DEGREE PROJECT IN ELECTRICAL ENGINEERING,

SECOND CYCLE, 30 CREDITS

STOCKHOLM, SWEDEN 2017

Distribution grid capacity for

reactive power support
EYSTEINN EIRÍKSSON

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ELECTRICAL ENGINEERING
Distribution grid capability for reactive
power support

Eysteinn Eiríksson

Master Thesis, 2017

KTH - Royal Institute of Technology
Abstract
The modern power system is changing at a rate faster than would have been expected
20 years ago. More and more conventional power plants will be shut down in favour of
distribution generation (DG). This is happening now with the trend of introducing renew-
able energy sources (RES) to the power system.

The grids were designed to transfer power from generating units connected to the high
voltage grids towards the end consumers connected to the low voltage grids. With changed
power mix, power flows in the system will change resulting in possible grid problems. One
of the main problems is keeping the voltage within operational limits of the system. When
the generation exceeds the consumption in a distribution network, the power will flow from
the low voltage network towards the high voltage network (reverse power flow) which will
cause the voltage to rise in the low voltage network. Reactive power support from DG can
be a valuable resource to mitigate the problem. Reactive power is necessary to operate
the power system. The main source of reactive power is synchronous generators. If this
source is shut down, the reactive power must come from another source.

This thesis investigates if DG could be used to support reactive power to the high
voltage transmission network to control the voltage. For this purpose, a distribution
system located close to Worms, Germany will be studied. This distribution system consists
of two MV feeders with high penetration of DG, mostly photovoltaic (PV) but also wind
turbines (WT). Consumption and generation measurement data was provided by the local
distribution system operator (DSO). A few reactive power control methods are introduced
and tested on this system. From the results, it is concluded that it is possible to provide
reactive power support from distribution networks and a voltage dependent reactive power
control can be used to this purpose.

1
1 Sammanfattning
Det moderna kraftsystemet förandras snabbara än vad som hade förväntats för 20 år
sedan. Fler och fler konventionella kraftverk kommer att stängas till fördel för distribu-
tionsgenering. Detta händer nu med trenden att introducera förnybara energikällor till
kraftsystemet.

Nätverket utformades för att överföra kraft från generatorer som är anslutna till högspän-
ningsnätet mot konsumenter anslutna till lågspänningsnätet. Med ändrad kraftblandning
kommer strömflödena i systemet att förändras vilket resulterar i eventuella nätproblem.
Ett av huvudproblemen är att hålla spänningen inom operativa gränser för systemet.
När generationen överstiger förbrukningen i ett distributionsnät, kommer strömmen att
strömma från lågspänningsnätet till högspänningsnätet vilket kommer att leda till att
spänningen stiger i lågspänningsnätet. Reaktivt kraftstöd från distributionsgenering kan
vara en värdefull resurs för att mildra problemet. Reaktiv effekt är nödvändig för att driva
elsystemet. Huvudkällan för reaktiv kraft är synkrona generatorer. Om den här källan
stängs av måste den reaktiva effekten komma från en annan källa.
Denna avhandling undersöker om distributionsgenering skulle kunna användas för att
stödja reaktiv kraft till högspänningsöverföringsnätet för att styra spänningen. För detta
ändamål studeras ett distributionssystem som ligger nära Worms, Tyskland. Detta distri-
butionssystem består av två MV-matare med med mycket distributionsgenerering, främst
solceller men även vindturbiner. Förbruknings- och generationsmätningsdata tillhan-
dahölls av den lokala distributionssystemoperatören. Några reaktiva effektstyrningsme-
toder introduceras och testas på detta system. Av resultaten dras slutsatsen att det
är möjligt att tillhandahålla reaktivt kraftstöd från distributionsnät och en spännings-
beroende reaktiv effektstyrning kan användas för detta ändamål.

2
2 Acknowledgement
I would like to thank my supervisors, Stefan Stankovic and Poria Hasanpor Divshali,
at the Department of Electric Power and Energy Systems for excellent supervision during
this project. I would also like to thank Lennart Söder at the Department of Electric Power
and Energy Systems for useful advice and for being my exminer. Lastly, I would like to
thank my friends and family for endless support throughout the progression of this project.

3
Contents

1 Sammanfattning 2

2 Acknowledgement 3

3 Introduction 7
3.1 Overview of the report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4 Background and literature review 8

4.1 Reactive power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.2 Importance of reactive power . . . . . . . . . . . . . . . . . . . . . 9
4.2 DG effect on distribution grid . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.3 Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.3.1 Voltage regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.3.2 Reactive power regulations . . . . . . . . . . . . . . . . . . . . . . . 11
4.3.3 Three phase connection . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.4 Voltage control in distributed networks . . . . . . . . . . . . . . . . . . . . 12
4.4.1 Voltage Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.4.2 Active power control . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.4.3 Reactive power control . . . . . . . . . . . . . . . . . . . . . . . . . 13

5 Model Description 16
5.1 General introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2 German electricity market . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.3 Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.4 Feeder 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.4.1 General information . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.4.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.5 Feeder 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.5.1 General information . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.5.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.6 Generation Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.6.1 Active Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.6.2 Reactive Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.7 Load Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.7.1 Active Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.7.2 Reactive Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.8 Reactive power capability . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

6 DigSilent Implementation 24
6.1 Power factors for small scale PV’s and loads . . . . . . . . . . . . . . . . . 24

7 Simulation Results 28
7.1 Active power flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
7.2 Base Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
7.2.1 Power flow at the primary substation . . . . . . . . . . . . . . . . . 29
7.2.2 Voltage profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4
 7.3 No reactive power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
7.4 Constant power factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.4.1 0.95 capacitive p.f. . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.4.2 0.95 inductive p.f. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7.5 Maximizing reactive power support . . . . . . . . . . . . . . . . . . . . . . 44
7.6 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
7.7 Maximizing reactive power provision . . . . . . . . . . . . . . . . . . . . . 48
7.8 Maximizing reactive power consumption . . . . . . . . . . . . . . . . . . . 54

8 Comparison 58

9 Conclusion 59

10 Future Work 60

Appendices 61

A Detailed diagram of feeder 1 61

B Detailed diagram of feeder 2 62

C List of Figures and Tables 63

5
Abbreviations and Symbols
DG Distributed Generation
HV High Voltage
MV Medium Voltage
LV Low Voltage
PV Photovoltaic
Q Reactive power
SUS Secondary Unit Substation (MV to LV)
WT Wind Turbine
p.f. Power factor
OLTC On-Load Tap Changer
p.u. Per unit
NLTC No-Load Tap Changer
DSO Distribution System Operator
cap. Capacitive power factor (generating Q)
ind. Inductive power factor (consuming Q)
RES Renewable Energy Sources

6
3 Introduction
Distribution generation sources (DG) are becoming an increasingly important factor in
the active power production in power systems. One of the main sources of reactive power
are synchronous generators. With increasing renewable energy sources (RES) connected
to the distribution grid, these synchronous generators will be shut down one by one. With
growing demand of stable electric power, reactive power importance is increasing. If one
of the main source of reactive power is going away, there is a need for another source of
reactive power. In this thesis, DG will be proposed as a new source of reactive power. It will
be investigated if and to what extent the distribution grid is capable of providing reactive
power support. One of the problems with a large penetration of DG in low voltage networks
is operating within voltage limits. DGs can cause reverse power flows and over-voltage in
the low voltage grid. Reactive power support can be used to mitigate this problem. The
main objective of this thesis is to investigate if DG is capable to provide reactive power
support for voltage control in the overlaying grids, i.e. high voltage transmission grid,
while still operating within voltage limits.
A real German distribution grid with a high penetration of DG (mainly photovoltaic
(PV) but also wind turbines (WT)) is used to analyse this problem. This distribution grid
was modelled in DigSilent PowerFactory where all simulations were done. A few reactive
power control methods will be proposed and tested on this system.

3.1 Overview of the report

First, a literature review of reactive power, DG and voltage control will be given in
Chapter 4. Then a description of the system will be given in Chapter 5. In Chapter 6.1, the
implementation in DigSilent will be explained and the base case investigated. Different
methods of reactive power control and the results are presented in Chapter 7. Finally,
conclusion and future work is discussed in Chapter 9 and 10.

7
4 Background and literature review
4.1 Reactive power
4.1.1 Basics
AC power systems produce and consume two types of electrical power, active and re-
active power. Active power is the true power given to any load. Active power is measured
in units of watts (W). Reactive power moves back and forth in the power system. It is
produced by inductive and capacitive loads. It only exists when there is a phase displace-
ment between voltage and current. It is measured in units of volt-ampere reactive (VAr).
The apparent power is the total power (combination of active and reactive power). It is
measured in units of volt-ampere (VA). [1]
Equation 4.1 shows the relationship between active, reactive and apparent power.
p
S= P 2 + Q2 (4.1)
where S is the apparent power, P is the active power and Q is the reactive power.
Another way to see the relationship is to look at the power triangle shown in Figure 1

Figure 1: Power Triangle

The angle difference between the voltage and current is denoted with φ. The current
can both lead and lag the voltage causing leading and lagging power factors. In this thesis,
we will talk about capacitive power factor when the DGs/loads are injecting reactive power
and inductive power factor when they are consuming reactive power.

8
4.1.2 Importance of reactive power
Importance of reactive power in power systems is increasing with growing demand for
electric power. Electric power must be generated in a stable, reliable and cost effective
way. Reactive power is an important factor to be able to do that. The main reasons why
reactive power is so important are:

4.1.2.1 Voltage Control

Reactive power is an important factor in controlling the voltage. Over voltage can
damage the insulation in equipments and low voltage can cause poor performance and
also overheating of the equipments because of possible bigger currents.
How the voltage can be controlled by reactive power will be explained in more detail in
Section 4.2.

4.1.2.2 Electrical Blackouts

If we look at the following equation:

P =U ·I (4.2)

Power equals voltage times current. If the voltage is poorly controlled, it can cause high
current which can over load the lines and cause blackouts. In order to control the voltage
correctly, reactive power is important like discussed in the previous chapter.

9
4.2 DG effect on distribution grid
High penetration of distributed generation (DG) such as photovoltaic (PV) and wind
turbines (WT), have caused new challenges such as voltage rise and reverse power flows.
This constantly growing use of DG in medium-voltage distribution networks will highly
impact the development of future electrical systems.
Distribution networks are normally built up in the following way: The HV transmission
network is connected to the MV distribution network via a primary substation. A number
of feeders are connected to this substation. These feeders are in general radially connected.
Some feeders are connected in a ring but then one line is usually disconnected with a switch
so that no loop flow can occur. Underground cables are mostly used in these feeders and
they have capacitive characteristics so they produce reactive power. Loads connected to
these feeders are mostly resistive, i.e. have a power factor very close to 1. Some big loads
are connected to the MV network and they are required to have a power factor within a
range of 0.9 inductive and 0.9 capacitive [2]. The voltage drop along the feeder can be
approximated by:
R ∗ PLoad + X ∗ QLoad
∆U ≈ (4.3)
UN
Where,
∆U Voltage change across the line
PLoad Active power consumption by the load (negative)
R Resistance of the line
QLoad Reactive power consumption by the load (negative)
X Reactance of the line
UN Nominal voltage
the voltage in distribution networks with no DG decreases therefore from the primary
substation to the end of the feeder.
When DGs are connected, the power flow can be reverse. So the voltage can be higher
at the end of the feeder than at the primary substation.

R ∗ (PLoad + PDG ) + X ∗ (QLoad + QDG )

∆U = (4.4)
UN
If the generation is a lot higher than the demand, the voltage rise can exceed the limits.
To avoid this, the reactive power in the generation units can be utilized. Normally, loads
have a power factor close to unity. PVs and WTs are usually connected through an inverter
that can adjust the active and reactive power almost freely but limited by the current
limit of the inverter. When a DG unit is operated inductively, it consumes reactive power
so QDG comes negative which lowers the voltage. Capacitive operation however injects
reactive power which increases the voltage. The impact that reactive power adjustments
have depends highly on the R/X ratio of the lines. In MV grids, the R/X ratio is usually
around 1, so active and reactive power have equal impact on the voltage rise.

10
4.3 Regulations
The VDE 4105 code has regulations that must be followed for DG installation in Ger-
many. [3]

4.3.1 Voltage regulations

Generation units connected to the MV grid in Germany are required to be able:

• to stay connected during fault

• to support the voltage by providing reactive power during the fault

• to consume the same or less reactive power after the fault clearance

4.3.2 Reactive power regulations

It is required that the capability of the inverters to feed in with a displacement up to
0.95 leading or lagging and if the plant power exceeds 13.8 kVA, a displacement up to 0.90
must be supported.

4.3.3 Three phase connection

Connections larger than 4.6 kVA must be three phase connected. Smaller connections
are allowed to be one phase connected.

11
4.4 Voltage control in distributed networks
In this chapter, a few control methods used in distribution grids will be discussed and
explained. As can be seen in equation 4.1, the voltage depends on the active and reactive
power, so by controling them, the voltage can be controlled. Active power controls and
reactive power controls will therefore be explained in this chapter. There are also more
direct methods to control the voltage that will be explained.

4.4.1 Voltage Control

4.4.1.1 OLTC for transformers
On-load tap changers (OLTC) are used to control the voltage directly, i.e. without
using active or reactive power control. Normally, the HV/MV transformers that connects
the distribution network to the transmission grid has an OLTC to control the voltage at
the MV busbar. However, this control could be more advanced. The OLTC controls the
voltage on the MV busbar. One or more feeders are connected to this busbar. If the DG is
uneven, e.g. on one feeder there is a lot of generation but on another feeder there is not as
much because of lack of DG capacity or clouds, then the voltage profile for these feeders
would be very different and this could cause under or over-voltage. An advanced approach
could be to monitor the voltage at various points in the distribution grid. The downside of
this approach however is additional cost. The MV/LV transformers normally have no-load
tap changers (NLTC). They have a fixed operation point and can only be changed on-site
when the transformer is disconnected. OLTC have a high advantage over NLTC. By using
ONTC on the MV/LV transformers, the LV network could operate almost independent of
the MV network. Then the voltage on the LV side could be set to below nominal voltage
during high in-feed and above nominal voltage during low in-feed. [4]

4.4.2 Active power control

One way to control the voltage is by controlling the active power. There are a few ways
that this can be done, for example:

4.4.2.1 Batteries
Batteries are built to store energy. When the DG is high and the voltage rises, batteries
can be used to store the extra energy and lower the voltage. Also, when the DG is low
and the voltage is low, batteries can use this extra energy to inject active power in to the
grid and prevent the voltage to lower further.

4.4.2.2 Active power curtailment

Active power curtailment can be used to avoid the disconnection of inverters due to
over-voltage tripping. Active power can be limited to a fixed percentage of the nominal
power or voltage-dependent. Active power curtailment can also be used to increase hosting
capacity [5]. An example of a voltage dependent curtailment can be seen in Figure 2 where
the curtailment starts at U=1.07 p.u. and the generation is completely shut down at U=1.1
p.u.

12
 Figure 2: Example of active power curtailment

4.4.3 Reactive power control

In this section, the reactive power control of DGs will be explained in more detail. As
has already been discussed in Chapter 4.2, the reactive power can be used to control the
voltage level. The voltage rise caused by the DG unit can be expressed as follows:
R · PDG + X · QDG
∆UDG ≈ (4.5)
UN
Generation units must be able to operate between 0.95 inductive and 0.95 capacitive. The
most common ways to control the reactive power are: constant Q, constant PF, cosφ(P ),
cosφ(U ) and Q(V). These methods are explained in the following sections.

4.4.3.1 Constant Q
Constant amount of reactive power is consumed or provided by the DG unit indepen-
dent of the voltage at the bus or the active power generated. This method is rather easy to
implement but the disadvantage is that this will unnecessary produce or consume reactive
power even when it is not needed.

4.4.3.2 Constant PF
With constant power factor, the DG unit is not consuming or providing reactive power
when there is no active power. However, it can still consume or provide reactive power
when it is not needed.

13
4.4.3.3 cosφ(P )
With this method, the power factor changes according to the active power generation.
This method improves the constant PF method as it can lower the reactive power con-
sumption or provision when the infeed decreases. There are still some cases when the
voltage is not at its upper limit but the units are consuming a lot of reactive power, e.g.
when clouds cover only part of the PV plants, causing the voltage to decrease but there
are some PVs generating maximum active power. An example of this controller can be
seen in Figure 3.

Figure 3: Example of cosφ(P ) control

4.4.3.4 cosφ(U )
This method is very similar as cosφ(P ) as the voltage is very dependent on the active
power. This will however partially fix the cloud situation mentioned in Section 4.4.3.3 as
the power factor is only controlled by the voltage. An example of cosφ(U ) control can be
seen in Figure 4.

14
 Figure 4: Example of cosφ(U ) control

4.4.3.5 Q(U)
This methods consumes/provides reactive power as a function of the voltage. Here
the reactive power is not a percentage of the active power, as in the cosφ methods. Now
it is possible to consume or generate reactive power even when there is no active power
production, e.g. at night if the voltage falls down below certain threshold it is possible to
inject reactive power to make the voltage rise. This characteristic is shown in Figure 5.

Figure 5: Example of Q(U ) control

The red dotted lines represent a certain threshold. So that when the voltage falls below
the lower threshold, reactive power is injected to fix that. Also, when the voltage rises
above the upper threshold, reactive power is consumed to lower the voltage. In between
the two thresholds there is a dead-band, where no reactive power is consumed or provided.
Of course, this is only an example of a Q(U) control. The dead-band could be wider or
nothing at all. Also, there could be some constant reactive power consumption or provision
within the dead-band.

15
5 Model Description
5.1 General introduction
In this thesis, a model was developed to study steady-state problems described in Chap-
ter 3. To simulate this model, the power system toolbox PowerFactory 15.2 by DIgSILENT
has been used. First version of this model was created by S. Geidel from Energynautics.
The model was improved and expanded by L. Hulsmann in his master thesis from KTH
[6]. Now this model has been further studied by me. Only the final version of this model
will be described in this thesis.

This model consists of two MV feeders connected to the same primary substation.
These feeders are a part of a German distribution grid. This Chapter will describes the
model in detail. First a summary of the German electricity market and then the location
and layout of the system.

5.2 German electricity market

Germany is the largest electricity market in Europe, it was opened for competition
in 1998. There is not a single system operator like in many other countries. Table 1
summaries the electricity market in Germany.

Table 1: German electricity market

Electricity Market
Distribution Distribution Voltage Transmission Transmission Voltage
Grid (km) Level (kV) Grid (km) Voltage Level # DSOs
1780856 <=110 36001 150, 220, 380 817

Customers Electricity Installed Installed

# TSOs (m) Production (TWh) Capacity (GW) Capacity RES (GW)
4 50.3 594.7 204.6 97.9
Source: BNetzA (2016) [7]

The share of RES capacity is 48% of the total installed capacity in the system. Of that,
around 97% of the RES are connected to the distribution grid. Wind power and PV have
the highest share of DG, and have a share of around 83% of total RES installed capacity.
PV are mostly connected to the distribution system, 65 % of PV generators are connected
to LV (230/400 V) and about 35 % to MV (11-60 kV). Only a few are connected to HV
(110 kV). 95% of wind DG are connected to MV network. [8]

16
5.3 Location
The distribution system is located 50 km south of Frankfurt, Germany as can be seen
in Figure 6.

Figure 6: Location of the system

There is one primary substation that connects the 110 kV grid to the 20 kV distribution
grid via two 45 MVA transformers. The two distribution feeders are shown in Appendices
A and B. They will be explained in more detail in the following sections.

5.4 Feeder 1
5.4.1 General information
There are 47 MV/LV secondary unit substations (SUS) at feeder 1. The furthest is
located at a distance of 22 km from the primary substation. A total number of 3700
customers are connected to the LV grids of this feeder and one 7.3 MW PV farm. The PV
farm is connected to the MV side and is located in fre08, which is 18 km from the primary
substation. A diagram of feeder 1 can be seen in Appendix A.

5.4.2 Parameters
The feeder consists mainly of underground cables apart from a few overhead lines
interconnecting villages. Information about this feeder is shown in Table 2. R/X ratios
along the feeder are given in Table 4.
The MV/LV transformers at the SUS’s are equipped with off-load tap changers. All tap
changers have three possible positions with one of the following settings: 20.8/20.0/19.2
kV to 400 V or 20.8/20.4/20.0 kV to 400V

17
 Table 2: Information about feeder 1

Number of substations 47
Average number of customers per substation 79
Total number of customers 3712
Number of small scale PV plants 278
Capacity of small scale PV plants 4.4 MW
Capacity of large scale PV plant 7.3 MW
Maximum load in 2015 3.7 MW

5.5 Feeder 2
5.5.1 General information
There are 39 MV/LV SUS’s at feeder 2. The furthest is located at a distance of 19.5 km
from the primary substation. A total number of 3020 customers are connected to the LV
grids of this feeder. There is one 9.6 MW wind farm connected to the MV grid in wah06
(18 km from the primary substation). A diagram of feeder 2 can be seen in Appendix B.

5.5.2 Parameters
The entire MV network consists of underground cables. The LV grids also consists
of underground cables with the exception of ofs07 where the LV grid mainly consists of
overhead lines. The R/X ratio along the feeder is given in Table 4. More information
about this feeder can be seen in Table 3. The transformers at the SUS’s are equipped with
off-load tap changers with the same possible tap positions as described in 5.4.2.

Table 3: Information about feeder 2

Number of substations 40
Average number of customers per substation 76
Total number of customers 3020
Number of small scale PV plants 196
Capacity of small scale PV plants 3.3 MW
Capacity of wind farm 9.6 MW
Maximum load in 2015 5.7 MW

Table 4: R/X ratios along the feeders

Feeder 1 & 2
R/X ratio in MV grid very close to the primary substation
0.9
(0-1 km for feeder 1, 0-8 km for feeder 2)
R/X ratio in MV grid further away from the primary substation
1.7
(1-22 km for feeder 1, 8-19.5 km for feeder 2)
R/X ratio in LV distribution grid 2.6 (mostly)
R/X ratio in LV grid - customer connections 8.2

18
 The secondary substations were modelled in the following way: The LV networks were
not modelled in detail. The total load and generation from each substation were aggregated
into a single load and a single generating unit as can be seen in Figure 7:

PV
P
20 kV 400 V 20 kV
 



Load
P

Figure 7: Aggregation of loads and PV’s

More information about how this aggregation was done is found in Section 5.6.1 for
PV’s and Section 5.7.1 for loads.

19
5.6 Generation Modelling
5.6.1 Active Power
There is only one large scale PV plant (the 7.3 MW PV plant in fre08). 15 minutes
average active power values are available from the DSO. These values are available to di-
rectly put in Power Factory. The small scale PV plants do not have smart meters, so the
same values are not available for them. The only data available for these plants are: the
PV capacity of every installation and the SUS to which it is connected. For the modelling,
the capacity of all PV’s connected to the same SUS is summed and one PV generator used
to represent all the small scale PV’s. The output of the large scale PV plant is used as
a reference, e.g. if the average power output of the large scale PV plant is 30 % of the
capacity, the output of the small scale PV plants would also be 30 % of their capacity.
This is a reasonable approximation for PV plants close to fre08. The village Freimersheim
is only around 1 km from fre08. The PV’s in feeder 2 are between 10 to 16 km from fre08
but due to lack of measurements, the large scale PV plant was also used as a reference in
feeder 2. The active power for the wind farm in wah06 is directly taken from measurements.

5.6.2 Reactive Power

The small scale PV’s should operate at a unity power factor according to the DSO.
However, because some of the LV networks are not included in the model, a unity power
factor was not suitable for the small scale PV’s. In order to approximate the final reactive
power at the feeding point, a value of 0.9985 inductive (VAr consuming) was chosen for
the small scale PV’s in feeder 2. In feeder 1, a power factor value of 0.9992 inductive was
found to be a good approximation for all small scale PVs. This is further explained in
Chapter 5.
The large scale PV plant in fre08 has a cosφ(U ) controller and 15 minute average mea-
surements values for the reactive power are available. The cosφ(U ) controller has the same
characteristics as in Figure 4.
The wind farm in wah06 operates at a constant power factor of 0.95 inductive (VAr con-
suming).

20
5.7 Load Modelling
5.7.1 Active Power
Three things are known about each customer from the DSO: The total yearly electricity
consumption, the SUS to which they are connected and the categorization of the customer.
Each customer is categorized either as a RLM (German: Registrierende Leistungsmessung,
English: Recorded power) or as a SLP (Standard Load Profile).

RLM customers are customers that consume more than 100 MWh per year. They
are obligated to measure and record their power consumption. For these customers, a 15
minute average value for the active power is available and can be used directly as an input
in Power Factory.

SLP customers are further categorized into household and industry customers. Stan-
dard load profile is used to model these customers. This is a 15 minute average values over
a large sample of customers of a common type. A typical household and industrial SLP
for one week is shown in Figure 8. The yearly electrical consumption for each customer
connected to the same SUS was summed up and then the SLP used to generate a 15
minute average load profile.
Example: There are 57 household customers connected to substation fre03, their total
yearly consumption is 207777 kWh. The standard load profile for household has a yearly
consumption of 3700 kWh. Then each 15 minute value is multiplied by 207777kW 3700kW h
h
. The
same thing was done for all industrial SLP customers. [6]

Figure 8: Standard Load Profile, residential vs. industrial

In feeder 1, there are 12 RLM customers and 3700 SLP customers.

In feeder 2, there are 20 RLM customers and 3000 SLP customers.

21
5.7.2 Reactive Power
The loads usually have a unity power factor but they can also be slightly inductive or
capacitive.
For the reactive power modelling a power factor of 0.94 inductive (VAr consuming) was
found fitting for all loads on feeder 2 and a power factor of 0.975 inductive was found
fitting for all loads on feeder 1.

5.8 Reactive power capability

As mentioned in Section 4.3, all generation units must be able to operate with a power
factor between 0.95 inductive and 0.95 capacitive. If the generation unit is generating
maximum active and reactive power, then the apparent power will be:
Pmax
= 0.95
Smax
Smax = 1.0526Pmax
So the PV inverter or the inverter in the wind turbine must be able to handle the current
when the apparent power is roughly 5% more than the maximum active power.
The maximum reactive power at every time is therefore:
p p
Qmax = Smax 2 − P 2 = 1.0526Pmax
2 − P2

Here, Qmax is dependent on the instantaneous active power. This is shown in Figure 9
where Smax is shown with bold blue half-circle and Pmax is shown with a red line.

Figure 9: Reactive power capability

22
 It is however complicated to implement this controller because the active power gener-
ation is not known beforehand and therefore Qmax is not known and is always changing.
For a more simpler approach, Qmax could be constant. For example when the generation
unit is generating maximum active power, the maximum reactive power is:
p p √
Qmax = Smax 2 2
− Pmax = (1.0526Pmax )2 − Pmax2 = 0.108Pmax = 0.3287Pmax

Now Qmax is constant, 32.87 % of Pmax . This maximum limit of the reactive power is
shown in Figure 9 with the black lines. In this project, this limit (32.87 % Pmax ) was used
when reactive power controllers were tested.

23
6 DigSilent Implementation
As mentioned in Section 5.6.2, a fitting power factor for small scale PV’s and loads had
to be approximated. In this chapter this will be explained further and a base case will be
presented.

6.1 Power factors for small scale PV’s and loads

The system was modelled in DigSilent Power Factory and all load flow calculations
done there. Reactive power measurements from the local DSO are available at the primary
substation. These measurements were used to find a fitting power factor for the loads and
generation units by comparing the measured and simulated reactive power flow at the
primary substation.
First, all loads and small scale PV’s are modelled with a unity power factor. The
large PV farm has the cosφ controller and the wind farm has a power factor of 0.95 VAr
consuming. The reactive power from each feeder at the primary substation is shown in
Figures 10 and 11 compared to the measured reactive power.

Figure 10: Reactive power from feeder 1, simulated and measured

24
 Figure 11: Reactive power from feeder 2, simulated and measured

The root mean square error (RMSE) between the measured and simulated values are:

Feeder RMSE
1 0.42 MVAr
2 0.97 MVAr

From these figures it can be seen that a unity power factor for both loads and gener-
ation is not a good approximation so there is need to find a fitting power factor to get a
more accurate model.

There is too much reactive power generated for both feeders when the power factor for
both loads and DG is unity. Reactive power consumption is therefore needed in order to
match the simulated values with the measurement data.
This was done with a trial and error method. Power factors were tried out and then
the reactive power flow at the primary substation compared to the measurement data.
The best outcome was obtained with the following values:

Feeder p.f. for loads p.f. for PVs

1 0.975 ind. 0.9992 ind.
2 0.94 ind. 0.9985 ind.

The comparison with these values can be seen in Figures 12 and 13.

25
Figure 12: Reactive power from feeder 1 whith corrected pf, simulated and measured

Figure 13: Reactive power from feeder 2 with corrected pf, simulated and measured

The root mean square error (RMSE) between the measured and simulated values are:

Feeder RMSE
1 0.23 MVAr
2 0.18 MVAr

26
 The LV networks that are not modelled clearly have an effect on the reactive power
flow in the system. In order to model this, a new generator was put on all substations
that have LV network not modelled (all SUS except the PV farm and wind farm (fre08,
wah06)). This generator models the LV network. Active power is taken from the real
generator and the power factor used to calculate Q that is consumed by this generator.
The new generator is added to the substation like shown in Figure 14.

PV
P
20 kV

Load
P

P = 0, Q 6= 0

Figure 14: Modelling of loads and PV units at SUS

This generator is added so that it is easier to try Q controllers. Then this new Q
generator does not change and different Q controllers can be implemented on the real
generator.

27
7 Simulation Results
In this chapter, a few different control methods will be tested on the system in order
to conclude how much reactive power support can be delivered from the two feeders.

− No reactive power support

− Constant power factor

These control methods have been tested and then a voltage dependent reactive power
method, Q(U), is used to maximize reactive power support.

7.1 Active power flow

The difference between load and generation from both feeders can be seen in Figures
15 and 16. Positive values mean that the generation exceeds the consumption. Two points
are marked in these Figures. The black circle marks the time when the generation minus
consumption is at minimum and the red circle marks the time when the generation minus
consumption is at maximum. These will be the two extreme scenarios used to check the
voltage in the system because this is when the voltages are most likely at maximum and
minimum in the network. More information about these scenarios can be seen in Table 5.

Figure 15: Difference between generation and load, Feeder 1

Figure 16: Difference between generation and load, Feeder 2

28
 Table 5: The scenarios for both feeders

Feeder 1 Feeder 2
Scenario
Time Gen [kW] Cons. [kW] Time Gen [kW] Cons. [kW]
Max Gen-Cons. Day 6, 14:15 9620 2248 Day 1, 05:15 9600 2306
Min Gen-Cons. Day 5, 20:45 0 2026 Day 5, 18:45 133 3765

7.2 Base Case

7.2.1 Power flow at the primary substation
The system was simulated without making any changes to it, i.e. the controller at the
PV farm was not changed and the constant power factor at the wind farm was not changed.
The data was from 6 days in 2015, from 18.06.2015 to 23.06.2015. The power flow at the
primary substation was recorded for these 6 days for both feeders and is shown in Figures
17 - 18, (positive means power going from the feeder towards the primary substation and
the HV grid).

Figure 17: Power flow from the primary substation, Feeder 1, Base Case

29
 Figure 18: Power flow from the primary substation, Feeder 2, Base Case

For feeder 1, when the large PV farm in fre08 starts to generate active power, it is
controlled to consume reactive power. During daytime the feeder is taking reactive power
from the HV grid.

For feeder 2, the wind farm is consuming reactive power, so most of the time the feeder
is taking reactive power from the HV grid.

The average active power losses were calculated for both feeders in the following way:
576
1 X i
PLosses = (PG−L + PPi S )
576 i=1

where PG−L
i
is the difference between generation and load at time i, taken from Figures
15 and 16. PPi S is the active power flow from the primary substation at time i, taken from
Figures 17 and 18. There are 576 quarters for these 6 days. The average active power
losses in this base case were calculated to be 27 kW for feeder 1 and 113 kW for feeder 2.

7.2.2 Voltage profiles

The voltage at each substation was also recorded for the 6 days. The two scenarios
in Table 5 were used for the voltage profiles. The voltage profile for these two scenarios
can be seen in Figures 19 - 21. In Figure 19 substation 1 is the primary substation and
substation 29 is wal01 which is the substation at the end of the feeder (see Appendix A).

30
 Figure 19: Voltage profile for two scenarios at feeder 1, Base Case

Two voltage profiles were made for feeder 2. First one that is from the primary sub-
station to the wind farm bus (wah06) and can be seen in Figure 20. The other goes to
the end of the feeder (ofs02), it can be seen in Figure 20. Buses one to five are the same
for both voltage profiles, bus one is the primary substation and bus five is mon01. See
Appendix B where voltage profile A is marked in red and voltage profile B is marked in
blue, the common part and other branches are coloured black.

Figure 20: Voltage profile for two scenarios at feeder 2 A, Base Case

In Figure 21 the voltage reaches maximum at mon01 (bus 5), that is because most of
the generation is at wah06, and wah06 is at the end of a branch that goes from mon01.
The voltage continues to rise in Figure 20 all the way to the wind farm bus wah06.

31
 Figure 21: Voltage profile for two scenarios at feeder 2 B, Base Case

A voltage profile with respect to time was made for the most vulnerable bus (the bus
that came closest to the voltage limits). In this case it was bus wah06 as can be seen in
Figure 21. The voltage profile can be seen in Figure 22.
The maximum voltage at bus wah06 was 1.072 p.u.

Figure 22: Voltage profile with respect to time for wah06, Base Case

32
7.3 No reactive power
The system was simulated with no source of reactive power, i.e. all the DGs are not
injecting or consuming any reactive power. The only change here from the base case is
that the controllers on the PV farm bus fre08 and the wind farm bus wah06 were taken
away. The generators added in Section 6.1 are however still consuming reactive power.
The power flow for both feeders for the six days can be seen in Figures 23 - 24.

Figure 23: Power flow from the primary substation, Feeder 1, zero Q

Figure 24: Power flow from the primary substation, Feeder 2, zero Q

33
 Even when there is no reactive power in the DG’s, there is still reactive power flowing
from the feeder to the primary substation, for feeder 1 the average reactive power provision
is 700 kVAr. For feeder 2 the average reactive power provision is 167 kVAr. This can be
explained by that the distribution network is mostly supplied with underground cables and
they work like capacitors, i.e. they provide reactive power. The average reactive power
at the primary substation, standard deviation and network losses can be seen in Table 6
and in Table 7, the voltage intervals (Umax and Umin ) can be seen for both feeders. The
voltage profile was made for the two scenarios in Figures 25 - 27. The red dashed line are
scenario 1 from the Base Case and the blue dashed line are scenario 2 from the Base Case.

Figure 25: Voltage profile for two cases at feeder 1, zero Q

Figure 26: Voltage profile for two cases at feeder 2 A, zero Q

34
 Figure 27: Voltage profile for two cases at feeder 2 B, zero Q

In scenario 2, the voltage profile doesn’t change from the base case. There is very little
generation in scenario 2 and therefore there is no reactive power consumption in the base
case and the voltage stays the same. In scenario 1 the voltage is higher when there is no
consumption of reactive power. Why this happens has been explained in Section 4.2.
A voltage profile with respect to time was made for the most vulnerable bus (the bus
that came closest to the voltage limits). In this case it was the wind farm bus wah06 as
can be seen in Figure 26. The voltage profile can be seen in Figure 28. The maximum
voltage at bus wah06 was 1.088 p.u.

Figure 28: Voltage profile with respect to time for the wind farm bus wah06, zero Q

35
 Table 6: Comparison between the base case and no reactive power method

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Table 7: Comparison between voltage intervals, base case and no reactive power method

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01

7.4 Constant power factor

7.4.1 0.95 capacitive p.f.
The system was simulated when all DG’s had a constant power factor of 0.95 capacitive
(providing reactive power). The power flow from the primary substation can be seen in
Figures 29 - 30.

Figure 29: Power flow from the primary substation, Feeder 1, constant p.f. 0.95 cap.

36
 Figure 30: Power flow from the primary substation, Feeder 2, constant p.f. 0.95 cap.

The mean values, standard deviations and average losses can be seen in Table 8 and
the voltage intervals can be seen in Table 9

Table 8: Comparison between Base Case, No Q and Constant p.f. 0.95 cap.

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Constant p.f. 0.95 cap. -1248 597 33 (+22%) -1350 832 113 (0%)

Table 9: Comparison between Base Case, No Q and Constant p.f. 0.95 cap., Voltage interval

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01
Constant p.f. 0.95 cap. 1.09 1.02 1.11 1.02

37
 The voltage profiles for the two scenarios in Table 5 can be seen in Figures 31 - 33.
The dashed lines in the Figures are from the base case. There is no generation in scenario
2 for feeder 1 and therefore the voltage profile doesn’t change at all because the reactive
power provision is zero on all DG’s as it was in the base case. On feeder 2, the generation
in scenario 2 is very small so the voltage profile changes a little bit.

Figure 31: Voltage profile for two scenarios at feeder 1, constant p.f. 0.95 cap.

Figure 32: Voltage profile for two scenarios at feeder 2 A, constant p.f. 0.95 cap.

38
 Figure 33: Voltage profile for two scenarios at feeder 2 B, constant p.f. 0.95 cap.

In this case, the voltage limit is violated on feeder 2 as can be seen in Figure 32.
A voltage profile with respect to time was made for the most vulnerable bus (the bus
that came closest to the voltage limits). In this case it was the wind farm bus (wah06) that
exceeded the voltage limits the most as can be seen in Figure 32 (bus 16). The voltage
profile can be seen in Figure 34. The maximum voltage at wah06 was 1.106 p.u.

Figure 34: Voltage profile with respect to time for wah06, constant p.f. 0.95 cap.

39
7.4.2 0.95 inductive p.f.
The system was simulated when all DG’s had a constant inductive power factor of 0.95.
The power flow can be seen in Figures 35 and 36. The mean values, standard deviation
and losses can be seen in Table 10 and the voltage intervals can be seen in Table 19.

Figure 35: Power flow from the primary substation, Feeder 1, constant p.f. 0.95 ind.

Figure 36: Power flow from the primary substation, Feeder 2, constant p.f. 0.95 ind.

40
 Table 10: Comparison between Base case, No Q and Constant p.f. methods

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Constant p.f. 0.95 cap. -1248 597 33 (+22%) -1350 832 113 (0%)

Constant p.f. 0.95 ind. -145 798 28 (+3.7%) 1039 1070 115 (+1.8%)

Table 11: Comparison between Base Case, No Q and Constant p.f. methods,
Voltage intervals

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01
Constant p.f. 0.95 cap. 1.09 1.02 1.11 1.02
Constant p.f. 0.95 ind. 1.07 1.02 1.07 1.01

Voltage profile for both cases and both feeders can be seen in Figures 37 - 39.

Figure 37: Voltage profile for two cases at feeder 1, constant p.f. 0.95 ind.

41
 Figure 38: Voltage profile for two cases at feeder 2 A, constant p.f. 0.95 ind.

Figure 39: Voltage profile for two cases at feeder 2 B, constant p.f. 0.95 ind.

The voltage profiles are almost identical to the base case. The reactive power flow on
feeder 2 is almost the same for this case and the base case because in both scenario 1 and 2
there is almost no PV generation and therefore almost no reactive power consumption by
the PVs, the wind farm has a power factor of 0.95 inductive in the base case and therefore
the reactive power flow is almost the same in feeder 2 for this case and the base case. On
feeder 1 in the base case, the PV farm has a cosφ(U ) controller like shown in Figure 4 and
in scenario 1, the voltage on the PV farm bus is 1.07 p.u. which means the power factor
from the PV farm is around 0.92 inductive. The PV farm is consuming more reactive
power in the base case than in this case but the rest of the PVs are consuming with a
power factor of 0.95 inductive in this case and nothing in the base case and therefore the
reactive power flow on feeder 1 is similar in this case and the base case. For scenario 2 on
feeder 1, there is no generation and therefore this case is identical to the base case on that
time.

42
 A voltage profile with respect to time was made for the most vulnerable bus.
Again, it was the wind farm bus wah06 as can be seen in Figure 37. The voltage profile
can be seen in Figure 40. The maximum voltage was 1.072 p.u.

Figure 40: Voltage profile with respect to time for wah06, constant p.f. 0.95 ind.

43
7.5 Maximizing reactive power support
In this section, a Q(U) control will be used to maximize the reactive power support.
In Figure 41, two examples of Q(U) control and the limits of the control can be seen.

Figure 41: Two examples of Q(U) control

The reactive power capability is from -32.87 %Pmax to 32.87 %Pmax represented by the
red lines. The blue line in Figure 41 is an example of a Q(U) control. If this control is
implemented in a DG unit, it will provide reactive power ( 30% of active power capacity)
until the voltage at the bus where the unit is connected reaches 0.96 p.u., then the reactive
power provision starts to decrease and when the voltage reaches 1.01 p.u. the reactive
provision is 20% of active power capacity. The pink line is another example of a Q(U)
control. This control is providing reactive power (20 % of active power capacity) until the
voltage reaches 0.91, then the reactive power provision starts to decrease and decreases
linearly and reaches 0 when the voltage reaches 1.1 p.u. This control method will first be
used to maximize the reactive power provision from both feeders and then to maximize
the reactive power consumption to both feeders.
A sensitivity analysis was made in order to decide from which DG unit the reactive power
support should come from. This analysis is explained in section 7.6.

44
7.6 Sensitivity Analysis
By controlling the reactive power in the DGs, the voltage profile changes. In order to
be able select which DG should contribute to the reactive power support the most, we need
to find out which DG has the least effect on the voltage. Therefore, a sensitivity analysis
was made in order to decide where the reactive power should come from, i.e. which DG
is the most sensitive to a change in reactive power consumption/provision. The analysis
goes like this:

− Set all DG power factors to 1, so there is no reactive power provision or consumption.

− Run a power flow and record the voltage at each substation.

− Put a constant 10 kVAr generation of reactive power at the first substation and
repeat step 2.

− Repeat step 3 for all the substation, one at a time.

The voltage change can be seen in Table 12 for feeder 1 and Table 13 for feeder 2.
From these results, the obvious conclusion is that the distance from the primary substation
has the most affect on the sensitivity of the substations.

45
 Table 12: Sensitivity, analysis Feeder 1

Name Distance from SS [m] ∆V [p.u.]∗10−5

wal01 22047 3.10
wal03 21664 3.04
wal02 21175 2.97
fre07 20290 2.85
fre02 20032 2.81
fre03 19475 2.73
fre06 19124 2.68
fre08 18422 2.58
fre05 17650 2.47
ket04 16734 2.35
ket02 16468 2.32
ket01 16011 2.25
ket06 15849 2.23
ket03 15622 2.19
ket05 15068 2.11
esl01 14632 2.05
esl03 14300 2.00
epp07 13078 1.91
epp06 13029 1.91
epp02 12714 1.86
epp03 12713 1.86
epp01 12372 1.81
din01 12562 1.76
flo02 11461 1.60
flo01 10601 1.55
flo05 11033 1.54
flo04 11783 1.48
ofl01 9086 1.33
ofl02 8759 1.29
ofl07 6884 1.02
ofl06 6307 0.93
ofl04 5985 0.89
ofl05 5371 0.80
ofl03 5067 0.76
ghm02 2802 0.45
ghm09 2420 0.39
ghm13 2462 0.38
ghm03 1412 0.31
ghm07 1671 0.26
ghm06 1643 0.25
ghm04 1360 0.21
ghm05 1200 0.19
ghm01 931 0.15
ghm11 380 0.06

46
 Table 13: Sensitivity analysis, Feeder 2

Name Distance from SS [m] ∆V [p.u.]∗10−5

ofs04 18939 2.90
ofs08 18656 2.86
ofs01 18608 2.85
wah06 18193 2.76
wah04 18110 2.75
ofs11 17788 2.73
ofs10 17625 2.71
ofs05 17451 2.68
ofs03 17218 2.65
wah03 17360 2.65
ofs09 16966 2.61
wah05 17031 2.60
ofs07 18118 2.57
wah02 16463 2.52
hos04 15739 2.43
hos02 15536 2.40
wah01 15559 2.39
hos03 15122 2.34
hos01 14655 2.27
mon14 13404 2.09
mon05 13196 2.05
mon02 12609 1.97
mon13 12322 1.93
mon08 11852 1.86
mon07 11531 1.81
mon12 11462 1.81
mon06 11252 1.77
mon21 11252 1.77
mon09 11899 1.76
mon03 11110 1.75
mon01 10601 1.68

These results will now be used to maximize the reactive power support with a Q(U)
controller.

47
7.7 Maximizing reactive power provision
In this section, a Q(U) control will be used to maximize the reactive power provision.
The sensitivity analysis performed in section 7.6 is used to determine the Q parameter for
U = 1.1 p.u.. This was done to determine how much reactive power can be provided
for the worst case scenario (when the voltages are at maximum). The procedure goes like
this: First the reactive power provision was raised on the strongest bus (ghm11 on feeder
1 and mon01 on feeder 2) until either the reactive power limits of the unit or the voltage
limits were met. Then the same was done for the next strongest bus (ghm01 on feeder 1
and mon03 on feeder 2) until either limits were met. This was done for all buses with DG
capacity. The results from this can be seen in Table 14.

Table 14: Q parameters for U=1.1 p.u.

Substation Q [% Pmax ] for U=1.1 p.u. Feeder

All PV’s on F1 32.87 1
Wind farm bus wah06 12 2
ofs01, ofs08, ofs01 0 2
Rest of SUS on F2 32.87 2

On feeder 1, all PV units can provide maximum reactive power when the voltage is
at maximum. On feeder 2, the PV units connected to the strongest 27 buses can provide
maximum reactive power. Then the wind farm bus reached the voltage limits when the
wind farm was providing reactive power of 12 % of its active power capacity. Then the
three remaining buses (ofs01, ofs08 and ofs04) could not provide any reactive power without
violating the voltage limits. Now we have Q parameters for one point in the Q(U) control
(when U=1.1 p.u.). Because all PV units can provide maximum reactive power when the
voltage is at maximum, it can be concluded that these units can provide maximum reactive
power for all voltages. So all PV units on feeder 1 can have a constant Q control and all
units can provide maximum reactive power all the time without violating the voltage
limits. On feeder 2 this is not the case. In order to find how the control for the units on
feeder 2 changes for different voltage values, Q parameters were defined for U = 0.9 p.u..
This was done with the same method, only now scenario 2 was used to determine the
parameters because in scenario 2 the voltage is at minimum. In Table 15 the Q parameter
for U = 0.9 p.u. can be seen for feeder 2.

Table 15: Q parameters for U=0.9 p.u. on feeder 2

Substation Q [% Pmax ] for U=0.9 p.u.

Wind farm bus wah06 32.87
ofs01, ofs08, ofs01 32.87
Rest of SUS on F2 32.87

For U = 0.9 p.u., all DG units on feeder 2 can provide maximum reactive power. Now
it can be estimated that the DG units connected to the first 47 buses can have a constant
Q control providing maximum reactive power at all times. The control for the wind farm
bus wah06 and the other three buses ofs01, ofs08 and ofs04 need to decrease the reactive
power provision at some voltage value so that the voltage limits are not violated.

48
 Now the Q parameters were found for these controllers at other voltage values (first
for U=1.0 p.u., then for U=1.01 p.u. and so on). The resulting controllers can be seen in
Figure 42 and in Tables 16 and 17 the resulting controllers can be seen.

Figure 42: Controllers for maximizing reactive power provision

The controller for the wind farm on wah06 is in red. The wind farm is providing 32.87
% of active power capacity until the voltage reaches 1.09 p.u. and then it decreases linearly
and reaches 12 % of active power capacity at U = 1.1 p.u..

Table 16: Controller for the wind farm bus wah06

Q @ U=0.9 - 1.09 p.u. Q @ U=1.1 p.u.

32.87 12

The controller for the PV units connected to ofs01, ofs08 and ofs04 is shown in green
in Figure 42. The units are providing 32.87 % of active power capacity until the voltage
reaches 1.03 then the Q provision decreases linearly and reaches 0 when U = 1.1 p.u..

Table 17: Controller for ofs01, ofs08 and ofs04

Q @ U=0.9 - 1.03 p.u. Q @ U=1.1 p.u.

32.87 0

The blue line in Figure 42 represents the controller for all PV units in feeder 1 and the
PV units connected to the strongest 47 buses in feeder 2. These units provide 32.87 % of
active power capacity all the time regardless of the voltage. The power flow and voltage
profiles can be seen in Figures 43 - 54.

49
Figure 43: Power flow from the primary substation, Feeder 1, maximizing reactive power
provision

Figure 44: Power flow from the primary substation, Feeder 2, maximizing reactive power
provision

The average Q values, standard deviation and active power losses can be seen in Table
18.

50
 Voltage profiles were made for the same two scenarios for both feeders. On feeder 1,
the voltage reaches 1.099 p.u. for scenario 1 on the PV farm bus fre08 (bus 22 in Figure
45). On feeder 2, the voltage reaches 1.099 on the wind farm bus wah06 (bus 16 in Figure
46). The voltage intervals for both feeders can be seen in Table 19.

Figure 45: Voltage profile feeder 1, Maximizing reactive power provision

Figure 46: Voltage profile feeder 2 A, Maximizing reactive power provision

51
 Figure 47: Voltage profile feeder 2 B, Maximizing reactive power provision

A voltage profile with respect to time was made for the bus that came closest to the
voltage limits. In this case it was bus wal01 on feeder 1. The maximum voltage was 1.099
p.u. However, the voltage didn’t go over 1.09 p.u. often (less than 2% of the time).

Figure 48: Voltage profile with respect to time for wal01, Maximizing reactive power
provision

52
 Table 18: Comparison between maximizing Q provision to other control methods

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Constant p.f. 0.95 cap. -1248 597 33 (+22%) -1350 832 113 (0%)

Constant p.f. 0.95 ind. -145 798 28 (+3.7%) 1039 1070 115 (+1.8%)

Max Q prov. -4512 128 129 (+378%) -4471 526 210 (+86%)

Table 19: Comparison between maximizing Q provision to other control methods, Voltage interval

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01
Constant p.f. 0.95 cap. 1.09 1.02 1.11 1.02
Max Q prov. 1.099 1.03 1.099 1.03

53
7.8 Maximizing reactive power consumption
In this section, a Q(U) controller will be used to maximize the reactive power con-
sumption to both feeders. First, the reactive power consumption from each DG unit was
found for when U=0.9 p.u. and then the possible reactive power consumption was found
for when U=1.1 p.u. for each DG unit. All the DG units can consume maximum reac-
tive power (32.87 % of active power capacity) so they all have a constant Q controller
consuming maximum Q all the time, regardless of the voltage.

Figure 49: Power flow from the primary substation, Feeder 1, maximizing reactive power
consumption

Figure 50: Power flow from the primary substation, Feeder 2, maximizing reactive power
consumption

The average Q values, standard deviation and active power losses can be seen in Table
20.

54
 Voltage profiles were made for the same two scenarios for both feeders. On both feeders,
the voltage decreases compared to the base case. The voltage intervals for both feeders
can be seen in Table 21.

Figure 51: Voltage profile feeder 1, Maximizing reactive power consumption

Figure 52: Voltage profile feeder 2 A, Maximizing reactive power consumption

55
 Figure 53: Voltage profile feeder 2 B, Maximizing reactive power consumption

A voltage profile with respect to time was made for the bus that came closest to the
voltage limits. In this case it was bus wal01 on feeder 1. The maximum voltage was 1.07
p.u.

Figure 54: Voltage profile with respect to time for wal01, Maximizing reactive power
consumption

56
 Table 20: Comparison between maximizing Q provision to other control methods

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Constant p.f. 0.95 cap. -1248 597 33 (+22%) -1350 832 113 (0%)

Constant p.f. 0.95 ind. -145 798 28 (+3.7%) 1039 1070 115 (+1.8%)

Max Q prov. -4512 128 129 (+378%) -4471 526 210 (+86%)

Max Q cons. 3216 133 88 (+226%) 4493 190 219 (+94%)

Table 21: Comparison between maximizing Q provision to other control methods, Voltage interval

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01
Constant p.f. 0.95 cap. 1.09 1.02 1.11 1.02
Max Q prov. 1.099 1.03 1.099 1.03
Max Q cons. 1.07 1.00 1.07 0.99

57
8 Comparison
In this section, the different methods will be compared. In Table 22 is the average
Q, standard deviation of Q and average losses for each method. In Table 23, the voltage
intervals for each method can be seen.

Table 22: Comparison between maximizing Q provision to other control methods

Feeder 1 [11.7 MW] Feeder 2 [12.9 MW]

Method
Average Q St. Deviation of Q Average Losses Average Q St. Deviation of Q Average Losses

[kVAr] [kVAr] [kW] (% of BC) [kVAr] [kVAr] [kW] (% of BC)

Base case -388 581 27 (0%) 887 1044 113 (0%)

No reactive power -700 131 26 (-3.7%) -167 184 100 (-12%)

Constant p.f. 0.95 cap. -1248 597 33 (+22%) -1350 832 113 (0%)

Constant p.f. 0.95 ind. -145 798 28 (+3.7%) 1039 1070 115 (+1.8%)

Max Q prov. -4512 128 129 (+378%) -4471 526 210 (+86%)

Max Q cons. 3216 133 88 (+226%) 4493 190 219 (+94%)

Table 23: Comparison between maximizing Q provision to other control methods, Voltage interval

Feeder 1 Feeder 2
Method UM ax UM in UM ax UM in
[p.u.] [p.u.] [p.u.] [p.u.]
Base Case 1.07 1.02 1.07 1.01
No reactive power 1.08 1.02 1.09 1.01
Constant p.f. 0.95 cap. 1.09 1.02 1.11 1.02
Max Q prov. 1.099 1.03 1.099 1.03
Max Q cons. 1.07 1.00 1.07 0.99

In the no reactive power method, all the DG units have a unity power factor. There
is however a small reactive power provision from the distribution grid (700 kVAr from
feeder 1 and 167 kVAr from feeder 2). This is because of the cables in the network that
have a capacitive effect and are providing reactive power. This can also explain why the
maximum consumption is lower than the maximum provision on feeder 1 although all DG
units are providing/consuming maximum Q in both cases. When all the DG units are
consuming reactive power, they first need to consume the reactive power provided by the
network and then they can consume reactive power from the HV transmission grid.

58
9 Conclusion
The increase of RES connected to the distribution network results in possible grid
problems. The main problem is the reactive power support which is the subject of this
thesis. The main source of reactive power support comes from conventional power plants
and with an increase share of RES, they are shutting down one by one so there is a need
to find another source that can provide reactive power support for voltage control.
In this project, a German distribution grid with a high penetration of DG has been
investigated, this distribution grid consists of two MV feeders. The system was simulated in
DigSilent PowerFactory and data from the local DSO used for modelling of consumption
and generation. The DG capacity is 11.7 MW and 12.9 MW in feeder 1 and feeder 2
respectively. The objective of this project was to analyse if, and to what extent, the
distribution grid is capable of providing reactive power support for voltage control in the
overlaying grids, i.e. high voltage transmission grid. The maximization of reactive power
support was done with a Q(U) controllers and sensitivity analysis was used to decide
from which DG unit most of the reactive power support should come from. However,
this sensitivity analysis was not needed when maximizing the reactive power consumption
because all the units could consume maximum reactive power without violating voltage
limits. The sensitivity analysis was neither needed when maximizing the reactive power
provision from feeder 1 for the same reason. The main conclusion is that it is possible
to provide reactive power support from distribution grids with high penetration of DG
to the overlaying grids for voltage control. The most effective way is to combine both
Q(U) controllers and constant Q controllers to the generation units like was done when
maximizing reactive power provision on feeder 2, where the wind farm wah06 and PVs on
three other substations (ofs01, ofs04 and ofs08) had Q(U) controllers and the rest of the
PVs had a constant Q controller (see Section 7.7). The active power losses increase when
maximizing the reactive power support. When maximizing the reactive power provision,
the active power losses increased 140% when both feeders are considered as the same
network and 120 % when maximizing the reactive power consumption. Although this
seems to be very high increase, the average losses are really small in the base case (1.9%
losses on feeder 1 and 6.3 % losses on feeder 2)

59
10 Future Work
There are a few things that could be investigated more and better. First of all, a few
improvements of the model could be made:

− Model all LV networks in full with individual SLP for each customer.

− Use more reference points for the PV modelling.

These changes would make the results more accurate and meaningful. To make these im-
provements, more data from the local DSO are needed.
In this project, the position of the on load tap changer (OLTC) in the primary substation
was not considered a variable. If it is, it can be used to lower the losses and also increase
the reactive power support. This report only investigates the impact of distributed gener-
ation on the power system. There are other things that will increase in the power system
in the future, things like electric vehicles (EVs) and battery banks. If batteries were in-
cluded in the system, it could be possible to provide the loads with batteries when there
is no generation. EV’s would also change the load profile significantly. It would be inter-
esting to add these things to the model and investigate the affects it has on the power flow.

60
 Appendices
A Detailed diagram of feeder 1
Primary Substation
ghm11 v

ghm01 v v v v v2802 m
ghm05 ghm03 ghm09 ghm02

ghm04 v v1643 m
ghm06

ghm07 v

ghm13 v

ofl08 v

ofl03 v

ofl05 v

ofl04 v

ofl06 v

ofl07 v

ofl02 v

epp03 epp07
ofl01 v v v

flo04 v v v v v v v13029 m
flo01 flo03 epp04 epp01 epp02 epp06

flo05 v

flo02 v

ket08 ket04
din01 v v v

esl03 v v v v v16468 m
ket05 ket03 ket01 ket02

esl01 v

18422 m 19475 m
fre05 v v v v v v v v v22047 m
fre08 fre06 fre03 fre02 fre07 wal02 wal03 wal01

61
 B Detailed diagram of feeder 2
Primary Substation
fld05 x
mon13 mon15 mon14 mon18

fld07 x x x x x 13666 m

mon11 x x x x x x x x x x x x 18193 m
mon03 mon07 mon08 mon02 mon05 wah01 wah02 wah05 wah03 wah04 wah06

mon01 x
@
@
mon06 x @@x x x 12968 m
mon21 mon09 mon17

mon12 x 11899 m

mon04 x x
mon10

hos01 x

hos03 x x
hos04 ofs08

hos02 x x

ofs09 x x x x x
18608 m
ofs03 ofs10 ofs07 ofs01
18118 m
ofs05 x

ofs11 x

ofs04 x

ofs02 x
19459 m

62
C List of Figures and Tables
List of Figures
1 Power Triangle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 Example of active power curtailment . . . . . . . . . . . . . . . . . . . . . 13
3 Example of cosφ(P ) control . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Example of cosφ(U ) control . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Example of Q(U ) control . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
6 Location of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
7 Aggregation of loads and PV’s . . . . . . . . . . . . . . . . . . . . . . . . . 19
8 Standard Load Profile, residential vs. industrial . . . . . . . . . . . . . . . 21
9 Reactive power capability . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
10 Reactive power from feeder 1, simulated and measured . . . . . . . . . . . 24
11 Reactive power from feeder 2, simulated and measured . . . . . . . . . . . 25
12 Reactive power from feeder 1 whith corrected pf, simulated and measured . 26
13 Reactive power from feeder 2 with corrected pf, simulated and measured . 26
14 Modelling of loads and PV units at SUS . . . . . . . . . . . . . . . . . . . 27
15 Difference between generation and load, Feeder 1 . . . . . . . . . . . . . . 28
16 Difference between generation and load, Feeder 2 . . . . . . . . . . . . . . 28
17 Power flow from the primary substation, Feeder 1, Base Case . . . . . . . . 29
18 Power flow from the primary substation, Feeder 2, Base Case . . . . . . . . 30
19 Voltage profile for two scenarios at feeder 1, Base Case . . . . . . . . . . . 31
20 Voltage profile for two scenarios at feeder 2 A, Base Case . . . . . . . . . . 31
21 Voltage profile for two scenarios at feeder 2 B, Base Case . . . . . . . . . . 32
22 Voltage profile with respect to time for wah06, Base Case . . . . . . . . . . 32
23 Power flow from the primary substation, Feeder 1, zero Q . . . . . . . . . . 33
24 Power flow from the primary substation, Feeder 2, zero Q . . . . . . . . . . 33
25 Voltage profile for two cases at feeder 1, zero Q . . . . . . . . . . . . . . . 34
26 Voltage profile for two cases at feeder 2 A, zero Q . . . . . . . . . . . . . . 34
27 Voltage profile for two cases at feeder 2 B, zero Q . . . . . . . . . . . . . . 35
28 Voltage profile with respect to time for the wind farm bus wah06, zero Q . 35
29 Power flow from the primary substation, Feeder 1, constant p.f. 0.95 cap. . 36
30 Power flow from the primary substation, Feeder 2, constant p.f. 0.95 cap. . 37
31 Voltage profile for two scenarios at feeder 1, constant p.f. 0.95 cap. . . . . 38
32 Voltage profile for two scenarios at feeder 2 A, constant p.f. 0.95 cap. . . . 38
33 Voltage profile for two scenarios at feeder 2 B, constant p.f. 0.95 cap. . . . 39
34 Voltage profile with respect to time for wah06, constant p.f. 0.95 cap. . . . 39
35 Power flow from the primary substation, Feeder 1, constant p.f. 0.95 ind. . 40
36 Power flow from the primary substation, Feeder 2, constant p.f. 0.95 ind. . 40
37 Voltage profile for two cases at feeder 1, constant p.f. 0.95 ind. . . . . . . . 41
38 Voltage profile for two cases at feeder 2 A, constant p.f. 0.95 ind. . . . . . 42
39 Voltage profile for two cases at feeder 2 B, constant p.f. 0.95 ind. . . . . . 42
40 Voltage profile with respect to time for wah06, constant p.f. 0.95 ind. . . . 43
41 Two examples of Q(U) control . . . . . . . . . . . . . . . . . . . . . . . . . 44
42 Controllers for maximizing reactive power provision . . . . . . . . . . . . . 49
43 Power flow from the primary substation, Feeder 1, maximizing reactive
power provision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

63
44 Power flow from the primary substation, Feeder 2, maximizing reactive
power provision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
45 Voltage profile feeder 1, Maximizing reactive power provision . . . . . . . . 51
46 Voltage profile feeder 2 A, Maximizing reactive power provision . . . . . . 51
47 Voltage profile feeder 2 B, Maximizing reactive power provision . . . . . . 52
48 Voltage profile with respect to time for wal01, Maximizing reactive power
provision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
49 Power flow from the primary substation, Feeder 1, maximizing reactive
power consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
50 Power flow from the primary substation, Feeder 2, maximizing reactive
power consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
51 Voltage profile feeder 1, Maximizing reactive power consumption . . . . . . 55
52 Voltage profile feeder 2 A, Maximizing reactive power consumption . . . . 55
53 Voltage profile feeder 2 B, Maximizing reactive power consumption . . . . 56
54 Voltage profile with respect to time for wal01, Maximizing reactive power
consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

64
List of Tables
1 German electricity market . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 Information about feeder 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 Information about feeder 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4 R/X ratios along the feeders . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5 The scenarios for both feeders . . . . . . . . . . . . . . . . . . . . . . . . . 29
6 Comparison between the base case and no reactive power method . . . . . 36
7 Comparison between voltage intervals, base case and no reactive power method 36
8 Comparison between Base Case, No Q and Constant p.f. 0.95 cap. . . . . . 37
9 Comparison between Base Case, No Q and Constant p.f. 0.95 cap., Voltage interval 37
10 Comparison between Base case, No Q and Constant p.f. methods . . . . . 41
11 Comparison between Base Case, No Q and Constant p.f. methods,
Voltage intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
12 Sensitivity, analysis Feeder 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 46
13 Sensitivity analysis, Feeder 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 47
14 Q parameters for U=1.1 p.u. . . . . . . . . . . . . . . . . . . . . . . . . . . 48
15 Q parameters for U=0.9 p.u. on feeder 2 . . . . . . . . . . . . . . . . . . . 48
16 Controller for the wind farm bus wah06 . . . . . . . . . . . . . . . . . . . . 49
17 Controller for ofs01, ofs08 and ofs04 . . . . . . . . . . . . . . . . . . . . . . 49
18 Comparison between maximizing Q provision to other control methods . . 53
19 Comparison between maximizing Q provision to other control methods, Voltage
interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
20 Comparison between maximizing Q provision to other control methods . . 57
21 Comparison between maximizing Q provision to other control methods, Voltage
interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
22 Comparison between maximizing Q provision to other control methods . . 58
23 Comparison between maximizing Q provision to other control methods, Voltage
interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

65
References
[1] Hadi Saadat, Power System Analysis WCB McGraw-Hill 1999

[2] TAB Mittelspannung 2008 Technical conditions for connection to the medium-voltage
network 2008

[3] VDE VDE-AR-N 4105 Generators connected to the low-voltage distribution network
- technical requirements for the connection to and parallel operation with low-voltage
distribution networks 2011

[4] Damir Jakus, Josip Vasilj and Petar Sarajčev, Voltage Control in MV Distribution
Networks Through Coordinated Control of Tap Changers and Renewable Energy Sources
2015

[5] Benoît Bletterie, Serdar Kadam and Julien Le Baut Increased hosting capacity by means
of active power curtailment 2016

[6] Leonard Hulsmann, Evaluation of two distribution grids in terms of PV penetration

limits and effectiveness of reactive power controls. KTH Master Thesis, 2016.

[7] BNetzA, 2016. Monitoring Report 2016. Bundesnetzagentur fur Elektrizitat, Gas,
Telekommunikation, Post und Eisenbahen, Bundeskartellamt.

[8] BMU, 2012. Distributed Generation in Germany: From policy planning to implemen-
tation to performance, presented at the Great Wall Renewable Energy Forum 2012,
Sino-German International Symposium on Renewable Energy and Distributed Gener-
ation, Beijing, December 10th 2012.

[9] Afshin Samadi, Ebrahim Shayesteh, Robert Eriksson, Barry Rawn, Lennart Söder,
Multi-objective coordinated droop-based voltage regulation in distribution grids with PV
systems. 2014

[10] Luis F. Ochoa, Andrew Keane, Gareth P. Harrison Minimizing the Reactive Sup-
port for Distributed Generation: Enhanced Passive Operation and Smart Distribution
Networks, 2011

[11] Sarina Adhikari, Fangxing Li and Huijuan Li P-Q and P-V Control of Photovoltaic
Generators in Distribution Systems 2015

66
TRITA TRITA-EE 2017:176
ISSN 1653-5146

www.kth.se