Kyriakos_Pantziris_-_MSc_Thesis_-_2014_-_Voltage_support_strategies_in_a_rural_LV_network_with_high_PV_penetration.

Kyriakos_Pantziris_-_MSc_Thesis_-_2014_-_Voltage_support_strategies_in_a_rural_LV_network_with_high_PV_penetration.
Voltage support strategies in a
rural low voltage network with
high photovoltaic penetration
Master of Science Thesis
K. Pantziris
Voltage support strategies in a rural
low voltage network with high
photovoltaic penetration
Master of Science Thesis
For the degree of Master of Science in Sustainable Energy Technology
at Delft University of Technology
Kyriakos Pantziris
Delft University of Technology – Electrical Sustainable Energy Department
DNV GL – Energy
May 2014
DELFT UNIVERSITY OF TECHNOLOGY
FACULTY OF
ELECTRICAL ENGINEERING, MATHEMATICS AND COMPUTER SCIENCE (EEMCS)
ELECTRICAL SUSTAINABLE ENERGY (ESE)
The undersigned hereby certify that they have read and recommend to the Faculty of
Applied Sciences (AS) for acceptance a thesis entitled
VOLTAGE SUPPORT STRATEGIES IN A RURAL LV NETWORK
WITH HIGH PV PENETRATION
by
KYRIAKOS PANTZIRIS
in partial fulfilment of the requirements for the degree of
MASTER OF SCIENCE IN SUSTAINABLE ENERGY TECHNOLOGY
May 27, 2014
Thesis committee members
Supervisors:
P. Bauer
P. Vaessen
Readers:
J. A. Ferreira
L. M. Ramirez Elizondo
i
Abstract
The rapidly increasing penetration of rooftop PV systems in rural LV distribution networks calls
for the attention of DNOs in order to secure end-user voltage range. In times of high
photovoltaic generation and low load consumption, voltage at PCCs may exceed the specified
upper limit and make PV inverters trip. This phenomenon hinders further PV integration in the
network although MV/LV transformer and conductors are by far not used up to their full
capacity yet.
In this thesis, voltage rise problem is analysed through load flows and simulations on a suitably
designed rural test network model implemented in PowerFactory software. The suggested local
voltage support strategies by the German directive VDE-AR-N 4105 and the recent European
standard EN 50438, namely PF(P) and Q(V), which require reactive power control capability of
photovoltaic inverters, are implemented, tested and analysed in order to check their
effectiveness and compare their behaviour. As active power curtailment capability is also already
required by some DNOs, a dynamic active power curtailment control algorithm is designed and
tested as well, taking into account the local load demand and the network’s feed-in limitations.
Afterwards, local battery storage is also incorporated in every PV system model and all the three
aforementioned strategies are tested and analysed again.
All studied strategies manage to mitigate the voltage rise problem up to a PV-integration level of
10 kW/household. However, their effectiveness is compared in terms of a set of evaluation
criteria for a range of PV-integration levels. Subsequently, the best candidate strategy, among
the ones studied, emerges through the help of experts’ opinion and a suitably designed overall
evaluation score number, for both the perspectives of a DNO and a PV system owner. It is
revealed that the overall preference of a DNO is for a solution which involves active power
curtailment and local storage, in contrast to the overall preference of a PV system owner for a
reactive power based strategy without storage.
Keywords: distribution network, low voltage, distributed generation, photovoltaic systems,
reverse power flow, voltage rise, inverters, voltage support, reactive power control, active
power curtailment, storage, photovoltaic integration, hosting capacity, PowerFactory
iii
Contents
Abstract ................................................................................................................................. i
Contents............................................................................................................................... iii
List of Figures ....................................................................................................................... vii
List of Tables..........................................................................................................................xi
List of Abbreviations ............................................................................................................ xiii
Acknowledgments ................................................................................................................ xv
1
2
3
Introduction ................................................................................................................... 1
1.1
Electric power systems and distributed generation history .......................................... 1
1.2
Photovoltaic systems ..................................................................................................... 3
1.3
PV grid integration ......................................................................................................... 4
1.4
Problem statement ........................................................................................................ 5
1.5
Project objective and research questions ..................................................................... 7
1.6
Research approach ........................................................................................................ 7
1.7
Watt connects................................................................................................................ 8
1.8
Thesis outline ................................................................................................................. 8
Regulations .................................................................................................................. 11
2.1
General ........................................................................................................................ 11
2.2
Voltage requirements .................................................................................................. 12
2.2.1
Supply voltage variations ................................................................................ 12
2.2.2
Flicker .............................................................................................................. 12
2.2.3
Voltage dips/swells.......................................................................................... 13
2.2.4
Voltage unbalance ........................................................................................... 13
2.3
Power factor - Reactive power capability.................................................................... 14
2.4
Active power curtailment ............................................................................................ 15
Voltage rise and mitigation solutions ............................................................................ 17
3.1
Voltage rise in a LV feeder ........................................................................................... 17
3.2
Solutions to voltage rise problem ................................................................................ 18
3.2.1
Grid reinforcement.......................................................................................... 18
3.2.2
MV/LV transformers with OLTC ...................................................................... 19
3.2.3
Reactive power control strategies .................................................................. 20
3.2.3.1 Fixed PF ............................................................................................ 21
iv
Contents
3.2.3.2 PF in terms of injected active power PF(P) ...................................... 21
3.2.3.3 Voltage-dependent reactive power Q(V) ......................................... 23
3.3
4
Active power curtailment (APC) ...................................................................... 25
3.2.5
Storage............................................................................................................. 27
Categorisation of strategies based on their communication requirements ............... 28
3.3.1
Local strategies ................................................................................................ 28
3.3.2
Decentralised strategies .................................................................................. 28
3.3.3
Central strategies............................................................................................. 29
Simulation setup........................................................................................................... 31
4.1
4.2
4.3
4.4
5
3.2.4
Rural test network model ............................................................................................ 31
4.1.1
External grid..................................................................................................... 33
4.1.2
MV/LV transformer ......................................................................................... 33
4.1.3
Feeders and conductor types .......................................................................... 34
4.1.4
Loads, PV generators and batteries................................................................. 35
Load model .................................................................................................................. 35
4.2.1
“Load profile” block ......................................................................................... 36
4.2.2
“Load data process” block ............................................................................... 37
4.2.3
“Load” block .................................................................................................... 38
PV-Battery system model ............................................................................................ 38
4.3.1
“Irradiance” block ............................................................................................ 39
4.3.2
“PV generation profile” block .......................................................................... 40
4.3.3
“Load Profile” blocks ....................................................................................... 41
4.3.4
“Voltage” block ................................................................................................ 41
4.3.5
“Active Power Control” block .......................................................................... 41
4.3.6
“Reactive Power Control” block ...................................................................... 42
4.3.7
“Battery Control” block ................................................................................... 45
4.3.8
“PV Generator” and “Battery” blocks.............................................................. 47
General simulation assumptions ................................................................................. 47
Implementation and test results ................................................................................... 49
5.1
High PV integration without voltage support .............................................................. 49
5.2
Voltage sensitivity analysis of test network ................................................................ 51
5.3
Reactive power control strategies ............................................................................... 53
5.3.1
PF(P) control mode .......................................................................................... 56
5.3.1.1 PF(P) without storage ....................................................................... 57
Contents
v
5.3.1.2 PF(P) with battery storage ............................................................... 58
5.3.2
Q(V) control mode ........................................................................................... 60
5.3.2.1 Q(V) without storage ........................................................................ 62
5.3.2.2 Q(V) with battery storage ................................................................ 63
5.3.3
5.4
5.5
6
Comparison of reactive power based voltage support strategies .................. 64
Dynamic active power curtailment strategies ............................................................. 68
5.4.1
DAPC without storage ..................................................................................... 68
5.4.2
DAPC with battery storage .............................................................................. 69
5.4.3
Comparison of the DAPC voltage support strategies ...................................... 70
Overall comparison ...................................................................................................... 72
5.5.1
Formulation of an overall evaluation criterion ............................................... 72
5.5.2
Relative weighting of the evaluation criteria .................................................. 74
5.5.3
Choosing the best strategy .............................................................................. 75
Conclusions and future work ........................................................................................ 77
6.1
6.2
Conclusions .................................................................................................................. 77
6.1.1
Main conclusions ............................................................................................. 77
6.1.2
Specific findings ............................................................................................... 77
Recommendations for future work ............................................................................. 79
Appendix A - OHL modelling................................................................................................. 81
Appendix B - DSL models’ code ............................................................................................ 83
Appendix C - DPL commands ................................................................................................ 85
Bibliography ........................................................................................................................ 89
vii
List of Figures
Figure 1.1: Conventional electric power system [2] ........................................................................ 1
Figure 1.2: Connection of distributed generators [2] ...................................................................... 2
Figure 1.3: Evolution of global cumulative installed capacity 2000-2013 [13] ................................ 4
Figure 1.4: Evolution of European PV cumulative installed capacity 2000-2012 [12] ..................... 5
Figure 1.5: PV contribution to the energy demand in the EU 27 in 2012 based on cumulative
installed capacity in 2012 [12] ......................................................................................................... 5
Figure 1.6: Voltage profile on an exemplary feeder with distributed generation [11].................... 6
Figure 1.7: Watt connects interactive table ..................................................................................... 8
Figure 2.1: Evolution of the German grid codes and regulations compared with PV deployment
(GW) [11] ........................................................................................................................................ 11
Figure 2.2: Reactive power capability in load reference frame [20] .............................................. 14
Figure 2.3: Reactive power control characteristic [20] .................................................................. 14
Figure 2.4: PF(P) characteristic according to VDE-AR-N 4105 [22] ................................................ 15
Figure 3.1: Simplified schematic of a PV generator and a load connected to a LV feeder ............ 17
Figure 3.2: Phasor diagram of voltage rise calculation .................................................................. 18
Figure 3.3: Fixed PF characteristic.................................................................................................. 21
Figure 3.4: PF(P) characteristic....................................................................................................... 22
Figure 3.5: Q(V) characteristic........................................................................................................ 23
Figure 3.6: P(V) characteristic ........................................................................................................ 25
Figure 3.7: Dynamic active power curtailment strategy for a feeder with 5 kW/household
maximum net generation............................................................................................................... 26
Figure 3.8: Categorisation of voltage support strategies based on their communication
requirements [15] .......................................................................................................................... 28
Figure 4.1: Single-phase diagram of the test rural network structure .......................................... 32
Figure 4.2: External Grid element connected to the MV busbar of the transformer .................... 33
Figure 4.3: Histograms and associated Weibull distribution of the average house distance for
rural, village and suburban networks [47] ..................................................................................... 34
Figure 4.4: Household load, farm load, PV generator and battery system connected to the last
terminal of the first feeder............................................................................................................. 35
Figure 4.5: Load frame ................................................................................................................... 36
Figure 4.6: Example of daily household load profile ...................................................................... 36
viii
List of Figures
Figure 4.7: Example of farm load profile ........................................................................................ 37
Figure 4.8: 3-phase load model in PowerFactory........................................................................... 38
Figure 4.9: PV-Battery system frame ............................................................................................. 39
Figure 4.10: Weekly irradiance profile used in the simulations ..................................................... 40
Figure 4.11: Fixed PF characteristic ................................................................................................ 43
Figure 4.12: PF(P) characteristic ..................................................................................................... 44
Figure 4.13: Q(V) characteristic ...................................................................................................... 44
Figure 5.1: Voltage profile of feeder 1 under low load and high PV generation conditions (10
kW/household and 27 kW/farm) for both conductor types .......................................................... 50
Figure 5.2: Voltage profile of feeder 2 under low load and high PV generation conditions (10
kW/household and 27 kW/farm) for both conductor types .......................................................... 50
Figure 5.3: Voltage measurements for the weakest point of the network under normal load
conditions and high PV integration (10 kW/household and 27 kW/farm) throughout the weekly
simulation ....................................................................................................................................... 51
Figure 5.4: Voltage sensitivities to P and Q variation for the terminals of feeder 1 ...................... 52
Figure 5.5: Voltage sensitivities to P and Q variation for the terminals of feeder 2 ...................... 53
Figure 5.6: Flowchart of the script executed to calculate the PV hosting capacity of the network
for different power factors of the PV generators .......................................................................... 54
Figure 5.7: Plot of the PV hosting capacity with respect to the power factor of a household PV
system ............................................................................................................................................ 56
Figure 5.8: Selection of the appropriate PF(P) characteristic ........................................................ 57
Figure 5.9: Comparison of the selected PF(P) characteristic with the hosting capacity limits in
case of maximum PV hosting capacity ........................................................................................... 57
Figure 5.10: Voltage at the weakest terminal in case of PF(P) strategy without storage .............. 58
Figure 5.11: Active and reactive power injection of a household PV system for a PV-integration
level of 10 kW/household .............................................................................................................. 58
Figure 5.12: Voltage at the weakest terminal in case of PF(P) strategy with battery storage....... 59
Figure 5.13: State of charge of one of the household battery systems throughout the weekly
simulation ....................................................................................................................................... 59
Figure 5.14: Sensitivity of Q(V) characteristic in the selection of its parameters: a) for the Vsp
parameter, b) for the Vmax parameter ............................................................................................ 60
Figure 5.15: Flowchart of the script executed to calculate the highest value of Vmax which allows
for each PV-integration level the use of Q(V) strategy .................................................................. 61
Figure 5.16: Voltage at the weakest terminal in case of Q(V) strategy without storage ............... 63
Figure 5.17: Comparison of the behaviour of Q(V) strategy at different terminals: a) daily voltage
measurements, b) active and reactive power injection ................................................................. 63
List of Figures
ix
Figure 5.18: Voltage at the weakest terminal in case of Q(V) strategy with battery storage ....... 64
Figure 5.19: Weekly grid losses comparison for the reactive power control strategies ............... 65
Figure 5.20: Weekly losses of all the battery systems ................................................................... 65
Figure 5.21: Maximum reactive power demand from the external grid ....................................... 66
Figure 5.22: Difference in reactive power requirement between PF(P) and Q(V) strategies ........ 67
Figure 5.23: Quality index comparison among the reactive power control strategies ................. 68
Figure 5.24: Voltage at the weakest terminal in case of DAPC strategy without storage ............. 69
Figure 5.25: Operation of DAPC strategy without storage for one of the PV systems connected in
the network .................................................................................................................................... 69
Figure 5.26: Voltage at the weakest terminal in case of DAPC strategy with battery storage ...... 70
Figure 5.27: Operation of DAPC strategy with battery storage for one of the PV systems
connected in the network .............................................................................................................. 70
Figure 5.28: Weekly grid losses comparison for the DAPC strategies ........................................... 71
Figure 5.29: Yield losses comparison ............................................................................................. 71
Figure 5.30: Overview of the performance of all strategies in the evaluation criteria: a) grid
losses, b) battery losses, c) maximum reactive power demand from the external grid, d) reactive
power performance index and e) yield loss ................................................................................... 73
Figure 5.31: Ranking of the studied voltage support strategies .................................................... 75
Figure 5.32: OEC comparison with respect to the PV-integration level according to the DNO’s
perspective ..................................................................................................................................... 76
Figure 5.33: OEC comparison with respect to the PV-integration level according to the PV system
owner’s perspective ....................................................................................................................... 76
Figure A.1: Geometry of overhead lines for a typical European LV network [48] ......................... 81
xi
List of Tables
Table 2.1: Interface protection settings in case of under-voltage ................................................. 13
Table 2.2: Interface protection settings in case of over-voltage ................................................... 13
Table 3.1: Operational intervals of PF(P) characteristic ................................................................. 22
Table 3.2: Parameters of PF(P) characteristic ................................................................................ 22
Table 3.3: Operational intervals of Q(V) characteristic .................................................................. 24
Table 3.4: Parameters of Q(V) characteristic ................................................................................. 24
Table 4.1: MV equivalent network parameters ............................................................................. 33
Table 4.2: Transformer characteristics .......................................................................................... 33
Table 4.3: Characteristics of the conductors available in the model ............................................. 34
Table 4.4: Line segments of the feeders and their corresponding lengths ................................... 35
Table 4.5: Inputs of "Load data process" block .............................................................................. 37
Table 4.6: Parameters of "Load data process" block ..................................................................... 38
Table 4.7: Solar irradiance data parameters [55] .......................................................................... 39
Table 4.8: Inputs of "Active Power Control" block......................................................................... 41
Table 4.9: Parameters of "Active Power Control" block ................................................................ 41
Table 4.10: Outputs of "Active Power Control" block .................................................................... 42
Table 4.11: Inputs of "Reactive Power Control" block ................................................................... 42
Table 4.12: Reactive power based control strategy selection ....................................................... 43
Table 4.13: Parameters of "Reactive Power Control" block .......................................................... 43
Table 4.14: Inputs of "Battery Control" block ................................................................................ 45
Table 4.15: Parameters of "Battery Control" block ....................................................................... 45
Table 5.1: Transformer and conductor loading for the worst case conditions (low load – high PV
generation) ..................................................................................................................................... 50
Table 5.2: PV hosting capacity of the network for the range of permitted power factors of the PV
generators ...................................................................................................................................... 55
Table 5.3: PV hosting capacity with respect to Vmax parameter..................................................... 62
Table 5.4: Selection of Vmax parameter for different PV-integration levels ................................... 62
Table 5.5: Differences of evaluation criteria .................................................................................. 72
Table 5.6: Relative weights of the evaluation criteria ................................................................... 74
Table A.1: Characteristics of the OHLs used in the model [48, 60] ................................................ 81
xii
List of Tables
Table A.2: Phase impedance matrix after Kron reduction (Ω/km) [48] ......................................... 81
xiii
List of Abbreviations
Abbreviation
Meaning
APC
active power curtailment
CHP
combined heat and power
DAPC
dynamic active power curtailment
DG
distributed generation (or distributed generator)
DNO
distribution network operator
DPL
DIgSILENT programming language
DSL
DIgSILENT simulation language
DSM
demand side management
DSO
distribution system operator
EN
European norm
FiT
feed-in tariff
GMR
geometric mean radius
HV
high voltage
HVDC
high voltage direct current
IEC
international electrotechnical commission
LV
low voltage
MV
medium voltage
OEC
overall evaluation criterion
OHL
overhead line
OLTC
on-load tap changer
PCC
point of common coupling
PF
power factor
PV
photovoltaic
QC
quality concept
RES
renewable energy sources
RMS
root mean square
STC
standard test conditions
TSO
transmission system operator
VDE
verband der elektrotechnik
VDEW
verband der elektrizitätswirtschaft
xv
Acknowledgments
First of all, I would like to express my gratitude to Peter Vaessen, and by extension DNV GL, for
the continuous guidance and monitoring throughout the completion of this thesis as well as for
the opportunity I was given to work on Watt connects project. Furthermore, I would like to
express my appreciation to my TU Delft supervisor, Pavol Bauer, for his trust and support.
Special thanks should be given also to my family for its endless support throughout these years.
Last but not least, I would like to thank Aspa for her patience and support.
1
Chapter 1
1 Introduction
1.1 Electric power systems and distributed generation history
Originally, public electricity supply was developed in the form of local generation feeding local
loads. The first complete electric power system was built by Thomas Edison; the historic Pearl
Street Station in New York city that served a number of factories, residences and street lighting
[1]. Thereafter, more individual small power systems were being built and operated by
independent companies providing electricity to a limited geographic region [2]. The systems
were isolated, without connections among them [3]. During the early years, up to around 1930,
this proved quite sufficient. However, it was then recognised that an integrated system was
needed to ensure an electricity supply that was both reasonably secure and economic. Improved
security resulted from the mutual emergency assistance that electric companies (often called
utilities) could provide. Improved economy resulted from the need for less generating reserve
capacity on each system [4].
It did not take very long for electric companies to realize that economies of scale caused a
dramatic lowering of costs [5]. An economy of scale simply means that it tends to be less
expensive to build and operate one large generator than several smaller ones [1]. This led to
large centrally located generators, built in areas close to large water reservoirs or near available
fuel supply routes.
Figure 1.1: Conventional electric power system [2]
As the electricity demand was increasing so did the need for transmitting larger amounts of
power over longer distances creating an incentive to use progressively higher voltage levels in
order to reduce current flow and therefore resistive losses in the lines [4]. Soon the familiar
2
Chapter 1: Introduction
power system structure was formed, generally split up into three parts: generation, transmission
and distribution (Figure 1.1). The transmission system interconnects all major generating
stations and main load centres in the system. It forms the backbone of the integrated power
system and operates at the highest voltage levels. Large industrial customers are often supplied
directly from the transmission system. The distribution system is the final stage in the transfer of
power to the individual customers and it is usually operated at medium and low voltage levels
[4].
However, from around 1990, there has been a revival of interest in smaller generating units
mainly based on renewable energy technologies and this has come to be known as distributed
generation (DG). The term describes electric power generation that is geographically distributed
or spread out across the grid, generally smaller in scale than traditional power plants and located
closer to the load, often on customers’ property [1].
There are various reasons for introducing these new types of production into the power system.
First, the open electricity market, which has been introduced in many countries since the early
‘90s, has made it easier for new players to enter the market. Second, in order to cope with the
environmental impact of conventional power plants and reduce the greenhouse gas emissions
the interest in renewable energy sources, such as sun and wind, is growing. Third, the margin
between the highest consumption and the likely available production is getting rather small for
some regions or countries. Building large conventional power stations is not always politically
acceptable for, among others, environmental reasons. It also requires large investments and can
take 10 years or longer to complete. Small-scale generation based on renewable sources of
energy does not suffer from these limitations. The total costs may be higher, but as the
investments can be spread over many owners, the financing may actually be easier [6].
Figure 1.2: Connection of distributed generators [2]
Distributed generation may be connected at a number of voltage levels from 120/230 V to 150
kV. Only very small generators may be connected to the lowest voltage networks, but large
1.2 Photovoltaic systems
3
installations of some hundreds of megawatts are connected to the busbars of high voltage
transmission systems (Figure 1.2) [2].
A wide variety of generating plant types is being connected to the medium and low voltage
distribution networks. Examples include the well-established technologies of small-scale hydro
generation, combined heat and power (CHP), wind turbines and PV systems, with the latter
being the centre of interest in this thesis.
1.2 Photovoltaic systems
Solar energy conversion into electricity takes place in a specially treated semiconductor device
that is called a solar cell. A solar cell is a unit that delivers only a certain amount of electrical
power. In order to use solar electricity for practical devices, which require a particular voltage or
current for their operation, a number of solar cells has to be connected together to form a solar
panel (also called PV module or PV panel). For higher generation PV panels are connected
together to form PV arrays [7].
Solar panels are only a part of a complete PV system. Their produced energy is transferred to the
load or to the electric grid by means of a subsystem that is generally referred to as the “balance
of system” (BOS). It encompasses all components of a PV system other than the PV panels and
may include the following [8, 9]:




supporting structures for mounting PV modules
power conditioning units, that adjust and convert the produced DC power to AC power
(inverters)
cables and protection devices, that allow a safe passage for current
storage devices that store PV generated electricity to be used when generation is not
sufficient
For many years, the major application of PV systems was off-grid for high-value, small electric
loads that were a long way from the nearest distribution network (e.g. vaccine refrigerators and
remote communication systems). PV systems have also been used as a power source for
satellites and space vehicles [3]. More recently, stimulated by financial support incentives, such
as Feed-in-Tariffs (FiT), their use as grid-connected distributed generators has increased
dramatically.
Most solar PV systems are installed on homes and businesses in developed areas. By connecting
to the local electricity network, owners can sell their excess or total energy production, feeding
it back into the grid. When solar energy is not available, electricity can be drawn from the grid.
Under a FiT regime, the owner of the PV system is paid by the local electricity provider for the
electric energy generated [10]. The photovoltaic modules may be roof mounted or incorporated
into the fabric of buildings in order to reduce overall cost and space requirements. Thus, these
PV installations (typically from 1 to 100 kW) are connected directly at customers’ premises and
so to the LV distribution network [2].
Large PV systems, with a capacity of hundreds of kilowatts (kW) to several megawatts (MW), are
usually connected in the MV distribution network. The solar panels of these systems are usually
4
Chapter 1: Introduction
mounted on frames on the ground. However, they can also be installed on large industrial
buildings such as warehouses, airport terminals or railway stations.
1.3 PV grid integration
Until recently, the PV market has been assisted in its development by financial support schemes,
such as Feed-in-Tariffs which have proven to be the most effective [11]. As a result, installations
of PV systems have grown over the past decade at a remarkable rate, even under difficult
economic circumstances (Figure 1.3). The majority of these PV systems are connected to the
grid, with off-grid ones accounting for less than 1 % of the installed PV capacity in Europe [12].
Europe remains the world’s leading region in terms of cumulative installed capacity, with almost
80 GW as of 2013. This represents about 60 % of the world’s cumulative PV capacity [13].
Figure 1.3: Evolution of global cumulative installed capacity 2000-2013 [13]
Europe’s market development is the result of a few countries that have taken the lead year after
year, with Germany showing a constant commitment from policymakers to support the
development of PV. Together with Italy the two countries have the largest portion of the
European installed PV capacity (Figure 1.4). Based on the capacity installed and connected to the
grid at the end of 2012, PV can currently provide a significant share of Europe’s electricity mix,
covering 2.6 % of the demand (Figure 1.5) and roughly 5.2 % of the peak demand [12]. The
increasing shares of PV electricity in the European energy mix is one of many factors that will
require modifications in the electricity system at national and European levels [11].
1.4 Problem statement
5
Figure 1.4: Evolution of European PV cumulative installed capacity 2000-2012 [12]
Figure 1.5: PV contribution to the energy demand in the EU 27 in 2012 based on cumulative installed
capacity in 2012 [12]
1.4 Problem statement
The continuously increasing installation of distributed photovoltaics in residential areas around
the world calls for detailed assessment of distribution grid impacts. Both photovoltaic generation
and domestic electricity demand exhibit characteristic variations on short and long time scales
6
Chapter 1: Introduction
and are to a large extent negatively correlated, especially at high latitudes (where electric
cooling devices are rarely used during midday hours in contrast with regions of lower latitudes).
With a more extensive integration of PV systems in residential areas, it is important to assure
that power quality and end-user voltages are not negatively affected [14].
On a distribution feeder, when the power consumed is lower than the power produced, a
reverse flow occurs. The electricity flows from the distributed generation to other consumers or
to higher voltage levels (Figure 1.6). As the distribution grid has not been built to host
distributed generation, reverse power flows introduce some issues that must be tackled. The
reverse flows may create grid bottlenecks that lead to voltage problems (voltage rise and
possible excess of the upper voltage limit) or equipment (lines or transformers) overload [11]:
Figure 1.6: Voltage profile on an exemplary feeder with distributed generation [11]
Voltage rise represents one of the main impacts of distributed generation [11]. In the case of
weak and/or long networks, voltage rise is common in times of low consumption and high power
feed-in by DG (Figure 1.6). Therefore, the possibility of upper voltage limit violation may limit the
PV-integration in a distribution network.
It is common to classify distribution grids in two different categories: urban and rural. Urban
grids supply many households in a small area; the transformer capacity and the number of
consumers powered are high and the length of the line is short. Rural grids are characterised by
much longer lines with lower transformer capacity and a lower number of powered households.
Overvoltage problems due to PV are virtually non-existent in urban networks as they are
characterised by a greater load and less space for PV. In rural grids, however, overvoltages are
more probable due to the longer line lengths and larger available space for PV installations [11].
1.5 Project objective and research questions
7
As mentioned before, most PV systems are connected to the LV distribution network. In
Germany, for example, about 70 % of the installed PV capacity is connected to the LV grid. South
German distribution grids, with a high concentration of PV systems, already experience gridintegration challenges related to PV. In some LV grids, the installed PV capacity can even exceed
the peak load by a factor of ten [15].
Taking into account the facts presented so far, the problem, which constitutes the motivation of
this project, is defined as follows:
“Voltage rise in a rural LV network with high PV penetration”
The problem has to be solved so as to simplify the integration of more RES in the power systems
and achieve a sustainable future energy supply.
1.5 Project objective and research questions
Aiming to achieve higher levels of PV integration by mitigating the voltage rise problem, new
studies, directives and standards demand from PV systems connected to the LV network to be
able to support the local voltage either by the provision of reactive power or by active power
curtailment.
The objective of this project is to test a selection of the suggested local voltage support
strategies and check their attributes, for a range of PV-integration levels, in a typical rural LV
network model. A series of research questions need to be answered:
i.
ii.
iii.
iv.
v.
What are the characteristics of a typical rural LV network?
What are the PV-integration limits of such a network and how much can they be
increased with the studied voltage support strategies?
What are the impacts of the studied voltage support strategies on grid operation and
how these depend on the PV-integration level?
How these impacts are influenced by the inclusion of battery storage in the PV systems?
Which strategy is the best candidate from the viewpoint of a DNO and a PV system
owner?
1.6 Research approach
The research approach is based on the following steps:
i.
ii.
iii.
iv.
v.
vi.
vii.
viii.
study of current regulations regarding the connection of DG in a LV network
study of the suggested voltage support strategies in literature
design of the appropriate network, load and PV-Battery system models
execution of a series of load flow calculations and simulations
analysis of the results
comparison of the studied strategies based on a set of technical evaluation criteria
weighted rating of the evaluation criteria by experts
estimation of the best voltage support strategy
8
Chapter 1: Introduction
1.7 Watt connects
Watt connects is a demonstration project on smart grids, initiated by three main partners, DNV
GL (formerly DNV KEMA), Alliander and TenneT. It consists of an interactive demonstration table
and simulation tools that create insight and enable new models, smart grid technologies and
research results to be tested, validated and presented to stakeholders. The most important
reason for using this procedure is to reduce the risk that a component, system or even an
entirely new technology fails, before it is used in the real world [16].
Figure 1.7: Watt connects interactive table
So far, simulations of power flows on the demonstration table for LV distribution networks,
which incorporate DG, have shown that, under specific conditions, severe over or under-voltages
can occur in the grid. One of the objectives of DNV GL, is to develop control algorithms that can
minimize problematic changes to voltage levels. Therefore, the objective of this research is well
suited in the Watt connects project and the validation of its results constitutes one of the
company’s future goals.
1.8 Thesis outline
Chapter 1 starts with a brief introduction on the transition electric power system is facing due to
the increasing penetration of DG. After a presentation of PV systems and the evolution of their
grid integration, the research problem is stated and the objective of the project as well as the
research questions and approach are defined.
Chapter 2 presents the requirements, set by the new regulations, regarding the connection of
distributed generators, such as PV systems, in the LV network.
Chapter 3 analyses the voltage rise problem and investigates the main voltage support strategies
which appear in literature as well as in new DG connection standards. Then, it categorises the
strategies based on their communication infrastructure requirements.
1.8 Thesis outline
9
Chapter 4 presents the details used for the design of the LV network, load and PV-Battery system
models. It finishes with a list of the assumptions been made.
Chapter 5 starts with an investigation whether the voltage rise problem is the limiting factor for
further PV integration in the LV network model. After an estimation of the PV-integration limits
without voltage support, it implements the selected voltage support strategies and tests their
attributes as the PV-integration level increases. Then, a comparison of the tested strategies
follows.
Chapter 6 ends the thesis report with the conclusions drawn from this research project and
outlines the recommendations for possible future work
.
11
Chapter 2
2 Regulations
2.1 General
Distribution grids must be managed to maintain the voltage level within specified limits and
deliver a high quality of power. Consumer electrical appliances are able to work only within
certain margins for voltage and frequency. To avoid their degradation, system operators have to
ensure that every consumer has access to secure and quality electricity [11].
In the past, system operators did not consider PV to be relevant for the electricity system. As a
result, they required every installation to switch off automatically and instantly as soon as a grid
problem (e.g. a voltage or frequency deviation) occurred. As distributed generation increased,
grid operators realised that new grid codes were urgently needed. The grid codes in some
countries have changed dramatically and now require that PV installations not only stay
connected during grid disturbances, but also that they be able to actively support system
operation. This step was being regarded more and more as absolutely necessary to guarantee
reliability and quality of electricity supply [11, 17].
New requirements were introduced progressively, first for installations connected to the HV
level, then to the MV level and finally to the LV grid. As an example, Figure 2.1 illustrates the
evolution of requirements for PV in Germany as PV deployment was increasing.
Figure 2.1: Evolution of the German grid codes and regulations compared with PV deployment (GW) [11]
12
Chapter 2: Regulations
Germany was one of the first European countries which adopted this new approach.
Consequently, the change has also been recognised by other countries as being trend-setting for
the new role that PV and distributed generation in general got to play [11, 17].
In the rest of this chapter focus is given on the German directive VDE-AR-N 4105 which PV
systems connected in the LV network have to follow from 1-1-2012. In addition, references will
be made to the European standards EN 50160 (Voltage characteristics of electricity supplied by
public electricity networks) and EN 50438 (Requirements for micro-generating plants to be
connected in parallel with public low-voltage distribution networks).
2.2 Voltage requirements
2.2.1 Supply voltage variations
According to the German directive, under normal operating conditions, the magnitude of the
voltage change at every PCC of the LV network, caused by all power generating stations
connected to this LV network, must not exceed the value of 3 % compared to the voltage when
these generating stations were not connected.
(2.1)
The European standard EN 50160 is also into force and requires the supply voltage variations not
to exceed ± 10 % of the nominal voltage Vn (for four-wire three phase systems Vn = 230 V
between phase and neutral). Specifically, during each period of one week 95 % of the 10 min
mean RMS values of the supply voltage shall be within the range of Vn ± 10 % and all 10 min
mean RMS values shall be within the range of Vn + 10 % / - 15 % [18].
2.2.2 Flicker
Voltage fluctuation causes changes of the luminance of lamps which can create the visual
phenomenon called flicker. Flicker is the impression of unsteadiness of visual sensation induced
by a light stimulus whose luminance or spectral distribution fluctuates with time. Above a certain
threshold flicker becomes annoying. The annoyance grows very rapidly with the amplitude of the
fluctuation. At certain repetition rates even very small amplitudes can be annoying [18].
The intensity of flicker annoyance, flicker severity, is evaluated by the following quantities:


short term severity (Pst) measured over a period of ten minutes with a flickermeter, a
specially designed instrument to measure any quantity representative of flicker [19]
long term severity (Plt) calculated from a sequence of twelve Pst-values over a two hour
interval, according to the following expression:
√∑
(2.2)
The European standard EN 50438 as well as the German directive VDE-AR-N 4105 state that
flicker created by power generating stations with rated currents lower than or equal to 16 A and
2.2 Voltage requirements
13
higher than 16 A but lower than or equal to 75 A per phase connected to the LV network should
comply with the relevant standards EN-IEC 61000-3-3 and EN-IEC 61000-3-11 respectively [20].
Furthermore, EN 50160 requires that under normal operating conditions, during each period of
one week the long term flicker severity Plt, caused by voltage fluctuation, should be less than or
equal to 1 for 95 % of the time [18].
2.2.3 Voltage dips/swells
Voltage dip is defined as the temporary reduction of the RMS voltage at a point in the electrical
supply system below a specified start threshold. Typically, a dip is associated with the
occurrence and termination of a short circuit or other extreme current increase on the system or
installations connected to it. For the purpose of EN 50160 the dip start threshold is equal to 90 %
of the reference voltage and the voltage dip duration is from half cycle (10 ms) up to and
including 1 min [18]. The interface protection settings required by EN 50438 and VDE-AR-N 4105
are presented in Table 2.1 [20, 21].
Table 2.1: Interface protection settings in case of under-voltage
Standard
Maximum disconnection time
EN 50438
VDE-AR-N 4105
1.5 s
0.2 s
Minimum operate
time
1.2 s
0.1 s
Trip value
230 V - 15 %
230 V - 20 %
Voltage swell is defined as the temporary increase of the RMS voltage at a point in the electrical
supply system above a specified start threshold. Voltage swells may appear between live
conductors or between live conductors and earth. Depending on the neutral arrangement, faults
to ground may also give rise to over-voltages between healthy phases and neutral. For the
purpose of EN 50160 the swell start threshold is equal to 110 % of the reference voltage and the
voltage swell duration is from 10 ms up to and including 1 min [18]. The interface protection
settings required by EN 50438 and VDE-AR-N 4105 are presented in Table 2.2 [20, 21].
Table 2.2: Interface protection settings in case of over-voltage
Standard
EN 50438
stage 1
stage 2
VDE-AR-N 4105
Maximum disconnection
time
3s
0.2 s
0.2 s
Minimum operate
time
0.1 s
0.1 s
Trip value
230 V + 10 %
230 V + 15 %
230 V + 10 %
2.2.4 Voltage unbalance
Voltage unbalance is the condition in a polyphase system in which the RMS values of the line-toline voltages (fundamental component), or the phase angles between consecutive line voltages,
are not all equal. The degree of the inequality is usually expressed as the ratios of the negative
and zero sequence components to the positive sequence component [18].
14
Chapter 2: Regulations
According to EN 50160, under normal operating conditions, during each period of one week, 95
% of the 10 min mean RMS values of the negative phase sequence component (fundamental) of
the supply voltage shall be within the range 0 % to 2 % of the positive phase sequence
component (fundamental).
VDE-AR-N 4115 puts a limit of 4.6 kVA to the allowed unbalance of power injection of the
generating stations connected to the LV network. Hence, a maximum plant power of 13.8 kVA
results when using single-phase, uncoupled inverters (3 x 4.6 kVA) only [21, 22].
2.3 Power factor - Reactive power capability
The requirement of the European standard EN 50438 is that the inverter based micro-generators
connected to the LV network should operate, under normal steady-state operating conditions,
across the statutory tolerance band of nominal voltage, between power factors of 0.9 leading
and 0.9 lagging, provided the output active power of the micro-generator is above 20 % of its
nominal value. When the active power output is less than 20 % of its nominal value the microgenerator should not exchange more reactive power than 10 % of its nominal active power
(Figure 2.2) [20].
Figure 2.2: Reactive power capability in load reference frame [20]
Figure 2.3: Reactive power control characteristic [20]
2.4 Active power curtailment
15
According to EN 50438 the reactive power control mode should be based on a configurable
characteristic curve defined by the DSO (Figure 2.3). The micro-generator shall be capable of
operating in the following control modes:



Q(V), voltage dependent reactive power control
fixed power factor (fixed PF)
PF(P), active power dependent power factor
Additional to the characteristic, the dynamic response of the control should be configurable. The
dynamics of the control should correspond with a first order filter having a time constant that is
configurable in the range of 3 s to 60 s. The time to reach 95 % of a new set point due to a
change in voltage will be 3 times the time constant [20].
The German directive VDE-AR-N 4105 requires from the inverters to feed in with power factors
up to 0.95lagging/leading from an apparent plant power of 3.68 kVA whereas, if the plant power
exceeds 13.8 kVA, even power factors up to 0.9 must be supported. It suggests that the fixed PF
method be used in case of generators with constant active power generation, like CHPs. When it
comes to generators with fluctuating generation, it recommends the use of droop-based
strategies such as PF(P) and Q(V). It is also more specific in the requirements of the PF(P)
characteristic (Figure 2.4).
Figure 2.4: PF(P) characteristic according to VDE-AR-N 4105 [22]
The respective inverter must feed in without phase shift up to half of its nominal active power.
Thereafter, it is to be steadily increased until it operates at full nominal power with the
maximum power factor (underexcited) valid for the respective plant [22].
2.4 Active power curtailment
The distribution grid operator should also be able to remotely limit the power of PV plants in
increments of no more than 10 % of the nominal power of the plant in the LV grid (in that
context, proven increments are 60, 30, or 0 % of the nominal power). Among others, conceivable
reasons for a power limitation include the operation of emergency generating units, a short term
16
Chapter 2: Regulations
overload of the superordinate medium-voltage or transmission grid or a system-endangering
frequency increase. This requirement of the German directive applies to all plants with more
than 100 kW of power and is otherwise comparable to that in the medium voltage directive [22].
In contrast, the Renewable Energy Sources Act (EEG), valid from the beginning of 2012, applies
the remote power limitation capability also to power plants with a nominal power of less than
100 kWp. However, operators of PV plants with less than 30 kWp are given the choice to skip
installing the device for remote power limitation if they accept in return a general limitation of
feed-in power to 70 % of the installed generator power [22].
17
Chapter 3
3 Voltage rise and mitigation solutions
One of the main impacts of the high number of PV systems and other distributed generators
along LV networks is voltage rise. This chapter gives a brief introduction on how PV power feedin can influence the voltage and presents some of the solutions available to overcome the
problem.
3.1 Voltage rise in a LV feeder
Without PV generators installed, voltage along distribution feeders would typically drop from
the substation to the remote end due to line impedances and loads. With integration of PV
generation, voltage profile improves as voltage drop across feeder segments reduces due to
reduced power flow through the feeder. However, if PV generation is greater than the local
demand at the connection point (CP) of the PV inverter, the surplus power flows back to the
grid. The excess power from PV systems may produce reverse power flow in the feeder (Figure
3.1) which would create voltage rise (Figure 3.2). Typical peak time of PV generation is noon,
when solar irradiance level is higher. Household demand, on the other hand, is typically lower at
this time of the day. LV distribution feeder may therefore experience voltage rise resulting from
low load demand and high PV generation [23].
Figure 3.1: Simplified schematic of a PV generator and a load connected to a LV feeder
18
Chapter 3: Voltage rise and mitigation solutions
b
a
o
δ
φ
I
c
VCP
VG
jIX
IR
Figure 3.2: Phasor diagram of voltage rise calculation
According to Figure 3.2 and the fact that the voltage angle between VG and VCP is normally very
small [24], the voltage rise along the feeder can be approximated as follows:
⇒
⇒
(3.1)
Where:
(
)
(3.2)
(
)
(3.3)
(3.4)
3.2 Solutions to voltage rise problem
Several solutions have been suggested until now in order to cope with the voltage rise
phenomenon at high PV penetration levels in LV networks:





grid reinforcement
use of MV/LV transformers with on-load tap changers (OLTC)
reactive power consumption by PV inverters
active power curtailment
storage
3.2.1 Grid reinforcement
According to equation (3.1), a decrease of the conductor impedance would result in lower
voltage rise. In order to achieve that decrease, conductors with larger cross section should be
used. That is actually the typical approach applied so far, prior to other solutions contributing to
the voltage control, which involved increasing the grid capacity by upgrading the distribution
3.2 Solutions to voltage rise problem
19
transformers to a larger power rating or by reinforcing the LV feeders by addition of parallel
conductors or replacement of old conductors with higher ampacity ones. However, this is a
rather costly procedure, especially when underground cables are used, and as a result DSOs try
to avoid or postpone these costs, since network elements such as cables and transformers are by
far not used up to their full capacity yet [25].
In Germany, for example, if a PV system cannot be interconnected due to technical reasons, the
distribution system operator is obliged, according to German law, to conduct necessary grid
reinforcement measures immediately. This is done by either replacing transformers in the grid or
reinforcing certain conductors. By law, the costs for these reinforcement measures must be
borne by the DSO [15].
3.2.2 MV/LV transformers with OLTC
According to EN 50160 the permissible voltage range for customers connected to low voltage
distribution grids is 230 V ± 10 % (voltage between line and neutral). Currently, voltage
regulation within the electricity grid is mainly limited to the OLTC of the HV/MV transformer.
Thereby, the voltage at the MV terminals of the transformer can be adapted to current network
conditions in order to keep the voltage within the permissible limits. The voltage level at the
HV/MV substation is usually set up to 104 % for the high load case. So far the tap changing of a
MV/LV transformer is mainly performed off-load and the voltage level may be as high as 106 %
[26]. With this setting aiming to cover a high load scenario, a voltage rise of only 4 % of the
nominal voltage is allowed until the upper voltage limit is reached under low load conditions and
high PV generation.
An effective way of controlling the voltage can be provided by a substation which features an
MV/LV transformer with OLTC. The voltage level at the LV bus of the transformer is no longer
dependent on the fluctuating voltage level at the MV side, but it may be controlled, for instance,
by means of a constant target value and a control range using the variable transformation ratio.
A tap change is required when the control range is exceeded. In this way, the allowable voltage
rise due to the feed-in of distributed generators would increase from currently 3 % up to 10 %
(upper limit set by EN 50160). The usage of this wider voltage band could allow such a high
increase of distributed generators that a restriction of new installation may no longer be caused
by voltage band limitations but by the exceeding of the ratings of existing network elements
[25].
Several OLTC control strategies exist. The more conventional ones use fixed voltage set points at
the transformers LV terminal, line-droop compensation or local power flow measurements,
while the more high-end ones utilize remote voltage measurement values from smart meters of
connected customers. The former could save the investment and operational costs for the
additional information and communication technology but may have other technical drawbacks
like unintended tap settings, due to a misinterpretation of locally measured values. The latter, on
the other hand, may require adaptations in the location of the measurement units when
possible changes within the LV network take place [27].
Pilot projects are already under development in order to assess the efficacy of the MV/LV
transformers with OLTC. In south-west Germany, for instance, a DSO has equipped one of its
20
Chapter 3: Voltage rise and mitigation solutions
rural MV/LV substations with a regulated 400 kVA transformer prototype in order to test its
benefits [28]. The general results were encouraging. However, analysing the data gathered
during the first weeks of operation showed that flicker problems can occur when the control
algorithm makes the tap-changer switch back and forth in short succession. Still, the solution is
not mature enough to be accepted as being feasible [29].
3.2.3 Reactive power control strategies
Going back to Figure 3.1 and equation (3.1), assuming that the maximum allowed voltage limit at
the connection point has been reached, if one wishes to further increase the amount of injected
active power without further increasing the voltage, the solution could be the use of a certain
amount of reactive power. Assuming that grid voltage remains unchanged:
(3.5)
Thus, one can calculate the change of reactive power required at the connection point, to keep
its voltage constant at the maximum voltage allowed, for a certain increase in the injected active
power. From the equation above:
(
)
(
)
⇒
⇒
(3.6)
From equation (3.6), one sees that as the factor R/X increases, more reactive power injection
from the external grid will be required to prevent over-voltage. Thus, the effectiveness of a
voltage control through reactive power management depends on the network characteristics.
The more inductive the network impedance, the easier the voltage can be controlled via reactive
power management [30].
A highly effective way to contribute to grid voltage support is to utilize the ability of modern PV
inverters to provide or absorb reactive power while feeding active power into the network. With
a reactive power consumption while active power is fed in, the voltage rise can be limited.
Modern PV inverters are usually able to operate between 0.9 lagging and leading power factors.
When their capacity is increased by around 11 % of their rated active power, the additional
capacity can be used for absorbing reactive power from the grid in order to decrease voltage [31,
32].
Various reactive power control methods, without requiring communication infrastructure, have
been proposed so far for the grid over-voltage limitation. These strategies can be mainly
grouped as:



fixed power factor (fixed PF)
PF in terms of injected active power, PF(P)
voltage-dependent reactive power, Q(V)
3.2 Solutions to voltage rise problem
21
Each of these methods is defined by using either a constant value and/or first order piecewise
equations that can be easily implemented in the inverter controllers and be modified even
remotely [33].
3.2.3.1 Fixed PF
According to the German directive VDE-AR-N 4105 the fixed PF method is more suitable for
generators with constant active power production, like CHPs. In case of generators with
fluctuating generation, it recommends the use of droop-based strategies such as PF(P) and Q(V).
Using the fixed PF method, the absorbed reactive power is proportional to the active power.
Thus, in case of PV systems, during low irradiance, the absorbed reactive power will also be as
low as the active power generation, by keeping the proportionality equal (Figure 3.3). At 100 %
power generation, the generator injects the maximum reactive power possible (Q lim), inductive
or capacitive.
Figure 3.3: Fixed PF characteristic
3.2.3.2 PF in terms of injected active power PF(P)
When the power production is low, the potential risk of the grid over-voltage becomes smaller
as well, since all produced real power may be consumed locally, without sending excessive
power to the MV network. In this case, the reactive power injection will be unnecessary as it
apparently creates additional network losses. The PF(P) method can improve this drawback
using a characteristic curve such as the one presented in Figure 3.4. A power factor versus active
power dependency is defined using a piecewise linear curve. The capacitive part of the curve can
be considered optional when only voltage rise is the problem of interest. By adjusting the
parameters of this curve the PF(P) control can be modified as required by the DSO.
22
Chapter 3: Voltage rise and mitigation solutions
Figure 3.4: PF(P) characteristic
The operational intervals which are presented in Figure 3.4 are explained in Table 3.1.
Table 3.1: Operational intervals of PF(P) characteristic
Interval
Description
A
Capacitive operation of the PV inverter; the main
objective is to increase the mains voltage.
B
Dead-band in which the controller is not injecting reactive
power for a predefined power range.
C
Inductive operation of the PV inverter; the objective is to
decrease the mains voltage.
The relevant parameters which define the PF(P) curve are explained in Table 3.2:
Table 3.2: Parameters of PF(P) characteristic
Parameter
Description
Pmin
Active power threshold below which the PV inverter should operate with the
minimum capacitive power factor.
Pdb,min
Minimum active power of dead-band interval. For lower active power values
overexcited (capacitive) inverter operation is chosen while for higher values
the PV inverter does not inject any reactive power.
Pdb,max
Maximum active power of dead-band interval. For lower values the PV
inverter does not inject reactive power while for higher values the
underexcited (inductive) inverter operation mode is chosen.
Pmax
Active power threshold above which the PV inverter should operate with the
minimum inductive power factor.
PFlim,cap
Minimum (capacitive) power factor value that the PV inverter operates.
PFlim,ind
Minimum (inductive) power factor value that the PV inverter operates.
3.2 Solutions to voltage rise problem
23
One property of this type of control is that inverters will inject reactive power regardless of the
location in the feeder. Since all inverters in the network are taking part, an overall better control
of the voltage is assumed. However, it may still be the case that the inverters might inject
reactive power to the network even though it may not be required (no significant voltage drop
or voltage rise situation).
3.2.3.3 Voltage-dependent reactive power Q(V)
The methods given so far support the grid voltage indirectly by using only local active power
measurement as input. In all these methods, it is assumed that the grid voltage level increases
with the produced active power from PV inverters, regardless of load variation. Nevertheless,
when the high irradiance level coincides with a significant power demand, then simply the
voltage rise may not reach to the critical value. The Q(V) method directly uses local voltage
information that is a consequence of the power production and consumption in the
neighbourhood [33]. This control needs to take also into account the standard tap changer
position of the MV/LV transformer.
The Q(V) control is normally implemented as in Figure 3.5. A reactive power versus voltage
dependency is defined using a piecewise linear curve. The optionally dead band, commonly used
in medium voltage applications, allows for a delay of reactive power injection in favour of active
power yield but can also be omitted [34]. The voltage at the inverter bus terminals is used as an
input value to the controller. This voltage may be computed as the averaged RMS value of the
three phases and expressed in per unit system. A low pass filter is added to the controller in
order to increase stability by making the controller slower (e.g. the inverter will not interact with
faster automatic voltage regulators) [29].
Figure 3.5: Q(V) characteristic
Analysing the Q(V) curve there are several defining parameters and intervals as shown in Figure
3.5. The operational intervals defined are explained in Table 3.3.
24
Chapter 3: Voltage rise and mitigation solutions
Table 3.3: Operational intervals of Q(V) characteristic
Interval
Description
A
Capacitive operation of the PV inverter; the main
objective is to increase the mains voltage.
B
Dead-band in which the controller is not injecting reactive
power for a predefined voltage range.
C
Inductive operation of the PV inverter; the objective is to
decrease the mains voltage.
The relevant parameters shown in the figure above are explained in Table 3.4:
Table 3.4: Parameters of Q(V) characteristic
Parameter
Vmin
Description
Voltage threshold below which the controller should apply maximum
capacitive reactive power at the inverter’s terminals.
Vdb,min
Minimum voltage of dead-band interval. For lower voltages overexcited
(capacitive) inverter operation is chosen while for higher values the PV
inverter does not inject any reactive power.
Vdb,max
Maximum voltage of dead-band interval. For lower values the PV inverter
does not inject reactive power while for higher values the underexcited
(inductive) inverter operation mode is chosen.
Vref
Reference voltage for dead-band selection. This parameter is chosen
according to the selected output voltage of the MV/LV transformer at the
secondary side (LV side) and in accordance with the voltage tap-setting. This
parameter has no other purpose but to correctly determine suitable values
for Vdb,min and Vdb,max.
Vmax
Voltage threshold above which the controller should apply minimum
(inductive) reactive power at the inverter’s output terminals.
Qmin
Qmax
Minimum reactive power generated by inverter. This parameter refers to the
underexcited reactive power operation of the inverter. The value can reach
up to the maximum underexcited reactive power capability of the inverter.
Maximum reactive power generated by inverter. This parameter refers to
the overexcited reactive power operation of the inverter. The value can
reach up to the maximum overexcited reactive power capability of the
inverter.
Parameter Vref may be chosen based on the rated PCC voltage and the position of the tap
changer on the MV/LV transformer.
The parameters Vmin and Vmax are usually chosen depending on the applicable lower and higher
voltage limits of the inverters. According to EN 50160 the 10 min mean RMS values of the supply
voltage should not to exceed ± 10 % of the nominal voltage Vn. When PV inverters are needed to
comply also with the German directive VDE-AR-N 4105, an additional limitation must be taken
3.2 Solutions to voltage rise problem
25
into account. The value of Vmax should be selected in such a way that the voltage change at the
PCC must not exceed 3 % compared to the voltage when distributed generating stations were
not connected.
Parameters Vdb,min and Vdb,max are defining the width of the voltage dead-band in which the Q(V)
control should not generate any reactive power. This region should restrict the inverters
injecting unnecessary reactive power when small variations, around the nominal prescribed
value, are present at the LV side of the distribution transformer. A too broad dead-band could
also have negative effects, since inverters closer to transformer station might not participate at
all in regulating the voltage, while inverters at the remote ends would provide maximum
reactive power.
3.2.4 Active power curtailment (APC)
Reactive power control may result in higher currents and losses and also in lower power factors
at the input of the feeders, especially in LV systems where voltages are less sensitive to reactive
power due to more resistive feeder characteristics. In addition, the apparent power of PV
inverters might have to be increased in order to provide reactive power without losing part of
their active power capability when reactive power has to be injected.
At a first glance, the option of active power curtailment could be seen as very attractive from
DSOs, as it postpones grid reinforcement and therefore could be considered, in some cases, as a
way to increase PV integration. In Germany, as discussed in paragraph 2.4, all PV systems
connected from 2012 onwards are required to provide a remote active power reduction
capability. In some other countries, such as Canada, reactive power cannot be injected by
inverters connected in the LV system, as it is forbidden for small power producers to interfere
with voltage control of the feeder using reactive power [35]. Thus, voltage rise mitigation
through reactive power absorption cannot be used in these cases.
P
Pav
Vl
Vh
V
Figure 3.6: P(V) characteristic
As in LV systems the relationship between voltage and active power is stronger than that with
reactive power, given the highly resistive line characteristics, droop-based APC methods have
been proposed [31, 35]. They are normally implemented as in Figure 3.6.
Up to a lower voltage threshold (Vl) no power curtailment is taking place and all the available
active power (Pav) is injected into the grid. When this threshold is exceeded, active power is
26
Chapter 3: Voltage rise and mitigation solutions
linearly reduced and from the time the voltage at the inverter terminals reaches the higher
voltage threshold (Vh) the PV system ceases its production [35].
The problem of this droop-based APC is that if the parameters of the P(V) curve are the same for
all PV inverters in a LV network, PV owners who live at the end of the LV feeders will be the first
to be affected by the curtailment while those closer to the MV/LV transformer may even not be
affected at all. The specific parameterization of the curve in order to equally share the curtailed
total power among all the PV owners makes this method not so easily exploitable.
To avoid unfair treatment and discrimination among PV producers a maximum guaranteed
active power injection could be offered to every household connected in the LV network. This
limitation could be determined for each LV network as the maximum amount of net generation
per household before technical problems, such as voltage rise, start to appear. Then the PV
power injection could be higher than the agreed limit by the amount of load consumption at the
PCC. This dynamic active power curtailment (DAPC) strategy is presented in Figure 3.7.
Analysis conducted in France showed that for MV rural feeders a 45 % PV capacity increase could
be achieved, if the loading level of the feeders were taken into account, for only 5 % loss of the
maximum possible PV energy generation throughout a year. Furthermore, for the same capacity
increase, this dynamic active power curtailment strategy led to one third of yearly energy losses
compared to the basic limitation [36].
Figure 3.7: Dynamic active power curtailment strategy for a feeder with 5 kW/household maximum net
generation
However, implementing such a strategy instead of reinforcing the network or allowing reactive
power injection would result to the customer’s decision whether to install a PV system of higher
capacity or not. In order to motivate the PV system owner to install a larger PV system and
produce more power than the value agreed with the DSO, self-consumption incentives could be
offered. Self-consumption mechanisms have recently been promoted in several European
countries. In some cases, pure net-metering schemes have been developed (such as in Belgium,
3.2 Solutions to voltage rise problem
27
Denmark and the Netherlands), while other countries (such as Germany) have favoured
mechanisms promoting an instantaneous consumption of the electricity produced [37, 38].
Either demand side management (DSM) techniques or storage solutions could lead the way to
an increase in self-consumption [11, 37, 39].
3.2.5 Storage
For mitigating voltage rise in a LV distribution feeder with PV, energy storage options have also
been proposed. The energy storage system can be a battery (bidirectional power flow) or a
controllable AC load (unidirectional power flow, e.g. heat pumps) [29]. When it comes to the
battery systems, two concepts in general stimulate the interest. The first concept relies on a
battery system which can provide voltage support for the whole feeder from a strategically
defined location while the second one is based on a number of distributed battery systems to
achieve the same target, for example, one battery system on every location with PV [40].
The centralised storage concept could face the voltage rise problem directly, as charge and
discharge control of the battery system could be based on voltage measurements on the most
critical nodes of the LV network, where the highest voltage rise is observed. The selection of the
position of this central battery system is also of major importance.
On the other hand, the distributed storage option could affect voltage rise only indirectly, as
charge and discharge control would be based on the instantly available photovoltaic generation
and load demand on the PCC [40]. The surplus of available PV generation can be stored in
batteries, which the PV owner will have in his property, in order to use it later in the day when
there would be no PV generation. In this way storage could act as the major facilitator for the
development of efficient self-consumption as it would increase the local consumption of PV
energy at the PCC without constraining the user in his consumption habits [41].
The difference of the concepts, regarding the direct and indirect treatment of voltage rise, lies
on their incentives. The DSO would install the battery system, at a strategically defined location,
so as to prevent voltage rise situations above the limit as well as to flatten the residual load
curve by significantly reducing the peak demand in the evening and avoiding excess reverse
power flow in the midday. In contrast, the incentive of a PV system owner to invest in the
installation of batteries in his property would be the potential profit from self-consumption.
In order for the concept of distributed storage to directly tackle the voltage rise problem it
should be combined either with reactive power absorption of PV inverters [40] or active power
curtailment [11]. This combination could positively affect both of the aforementioned strategies.
In the first case (reactive power absorption of PV inverters), it would result in less reactive power
demand by the LV network, which otherwise would rise up as the PV penetration increases.
Already, utilities in the south-western United States have started to encounter power factor
violations of the operating rules, laid down by the regional transmission organizations (RTO) and
independent system operators (ISO) who have oversight over their systems, and may incur fines
for operating their systems outside of the prescribed conditions [42].
In the second case (active power curtailment), storage would result in less PV generation being
curtailed, thus higher revenue for the PV system owner and larger interest from non-owners to
28
Chapter 3: Voltage rise and mitigation solutions
install their own PV systems. This would lead to both higher PV penetration and larger rates of
self-consumption in the LV networks.
The main disadvantage of the storage solution is the high cost of batteries. However, a second
trend in our society is a fast increasing market share of electric and hybrid cars. This has several
advantages: a massive increase in batteries will take place in the future resulting in lower prices
and an additional market for used batteries will appear. Batteries of electric and hybrid cars with
reduced capacity could still be usable for domestic applications [43].
3.3 Categorisation of strategies based on their communication requirements
Based on their communication requirements, three types of voltage support strategies can be
distinguished: local, decentralized and central (Figure 3.8) [15].
Figure 3.8: Categorisation of voltage support strategies based on their communication requirements [15]
3.3.1 Local strategies
Local control strategies do not require communication devices. The distributed generator reacts
to specific grid situations according to predefined parameters and droop functions, as well as
measurements (e.g. voltage or frequency) at its PCC.
PV inverters connected to distribution grids, using their active and reactive power control
capabilities, can contribute to lowering their impact (in terms of voltage rise) on the grid in times
of high solar irradiance. They reduce the need for additional grid reinforcement measures and
do not require any additional information and communication infrastructure. Therefore, local
voltage control strategies can be easily integrated into the overall grid operation [15].
3.3.2 Decentralised strategies
Decentralized strategies can be achieved via the coordination of several active system
components, automated and without regulation by the grid control centre of the system
3.3 Categorisation of strategies based on their communication requirements
29
operator. However, not only local measurements but also an information exchange is required
among single controllable entities, such as distribution substations and PV inverters, to increase
overall system performance.
3.3.3 Central strategies
In contrast to decentralized voltage support strategies, where subsets of the distribution system
are controlled independently, central control aims for coordinated control of the complete
system from the distribution system control centre. It thus requires a set of information, with
which to establish the current system status, as well as knowledge of the boundaries in which
the system needs to operate.
Decentralised and central approaches require significant investment in sensors, communication
equipment and control systems, which makes their application to massive DG conditions difficult
to implement [44]. As priority seems to be given to local voltage support strategies, in this thesis
only this kind of strategies, based on measurements on the PCC of each PV-battery system, are
going to be implemented and tested.
31
Chapter 4
4 Simulation setup
The voltage rise problem in a low voltage network caused by the high penetration of PV systems
and the suggested local voltage support strategies, presented in chapter 3, are analysed by a
series of RMS simulations as well as load flow calculations. The overall simulation model consists
of three main components, which are presented in detail in the following paragraphs:



the rural test network model
the load models
the PV-Battery system models
4.1 Rural test network model
In order to test the performance of the studied voltage support strategies in a LV distribution
network, a suitable and realistic test network model has to be implemented. The focus of this
thesis is on rural distribution networks because, on the one hand, the grid structures there are
generally rather weak (low level of interconnection, long distances, low load density, low short
circuit power) and, on the other hand, they are confronted with vast available spaces for
distributed generation [45]. Moreover, rural areas often have a high percentage of farms which
offer even more generous roof space for large PV arrays making the chance of violating voltage
limitations even higher [46].
Based on statistical investigations of LV distribution networks in Southern Germany [47], a rural
test network model is derived by appropriately selecting its characteristic parameters. Important
parameters describing the electrical characteristics of LV distribution networks are [34]:




the distance between neighbouring house connections
the typical number of households and feeders as well as the number of households per
feeder
the transformer power per household
the conductor type (cable, overhead line)
The test network model is implemented in DIgSILENT PowerFactory software. It consists of the
external grid, a MV/LV transformer, two feeders, house and farm loads, PV generators and
batteries. Figure 4.1 shows the single-phase diagram for the rural network structure.
32
Chapter 4: Simulation setup
Figure 4.1: Single-phase diagram of the test rural network structure
4.1 Rural test network model
33
4.1.1 External grid
The LV network is connected to the MV network via the transformer. The parameters of the MV
equivalent network of a typical European LV network, based on the Cigre LV benchmark [48], are
presented in Table 4.1.
Table 4.1: MV equivalent network parameters
Nominal system voltage (line to line)
Short circuit power
R/X ratio
20 kV
100 MVA
1
The same values are used for the “External Grid” element of the model (Figure 4.2).
Figure 4.2: External Grid element connected to the MV busbar of the transformer
4.1.2 MV/LV transformer
According to a statistical analysis of the LV distribution networks in southern Germany [32, 47],
the average transformer power is around 12 kVA per household. Moreover, the most common
rated powers of the transformers used are 100 and 160 kVA. Based on the number of
households (10) and farms (2) connected to the test network model, the selected rated power
for the transformer used in the model is 160 kVA. All the characteristics of the transformer
element are presented in Table 4.2 [49].
Table 4.2: Transformer characteristics
Rated power
Rated voltage (primary/secondary)
Short circuit voltage
Copper losses
Iron losses
Connection
160 kVA
20/0.4 kV
4%
2000 W
200 W
Dyn11
34
Chapter 4: Simulation setup
4.1.3 Feeders and conductor types
Two feeders are connected to the LV busbar of the MV/LV distribution transformer of the rural
test network model. The conductor types, which are used in Europe, can be either cables or
overhead lines (OHL). More specifically, in southern Germany both types have the same share in
rural LV networks [47]. Thus, either OHLs or cables can be selected in the test network model
created. The characteristics of the available conductors are presented in Table 4.3. More details
on OHL modelling in PowerFactory software can be found in Appendix A - OHL modelling.
Table 4.3: Characteristics of the conductors available in the model
Type
Material
OHL
Cable
Aluminium
Aluminium
Size
[mm2]
70
120
Rated current
[A]
270
265
Impedance
[Ω/km]
0.492 + j0.285
0.254 + j0.069
R/X impedance
ratio
1.7
3.7
According to the same statistical analysis, the distance between neighbouring house connections
in the case of rural LV networks in southern Germany varies between around 25 and 150 m
(Figure 4.3) [47].
Figure 4.3: Histograms and associated Weibull distribution of the average house distance for rural, village
and suburban networks [47]
The lengths of the line segments which are chosen for the test network model are presented in
Table 4.4.
4.2 Load model
35
Table 4.4: Line segments of the feeders and their corresponding lengths
Line segment
F1_LTR1
F1_L12
F1_L23
F1_L34
F1_L45
F2_LTR1
F2_L12
F2_L23
F2_L34
F2_L45
Length
[m]
20
60
80
80
80
50
60
60
60
60
4.1.4 Loads, PV generators and batteries
Five households are connected to each feeder. Additionally, two farms are connected to the first
feeder (upper one - Figure 4.1) on terminals 3 and 5, as shown, in more detail, in Figure 4.4. All
these households and farms constitute the loads of the test network model. The load models
used are further explained in paragraph 4.2.
Ten PV-Battery systems, each one connected to one of the ten PCCs available, constitute the
distributed PV generation and storage of the test network model. The components,
characteristics and control modes of the PV-Battery system model are further described in
paragraph 4.3.
Figure 4.4: Household load, farm load, PV generator and battery system connected to the last terminal of
the first feeder
4.2 Load model
The frame of the load model is shown in Figure 4.5. Each of its blocks is explained in the
following sections.
36
Chapter 4: Simulation setup
Figure 4.5: Load frame
4.2.1 “Load profile” block
Measurements of electric power consumption in a household, gathered by Électricité de France
(EDF) between December 2006 and November 2010, constitute the source of the load profiles
used in the case of households [50]. Ten different weekly load profiles, with 15 min average
values, corresponding to the summer season, were assigned randomly to the simulated
household loads. Figure 4.6 presents an example of a daily load profile used.
Figure 4.6: Example of daily household load profile
In the majority of the load profiles under consideration there was more intensive electricity
consumption in the evening than during the day. This behaviour corresponds to families with
employed adults and children going to school. These types of families constitute a high
4.2 Load model
37
percentage of the inhabitants in rural single family houses [51]. Thus, the load profiles used can
be assumed as a good calculation basis for an important group of users of PV and PV-Battery
systems in the simulation scenarios considered.
In the case of farms, daily load profiles were derived from the VDEW standard load profile for
German farms, with 15 min average values [52]. Two variations were created, which slightly
differ in time and peak power values, and each one was assigned to one farm load. Both farm
loads were assumed inductive with a constant power factor of 0.95.
Focusing on the region of southern Germany and more specifically in Bavaria, the average herd
size of Bavarian dairy farms is 30.9 cows [53]. Farms in this region have been mostly traditional
family farms. Based on this traditional operation the annual energy consumption was considered
high and a value of around 1,000 kWh/cow was assumed [54]. Thus the original VDEW load
profile was suitable time shifted and scaled for peak load demands of 7 kW (Figure 4.7) and 8
kW, creating two different farm load profiles which correspond to an annual energy
consumption of around 28,000 kWh and 32,000 kWh respectively.
Figure 4.7: Example of farm load profile
4.2.2 “Load data process” block
The role of the “Load data process” block is to allow the user suitably edit the power
consumption of the load by scaling the values defined by the load profiles and even set or
change the power factor of the load. The inputs and parameters required are presented in Table
4.5 and Table 4.6.
Table 4.5: Inputs of "Load data process" block
Name
P
Q
Unit
MW
MVAr
Description
Active power consumption as defined by the load profile
Reactive power consumption as it may be defined by the load profile
38
Chapter 4: Simulation setup
Table 4.6: Parameters of "Load data process" block
Name
scale
Unit
-
Description
ExtCtrl
-
Activation/deactivation of externally controlled (by load profile)
reactive power consumption
PF
-
Power factor value in the case of internally defined reactive power
consumption
Scale factor
The active and reactive power outputs of the block (Pext, Qext) are defined as follows:
(4.1)
{
(
)
(4.2)
The exact DSL model block definition can be found in Appendix B - DSL models’ code.
4.2.3 “Load” block
The “Load” block definition has as inputs the externally defined active (Pext) and reactive (Qext)
power consumption of the load as calculated by equations (4.1) and (4.2). The built-in load
model in PowerFactory, whose equivalent circuit is presented in Figure 4.8, is associated with
the “Load” block definition.
Figure 4.8: 3-phase load model in PowerFactory
4.3 PV-Battery system model
The frame of the PV-Battery system model is shown in Figure 4.9. The sections that follow
explain one by one the blocks which constitute the overall model.
4.3 PV-Battery system model
39
Figure 4.9: PV-Battery system frame
4.3.1 “Irradiance” block
Global solar irradiance data on an inclined surface are used as an input to the PV-Battery system
model. The corresponding city of the data is Munich. PV panels are assumed to face directly into
South with a 30o vertical inclination. More details are presented in Table 4.7.
Table 4.7: Solar irradiance data parameters [55]
Provider
Site latitude
Site longitude
Elevation
Tilt angle
Azimuth angle
Albedo of the ground
Year
Time resolution
MINES ParisTech - Armines
48.15o
11.58o
519 m
30o
180o
0.2
2005
15 min
As solar irradiance shows higher values during summer, the PV generation is also higher and
voltage rise problem more intense. Hence, a week of June is selected as the study period of the
simulation. Thus, irradiance data from 13/06/2005 to 19/06/2005 in W/m 2 define the output of
the irradiance block during the executed simulation (Figure 4.10).
40
Chapter 4: Simulation setup
Figure 4.10: Weekly irradiance profile used in the simulations
4.3.2 “PV generation profile” block
In order to create a PV generation profile (see also Appendix B - DSL models’ code) for each PV
system connected to the network, the irradiance profile as well as a selected maximum output
power for each PV generator is required. “PV generation profile” block definition has as input
the irradiance (I) from the “Irradiance” block and as output the available active power (Pav) in per
unit. The per unit power base (Sb) of its PV system is defined as the nominal apparent power
(SPV,nom) of the static generator (see paragraph 4.3.8) associated with the specific PV-Battery
system model.
(4.3)
Assuming maximum active power generation under 1000 W/m2 irradiance conditions, the
available active power in p.u. is calculated as follows:
(4.4)
For every PV static generator element the nominal apparent power (SPV,nom) in kVA has been
selected such that:
(4.5)
Where:
PFlim is the power factor operating limit of the PV generator
Pmax is the maximum active power capability of the PV generator in kW
4.3 PV-Battery system model
41
4.3.3 “Load Profile” blocks
Each of the PV-Battery systems is associated with either a household only or a household and a
farm load. Hence, the corresponding load profiles serve as input to the active power controller
and battery controller. More information on these load profiles are given in paragraph 4.2.1.
4.3.4 “Voltage” block
The Voltage block is associated with the voltage measurement device which is used to measure
the AC voltage at each terminal. The measured positive sequence voltage (v) in p.u. is then fed
as a signal into the active and reactive power controllers as well as to the battery controller.
4.3.5 “Active Power Control” block
The role of the “Active Power Control” block is to define the output active power injection of the
PV generator, based on the available PV power (Pav) as well as the possible limitations, control
modes and parameters specified by the user. The inputs and parameters required are presented
in Table 4.8 and Table 4.9. The exact DSL model can be found in Appendix B - DSL models’ code.
Table 4.8: Inputs of "Active Power Control" block
Name
Pav
Unit
p.u.
Description
Available PV power as calculated by the “PV generation profile” block
Pload,A
MW
Household load demand (active power)
Pload,B
MW
Farm load demand (active power)
Pbat
MW
Battery power (positive when charging)
v
p.u.
Voltage at PV generator’s terminals
Table 4.9: Parameters of "Active Power Control" block
Name
Unit
SPV,nom
MVA
DAPC
-
Pcurt,set
MW
T
s
Description
Nominal apparent power of the PV static generator
Activation/deactivation of dynamic active power curtailment mode
Active power limitation set by the DSO in case of DAPC
Time constant of the first-order lag filter (PT1)
Parameter DAPC enables the dynamic active power curtailment mode (see paragraph 3.2.4)
when it is set to 1 and disables it when it is set to 0. If DAPC mode is enabled, then an upper limit
in the active power injection is set, defined by the following equation:
(4.6)
Where:
{
If DAPC is disabled, then all the available active power (Pav) is injected into the grid.
(4.7)
42
Chapter 4: Simulation setup
The time constant (T) required is associated with the first-order lag filter (PT1) which is
incorporated in the Active Power Control model (see Appendix B - DSL models’ code). Such a
filter should be applied after power references in order to make sharp references slow down and
avoid output oscillations caused by iterations [31].
The outputs of the block are the active power (P) in p.u. that is going to be injected into the LV
network and the direct-axis current (id) in p.u. which serves as one of the inputs to the PV static
generator (see paragraph 4.3.8).
Table 4.10: Outputs of "Active Power Control" block
Name
P
id,ref
Unit
p.u.
p.u.
Description
Active power injection into the grid
Direct-axis current reference for the PV static generator
In order to work in the dq0 rotating frame a Park Transformation would be necessary so as to
convert the abc voltage signal. Because of the symmetrical and balanced conditions assumed
throughout the simulations and as the dq0 rotating frame is synchronised with the grid voltage
(no rotating mechanical parts in a PV inverter), if at t=0 the alignment of the frame is selected
properly, so that the d-axis coincides with phase A, then all the q and 0 components are zero and
the d component has the magnitude of the positive sequence voltage. Thus:
(4.8)
In general, the active power (P) is related to the d and q-axis voltage and current components
through the following equation:
(4.9)
Thus, as a result of equation (4.8) and (4.9):
(4.10)
4.3.6 “Reactive Power Control” block
The role of the “Reactive Power Control” block is to define the reactive power injection of the PV
generator based on the control mode, parameters and possible limitations specified by the user.
The inputs required are presented in Table 4.11.
Table 4.11: Inputs of "Reactive Power Control" block
Name
P
v
Unit
p.u.
p.u.
Description
Active power injection into the grid
Voltage at PV generator’s terminals
4.3 PV-Battery system model
43
The three reactive power based voltage support strategies, presented in chapter 3, are
implemented and the user is able to select the one preferred by changing the appropriate
parameter (mode) as shown in Table 4.12.
Table 4.12: Reactive power based control strategy selection
mode
0
1
2
Control strategy
fixed PF
PF(P)
Q(V)
The rest of the parameters that need to be defined in the Reactive Power Control model are
shown in Table 4.13.
Table 4.13: Parameters of "Reactive Power Control" block
Name
Unit
Description
PFfixed
-
Fixed power factor value of the fixed PF mode
PFlim
-
Power factor operating limit of the PV generator
Psp
-
Ratio of the maximum active power of the PV generator over which
the PV system starts absorbing reactive power when the PF(P) mode
is selected
Vsp
p.u.
Voltage set-point over which the PV system starts absorbing reactive
power when the Q(V) mode is selected
Vmax
p.u.
Voltage threshold above which the PV system should absorb the
maximum possible reactive power when the Q(V) mode is selected
T
s
Time constant of the first-order lag filter (PT1)
Figure 4.11 presents the implemented characteristic of the fixed PF strategy. Only the inductive
part of that of Figure 3.3 is considered as only voltage rise is the problem of interest in this
thesis.
Figure 4.11: Fixed PF characteristic
The equations related to the characteristic of Figure 4.11 are:
44
Chapter 4: Simulation setup
(
)
(
(4.11)
)
(4.12)
The selected power factor (PFfixed) should be within the operational limits of the PV generator.
Thus, for the needs of this study:
(4.13)
In case the controller is set to the PF(P) control mode then the characteristic presented in Figure
4.12 is enabled. This strategy has been extensively described in paragraph 3.2.3.2. However, as
shown in Figure 4.12, only the inductive part is considered for the needs of this thesis. Equation
(4.14) describes the PF(P) characteristic used:
(4.14)
{
(
)
Figure 4.12: PF(P) characteristic
The last control mode is associated with the Q(V) strategy and it is implemented as shown in
Figure 4.13.
Figure 4.13: Q(V) characteristic
4.3 PV-Battery system model
45
It is based on the following equation:
(
)
(4.15)
{
Where, Qlim is a function of instantaneous active power injection (P) and power factor limit (PFlim)
of the PV static generator:
(
)
(4.16)
The output of the block is the quadrature-axis current (iq) in p.u. which is connected to the
corresponding input (iq,ref) of the PV static generator (see paragraph 4.3.8). Regardless of which
control mode is used, the reactive power injection of the PV generator is related to the d and qaxis voltage and current components through the following equation:
(4.17)
And as a result of equation (4.8) and (4.17):
(4.18)
The exact DSL model can be found in Appendix B - DSL models’ code.
4.3.7 “Battery Control” block
The role of the “Battery Control” block is to manage the charging and discharging procedure of
the battery associated with it. The inputs and parameters required are presented in Table 4.14
and Table 4.15.
Table 4.14: Inputs of "Battery Control" block
Name
Pav
Unit
p.u.
Description
Available PV power as calculated by the “PV generation profile” block
Pload,A
MW
Household load demand
Pload,B
MW
Farm load demand
v
p.u.
Voltage at battery’s terminals
Table 4.15: Parameters of "Battery Control" block
Name
Unit
Description
EnableStorage
-
C
MWh
Energy capacity of the battery
Sbat,nom
MW
Nominal apparent power of the Battery static generator
SPV,nom
MW
Nominal apparent power of the PV static generator
T
s
Activation/deactivation of the battery
Time constant of the first-order lag filter (PT1)
46
Chapter 4: Simulation setup
Battery capacity (C), for each household and farm, is selected in such a way that part of PV
power generated is stored for later consumption within the day. Unlike autonomous dwellings,
for grid-connected ones, a battery system should not necessarily cope with a long period of low
PV generation because the grid is available as a back-up. Therefore, the target is on a battery
system which makes daily shifts of energy exchange. As no design rules for this purpose are
available in the literature [43], battery capacity for each household and farm in the model is
selected in such a way that it covers the average electricity demand of the time period between
19:30 and 07:00, which is a period of low or no irradiance.
The objective of the energy management of the battery system is to store the additional PV
energy (charge) when PV generation exceeds load demand and to provide energy (discharge)
when load demand exceeds PV generation. This functionality is limited by the capacity of the
battery system and the maximum charge/discharge rate (equation (4.19)). When the battery is
fully charged, the PV energy has to be injected directly into the grid. This energy is not used for
local consumption. When the battery is fully discharged the required electricity is provided by
the public grid.
As lithium-ion technology is assumed for the batteries, to prevent the effects of battery ageing
the active state of charge (SOC) range is limited to 60 % of the initial battery capacity (SOC range:
20 – 80 %). According to the modelling assumptions a constant efficiency (nbat) of 95 % is
considered for both charging and discharging states [41, 43]. Furthermore, the maximum rate of
charge/discharge is limited to 1 C [56]. Hence, the battery power (Pbat) in p.u. is:
(4.19)
{
Where:
(4.20)
∫
(4.21)
The coefficient a used in equation (4.21) is defined as follows:
{
(4.22)
4.4 General simulation assumptions
47
The other outputs of the block, the direct-axis current (id) and the quadrature-axis current (iq) in
p.u. serve as input to the Battery static generator (see paragraph 4.3.8) and are calculated, using
equations (4.19), (4.8), (4.9) and (4.17) as follows:
(4.23)
The exact DSL model can be found in Appendix B - DSL models’ code
4.3.8 “PV Generator” and “Battery” blocks
In PowerFactory a static generator can serve as a model of any kind of static (no rotating)
generator, such as:






photovoltaic generators
fuel cells
storage devices
HVDC terminals
reactive power compensators
wind generators
In the case of the PV-Battery system model, the two static generators used serve as the interface
of the PV and battery system, respectively, to the LV network. Their inputs are the direct and
quadrature axis reference currents which are associated with the active and reactive power
injected into the grid, as defined by their corresponding controllers.
4.4 General simulation assumptions
The following general simulation assumptions apply:








Household and farm loads are assumed to be balanced and 3-phase connected. No voltage
dependency is taken into account for loads as well.
No daily variation was assumed for farm load profiles.
Every PV generator and battery system is considered to be 3-phase connected.
Irradiance is considered the same for all household and farm PV systems.
The maximum active power generation for all the household PV systems is the same. The
same applies to farm PV systems although at different level than that of households.
No inverter and cable losses are taken into account for the PV-Battery systems.
The efficiency of the PV panels is regarded constant and based on STC conditions.
Voltage at the MV side of the transformer is assumed constant thus no voltage variation in
the MV grid is taken into account
49
Chapter 5
5 Implementation and test results
In this chapter the LV network model is firstly tested under high PV integration to verify that the
voltage rise problem indeed appears. Then the local voltage support strategies, presented in
chapter 3, are activated and tested one by one, as well as their combination with battery
storage, in order to check their effectiveness and behaviour for different PV-integration levels.
5.1 High PV integration without voltage support
Referring to a study performed by a German DSO, the average installed power for PV systems on
residential rooftops is 10 kWp, whereas on farm buildings it is 27 kWp [34]. Assuming that every
household and farm PV system of the test network model can produce power up to 10 kW and
27 kW respectively, it should be checked whether the voltage at the network’s terminals violates
or not the limit set by the German directive VDE-AR-N 4105.
Since off-load tap of LV distribution transformer is usually adjusted by estimating voltage drop
with maximum nominal load, only low load condition should be considered to investigate
voltage rise [57]. As the MV level is not part of the simulation’s interest, the voltage on the MV
side of the transformer is set fixed to 1 p.u.. It is also assumed that the maximum voltage at each
LV terminal must not exceed 1.03 p.u. in order to fulfil the requirements.
Thus, in order to cover the worst case scenario in terms of voltage rise, maximum PV generation
and low load conditions (base load of load profiles) are assumed, as a minimum load would
always be present in the system. Figure 5.1 and Figure 5.2 present the voltage profile along each
feeder of the LV network for both conductor types and for this worst case load flow calculation.
It is observed that voltage rise, for both feeders, is more intense when OHLs are used. Feeder 1
exhibits higher voltage rise than feeder 2 because of the higher total PV generation, due to the
farm PV systems that are connected into it. Its last three terminals clearly exceed the 3 % limit in
the case of OHL, while only the last two terminals exceed the limit in the case of cables.
Although none of the terminals of feeder 2 exhibits voltage higher than 1.03 p.u. when cables
are used, in the case of OHLs its last two terminals experience voltage rise slightly above the
limit.
The transformer and conductor loading for this worst case scenario are presented in Table 5.1.
As it is observed they do not constitute limiting factors for this operating condition of the
network under study.
50
Chapter 5: Implementation and test results
Voltage Profile - Feeder 1
1.06
voltage [p.u.]
1.05
1.04
1.03
1.02
1.01
1
OHL
Cable
20 (T1)
80 (T2)
160 (T3)
240 (T4)
distance of terminal from transformer [m]
320 (T5)
Figure 5.1: Voltage profile of feeder 1 under low load and high PV generation conditions (10
kW/household and 27 kW/farm) for both conductor types
Voltage Profile - Feeder 2
1.06
OHL
Cable
voltage [p.u.]
1.05
1.04
1.03
1.02
1.01
1
50 (T1)
110 (T2)
170 (T3)
230 (T4)
distance of terminal from transformer [m]
290 (T5)
Figure 5.2: Voltage profile of feeder 2 under low load and high PV generation conditions (10
kW/household and 27 kW/farm) for both conductor types
Table 5.1: Transformer and conductor loading for the worst case conditions (low load – high PV
generation)
Conductor
OHL
cable
Transformer loading
78.5 %
79.5 %
Maximum conductor loading
41.5 %
42.8 %
It is also a matter of interest whether the voltage limit is violated not only under low load
conditions but also under normal load conditions. For this purpose, weekly simulations are
executed, for both conductor types, using the load profiles presented in paragraph 4.2.1 and the
irradiance data presented in paragraph 4.3.1. As observed in Figure 5.1, the weakest point of the
5.2 Voltage sensitivity analysis of test network
51
test network is the fifth terminal of the first feeder. Thus, voltage measurements for this
terminal are shown in Figure 5.3.
Voltage - F1_T5
1.06
OHL
cable
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.3: Voltage measurements for the weakest point of the network under normal load conditions and
high PV integration (10 kW/household and 27 kW/farm) throughout the weekly simulation
The results show that the 3 % limit set by the German directive is clearly exceeded during the
simulations, when the irradiance observed is high (see Figure 4.10). It is also clear that the
problem is more intense when overhead lines are used. Therefore, voltage support strategies
are required to keep voltage under the limit when PV integration in the test LV network is high.
The level of PV integration over which voltage rise problem appears is studied in paragraph 5.3.
5.2 Voltage sensitivity analysis of test network
Before implementing the studied voltage support strategies suggested, which are based on
active and reactive power control, it would be useful to check the voltage sensitivity of the test
network to active and reactive power variations. Both cases of overhead lines and cables are
tested.
In general, voltage sensitivity analysis can be used for estimating the voltage variation due to a
small change in active or reactive power injection at a certain location. Voltage sensitivity
matrices for a network can be derived for both active and reactive power by solving non-linear
load flow equations using the Newton–Raphson algorithm, which provides a linear model
around the given operating point. The same matrices can be used to identify critical locations in
relation to load/generation conditions. Elements of resultant sensitivity matrices give the most
effective places to support voltage by regulating Q and P at related nodes. Moreover P-V and QV sensitivity matrices can be compared to each other for a specific network to determine which
system parameter input (P or Q) has dominant impact on grid voltage. The most remote node in
the feeder, presents the highest sensitivity value, thus it is the most critical location for active
power injection in relation to voltage variation [31, 40].
One can therefore calculate the expected small changes in angle (θ) and amplitude (V) of the
voltage for a given small change in the active and reactive power values:
52
Chapter 5: Implementation and test results
[
]
[
] [
]
[
]
(5.1)
Where:
J is the system Jacobian matrix which is updated at each load flow iteration, until convergence
tolerance is satisfied. Solving equation (5.1) for Δθ and ΔV the sensitivity matrix S arises:
[
]
[
]
[
] [
]
(5.2)
The voltage sensitivity matrix is composed of four sub-matrices with partial derivatives that
portray the variation in the voltage magnitude and angle of the buses due to variations in active
and reactive power at each bus. The diagonal elements of
and
represent the voltage
variation at a bus due to a variation of active and reactive power respectively at the same point.
The non-diagonal elements describe the voltage variation at a bus due to the variation in active
and reactive power at a different point on the network.
Applying this technique on the test network model, voltage sensitivities are calculated.
Regarding the worst case condition of voltage rise problem, the operating point is selected the
same as that of the previous paragraph (low load demand – 10 kW/household and 27 kW/farm
PV generation).
Figure 5.4 and Figure 5.5 plot the voltage sensitivity magnitudes of the diagonal elements of
and
for the two types of conductors (OHL and cable) that have different R/X impedance
ratios (Table 4.3). We should recall from equation (3.6) that, as the R/X ratio increases, higher
values of reactive power injection from the external grid will be required to prevent overvoltage.
Thus, the effectiveness of a voltage support strategy through reactive power management in the
test network is higher in case overhead lines are used as conductors.
Voltage sensitivities - Feeder 1
sensitivity [p.u./MW - p.u./MVAr]
1
OHL dv/dP
OHL dv/dQ
Cable dv/dP
Cable dv/dQ
0.8
0.6
0.4
0.2
0
20 (T1)
80 (T2)
160 (T3)
240 (T4)
distance of terminal from transformer [m]
320 (T5)
Figure 5.4: Voltage sensitivities to P and Q variation for the terminals of feeder 1
5.3 Reactive power control strategies
53
Voltage sensitivities - Feeder 2
sensitivity [p.u./MW - p.u./MVAr]
1
0.8
OHL dv/dP
OHL dv/dQ
Cable dv/dP
Cable dv/dQ
0.6
0.4
0.2
0
50 (T1)
110 (T2)
170 (T3)
230 (T4)
distance of terminal from transformer [m]
290 (T5)
Figure 5.5: Voltage sensitivities to P and Q variation for the terminals of feeder 2
With bigger impedance (OHL case, see Table 4.3), there is always higher voltage impact by both
active and reactive power variations. It is also observed from Figure 5.4 and Figure 5.5 that
is
higher than
at the terminals which are closer to the transformer. This happens because of
their shorter distance to the transformer. Therefore, short-circuit reactance of the transformer
becomes dominant over the short-circuit resistance at these locations. As the distance to the
transformer increases along the feeder, line resistance contributes more on the impedance so
that the active power control becomes more effective on the voltage support than the reactive
power control.
5.3 Reactive power control strategies
Recalling the German directive, PV inverters should be capable to feed in with power factors up
to 0.95 from an apparent PV system power of 3.68 kVA, while if PV system power exceeds 13.8
kVA, even power factors up to 0.9 must be supported. Thus during the simulation all the
household PV systems can be controlled to absorb reactive power with a power factor up to 0.95
whereas the two farm PV systems can be controlled up to 0.9.
It should be mentioned here that, because of the statistics presented in paragraph 5.1, a ratio
2.7:1 is considered for the assigned maximum active power injections of farm and household PV
systems throughout all the simulations and load flow calculations of this thesis. Selecting, for
example, a maximum PV power of 6 kW for a household PV system means that a farm PV system
has a capability of generating up to 16.2 kW (2.7 · 6 kW). Therefore, from this point forward,
when a value is mentioned for the maximum active power feed-in of a household PV system, it
would mean that a farm PV system injects 2.7 times this value.
One question that arises is:
“What is the PV hosting capacity of this test LV network depending on the power factor under
which the maximum active power of each PV generator is fed into it?”
54
Chapter 5: Implementation and test results
Figure 5.6: Flowchart of the script executed to calculate the PV hosting capacity of the network for
different power factors of the PV generators
5.3 Reactive power control strategies
55
In order to answer this question, a script is written in DPL (see Appendix C - DPL commands) and
its flowchart is presented in Figure 5.6. The script takes into account the worst case scenario,
thus low load conditions are assumed.
PV hosting capacity is defined as the maximum PV integration for which the network still
operates satisfactorily [30]. It is determined in such a way that PV power production per
household and farm is increased until a limiting factor reaches its corresponding limit levels [33].
Regarding the worst-case condition in the sense of voltage rise, low load conditions are
assumed. The PF is decreased from 1 to 0.95 for households with a 0.005 step and from 1 to 0.9
for farms with a 0.1 step. Thus eleven measurements in total are taken. For each PF set the
active power injection of each household PV system is increased with a 0.1 kW step (0.27 kW for
farm PV systems) and PV hosting capacity of the LV network is determined as soon as either:


the voltage at a terminal exceeds 1.03 p.u., or
the loading of the transformer or a line section exceeds 100 % of its rated power
This method to estimate the maximum PV integration in the network has the advantage of
creating a uniform distribution of PV power across the entire LV network. Practical cases may
differ from this situation though. The script was executed for both OHL and cables usage. The
results are presented in Table 5.2 and plotted in Figure 5.7.
Table 5.2: PV hosting capacity of the network for the range of permitted power factors of the PV
generators
Power Factor
Household Farm
1
1
0.995
0.99
0.990
0.98
0.985
0.97
0.980
0.96
0.975
0.95
0.970
0.94
0.965
0.93
0.960
0.92
0.955
0.91
0.950
0.9
PV hosting capacity [kW] (OHL)
Household Farm Total Gain
5.6
15.1 75.0
0%
6.4
17.3 85.8
14%
6.9
18.6 92.5
23%
7.3
19.7 97.8
30%
7.7
20.8 103.2 38%
8.1
21.9 108.5 45%
8.5
23.0 113.9 52%
8.9
24.0 119.3 59%
9.4
25.4 126.0 68%
9.9
26.7 132.7 77%
10.4
28.1 139.4 86%
PV hosting capacity [kW] (Cable)
Household Farm Total Gain
9.0
24.3 120.6 0%
10.4
28.1 139.4 16%
11.2
30.2 150.1 24%
11.9
32.1 159.5 32%
12.0
32.4 160.8 33%
11.9
32.1 159.5 32%
11.8
31.9 158.1 31%
11.7
31.6 156.8 30%
11.6
31.3 155.4 29%
11.5
31.1 154.1 28%
11.4
30.8 152.8 27%
The results show that in case of overhead lines the test LV network can host up to 86 % more PV
capacity with reactive power based voltage support strategies than without support. Voltage rise
is the limiting criterion for all the hosting capacity levels. When cables are used as conductors,
reactive power injection relocates the limiting criterion of hosting capacity from voltage towards
transformer loading from a certain hosting capacity upwards. This shift happens at
approximately 12 kW per household PV system which leads to an increase of only 33 % in total
PV capacity of the test network compared to the unity power factor case. However, maximum
PV hosting capacity in case of cables is around 15 % more than in case of overhead lines.
56
Chapter 5: Implementation and test results
180
OHL
cable
OHL
cable
PV hosting capacity per household [kW]
28
26
24
22
20
160
140
120
18
100
16
14
80
12
10
60
8
40
6
4
Total PV hosting capacity [kW]
30
20
2
0
0.95 0.955
0.96 0.965
0.97 0.975
0.98 0.985
0.99 0.995
1
0
Figure 5.7: Plot of the PV hosting capacity with respect to the power factor of a household PV system
Line loading cannot be a limiting factor in this network, under the simulated operating
conditions, as the thermal rating of both conductor types used is higher than that of the
transformer, thus in any studied case the transformer would be overloaded first.
From the analysis so far it is becoming clear that voltage rise problems are more intense in case
overhead lines are used, comparing conductors of almost equal current rating (Table 4.3).
Hence, for the rest of the analysis and simulations executed, implementing the studied voltage
support strategies, only the case of overhead lines is considered.
5.3.1 PF(P) control mode
In order to check the effectiveness and behaviour of the PF(P) voltage support strategy, the
corresponding control mode is activated in the reactive power controllers of the PV-Battery
systems. Then a series of simulations, under normal load conditions, are executed for PV
integration levels from 6 up to 10 kW/household, with a step of 1 kW/household.
Following the German directive, the Psp parameter (see Table 4.13) of the PF(P) characteristic,
presented in Figure 4.12, is set to 0.5. This means that when the PV power production exceeds
half of the PV system’s maximum power, the power factor starts linearly decreasing up to the
power factor limit (see paragraph 5.3) and reactive power is absorbed.
It should be noticed that the selection of the value for the Psp parameter can influence the
behaviour of the PF(P) strategy. For example, in the case of the test LV network under study, if
the active power threshold, over which PV generators start absorbing reactive power, is selected
larger than 5.6 kW, then the resulting characteristic could still cause voltage rise problems in
case of low irradiance and low load conditions, but it would be efficient for higher values of
irradiance. This case is presented in Figure 5.8, for 9 kW/household PV-integration level. The red
line represents the problematic characteristic whose threshold is set to 7 kW and the green one
the suggested by the German regulation whose threshold is set to 4.5 kW (Psp = 0.5). The blue
markers indicate the hosting capacity limit as given in Table 5.2.
5.3 Reactive power control strategies
57
PF(P) characteristics
hosting capacity limit
good characteristic
problematic characteristic
1
PF
0.99
0.98
0.97
0.96
0.95
0
1
2
3
4
4.5 5
5.6 6
Power [kW]
7
8
9
10 10.4 11
12
Figure 5.8: Selection of the appropriate PF(P) characteristic
In order to test the validity of the PF(P) voltage support strategy in case of maximum PV hosting
capacity (10.4 kW/household), a characteristic with Psp = 0.5 and Pmax = 10.4 kW is compared
with the hosting capacity limits as presented in paragraph 5.3. It is observed that every point of
the suggested PF(P) characteristic is below the hosting capacity limits which makes the PF(P)
strategy suitable for all the tested PV-integration levels.
PV hosting capacity - PF(P) characteristic
hosting capacity limit
PF(P) characteristic
1
PF
0.99
0.98
0.97
0.96
0.95
0
2
4
5.2 5.6 6
Power [kW]
8
10 10.4
12
Figure 5.9: Comparison of the selected PF(P) characteristic with the hosting capacity limits in case of
maximum PV hosting capacity
5.3.1.1 PF(P) without storage
As resulted by the series of simulations, PF(P) strategy manages to keep the voltage below the
limit set by the German regulation. Figure 5.10 presents the voltage of the weakest terminal for
6 and 10 kW/household PV-integration levels.
58
Chapter 5: Implementation and test results
Voltage - F1_T5 - PF(P) without storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.10: Voltage at the weakest terminal in case of PF(P) strategy without storage
The active and reactive power injection by a household PV system, for a maximum active power
of 10 kW, is presented in Figure 5.11. It can be noticed that when the active power generated by
a PV system is below 5 kW (half of the maximum active power) no reactive power is absorbed.
Active and Reactive Power by a household PV System
12
P
Q
10
power [kW - kVAr]
8
6
4
2
0
-2
-4
0
24
48
72
96
120
144
168
time [h]
Figure 5.11: Active and reactive power injection of a household PV system for a PV-integration level of 10
kW/household
5.3.1.2 PF(P) with battery storage
The same series of simulations are executed again but now the battery storage option is also
activated. The results concerning the voltage at the weakest terminal are presented in Figure
5.12, for 6 and 10 kW/household PV integration levels.
5.3 Reactive power control strategies
59
Voltage - F1_T5 - PF(P) with battery storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.12: Voltage at the weakest terminal in case of PF(P) strategy with battery storage
Long time periods when the voltage is very close to 1 p.u. are observed, which indicate battery
charging and discharging states. When batteries are fully charged there is still some voltage rise
which is kept below the limit via the reactive power absorption of PV systems. The active and
reactive power injection by a household PV system, for a maximum active power of 10 kW, is the
same as the one presented in Figure 5.11.
Figure 5.13 presents an example of the state of charge of a battery system during a weekly
simulation. It is shown that the target of daily shifts of energy exchange, as discussed in
paragraph 4.3.7 is achieved.
SOC - F2_Bat1 - P PV,max = 8kW
100
SOC [%]
80
60
40
20
0
0
24
48
72
96
120
144
168
time [h]
Figure 5.13: State of charge of one of the household battery systems throughout the weekly simulation
60
Chapter 5: Implementation and test results
5.3.2 Q(V) control mode
The next voltage support strategy that is tested is Q(V). The corresponding control mode is
activated in the reactive power controllers and a series of simulations, under normal load
conditions, are executed for the same PV-integration levels as with the PF(P) strategy (6 – 10
kW/household).
As stated by the German regulation, the droop characteristic of the Q(V) strategy should be
provided by the authorised distribution system operator. Therefore, a droop characteristic
specific for the LV network under consideration should be specified. First, it should be studied
how the selection of the parameters Vsp and Vmax can influence the behaviour of the Q(V)
strategy.
Given a voltage measurement (v), the amount of reactive power that should be absorbed
depends of the selection of parameters Vsp, Vmax as explained in Figure 5.14. For the first
parameter, a value larger than or equal to the network’s nominal voltage can be selected, thus
Vsp ≥ 1 p.u., creating a dead band which delays the reactive power consumption, if it is
unnecessary. Increasing Vsp results to lower reactive power absorption ( | | | | ) as shown in
Figure 5.14a. For the second parameter, a value lower than or equal to the limit specified by the
German guideline has to be selected, thus Vmax ≤ 1.03 p.u. Figure 5.14b shows that decreasing
Vmax results to higher reactive power consumption ( | | | | ).
Q
0
Q
Vsp Vsp’ v
Vmax
0
Vsp
v Vmax’ Vmax
V
q’
q
V
q
q’
-Qlim
-Qlim
a)
b)
Figure 5.14: Sensitivity of Q(V) characteristic in the selection of its parameters: a) for the Vsp parameter,
b) for the Vmax parameter
Parameter Vsp is set to one, thus no dead-band is used, so as not to burden the whole reactive
power load to the PV systems at the end of the feeders, where voltage would be higher in case
of reverse active power flow. When it comes to parameter Vmax, the aim is to select the highest
value possible so as to avoid a possible excessive demand of reactive power, which leads to
losses and is undesirable.
Wanting to specify the highest value of Vmax which allows for each PV-integration level the use of
the Q(V) strategy, a script is written in DPL (see Appendix C - DPL commands) and its flowchart is
presented in Figure 5.15. The script takes into account the worst case scenario, thus low load
conditions are assumed. The results are presented is Table 5.3.
5.3 Reactive power control strategies
61
Figure 5.15: Flowchart of the script executed to calculate the highest value of V max which allows for each
PV-integration level the use of Q(V) strategy
62
Chapter 5: Implementation and test results
Table 5.3: PV hosting capacity with respect to Vmax parameter
Vmax
[p.u.]
1.030
1.029
1.028
1.027
1.026
1.025
1.024
1.023
1.022
1.021
1.020
1.019
1.018
1.017
1.016
PV hosting capacity
[kW/household]
8.6
8.7
8.8
8.8
8.9
9.0
9.1
9.2
9.3
9.4
9.5
9.5
9.6
9.6
9.6
Vmax
[p.u.]
1.015
1.014
1.013
1.012
1.011
1.010
1.009
1.008
1.007
1.006
1.005
1.004
1.003
1.002
1.001
PV hosting capacity
[kW/household]
9.7
9.7
9.7
9.8
9.8
9.9
9.9
9.9
9.9
10.0
10.0
10.0
10.1
10.1
10.1
The results of Table 5.3 are also verified through simulations with low load conditions and finally
the selected values for Vmax, for each PV-integration level, are specified (Table 5.4). It can be
observed that for a PV-integration level of 10 kW/household, a very steep Q(V) characteristic is
required, which is logical as the network reaches close to its maximum possible hosting capacity
for the specified power factor limits of PV systems connected to it (see paragraph 5.3).
Table 5.4: Selection of Vmax parameter for different PV-integration levels
PV-integration level
[kW/household]
6
7
8
9
10
Vmax
[p.u.]
1.030
1.030
1.030
1.023
1.003
5.3.2.1 Q(V) without storage
The target to maintain the voltage at every terminal below 1.03 p.u., using the Q(V) voltage
support strategy, is achieved as it is shown in Figure 5.16, which plots the voltage of the weakest
terminal for 6 and 10 kW/household PV-integration levels.
A characteristic of the Q(V) strategy is that not all PV systems contribute the same to the
reactive power consumption that is required to keep the voltage under the specified limit. That
happens because, in case of reverse power flow, the voltage at connection points closer to the
MV/LV transformer is lower than the voltage at the end of the feeder. Thus, PV systems at the
end of the feeder may consume the maximum of their reactive power capability while PV
systems closer to the transformer may consume a very small amount of reactive power. This fact
is presented in Figure 5.17. It is clear that due to lower voltage magnitude at terminal 2 than at
5.3 Reactive power control strategies
63
terminal 4 of feeder 1, the reactive power consumption of the PV system of terminal 2 is also
lower than that of terminal 4.
Voltage - F1_T5 - Q(V) without storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.16: Voltage at the weakest terminal in case of Q(V) strategy without storage
P and Q - F1_T2 vs F1_T4 - Monday
Voltage - F1_T2 vs F1_T4 - Monday
10
1.06
Feeder 1 - Terminal 2
Feeder 1 - Terminal 4
1.05
power [kW - kVAr]
voltage [p.u.]
1.04
1.03
1.02
1.01
1
6
4
2
0
-2
0.99
0.98
P
Q, F1_T2
Q, F1_T4
8
0
2
4
6
8
10 12 14
time [h]
16
a)
18
20
22
24
-4
0
2
4
6
8
10
12 14
time [h]
16
18
20
22
24
b)
Figure 5.17: Comparison of the behaviour of Q(V) strategy at different terminals: a) daily voltage
measurements, b) active and reactive power injection
5.3.2.2 Q(V) with battery storage
The same series of simulations are executed activating the battery storage option. The results
concerning the voltage at the weakest terminal are presented in Figure 5.18, for 6 and 10
kW/household PV-integration levels.
64
Chapter 5: Implementation and test results
Voltage - F1_T5 - Q(V) with battery storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.18: Voltage at the weakest terminal in case of Q(V) strategy with battery storage
5.3.3 Comparison of reactive power based voltage support strategies
Both reactive power based voltage support strategies, with and without battery storage, succeed
in mitigating voltage rise problem up to a PV-integration level of 10 kW/household. However,
when increasing the PV integration level of the test LV network, a question regarding grid losses
arises:
“How grid losses are influenced by the increase of PV-integration level?”
Figure 5.19 presents grid losses during the weekly simulation, which consist of the transformer
and line losses, for PV-integration levels from 6 - 10 kW/household, using PF(P) and Q(V) voltage
support strategies, with and without battery storage. In the same figure, grid losses are plotted
for the case that no voltage support is provided. It is shown in paragraph 5.3 that up to a PVintegration level of 5 kW/household no voltage support is actually required. However,
comparing the resulting curves, for PV-integration levels greater than 6 kW/household,
differences in grid losses are observed.
PF(P) and Q(V) strategies without storage exhibit greater grid losses than the case no voltage
support strategy is activated. This difference is due to the increased reactive power flow in the
network caused by the reactive power control of PV systems in order to mitigate voltage rise. It
is observed that for a range of 6 – 9 kW/household, Q(V) strategy leads to lower grid losses than
PF(P) strategy. This is reversed for the PV-integration level of 10 kW/household, when Q(V)
strategy shows higher losses mainly due to its steep characteristic curve that requires high
reactive power absorption from all PV systems. Although the same tendency appears between
PF(P) and Q(V), when battery storage is also enabled, grid losses are lower than the case no
voltage support strategy is activated because of self-consumption that lowers the active power
flow in the network.
5.3 Reactive power control strategies
65
Grid Losses - PF(P) vs Q(V) with and without battery storage
140
PF(P)
Q(V)
PF(P) with battery
Q(V) with battery
without voltage support
grid losses [kWh]
120
100
80
60
40
0
1
2
3
4
5
6
7
PV integration level [kW/household]
8
9
10
Figure 5.19: Weekly grid losses comparison for the reactive power control strategies
However, apart from grid losses, significant battery losses also appear in case battery storage is
used (Figure 5.20). They are independent of the reactive power strategy used as the
charging/discharging control algorithm depends only on the local active power generation and
consumption (see paragraph 4.3.7). It seems that the reduction of grid losses, in case battery
storage is used, is by far less than the increase in battery losses, thus the LV network seems to be
less efficient. Nevertheless, more information would be required regarding losses caused by the
reactive power flow in other voltage levels of the transmission and distribution system.
Battery Losses
110
battery losses [kWh]
105
100
95
90
85
80
6
7
8
9
PV integration level [kW/household]
10
Figure 5.20: Weekly losses of all the battery systems
The reactive power consumed additionally enlarges the loading of the grid equipment and
demands measures for compensation. The more reactive power is consumed by distributed
generators, such as PV systems, the higher the need from transmission system operators (TSO)
to provide the lack of reactive power in their grids. Therefore, an effective concept of reactive
66
Chapter 5: Implementation and test results
power consumption is needed in order to increase the hosting capacity of distribution networks
without causing additional measures in other places or network levels.
Figure 5.21 shows the trend of the maximum reactive power demand from the external grid,
using PF(P) and Q(V) strategies, with and without storage, as the PV-integration level increases.
Q(V) strategies exhibits better results than PF(P) ones especially in the range of 6 – 9
kW/household. It should be noted that the increase in maximum reactive power demand from
the external grid throughout the simulations reached up to around 700 % comparing with the
case that no voltage support is considered. This fact may lead to the requirement of
supplemental reactive power compensation systems, which means additional investments from
the power system operators.
Maximum reactive power demand from the external grid
70
PF(P)
Q(V)
PF(P) with battery
Q(V) with battery
without voltage support
reactive power [kVAr]
60
50
40
30
20
10
0
0
1
2
3
4
5
6
7
PV integration level [kW/household]
8
9
10
Figure 5.21: Maximum reactive power demand from the external grid
Besides the increase in hosting capacity, and eventually in PV energy yield, through the use of
these reactive power based voltage support strategies, it is also important how much reactive
power in time has to be provided to reach a certain PV-integration level. A way to evaluate this
criterion can be through a quality index defined as follows [34]:
∫
∫
(5.3)
The numerator is the total PV energy yield of all PV systems of the network throughout the
simulation and the denominator is the corresponding apparent “energy”:
∑
(5.4)
5.3 Reactive power control strategies
67
∑√
(5.5)
An example of the difference in reactive power consumption by a PV system connected in the
network, between the PF(P) and Q(V) strategies is shown in Figure 5.22. The apparent power,
whose integral is used in the denominator of the quality index, is also plotted in the same figure.
It can be observed that, for this PV-integration level, PF(P) strategy consumes more reactive
power than Q(V) one, while both succeed in keeping the voltage below the permitted limit.
Apart from the increased losses caused to the grid, costs for the compensation of reactive power
may also be relevant, if the DNO has to compensate its reactive power flows. Furthermore, the
more reactive power is consumed by a PV inverter the more it is loaded thus its lifetime can be
negatively influenced [58].
P,Q and S - F1_PVS4 - P max = 8 kW - Sunday
10
P
Q, PF(P) strategy
Q, Q(V) strategy
S, PF(P) strategy
S, Q(V) strategy
power [kW - kVAr - KVA]
8
6
4
2
0
-2
-4
0
2
4
6
8
10
12
14
time [h]
16
18
20
22
24
Figure 5.22: Difference in reactive power requirement between PF(P) and Q(V) strategies
Figure 5.23 plots the quality index for the PV-integration range of 6 – 10 kW/household. Because
of the dependence of reactive power consumption only on active power production, the quality
index is the same and constant in case of PF(P) strategies. In contrast, the reduction of the
quality index in case of Q(V) strategies reveals that the ratio of reactive power demand over time
per active power generation over time gets higher as PV-integration level and eventually energy
yield increases.
68
Chapter 5: Implementation and test results
Quality index
1.02
PF(P)
Q(V)
PF(P) with battery
Q(V) with battery
1.01
quality index
1
0.99
0.98
0.97
0.96
0.95
0.94
6
7
8
9
PV integration level [kW/household]
10
Figure 5.23: Quality index comparison among the reactive power control strategies
Q(V) strategies present higher quality indices than PF(P) ones for a range of 6 – 9 kW/household,
although this is reversed for the PV-integration level of 10 kW/household. Moreover, the quality
index is higher for the Q(V) strategy with battery storage than for the same strategy without
storage as at times battery systems are in charging state, voltage at their terminals is lower and
thus also lower reactive power absorption is demanded by the reactive power controllers.
5.4 Dynamic active power curtailment strategies
DAPC control mode, as presented in paragraph 4.3.5, is activated in the active power controller
of the PV-Battery systems, while the reactive power controller is set to unity fixed power factor
(no reactive power is absorbed by PV systems). The effectiveness and behaviour of DAPC voltage
support strategy is then tested through, a series of simulations, under normal load conditions,
for PV integration levels from 6 - 10 kW/household, with a step of 1 kW/household.
The DPL script, whose flowchart is presented in Figure 5.6, is executed again, but now only under
unity power factor and no load conditions, in order to estimate the highest possible value for
Pcurt,set parameter, as defined in Table 4.9. This parameter is selected to match the maximum net
generation per household over which voltage rises above the limit at the weakest terminal of the
network. The estimated value for the maximum net generation of the studied network is 5
kW/household.
5.4.1 DAPC without storage
The target to maintain the voltage at every terminal below 1.03 p.u., using the DAPC voltage
support strategy, is achieved as it is shown in Figure 5.24, which plots the voltage of the weakest
terminal for 6 and 10 kW/household PV integration levels.
5.4 Dynamic active power curtailment strategies
69
Voltage - F1_T5 - DAPC without storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.24: Voltage at the weakest terminal in case of DAPC strategy without storage
An example of the operation of the DAPC control mode is presented in Figure 5.25. The dynamic
limitation curve is the “load demand” curve offset by the net generation limit. In this way, the
area between the “ideal PV generation” and “actual PV generation” curves corresponds to the
PV energy loss due to the power curtailment.
DAPC without storage - F1_T1 - P PV,max = 8kW - Monday
12
load demand
net generation limit
dynamic limitation
ideal PV generation
actual PV generation
10
power [kW]
8
6
4
2
0
0
2
4
6
8
10
12
14
time [h]
16
18
20
22
24
Figure 5.25: Operation of DAPC strategy without storage for one of the PV systems connected in the
network
5.4.2 DAPC with battery storage
Activating the battery storage option the same series of simulations are executed again. The
results concerning the voltage at the weakest terminal are presented in Figure 5.26, for 6 and 10
kW/household PV integration levels. As it turns out, the target to limit the voltage below 1.03
p.u. is achieved.
70
Chapter 5: Implementation and test results
Voltage - F1_T5 - DAPC with battery storage
1.06
6 kW/household
10 kW/household
1.05
voltage [p.u.]
1.04
1.03
1.02
1.01
1
0.99
0.98
0
24
48
72
96
120
144
168
time [h]
Figure 5.26: Voltage at the weakest terminal in case of DAPC strategy with battery storage
An example of the operation of the DAPC control mode with battery storage is presented in
Figure 5.27. In this case, the dynamic limitation curve is the “load demand” curve plus the
opposite of the charging part of the “battery power” curve offset by the net generation limit.
The area between the “ideal PV generation” and “actual PV generation” curves corresponds to
the PV energy loss due to the power curtailment and as it can be observed, it is smaller than that
of Figure 5.25.
DAPC with battery storage - F1_T1 - P PV,max = 8kW - Monday
14
12
10
8
power [kW]
6
4
2
0
load demand
battery power
net generation limit
dynamic limitation
ideal PV generation
actual PV generation
-2
-4
-6
-8
0
2
4
6
8
10
12
time [h]
14
16
18
20
22
24
Figure 5.27: Operation of DAPC strategy with battery storage for one of the PV systems connected in the
network
5.4.3 Comparison of the DAPC voltage support strategies
Both DAPC voltage support strategies, with and without battery storage, succeed in mitigating
voltage rise problem. However, as in the case of reactive power based strategies, they are
compared to each other based on a set of selected criteria.
5.4 Dynamic active power curtailment strategies
71
The first comparison is related to grid losses and the results are presented in Figure 5.28. Both
alternatives exhibit grid losses less than the case without voltage support due to the curtailed
active power which would cause losses by flowing in the grid. Because of the self-consumption
scheme, when battery storage is activated, grid losses are even lower than the case without
storage. On the other hand, battery losses appear as a result of batteries’ energy exchange (see
Figure 5.20).
Grid Losses - DAPC with and without battery storage
140
DAPC
DAPC with battery
without voltage support
grid losses [kWh]
120
100
80
60
40
0
1
2
3
4
5
6
7
PV integration level [kW/household]
8
9
10
Figure 5.28: Weekly grid losses comparison for the DAPC strategies
As a result of the active power curtailment, a part of the possible PV energy yield, throughout
the study period, is consequently curtailed. Yield loss, which is defined in equation (5.6), is
plotted for both DAPC alternatives, in Figure 5.29.
(5.6)
It is observed that as PV-integration increases, yield losses become rather significant. The
differences between the cases with and without storage lie on the availability of the batteries to
store a portion of the energy that otherwise would be curtailed.
Yield loss - DAPC with and without battery storage
10
yield loss [%]
8
6
4
2
DAPC
DAPC with battery
0
6
7
8
9
PV integration level [kW/household]
Figure 5.29: Yield losses comparison
10
72
Chapter 5: Implementation and test results
5.5 Overall comparison
It is clear from the analysis so far that the studied voltage support strategies perform differently
in the following evaluation criteria:





grid losses
battery losses
maximum reactive power demand
reactive power performance (quality index)
yield loss
An overview of the performance of all the strategies in the criteria above is presented in Figure
5.30. A difficulty arises in the overall comparison among them as there is not a unique one which
performs the best in all the criteria.
5.5.1 Formulation of an overall evaluation criterion
When there are more than one objective that a solution method is expected to satisfy, then the
need for an overall evaluation criterion (OEC) arises [59]. In engineering an overall index is not so
common because of the following difficulties:



Units of measurement – the criteria of evaluations in engineering and science are
generally different and the same applies to their units.
Relative weighting – Not all the criteria of evaluation are of equal importance.
Sense of the quality characteristic (QC) – The quality characteristic indicates the
direction of desirability of the evaluation numbers. Depending on the criteria and how
they are measured, QC can be “bigger is better”, “smaller is better”, or “nominal is the
best”. Unless the quality characteristics of different criteria are the same, the evaluation
numbers cannot be readily combined.
In the case of the criteria set for the evaluation of the studied strategies all the above difficulties
are valid. This situation is summarised in Table 5.5.
Table 5.5: Differences of evaluation criteria
Criterion
Grid losses
Battery losses
Maximum Q demand
Q performance
Yield loss
Symbol
LG
LB
Qmax
PQ
LY
Unit
kWh
kWh
kVAr
%
Relative weighting
WLG
WLB
WQmax
WPQ
WLY
Quality characteristic
smaller is better
smaller is better
smaller is better
bigger is better
smaller is better
5.5 Overall comparison
PF(P)
Q(V)
73
DAPC
PF(P) with battery
Q(V) with battery
DAPC with battery
without voltage support
Grid Losses
Grid Losses
140
140
130
120
grid losses [kWh]
110
100
90
grid losses [kWh]
120
80
100
80
60
70
40
60
6
7
8
PV integration level [kW/household]
50
6
100
7
Maximum reactive power demand from the external grid
70
8
9
10
PV integration level [kW/household]
60
80
reactive power [kVAr]
battery losses [kWh]
10
a)
Battery Losses
40
60
40
20
0
50
40
30
20
10
6
7
8
9
PV integration level [kW/household]
0
10
6
b)
7
8
9
PV integration level [kW/household]
10
c)
Yield loss
Reactive power performance index
10
1.02
1.01
8
1
yield loss [%]
reactive power performance index
9
0.99
0.98
0.97
6
4
0.96
2
0.95
0.94
6
7
8
9
PV integration level [kW/household]
10
0
6
7
8
9
PV integration level [kW/household]
10
d)
e)
Figure 5.30: Overview of the performance of all strategies in the evaluation criteria: a) grid losses, b)
battery losses, c) maximum reactive power demand from the external grid, d) reactive power performance
index and e) yield loss
Therefore, what is needed is a properly formulated OEC number representing the overall
performance of the tested strategies. In order to combine the different criteria, they must first
be normalised and weighted accordingly. Then, for each PV-integration level i and for each
strategy j, an OEC is formulated as follows:
74
Chapter 5: Implementation and test results
(
(
(
{
{
}
}
{
}
)
{
{
{
}
}
(
}
{
}
)
{
{
}
{
}
}
{
}
{
}
{
)
}
(5.7)
)
Before all evaluation criteria can be combined, their QCs must all be the same. By choosing an
OEC number with a “bigger is better” QC, all the criteria with “smaller is better” QC are properly
converted to the desired QC by subtracting the normalising fraction from 1. The numerator in
each term is calculated by subtracting the smaller magnitude of all the strategies from the one
associated with the specific strategy whose OEC is being calculated. The denominator is the
positive difference between the best and the worst magnitude for the criteria.
An average OEC is then calculated for each strategy, taking into account the OEC values of each
PV-integration level, as follows:
∑
(5.8)
5.5.2 Relative weighting of the evaluation criteria
In order to calculate the relative weights of the evaluation criteria, eight independent experts
(DNV GL – Energy consultants, a DNO consultant, a TU Delft professor and a TU Delft PhD
candidate) were asked to determine the ranking of the various criteria. This procedure was
conducted through the use of a multicriteria-analysis software which combines a pairwise
comparison method and a “Fuzzy Logic” algorithm to determine and quantify the comparative
importance of the criteria.
The experts were asked to determine the ranking of the criteria twice as there are two main
stakeholders involved in the operation of the LV network with connected PV systems, namely
the DNO and PV system owners. The relative weights of the evaluation criteria were thus
calculated for both the perspectives of a DNO and a PV system owner (Table 5.6).
Table 5.6: Relative weights of the evaluation criteria
Criterion
Grid losses
Battery losses
Maximum Q demand
Q performance
Yield loss
Relative weights [%]
DNO
PV system owner
23.3
12.6
12.6
25.9
24.4
13.8
24.2
19.6
15.5
28.2
5.5 Overall comparison
75
5.5.3 Choosing the best strategy
As it is indicated by the weights obtained, the choice of the best strategy heavily depends on the
perspective of the involved stakeholders. Grid losses, for example, are a major issue for a DNO
while they are of minor importance for a PV system owner. The reverse is valid for battery losses
while the performance of a reactive power based strategy is of common interest.
Figure 5.31 presents the ranking of the strategies based on their average OEC value, as it is
calculated from equation (5.8). It is observed that the best voltage support strategy for a DNO is
DAPC with battery storage while for a PV system owner this strategy ranks in the second lowest
position. Q(V) strategy seems to be superior for an owner of a PV system but it ranks fourth in
the DNO’s preference. Thus a conflict of interests seems to arise.
Average OEC
100
90
PF(P)
Q(V)
DAPC
PF(P) with battery
Q(V) with battery
DAPC with battery
80
70
OEC
av
60
50
40
30
20
10
0
DNO
PV system owner
perspective
Figure 5.31: Ranking of the studied voltage support strategies
An interesting observation is how the overall performance of each strategy varies according to
the PV-integration level. Figure 5.32 and Figure 5.33 present the OEC values of the studied
strategies with respect to the PV-integration level for a DNO’s perspective and a PV system
owner’s perspective respectively.
For the former, DAPC strategy, with and without battery storage, gathers a high score
considering the whole studied PV-integration range. Q(V) strategies, while comparable with
DAPC ones at lower PV-integration levels, they present a declining trend, as the integration
increases, and finally gather the lowest scores at 10 kW/household. These are even lower than
the ones of PF(P) strategies, which are the most unfavourable of all for the range of 6-9
kW/household. For the same range, the DNO clearly favours the option of battery storage,
considering the same type of strategy.
On the other hand, a PV system owner would prefer to avoid the battery storage option,
considering the same type of strategy. Q(V) strategies, while generally the most preferable ones,
show a downward trend and finally gather the lowest scores at 10 kW/household. On the
contrary, PF(P) alternatives, while generally the least favourable, tend to be the predominant
choice at 10 kW/household. Finally, when it comes to DAPC strategies, they are in the second
place from 6 to 8 kW/household but present rather competitive at 9 and 10 kW/household.
76
Chapter 5: Implementation and test results
OEC - DNO's perspective
100
90
PF(P)
Q(V)
DAPC
PF(P) with battery
Q(V) with battery
8
PV integration level [kW/household]
9
DAPC with battery
80
70
OEC
60
50
40
30
20
10
0
6
7
10
Figure 5.32: OEC comparison with respect to the PV-integration level according to the DNO’s perspective
OEC - PV system owner's perspective
100
90
PF(P)
Q(V)
DAPC
PF(P) with battery
Q(V) with battery
DAPC with battery
80
70
OEC
60
50
40
30
20
10
0
6
7
8
PV integration level [kW/household]
9
10
Figure 5.33: OEC comparison with respect to the PV-integration level according to the PV system owner’s
perspective
77
Chapter 6
6 Conclusions and future work
In this chapter the conclusions extracted from this study as well as some recommendations for
future work are presented.
6.1 Conclusions
6.1.1 Main conclusions
The test and comparison of the studied voltage support strategies led to the following main
conclusions:




All studied strategies manage to mitigate the voltage rise problem, in the test network,
for the range of 6-10 kW/household PV integration, achieving 86% more PV hosting
capacity than the case without voltage support.
The comparison of all the three type of strategies studied, with and without the storage
option, and the selection of the best candidate proves not to be an easy procedure. Each
strategy may be more appropriate regarding a specific criterion for a specific PVintegration level but may be disadvantageous considering another criterion and another
PV-integration level. The viewpoints of the main stakeholders involved (DNO, PV system
owner) are also different regarding the relative importance of each criterion.
Through the help of energy experts from various professional positions, the overall
evaluation criterion, which is introduced in the form of a score number, reveals the
overall preference of a DNO for DAPC strategy with battery storage in contrast to the
overall preference of a PV system owner for Q(V) strategy without storage.
The preference of each stakeholder varies with respect to the PV-integration level.
Generally, a DNO seems to favour active power curtailment and local battery storage
while a PV system owner would prefer a reactive power based strategy without storage.
6.1.2 Specific findings
Integration of PV systems in LV networks:

Power systems face a transition from centralized to distributed generation and
specifically renewable generation, such as PV, due to electricity market opening,
environmental awareness and increasing electricity demand. Most PV systems are
connected to the LV distribution network, especially in rural areas where vast available
space is offered on rooftops of houses and farms. However, rural networks are generally
rather weak (low level of interconnection, long distances, low load density, low short
circuit power) and thus more prone to voltage rise situations when the power consumed
is lower than the power produced.
78
Chapter 6: Conclusions and future work


Grid codes and regulations set the limits to voltage variations in order to secure power
quality and define the requirements for the connection of DG in the LV network. The
possible violation of the upper voltage limit, in cases of low consumption and high
generation, places an obstacle to the integration of more RES in the power systems and
consequently hinders the global target to achieve a sustainable future energy supply.
Load flow calculations, on a suitably designed rural LV network model, show that for
conductors (OHL and cables) with the same ampacity the problem is more intense in the
case OHLs are used instead of cables. Conductor and transformer loadings do not
initially constitute a problem in further PV integration.
Voltage sensitivity of a LV network on active and reactive power:



The implementation of local voltage support strategies, based on reactive power
absorption by PV inverters and active power curtailment, are suggested by the new
connection standards.
Voltage sensitivity analysis on active and reactive power on the test network’s terminals
shows that, comparing the two types of conductors, both sensitivities present higher
values in the case of OHLs than in the case of cables. Thus both active and reactive
power based voltage support strategies are more effective in mitigating the voltage rise
problem when OHLs are used.
For both conductor types voltage sensitivity in active power is lower than the one in
reactive power for terminals closer to the transformer as short-circuit reactance
becomes dominant over the short-circuit resistance at these locations. As the distance to
the transformer increases along the feeder, line resistance contributes more on the
impedance and as a result voltage sensitivity in active power is higher than the one in
reactive power for the terminals which are deeper in the feeder. Hence, reactive power
based voltage support strategies are more effective in case of PV systems connected
close to the transformer while for PV systems connected deeper in a LV feeder active
power based strategies seem more suitable.
Reactive power based voltage support strategies and battery storage:




By allowing PV inverters to absorb reactive power a significant increase in the network’s
PV hosting capacity can be achieved while keeping the voltage level under the specified
upper limit. For the test rural LV network this increase reached up to 86 %. Reactive
power absorption can relocate the limiting criterion of hosting capacity from voltage
towards transformer or conductor loading limitation from a certain hosting capacity
upwards, as it was shown in the case of cable usage on the test network.
The parameterisation of PF(P) characteristic needs attention as it may lead to violation
of upper voltage limit in low irradiance conditions while being effective in higher
irradiance, if it is not carefully designed.
Using PF(P) strategy, inverters will absorb reactive power regardless of their location in
the feeder. As a result inverters might absorb reactive power even though it may not be
required (no significant voltage rise situation).
Q(V) characteristic also requires attention in its parameterisation so as not to lead to
excessive unnecessary reactive power demand.
6.2 Recommendations for future work



79
Using Q(V) strategy, not all PV systems contribute the same to the reactive power
consumption that is required to keep the voltage under the specified limit. That happens
because, in case of reverse power flow, the voltage at connection points closer to the
MV/LV transformer is lower than the voltage at the end of the feeder. Thus, PV systems
at the end of the feeder may consume the maximum of their reactive power capability
while PV systems closer to the transformer may consume a very small amount of
reactive power.
The performance of Q(V) strategy in all criteria (grid losses, maximum reactive power
demand from the external grid, quality index) appears better than that of PF(P) one up
to a certain PV integration level but gets worst in most criteria as the network reaches
close to its maximum possible hosting capacity for the specified power factor limits of PV
systems connected to it.
The presence of battery storage in each PV system connected to the network improves
self-consumption and as a result grid losses and reactive power demand decrease.
However, battery losses are introduced.
Dynamic active power curtailment strategies and battery storage:


DAPC strategy succeeds in mitigating the voltage rise problem at the cost of yield loss which
becomes rather significant as the PV-integration level increases. Grid losses appear less
than the case without voltage support due to the curtailed active power which would cause
losses by flowing in the grid.
Because of the self-consumption scheme, when battery storage is activated, grid losses are
even lower than the case without storage. On the other hand, battery losses appear as a
result of batteries’ energy exchange.
6.2 Recommendations for future work
Possible tasks suggested for future work can be:





Validation of the results on Watt connects interactive table in order to incorporate the
(unpredictable) customer’s behaviour.
Substitution of the designed LV test network with actual ones and comparison of the
results in order to investigate the range of the possible differences.
Prolongation of the time period of study to a full year to include the seasonal variation
of PV generation and its impact on the evaluation criteria.
Economic evaluation of the studied voltage support strategies, maybe the most crucial
deciding factor for the selection of the best candidate strategy.
Implementation and test of other voltage support solutions like OLTC transformers and
storage at the distribution substation level or other strategically defined location.
81
Appendix A - OHL modelling
The physical geometry of overhead lines for a typical European LV network is shown in Figure
A.1 and the specifications of the model used are given in Table A.1 [48, 60].
Figure A.1: Geometry of overhead lines for a typical European LV network [48]
Table A.1: Characteristics of the OHLs used in the model [48, 60]
Material
Size
[mm2]
Al
70
Outer
diameter
[mm]
10.5
GMR
[mm]
DC-Resistance
[Ω/km]
Ampacity
[A]
a
[m]
b
[m]
3.98
0.4367
270
8
0.3
In PowerFactory’s environment, in order to model the selected OHLs, the “TypTow” tower
geometry type is used and the electrical parameters are provided by the phase impedance
matrix. As neutral wires are not included in this type the phase impedance matrix after Kron
reduction is used (Table A.2) [48].
Table A.2: Phase impedance matrix after Kron reduction (Ω/km) [48]
Phase
A
B
C
A
0.616 + j0.588
0.131 + j0.306
0.141 + j0.245
B
0.131 + j0.306
0.628 + j0.566
0.147 + j0.276
C
0.141 + j0.245
0.147 + j0.276
0.650 + j0.527
Regarding the option of cable usage, the already available type, in PowerFactory’s library, NA2XY
3x120 mm2, is chosen in order to match the ampacity of the conductor used as OHL, for loading
comparison purposes throughout the simulations.
83
Appendix B - DSL models’ code
“Load Data Process” block definition:
model Pext,Qext = 'BlkDef Load Data
Process'(P,Q;;scale,PF,ExtCtrl_Q,base_load,P_base,Q_base;)
Pext=scale*select(base_load, P_base, P)
Qext=scale*select(base_load, Q_base, select(ExtCtrl_Q, Q,
P*tan(acos(PF))))
“PV generation profile” block definition:
model P_av = 'BlkDef PV generation profile'(I;;PF_lim;)
inc(P_av)=0
inc(I)=0
P_av=lim((I/1000)*PF_lim,0,PF_lim) ! available active power in p.u.
“Active Power Control” block definition:
model id_ref,P = 'BlkDef Active Power
Control'(P_av,P_load_A,P_load_B,P_bat,u;x;S_nom,DAPC,P_curt_set,T;P_load
,P_curt_lim,P_lim,P_bat_charge)
inc(P)=0
inc(x)=0
P_load=P_load_A+P_load_B
P_bat_charge=select(P_bat>0,P_bat,0)
P_curt_lim=(P_curt_set+P_load+P_bat_charge)/S_nom
P_lim=select(DAPC<0.5,P_av,P_curt_lim)
P=lim(P_av,0,P_lim)
limits(T)=(0,)
x.=(P-x)/T
id_ref=x/u
“Reactive Power Control” block definition:
model iq_ref = 'BlkDef Reactive Power
Control'(u,P;x;mode_Q,PF_fixed,PF_lim,u_max,u_sp,P_sp,T;Q,Q_0,Q_1,Q_2,Q_
lim,PF_P)
inc(Q)=0
inc(x)=0
Q_0=select(P<0.2*PF_lim,0,-P*tan(acos(PF_fixed)))
PF_P=lim(((PF_lim-1)*(P-P_sp*PF_lim))/(PF_lim*(1-P_sp))+1,PF_lim,1)
Q_1=-P*tan(acos(PF_P))
Q_lim=-P*tan(acos(PF_lim))
Q_2=lim((u-u_sp)/(u_max-u_sp)*Q_lim, Q_lim,0)
Q=select(mode_Q<0.5,Q_0,select(mode_Q<1.5,Q_1,Q_2))
limits(T)=(0,)
x.=(Q-x)/T
iq_ref=-x/u
84
Appendix B - DSL models’ code
“Battery Control” block definition:
model P_bat,id_ref,iq_ref = 'BlkDef Battery
Control'(u,P_av,P_load_A,P_load_B;x1,x2;EnableStorage,S_PV_nom,C,S_bat_n
om,T;P_load,P_bat_av,SOC,eff)
inc(x1)=20
inc(x2)=0
inc(P_bat)=0
P_load=P_load_A+P_load_B
P_bat_av=lim(P_av*S_PV_nom-P_load,-C,C)
eff=select(P_bat_av>0,0.95,1/0.95)
x1.=eff*P_bat_av/3600/C*100
SOC=select(EnableStorage<0.5,20,limstate(x1,20,80))
P_bat=select(EnableStorage<0.5,0,select(SOC>20.and.SOC<80,P_bat_av,0))
x2.=(P_bat/S_bat_nom-x2)/T
id_ref=-x2/u
iq_ref=0
85
Appendix C - DPL commands
Hosting capacity depending on the PF settings:
double P_set,P_max, v_max, line_maxload, trafo_maxload, a, b, PF_f,PF_h,
volt_F1, volt_F2, load_F1, load_F2, load_trafo;
int nso, err, j, check_volt_lim, check_load_lim;
set s;
object o;
string str1;
j=MyRes.Clear();
v_max=1.03;
line_maxload=100;
trafo_maxload=100;
s=PV_Set.All();
nso=s.Count();
!
!
!
!
voltage limit [p.u.]
maximum line loading [%]
maximum transformer loading [%]
set of PV static generators
PF_f=1.00;
! initial PF for
PF_h=1.00;
! initial PF for
P_set=0.005;
! initial active
PV System
while (PF_f>0.89)
! lower limit of
{
P_max=P_set;
do
{
o=s.First();
for (o=s.First(); o; o=s.Next())
{
str1=o:loc_name;
a=strstr(str1, 'F1_PV3');
b=strstr(str1, 'F1_PV5');
if (a+b>=-1)
{
o:pgini=2.7*P_max;
! for a
production is 2.7 times higher
o:cosgini=PF_f;
}
else
{
o:pgini=P_max;
o:cosgini=PF_h;
}
o:pf_recap=1;
o:iv_mode=0;
}
ResetCalculation();
err=Ldf.Execute();
if (err)
{
output('Load Flow Problem');
exit();
}
else
{
farm PV systems
household PV systems
power production of a household
PF for the while loop
farm PV system active power
86
Appendix C - DPL commands
volt_F1=F1_Bus:m:u1;
volt_F2=F2_Bus:m:u1;
load_F1=F1_Line:c:loading;
load_F2=F2_Line:c:loading;
load_trafo=Trafo:c:loading;
check_volt_lim=volt_F1>=v_max.or.volt_F2>=v_max;
check_load_lim=load_F1>=100.or.load_F2>=100.or.load_trafo>=100;
if (check_volt_lim.and..not.check_load_lim)
{
P_set=P_max;
}
else
{
if (check_load_lim)
{
P_set=P_max-0.001000;
}
else
{
P_max+=0.000100;
}
}
}
}
while
(volt_F1<v_max.and.volt_F2<v_max.and.load_F1<100.and.load_F2<100.and.loa
d_trafo<100)
P=P_max*1000-0.1;
pf=PF_f;
MyRes.Write();
PF_f-=0.01;
PF_h-=0.005;
}
MyRes.Flush();
j=MyRes.Draw();
Export:pResult=MyRes;
Export:iopt_exp=6;
Export:f_name='C:\Thesis\Simulations\PF_Pmax.csv';
Export.Execute();
Calculation of Vmax for Q(V) strategy:
double P_set,P_max, v1, v2, line_maxload, trafo_maxload, a, b,
volt_F1, volt_F2, load_F1, load_F2, load_trafo,
S_nom_h, S_nom_f, Q_lim_h, Q_lim_f,q_max_h,q_max_f;
int nso, err, j, check_volt_lim, check_load_lim;
set s;
object o;
string str1;
ResetCalculation();
j=MyRes.Clear();
line_maxload=100;
trafo_maxload=100;
s=PV_Set.All();
nso=s.Count();
! maximum line loading [%]
! maximum transformer loading [%]
! set of PV static generators
q_max_h=PF_lim_h*tan(acos(PF_lim_h));
q_max_f=PF_lim_f*tan(acos(PF_lim_f));
Appendix C - DPL commands
v1=1.0000;
characteristic
v2=1.0300;
characteristic
P_set=0.0050;
household PV System
87
! define v_sp (=v1) parameter of Q(V)
! initial v_max (=v2) parameter of Q(V)
! initial active power production of a
while (v2>v1)
{
P_max=P_set;
S_nom_h=P_max/PF_lim_h;
S_nom_f=2.7*P_max/PF_lim_f;
Q_lim_h=-P_max*tan(acos(PF_lim_h));
Q_lim_f=-2.7*P_max*tan(acos(PF_lim_f));
do
{
o=s.First();
for (o=s.First(); o; o=s.Next())
{
str1=o:loc_name;
a=strstr(str1, 'F1_PV3');
b=strstr(str1, 'F1_PV5');
if (a+b>=-1)
{
o:sgn=S_nom_f;
o:pgini=2.7*P_max;
! for a farm PV system active power
production is 2.7 times higher
o:Pmax_uc=S_nom_f;
o:P_max=S_nom_f;
o:cosgini=1;
o:pf_recap=1;
o:iv_mode=2;
o:ddroop=((v1-v2)*S_nom_f*100)/Q_lim_f;
o:usetp=v1;
o:q_min=-q_max_f;
o:q_max=q_max_f;
}
else
{
o:sgn=S_nom_h;
o:pgini=P_max;
o:Pmax_uc=S_nom_h;
o:P_max=S_nom_h;
o:cosgini=1;
o:pf_recap=1;
o:iv_mode=2;
o:ddroop=((v1-v2)*S_nom_h*100)/Q_lim_h;
o:usetp=v1;
o:q_min=-q_max_h;
o:q_max=q_max_h;
}
}
ResetCalculation();
err=Ldf.Execute();
if (err)
{
output('Load Flow Problem');
exit();
}
else
{
88
Appendix C - DPL commands
volt_F1=F1_Bus:m:u1;
volt_F2=F2_Bus:m:u1;
load_F1=F1_Line:c:loading;
load_F2=F2_Line:c:loading;
load_trafo=Trafo:c:loading;
check_volt_lim=volt_F1>=v_max.or.volt_F2>=v_max;
check_load_lim=load_F1>=100.or.load_F2>=100.or.load_trafo>=100;
if (check_volt_lim.or.check_load_lim)
{
P_set=P_max;
}
else
{
P_max+=0.0001;
S_nom_h=P_max/PF_lim_h;
S_nom_f=2.7*P_max/PF_lim_f;
Q_lim_h=-P_max*tan(acos(PF_lim_h));
Q_lim_f=-2.7*P_max*tan(acos(PF_lim_f));
}
}
}
while
(volt_F1<v_max.and.volt_F2<v_max.and.load_F1<100.and.load_F2<100.and.loa
d_trafo<100)
Pmax=P_max*1000-0.1;
v_2=v2;
MyRes.Write();
v2-=0.0010;
}
MyRes.Flush();
j=MyRes.Draw();
Export:pResult=MyRes;
Export:iopt_exp=6;
Export:f_name='C:\Thesis\Simulations\Q_V_Hosting_Capacity.csv';
Export.Execute();
89
Bibliography
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
A. von Meier, Electric power systems: a conceptual introduction. Hoboken, New Jersey:
Wiley, 2006.
N. Jenkins, G. Strbac, and J. B. Ekanayake, Distributed generation. London, United
Kingdom: The Institution of Engineering and Technology (IET), 2010.
A.-M. Borbely and J. F. Kreider, Distributed generation: the power paradigm for the new
millennium: CRC Press LLC, 2001.
P. Kundur, Power system stability and control: McGraw-Hill, Inc., 1994.
J. D. Bouford and C. A. Warren, "Many states of distribution," Power and Energy
Magazine, IEEE, vol. 5, pp. 24-32, 2007.
M. H. Bollen and F. Hassan, Integration of distributed generation in the power system:
Wiley-IEEE Press, 2011.
M. Zeman, "Photovoltaic systems," in Solar cells, ed.
B. M. Wilamowski and J. D. Irwin, The industrial electronics handbook, Power Electronics
and Motor Drives, 2nd ed.: CRC Press, 2011.
A. Malla and A. Niraula, "Importance of balance of system in a solar PV application."
EPIA and Greenpeace, "Solar generation: solar photovoltaic electricity empowering the
world," 2011.
EPIA, "Connecting the Sun," 2012.
EPIA, "Global market outlook for photovoltaics 2013-2017," 2013.
EPIA, "Market report 2013," 2013.
J. Widén, E. Wäckelgård, J. Paatero, and P. Lund, "Impacts of distributed photovoltaics
on network voltages: Stochastic simulations of three Swedish low-voltage distribution
grids," Electric Power Systems Research, vol. 80, pp. 1562-1571, 2010.
J. von Appen, M. Braun, T. Stetz, K. Diwold, and D. Geibel, "Time in the sun: the
challenge of high PV penetration in the German electric grid," Power and Energy
Magazine, IEEE, vol. 11, pp. 55-64, 2013.
(30/04/2014). Watt connects. Available: http://www.wattconnects.com/
T. Degner, G. Arnold, M. Braun, D. Geibel, W. Heckmann, and R. Brundlinger, "Utility
scale PV systems: grid connection requirements, test procedures and European
harmonisation," Photovoltaics International, 2009.
"EN 50160 Voltage characteristics of electricity supplied by public electricity networks,"
ed, 2010.
"EN-IEC 61000-3-3 Electromagnetic compatibility (EMC) - Part 3-3: Limits - Limitation of
voltage changes, voltage fluctuations and flicker in public low-voltage supply systems,
for equipment with rated current ≤ 16 A per phase and not subject to conditional
connection," ed, 2013.
"EN 50438 Requirements for micro-generating plants to be connected in parallel with
public low-voltage distribution networks," ed, 2013.
B. I. Craciun, T. Kerekes, D. Sera, and R. Teodorescu, "Overview of recent Grid Codes for
PV power integration," in Optimization of Electrical and Electronic Equipment (OPTIM),
2012 13th International Conference on, 2012, pp. 959-965.
SMA, "PV Grid Integration," 2012.
M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, "Distributed energy storage for mitigation
of voltage-rise impact caused by rooftop solar PV," in Power and Energy Society General
Meeting, 2012 IEEE, 2012, pp. 1-8.
T. A. Short, Electric Power Distribution Handbook: CRD Press, 2004.
90
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
Bibliography
D. Geibel, T. Degner, T. Reimann, B. Engel, T. Bulo, J. P. Da Costa, et al., "Active
intelligent distribution networks — Coordinated voltage regulation methods for
networks with high share of decentralised generation," in Integration of Renewables into
the Distribution Grid, CIRED 2012 Workshop, 2012, pp. 1-4.
T. Degner, G. Arnold, T. Reimann, P. Strauß, B. Engel, and M. Breede, "Increasing the
photovoltaic-system hosting capacity of low voltage distribution networks," presented at
the 21st International Conference and Exhibition on Electricity Distribution (CIRED),
Frankfurt, 2011.
T. Stetz, F. Marten, and M. Braun, "Improved Low Voltage Grid-Integration of
Photovoltaic Systems in Germany," Sustainable Energy, IEEE Transactions on, vol. 4, pp.
534-542, 2013.
C. Korner, M. Hennig, K. Handt, and R. Schmid, "Gaining experience with a regulated
distribution transformer in a smart grid environment," in Integration of Renewables into
the Distribution Grid, CIRED 2012 Workshop, 2012, pp. 1-4.
A. Constantin, R. D. Lazar, and S. B. Kjær, "Voltage control in low voltage networks,"
Danfoss Solar Inverters A/S2012.
B. Bletterie, A. Gorsek, B. Uljanic, B. Blazic, A. Woyte, T. Vu Van, et al., "Enhancement of
the Network Hosting Capacity – Clearing Space for/with PV," presented at the 25th EU
PVSEC, Valencia, Spain, 2010.
E. Demirok, D. Sera, R. Teodorescu, P. Rodriguez, and U. Borup, "Evaluation of the
voltage support strategies for the low voltage grid connected PV generators," in Energy
Conversion Congress and Exposition (ECCE), 2010 IEEE, 2010, pp. 710-717.
G. Kerber, R. Witzmann, and H. Sappl, "Voltage limitation by autonomous reactive power
control of grid connected photovoltaic inverters," in Compatibility and Power Electronics,
2009. CPE '09., 2009, pp. 129-133.
E. Demirok, P. Casado Gonzalez, K. H. B. Frederiksen, D. Sera, P. Rodriguez, and R.
Teodorescu, "Local Reactive Power Control Methods for Overvoltage Prevention of
Distributed Solar Inverters in Low-Voltage Grids," Photovoltaics, IEEE Journal of, vol. 1,
pp. 174-182, 2011.
D. Mende, Y. T. Fawzy, D. Premm, and S. Stevens, "Increasing the hosting capacity of
distribution networks for distributed generation using reactive power control Potentials and Limits," presented at the 2nd Solar Integration Workshop, Lisbon,
Portugal, 2012.
R. Tonkoski, L. A. C. Lopes, and T. H. M. El-Fouly, "Coordinated Active Power Curtailment
of Grid Connected PV Inverters for Overvoltage Prevention," Sustainable Energy, IEEE
Transactions on, vol. 2, pp. 139-147, 2011.
C. Gaudin, A. Ballanti, and E. Lejay, "Evaluation of PV curtailment option to optimize PV
integration in distribution network," in Integration of Renewables into the Distribution
Grid, CIRED 2012 Workshop, 2012, pp. 1-4.
EPIA, "Self consumption of PV electricity," ed, 2013.
SMA, "SMA smart home: The system solution for more independence," ed.
G. G. Pillai, G. A. Putrus, and N. M. Pearsall, "The potential of demand side management
to facilitate PV penetration," in Innovative Smart Grid Technologies - Asia (ISGT Asia),
2013 IEEE, 2013, pp. 1-5.
F. Marra, Y. T. Fawzy, T. Bulo, and B. Blazic, "Energy storage options for voltage support
in low-voltage grids with high penetration of photovoltaic," in Innovative Smart Grid
Technologies (ISGT Europe), 2012 3rd IEEE PES International Conference and Exhibition
on, 2012, pp. 1-7.
M. Braun, K. Büdenbender, D. Magnor, and A. Jossen, "Photovoltaic self-consumption in
Germany - Using Lithium-Ion storage to increase self-consumed photovoltaic energy,"
presented at the 24th European Photovoltaic Solar Energy Conference, Hamburg,
Germany, 2009.
Bibliography
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
91
C. A. Hill, M. C. Such, C. Dongmei, J. Gonzalez, and W. M. Grady, "Battery Energy Storage
for Enabling Integration of Distributed Solar Power Generation," Smart Grid, IEEE
Transactions on, vol. 3, pp. 850-857, 2012.
G. Mulder, F. D. Ridder, and D. Six, "Electricity storage for grid-connected household
dwellings with PV panels," Solar Energy, vol. 84, pp. 1284-1293, 2010.
P. M. S. Carvalho, P. F. Correia, and L. A. F. Ferreira, "Distributed Reactive Power
Generation Control for Voltage Rise Mitigation in Distribution Networks," Power
Systems, IEEE Transactions on, vol. 23, pp. 766-772, 2008.
T. Bülo, G. Bettenwort, D. Geibel, T. Degner, A. Seibel, J. P. d. Costa, et al., "Voltage
control in active, intelligent distribution networks," presented at the 27th EU PVSEC,
2012.
G. Arnold, M. Braun, T. Reimann, T. Stetz, and B. Valov, "Optimal reactive power supply
in distribution networks - Technological and economic assessment for PV-systems,"
presented at the 24th EU PVSEC, Hamburg, 2009.
G. Kerber and R. Witzmann, "Statistical Distribution Grid Analsysis and Reference
Network Generation," ew, vol. 107, pp. 22-26, 2008.
K. Strunz, "CIGRE-EU-LV-Benchmark," 2013.
S. Electric, "Use and maintenace of oil-immersed distribution transformers - Elvim
range," ed.
G. Hébrail and A. Bérard, "Individual household electric power consumption Data Set ",
É. d. France, Ed., ed: UCI Machine Learning Repository, 2012.
M. Braun, K. Büdenbender, T. Stetz, and U. Thomas, "Activation of energy management
in households," presented at the ETG Congress, Düsseldorf, 2009.
O. Kalab, "Standardisierte Lastprofile," ed: Wirtschaftskammer Oberösterreich.
J. Mačuhová and B. Haidn, "Labour input on Bavarian dairy farms with conventional or
automatic milking," presented at the International conference of agricultural
engineering, Valencia, Spain, 2012.
K. Bonkoß, J. Neiber, and D. S. Neser, "Energy savings in dairy farming," ed: Bayerische
Landesanstalt für Landwirtschaft, 2012.
(28/10/2013).
Solar
radiation
data.
Available:
http://www.sodais.com/eng/services/services_radiation_free_eng.php
A. U. Schmiegel, P. Knaup, A. Meissner, C. Jehoulet, H. Schuh, M. Landau, et al., "The Solion system: a stationary PV battery system," presented at the 5th International
Renewable Energy Storage Conference (IRES), 2010.
E. Demirok, D. Sera, R. Teodorescu, P. Rodriguez, and U. Borup, "Clustered PV inverters
in LV networks: An overview of impacts and comparison of voltage control strategies," in
Electrical Power & Energy Conference (EPEC), 2009 IEEE, 2009, pp. 1-6.
T. Stetz, M. Kraiczy, M. Braun, and S. Schmidt, "Technical and economical assessment of
voltage control strategies in distribution grids," Progress in Photovoltaics: Research and
Applications, vol. 21, pp. 1292-1307, 2013.
R. K. Roy. (04/04/2014). Overall Evaluation Criteria (OEC). Available: http://nutekus.com/wp-oec.html
Suedkabel, "Overhead Line Conductors," ed.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement