K. Model reduction and control of electronic interfaces of voltage dip proof dg units. H. 610 June 2004.
Italy. and L. Turin.. Provoost. [63] G. Kling. W. IEEE Power and Energy Magazine. Myrzik. Barcelona. 1 (2003). Robinson. PSS/E 25. W. 1517 February 2005. [60] M. and W. [62] P. Impacts of distributed generation penetration levels on power systems transient stability. Rahman. van der Sluis. IEEE Computer Applications in Power. and L. J. the Netherlands. 3. 1215 May 2003. [58] M. Microhydro in Papua New Guineathe experiences of an electrical engineer. Inc. [59] M. Rabinowitz. CIRED 2003. G. Spain. L. Implementation of renewable electrical energy generation in an urban distribution network: Impacts of energy storage and demand growth. [57] S. Paap. 5 pages.Bibliography 147 [53] Power Technologies. 1541–1545. Green power: What is it and where can we ﬁnd it?. Kling. Reza. P. pp. 1819 March 2004. M. in 17th International Conference on Electricity Distribution. IEEE Power Engineering Review. Spain. Schavemaker. M. G. J. December 1997. H. A research program on intelligent power systems: Self controlling and self adapting power systems equipped to deal with the structural changes in the generation and the way of consumption. van der Sluis. . Demystifying power system oscillations. 4–6. 20 (2000). pp. Jokic. Price. Schavemaker. van der Sluis. in 2000 IEEE Power Engineering Society Summer Meeting. 6 pages. [54] A. CIRED 2005. P. IEE Power Engineering Journal. Kling. W. [61] M. Technologies for energy storagepresent and future: Flow batteries. pp. Barcelona. van Voorden. Self controlling autonomous operating power networks. in 18th International Conference on Electricity Distribution. 9 (1996). Schavemaker. 273–280. A. 6 pages. Power systems of the future (part 4). C. Morren. H. (1988). and L. Rogers. L. P. United Kingdom. van der Sluis. P. [56] M. Schavemaker. Reza. H. Reza. Slootweg. L. H. [55] F. Schenectady.4 Online Documentation. 30–37. London. pp. vol. CIRED 2003. US. Kling. Reza. in IEEE Young Researcher Symposium: Intelligent Energy Conversions. A. 1620 July 2000. 1215 May 2003. 30–35. pp. in 2005 IEE 3rd Reliability of Transmission and Distribution Networks. 69 June 2005. A. in 17th International Conference on Electricity Distribution. 4 pages. and L. A. J. Power electronic interfaced DG units: Impact of control strategy on power system transient stability. 6 pages. Ishchenko. Delft. L.
November 2003. 937–942. Overview of new energy storage systems for an improved power quality and load managing on distribution level. [72] N. 1 (2006). vol. 868–873. pp. 610 June 2004. 4. Sheth and M. 5 pages. pp. USA. Executive overview: Energy storage options for a sustainable energy future. Shahidehpour. Denver. Control and Instrumentation IECON 97. in 2005 IEEE Power Engineering Society General Meeting. and R. Publ No. de Haan. C. Arnhem. Dragu. pp. Illinois. in 2004 IEEE Power Engineering Society General Meeting. Slootweg and W. [68] S. pp. Modeling new generation and storage technologies in power system dynamics simulations. S. [65] S. Belmans. in 2002 IEEE Power Engineering Society Summer Meeting. IEEE Transactions on Energy Conversion. 914 November 1997. 482). Schwartz. Geothermal energy in power systems. G. Schainker. in 2005 IEEE Power Engineering Society General Meeting. USA. 2. USA. L. pp. Patterson. [73] Tennet. Illinois. Kling. 2. L. 11 (1996). Swenson. A kinetic energy tidal generator in the Northern Territory Results. Amsterdam. M. 1216 June 2005. Wind Power: Modelling and Impact on Power Systems Dynamics. Chicago. Kling. 610 June 2004. Belmans. .148 Bibliography [64] R. 2309–2314. the Netherlands. [71] J. in 2004 IEEE Power Engineering Society General Meeting. [67] T. 1821 June 2001. R. Technische gegevens van componenten van het door TenneT beheerde 380 kV. 658–665. Slootweg. vol. 1910–1915. pp. D. 2. G. PhD Thesis. Burns.en 220 kVnet situatie per 1 november 2003. J. 2125 July 2002. Renewable biomass energy. Sriram and M. Small hydro as green power. vol. pp. [74] A. in 2002 IEEE Power Engineering Society Summer Meeting. and J. vol. W. B. T. CIRED 2001. Van and R. 1216 June 2005. Delft University of Technology. 2. Shahidehpour. Colorado. [70] J. 503– 508. V. Distributed generation overview: current status and challenges. Polinder. [69] J. M. Tuckey. G. Pegallapati. vol. and W. Shahidehpour. 178–189. Craenenbroeck. International Review of Electrical Engineering (IREE). pp. Sels. [75] T. Impacts of distributed generation on power system transient stability. the Netherlands. [66] F. in 23rd International Conference on Industrial Electronics. Schoenung and C. 1883–1890. V. H. 2125 July 2002. H. Chicago. pp. 2003. Slootweg. 1972–1977. and M. in 16th International Conference and Exhibition on Electricity Distribution (IEE Conf. Utility energy storage applications studies.
[79] T. Williamson. M. Xue. Proceedings of the IEEE. McGrawHill Inc. V. Belmans. P. G. 127–140. T. 400–412. L. [77] G.. [82] Y. 26 July 2001. 117–120. 69–72. in 2001 European Wind Energy Conference and Exhibition. M. 1719 November 1993. [80] K.Clean Power 2001. Van. 3 (1988). The development of shoreline wavepower in the uk. V. Nijs. Voltage ﬂuctuations on distribution level introduced by photovoltaic systems. Kitz. J. H. Weedy.. A simple direct method for fast transient stability assessment of large power systems. [81] A. Whittaker. 6 (2005). D. Gallup. 21 (2006). pp. International Energy Journal. R. Geothermal power technology. ﬁrst ed. Are wind turbines growing too fast?. RibbensPavella. . 89 (2001). R. in International Conference on Renewable Energy . Interconnection of distributed generators and their inﬂuences on power system. [78] B. M. and K. Gunderson. IEEE Transactions on energy conversion. pp. pp. 202–209. 1783–1792. Van.Bibliography 149 [76] T. pp. and J. pp. IEEE Transactions on Power Systems. van Kuik. T. and M. Woyte. T. A. 1994. Electric Power Systems. and R. pp. van Cutsem. Hamblin. Driesen. J. Belmans.
150 Bibliography .
SeptemberOctober 2006. F. Athens. W. van der Sluis. L. L. W. H. D. G. 2. A. P. 8 pages. Reza. Stochastic Transient Stability Analysis of Power Systems with Distributed Energy Systems. L. 6 pages.In Proceedings of IEEE Young Researcher Symposium. Reza. 6. Reza. Reza. 6 pages. Rodriguez. M. L. C. 2728 April 2006. L. In Proceedings of IEEE PowerTech Conference. O. 3. L. O. Kling. L. L. 7. Schavemaker. Atlanta. P. 8 pages. Amsterdam. P. Vol. Gent. H. 5. W. W. Kling. Georgia. Amsterdam. USA. L. Dynamic Stability of Power Systems with Power Electronic Interfaced DG.Scientiﬁc Contributions Publications 1. M. van der Sluis. Russia. 29 October1 November 2006. W. M. Schavemaker. The Netherlands. Kling. Schavemaker. G. Belgium. A. Reza. Schavemaker. A Transient Stability Studies of a ’VerticaltoHorizontal’ Transformation of Power Systems on a Real Time Digital Simulator. pp. P. 6 pages. W. 1516 April 2005. 4. Gibescu. Papaefthymiou. Kling. St Petersburg. Viawan. H. M. 2327 June 2005. P. H. L. 1618 November 2005. In Proceedings of International Conference on Future Power Systems (FPS) 2005. 479493. No. Sudarmadi. Transient Stability Impacts of a ’VerticaltoHorizontal’ Transformation of Power Systems. P. H. . In Proceedings of International Conference on Future Power Systems (FPS) 2005. Dominguez. Maintaining the power balance in an ’Empty Network’. Reza. Maintaining the power balance in an ’Empty Network’. W. 5. Kling. 16. Kling. A. M. Investigating Transient Stability Impacts of a ’VerticaltoHorizontal’ Transformation of Power Systems. Kling. Dominguez. M. van der Sluis. 6 pages. In Proceedings of CIGRE Symposium. The Netherlands. Greece. European Transactions on Electrical Power. Schavemaker. M. Papaefthymiou. Reza. In Proceedings of IEEE Power Systems Conference and Exposition. 1618 November 2005. M.
H. The Netherlands. L. M. J. J. Denver. van der Sluis. H. In Proceedings of the 4th IASTED International Conference on Power and Energy Systems (EuroPES) 2004. W. In Proceedings of IEEE Power Engineering Society. Reza. Delft. 16. L. L. J. Morren. Schavemaker. M. Paap. W. 1517 February 2005. 1819 March 2004. 2830 June 2004. van der Sluis. L. Barcelona. Slootweg. Kling. Italy. Bologna. van der Sluis. M. Kling. Schavemaker. P. In Proceedings of CIRED 2003 Conference. L. Schavemaker. 910 October 2003. J. M. Kling. 4 pages. Delft. General Meeting. M. van der Sluis. 4 pages. Kling. van der Sluis. Reza. L. P. Reza. Investigating Network Constraints (Congestions) of HorizontallyOperated Power Systems with Stochastic Distributed Generation. P. 15. L. L. 6 pages. 1215 May 2003. 4 pages. L. M. Reza. Schavemaker. C. Papaefthymiou. Investigating Impacts of Distributed Generation on Transmission System Stability. G. London. Slootweg. Papaefthymiou. Rhodes. Potential for Transmission Lines Losses Reduction in Electrical Power System Operation with Distributed Generation. P. Greece. Kling. The Netherlands. 6 pages. Impact of Distributed Generation Grid Connection Strength on Power System Transient Stability. L. P. Colorado. Morren. 611 June 2004. Reza. 1517 February 2005. Implementation of Renewable Electrical Energy Generation in an Urban Distribution Network: Impacts of Energy Storage and Demand Growth. Slootweg. Schavemaker. Schavemaker. W. van Voorden. G. Kling. Kling. In Proceedings of IEEE PowerTech Conference. M. van der Sluis. Impacts of Converter Connected Distributed Generation on Power System Transient Stability. W. L. 6 pages. J. P. H. 2326 June 2003. H. Schavemaker. G. London. Reza. 9. H. Spain. M. van der Sluis. Power Electronic Interfaced DG Units: Impact of Control Strategy on Power System Transient Stability. Schavemaker. 7 pages. United Kingdom. A. L. Reza. In Proceedings of 3rd Reliability of Transmission and Distribution Networks (RTDN) 2005. 11. In Proceedings of IEEE Young Researcher Symposium: Intelligent Energy Conversions. 12. United Kingdom. H. L. G. G. In Proceedings of 3rd Reliability of Transmission and Distribution Networks (RTDN) 2005. In Proceedings of 8th Indonesian Student Scientiﬁc Meeting 2003 Conference. L. G. 13.152 Scientiﬁc Contributions 8. W. 10. 14. A Research Program on Intelligent Power Systems: Self Controlling and Self Adapting Power Systems Equipped to Deal with the Structural Changes in the Generation and the Way of Consumption. 6 pages. Impacts of Distributed Generation Penetration Levels on Power Systems Transient Stability. Schavemaker. L. H. Reza. W. van der Sluis. M. P. Reza. M. G. P. W. USA. . H. H. van der Sluis. J. P. Slootweg. L.
17. P. C. The Impact of Distributed Generation on Voltage Dip and Overcurrent Protection Coordination. 2327 June 2005. pp. M. M. Papaefthymiou. St Petersburg. Talks 1. Tsanakas. 2006 (poster). 8 pages. P. 2. Solving Network Constraints (Congestions) by Investigating Impact of Phase Shifters on the Flows. 29 October1 November 2006. Indonesia. a as. 7 pages. Brussels. Reza. In Proceedings of IEEE PowerTech Conference. M. DC Interconnection between Java and Sumatera. L. L. United Kingdom. G. A. van der Sluis. 1821 July 2006. Bali. van der Sluis. Atlanta. Spain. Bali. ABB Corporate Research. M. Joelianto. M. E. 1517 February 2005. Russia. Panel Discussion on “Reliable Power . 6 pages. Georgia. 2006. van der Sluis. V¨ster˚ Sweden. Mostavan. Tsanakas. Paap. Discrete Event Controller Design of a Solar PV Water Pumping System using Signal Interpreted Petri Net. M. 20. London. Barcelona. L. 2528 June 2002. W. L. Reza. A. M. 189193. Belgium. Grece. In Proceedings of International Conference on Future Power Systems (FPS) 2005. Reza. Sudarmadi. H. A. Stochastic Modelling and Analysis of Horizontally Operated Power Systems with High Wind Energy Penetration. 1618 November 2005. 1821 July 2006. May 10. 6 pages. USA.Scientiﬁc Contributions 153 In Proceedings of CIRED 2003 Conference. Indonesia. G. 6 pages. D. A.Europes Strength”. Schavemaker. Papaefthymiou. ‘Empty Network’. In Proceedings of the 6th Asian Control Conference. L. . In Proceedings of IEEE Power Systems Conference and Exposition. Crete. F. G. Kling. 1215 May 2003. May 18. A. 22. Reza. van der Sluis. Schavemaker. In Proceedings of 3rd Reliability of Transmission and Distribution Networks (RTDN) 2005. 19. E. Joelianto. Mostavan. Reza. van der Sluis. A Discrete Event Controller for PVWind Hybrid Energy System using Signal Interpreted Petri Net. in Indonesia. 6 pages. The Netherlands. In Proceedings of the 2nd IASTED International Conference Power and Energy Systems (EuroPES) 2002. Impacts of Distributed Generation and Distribution Systems on Power System Transient Stability. 18. 23. Viawan. L. 6 pages. Reza. Reza. Amsterdam. H. Reliability Assessment of HV/MV Transformer Links for Distributed Power Systems Planning. In Proceedings of the 6th Asian Control Conference. 21.
An Example of Utilizing a Stochastic Approach to Elaborate the Investigation of the Impacts of Distributed Generation on Power Systems. 7. March 45. Germany. 2005. Delft. 6. The 8th Indonesian Student Scientiﬁc Meeting. Inherently Stable Transmission Systems on a VerticaltoHorizontal Transformation of Power Systems with Distributed Generations. The Netherlands. Hamburg. The 9th Indonesian Student Scientiﬁc Meeting. Germany. Arnhem. 2004. The Netherlands. Developing Indonesian Electricity with Distributed Generation. . The Netherlands.154 Scientiﬁc Contributions 3. 5. Aachen. Lunch Lecture KEMA. 2003. October 79. Potential of RenewableEnergyBased Decentralized Generation in Power Network for Carbon Dioxide Emission Reduction. Potential for Distributed Generation in Indonesian Electrical Power Systems. 4. 2004. 2006. Delft. October 21. The 3rd Indonesian Applied Mathematics Society in The Netherlands (IAMSN) Seminar on Applied Mathematics. February 16. October 910. International Conference on Indonesia Toward 2020.
My special thanks also go to my daily supervisor dr.ir. A.Acknowledgment This thesis describes my work that was performed during my appointment at the Electrical Power System (EPS) Research Group. the chairman of the IOPEMVT program committee. the Netherlands. The ’Intelligent Power Systems’ research program is conducted by the Electrical Power Systems and the Electrical Power Processing Research Groups of the Delft University of Technology (TU Delft) and the Electrical Power Systems and Control System Research Groups of the Eindhoven University of Technology (TU Eindhoven). Schavemaker mentioned earlier and furthermore ir. For this. H. H. Antal (Technische Universiteit Eindhoven. H. and especially for giving me a great opportunity to do research in the EPS Research Group. Gibescu (Madeleine) for fruitful discussions and encouragement during the research. J.dr. ir. P.ir. Faculty of Electrical Engineering. M. prof. M. Myrzik (TU Eindhoven’s Electrical Power Systems). Boltje (SenterNovem). W. Upon completion.dr. Antal. Mathematics and Computer Science. Belgium) and prof.ir R. I would like to thank again prof. my appreciations should go to prof. SenterNovem is an agency of the Dutch ministry of Economic Aﬀairs. This research has been performed within the framework of the research program ’intelligent power systems’ of the IOPEMVT (Innovatiegerichte Onderzoeksprogramma’s . I am indebted to a number of people for their direct or indirect contributions.dr. de Haan (TU Delft’s Electrical Power Processing). prof. M. Ferreira (TU Delft’s Electrical Power Processing). ﬁnancially supported by SenterNovem. M.ElektroMagnetische VermogensTechniek) Program. the program coordinator.ir. Kling (Wil) and prof. L. van der Sluis. Delft University of Technology (TU Delft).dr.dr. and dr. Many encouraging discussions were done during the appointment with highly experienced people in the ﬁeld. J. van der Sluis (Lou) for being the promotor and the copromotor of my PhD research. emeritus). G. Belmans (Katholieke Universiteit Leuven. Kling. O’Malley (University College Dublin. Blom (Technische Universiteit Eindhoven). prof. A. M. prof. I would like to specially thank prof.ir.ir. S.ir.ir. Schavemaker (Pieter) and dr. van den Bosch (TU Eindoven’s Control . dr. and the leaders and supervisors of the project. L. J. prof. the Netherlands. J.ir. prof. W. W.ir P. In this respect. Ireland) of the promotion committee for their time to carefully read the manuscript.ir.
I would like to acknowledge Yogi Erlangga. I would like also to acknowledge Sheikh Sharief and the Suﬁ brothers and sisters in Rotterdam. and Bob Paap. and I Made Ro Sakya.156 Acknowledgment Systems). Dedy Wicaksono. Barbara Slagter. It was so often that these discussions paved the way to better understand many aspects that were and are still beyond my knowledge. Muhamad Reza Delft. my sister Mia Miranti and my brother Muhammad Lukman for their sincere support and prayers so that I could make my way up to where I am now. Most of all. who was kind enough to share his knowledge and experience with me. Arjan van Voorden. Didik Sudarmadi. the Ph.D. who can erase my problems just by smiling and laughing. Many thanks also go to those who support me with their hospitality and friendship during my stay in the Netherlands. Datuk Ary Samsura. and the younger Averrouz Mostavan. Ezra van Lanen. Slootweg’s thesis. with whom I discussed scientiﬁc subjects and technical experiences we faced in the IOPEMVT project. Johan Morren. Special thanks should go to my wife Novi Ineke Cempaka Wangi for her endless support and encouragement throughout the diﬃculties I was facing in this work. the EPSmates. with whom I experienced a highly stimulating and pleasant working atmosphere during my stay at the EPS group. and discussed our hopes and dreams in the future. A. Cai Rong. and Sjef Cobben. the Creator who creates all these nice people and the pleasant opportunities. Damen (TU Eindhoven’s Control Systems). Boy Fadhilah. Marjan Popov. H. students. Uly Nasution. Frans Provoost. Roald de Graaﬀ. Andrej Jokic. G. Jody Verboomen. Ikshan Rashad and Yusuf Maury who sincerely helped me with practical things from repairing bikes to reinstalling crashed computers. with whom I shared interests in many subjects from the very technical to the philosophical and spiritual ones. I am also grateful to my grand parents and my parentsinlaw for their constant support and encouragement. with whom I shared colorful and enjoyable experiences on life and spirituality.ir. Anton Ishchenko. as well as to my son Muhammad Rifqi Aulia Yahya. My thanks are also to George Papaefthymiou. since the names would make a couple of pages! Nevertheless. for the scientiﬁc discussions and the technical feedback from the ‘real world’.ir. Special thanks are also once more to George Papaefthymiou.D. The students Alejandro Dominguez and Cristovo Rodriguez are acknowledged for their contributions to this research. I wish to deeply thank my parents. I found it not easy to mention all their names. Jan Heydeman. A. Johan Vijftigschild. my oﬃcemate. I ultimately praise and thank Allah the Almighty. students in Delft. Bart Ummels. Diah Chaerani and Anita Pharmatrisanti with whom I shared experiences as Indonesian Ph. Ralph Hendriks. Boukje Ypma. Summer 2006 . Tirza Drizi. J. and also becoming nice friends for chatting during our interaction in Delft. The initial part of this thesis is based to some extent on dr. and dr. all in all. Zulﬁkar Dharmawan.
he obtained a B. From 19941997 he was awarded fellowship by ToyotaAstra Foundation. In 2000 he was awarded the Best Grade Average of the Master of Science International Program 19982000. In 19971998 and 20002002 he was with the Electrical Power System and Distribution Laboratory. Indonesia for his B. the Netherlands. he joined Electrical Power System Research Groups. study.Sc.Sc.Sc. From 1998 until 2000 he attended Delft University of Technology (TU Delft). Indonesia. an agency of the Dutch Ministry of Economic Aﬀairs. Department of Electrical Engineering. Manila. chaired by Professor Lou van der Sluis. and in PT Tesla Daya Elektrika. and Pertamina. Bandung. from where he received an M. Indonesia (2001). Indonesia in 1993. study. Muhamad Reza is married to Novi Ineke Cempaka Wangi and has a son. ITB. the capital of the province of West Java. a as. in Bandung. Indonesia (1997). for three months. In 1997. TU Delft. In 1997 he was awarded Ganesha Prize as the best ITB student Year 1997 (highest in the university). Indonesia (2001). at the Faculty of Electrical Engineering. degree in Electrical Engineering from Bandung Institute of Technology (ITB) with honor (cum laude). Since February 2002. His work experience include student internships in Schlumberger Wireline. Indonesia (2001). From 19982000 he was awarded a fellowship within the Highly Talented Indonesian Student (Thalis) Program for his M. the Philippines (1996).Biography Muhamad Reza was born on November 4. at TU Delft. 1974. degree in Electrical Engineering with honor (cum laude). V¨ster˚ Sweden. He ﬁnished the secondary school education at SMA Negeri 3 Bandung. and in 1998 he received a grant from Bandoengsche Technische Hogeschool Fonds (BTHF) where he used it for working on a research activity at the Electrical Power System Research Group (EPS). Mathematics and Computer Science as a PhD student within the framework of the research program ’Intelligent Power Systems’ supported ﬁnancially by SenterNovem.Sc. He is currently working as a Research & Development scientist in power systems at ABB Corporate Research. Muhammad Rifqi Aulia Yahya. where he involved in some projects with PLN. . YPF Maxus.