Professional Documents
Culture Documents
Ioannis-Thomas K. Theologitis
Degree project in
Electric Power Systems
Second Level
Stockholm, Sweden 2011
XR-EE-ES 2011:011
Comparison of existing PV models and
possible integration under EU grid
specifications
Ioannis-Thomas K. Theologitis
Ioannis-Thomas Theologitis
Royal Institute of Technology (KTH), Sweden
©2011
School of Electrical Engineering
Kungliga Tekniska Högskolan
SE-100 44 Stockholm
Sweden
The author is officially enrolled to the Sustainable Energy Engineering Master Program
(SEE) and belongs to the School of Industrial Engineering and Management and the
Department of Energy Technology.
The picture of the front cover is taken from the second edition of the book “Planning &
Installing Photovoltaic Systems – A guide for installers, architects and engineers”, published
by Earthscan and copyrighted by The German Energy Society (Deutsche Gesellshaft für
Sonnenenergie (DGS LV Berlin BRB) in 2008. ISBN-13: 978-1-84407-442-6
“Αποσκότισόν με”
-Διογένης-
It was the answer to Great Alexander, when he stood in front of Diogenes and asked him what favour
he needs. Diogenes, as a cynic philosopher, answered this phrase to Alexander implying that he was
blocking the sun with his body. Cynics believed that the happiness is hidden in simple things as the
energy and warmth of the sun and not in material goods.
MSc Thesis Project Abstract
KTH, June 2011
Abstract
This master thesis investigates the capabilities of a generic grid-connected photovoltaic (PV)
model that was developed by DIgSILENT and is part of the library of the new version of
PowerFactory v.14.1. The model has a nominal rated peak power of 0.5 MVA and a designed
power factor cosφ=0.95. A static generator component, which includes the PV array, the DC
bus with the capacitor, the inverter and the control frame, is used to model the PV system.
The PV array is considered to operate at the MPP and the generator with cosφ=1.
The thesis begins with a short review of the current status of the PV sector, focusing mostly
on the types of PV systems and the necessary components that are used in grid-connected
systems. Since the PV inverter is the key component, special reference is made to the
different technologies applied and to the multifaceted role that inverters should play
nowadays supporting the grid’s stability. Technical restrictions and requirements are
presented highlighting primarily the German Grid Code for the MV network, which is the
benchmark for the analysis of the role and behaviour of the PV model in question. Germany
is regarded a very good example to base the study on due to its leading position and
experience in the renewable area and its thorough grid specifications.
The main part of the report includes a detailed description of the structure of the generic
model, presenting the operating procedure of its components as well as model assumptions
and simplifications. Various simulations in variable solar irradiation, frequency and voltage
conditions are performed in order to conclude in its capabilities. The static voltage support
is investigated under cloud effect situation where the changes in active power output of the
PV array can influence the voltage stability of the grid at the PCC. The active power control
is examined by forcing the grid frequency to deviate beyond specified limits and observing
the active power output results. At last, the dynamic voltage support capability (LVRT) is
examined by simulating four different short circuit events creating four different voltage
dips. The ability of the PV inverter to stay connected and to provide reactive current when
necessary is seen. The external grid component is designed to represent a strong grid.
The results showed that the model is capable for active power reduction and LVRT
behaviour. However, the absence of reactive power control makes it inapplicable for static
voltage support. Thus, a PI controller is implemented in order to supply constant reactive
power in steady state operation and support the grid stability.
At last two different interconnections were built using a slightly modified version of the
same generic model with a rated power 1 MVA. The control scheme remained the same.
Both configurations were examined statically and dynamically and their results were
compared. Small differences were found in terms of reactive power consumption/injection
at the PCC.
Keywords: Grid-connected Photovoltaic, PV inverter, Grid codes for PV, PV model, DIgSILENT
iv
MSc Thesis Project Sammanfatning
KTH, June 2011
Sammanfatning
Examensarbetet inleder med en kort genomgång av det nuvarande läget av PV sektorn, som
fokus för det mesta på PV system sorter och de viktiga beståndsdelarna som användas i
nätanslutna system. Eftersom PV strömväxlaren är den viktigaste beståndsdelen, är särskild
hänvisning görs till de olika tillämpade tekniker och den mångfacetterade roll som
växelriktare bör spela nuförtiden stödja nätets stabilitet. Tekniska begränsningar och krav
presenteras för att belysa främst på den tyska GC för MV nätet, vilket är utgångspunkten för
analysen av den roll och beteende av PV modellen i fråga. Tyskland anses ett mycket bra
exempel att basera studien på grund av sin ledande ställning och erfarenhet inom förnybar
området och dess grundliga specifikationer nätet.
Resultaten visas att modellen har kapacitet för aktiv effekt minskning och LVRT beteende.
Men gör det saknas styrning av reaktiv effekt inte tillämpas under statisk spänning stöd.
Därför är en PI-regulator som genomförs för att leverera konstant reaktiv effekt i konstant
drift och support för stabila nät.
Äntligen två olika sammankopplingar byggdes med en något modifierad version av samma
generiska modell med en nominell effekt 1 MVA. Kontrollschemat förblevs densamma. Båda
konfigurationerna undersöktes statiskt och dynamiskt och resultaten jämfördes. Små
skillnader fanns i form av reaktiv effekt förbrukning / insprutning i PCC.
Nyckelord: Nätanslutna solcell, PV strömväxlare , Nät koder för PV, PV modell, DIgSILENT
v
MSc Thesis Project Acknowledgements
KTH, June 2011
Acknowledgements
At first I wish to thank all the people, who, in whichever way, assisted me to complete this
interesting project for my master thesis and made this period an important benchmark for
my future professional expectations. From Energynautics GmbH1, Dr Thomas Ackermann,
who gave me the opportunity to complete the thesis in his company, Dr Eckehard Tröster,
for his patience with all my questions, his valuable advices and insight that gave direction to
my work, Rena Kuwahata, who was the initial contact with the company and the person that
facilitated my work and life in the new environment, Dr Nis Martensen and Stanislav
Cherevatskiy, who shared their experiences in the field whenever those were asked for and
in general I wish to thank all the rest of the personnel, who were part of my everyday life
the last five months, ensuring a friendly and highly professional environment.
Furthermore, I would like to thank Prof. Lennart Söder for the fruitful pre-presentation and
his valid comments and of course Giannis Tolikas for undertaking the translation of the
abstract to Swedish. Since it is likely that I forget some people that offered their helped for
the completion of this project, I feel obliged to thank them as well.
Special thanks should be paid to Panagiotis Giagkalos and Kyriakos Liotsios, who were my
classmates, colleagues, but most of all my friends during the last two years of this master. It
is important to realize that anytime you can find people that you can count on. May this
friendship lasts and don’t leave time and distance to wear it, rather strengthen it through
personal or professional common experiences.
To Angela Maria Castaño Garcia, for this beautiful journey that still goes on. Her support
during this time was more that I could ask for. As far as the thesis concerned, her
contribution and effort to the final format of the report was significant.
Such moments, I feel the need to give space and mention all the people that left something
valuable to me. Old friends from Greece that never forget, new friends from different parts
of the world, people I met for short period, all of them are the people that with one way or
another made this time worth living it again. You are my personal ark. Thank you all and
wish you the best.
1 http://www.energynautics.com/
vi
MSc Thesis Project Table of Contents
KTH, June 2011
Table of Contents
ABSTRACT......................................................................................................................... IV
SAMMANFATNING ........................................................................................................... V
ACKNOWLEDGEMENTS .................................................................................................... VI
TABLE OF CONTENTS ...................................................................................................... VII
LIST OF FIGURES .............................................................................................................. IX
LIST OF TABLES ............................................................................................................... XII
NOMENCLATURE ........................................................................................................... XIII
1 INTRODUCTION......................................................................................................... 1
1.1 THE DRIVING FORCE ....................................................................................................... 2
1.2 OVERVIEW OF THE THESIS REPORT ................................................................................ 4
1.3 OBJECTIVES .................................................................................................................... 5
1.4 LIMITATIONS .................................................................................................................. 6
2 BACKGROUND .......................................................................................................... 8
2.1 PV SYSTEMS – OVERVIEW .............................................................................................. 8
2.1.1 I-V CHARACTERISTICS ............................................................................................. 9
2.2 GRID-CONNECTED PV SYSTEMS ................................................................................... 11
2.3 PV INVERTER ................................................................................................................ 12
2.3.1 WHAT IS AVAILABLE – CURRENT STATUS ............................................................ 13
2.3.2 ISSUES WHEN CHOOSING INVERTER ................................................................... 16
2.3.3 ADDITIONAL REQUIREMENTS – ANCILLARY FUNCTIONS .................................... 18
2.4 LOW VOLTAGE RIDE THROUGH (LVRT) REQUIREMENT ............................................... 18
2.4.1 REACTIVE POWER AND ITS IMPORTANCE ........................................................... 19
2.5 GRID REQUIREMENTS FOR PV SYSTEMS ...................................................................... 19
2.5.1 THE NEW GERMAN GRID CODE ........................................................................... 20
2.5.2 THE SITUATION IN THE REST OF EUROPE ............................................................ 25
2.5.3 FURTHER INTERNATIONAL AND EUROPEAN REQUIREMENTS FOR PV ............... 25
3 METHODOLOGY ...................................................................................................... 27
3.1 DESCRIPTION OF THE TOOLS ........................................................................................ 27
3.2 WAYS FOR SIMULATING PV WITH POWERFACTORY .................................................... 27
4 MODEL DESCRIPTION .............................................................................................. 31
4.1 THE BASE MODEL ......................................................................................................... 31
4.2 THE PV GENERATOR ..................................................................................................... 33
4.2.1 THE CONTROL FRAME OF THE PV GENERATOR ................................................... 36
4.3 INVESTIGATION UNDER GERMAN GCS......................................................................... 48
4.3.1 STEADY STATE CONDITION .................................................................................. 48
vii
MSc Thesis Project Table of Contents
KTH, June 2011
viii
MSc Thesis Project List of Figures
KTH, June 2011
List of Figures
ix
MSc Thesis Project List of Figures
KTH, June 2011
x
MSc Thesis Project List of Figures
KTH, June 2011
FIGURE 8.12: BEHAVIOUR OF THE SECOND INTERCONNECTION IN 50% VOLTAGE DIP .......... 95
FIGURE 8.13: BEHAVIOUR OF THE SECOND INTERCONNECTION IN 20% VOLTAGE DIP .......... 95
xi
MSc Thesis Project List of Tables
KTH, June 2011
List of Tables
xii
MSc Thesis Project Nomenclature
KTH, June 2011
Nomenclature
AC Alternate Current
AI Anti-Islanding
AM Air Mass
ASTM American Society for Testing and Materials
BDEW Bundesverband der Energie-und Wasserwirtschaft (Federal Association of
Energy and Water)
CSI Current Source Inverter
DC Direct Current
DIgSILENT DIgital SImuLator for Electrical NeTwork
DIN Deutsches Institut für Normung (German Standardisation System)
DSL Dynamic Simulation Language (DIgSILENT Simulation Language)
eEURO European efficiency
EMI Electromagnetic Interference
EN European Normalization
EU European Union
FB Full Bridge
FiT Feed in Tariff
FRT Fault ride through
GCs Grid Codes
HV High Voltage
IEA International Energy Agency
IEC International Electrotechnical Commission
IEEE Institute of Electrical and Electronic Engineers
IGBT Insulated Gate Bipolar Transistor
IK or ISC Short circuit current
Impp Current at maximum power point
MOSFET Metal Oxide Semiconductor Field Effect Transistor
LV Low Voltage
LVRT Low Voltage Ride Through
MPP Maximum Power Point
MV Medium Voltage
NPC Neutral Point Clamped
PCC Point of Common Coupling
PF Power Factor
PLL Phased Locked Loop
p.u. per unit
PV Photovoltaic
PVPS Photovoltaic Power Systems
PWM Pulse Width Modulation
RET Renewable Energy Technology
RMS Root Mean Square
xiii
MSc Thesis Project Nomenclature
KTH, June 2011
xiv
MSc Thesis Project Introduction
KTH, June 2011
1 Introduction
In the coming decades and taking into account the continuous population growth, the
energy demand will probably double [1], if not more, bringing the societies to the brink of
energy shortage. Even if the improvement of energy efficient technologies is significant, the
future demand will not be able to be balanced if new sources would not be introduced and
innovative technologies (either passive or active) would not be exploited. Due to the use of
fossil fuels, major side-effects both in the environment and social life, have already caused
an outburst, sounding the alarm for cleaner and carbon-free energy sources. The renewable
energy sources, as carbon-free sources, appear to be a feasible alternative to conventional
fuels. This shift is not a current phenomenon and mankind has already exploited the sun,
the wind, the water and the earth to produce clean energy in order to address its needs and
provide better quality of services.
Since 1997 and the introduction of the White Paper by the European Commission, the
formation of a renewable energy policy has begun. The overall objective was to reduce the
dependence on fossil fuel imports and increase the security of supply moving towards a low
carbon economy [2]. Over these years the orientation of the EU has changed from indicative
targets, referring to electricity and transport fuel, to specific targets that are legally
established by a legislation pattern. What is more, change has occurred towards redefinition
of the infrastructure policy that plays a key role to the growth of Renewable Energy
Technologies (RET) [3]. Nowadays, it is common belief from all the stakeholders involved
(government, producer and the end user) that the benefits of a society, where renewable
sources account increasingly to the consumption needs, are multiple. Strengthening the
national and local economy, jobs creation, better life quality and of course less harmful
contribution to the environment are some of the strongest arguments in favour of RET.
In 2001, the EU “Renewable Electricity Directive” together with the “Biofuels Directive” that
was signed two years after, set quite ambitious goals for the member states by 2010.
Unfortunately, the 21% of renewable electricity production was met only by very few
countries (i.e. Denmark, Germany, Poland et al.), however it gave boost to many economies
and the sector of renewable energies experienced significant growth with the ‘electricity
production’ enjoying the biggest share [3]. In the year 2009 almost 61% of the new
electricity generating capacity that was connected to the grid in the EU was from RET [4],
while in 2010 the total electricity share in the EU reached 18.5%. This number, even if it is
promising, is still far from the 37% that the Member States have set for 2020. Nevertheless,
today the conditions are more favorable for higher and faster growth rates. There is a much
more organized research field dedicated to the renewable technologies, better and more
flexible legislation framework with number of incentives and support mechanisms and also
a more open-minded industry.
It is inevitable though that with changes, new problems are acquired and need to be
addressed. The high penetration of renewables in the electricity supply system could create
1
MSc Thesis Project Introduction
KTH, June 2011
issues of instability to the grid, which is not designed to receive such integration. The
problem requires an immediate attention in order to meet the target of 2020 or even
exceed it. There are researches [4] that present scenarios for 2030 and 2050, when
renewable electricity supply could be 68% and 100% respectively. It is understood then that
more modern electricity grid system should be adopted, which means expansion of the grid
but more important implies urgent modification of the already existing one.
Improved technical specifications, the so-called Grid Codes (GCs), which will ensure the
proper and safe function of the electrical grid should be introduced and facilitate the
interconnection of electricity systems and the reinforcement of the grid. Problems like
bottleneck in the grid should be overcome so as grid operators to exchange kWh when
excess of electricity is produced from one and is needed by another one. Deregulation-
based energy market for the support of the distributed generation can be the compass of
the reformation of energy scenery.
Among the RET, the lights have turned to wind turbines and solar PV technologies. Taking
into consideration the fact that countries such as Germany and Spain enjoy a leading
position in the renewable energy sector, tangible results can be withdrawn about the
situation in Europe in general if those countries used as study cases. In figure 1.1 the
aggregated renewable energy capacity in Germany is shown, proving that mainly the focus
is on the wind technology and PV. As far as the PV installations concerned, there is a
dramatic change since 2004, when the feed-in tariff policy mechanism and relevant
subsidies have been in effect. Only in 2008 around 1.5 GWp were installed in Germany [5],
while in September of 2010 the total number of installed capacity was 15 GWp, which is the
30% of the total RET installed and the 37.5% of the minimum load of electricity in 2009 as is
seen in figure 1.1.
2
MSc Thesis Project Introduction
KTH, June 2011
In 2009, an improved version of the feed-in law was introduced specifying that for new PV
power plants the feed-in tariffs will be reduced from 8% to 10% per year. The main reason
for this change was to force the reduction of the investment cost in PV systems and lead to
grid parity [5].
From the above, it is understood that the conditions are very favourable for the expansion
of RET and especially of PV systems, which is the concern of this project. The Thesis focuses
on grid connected PV systems and their advantage as a power generation unit. The
tendency in industrialized countries is to connect the PV systems to the grid since there is
almost everywhere an electrical network available. Figure 1.2 illustrates this tendency. The
multifunctional role of the PV system and specifically of the grid tied inverter is highlighted
in the Thesis. PV inverter is the main component of the system and is responsible for the
power injection to the grid. Until now, its conventional role was to convert the DC power to
AC power and feed-in the maximum possible active power to the grid. Moreover, in case of
a grid failure it was designed to disconnect until the conditions stabilize again to reconnect.
However, the high penetration of photovoltaics to the distribution network has raised new
requirements for the modern PV inverters. Their role has become much more significant not
only for the PV system but now also for the grid that is connected to.
Figure 1.2: Cummulative installed grid connected and off grid PV power in 26 countries that participate in the
IEA PVPS [6]
The phenomenon of PV being one of the fastest growing sector in the RET industry is not
only European but worldwide. In figure 1.3 is obvious that there is a steady rise in PV
production around the world, which is being followed by a steady rise in PV installations,
which can be seen in figure 1.4.
3
MSc Thesis Project Introduction
KTH, June 2011
The new setting that is being shaped because of the reasons mentioned so far makes the PV
field even more interesting and every study around on-grid systems more challenging.
The project in question is part of the initiative of Energynautics GmbH2 to fulfil a study
concerning modelling and simulating large scale PV systems in relation to their impact on
4
MSc Thesis Project Introduction
KTH, June 2011
the power system. The project’s objectives are limited due to tight time constraints and are
presented below in the same chapter. Nevertheless, the thesis tries to cover important
aspects in theory and present realistic results through iterative simulations, aggregating
some knowledge as far as the characteristics of larger penetration concerned.
In the following chapter the theoretical approach is undertaken and basic background is
presented. A brief overview of PV systems and some important characteristics are included
and give place to a deeper analysis of the PV inverter and its modern role. A major part of
this chapter is being covered by the reference of current GCs, principally those of Germany.
In the third chapter the methodology of the study is explained as well as the tool that was
used to model the PV and perform simulations.
Chapter four is dedicated to the model with capacity 0.5 MVA that is used in this thesis and
built by the company DIgSILENT. The control system is explained thoroughly and the choice
of the configuration, inputs, outputs and parameters is justified. Its ability to be used as a
generic model for PV systems that comply with the German GCs is investigated through
simulations.
1.3 Objectives
The main aim of this thesis, presented also above, is to group together some basic
knowledge, concerning the requirements that a grid connected photovoltaic system should
fulfill in order to comply with certain codes, particularly with the German GCs, and how the
integration of such decentralized power systems affect the behavior of the distribution
network. More specifically, the objectives set for this project are:
5
MSc Thesis Project Introduction
KTH, June 2011
and mainly the provisions of the new German Transmission Code for Medium
Voltage networks.
• Give a description of the tool used to model the full PV system, meaning the
network, the PV generator, the inverter and the control system. The tool used for
the simulations is the PowerFactory of DIgSILENT.
• Train for the use of PowerFactory and understand number of features that are useful
for modeling and simulating PV systems.
• Understand and analyze a generic PV model and its compliance with the German
GCs. Furthermore, examine, through simulations the shortcomings of the model.
• Create two different interconnected PV configurations with the use of the modified
PV model and study their behavior in case of faults.
1.4 Limitations
In the thesis, due to the broadness of the subject, several limitations were delineated with a
sole purpose of presenting some facts in a conceivable and precise way and not correlating
general cognition without any clear purpose. Some important limitations, following the
objectives above, are:
• No stand alone systems are examined. The models are grid tied PV systems,
meaning that no energy storage is taking into account.
• The attention is turned more to the electrical grid, meaning that the behaviour of
the grid and the role of the inverter in this part are the basic considerations. No
study is carried out as far as the suitable choice of an inverter according to the PV
systems arrangement (number of modules per strings, number of strings per
inverter or shading phenomena).
• No real grid is taken into account for the study rather than an external grid
component from the software used, where only the short-circuit power, the
voltage factor and R/X ratio are defined. The grid is assumed strong and no
comparison with weak grids is taking place.
6
MSc Thesis Project Introduction
KTH, June 2011
• The purpose of the study of the two different configurations is not to conclude in
an optimum design.
7
MSc Thesis Project Background
KTH, June 2011
2 Background
The stand-alone systems, which are usually implemented in rural and remote areas
in developing countries where no access to the grid is possible. However, the low
cost production and innovative ideas have led to numerous of applications in
industrialized countries as well (e.g. roof top systems, PV-glazing, solar traffic
lighting, solar parking ticket machines, solar chargers, telecom et al.). Stand-alone
systems are usually supported by storage systems (e.g. batteries) in order to meet
the load in times when the solar irradiation is not enough for the PV to cover the
whole need.
8
MSc Thesis Project Background
KTH, June 2011
9
MSc Thesis Project Background
KTH, June 2011
The maximum power point (MPP) is the point where the PV cell, module or array
supplies the maximum possible power. At this point the voltage and current are
defined as maximum power voltage (Vmpp) and maximum power current (Impp)
respectively. The MPP is given in peak watts (Wp) and is strongly affected by the
irradiance level as well as the operating temperature of the PV.
The short-circuit current (IK) is the maximum current that can flow from a PV when
the voltage across the terminals is zero, meaning that are either connected to each
other or an abnormal low-resistance connection has occurred. The short-circuit
current is strongly affected but the incoming irradiation as it is seen in figure 2.3 and
is approximately 5 to 15 per cent higher than the Immp [8]. Typical values of short-
circuit current of various PV modules and under STC can be found in the
specifications of the product [10].
The open-circuit voltage (VOC) is the voltage between the two terminals of the PV
when no external load in connected to it. The VOC is influenced by the operating
temperature of the PV array which is of course linked to the ambient temperature.
This can also be seen in figure 2.3. Typical values of open-circuit voltage can also be
seen in [10].
The STC is a standard test of uniform conditions related to IEC 60904/DIN EN 60904
standards, which categorize the PV modules according to their I-V characteristics [8]. In
brief, those are: vertical irradiance E of 1000 W/m2, cell temperature T of 25°C with a
tolerance of ± 2°C and a defined air mass AM =1.5. AM defines the shape of solar light
spectrum, in an approximate way, at a specific position of sun in comparison to that at
zenith at sea level, which considers to be 1. AM increases as the zenith - sun angle increases
since the light passes through “more atmosphere” and the attenuation (scattering and
absorption is greater.
Figure 2.3: The effect of solar radiation and temperature on the I-V curve [8]
10
MSc Thesis Project Background
KTH, June 2011
A PV system, which is an interconnection of PV modules in series and in parallel, has its own
I-V curve depending on how many PV modules are connected in series (strings) and how
many are connected in parallel. Below it can be seen how the characteristic is formed by
adding PV modules to the system. Furthermore, figure 2.4 points out that only PV modules
with the same electrical characteristics are used in the interconnection in order to avoid
power losses in the final system.
Figure 2.4: The effect of the interconnection of PV modules on the I-V curve
Grid connected systems were explained in the previous section 2.1 as one of the two types
of PV systems. In this part of the chapter a more detailed reference will be attempted since
they cover the highest share of the installed photovoltaic capacity, either as on-grid
domestic systems or power plants. Such PV systems consist mainly of the following
components [8], [11], [12]:
11
MSc Thesis Project Background
KTH, June 2011
The PV modules that interconnect together forming the desirable system. The PV
array is basically the generator of the system and specifically the static generator as
it will be presented later on, since there is no rotating part.
The mounting system, which for PV power plants is a stone or concrete pad
foundation with metal or timber frames attached on it. The mounting system should
above all ensure the designed angle of the PV system with the sun’s incident
irradiation. When the system is implemented in an open field, sometimes the
mounting should be within some requirements for environmental reasons.
The DC cabling.
The PV combiner/junction box, which is the place where all the strings are
connected together and end up to the main DC cable. This box contains also
important safety components as string diodes, fuses, isolations and the DC main
switch to protect the system and the maintainer from accidents in case of faults.
These protections together with the equivalent ones from the AC side are also found
in literature as the balance of the system.
The PV inverter, which transforms the DC current to AC and supplies it to the grid
(mostly distribution), fully synchronised in frequency and voltage with it. The
significance of this component is high and the full modern role is described in the
following section of the chapter.
The meter cupboard, which is the system’s data monitor involving supply and feed
meter, displaying the flow energy between the PV and the grid or/and the load.
In large PV power plants, there can be additional components that improve the efficiency of
the system or ensure better control and monitoring. For instance, cooling pipes in the back
of the PV array to reduce the operating temperature and increase the MPP or remote
monitors that allow real-time performance, output values and potential faults to be
displayed in a the owner’s computer are some examples of such components.
2.3 PV inverter
The PV inverters that are studied in this thesis are the grid-tied ones. They are also found as
utility-intertie or synchronous inverters. The labelling exists, because there is often a
misunderstanding that there are no inverters for stand-alone systems or that both types of
inverters are the same. In fact there are and they can connect to the grid, but only to import
power from the grid and not supply [11].
12
MSc Thesis Project Background
KTH, June 2011
Three-phase inverters refer to larger systems, which is mostly the case for on-grid PV
systems and are connected of course to a three-phase supply system. The basic
three-phase inverter consists of three single line inverters, which are connected to
each load terminal. So, it is not actually a true three-phase inverter and this is
because a three-wire topology will require relatively high DC voltage values (around
600 V for a 400 V three-phase grid) and is limited to 1000 V due to safety reasons in
installation procedures. Also the monitoring and control for islanding requirement
becomes more difficult in relation to three single phase connections [14].
Figure 2.5: Principle of connecting PV systems to the grid with a single-phase and three-phase inverter [8]
According to the size and the application, inverters can be central, string, multistring and
module kind [11], [14].
Central inverters are connected with more than one or all the parallel strings of PV
modules and can be of some kW until one MW of power range.
String inverters are connected to each string of the PV array seperately and their
range of power is a few KW (0.4 - 2 kW).
Multi-string inverters are a rather new concept according to which, several strings
of different configation (different PV modules) and orientation can be connected
13
MSc Thesis Project Background
KTH, June 2011
together. For this reason, necessary DC-DC converters are used to provide the same
output signal to the input of the multi-string inverter. Multi-string inverters increase
the efficiency of the system since every string can track its own MPP. Their range
varies from 1.5 kW to 6 kW.
Taking into consideration that the PV modules produce DC power at a low voltage, the
system’s output requires some adjustment to be fed as AC power at the votage of the grid
as cited before. The inverters used for this adjustment and apply diferent operation-
principle are [8], [13]:
Line-commutated. Such invertrs use switching devices (thyristor bridge or IGBT) that
control the switch-on time only. The switch-off time is done by reducing the circuit
current to zero by using the voltage of the grid. The name line-commutated
represents exactly this grid controlled dependance, meaning the inverter uses the
voltage of the grid to decide the turn on and turn off time of these thyristors. One
disadvantage is that they produce a square wave current output, which introduces
undesirable harmonic components, which can be reduced by the use of filters. This
principle is used today less especially in single phase inverters.
Self-commutated. Such inverters are more complicated and use switching devices
(IGBT and MOSFET) that can control the switch-on and switch-off time and adjust
the output signal to the one of the grid. The self-commutated inverters are the
predominant technology in PV power sources because of their ability to control the
voltage and current output signal (AC side), regulate the power factor and reduce
the harmonic current distortion. Especially, since the role of PV inverter has become
more vital, this operation principle is offering the capabilty to cover the multiple
services and increase the resistance to the grid disturbances. Depending on the type
of pulse they control, either voltage or current, self-commutated are divided to
voltage source and current source inverters.
• Voltage source inverters (VSI). VSI realize the DC side as a constant voltage
source and the output current is changing with the load. For this reason is
normally connected to the grid with an inductance so as not to supply with
current infinitely when there is not voltage or phase match between
inverter and grid.
Self-commutated inverters produce very good sine wave outputs with the use PWM
technic and low pass filters [8].
14
MSc Thesis Project Background
KTH, June 2011
Another basic criterion for categorizing PV inverters is whether or not use galvanic isolation
(transformer) to connect to the grid. There are many advantages and disadvantages in each
type to be considered, with Electromagnetic Interference (EMI) being one of the most
important issue. Inverters with low-frequency transformers (50 Hz) or high frequency
transformers (10 kHz to 50 kHz) have the DC circuit seperated from the AC circuit, offering
recuction of EMI. However, the big size especially when using low frequency transformers,
the lower efficiency of the inverter due to transformer losses and the extra cost turn the
attention to transformless topologies and their improvement to work in higher power
ranges than today [8]. Transformless topologies still need more innovative and complicated
solutions to become competitive especially in terms of electrical safety. Furthermore, in
cases when the the DC output of the PV system is not as the one of the grid or higher, a
step-up DC-DC converter is needed. Thus, part of the losses that were avoided from not
using a transformer are compensated by the use of the converter. Nevertheless, almost all
the typical applied inverter structures today need a boosting and require a DC-DC converter
[14].
In general, there are numerous different topologies of inverters that could apply in grid
connected systems. Today, big manufacturing companies promote their own designs and
variations, which are derivatives of two main converter families [14]:
H-bridge or FB topology. Figure 2.6 shows the original structure of this topology,
which is used in the most typical complete PV structures nowadays. Based on that
many designs have been patented offering a wider range of choices (e.g. H5 Inverter
of SMA, HERIC Inverter of Sunways, REFU Inverter etc).
NPC topology. It is a more modern topology and the one that was connected to the
grid without transformer. In general and in comparison with the FB topology, NPC
can produce lower switch losses and harmonics, improving the efficiency of the
inverter [15]. However, is rather unbalanced and require double voltage input in
comparison with the FB topology [14].
15
MSc Thesis Project Background
KTH, June 2011
Efficiency. This is a basic issue in every system, but mostly in PV systems where the
highest energy yield is the priority. The current efficiency of the inverters is very high
in every topologies, reaching the 92% and 94% in inverters with transformers and
even higher without galvanic isolation. As rule of thumb an improvement of 1% can
result in 10% more power output over a year [16]. Standby power losses during
periods of negligible load need to be assessed, because they affect the overall
efficiency. Since inverters operate at different efficiencies depending on the load,
every inverter is expressed with different efficiency curve. A reliable method to
evaluate the overall efficiency of the inverter is the European Efficiency standard or
else eEURO. This standard takes into consideration the amount of time (in percentage)
that the inverter is expected to work at partial load/level of irradiation. Even if this
standard is valid for irradiance levels of Central Europe, it is a sufficient way to
compare different inverters [11]. The euro-efficiency is defined by (2.1)
Explaining a factor of the above component e.g. 0.03, it means that the inverter is
operating at 5% for a duration of 0.03 and the total operating time.
Power Quality. This issue is actually a general requirement for all grid-connected
inverters and not only PV ones. It refers to variations in voltage magnitude,
frequency limits, harmonic content in the waveforms and other parameters. The
THD limits are set by international and local standards (usually less than 8%). The
16
MSc Thesis Project Background
KTH, June 2011
total harmonic distortion is given by (2.2) [17]. In general the harmonic content must
be low to protect both loads and utility equipment. The waveform and power factor
must be acceptable to the utility.
40
∑X
h=2
2
h
THD = (2.2)
X1
X h = harmonics
X 1 = fundamental
Compatibility with the array. Both array and inverter need to be compatible and the
inverter should be able to withstand the maximum array current and voltage. The
VOC of the array should also be well within the inverter’s tolerable voltage range. The
MPP range of the inverter should also match the operating voltage of the array.
These compatibility issues can be seen in figure 2.7. As far as the MPP tracker
requirements concerned, those should be of high efficiency during steady state, fast
tracking in sudden changes of solar radiation and stable operation at very low
irradiation levels [14].
Lightning and voltage impulse protection. These must comply with local provisions.
17
MSc Thesis Project Background
KTH, June 2011
Other. Issues like size, weight, construction and materials, protection against local
weather conditions, terminals, and instrumentation should be addressed in
conjunction with local rules.
• Voltage control
• Active power control
• Reactive power compensation
• Harmonic compensation
• Fault ride-through
Those new services of the PV inverters will be presented more deeply later on in the “new
German Grid Code” section.
In general, the definition of LVRT or FRT includes the requirements that a power generating
unit tied to the grid should meet, in case of a voltage dip due to a fault or sudden load
change in the grid. The impact of the voltage dip can be described according to the voltage
level reduction and its duration.
In this case, the power generation unit is a PV system, connected with the grid through the
inverter, which is in fact the device that should be capable for LVRT. The possible scenarios
during a voltage dip (or power dip) are:
Immediate disconnection from the grid, when the fault occurs and throughout its
duration. The inverter shall reconnect again after the fault is cleared.
Stay connected but support the grid only with reactive power (reactive current)
during the fault. After the clearance of the fault the unit should consume the same
reactive power as before the fault.
From the above three scenarios, the one that will be applied depends on the decision of the
grid operator under respective grid codes. Until recently PV systems were designed with the
sole purpose to provide the best possible active power to the grid. Therefore, the PV
inverter topologies and control systems were orientated mostly to the MPP tracking and
18
MSc Thesis Project Background
KTH, June 2011
even if they were designed to provide reactive current, during a fault they were
disconnecting from the grid under technical standards as IEEE 1547 and VDE 0126-1-1 [17].
However, due the extensive PV penetration into the grid, as described in the introduction
chapter, the requirements have been modified. The PV inverters should stay connected to
the grid and support it with reactive power, when needed, contributing to the power quality
and prevent voltage instabilities. The disconnection-reconnection scenario is aggravating for
the components of the system reducing possibly their lifetime or causing even greater
instabilities to the grid especially in large scale integration. Moreover, after disconnection
the PV unit will be connected again when the grid is stabilized, meaning that the time that is
off the grid the loss of active power could be of great significance especially on grids where
the share of PV power is high.
In PV systems the importance became obvious after the growth of PV systems connected to
the LV and MV as well. The existing grid was not designed for such high penetration of
interconnected PV and a violation of voltage limits in times of high solar irradiation was
possible. For this reason PV inverters should be able to provide reactive power in order to
reduce the voltage rise along the feeder [18]. As mentioned above, in case of voltage
collapse the inverter should be able to provide reactive current and stabilize the grid within
some time frames defined by grid codes. Nevertheless, reactive power compensation for
the PV inverters hasn’t been part of many local GCs. German utilities, though, have defined
and analysed the supply methods of reactive power, which are stated below the “the new
German Grid Code” section.
The specificity of the PV systems to be close to the location where the reactive power could
be needed is an advantage considering the fact that reactive power does not travel far in
comparison to the active power [19].
The need of improved and new technical specifications, as stated in previous parts of this
report, is the requirement that will ensure smoother penetration of PV systems, without
compromising the power quality and stability of the grid. The GCs represent these
requirements and address to network operators, project designers as well as component
manufacturers (mostly PV inverter manufacturers) in order to design their products
19
MSc Thesis Project Background
KTH, June 2011
according to some uniform guidance [20]. Photovoltaic power systems affect mostly the low
and medium-voltage network and only approximately 1% of the high voltage network is fed
by PV power [5], meaning that the demand for grid stability reflects the low and the
medium voltage networks.
The growth of the renewable generation and the expansion of distributed generators, have
aroused awareness to many countries. However the introduction of effective grid codes is a
rather difficult task with problems that cause significant delays in the process [14]. Some of
those problems are the different features among the different generators, the void
legislation pattern and the lack of production management in the field. The PV industry is
even more sensitive to such problems, because of the wide range of different PV inverter
technologies and designs and their multitask role in comparison to the conventional one
they had until now [14].
Due to the different grid characteristics there are many different GCs that have been
introduced around the world. Countries as China, Australia and India have different
requirements among them. Even inside Europe there are many differences. Concerning the
PV field the requirements usually follow the requirements of wind power systems or the
general provisions that apply to the generators that produce electricity close to the end
users of power (distributed generation).
20
MSc Thesis Project Background
KTH, June 2011
Table 2.1: New requirements for grid tied generators [22]
Under continuous operation and when the system operator requires it, the unit must be
capable to participate in the static voltage support in order to keep the voltage within
acceptable limits when slight deviations occur. The participation is has to do with reactive
power injection capabilities which are described below. These voltage limits are different in
every level of voltage but are usually between +12% and -13% of the nominal voltage [23].
The IEEE-1547 standard requires for the utility interactive inverters +10% and -12% at the
PCC [24].
Active power control or active power throttling [21] or active power derating [22] or active
power curtailment [14], [23] as it is found in the literature refers to the ability of the
generating plant to reduce/adjust its power output as required by the network operator or
even disconnect the PV plant in order to avoid potential dangers regarding the stability of
the system and human personnel. Some cases of controlling the active power could be:
unsafe system operation, unintentional islanding, frequency deviation or maintenance after
a grid failure.
In Germany the active power is required to be changed with a ramp rate of 10% per minute,
or smaller, of the rated active power capacity until any level is necessary. However, it
cannot be lower than 0.1 p.u [14]. The power plant should not disconnect from the grid for
any setpoint over 10%. The control is implemented normally in two ways [20][21]:
21
MSc Thesis Project Background
KTH, June 2011
Manually, with the use of an adequate signal by the operator, which represents a
setpoint (e.g. 100%, 60%, 30% and 0%). There is no physical interference in the
control unit, only the use of the control signal.
As far as the first way concerned, the control unit should follow the below figure 2.8.
According to this figure the PV system should reduce the power output when the frequency
exceeds the value 50.2 Hz. The slope or gradient of reduction should be 40% of the
instantaneous last value of power just before the 50.2 Hz. Besides the upper frequency
limitation, the value 50.05 is the lower limit below which the PV system can increase again
the active power feed-in. The grey areas in the figure set where the plant should disconnect
from the grid, what is to say below 47.5 Hz and above 51.5 Hz.
Figure 2.8: Active power control requirement for grid-tied generators [20][21]
The reactive power was discussed in a different part of this thesis, in an attempt to
emphasize its importance to the safe operation of the grid. The German GCs state that the
generating units should be able to provide reactive power support in every operating point
by adjusting the power factor at the PCC, at least in a value of 0.95 both leading and lagging
for all the power levels. The investigated reactive power supply methods are [18][20][25]:
22
MSc Thesis Project Background
KTH, June 2011
A Q(U) droop function, where the provided reactive power depends on the voltage
at the PCC.
Nowadays, PV systems are mainly designed to provide active power, since reactive power
contributes to losses in the lines, transformers and inverter. For this reason and in order to
comply with the above requirement the inverter should be oversized. What is to say, taking
into account the above power factor of 0.95 an inverter able to supply 475 KW should be of
500 kVA apparent power.
When referring to dynamic voltage support, it is simply implied the requirements what a PV
system should fulfil under fault conditions and grid disturbances and also which should be
its behaviour after the restoration of the fault. These requirements are [14], [20]-[22]:
23
MSc Thesis Project Background
KTH, June 2011
Reactive current injection requirement. During the fault and as described above
the PV should support the grid (voltage support) by feeding-in reactive power or
absorbing. In the German GCs the required reactive current is defined as presented
in figure 2.11.
Figure 2.11: Reactive current injection requirements in the event of network faults [20]
24
MSc Thesis Project Background
KTH, June 2011
(asymmetrical fault), the reactive current should not cause voltage increment above 10% of
the nominal voltage in the non faulty phases.
IEC 61727 – Characteristics of Utility Interface. This standard is more specific for
PV systems and refers to on-grid systems operating in parallel with the utility and
utilize static non-islanding inverters and also to PV systems interconnected to the
distribution system. A more specific standard, IEC 62116, has been implemented
also, defining the testing procedures of AI measures that cited in the IEC 61727.
25
MSc Thesis Project Background
KTH, June 2011
The above parameters should be fulfilled during the 95% of the testing period,
while for 5% of the period other wider ranges apply. The specific variations in
voltage and frequency are describes in the local GCs.
26
MSc Thesis Project Methodology
KTH, June 2011
3 Methodology
The method applied in this thesis and meeting the objectives was based on creating a
sufficient background by continuous literature review of a number of scientific papers,
articles, reports and books. The necessary information were filtered and used to provide a
theoretical overview over the grid connected photovoltaic systems and support with
discussion the simulation results. Furthermore, during this bibliographical research the first
acquaintance with the modelling tool was made, by studying tutorials and useful parts of
the technical manual.
Since the theoretical needs have been satisfied, the practical part was initiated that covers
an examination of a PV model of DIgSILENT at first, a modification of it and a development
of two interconnected systems.
The study that was carried out in this project was a simulation study of a PV on-grid system.
The model as well as the simulation was performed using the PowerFactory tool of
DIgSILENT. The respective company has applied years of experience in modelling power
systems and the simulation tool is considered to be one of the most powerful in the field.
DIgSILENT provides the ability to the user to simulate load flow, RMS fluctuations and
transient events in the same environment. PowerFactory has a quite comprehensive library
of models for electrical power system components such as generators, motors, relays etc, as
well as many passive network elements such as lines, terminals, transformers etc. Those
built-in models can correspond to predefined types that are part of the library or user-
defined data types. What is more, it is possible to create new models with DSL and by using
mathematical formulas that describe the behaviour of the model.
The version used is the latest one (version 14.1), which very more adequate for distributed
generation modelling. Load flow studies and RMS dynamic simulations were the functions
that used in this thesis.
Until now three methods have been used to model a PV system itself with PowerFacory,
regardless the type of control. In each method different elements/components are used
27
MSc Thesis Project Methodology
KTH, June 2011
depending on the user’s experience and the type of study that is performed. Below those
methods are deployed in short.
DC Current Source
DC Voltage Source
In fact DC voltage source is used to model PV systems with storage requirements and not to
represent the PV array itself. The battery can be considered as a source of real voltage. The
source voltage represents the open circuit voltage between its terminals [26]. An example
from the PowerFactory library is shown in figure 3.2.
28
MSc Thesis Project Methodology
KTH, June 2011
PWM converter
As seen in figure 3.2 and more clearly in figure 3.3, PWM converter can be used to model a
PV inverter. The PV array is again modelled as a DC current source connected to a DC
terminal and a PWM PV inverter is used to create a sine wave. This converter represents a
self-commutated, voltage sourced converter [26].
Static generator
The PV system is modelled as a static generator, since there is no rotating part in the PV
array. The static generator is used for other similar generators, without rotor, such as fuel
cells, storage etc. Since the generator is connected to an AC terminal, it includes the PV
array, the DC terminal and the PV inverter, which are normally presented as DSL user-
defined models in the general frame. A more detail reference is done in the following
29
MSc Thesis Project Methodology
KTH, June 2011
chapters, because in this Thesis the component used for modelling the PV system is the
static generator.
30
MSc Thesis Project Model Description
KTH, June 2011
4 Model Description
The basic PV system that is analysed in this Thesis is developed by a static generator. It is a
generic model that was built by DIgSILENT as part of a past study and is available in the
newest version of the PowerFactory tool. The template consists of the PV generator with
number of control systems and design features, which are integrated in it and also a LV
terminal of nominal voltage 0.4 KV that the generator is connected with. The capacity of the
system is 0.5 MW.
The model is being examined thoroughly and its features are being presented below. Some
additional information about the DSL code and the parameters used are found in the
Appendix. The model is being analyzed in accordance with the German GCs and its
possibility to serve the needs of PV on-grid systems in Germany. Below in figure 4.1 the
system-model is pasted and highlighted inside the red box. The rest of the configuration,
which includes an external grid, a MV bus bar of 33kV nominal voltage and a step up
transformer of 0.5 MVA rated power, were just built in order to serve the needs of the
examination.
31
MSc Thesis Project Model Description
KTH, June 2011
The external grid that is used in the system is a component of the program. The values that
are used for the calculations in the study are the minimum short circuit values as seen in
figure 4.2. In general, the minimum values are used to determine where to set the fault
pickup level. The minimum short circuit current is the smallest current that can run at a
given point and the circuit breaker should be able to sense that fault at that point [27]. The
maximum values of short circuit currents are calculated to determine the breaking capacity
of the circuit breakers. Both minimum and maximum values are defined by the IEC standard
(IEC 60909) [27].
The assumption that the short circuit power is 30 times higher than the capacity of the solar
power is used [28]. This value is considered a good estimation. In order to determine how
much of PV capacity can be installed in a certain grid, load flow studies are necessary to
check the voltage rise at the PCC.
From this value and using (4.1) [29], the initial short circuit current is calculated
automatically by the program. Figure 4.3 is a simple representation of a short circuit.
32
MSc Thesis Project Model Description
KTH, June 2011
''
S kQ
S ''
kQ = 3U I ''
nQ kQ ⇔I ''
kQ = (4.1)
3U nQ
where,
The c-factors, or else voltage factors, are according to IEC 60038 for MV up to 35 kV [27],
[29]. Voltage factors refer to voltage regulation and imply that the pre-fault voltage
(nominal) would be approximately 5% lower than the no-load voltage. The voltage factors
cmax and cmin define the allowance for system voltages. Here the cmin value is used, which is
used for minimum currents. As far as the R/X ratio concerned, based on the conclusions of
[30], at low values (<0.4) reactive power is more effective for voltage regulation in
distribution networks, while for values above 1.8 active power has a larger impact. In this
case a value of 0.3 is assumed.
The PV generator under normal steady-state operation and flow injects 448.84 kW and 0
kVar as seen in figure 4.4, implying PF=1 at the point of connection with LV terminal. The
active power is at the MPP and is defined by the parameters of the PV array and the data
sheets of the PV modules used in the array. In table 9.1 the values of Vmpp and Impp of the PV
modules are given for STC and are 35V and 4.58A respectively. Taking into account that 20
modules are per strings and 140 modules in parallel then the following calculation gives the
input active power result.
(35 V ⋅ 20 modules series ) ⋅ (4.58 A ⋅140 modules parallel ) = 700 ⋅ 641.2 = 448.84 kW (4.2)
33
MSc Thesis Project Model Description
KTH, June 2011
Figure 4.4: PV generator power flow characteristics under normal steady-state operation
The active power operation limits refer to the inverter (AC side), which is able to inject 475
kW, due to the 0.95 PF that the PV generator is designed to operate. Concerning the
reactive power limits of the inverter, those are defined by the capability curve, which is
implemented in the generator, shown in figure 4.5 and explained below.
34
MSc Thesis Project Model Description
KTH, June 2011
In the above graph a, the complete capability curve of the PV generator (or PV inverter) is
seen, which includes manufacturing constraints and limitations defined by the operator and
the GCs. The x-axis is the active power in p.u. values, while the y-axis suggests the reactive
power in p.u. values.
To begin with, the blue line is the power limit of the inverter. The inverter cannot operate
outside this curve since it is limited by the nominal power of the generator (0.5 MVA in this
case). Thus, it is considered to be a manufacturing constrain. I should be stated that the
injected power is limited by the nominal current of the inverter, meaning that is impossible
to operate at maximum active and reactive power at the same time. As far as the black line
concerned, it is the limit of the active power injection by the engine itself due to the
designed power factor (0.95 in this case), as explained before.
Finally, the red lines are the maximum values (limits) of possible reactive power injection.
Those limits are defined by the manufacturer and here are given for three voltage levels as
it is seen in graph b of the same figure. The nominal AC voltage (green lines in graph b) and
the two voltage limits, maximum (blue lines in graph b) and minimum (red lines in graph b)
for normal operation. The lines of graph b are drawn according to the values of the matrices
in figure 4.6.
The inverter can supply reactive power within the limits (-Qmax and +Qmax) according to the
system operator, the control system and method that is used. In the green marked area in
35
MSc Thesis Project Model Description
KTH, June 2011
graph a, the requirements according to the medium voltage GC is shown. The inverter
should supply reactive power within the defined PF limits and of course the maximum limits
(the capability) should cover this area.
Concerning the use of the reactive capability curve, it is used by setting the local voltage
control to a setpoint (0.95 p.u., 1 p.u. or 1.05 p.u.) and running load flow calculations
considering the reactive power capability settings. The static generator tries then to reach
the voltage setpoint by feeding or consuming reactive power until either the voltage
reaches the setpoint or the reactive power reaches its limit. The reactive power capability
curve is expressed in p.u. values, so it is scaled according to the rated power of the static
generator.
The area that is highlighted in red circle in figure 4.5 and figure 4.6 is a situation where the
reactive power is limited in a certain active power value and voltage due to the nominal
apparent power (0.5 MVA) and the PF (0.95) of the generator.
As mentioned before, the inverters are not designed to provide reactive power. For this
reason they are normally over-sized to provide some amount of reactive power even in
maximum active power production in order to compensate for the absorbance by the
transformers. Under no solar conditions, they can provide voltage support at no extra cost
[31].
36
MSc Thesis Project Model Description
KTH, June 2011
In the second chapter of this Thesis the influence of the solar irradiation to the PV array
current and consequently to the power output was presented. The existence of this slot is to
to comprehend with all the potential change of irradiance (dE) per second and integrate
them over a period of time (see figure 4.8).
37
MSc Thesis Project Model Description
KTH, June 2011
This is of great importance when examining ramp rates and dynamic behaviours of the PV
system during cloud effects. Below in figure 4.9 a simple case of solar radiation increment is
presented, while in figure 4.10 its effect on the PV output characteristics are seen.
38
MSc Thesis Project Model Description
KTH, June 2011
From the above, it is clear the high effect solar irradiance has on the PV array current (~ 70 A
more). On the other hand the voltage output and the Vmmp of the array are not influenced
much from this rise (less than 10 V). Therefore, as expected the PV power output increases
when the solar irradiation increases. This is shown in figure 4.11. This result is not entirely
correct, because with the rise of the incoming radiation, the operating temperature of the
PV array is rising also, especially when there are no cooling mechanisms (fans, heat
exchangers etc). In fact the PV power output will start decreasing slightly after a point, since
the array voltage will decrease.
In this case and under normal operation, no shading effects are taken into account, so the
constant value of this slot is irradiance of 1000 W/m2, which corresponds to the STC
39
MSc Thesis Project Model Description
KTH, June 2011
irradiance. The upper limit of the integrator is set to 1400 W/m2, which is related to the
mean solar power above Earth atmosphere (1366 W/m2) [32].
Temperature (Slot 2)
Similar reasoning follows the temperature slot. Temperature is the second important factor
that influences the power output of the PV system by affecting the voltage of the array. An
integrator is used also here for the potential temperature changes in the cell/module per
second. Such changes can be seen in figure 4.12. The effect in the voltage output is
presented in figure 4.13. However the temperature that is used here is 25 oC which is the
STC temperature.
40
MSc Thesis Project Model Description
KTH, June 2011
This slot consists of five other blocks which are presented in figure 4.14. The main
component of this model is the one that is enclosed in blue contour, which is the model
description of one PV module of the whole array. Then according to the way that is
interconnected with identical modules the final outputs of the PV array are calculated,
which are the array current and the array voltage at MPP, as described in (4.2). The inputs
are the operating temperature “theta” and the irradiance “E” that are defined in slots 1 and
2 as seen above, as well as the voltage at the DC bus bar, meaning the operating voltage of
the PV array, which is denoted as Uarray here. In brief, the voltage is passing through a low
pass filter to attenuate the high frequency signals in case of abnormal operation, so under
normal conditions is deactivated. The “filtered” voltage then is devided by the number of
modules that are connected in series in order to achieve the voltage per module.
Inside the PV module an algorithm is used and presented in Appendix (figure 9.1) by
captions taken from DIgSILENT. This algorithm is calculating the voltage and current at MPP
taking into consideration the temperature and solar irradiation dependency. Since all the
parameters are given at STC, corrections should be made in all voltages (VOC and Vmpp) and
currents (Ik and Impp) according to the operating temperatures of the module. The factors au
and ai from table 9.1 are provided by the module manufacturer and are used in the
following equations:
41
MSc Thesis Project Model Description
KTH, June 2011
These correction factors are used to determine with the following equations in figure 9.1 all
the voltages and currents when the PV array operates in real temperatures, in order to
determine undesired voltage and current levels. For instance, early in the morning or in cold
places the Voc is much bigger than in times with high irradiance and high temperatures. Such
voltage ranges can be harmful for the inverter. The calculation procedure is based on the
modifications of ASTM E1036-96 [33].
The equivalent circuit for the ideal solar cell that is used as a model for calculating the
current output is shown in figure 4.15. The equations, found in [34], that follow the below
figure are used in the DSL code to determine the current output. Such models of course
have a low approximation quality, since important parameters as Rs and Rp that affect the
efficiency of the cell are omitted. Rs refers to losses due to poor conductivity and poor
external connection, while Rp refers to losses due to poorly rectifying devices and has to do
with leakage of current through the cell and around the edges [35].
I = I ph − I 0 (e UT
− 1) (4.5)
I ph − I + I 0
U = U T ln( ) (4.6)
I0
−U oc
I0 = Ik ⋅ e UT
(4.7)
I ph = I k (4.8)
42
MSc Thesis Project Model Description
KTH, June 2011
where,
The slot consists of four blocks, see figure 4.16, and represents the DC bus to which the PV
array and the DC side of the inverter are connected. There are two inputs and one output
from the model. The one input is the Iarray that comes from the PV model. The other one is
the active power pist measured in slot 4. The output is the input DC voltage of the PV
inverter. This values is denoted as udc or Uarray and is the value that enters the PV module
model (slot 3).
The function of this slot is rather simple. The active power measured in the connection point
of the static generator is devided with the udc and the current that runs the DC bus is
calculated in A. The magnitude of units is not the same, so attention should be paid for
transforming the MW to A. The DC current is now subtracted from the PV array current in
order to find the differential current that runs in the capacitor, which is connected in
parallel with the DC bus. This current is transformed in p.u. using the nominal current as
base. The nominal current is not known, however is calculated knowing the nominal DC
voltage and the nominal power of the PV. The p.u. current enters an integrator in order to
calculate the voltage across the capacitor, which is the voltage of the DC bus and input of
the inverter. The calculation in the integrator is similar to the one shown in (4.9) and is valid
for any capacitor. At last the voltage is transformed to V units.
43
MSc Thesis Project Model Description
KTH, June 2011
du (t )
i (t ) = C (4.9)
dt
AC Voltage (Slot 7)
This slot represents one of the main requirements that the PV inverters should meet
according to the German GCs. It is described by the equations in figure 9.2, which are
implemented in the “over frequency power reduction” block that is seen in figure 4.17. The
equations follow the needs described in the analysis of the new German GC and figure 2.8
and its function is examined thoroughly later on in the 4.3.2 paragraph. It has one input,
which is the frequency measurement from slot 5 and one output pred, which is an input to
the main Controller. The frequency, which passes through a filter as well, is the variable that
will trigger the function, when excess of active power in the grid will cause over frequency
situation.
44
MSc Thesis Project Model Description
KTH, June 2011
Controller (Slot 9)
The Controller is the main part of the PV control frame. In figure 4.18 is seen that consists of
several different blocks. The Controller has four inputs and two outputs. The outputs are the
two components idref and iqref, which will be inputs in the static generator slot where, in
fact, will be used by the PV inverter to modulate and control the active and reactive power
respectively.
Starting with the active power control part, the vdcref is the value that was calculated by the
PV array model and is denoted by Umpp-array. This value is the desired voltage at MPP for the
input of the inverter (DC side). This value passes first through a low-pass filter to atenuate
high-frequency components and then ends to a lower limit block. Inside that block (Max in
the figure), it is compared with the minimum operating value of the inverter U_min, which is
set to be 333 V. According to figure 2.7 the inverter has a low limit voltage on the DC side
below which is turned-off (turned-off voltage level). By this comparison inside that block the
voltage that is chosen vdcref0 is above the U_min. This value is then compared (subtracted)
with the actual voltage of the DC side of the inverter, udc, which is here seen as vdcin and
also with the value dvdcref, which is the difference vdcref0 - vdcin. The result of this
comparison is denoted with dp and is also a voltage that passes through a low-pass filter.
Finally, the dpd value enters a PI controller with proportional gain Kp and integration time Tip
and the id component that regulates the active power is calculated. The PI controller is
limited by two parameters, id_min and id_max, and the variable pred from the slot 8. The
45
MSc Thesis Project Model Description
KTH, June 2011
parameters represent the minimum and maximum active current limits respectively, while
pred represents the reduction due to overfrequency. Under normal operation pred is equal
with 1.
On the other hand and as far as the reactive power control part concerned, the uac value
that is measured by the voltage measurement device as described above, passes through a
low-pass filter and is compared with a reference value, uac0 which is the voltage in the
steady state condition (no fault). Both values refer to the voltage on the AC side of the
inverter. The comparison (difference) gives duac, which shows the voltage deviation or
voltage dip in the connection point of the PV generator. This voltage change is the input of
the “Reactive Power Support” block that follows the principle of figure 2.11. The equations
that seen in figure 9.4 define a deadband of 10% of the nominal voltage and determine the
iq component as in (4.10)
iq = K duac (4.10)
The factor K is what is denoted as droop in the parameters table 9.4. The definition of iq is
written according to the Transmission Code 2007 and the System Service Ordinance
SDLWindV. The “Reactive Power Support” block is limited by the maximum reactive current
(iq_max) and minimum reactive current (iq_min).
At last the calculated values of iq and id together with duac enter the current limiter block in
which the reference values of these components are calculated. The limiter sets the
maximum allowed value of ablosute current and the maximum absolute value or reactive
current in normal operation, as limits.
46
MSc Thesis Project Model Description
KTH, June 2011
This slot contains a PLL device from the library of PowerFactory. PLL is a closed loop
structure, which contains an internal oscillator that is synchronized by being phased-locked
to some particular grid power signal. Normally, as well as in this case, this element is able to
measure the frequency and phase of a voltage in the system. The measurement point is the
LV bus bar as shown before is slot 5. The basic structure of a PLL is seen in figure 4.19 and
explained shortly below.
47
MSc Thesis Project Model Description
KTH, June 2011
Loop filter is a low-pass filter to cut the high-frequency AC components from the
phase detector.
With this slot the static generator component that was described in the section 4.2 (The PV
generator) is matched.
In this part of the Thesis number of simulations were run in order to investigate the
compliance of the model with the German GCs. The model was tested for its “static voltage
support” and “reactive power injection” behaviour under normal opearation, it’s active
power control capability, FRT compliance and dynamic support. Each study is run seperately
and presented below.
The methods of providing reactive power are four. Here, is examined if the PV generator is
able to fulfill this requirement and with which of the four ways is able to provide the
reactive power. The power factor limits (0.95leading to 0.95lagging), set by the GCs, are taken
into consideration.
48
MSc Thesis Project Model Description
KTH, June 2011
After a closer look to the control frame of the PV model that was presented above, it is easy
to see that there is no Q control function in the system. Only the active power is controlled
according to the German codes and the reactive power supply during the fault (dynamic
support). This can be proved by running a simple simulation, which is presented below.
The simulated event is a parameter event, where the solar irradiation was set to drop
(maybe due to a cloud effect) to 50%. This drop that is seen in figure 4.20 of course implies
drop to the active power output of the PV array. The drop starts at the 5 sec and the cloud
effect lasts for 5 seconds as well.
To perform the investigation, a contant power factor of 0.95 inductive (lagging) was set in
the PV generator as seen in figure 4.21 in the load flow tab. The choice of the method fixed
power factor is random and serves the purpose of the study. The generator in a simple load
flow with full capacity (448.84 kW) is injecting 147.53 kVar of reactive power. The expected
result would be that with the active power drop, the reactive power would drop as well in
order to maintain the fixed power factor constant during the voltage variation. If this way of
providing reactive power is not part of the control system, then at least the power factor
should be in compliance with the GC limits to justify the presence of another type of control
(fixed Q, cosφ (P) function or Q (U) droop function). However, the simulation showed
different results, which are presented in figure 4.22.
49
MSc Thesis Project Model Description
KTH, June 2011
The active power decrement and the reactive power change at the PCC were plotted in
figure 4.22 for the duration of the event, to observe if the power factor was kept constant.
Furthermore, the voltage deviation due to active power change is seen in figure 4.23.
Figure 4.22: Active and reactive power change during a cloud effect
50
MSc Thesis Project Model Description
KTH, June 2011
Judging by the graphs and the two different points A and B, it is concluded that the power
factor doesn’t remain constant. Moreover, performing the necessary calculations with the
use of (4.11) and (4.12), it is found that no Q control is implemented in the system.
active power ( P )
power factor (cos ϕ ) = (4.12)
apparent power ( S )
Calculations:
4.999 s point A: S = 0.8982 + 0.2432 = 0.93 p.u.
0.898
cos ϕ = = 0.966
0.93
The power factor in point B is found 0.818, which is outside the specified limits, meaning
that the model is not capable for static voltage support according to the German GCs. The
voltage change is almost insignificant as seen in figure 4.23, however the reactive power
supply is not according to the GCs. The voltage starts slightly above 1 p.u. due to the fixed
power factor.
Furthermore, plotting the active and reactive power behaviour in the LV terminal, where the
generator is connected, the same behaviour is observed. As seen in figure 4.24 at the right
plot, which is a zoom version of the reactive power (green line) of the left plot, the injection
is very small of rate of 10-4, meaning that the results in the MV bus bar are influenced from
51
MSc Thesis Project Model Description
KTH, June 2011
the transformer, which absorbs reactive power regardless of the loading [36]. This
stengthens the argument that there is no Q control for the normal operation of the PV
generator.
The same results were observed when the solar radiation was set to increase. The active
power increased, while the reactive power was decreased. The voltage decreased slightly as
well.
In order to create a frequency change with PowerFactory the possibilities are more than
one. One way is to use an AC voltage source component with an adequate frame instead of
the external grid and change the frequency, which is an input value in this component.
Another possibility would be to disable the PLL, initialize the frequency using a DSL
command, inc(), and then change it through a parameter event. However in this case a third
solution was chosen which doesn’t require any changes in the model. Based on the
consideration that the grid is like an AC synchronous generator, by changing the speed of
the rotor the frequency is changed. As seen in figure 4.25 a parameter event was set and
the speed was changed from 1 p.u. to 1.04 p.u. The change occurs in the 7th second and the
simulation runs for 10 seconds.
52
MSc Thesis Project Model Description
KTH, June 2011
As seen in figure 4.26 the 0.4 p.u. change in the speed of the rotor created an overfrequency
of 2 Hz, which is enough in order to evaluate the active power reduction, since the upper
limit as shown before is 50.2 Hz.
Indeed, when the frequency changed the power reduction came into operation and the
value pred changed from 1 p.u. (no reduction) to 0.281 p.u. (71.9% reduction) as seen in
figure 4.27. According to the theory, the reduction step is 40% of the active power per Hz.
In this study the excess in frequency is 52-50.2=1.8 Hz, meaning that the reduction should
be of 1.8 ⋅ 40%=72%.
53
MSc Thesis Project Model Description
KTH, June 2011
Figure 4.27: The active power reduction of the generator due to overfrequency
The reduction of around 72% in active power is seen when perform power measurements in
LV and MV buses in figure 4.28 and figure 4.29 respectively. The reactive power is not
affected, appart from some slight increment in the MV bus due to the transformer
absorbance.
Figure 4.28: The active and reactive power values in the LV bus during the overfrequency event
54
MSc Thesis Project Model Description
KTH, June 2011
Figure 4.29: The active and reactive power values in the MV bus during the overfrequency event
Table 4.1: Voltage dip tests for generating units type-2 [37]
55
MSc Thesis Project Model Description
KTH, June 2011
Table 4.2: Tests performed with the PV model
For achieving voltage drops of certain percentage a short-circuit event is simulated with the
fault impedance to be adjusted properly. To avoid multiple iterations and testing of
different fault resistances, in order to acquire the desirable voltage dip, an approximate
method was used. This method follows figure 4.31, which is taken from [29] and adjusted.
Figure 4.31: Equivalent plan of a grid with fault (a) and the electrical circuit representation (b)
The above figure 4.31 is described by (4.13) and the values seen in figure 4.2.
56
MSc Thesis Project Model Description
KTH, June 2011
cU nQ cU nQ
''
I kQ = ⇔ ZQ = '' (4.13)
3Z Q 3I kQ
Test 1
Test 1 refers to a voltage dip of 100%. That means that the voltage at the PCC in the MV bus
is 0 p.u. The duration of the fault is 150 ms following figure 4.30. The results are seen below
in figure 4.32. The main concern and requirement for the PV inverters is to observe the
ability of reactive power injection and see the voltage level in the LV bus, where the PV
generator is connected. Each graph from the figure is explained below.
57
MSc Thesis Project Model Description
KTH, June 2011
Graph a: The active power at the PCC drops to 0 during the fault, while before and after the
fault the value is 0.898 p.u., which corresponds to 0.449 MW (the base is 0.5 MW).
Moreover, the reactive power in the MV in non-fault state is found -0.049 p.u. (24.5kVar),
meaning that the transformer is absorbing reactive power from the grid. During the fault
the reactive power is 0, since it is compensated by the PV generator, which provides a
constant reactive current. This function is seen clearer in graph c.
Graph b: The voltage in the MV bus drops to 0 due to the pure short-circuit fault. However,
in the LV bus (the AC side of the inverter) the voltage is 0.057 p.u., which is around 22.8 V.
Before and after the fault both buses obtain values close to nominal (0.997 p.u.).
Considering these values and based on figure 2.11 and the “Reactive Power Support” block
in figure 4.18 the following simple calculations can be performed:
duac = uac before fault − uac during the fault = 0.997 p.u. − 0.057 p.u. = 0.94 p.u. (4.14)
and since the droop parameter is 1 (table 9.4) the reactive current that is injected from the
PV generator is iq=0.94 p.u, which leads to
qduring the fault = uac during the fault ⋅ iq during the fault = 0.057 ⋅ 0.94 = 0.053 (4.15)
Graph c: The PV generator before and after the fault supplies the LV bus with 0.898 p.u.
active power, which corresponds to 0.449 MW. This value is the same with the one in the
MV bus, which implies that the losses are insignificant, since there is only one transformer
58
MSc Thesis Project Model Description
KTH, June 2011
and no lines in this simple grid plan. During the fault the real power is 0. Furthermore, and
only during the fault the, PV injects 0.053 p.u. reactive power as calculated above in (4.15),
which is 26.5 kVar.
Graph d: The voltage change in the DC side of the PV inverter is seen. In a voltage dip of
100%, the PV inverter obtains the Voc of the array, which is 876 V based on table 9.1
(Appendix), where 20 (modules in series) x 43.8 V (open circuit of a module UI0)=876 V.
In conclusion, it can be said that the German grid code requirements are fulfilled during the
100% voltage dip. The PV generator is able to remain connected to the LV bus and provide
reactive current for the whole duration of the fault. The response time is instant after the
fault according to the needs. After the clear of the short-circuit the voltage stabilizes almost
directly in compliance with the GCs.
Similar conclusions are found in the other tests. The results are seen in the graphs and more
clearly in the aggregative table 4.4 below.
Test 2
Test 2 refers to a voltage dip of 80%. That means that the voltage at the PCC in the MV bus
is 6.6kV. The duration of the fault is 550 ms. The results are seen below in figure 4.33.
59
MSc Thesis Project Model Description
KTH, June 2011
Test 3
Test 3 refers to a voltage dip of 50%. That means that the voltage at the PCC in the MV bus
is 16.5kV. The duration of the fault is 1000 ms. The results are seen below in figure 4.34.
Test 4
Test 4 refers to a voltage dip of 20%. That means that the voltage at the PCC in the MV bus
is 26.4kV. The duration of the fault is 1500 ms. The results are seen below in figure 4.35.
60
MSc Thesis Project Model Description
KTH, June 2011
Seeing the results of the above table, interesting conclusions can be drawn. Starting with
the most expected outcomes, when the voltage drop becomes bigger the active power
injection of the PV generator is less, while in a pure three-phase fault (100% voltage dip) the
injected active power is 0. Also the voltage in the DC bus bar increases reaching the VOC at
100% voltage dip. As far as the reactive current injection and the voltage level in the
connection point of the PV generator concerned, which is the actual purpose of this
investigation, it is seen that the reactive current injection is bigger when the voltage dip is
bigger. If those results seen in conjunction with figure 2.11, moving further from the
deadband of 10%, meaning bigger voltage dip, the reactive current should be bigger.
Therefore, the results are in accordance with the GCs. On the other hand, the reactive
power injection is dependent on two inversely proportional factors, the voltage level and
61
MSc Thesis Project Model Description
KTH, June 2011
the reactive current. Thus, the maximum value should be at 50%, which is the case as seen
in the table 4.4.
The reactive current injection and LVRT requirements are fulfilled in each of the 4 tests that
the PV model was examined. The voltage stabilizes almost instantly after the fault clearance
ensuring that the PV is capable of dynamic voltage support.
4.4 Summary
During chapter 4 the PV model was investigated according to the German GCs. The
investigation included static and dynamic behaviour and results were presented for both LV
and MV buses. Summarizing the capabilities of this generic model, it could be said that the
generic model is equipped with basic functions in order to address the new requirements.
Active power reduction and dynamic voltage support are met as seen from the results.
However, the lack of reactive power control during normal operation for static voltage
support is considered to be a significant shortcoming.
As far as the ideal solar cell model that the PV array calculations were based on concerned,
for such studies, where the focus is on the impact on the power system, is acceptable. If the
study requires deep analysis and very detailed results, changes should be done and different
solar cell models should be considered such as with Rs or with Rs and Rp or two diode
models.
At last, an issue that was not mentioned during the analysis is that the model has no real
MPP logic implemented. It is assumed that the model works at MPP and a simplified
approximation is used only for the MPP voltage and current by taking into account the
dependency on the irradiance and the temperature with the use of correction factors. For
better performance, the control system could be improved by adopting a more complicated
technique. One way could be adding a block (controller), which will control the vdref signals
as it is seen in figure 4.36. The vdref is the desired voltage at MPP for the input of the
inverter (DC side).
62
MSc Thesis Project Model Description
KTH, June 2011
63
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
It was concluded before that one of the most important weaknesses of the model is the lack
of ability to respond properly to active power variations caused normally by solar irradiation
variations, an attempt was made at this point to implement such function. The approach
that was followed is consistent with simplified methods that used in other PV models built
by DIgSILENT and especially in generic wind turbine models. In figure 5.1 the control
addition is seen inside the red dashed lines and is a constant Q operation mode.
The principal that is used is similar to the one used for the active power control. The control
is done through a PI controller and the switching between fault and non fault current is in
the "Current Limiter" block. The switching for the two operations is seen in figure 5.2, while
the parameters used for the control are seen in table 5.1.
64
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Table 5.1: Parameters for the constant Q control that added in the “Controller” model
Constant Q control
Parameter Symbol Value
Reactive Power Control Gain [p.u.] Kq 0,6
Reactive Power Control Time Constant [s] Tq 0,5
Measurement delay [s] Tmq 0,001
Maximum Reactive Power Limit (lagging) Qmax 0,296
Minimum Reactive Power Limit (leading) Qmin -0,296
The selected reactive power limit values in the above table were based on the capability
curve of the generator and the manufacture’s settings as presented in figure 4.6 for voltage
0.95 p.u. (worst case). Concerning the values of gain, Kq, and time constant Tq those were
defined after a number of iteration and checking the response of the control to be as close
as possible to the desired value Q that was set. It should be noted that the tuning is not the
optimum one and the trial-and-error method is not always the most accurate one. Tuning
the controller can be rather complicated depending on the application; however for this
study the method is sufficient.
Figure 5.2: The switching function written in DSL inside the current limiter
In order to test the effectiveness of the implemented control a similar method as in the
paragraph 4.3.1 is used. At first through the static generator the method of supplying
constant reactive power is set as seen in figure 5.3. The value of 93.5 kVar (0.187 p.u.)
corresponds to a PF of 0.98, meaning that is within the required limits. Then, a parameter
event is set, changing the instantaneous solar irradiation, which leads to active power
output change. This change is seen in figure 5.4, where it is implied that there is first an
irradiation drop and then an increment above 1000 W/m2, since it is assumed that in steady
state the PV array operates under STC conditions.
65
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
The results at the bus where the generator is connected are presented in figure 5.5 and
should be seen in comparison with the ones presented in figure 4.24, where there was no
control. The response of the PV generator to the active power drop is almost instant. There
is a small spike at the moment of the event and then the controller tries to regulate the
reactive power supply stabilizing it at 0.187 p.u. (the set value) after a period of one second.
The same response is seen during the increment of the solar irradiation.
66
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
The spikes in figure 5.5 at the moment of the event change the reactive power within a
value of power of 10-5, meaning that there is no actual effect that violates the required PF
limits. Figure 5.6 show the voltage variations in the LV bus. The change of 10-3 p.u. during
the changes is insignificant.
67
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
dynamic simulations and consequently change the active power output. The modified static
generator (inverter) was based taking into consideration a real commercial central PV
inverter, at least as far as the maximum input and output values of voltage and current
concerned. The full technical data of this inverter can be found in [38].
In the parameters of the “Photovoltaic model” (slot 3 and table 9.1) the number of
modules in parallel, parameter “nParallelModules”, was set to 290 instead of 140. All
the other values remained unchanged considering the same type of modules in the
modified PV array. The modified array has still VMMP=700 V but the current now is
IMMP=1328.2 A. Judging by [38] the values remain within acceptable and realistic
limits and give maximum power output of 929.740 kW. The change through the
parameters was necessary for the RMS simulations.
The rated power was changed in the static generator’s basic data as well. The rated
power was set 1 MVA and with the PF 0.95 of the engine the active power
operational limits were set to 950 kW in the capability curve. The active power in
steady state load flow under normal conditions was set 929.740 W, which is the
value defined by the above new configuration. The generator is assumed again to
work at MPP. The change in the static generator data was necessary for the load
flow calculations.
In the parameters of the “DC Busbar and Capacitor” model, table 9.2, the capacity
was set to double than the previous one, meaning 0.0344 and the rated power
value, Pnen, that is used to define the voltage input of the inverter was set to 1 MW.
The rated power of the transformer was also changed to 1 MVA in order to satisfy
the needs of the new generator.
The modified generator fulfils the same requirements and behaves the same under both
static and dynamic conditions since no interventions took place in the basic control systems
and no active or passive components were added to the frame of the model. However, the
adjustment “hides” an important assumption. The maximum reactive power limits of the
new inverter remained the same. In reality it is most likely that they would be different.
The modified generator and the transformer are used to create two different models of 20
MVA each. The first model is set by connecting 20 times the generator in series, through
lines. Then the whole plant is connected to the MV bus via a line. In the second set up
another method is applied by connecting the generator 20 times to a common MV bus, like
a star. Since, different number and types of AC lines are used the behaviour of each plant is
examined and the results are compared. Draft designs of both set ups are seen in figure 5.7
and figure 5.8.
The external grid in both cases was modified in terms of short-circuit power value, which
was set 600 MVA. The same assumption of 30 times more than the power plant’s capacity
was made as before.
68
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
3
Figure 5.8: The second set up of the PV power plant of 20 MVA
3 The figure 5.7 and figure 5.8 were made using symbols from figures used in [8]
69
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
The amount of nominal current that is carried by each cable defines the type and the
diameter of the cable. In general terms, more current requires a bigger cable (in
diameter/cross-sectional size) to support it. Different diameter means different impedance,
70
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
capacitance and inductance of the cable. So it is vital to conclude to the best possible choice
of cables for the PV system. In this case the selection of cables was made from the types of
lines that the library of PowerFactory offers and according to the nominal current that they
support. Table 5.2 shows all the lines used in the configuration, the nominal current of those
lines and the nominal current that the lines should support.
Aluminium cables were used (Al), which together with the copper ones (Cu) are the most
common types. However, no study for the best type (material) of cable was conducted in
this thesis. The column “nominal current” from table 5.2 was filled from the data of each
type of cable used, while the “current support without voltage drop” was filled using (5.1),
which defines the nominal current of each PV generator that the line should be able to
support. Thus, based on where the line is used in the configuration, meaning which
generators connects, the nominal current (Iflowed) is multiplied with a factor of 1 or 2 or 3 etc
as explained at the beginning of this paragraph 5.2.2. The citation “without voltage drop” is
used, in order to mark the fact that no voltage drop effect across the line was taken into
account due to impedance change.
1MW
I flowed = (5.1)
3 ⋅ 20kV
Table 5.2: Lines used in the first configuration
71
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Before performing similar studies, as presented in chapter 4 for the generic model, load
flow calculations were conducted. The purpose was to investigate the behaviour of the
power plant when the active power output decreases in conjunction with the voltage
deviation at the PCC in a strong grid. By power plant, the 20 PV generators, the transformers
and the lines as a whole unit are referred. The decrement reflects the active power output
dependence on the solar irradiation. In this case it was assumed that all the PV generators
undergo the same reduction step of active power, which is 20% as seen in the first column
of table 5.3. The results of active and reactive power were summarized, showing how the
power plant is behaving in order to maintain approximately the same level of voltage at the
PCC. The maximum voltage deviation is considered to be +/- 5%. The relevant Q curve in p.u.
values is drawn based on the results of the table and shown in figure 5.10 presenting the
reactive power injection ability in steady state operation.
Judging by the results of the above table, it can be concluded that the more active power
the PV plant unit is injecting, the more reactive power consumes from the grid in order to
maintain approximately the same voltage level around 1 p.u. When the PV power plant is
operating at 20% and 0% injects reactive power to the grid due to the consumption of the
transformers and lines. As seen in figure 5.10 the power plant starts consuming reactive
power when more than 0.35 p.u. active power is injected, which is almost 7 MW. The PF
was around 0.99 in all cases.
At this point it should be pointed out that this model and also the second configuration are
simple interconnected set-ups. They should not be confused with aggregative models, since
for such purpose further assumptions and changes should be applied that are not covered in
this thesis. However, when performing aggregative techniques load flow studies are
72
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
necessary in order to define the capability curve of the aggregative model at the PCC and
under normal operation. Lines and transformers as mentioned before in this report affect
the reactive power limits at the PCC regardless of the capability curve of each generator and
for this reason, the “new” limits should be defined. A valid way to estimate the reactive
power margins is to use the U-Q curve method as described in [28]. Briefly, the method
includes load flow calculations in different active power outputs and in various voltages in
the respective bus (PCC). Then, the reactive power versus the voltage is plotted. For steady
state operation, the reactive power limits at 0.95 p.u. and 1.05 p.u. voltage are the ones of
interest.
Furthermore, according to the German GCs the aggregative model should provide reactive
power within 0.95underexcited to 0.95overexcited limits. Thus, the obtained limits from the U-Q
curves should include the reactive power values at 0.95underexcited to 0.95overexcited area as in
figure 4.5. In a different way power factor corrections should be made by connecting
capacitors (that supply reactive power) or inductors (that consume reactive power).
All the generators used in the configuration follow the same control frame. That means that
the static voltage support addition that was presented in section 5.1 counts for this case as
well. The results should be the same as before since the control refers to the point of
connection of the PV generator, which is the LV bus and no change was made there. Notice,
that in an aggregative model the static voltage control should be examined at the PCC. That
means that first the total influence of the equivalent distribution network (including lines,
transformers etc), as that is seen at the PCC, on the reactive power should be calculated.
73
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Similarly, no change took place in the active power control blocks, meaning that identical
results were found after performing the same simulation as in 4.3.2. Assuming the same
power reduction in each PV generator consequently means approximately the same
reduction at the PCC in p.u. values.
The dynamic voltage support was examined using the same method as described in the
4.3.3 chapter. However, due to the fact the grid is different the fault impedances are
different as well. The values of resistance and reactance that were used in this case to
create the desired voltage dips are presented in table 5.4.
After running the simulation the results were summarized in table 5.5. The relative graphs
taken by PowerFactory are found in the Appendix in the 8.3.1 section.
Table 5.5: Aggregation of the results for dynamic voltage support-first case
The above table presents similar results with the ones found in chapter 4.3.3. When the
voltage drop becomes bigger the active power injection of the PV generator is less. In
voltage dip of 100% the active power injection is 0. The reduction of the active power in the
LV bus is followed by a power reduction at the PCC as it is seen in the 6th column. As far as
the reactive current injection and the voltage level at the connection point of the PV
generator concerned, the results show that the reactive current injection is bigger when the
voltage dip is bigger as found also in the previous LVRT analysis. Furthermore, the reactive
74
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
power injection by each PV generator has again its maximum value at 50% due to the
dependence on the two inversely proportional factors, the voltage level and the reactive
current. This behaviour is repeated at the PCC for the whole power plant as it is seen in the
last column of table 5.5.
The LVRT requirements are fulfilled in each of the 4 tests that the PV configuration was
examined. The voltage stabilizes almost instantly after the fault clearance ensuring that the
PV is capable of dynamic voltage support (short-circuit support).
75
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
76
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
A simple load flow study as in the first case was carried out using the same levels of active
power output. The results are seen in table 5.7 and the relevant graph in figure 5.12.
Judging by the results of the above table, the same conclusions can be drawn for the
behaviour of the second PV interconnection. More active power by the PV plant unit results
to more reactive power consumption from the grid in order to maintain approximately the
same voltage level around 1 p.u. Again at 20% and 0% of operating ability the unit injects
reactive power to the grid due to the consumption of the transformers and lines. In figure
5.12 is seen that the power plant starts consuming reactive power when more than 0.30
p.u. active power is injected, which is almost 6 MW. The PF was around 1 in all cases.
77
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Like in the first case the results were the same as in section 5.1
Like in the first case the results were the same as in the section 4.3.2.
In order to facilitate the comparison between the two cases, the same voltage dips were
simulated in this model as well. The grid is considered identical with the one in the first
configuration, meaning that the values of table 5.4 were used for the fault impedance. The
LVRT results are presented in table 5.8, while the graphs are found in section 8.3.2.
Table 5.8: Aggregation of the results for dynamic voltage support-second case
The above table verifies that the second configuration fulfils the LVRT requirement. At the
PCC the maximum value of reactive power injection is found at 50% voltage dip.
78
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Table 5.9: Load flow results of both cases
The results can also be compared easily by observing the curves in figure 5.13. From the
graph is obvious that in the first case the plant starts injecting reactive power at 7 MW,
while in the second case, when the plant injects 6 MW, proving that the consumption of
reactive power of the lines in the first case is greater.
Similar conclusions can be drawn looking the injected reactive power by both set-ups in case
of voltage dips. Table 5.10 shows, that even if the injected reactive current and power by
each generator is the same in both configurations, the total reactive power behaviour at the
PCC differs, showing larger consumption of reactive power in the first case.
79
MSc Thesis Project Further Analysis & Discussion
KTH, June 2011
Table 5.10: Reactive power supply of both cases at PCC in several voltage dips
Voltage dip [%] Injected reactive power by the power plant [p.u.]
80
MSc Thesis Project Conclusions
KTH, June 2011
6 Conclusions
This chapter summarizes the findings of the presented master thesis project “Comparison of
existing PV models and possible integration under EU grid specifications”. Solar power and
especially grid-tied photovoltaic power is expected to play a significant role as an energy
source among other renewables like wind power and hydropower that enjoy the leading
positions. Europe is turning attention to PV power and Germany as a strong player in the
area will influence things notably. PV applications still live their pre-mature days and there
are many, mostly technical, aspects to be studied and improved. Efficiency of the
components, inverter topologies, control schemes and aggregation techniques are some of
the major research areas.
This thesis report focuses on the static and dynamic behaviour of on-grid PV systems and
their interaction with the power grid. A generic model built by DIgSILENT is selected to be
presently examined. The PV system is modelled by a static generator and the control
scheme is implemented in it. The model of the PV array is based on the ideal model of a PV
cell and takes into consideration the voltage and current correction factors based on the
operating temperature and solar radiation.
The control is structured according to the requirements of the German GC for the MV
distribution network. Active power reduction requirement is effectively adjusted and
operates in case of over-frequency events. The reduction is occurred in less than 50 msec.
The LVRT requirement is tested under four different voltage dips of different duration each
according to the German technical guidelines for type-2 generating units. The results
support the capability of the PV model in question to remain connected when a voltage dip
occurs and provide reactive current when it is needed according to the GC. Thus, the grid
stability is improved at the PCC, there is no loss of active power since active power is
provided again the moment the grid is stabilized and no burden on the lifetime of the
components due disconnection-connection and grid is occurred.
As far as the static voltage support concerned, initially the model had no relevant control.
For this reason, a PI controller is implemented to the main control scheme providing
reactive power with the constant-Q method. The controller shows sufficient behaviour
when changes of the irradiance take place, which cause changes in the active power supply
affecting the voltage at the terminal that the generator is connected. However, the need of
more proper tuning is necessary. The switch between static voltage support and dynamic
voltage support in case of a fault is inside the current limiter block and ensures reactive
power support in any occasion.
The study of the two interconnected models shows that the choice of the configuration and
consequently the choice of lines, affects the reactive power at the PCC. This is of great
importance in aggregative studies, where the equivalent line should have the characteristics
(capacitance, impedance etc) of all the lines of the network that has been substituted.
81
MSc Thesis Project Conclusions
KTH, June 2011
Judging by the objectives set at the beginning of this project, it can be said that all of them
where addressed during this study and analysed to the extent of the time constraints given
for such projects. However, in such multifaceted areas there are numerous of issues that
require deeper analysis and further research. Some important issues concerning this model
and in general the “PV on-grid systems” field are:
The PV array model is based on the ideal model of PV cell making the results of
voltage and current output rather simplified. Cell model with one or two diodes and
taking into consideration the Rs and Rp resistances would provide more accurate and
tangible results.
MPP logic would have been also an issue of concern for future studies. Besides the
simplified techniques for correcting the voltage and current in changeable irradiance
and temperature, more advanced methods could be applied e.g. MPP tracking with
fuzzy logic.
Research studies should be done in terms of power quality. So far, there are no
results for the harmonic distortion or possible flicker effects that this PV model can
cause.
The model has been tested under the German GCs, which are considered being the
most detailed. However, it would be interesting to examine the model under other
GCs especially nowadays that countries like France and Italy have increased their
installed capacity significantly. Regarding the Spanish GCs, some studies have been
conducted by [28].
The model has been tested only under voltage dip-types of grid fault. Additional
studies under other types of faults e.g. unsymmetrical fault scenarios can be
conducted.
Rounding up the conclusions of this thesis, in response to the fact that policies and
incentives have brought PV systems to the fore, the research area has to be expanded.
Better solar radiation forecasting and cloud effect studies can be initiated, aggregative
studies can be undertaken, ant-islanding control schemes can be improved and numerous
other areas can be evolved in order to thrust the PV market, encourage the high PV
penetration, while securing the stability and normal operation of the power system.
82
MSc Thesis Project References
KTH, June 2011
7 References
[1] Boyle G., Renewable Energy – Power for a Sustainable Future, Oxford University
Press, New York, USA, 2004
[2] European Union, “Renewable energy: White Paper laying down Community strategy
and action plan”, Summaries of EU legislation, available at
http://europa.eu/legislation_summaries/other/l27023_en.htm , last modified
13.03.2001, as accessed 07.03.2011
[3] European Commission, “Renewable Energy: Progressing towards the 2020 target”,
Progress report, Communication from the Commission to the European Parliament
and the Council, European Commission, Brussels, 31.1.2011
[4] Ackermann T., Kuwahata R., Troester E., “European Grid Study 2030/2050”,
Commissioned by Greenpeace International study, Energynautics GmbH, Germany,
18.01.2011
[5] Brown M., “Integrating PV in Local Distribution Systems -Germany-”, IEA PVPS Task
14 Meeting, Golden, CO, USA, 01.12.2010, available at
www.nrel.gov/eis/pdfs/iea_task_14_workshop_braun.pdf , as accessed 25.02.2011
[7] Jager-Waldau A., “PV Status Report 2010”, Scientific and Technical report, European
Commission Joint Research Center, Italia, August 2010
[8] Earthscan, Planning & Installing Photovoltaic Systems – A guide for installers,
architects and engineers, Second Edition, Earthscan, UK, 2008
83
MSc Thesis Project References
KTH, June 2011
[11] Falk A., Dürschner C., Remmers K.H. Solarpraxis AG, Photovoltaics for Professionals –
Solar Electric Systems – Marketing, Design and Installation, Beuth Verlag GmbH
Berlin Wien Zürich in association with Earthscan, Berlin, Germany 2007
[13] Eltawil Mohamed A., Zhengming Z., “Grid-connected photovoltaic power systems:
Technical and potential problems – A review”, Elsevier, Vol. 14, Issue 1 pp. 112-129,
January 2010
[14] Teodorescu R., Liserre M., Rodriguez P., Grid Converters for Photovoltaic and Wind
Power Systems, Wiley, United Kingdom, 2011
[15] Ma Lin, Kerekes T., Teodorescu R., Xinmin J., Floricau D., Liserre M., “The High
Efficiency Transformer-less PV Inverter Topologies Derived From NPC Topology”, IEEE
Xplore, 2009, available at http://vbn.aau.dk/files/18747465/Publication, as accessed
10.03.2011
[16] Wenham R. Stuart, Green A. Martin, Watt E. Muriel, Corkish R., Applied
Photovoltaics, Earthscan, United Kingdom, 2007
[17] Benz H. Ch., W.-Toke Franke, Fuchs W. F. “Low Voltage Ride Through Capability of a
5k W Grid-Tied Solar Inverter”, 14th International Power Electronics and Motion
Control Conference, Wuhan, China, 17-20.05.2010
[18] Brown M., Stetz T., Reimann T., Valov B., Arnold G., “Optimal reactive power supply
in distribution networks-Technological and economical assessment for PV-Systems-”,
24th European Photovoltaic Solar Energy Conference and Exhibition, Hamburg,
Germany, 21-25.09.2009
[19] OAK Ridge National Laboratory, “Reactive Power and Importance to Bulk Power
System”, available at
http://www.ornl.gov/sci/decc/RP%20Definitions/Reactive%20Power%20Overview_j
peg.pdf , as accessed 13.03.2011
[20] Troester E., “New German Grid Codes for Connecting PV Systems to the Medium
Voltage Power Grid”, 2nd International Workshop on Concentrating Photovoltaic
Power Plants: Optical Design, Production, Grid connection, Darmstadt, Germany, 09-
10.03.2009
[21] Notholt A., “Germany’s New Code for Generation Plants connected to Medium-
Voltage Networks and its Repercussion on Inverter Control”, International
Conference on Renewable Energies and Power Quality (ICREPQ’09), Valencia, Spain,
15-17.04.2009
84
MSc Thesis Project References
KTH, June 2011
[22] Borup Uffe, “Features of modern grid interconnection devices”, Danfoss, PV ERA NET
The Photovoltaic European Research Area Network, available at http://www.pv-
era.net/doc_upload/documents/179_5Featuresofmoderngridinterconnectiondevices
.pdf , as accessed 27.02.2011
[23] Christiansen W., Johnsen T. D., “Analysis of requirements in selected Grid Codes”,
academic report, Technical University of Denmark (DTU), January 2006, available at
http://www.frontwind.com/Analysis%20of%20the%20requirements%20in%20select
ed%20Grid%20Codes.pdf , as accessed 08.03.2011
[24] Enphase Energy, “Calculating ac line voltage drops”, Application Note, 11.10.2010,
available at
http://www.enphaseenergy.com/downloads/EnphaseAppNote_Vdrop_Calculations.
pdf , as accessed 10.03.2011
[25] Brown M., Stetz T., Buedenbender K., “Integration of Photovoltaic in Distribution
Systems”, Electric Cars and integration of Renewable Energy at the 2020 horizon,
Lyngby - DTU, Denmark, 10-12.03.2010
[26] Hernanz R., Campayo M., Belver Z., Pérez P., Gogeascoechea M., Sáiz S., “Dynamic
Simulation of a Photovoltaic Installation”, International Conference on Renewable
Energies and Power Quality (ICREPQ’09), Valencia, Spain, 15-17.04.2009
[27] Prakash M., “Short Circuit Calculations using SIMARIS design software”, Siemens
publications, available at http://www.automation.siemens.com/mcms/electrical-
design-software/de/service-
support/veroeffentlichungen/Documents/Short_Circuit_Calculations_with_SIMARIS
_design.pdf , as accessed 07.04.2011
[28] Alvaro R., “System aspects of large scale implementation of a photovoltaic power
plant”, Master Thesis, KTH, Stockholm, Sweden, March 2011
[29] Oeding D., Oswald B.R., Elektrische Kraftwerke und Netze4, Springer-Verlag, Berlin
Heidelberg, 2004
[30] Blazic B., Papic I., “Voltage profile support in distribution networks – influence of the
network R/X ratio”, Power Electronics and Motion Control Conference 13th (EPE-
PEMC), available at IEEE Xplore, 30.09.2008
[31] Papandrea F., “Grid Integration Aspects of Large Solar PV Installations: LVRT
Capability and Reactive power/Voltage support Requirements”, Master Thesis,
Politecnico di Torino, Luglio, Italy, 2010
85
MSc Thesis Project References
KTH, June 2011
[32] Gueymard A. Christian, “The sun’s total and spectral irradiance for solar energy
applications and solar radiation models”, Elsevier, Vol. 76, Issue 4 pp. 423-453, April
2004
[33] Marion B., “Validation of a Photovoltaic Module Energy Ratings Procedure at NREL”,
NCPV Program Review Meeting 2000, Denver, Colorado, 16-19.04.2000
[34] Wagner A., Photovoltaik Engineering – Handbuch für Planung, Entwicklung und
Anwendung5, Springer-Verlag, Berlin Heidelberg, 2009
[35] Nelson J., The Physics of Solar cells, Imperial College Press, London, UK, 2003
[36] Radhakrishna C., “Reactive Power Management”, Lecture material, South Asia
Regional Initiative for Energy (SARI/Energy), Sri Lanka, available at http://www.sari-
energy.org/PageFiles/What_We_Do/activities/CEB_Power_Systems_Simulation_Trai
ning,_Colombo,_Sri_Lanka/Course_ppts/lecture_41.pdf , as accessed 07.04.2011
[38] SMA Solar Technology, Sunny Central 1000MV, Technical data, available at
http://www.sma.de/en/products/solar-inverters/sunny-central/sunny-central-
800mv-1000mv-1250mv.html , as accessed 04.05.2011
Throughout this report and especially from chapter 4 until chapter 9, figures, graphs, tables,
parameters and parts of the software environment PowerFactory of DIgSILENT have been
used as captions. There is no specific bibliographic reference, however it is mentioned several
times that the subject of test and the results are products of the simulation tool in question.
86
MSc Thesis Project Appendix
KTH, June 2011
8 Appendix
PV Array
Parameter Symbol Value
Open circuit voltage of module in STC [V] UI0 43,8
MPP voltage of module in STC [V] Umpp0 35
MPP current of module in STC [A] Impp0 4,58
Short-circuit current of module in STC [A] Ik0 5
Temperature correction factor (voltage) [1/K] au -0,0039
Temperature correction factor (current) [1/K] ai 0,0004
Number of modules connected in series [-] nSerialModules 20
Number of modules connected in parallel [-] nParallelModules 140
Time constant of module [s] Tr 0
87
MSc Thesis Project Appendix
KTH, June 2011
Table 8.4: Parameters in main Controller slot
Controller
Parameter Symbol Value
Gain of the active power PI controller [-] Kp 0,005
Integration time constant of the active power PI controller [s] Tip 0,03
Measurement delay [s] Tr 0,001
Time delay MPP-Tracking [s] Tmpp 5
Deadband for AC voltage support [p.u.] deadband 0,1
Static for AC voltage support [-] droop 1
i_EEG = 0 according to TC2007; i_EEG = 1 according SDLWindV
[-] i_EEG 1
Minimum active current limit [p.u.] id_min 0
Minimum allowed DC - voltage [V] U_min 333
Minimum reactive current limit [p.u.] iq_min -1
Maximum active active current [p.u.] id_max 1
Maximum reactive active current [p.u.] iq_max 1
Maximum allowed absolute current [p.u.] maxAbsCur 1
Maximum absolute reactive current in normal operation [p.u.] maxIq 1
88
MSc Thesis Project Appendix
KTH, June 2011
89
MSc Thesis Project Appendix
KTH, June 2011
Figure 8.2: Main part of DSL code in the active power reduction block
Figure 8.4: The DSL code in the reactive power support block
90
MSc Thesis Project Appendix
KTH, June 2011
91
MSc Thesis Project Appendix
KTH, June 2011
92
MSc Thesis Project Appendix
KTH, June 2011
93
MSc Thesis Project Appendix
KTH, June 2011
94
MSc Thesis Project Appendix
KTH, June 2011
95