VSC HVDC in Meshed Networks (Master Thesis Report)

Master Thesis Report
Doc. title Revision Page
VSC HVDC in meshed networks 01 1/118
Summary:
This thesis investigates some advantages of VSC HVDC in meshed networks. A load flow
model that includes a VSC HVDC model is programmed. This load flow model is used
throughout the simulations.
The beneficial influence of VSC HVDC, when operated in parallel with an AC line, has already
been proven. A VSC HVDC link operated in parallel with an AC line in a simple two-bus
network has been investigated by ABB. This is extrapolated to small meshed networks. The
same conclusions are valid here: static voltage-power relationships improve with the inclusion of
VSC HVDC.
The main part of the thesis is to verify the feasibility, technical as well as economical, of VSC
HVDC on loss minimization in meshed networks. The losses of the different components of a
VSC HVDC link are studied and the impact of the overall installation on loss reduction in
meshed networks is verified through simulations. Simulations are performed on two networks:
the small IEEE 14 bus network and the larger IEEE 118 bus network. The networks are
investigated in their original state and in a state of increased loading. Finally, an economical
analysis is performed to investigate the economical feasibility of VSC HVDC on loss reduction in
meshed networks.
The main conclusion of this work is that VSC HVDC is technically feasible to reduce losses in
meshed networks for the networks under study. From an economical point of view however, it
is more beneficial for the TSO to keep the original network instead of replacing an AC line by a
VSC HVDC link. When VSC HVDC is included, the benefits of VSC HVDC do not outweigh the
costs.
TABLE OF CONTENTS
1 INTRODUCTION ................................................................................................................................... 4
1.1 PURPOSE ......................................................................................................................................... 4
1.2 SCOPE ............................................................................................................................................. 4
1.3 STRUCTURE ..................................................................................................................................... 5
2 ASPECTS OF VSC HVDC .................................................................................................................... 6
2.1 THE ROLE OF VSC HVDC IN THE FUTURE TRANSMISSION GRID ........................................................ 6
2.2 OUTLOOK OF VSC HVDC LINK ......................................................................................................... 7
2.3 CONVERTER TOPOLOGY ................................................................................................................... 8
2.4 ACTIVE AND REACTIVE POWER FLOW CONTROL .............................................................................. 10
2.5 CAPABILITY CHART ......................................................................................................................... 12
2.6 FOUR QUADRANT OPERATION......................................................................................................... 13
2.7 LCC HVDC VERSUS VSC HVDC ................................................................................................... 13
2.8 APPLICATIONS OF VSC HVDC........................................................................................................ 16
3 VSC HVDC LOAD FLOW MODEL ..................................................................................................... 18
3.1 LOAD FLOW CALCULATION [31] ....................................................................................................... 18
3.2 VSC HVDC MODEL USED IN LOAD FLOW CALCULATION .................................................................. 20
3.3 OUTLOOK OF LOAD FLOW CALCULATION INCORPORATING VSC HVDC............................................. 23
3.3.1 The Function HVDC_newtonpf()........................................................................................... 25
3.4 COMPUTATIONAL COMPLEXITY OF EXTENDED LOAD FLOW PROGRAM ............................................... 26
4 INFLUENCE OF VSC HVDC ON VOLTAGE STABILITY ................................................................. 29
4.1 ASPECTS ON VOLTAGE STABILITY ................................................................................................... 29
4.2 VOLTAGE STABILITY IMPROVEMENT WITH VSC HVDC: ABB TEST CASE .......................................... 33
4.3 TESTING OF THE ABB CASE WITH OWN MODEL OF VSC HVDC ......................................................... 36
4.4 VOLTAGE STABILITY IMPROVEMENT WITH VSC HVDC: SMALL MESHED NETWORKS.......................... 37
5 LOSSES OF VSC HVDC LINK ........................................................................................................... 42
5.1 OUTLOOK OF VSC HVDC LINK ....................................................................................................... 42
5.2 LOSSES IN CONVERTER STATION .................................................................................................... 43
5.2.1 Valve Losses......................................................................................................................... 43
5.2.2 Converter Transformer Losses ............................................................................................. 52
5.2.3 AC Harmonic Filter Losses ................................................................................................... 54
5.2.4 High-Frequency Blocking Filter Losses ................................................................................ 56
5.2.5 Phase Reactor Losses.......................................................................................................... 56
5.2.6 DC Capacitor Losses ............................................................................................................ 56
5.2.7 DC Smoothing Reactor Losses ............................................................................................ 57
5.2.8 DC Harmonic Filter Losses ................................................................................................... 57
5.2.9 Auxiliary and Other Equipment Losses ................................................................................ 57
5.3 CONVERTER STATION EQUIVALENT LOSS MODEL ............................................................................ 58
5.4 LOSSES OF DC CABLE.................................................................................................................... 60
6 IMPACT OF VSC HVDC ON LOSSES IN MESHED NETWORKS ................................................... 62
6.1 POTENTIAL ADVANTAGE OF VSC HVDC IN MESHED NETWORKS: REDISTRIBUTION OF POWER .......... 62
6.2 ADVANTAGE OF VSC HVDC IN MESHED NETWORKS: BOOSTING THE VOLTAGE ................................ 64
6.3 PROPER ACTIVE POWER FLOW CONTROL OF THE VSC HVDC LINK ................................................. 64
7 RESULTS OF INCORPORATING VSC HVDC IN MESHED NETWORKS ....................................... 66
7.1 INSERTING A NEW LINK IN A MESHED NETWORK: AC VERSUS VSC HVDC........................................ 66
7.1.1 Inserting a New AC Line ....................................................................................................... 67
7.1.2 Inserting a New VSC HVDC Link.......................................................................................... 68
7.1.3 Conclusion ............................................................................................................................ 69
7.2 REPLACING AN AC OVERHEAD LINE IN A MESHED NETWORK BY A VSC HVDC CABLE ...................... 70
7.3 EFFECT OF VSC HVDC ON A LARGER NETWORK: THE IEEE 118 BUS NETWORK ............................. 72
7.3.1 VSC HVDC Link between Buses 25 and 27 ......................................................................... 72
7.3.2 VSC HVDC Link between Buses 69 and 75 ......................................................................... 73
7.3.3 VSC HVDC Links between Buses 25-27 and 69-75 and Their Interaction .......................... 74
7.3.4 Conclusion ............................................................................................................................ 75
7.4 EFFECT OF VSC HVDC ON HEAVILY LOADED NETWORK ................................................................. 75
7.4.1 IEEE 14 Bus System ............................................................................................................ 75
7.4.2 IEEE 118 Bus System .......................................................................................................... 83
8 ECONOMICAL ANALYSIS ................................................................................................................ 87
8.1 DATA USED IN ECONOMICAL ANALYSIS ............................................................................................ 87
8.1.1 AC Data ................................................................................................................................ 87
8.1.2 VSC HVDC Data ................................................................................................................... 88
8.1.3 Ancillary Services Data ......................................................................................................... 88
8.2 FIRST SCENARIO: ANNUAL LOAD INCREASE ..................................................................................... 89
8.2.1 IEEE 14 Bus Test System .................................................................................................... 89
8.2.2 IEEE 118 Bus Test System .................................................................................................. 92
8.3 SECOND SCENARIO: EACH YEAR THE SAME LOADING PROFILE ........................................................ 92
8.3.1 IEEE 14 Bus Test System .................................................................................................... 93
8.3.2 IEEE 118 Bus Test System .................................................................................................. 93
8.4 CONCLUSION ................................................................................................................................. 94
9 CONCLUSIONS .................................................................................................................................. 95
10 FUTURE WORK .............................................................................................................................. 96
11 SUMMARY ...................................................................................................................................... 97
12 REFERENCES .............................................................................................................................. 100
13 ENCLOSURES .............................................................................................................................. 105

13.1 WORLDWIDE VSC HVDC PROJECTS AND THEIR BASIC PARAMETERS [80].................................. 105
13.2 DESCRIPTION OF THE VSC HVDC MATPOWER CODE IN MATLAB .......................................... 106
13.2.1 The Function runpf() ........................................................................................................... 106
13.2.2 The Function HVDC_losses() ............................................................................................. 106
13.2.3 The Function HVDC_Limits_Converters() .......................................................................... 106
13.2.4 The Function HVDC_newtonpf()......................................................................................... 107
13.2.5 The Function HVDC_CalculatePowers() ............................................................................ 107
13.2.6 The Function HVDC_Mismatches() .................................................................................... 108
13.2.7 The Function Convert_Jacobean() ..................................................................................... 108
13.2.8 The Function HVDC_Jacobean()........................................................................................ 109
13.2.9 The Function HVDC_Updating() ......................................................................................... 110
13.2.10 The Function HVDC_Limits() .......................................................................................... 110
13.2.11 The Function HVDC_PQpower() .................................................................................... 110
13.2.12 The Function updategen() ............................................................................................... 111
13.2.13 The Function HVDC_printpf().......................................................................................... 111
13.3 PARALLEL OPERATION OF AN AC LINK AND A VSC HVDC LINK .................................................. 112
13.4 DERIVATION OF A VOLTAGE SECURITY INDICATOR ...................................................................... 113
13.5 REACTIVE POWERS INJECTED BY CONVERTERS DURING GENERATION OF PV-CURVES ................ 114
13.6 HARMONIC SPECTRA ................................................................................................................ 115
13.7 IEEE 14 BUS NETWORK ........................................................................................................... 117
1 INTRODUCTION
1.1 Purpose
The purpose of this report is to study the influence of Voltage Source Converter High-
Voltage Direct-Current (VSC HVDC) technology in meshed networks. VSC HVDC is
also called HVDC Light by ABB.
Two main advantages of the inclusion of VSC HVDC in meshed networks are treated in
detail: the effect of VSC HVDC on voltage stability and the effect of VSC HVDC on
network loss reduction.
A first purpose of the thesis is to prove the beneficial influence of VSC HVDC on the
voltage profiles of the load buses in a meshed network.
In a second part, it is the aim to prove that network losses can be reduced when VSC
HVDC is included in a meshed network. This technical analysis is extended with an
economical analysis to investigate if loss minimization is also beneficial from a financial
point of view.
The target groups for this report are power system engineers and transmission system
operators who plan to insert a VSC HVDC link in their transmission network.
1.2 Scope
The steady-state relation between bus voltages and associated bus powers improves
with the inclusion of VSC HVDC. To avoid voltage collapse, the Transmission System
Operator (TSO) wants to know how far his system is from the steady-state voltage
stability limit. A simple voltage security indicator is developed to provide this information.
This report also describes the influence of VSC HVDC on total network losses in meshed
networks. When including a VSC HVDC link in a meshed network, the converters and
the DC cable introduce extra losses. On the other hand however, VSC HVDC has some
properties that have an impact on network losses in the surroundings of the link. The
active and reactive power flow controllability can decrease network losses in the
neighborhood of the DC link. A minimum in total network losses has to be obtained.
The optimal loading of the VSC HVDC link is investigated. If the TSO plans to operate
his grid to minimize total network losses, he has to know how to operate the VSC HVDC
link; he has to know how much active power has to be sent over the DC link to minimize
network losses.
The VSC HVDC technology and its influence on the surrounding AC network described
in this report are based upon information gathered from different scientific sources
(IEEE, ABB ). Whenever information is used that can not be found on public domain,
this is explicitly stated.
This report does not describe the following aspects:

VSC HVDC in radial networks
Multi-terminal VSC HVDC in meshed networks
Dynamic effects of VSC HVDC
These topics could be covered in future works.
1.3 Structure
This report has the following structure.

Section 1 Introduction (this section) describes the purpose and scope for this report.
Section 2 Aspects of VSC HVDC describes the increasing role of VSC HVDC in the
future transmission grid and describes the operational characteristics of a VSC HVDC
link. Differences between the different HVDC technologies are also explained.
Applications of VSC HVDC are also highlighted.
Section 3 VSC HVDC Load Flow Model outlines the principles of a load flow
calculation. The inclusion of a VSC HVDC model in load flow calculations is explained in
detail.
Section 4 Influence of VSC HVDC on Voltage Stability describes the beneficial
influence of VSC HVDC on the voltage profiles.
Section 5 Losses of VSC HVDC Link analyses the different components of a VSC
HVDC link in detail and specifies their losses. The loss model used in the load flow
calculation is also explained.
Section 6 Impact of VSC HVDC on Losses in Meshed Networks illustrates the impact
of a VSC HVDC link on the losses in a meshed network. Both active and reactive power
flow control have an effect on total system losses.
Section 7 Results of Incorporating VSC HVDC in Meshed Networks covers the
different simulation cases with their corresponding results. Two networks are studied:
the IEEE 14 bus system and the larger IEEE 118 bus system. The meshed networks
are studied in their original state and in a state where the system load is increased.
Section 8 Economical Analysis verifies the economical feasibility of VSC HVDC on
loss reduction in meshed networks.
Section 9 Conclusions concludes the foregoing work.
Section 10 Future Work gives some ideas for future works.
Section 11 Summary is a summary of the Master Thesis.
Section 12 References specifies source material and further reading.
In section 13 Enclosures are supplementing information.
2 ASPECTS OF VSC HVDC
New developments in the transmission grid result in the need for new technologies.
VSC HVDC could be playing an important role here in the future. This aspect is treated
in 2.1. The main components of a VSC HVDC link are outlined in 2.2. The most
important component of a VSC HVDC link is the voltage source converter. Different
VSC converter topologies are discussed in 2.3. Active and reactive power flow
controllability is the main advantage of VSC HVDC. However, there is a limit on the
amount of active and reactive power that can be exchanged between the converters and
the AC grid. The capability chart, as discussed in 2.4, illustrates the limits on power flow
control. VSC HVDC can operate in all four quadrants in the power plane as described in
2.5. Advantages and disadvantages of both VSC HVDC and the older LCC HVDC are
described in 2.7. To conclude this chapter, applications of VSC HVDC are discussed in
2.8.
2.1 The Role of VSC HVDC in the Future Transmission Grid
The scene of electric power systems has changed drastically in the last couple of years.
Different political, technological and socio-ecological aspects are the cause of this
evolution. Throughout Europe, there are two main factors contributing to this evolution:
the liberalization of the electricity markets and the interest in renewable energy,
especially wind energy, with many large projects planned [1].
Before the deregulation of the electricity market, each national grid was controlled by a
vertically integrated company. Power flows in the grid were well known. The centrally
coordinated utility was able to control power flows via connecting or disconnecting lines,
changing the tap of tap-changing transformers and inserting capacitor banks for reactive
power compensation. The operator then had control over both the transmission system
and the generators and could give the generators signals to change their operating
point. Thus, active and reactive power flows could be controlled by the same company.
The connections between the different national grids were built for three main reasons:
back-up, stability (maintaining synchronism) and long term contracts between two
national grids. With the liberalization of the electricity market, generation and
transmission are unbundled. Power is now exchanged on power exchange markets and
the power flows follow the price differences of the different areas (former national grids).
Power is thus exchanged via the borders of national grids although these lines were not
originally built for this reason. The real power flows are determined by Kirchoffs laws
however. Because of the low cross-border transfer capacity between countries, this can
lead to congestion of transmission lines.
On the other hand, the increase in renewable energy raises another problem that
transmission system operators have to deal with. With the increased attention in
renewable energy, new power flows are introduced in the network. Because of the
uncertain character of renewable energy, for example wind energy, this leads to
unexpected flows in the network. This raises the uncertainty in the transmission grid and
can also lead to congestion of the lines.
Previous problems can be overcome by installing new transmission lines to lighten the
load on congested lines. Especially new lines interconnecting different areas are very
interesting. However, new overhead lines are not installed because of political and
socio-ecological unwillingness. The problem with new overhead lines is the
environmental and visual impact. The visibility of an overhead line can also raise the
concern about electromagnetic fields with the general public.
The previous trends in electric power systems open the way for other alternatives in the
transmission grid. New technological developments are getting more attention. An
overview of the technical developments in the future transmission grid is presented in [2].
A special group of new technologies are the power flow controlling devices, consisting of
phase shifting transformers, Flexible Alternating Current Transmission Systems (FACTS)
and High-Voltage Direct-Current (HVDC) technology [3]. These devices are able to
reduce loop flows, lighten bottlenecks and allow the TSO to operate his grid closer to the
limits. Their dynamic behaviour also plays a beneficial role in the damping of oscillations
and the stability of the voltage in the surrounding grid.
This thesis focuses on HVDC, Voltage Source Converter (VSC) HVDC more specifically.
The insertion of VSC HVDC in meshed networks is investigated. Transmission networks
are meshed networks, contrary to distribution networks, that have a radial configuration.
The last few years, VSC HVDC has become an attractive option for bulk power transport
in meshed grids. Advantages are the high controllability of active and reactive power at
the converters terminals and the ability to increase the stability of the surrounding AC
system. Active power flow controllability can play an important role in the liberalized
electricity market. VSC HVDC has the ability to make contract flows and physical flows
equal. Disadvantages of VSC HVDC are the higher losses, the higher cost and the fact
that the technology is still fairly new, an aspect that can hold TSOs back from installing a
VSC HVDC link. The TSO can control his grid in an optimal way by minimizing network
losses, increasing security margins, increasing cross-border capacity Only
minimization of network losses is studied in this thesis.
2.2 Outlook of VSC HVDC Link
A VSC HVDC link consists of two voltage source converters. These VSCs are
connected either directly to each other (back-to-back configuration) or via a high-voltage
DC cable (full VSC HVDC). Active power is transported from the rectifier to the inverter
station. A schematic representation of a VSC HVDC link is represented in Figure 2-1.
Figure 2-1: A VSC HVDC transmission scheme

The reactance Xconv depicted in Figure 2-1 is the combined reactance of the phase
reactor and the converter transformer. UL is the voltage of the bus connected to the
converter terminal. The AC network is represented by an impedance ZL. On the DC
side, the voltage is expressed as Ud, the current as Id and the resistance as Rd.
2.3 Converter Topology
The operational behavior of a converter is similar to a generator; the VSC delivers a

sinusoidal voltage wave form of which the amplitude, the phase and the frequency can
be controlled independently of each other. From the DC voltage, the converter builds an
AC sinusoidal waveform using IGBT switches and the pulse width modulation (PWM)
technique [4]. For example, the three-phase two-level converter depicted in Figure 2-2,
Ud
can build a sinusoidal voltage using two different DC voltages: . The associated
2
voltage output is illustrated in Figure 2-3. A capacitor of infinite capacitance is assumed;
there is no voltage ripple present.
Figure 2-2: Three-phase two-level converter and associated AC voltage waveform for
one phase (fundamental frequency switching)
Figure 2-3: Single-phase AC voltage output for a two-level converter with PWM switching
Filtering the PWM waveform and obtaining a sinusoidal voltage pattern at fundamental
frequency is accomplished by the phase reactor [5].
A three-phase two-level converter is used in The Hellsjn HVDC Light project (3 MW,
10 kV rating) [6], the Gotland HVDC Light project (50 MW, 80 kV rating) [7], the
Directlink HVDC Light project (3 x 60 MW, 80 kV rating) [8] and the Estlink HVDC
Light project (350 MW, 150 kV rating) [9].
Another configuration used in VSC HVDC transmission is the three-phase three-level

Neutral Point Clamped (NPC) converter topology. Here the sinusoidal waveform can be
built using three levels of DC voltage: U d and zero. This topology is illustrated in
Figure 2-4 and Figure 2-5. A capacitor of infinite capacitance is assumed; there is no
voltage ripple present.
Figure 2-4: Three-phase three-level NPC converter and associated AC voltage

waveform for one phase (fundamental frequency switching)
Figure 2-5: Single-phase AC voltage output for a three-level NPC converter with PWM
switching
In the three-level NPC configuration there are twice as many valves used in comparison
to the two-level configuration. In the two-level converter topology, each valve has to be
able to withstand the voltage between the two DC terminals. In the three-level topology,
this is not the case. In normal operation, each valve in a three-level converter topology
has to be able to withstand only 50% of the terminal-to-terminal DC voltage [4]. The AC
voltage generated by the three-level converter contains fewer harmonics than the
voltage generated by the two-level converter. In order to further reduce harmonic
content, multi-level converter topologies are possible [4].
The three-level Floating Capacitor (FC) topology is an alternative to the three-level
neutral point clamped topology. In the floating capacitor topology, the additional voltage
step is achieved by the inclusion of a separate DC capacitor in each phase. This
topology generates the same AC waveform as the NPC topology.
A three-level configuration is used in the Cross Sound Cable HVDC Light project (330
MW, 150 kV rating) [10], [11] and the Murraylink HVDC Light project (220 MW, 150
kV rating) [12].
The PWM technique allows the VSC to build a waveform with variable amplitude and
variable phase angle, independent of each other. This feature determines active and
reactive power transferred, as explained in the following paragraph.
2.4 Active and Reactive Power Flow Control
A one-phase equivalent circuit diagram of one converter station of a VSC HVDC link is
shown in Figure 2-6.
Figure 2-6: Circuit diagram of a VSC HVDC converter station
The total converter station consists of valves, phase reactor, filter bus and power
transformer.
The fundamental base apparent power at the filter bus between the converter reactor
and the AC filter is:
*
Sb = P + jQ = 3.U F .I R 2-1
The active power component between converter and filter bus is given by:
U F .U C
P= .sin
X phase reactor 2-2
Whereas the reactive power component between converter and filter bus is given by:
U F .(U F U C .cos )
Q=
X phase reactor
2-3
where is the phase angle difference between the filter bus voltage UF and the
converter voltage UC.
Formulas 2-2 and 2-3 assume a lossless phase reactor.
Control of active power flow by VSC HVDC is accomplished by changing the phase
angle difference and control of reactive power flow is realized by changing the voltage
amplitude difference between converter bus and filter bus. The converters PWM control
can change the converters voltage angle and amplitude. Thus, active and reactive
power can be controlled independently [6].
Concerning reactive power flow control, two control methods are possible [13]. The VSC
HVDC converter can inject a predefined amount of reactive power (reactive power
control). It is however also possible to hold the voltage at the bus, connected to the
associated converter, constant (AC voltage control). The load flow calculation then
solves how much reactive power needs to be injected/absorbed by the converter to hold
the bus voltage at the predefined value.
Reactive power exchange between the converter and the AC grid depends on the
converter voltage relative to the AC grid voltage. To vary reactive power, the converter
voltage needs to be varied between the upper and the lower converter voltage limit. For
a two-level converter topology using sinusoidal PWM (SPWM), upper and lower voltage
limits are given by the modulation index M given by:
U C peak
M= 2-4

Ud .
4
with UC and Ud defined as in Figure 2-6.

A guideline for the maximum modulation index is 0.86 and for the minimum modulation
index is 0.77 (data from ABB Corporate Research Sweden). With lower modulation
index, the harmonic content increases. For the 150 kVDC HVDC Light configurations,
this results in a maximum fundamental AC output phase-to-neutral voltage of 72 kV and
a minimum fundamental AC output phase-to-neutral voltage of 64 kV. This is a margin
of 4 kV around 68 kV, or 6%. HVDC Light however uses PWM with third harmonic
injection instead of SPWM. This is the reason why converters with a DC voltage of 150
kV are able to generate a phase-to-phase voltage of 195 kV on the AC side [14]. In
Figure 2-7 however (see next paragraph), the VSC HVDC converter can still inject
reactive power if the AC grid voltage is 1.1 pu. This is only possible with converter
voltages higher than 1.1 pu. The capability chart in [17] shows that is it possible to inject
a maximum amount of reactive power of 0.4 pu when the AC voltage is already 1.1 pu.
Also, the tap-changing transformer between the AC grid and the filter bus allows for an
optimal control of the filter bus voltage, so that more reactive power can be injected. A
compromise is chosen here; in the following calculations, data from [14] is used and
lower and upper converter voltage limits were chosen to be 0.9 and 1.1 pu respectively.
These limits are also used in [32]. However, the user can change these limits to other
desired values.
Reference [15] uses the same basic principles for the load flow calculation model of a
VSC HVDC link as is described in chapter 3. A VSC converter is also represented by a
complex voltage source behind a reactance. The converter voltage limits used are also
0.9 and 1.1 pu for the AC voltage produced by the VSC.
2.5 Capability Chart
The operating area of a VSC HVDC link is defined by its capability chart. The capability
chart is a PQ-diagram defining the valid operating points of a VSC HVDC link. The
following limits apply to a VSC HVDC link [16]:
- The maximum current through the IGBTs. This results in a maximum MVA circle in
the power plane where maximum current and actual AC voltage are multiplied.
Thus, if the AC voltage decreases, the MVA circle shrinks.
- The maximum DC voltage level. This results in a maximum reactive power arc in the
power plane [17]. The reactive power injection is dependent on the difference of the
AC voltage the VSC can generate from the DC voltage and the actual AC voltage.
The larger the difference, the more reactive power can be injected. This is very
useful for the stability of the network surrounding the VSC HVDC link. If the AC
voltage of the grid drops, more reactive power can be injected in the network by the
VSC HVDC to support the voltage of the grid.
- The third limit is the maximum current through the DC cable. This results in a
maximum active power transferred over the link.
The complete capability chart including the three limiting factors is shown in Figure 2-7.
Figure 2-7: Capability chart of VSC HVDC
VSC HVDC can almost instantaneously take any operating point in the capability chart.
Power flow reversals are also possible. This is because VSC HVDC does not change
the DC voltage polarity but instead changes the DC current direction [16].
Because reactive power exchange between the converter and the grid is determined by
the converter upper and lower voltage limits and the fact that these limits are
incorporated in the calculations, reactive power limits determined by the maximum DC
voltage level are always fulfilled.
2.6 Four Quadrant Operation
VSC HVDC can change active and reactive power independently of each other. Active
power can be reversed and both reactive power injection and absorption are also
possible. This allows the VSC HVDC to operate in four quadrants, as depicted in the
capability chart in Figure 2-7. Control of active power is achieved via the phase angle
difference over the phase reactor. This is illustrated in Figure 2-8 (left).
If the converter voltage UC leads the AC network voltage UF, active power is injected in
the AC network (inverter). U is the voltage drop over the phase reactor. On the other
hand, if the converter voltage lags the AC network voltage, active power is absorbed by
the converter (rectifier).
Figure 2-8: Active and reactive power flow control of VSC HVDC link
Reactive power flow depends on the difference in voltage amplitude between the
converter voltage and the AC network voltage. This is illustrated in Figure 2-8 (right). If
the amplitude of the converter voltage is higher than the AC network voltage, the
converter is injecting reactive power (capacitive operation). On the other hand, if the AC
network voltage is higher than the converter voltage, the converter is absorbing reactive
power (inductive operation).
From the previous analysis, it is clear that the phase reactor plays an important role in a
VSC HVDC converter station.
2.7 LCC HVDC versus VSC HVDC
There are two types of HVDC systems in existence: the LCC (Line Commutated
Converter) HVDC and the VSC HVDC. ABB refers to both technologies as HVDC
Classic and HVDC Light respectively. LCC HVDC uses thyristors for the conversion
from AC to DC and vice versa and VSC HVDC uses IGBTs with anti-parallel diodes
instead. LCC HVDC is a current source converter (CSC) topology; there exists a
constant current on the DC side. VSC HVDC is a voltage source converter topology with
a constant DC voltage. Power flow reversal with LCC HVDC is done by changing the
polarity of the DC voltage. With VSC HVDC this is done by changing DC current
polarity. This power flow reversal capability with LCC HVDC is limited: the maximum
ramping speed of the NorNed LCC HVDC cable is fixed at 20 MW/min leading to a full
power reversal in 1 hour [18]. With VSC HVDC, power can be changed almost
instantaneously; full power reversal of DC power flow can be made within a voltage cycle
(1/50 or 1/60 sec) if necessary [19].
The thyristor switches can be turned on, but they need the current to go through zero to
turn off. The switching frequency is thus 50 Hz. IGBTs on the other hand can be
switched on and off very fast, resulting in very high switching frequencies (kHz range).
The higher switching frequency of VSC HVDC results in a lower AC harmonic filtering
compared to LCC harmonic filtering.
LCC HVDC converter stations always consume reactive power. This is because the
current lags the voltage due to the firing of the thyristor switches. At normal operation, a
converter consumes a reactive power of approximately 50% of the transmitted active
power. This is a disadvantage for the surrounding network, because reactive power has
to be delivered to the LCC HVDC connection. This is accomplished by installing
capacitor banks or Static Var Compensators (SVC) at the end terminals of the LCC
HVDC link [20], [21]. 35% of the needed reactive power is delivered by the AC harmonic
filters and 15% is delivered by capacitor banks. LCC HVDC has limited reactive power
flow control by means of switching of filters and shunt banks and to some level by firing
angle control. But this control requires additional equipment. With VSC HVDC it is
possible to control both active and reactive power independently, and this is done almost
instantly.
An LCC HVDC converter station and associated power flows in comparison to a VSC
HVDC converter station and associated power flows are illustrated in Figure 2-9.
Figure 2-9: LCC HVDC (left) versus VSC HVDC (right)
LCC HVDC can be used with overhead lines and underground cables. VSC HVDC on
the other hand is always used with underground cables. One exception is the Caprivi
link interconnector, which uses overhead lines instead [22]. The reason why
underground cables are used is a reduced environmental impact and the fact that VSC
HVDC can not withstand DC faults; DC line-to-line and line-to-ground faults are critical.
The DC capacitor is short circuited and at the same time the converter bridge acts as a
diode bridge and contributes to the short circuit current on the DC side. The fault current
lasts until the AC circuit breakers trip. Therefore, applications where the VSC can be
most suitable for HVDC include back-to-back stations where no DC link exists or HVDC
transmission using cables instead of overhead lines, where the risk of faults on the DC
side is reduced. If a DC underground cable is used, a fault is probably a permanent fault
and the converter stations should be shut down. When DC overhead lines are used, the
fault is probably a temporary fault (e.g. lightning).
A comparison of a classical AC line, an LCC HVDC and a VSC HVDC transmission is
presented in Table 2-1 [3].
Table 2-1: Comparison between different transmission topologies
Benefits of VSC HVDC on voltage stability and rotor angle oscillation damping are
discussed in [16]. These benefits arise because of the fast active and reactive power
flow control of the VSC HVDC converters.
When inserting LCC HVDC in an AC system, the AC system needs to be strong enough
to support the LCC HVDC installation. When a temporary fault occurs in the AC system
connected to a converter, a commutation failure can occur, interrupting power flow. The
switching operation of the thyristors relies on a stable voltage and frequency. This is
expressed in the short-circuit ratio (SCR), given by the following formula [23]:
MVAAC
SCR = 2-5
MWDC
with MVAAC the short-circuit capacity of the connected AC system in MVA and MWDC the
rating of the converter terminal in MW. The effective short-circuit ratio (ESCR) includes
the effects of AC side equipment associated with the DC link: filters, shunt capacitors,
etc., which reduce the short-circuit capacity of the network. Traditionally, the AC system
strength has been classified as follow:
High, if ESCR is greater than 3
Low, if ESCR is between 2 and 3
Very low, if ESCR is less than 2
The above classification of AC system strength gives an idea of the interactions that can
occur when connecting an LCC HVDC to a weak AC system. Following problems can
arise when a DC system is connected to a weak AC system [24]:
High dynamic overvoltages
Voltage instability
Harmonic resonance
Voltage flicker
With VSC HVDC, the short-circuit capacity of the surrounding AC network is not
important; VSC HVDC has no problem feeding a passive network; this is a network
without synchronous machines. VSC HVDC has another advantage: loss of active
power has no influence on the AC voltage because the converter can control the
reactive power independently from the active power. Another advantage of both LCC
HVDC and VSC HVDC is that they do not contribute to the fault current [21].
VSC HVDC can help the transmission operator in maintaining a safe operating state of
his network. His network is characterized by the N-1 criterion. Losing the most
important component of his network may not cause the transmission system to black out.
If a generator or an important line trips, VSC HVDC can rapidly change its operating
point and temporary strengthen the grid [16].
The ability of VSC HVDC to produce the AC voltage and frequency by itself allows it to
be used in a black-start situation. When the grid suffers from a black-out, the VSC
HVDC link is available to restore the grid. LCC HVDC has no black-start capability
because it requires short-circuit capacity.
The main drawbacks of VSC HVDC in contrast to LCC HVDC are its high losses. LCC
HVDC losses are typically 0.6%-0.7% of the rated HVDC transmission capacity per
station at rated load [25]. VSC HVDC losses are typically 1.6%-1.7% of the rated
transmission capacity per station at rated load [26].
The largest projects installed to date are the Itaipu HVDC Classic system in Brazil
(6300 MW, 600 kV DC rating) [27] and the Estlink HVDC Light system between
Estonia and Finland (350 MW, 150 kV DC rating) [9]. Higher ratings are possible with
HVDC Light however, the largest being the M9 configuration (1140 MVA, 300 kV DC)
[5].
2.8 Applications of VSC HVDC
Situations where HVDC is preferred over AC are the following [20]:

Connecting two asynchronous systems. This is not possible with AC. HVDC can
connect asynchronous grids or even grids with different frequency.
Long cable connections. AC cables generate a capacitive current, thus lowering
the available capacity for active power transfer. DC cables dont have capacitive
charging currents, all the capacity of the cable can be used for active power
transfer.
Long distance transfer of power. The converter losses of HVDC are high. But
line losses are lower with DC than with AC. There exists a break-even distance
where the losses with HVDC become lower than the losses with AC.
Because of the advantages discussed in the previous paragraph, VSC HVDC is used in
the following applications [4], [28]:
Connecting wind farms. VSC HVDC can handle the variable frequency in the
wind farm and the wind farm is also electrically isolated from faults in the shore
grid.
Underground power links. When it is (nearly) impossible to obtain a permit for an
AC overhead line, VSC HVDC can play an important role.
Powering islands. VSC HVDC with underground sea cables is perfect for this
application.
Offshore platforms. VSC HVDC with underground sea cables is perfect for this
application. An example is the power delivery to the Troll A gas platform with an
HVDC Light with a rating of 84 MW and 60 kV DC [29].
Improved grid reliability.
City centre infeed. City centres are characterized by a high power demand and
significant network constraints. VSC HVDC can play an important role here [23].
Small, isolated remote loads. Places that are located far away from the grid and
that dont have their own generation can be powered via VSC HVDC [30].
Multi-terminal systems. With VSC HVDC, three or more converter stations can
make up a HVDC grid a lot easier than with LCC HVDC. Because the VSC
HVDC converters are voltage controlled, there is no need to balance the currents
like in LCC HVDC.
The increased attention in VSC HVDC in the future transmission grid has been outlined.
With its superior behavior in comparison to LCC HVDC, the interest in VSC HVDC will
continue to grow in the next years. The main advantage of VSC HVDC in comparison to
LCC HVDC is its ability to operate in all four quadrants. Other advantages and
disadvantages of VSC HVDC and LCC HVDC have been discussed in this chapter.
An overview of the existing VSC HVDC projects worldwide, their technical characteristics
and the reason why there is opted for VSC HVDC is given in Enclosure 13.1.
3 VSC HVDC LOAD FLOW MODEL
Load flow calculations result in a steady-state operating condition of the network. The
basic principles of a load flow calculation are outlined in 3.1. The load flow calculations
are adjusted to cope with the inclusion of one or more VSC HVDC links. The load flow
model for a VSC HVDC link and its parameters are highlighted in 3.2. The extra
calculations and their cooperation with the original load flow program are treated in detail
in 3.3. In a last section, the computational complexity of the extended program is
investigated. It is also checked if the behavior of the VSC HVDC model is as expected.
3.1 Load Flow Calculation [31]
Load flow calculations are performed when the user is interested in the state of the
network. A load flow is the steady-state solution of power flows through a network. The
network is represented by buses connected through lines. Input variables to the load
flow are:
- Generation properties (Pgen, Qgen, U)
- Load properties (Pload, Qload)
- Line properties (Rline, Xline, Bline)
- Type of bus
There are four unknown parameters per bus: the active power P, the reactive power Q,
the voltage amplitude U and the voltage angle . Two of these are defined for each type
of bus. There are three types of buses used in a load flow calculation:
- Slack or swing bus
- PV-bus
- PQ-bus
The slack bus is specified by a voltage magnitude and a voltage angle. This bus
supplies the losses to the network, which are unknown before the load flow calculation.
There is only one slack bus in a power network. PV-buses represent generators. These
are specified by a known active power injection and a known voltage magnitude. Often,
the minimum and maximum reactive powers of a generator are limited. If a generator
reaches this limit, the PV-bus is changed into a PQ-bus with Q the limited reactive power
of the generator. PQ-buses are specified by a known active and reactive power. These
buses represent loads.
Solving the network is done by iterative methods, for example the Newton-Raphson
method. The equations to solve are:
I = Y . U 3-1
*
S = U . I 3-2
with Y the admittance matrix and S the apparent power ( S =P+jQ). Only the Newton-
Raphson method is used here. Further information on other iterative methods used in
load flow calculations can be found in [31]. The working principle of the Newton-
Raphson method is illustrated in Figure 3-1.
Figure 3-1: The Newton-Raphson method used in load flow calculations
For buses p and q connected by a line, characterized by an admittance Ypq=Gpq+jBpq, it

follows from 3-1 and 3-2 for bus p:
n
( )
Pp (U , ) = U p G pq cos ( p q ) + B pq sin ( p q ) U q
q =1 3-3
n
( )
Q p (U , ) = U p G pq sin ( p q ) B pq cos ( p q ) U q 3-4
q =1
It follows that there are two non-linear simultaneous equations for each bus, except for
the slack bus, which is completely specified.
In each iteration step the difference between specified and calculated powers is
calculated and if the difference is smaller than a predescribed tolerance, the load flow
stops and convergence is reached. So, convergence is established if:
P = Pspecified P (U , ) tolerance
3-5
Q = Qspecified Q (U , ) tolerance
3-6
If convergence is not reached, the next values of V and are calculated using the
P
Newton-Raphson algorithm. If F ( X ) = and X = th
, then for the k iteration:

Q U
F ( X )( k ) = [ J ( k ) ].X ( k ) 3-7
or
P ( k ) P
(k )

P ( k )
U

.
(k )
(k ) = (k ) (k )
Q Q Q U
(k )

U
3-8
where J is the Jacobean matrix.
This process is repeated until P and Q are small enough. From the solution of U
and , the following results are obtained:
the active and reactive powers (formulas 3-3 and 3-4);
the losses in the lines;
the active and reactive power that has to be provided by the slack bus;
the reactive power that has to be provided by the PV-generators.
Thus, the steady-state operating condition of the network is found.
3.2 VSC HVDC Model Used in Load Flow Calculation
The VSC HVDC model used in the load flow calculation is based on [32]. A detailed
paper from the same authors on the inclusion of VSC HVDC in a Newton-Raphson
power flow algorithm is found in [33]. Further information on the implementation of multi-
terminal VSC HVDC can be found in references [34] and [35]. When implementing
FACTS devices in a power flow calculation using the same basic principles as described
here, the user is referred to [36]. The schematic representation of a VSC HVDC link
between two buses k and m is depicted in Figure 3-2.
Figure 3-2: Schematic representation of a VSC HVDC link used for load flow calculation
Each converter station is connected to a bus of the surrounding AC network via a

reactance. The reactance represents the combined reactance of phase reactor and
power transformer. Active and reactive power is controlled independently at each
converter station. The active power flow has one constraint however: the power leaving
the inverter station must equal the power flowing into the rectifier station minus the
losses. The losses consist of the losses in the rectifier station, DC cable and inverter
station. Any difference in active power between the two converter stations means that
the DC voltage rapidly increases/decreases [6]. To achieve this power balance one
converter station is defined to control the DC voltage while the other converter station is
controlling the active power transferred over the DC link [13]. A more detailed
description of the different types of losses in a VSC HVDC installation can be found in
chapter 4. The VSC HVDC link can be represented by two shunt-connected voltage
sources linked together by an active power flow constraint. Each voltage source is
connected to the associated AC bus through the combined reactance of phase reactor
and converter transformer. An equivalent circuit is shown in Figure 3-3.
Figure 3-3: Equivalent circuit of a VSC HVDC link used for load flow calculation
The complex voltage sources representing the converter stations are:

U c1 (cos c1 + j sin c1 ) 3-9
for the rectifier and
U c 2 (cos c 2 + j sin c 2 ) 3-10
for the inverter.
The constraining power flow equation for the active power through the VSC HVDC link
is:
Re {U c1.I r1* + U c 2 .I r 2* + Ploss rect + Ploss inv } = 0 3-11

for the back-to-back configuration and
Re {U c1.I r1* + U c 2 .I r 2* + Ploss rect + Ploss inv + Ploss DC cable } = 0 3-12
for the full VSC HVDC incorporating a DC cable.

For the active and reactive power flow equations at a bus connected to a converter, refer
to 3-3 and 3-4 respectively (cfr. 2-2 and 2-3 respectively). During the load flow
calculation, the Jacobean matrix (as defined in 3-7 and 3-8) is increased with extra rows
and columns to cope with the inclusion of the VSC HVDC link(s). The dimensions of the
Jacobean increase as:
[ # buses x # buses ] [ (# buses + 4.# HVDC links) x (# buses + 4.# HVDC links) ]
The extended Jacobean matrix for one VSC HVDC link in a 3 bus network looks like this:
. . . . . . . . . . . .
. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .
. . . . . . . . . . . .
= .
. . . . . . . . . . . .
P . . . . . . . . . . c1
rect
Qrect . . . . . . . . . . U c1
P . . . . . . . . . . c 2
inv
Qinv . . . . . . . . . . U c 2
There are 2 mismatch equations for every converter station.

Load flow calculations and VSC HVDC programming are performed in Matpower [37].
The following data is added using one or more VSC HVDC links (each row of the matrix
hvdc contains the data of one VSC HVDC link):
HVDCsend: shunt converters sending bus (rectifier)
HVDCrec: shunt converters receiving bus (inverter)
Rd: DC cable resistance of one cable of the VSC HVDC link [] (zero for BTB)
Prect: specified active power flow over VSC HVDC link [MW]
VvrHi: upper limit for voltage source magnitude (1.1 pu) [pu]
VvrLo: lower limit for voltage source magnitude (0.9 pu) [pu]
Sending and receiving bus each need the following set of information:
Xconv: combined inductive reactance of transformer and phase reactor [pu]
TarVol: target nodal voltage magnitude [pu]
VSta: control status for nodal voltage magnitude: 1 is on, 0 is off
Qrect/Qinv: specified reactive power flow [MVAr]
QSta: control status for specified reactive power flow: 1 is on, 0 is off
Vvr: initial condition for the shunt source voltage magnitude (1 pu) [pu]
Tvr: initial condition for the shunt source phase angle (0 rad) [rad]
As can be seen, the user can choose between reactive power flow control (QSta) and
AC voltage control (VSta).
Each converter is characterised by an upper and a lower voltage limit for the VSC. This
has an impact on the reactive power capability of the VSC HVDC link. When voltage
limits are reached, reactive power injection/absorption by the VSC HVDC has to be
decreased depending on the situation.
The capability chart found in [5] is used in the load flow calculation.
3.3 Outlook of Load Flow Calculation Incorporating VSC HVDC
A flowchart of the load flow program that is extended to cope with the inclusion of one or
more VSC HVDC links is given in Figure 3-4. A detailed description of the Matlab code
is available in Enclosure 13.2.
Figure 3-4: Flowchart of load flow program
Blocks that are part of a normal AC load flow calculation are white. Coloured blocks deal
with the inclusion of VSC HVDC in the AC network. The Newton-Raphson iteration is
stopped when convergence is reached or when the maximum number of iterations is
reached. A convergence tolerance of 10-8 on a per unit power mismatch is used and a
maximum number of iterations of 100.
The function to solve a load flow is the function runpf(casename) with casename the
data file of the network that has to be solved. A data file consists of the following
matrices:
bus: specifies bus number, type of bus and active and reactive load at each bus.
gen: specifies generator active power, reactive power limits and generator
voltage.
branch: specifies resistance, inductance and susceptance of the branches.
The data file has to be extended with a matrix including the VSC HVDC parameters as
described in paragraph 3.2. The matrix hvdc has the following structure:
HVDCsend HVDCrec Rd Prect VvrHi VvrLo Xconv1 TarVol1 VSta1 Qrect QSta1 Vvr1 Tvr1 Xconv2 TarVol2 VSta2 Qinv QSta2 Vvr2 Tvr2
with indices 1 and 2 representing rectifier and inverter respectively.

In the function runpf(), the first calculations that are performed are:
Calculating the losses of the VSC HVDC link. For a specified VSC HVDC
configuration, the losses of the converter station are calculated based on the loss
model of Sdra Lnken provided by ABB Corporate Research Sweden (see
paragraph 5.3). The converter station losses in this model are calculated based
on the converters active and reactive power exchange with the AC grid:
Ploss rect = f ( Prect , Qrect )
Ploss inv = f ( Pinv , Qinv )
Losses of the rectifier are calculated using Prect and Qrect at the converter
terminals. Losses of the inverter are calculated differently because Pinv is not
known in advance; we do not know the inverter loss unless we first know Pinv. In
a first iteration step, the efficiency of the inverter is assumed to be the same as
the efficiency of the rectifier. Pinv can then be calculated. With this value of Pinv
and the known value of Qinv, the inverter losses can be determined. With this
new value of inverter losses, a new Pinv can be obtained. The losses are then
again calculated using Pinv and Qinv. Total converter losses consist of rectifier
losses, DC cable losses and inverter losses.
Checking the limits on the capability chart. It is checked here if the operating
point of the VSC HVDC link is within the capability chart limits as defined in [5].
It has to be noticed that the reactive powers used in the previous two functions are the
powers corresponding to Qrect and Qinv. If the user desires AC voltage control instead of
reactive power control, these values are not the exact values. Exact reactive powers are
then determined by the load flow calculation. The user is advised to give a good
approximation of the reactive powers needed to maintain the voltage of the AC bus to a
desired value. However, exact losses and capability chart fulfilments are calculated later
in the load flow calculation with the exact reactive powers.
The calculations that follow are part of a loop. This loop deals with the reactive power
limits of the generators. Calculations are performed and when it is noticed at the end of
the calculations that one or more generators have reached their reactive power limit, this
is adjusted in the matrices bus and gen. The corresponding generator(s) is (are)
changed from a PV-bus to a PQ-bus.
Following steps are:
Building the admittance matrix using the branch properties.
Calculating the specified bus powers using the generation and load powers.
In the following calculation step, the Newton-Raphson algorithm is performed. Outputs
of the function HVDC_newtonpf() are the voltages of the buses in the AC network, the
voltages of the VSC HVDC converters and the number of iterations. The function
HVDC_newtonpf() is treated in detail here.
3.3.1 The Function HVDC_newtonpf()
Power flows in the network are calculated using the initial voltages and the admittance
matrix. From 3-1 and 3-2 it is clear that:
S = U .( Y . U )* 3-13
These power flows are updated with the power flows between the VSC terminals and the
AC network. The VSC HVDC power flows are calculated according to 2-2 and 2-3. The
difference between specified powers and calculated powers is then verified. Active and
reactive power mismatches for the slack bus and reactive power mismatches for the PV-
generators are put equal to zero. The mismatch vector is then updated with the power
mismatch equations for the VSC converter terminals. There are 2 extra equations for
each converter terminal, thus 4 extra equations for one VSC HVDC link (Prect, Qrect,
Pinv and Qinv). The specified active power for the rectifier is Prect and the specified
active power for the inverter is Prect minus the losses of the 2 converter stations and the
DC link.
In the following step, it is checked whether convergence is reached. If the infinity norm
of the power mismatches on a per unit base is smaller than 10-8, convergence is
reached.
If convergence is not reached, a loop is entered. This loop is the actual Newton-
Raphson algorithm. The loop can only be abandoned if convergence is reached or if the
maximum number of iterations is reached (100 iterations). First of all, the Jacobean
matrix is constructed. The dimensions for the Jacobean matrix are:
size([ J ]) = [ # buses x # buses ]
In a following step, the active and reactive power equations for the slack bus and the
reactive power equations for the PV-buses are removed. This is because the slack bus
is completely specified (U and ) and the voltage of the PV-buses is specified.
The Jacobean is then updated with the VSC HVDC mismatches. Dimensions increase
as
[ # buses x # buses ] [ (# buses + 4.# HVDC links) x (# buses + 4.# HVDC links)]
The derivatives of VSC HVDC that are added to the Jacobean can be found in [32]. It is
then checked whether one or more converter terminals is connected to a slack or PV-
bus. If this is the case, the associated mismatch equations have to be removed.
From the Jacobean matrix, the voltage amplitude and voltage angle mismatches are
obtained using formulas 3-7 and 3-8.
Subsequently, voltages of the AC buses and voltages of the VSC converters are
updated. The limits of the VSC converters are checked. If the upper or lower voltage
limit is reached, the converter voltage is maintained to this limit.
From these new voltage values, calculated powers in the network can be obtained. All
the previous calculations are run again until convergence is reached. The function
HVDC_newtonpf() is then abandoned.
After the Newton-Raphson algorithm, converter voltage limits are checked. If

convergence was not reached and if the rectifier or inverter voltage is equal to the upper
limit, the reactive power injected by the rectifier or inverter has to be decreased. This is
comparable to the principle of a generator when a PV- bus is changed to a PQ-bus. If
the converter voltage becomes too high, it is maintained at its upper limit but
consequently the reactive power injected by the converter has to be decreased. If the
VSC converter is in AC voltage control operating mode, the voltage at the AC bus
connected to the VSC converter can not be maintained because the injected reactive
power from the VSC decreases. With the new value of injected reactive power, losses
are calculated again and the Newton-Raphson algorithm is run again. The same is
applied if converter voltages are at their lower limit. In that case, the absorbed reactive
power by the VSC converter has to be decreased.
The last step in the load flow calculation is to obtain the power flows in the network.
Powers between the buses in the AC network are calculated using 3-13. Active and
reactive power between a converter and the associated AC bus are calculated according
to 2-2 and 2-3.
The load flow calculation thus far is performed with the values of Qrect and Qinv that the
user inserted. However, it is also possible that one or both converters operate in AC
voltage control mode. The previous load flow calculation solved how much reactive
power needs to be injected by the converters to maintain that voltage. However, the
losses of the converters need to be calculated again with the correct values of
injected/absorbed reactive power. If it is noticed that the user desires AC voltage
control, the load flow is run again with the correct values of converter terminal reactive
power. This results in a correct value of converter losses and a correct value of the
working points in the capability chart.
The final step in the load flow calculation is the control of generator reactive power limits.
If reactive power limits of generators are violated, the associated PV-bus is changed to a
PQ-bus and the whole load flow is run again.
3.4 Computational Complexity of Extended Load Flow Program
As an example, the load flow calculation is performed on the simple 5 bus network as
depicted in Figure 3-5. This 345 kV network in is extracted from [32]. Data of this simple
5 bus network can be found in Table 3-1. Solving this original network takes 3 iterations.
A VSC HVDC link replaces the AC line between buses 3 and 4 and it tries to maintain
the voltage at these buses to a value of 1 pu. A VSC HVDC link of 190 MVA and 150
kVDC is chosen (HVDC Light M4, see chapter 4) with a length of the DC cable of 75 km.
The active power flow over the VSC HVDC link is varied between 0 MW and 180 MW.
In this way, capability chart limits stay fulfilled during the simulations. The number of
iterations needed is illustrated in Figure 3-6. When the active power flow over the VSC
HVDC link reaches an excessive value, it is noticed that the number of iterations
increases significantly.
42 40 19 19
1 3 4
24
28 7
89
28
24
87 7
55 53
2 5
Figure 3-5: Original 5 bus network (active power flows in MW)
generated desired active load reactive load voltage after

bus type
power [MW] voltage [pu] [MW] [MVAr] load flow [pu]
1 slack - 1.06 - - 1.06
2 PV 40 1.00 20 10 1.00
3 PQ - - 45 15 0.987
4 PQ - - 40 5 0.984
5 PQ - - 60 10 0.972
Table 3-1: Properties of 5 bus network
40
35
30
25
# of iterations [-]
20
15
10
0
0 20 40 60 80 100 120 140 160 180
Power through VSC HVDC link [MW]
Figure 3-6: Number of iterations that the extended load flow calculation needs
The voltages at buses 3 and 4 during the simulations are plotted in Figure 3-7. To
maintain the voltage at bus 3 at a value of 1 pu when active power over the DC link is
increasing, the reactive power injected by the rectifier has to increase. After 106 MW of
active power is sent through the DC link, it becomes impossible for the rectifier to
maintain the voltage at bus 3 constant: the converter is not able to inject enough reactive
power. The converter is at its upper voltage limit of 1.1 pu. The rectifier tries to inject a
maximum amount of reactive power to hold the voltage at bus 3 as high as possible.
The voltage at bus 3 keeps dropping though.
The voltage at bus 4 can be held to a value of 1 pu during the simulations. When the
active power sent over the DC link is low, reactive power needs to be injected by the
inverter. If more than 77 MW of power is sent over the DC link, the inverter starts to
absorb reactive power.
From the results it is clear that the behavior of the VSC HVDC model is as desired.
Figure 3-7: Voltage profile and reactive power injection at bus 3 (upper) and at bus 4
(lower) in function of power through the DC link
A VSC HVDC model has been included in the load flow calculations. The working
principles of the model are described in detail. The VSC HVDC model can operate in
one of two modes: either AC voltage control mode or reactive power control mode.
Capability chart limits, converter voltage limits and losses of the converter stations and
DC cable are included. The loss model used is the Sdra Lnken model, which is
analyzed in chapter 5. The computational complexity of the extended load flow program
increases in comparison to the original load flow program. The model has been tested
on a five bus network and results are as expected.
4 INFLUENCE OF VSC HVDC ON VOLTAGE STABILITY
Including a VSC HVDC link in parallel to an AC line in a network has a positive effect on
the voltage profiles of the buses connected to the VSC HVDC link. First, some aspects
on voltage stability are treated. In a following paragraph, results obtained by ABB are
investigated. In this case, two buses are connected via a VSC HVDC link in parallel with
an AC line. An analytical derivation of this simple network leads to some interesting
conclusions. In a final paragraph, the effect of VSC HVDC on the voltage profiles in a
small meshed network is investigated. It is checked if results are comparable to the
simple ABB test case.
4.1 Aspects on Voltage Stability
A power system may suffer from rotor angle, frequency or voltage instability. Steady-
state voltage stability is described here. Electrical power networks worldwide suffer from
an increased loading profile. This load increase results in a decrease of the bus
voltages in the network. Some basic principles of voltage instability are outlined in [38].
The following simple example extracted from this reference explains the basic principles
of voltage instability. In Figure 4-1 a source (with voltage E ) is feeding a load (with
voltage U ) via a transmission line with reactance X. The load is represented by an
active power P and a reactive power Q; both powers are absorbed by the load.
Figure 4-1: Two-bus system
Active and reactive powers are given by:

E.U
P= sin 4-1
X
U 2 E.U
Q= + cos 4-2
X X
Solving 4-1 and 4-2 with respect to U and leads to:
E2 E4
U= QX X 2 P 2 XE 2Q 4-3
2 4
From the formula it follows that for certain active and reactive powers P and Q
consumed by the load, there are two solutions for U. U as a function of P and Q is
illustrated in Figure 4-2. The variables U, P and Q are made dimensionless in the figure.
The parameter defining the nose curves in the figure is tan=Q/P.
Figure 4-2: Load voltage in function of active and reactive power
A negative tan corresponds to a load injecting reactive power. From Figure 4-2 a
graph of the voltage U in function of the active power P can be extracted with tan as a
parameter. This is represented in Figure 4-3.
Figure 4-3: Nose curves
From 4-3 it follows that for a certain P and Q, there are two solutions for U. This is
illustrated in Figure 4-3. When the active power consumed by the load increases (with
constant tan), the two solutions for U move closer to each other. At the point of
maximum loading (the nose point or Point of Maximum Loadability PML), the two
solutions coalesce. This point is called the steady-state voltage stability limit. In an
electrical network, for a certain load P, the operating point is always the upper point of
the PV-characteristic. The lowest voltage has a higher current for the same power and
thus higher losses. As a consequence, this lower operating point is avoided. It is clear
from Figure 4-3 that a better voltage profile exists for a load injecting reactive power.
This can be accomplished by capacitor banks installed at the load. However there is a
drawback in this case: too much reactive power injection at the load brings the PML
U
closer to normal operating values ( 1 ).
E
Before moving on, some definitions are in place.
The definition of voltage stability given by the IEEE [39] is the ability of a system to
maintain voltage so that when load admittance is increased, load power will increase,
and so that both power and voltage are controllable.
Also according to the IEEE, voltage instability is defined when a disturbance, increase in
load, or system changes causes voltage to drop quickly or drift downward, and operators
and automatic system controls fail to halt the decay. The voltage decay may take just a
few seconds or ten to twenty minutes. If the decay continues unabated, voltage collapse
will occur.
Voltage collapse is defined as the process by which voltage instability leads to the loss
of voltage in a significant part of the system.
The disturbance can increase X and/or decrease E in Figure 4-1. Steady-state load
increase is further investigated in this chapter.
For certain tan (lagging) and certain load admittance, an operating point on the upper
part of the PV-curve is established. When increasing load admittance, the operating
point shifts to the right: increased load power but lower load voltage. When the PML is
reached, a further increase in load admittance lowers the power delivered to the load
and the load voltage keeps dropping. This can be interpreted as a voltage collapse. An
animation illustrating the previous analysis is presented in [40].
The load behavior plays an important role in voltage stability analysis. Static voltage-
power relationships are represented by a constant load (voltage-independent), a
polynomial or an exponential load model. In the load flow calculations that follow,
voltage-independent loads are considered.
A power system can be modeled by the following equations [38], [41], [42]:
x = f ( x, y , ) 4-4
0 = g ( x, y , ) 4-5
where x are the state variables, y the algebraic variables (voltage amplitudes and
phases), f the differential equations and g the algebraic equations for the active and the
reactive power balances at each bus. is a loading factor. The generator and load
powers are increased according to:
PGi = PGi 0 (1 + .kGi ) 4-6
for a generator bus and

PLi = PLi 0 (1 + .k Li ) 4-7
QLi = QLi 0 (1 + .k Li ) 4-8
for a load bus. PGi0, PLi0 and QLi0 are the initial powers. kGi and kLi are multipliers to
designate the rate of change in generation and load respectively at bus i as changes.
The Jacobean of 4-4 and 4-5 is
f f
x y
4-9
g g
x y

Bifurcation points arise when a small variation in parameters results in a change of the
qualitative structure of the system. A Saddle-Node Bifurcation SNB is when two
equilibria coalesce and then disappear. At this point, the Jacobean in 4-9 is singular.
Previous analysis can be extended to the Jacobean of the load flow calculation:
P P
P U
Q = Q .
Q U
4-10

U
When this Jacobean becomes singular, a SNB is reached. When voltage-independent

loads are used, this SNB also coincides with the PML [41]. This condition of singularity
can be monitored through the determinant, the smallest singular value or the eigenvalue
closest to the origin.
When using a load flow program as Matpower for bifurcation analysis, a problem
arises. The Jacobean becomes singular near the PML. Methods to get around the PML
in a load flow calculation are called continuation methods. In [43], Matpower is used
for load flow calculations and PV-curves are implemented. The continuation power flow
method is described in [44]. A Continuation Power Flow CPF method described in [45]
is treated here.
The power flow equations are as in 4-6, 4-7 and 4-8. The CPF starts from a known
solution and uses a predictor-corrector method to find solutions at higher load levels.
The principle is illustrated in Figure 4-4. The predictor calculates the tangent vector and
chooses a step size. In a following step, the corrector corrects the approximate solution
obtained by the predictor. To check if the critical point has been passed, the sign of the
tangent component of (d) should be observed. Before the PML d is positive, at the
PML d is zero and after the PML d is negative.
As a result, the PV-curve can be obtained and the singularity of the load flow Jacobean
is avoided. In Figure 4-4, the critical point corresponds to the PML.
Figure 4-4: The predictor-corrector scheme used in the continuation power flow
For the TSO, the knowledge of the location of the PML is important. For a certain
loading profile on his network, the TSO wants to know how far away from the PML his
network is. Many indices to detect the proximity to the static voltage collapse point have
been invented. An overview of indices based on the load flow Jacobean is given in [46].
In paragraph 4.3, a voltage stability index to be used in small meshed networks is
explained in detail.
4.2 Voltage Stability Improvement with VSC HVDC: ABB Test Case
The effect of a VSC HVDC link on the PV-curve of a two-bus network has been
investigated by ABB [47], [16]. The main results are presented here. The two-bus
network with its parameters is presented in Figure 4-5.
Figure 4-5: Two-bus network used in voltage stability analysis
The network consists of an AC line in parallel with a VSC HVDC link. A generator is
feeding the network over a reactance Xt. The load is an active load PL. Both nodes are
characterized by a line susceptance B. The parameters defining the network together
with the equations and the assumptions are presented in Enclosure 13.3. The current
rating of the converter is given by Iconv.
ABB describes three different control methods to operate the VSC HVDC link in this
situation:
P-control: as much active power as possible is transferred over the DC line.
Transferred active power is given by:
Pdc = I conv .min(U1 ,U 2 ) 4-11
Reactive power injected at both ends of the VSC HVDC link is zero.
Q-control: A maximum of reactive power is injected at both terminals. Reactive
power injection is given by:
Q1dc = I conv .U1 4-12
Q2 dc = I conv .U 2 4-13
The sign indicates that reactive power is injected in the AC network. Active
power transferred over the DC link is zero. This mode of operation corresponds
to the inclusion of a STATCOM at each end of the AC line.
Maximum converter voltage limits are not taken into account in this analysis.
These limits may put further constraints. Only the maximum MVA circle, as
defined in the capability chart, is taken into account.
In the simulations, when the bus voltage is higher than 1 pu, it is assumed that
the VSC can be operated to hold the voltage at a value of 1 pu. In this case, a
value of 1 pu is plotted in the PV-curve.
Mixed control: the optimum active power is calculated as:
Pdc = I conv .min(U1 ,U 2 ).sin(1 2 ) 4-14
and as a consequence the reactive power is calculated as:
Qidc = ( I conv .U i ) 2 Pdc 2 4-15
for i = 1, 2.
The sign is chosen so that reactive power is injected into the network.
In the mode of mixed control, the optimal active and reactive powers are derived as
illustrated in Figure 4-6. In the figure =1-2. In this plane, active and reactive power
flow equations are visualized. The blue circle arc represents the active and reactive
powers sent over the AC line. The active and reactive powers sent over the DC link
have to be added. This is represented by the green MVA circle of the inverter (cfr.
capability chart), which is added to the radius of the blue circle representing the receiving
end of the AC line. In the case of the full lines in the figure, a lot of active power is sent
over the DC link and little reactive power. However, as illustrated in the case of the
dotted lines, when less active power is sent over the DC link and thus more reactive
power, the overall transferred active power sent over the combination of the AC and the
DC line increases. In the optimal case, active and reactive powers sent over the DC line
are given by 4-14 and 4-15. Thus, in the mixed control mode, as much active power as
possible is sent over the combination of the AC and the DC line.
Figure 4-6: The power circle plane
The PV-curve for bus 2 is given in Figure 4-7.
1.2
1.1
Load Voltage [pu]
0.9
Q Control
0.8 P Control
Mixed control
0.7 No DC-link
0 0.2 0.4 0.6 0.8 1
Load Power [pu]
Figure 4-7: PV-curve and positive influence of VSC HVDC
As is clear from the figure, PV-curves have a better characteristic when a VSC HVDC
link is inserted. When the DC link operates in P-control, voltages are higher for the
same load as compared to network with only the AC line. The PML also shifts to the
right, which is a positive effect. When operating in Q-control or mixed control mode, the
VSC holds the voltage at bus 2 to a value of 1 pu. When it is not possible to inject
enough reactive power to hold the voltage at 1 pu, the voltage starts dropping. In these
modes of operation, the PML is also shifted to the right. The best result is obtained when
the VSC HVDC link operates in the mixed control mode.
4.3 Testing of the ABB case with own model of VSC HVDC
The simple system from Figure 4-5 is tested here with the VSC HVDC model
implemented in Matpower. Parameters of the network are chosen the same. The
parallel VSC HVDC link also has a rating of about 10 times smaller than the rating of the
network. A base MVA of 100 MVA is used, the rating of the VSC HVDC is thus 10 MVA.
Losses in the DC link are also assumed to be zero. The effect of converter voltage limits
however will be shown. They play an important role in the simulations.
The generator in this case is the slack bus. Bus 1 is a PQ bus with both active and
reactive power equal to 0; bus 2 is a PQ bus with varying active power and reactive
power equal to 0. Phase reactor and transformer reactor values are based on [14].
When operating in P-control mode, 10 MW is sent through the DC link and reactive
powers at both stations are 0. In Q-control mode, 10 MVAr is injected at both
converters. Mixed control is not possible in this case. In the ABB test case, bus 2 has a
voltage angle of 0 degrees and is the reference. As a consequence, there is one
unknown parameter less in the determination of the optimal active and reactive powers
as defined in 4-14 and 4-15. In this load flow calculation, only the slack bus can have a
reference voltage angle of 0 degrees.
The PV-curves can only be generated until the PML (until convergence is lost). This is in
contrast to the ABB test case. In that case, load admittance is increased and load can
be increased even after the PML has been reached. The result presented in Figure 4-8
is without converter voltage limits. Keeping the voltage at buses 1 and 2 to a value of 1
pu is not even possible, both converters have to absorb more than 10 MVAr.
1.35
1.3
1.25
1.2
Load Voltage [pu]
1.15
1.1
1.05 no VSC HVDC link

P-control
1 Q-control
0.95
0.9
0.85
0 0.2 0.4 0.6 0.8 1
Load Power [pu]
Figure 4-8: PV-curves of simple two-bus system with own VSC HVDC model
Results are comparable to Figure 4-7. With more realistic values of the converter lower
and upper voltage limit of 0.9 and 1.1 pu respectively, the VSC HVDC can not even
operate in P-control or Q-control. For example, at a load of 0.5 pu and no active power
through the DC link, the rectifier can only inject a maximum value of 9 MVAr, because
the inverter then already has to absorb a maximum of 10 MVAr. This illustrates the
importance of the converter voltage limits on the results. The example in the following
paragraph deals with converter voltage limits of 0.9 and 1.1 pu and an appropriate
control method is described that handles the converter voltage limits.
4.4 Voltage Stability Improvement with VSC HVDC: Small Meshed Networks
The simple example in the previous paragraph gives some insight on the positive
influence of a VSC HVDC link when operated in parallel with an AC line. A small
meshed grid is studied in this paragraph. A VSC HVDC is included to improve the
voltage stability of the network. Results should be comparable to the results described
in the previous paragraph. However, there are some differences in the simulation setup:
A full VSC HVDC link is included. Losses in the converter stations and in the DC
link are taken into account;
The AC lines have a resistance and associated losses;
Network loads are not purely active loads;
The VSC HVDC link does not work in Q-control mode. If the TSO desires
injection of reactive power to support the voltage, other means are to his
disposal: STATCOMs, SVCs;
DC-voltage limits are taken into account. An upper and lower converter voltage
limit of 1.1 pu and 0.9 pu respectively are implemented. These limits can never
be exceeded;
The VSC HVDC load flow model is implemented with voltage control. The
assumption made in the previous paragraph where the bus voltage was set to 1
pu even if the voltage was higher, is dropped. The exact amount of reactive
power needed to maintain a voltage at a desired value can be calculated;
The influence of the VSC HVDC link on buses not directly connected to the VSC
HVDC link is also investigated;
It is not the purpose to deliver as much active power as possible to the load. The
load in these simulations is fixed. Therefore, the mixed control mode of operation
has no meaning here.
The mode of operation here has some similarities with the mixed control mode
however. The exact mode of operation is explained later.
The generation of PV-curves is currently not implemented in Matpower. To generate
the (upper part of the) PV-curve of a load bus, the load is increased (with constant tan)
until the load flow Jacobean becomes singular. To give an idea of how far the system
state is from the PML, a voltage security indicator is implemented. The indicator is
based on the linearization of the meshed network to a two-bus network [49]. This is the
reason why a small network is studied. Larger networks need better indicators. Voltage
stability studies in larger networks are more precise using continuation power flow
methods for the generation of the PV-curves and using voltage stability indices
described in [46]. This is out of the scope of this study however.
The indicator is derived as follow. In the two-bus system, sending end powers (PS and
QS) and receiving end powers (PR and QR) are related by:
PS = PL + PR 4-16
QS = QL + QR 4-17
with PL and QL the line losses. With the sending end bus voltage being equal to 1 pu
and with R and X the resistance and reactance respectively of the line connecting
sending and receiving bus, an analytical derivation leads to two possible values for PS
(similarly for QS). Assuming a discriminant greater than zero in the solutions for PS (and
QS), this leads to the following condition for the voltage security indicator:
indicator = 4(( XPR RQR ) 2 + XQR + RPR ) 1 4-18
A complete derivation is presented in Enclosure 13.4. The indicator varies with

increasing load from 0 to a maximum value of 1. At the value of 1, the PML is reached.
The network studied is presented in Figure 3-5. Properties are summarized in Table 3-1.
PV-curves and associated voltage security indicators of load buses 3, 4 and 5 are
presented in Figure 4-9. During the load increase, for each MW load increase, the PV-
generator at bus 2 delivers 0.5 MW. The slack bus delivers the other 0.5 MW and the
losses in the network. During the generation of the PV-curves, generator limits are not
taken into account.
As can be seen in the figures, the indicator reaches a value of 1 at the PML. It is clear
from the figures that this small network will never operate near the PML, the voltages are
too low.
Subsequently, a VSC HVDC link is inserted between buses 3 and 4. The link is
operated in parallel with the original AC line. From the original power flows in the
network, it is clear that the (smallest) M1 HVDC Light configuration with a rating of 101
MVA and 80 kV is the best option. The voltages at buses 3 and 4 are tried to be
maintained at a value of 1 pu. From the capability chart in [14], it follows that the
maximum reactive power that can be injected by a VSC terminal is 0.5 pu of the HVDC
Light rating. The operating mode of the VSC HVDC link is as follow:
The voltage at the terminal buses is tried to be maintained at 1 pu;
If this is not possible because of upper converter voltage limits, a maximum
amount of reactive power is injected by the VSC to obtain the best possible
voltage profile. The limit of the maximum amount of injected power is 50 MVAr
for the M1 configuration;
This leaves a maximum of 85 MW for the active power transfer over the link. 85
MW is sent over the link in all cases in the following simulations.
The simulations are stopped when convergence is lost (a maximum of 100 iterations is
used).
PV-curves for buses 3, 4 and 5 are presented in Figure 4-10. The reactive powers
injected by the rectifier (at bus 3) and inverter (at bus 4) during the simulations are
presented in Enclosure 13.5. As stated earlier, it is clear that more reactive power can
be injected when the AC voltage is low. This is because reactive power exchange
between converter bus and AC bus depends on the difference in voltage amplitude
between the two buses.
Only the voltage at bus 4 can be maintained to a value of 1 pu when bus 4 is lightly
loaded. When the upper converter voltage limit of 1.1 pu is reached, as much reactive
power as possible is injected by the VSC HVDC converters.
The conclusion of the simulations is obvious: static voltage characteristics improve with
the inclusion of VSC HVDC. Using VSC HVDC, a higher voltage results for the same
load compared to the original case. The PML is also shifted to the right, which is a
positive effect. Results are thus comparable to the simple test case studied by ABB.
Future work could cover one or more of the following aspects:
Studying larger networks. A CPF method could be implemented in Matpower.
Different voltage stability indicators could be compared.
Other configurations of VSC HVDC in meshed networks. Insertion of a VSC
HVDC link a meshed grid is always some kind of parallel connection between the
AC lines and the VSC HVDC link. If the VSC HVDC link is not operated directly
in parallel with an AC line, the best operating mode could be searched for.
Investigating how far in a large network the positive effect of the VSC HVDC
link on voltage stability is visible.
The effects of VSC HVDC on the static voltage-power relationship have been
investigated. Results obtained in the small meshed grid are comparable to results
obtained in the simple ABB test case. For both cases, the same conclusions are valid:
Using VSC HVDC in parallel with an AC line, voltages are higher for the same
load;
The PML is shifted to the right.
Figure 4-9: PV-curves of load buses 3, 4 and 5

Figure 4-10: PV-curves with and without the inclusion of a VSC HVDC link
5 LOSSES OF VSC HVDC LINK
The VSC HVDC link is analyzed in detail. A description of every component is given. A
loss evaluation of every component is performed. The converter valves are the most
important component of a VSC HVDC substation. Losses in this component, both
switching and conduction losses, are treated in detail. The different converter valve
topologies and their impact on valve losses are also highlighted. The future looks bright
for loss reduction in valves. Some future developments in valve technology are
investigated. Paragraph 5.3 describes the Sdra Lnken loss model. The loss
behaviors of all the converter station components are combined in this model. The
Sdra Lnken loss model is used in the load flow model described in chapter 3. A final
paragraph studies the losses in the DC cable.
5.1 Outlook of VSC HVDC Link
One side (rectifier or inverter side) of a VSC HVDC link is depicted in Figure 5-1.
Figure 5-1: Outlook of one side of a VSC HVDC link
The losses of the link can be calculated as the sum of the losses of the different
components. The components of a VSC HVDC link are the following:
- Converter transformer
- AC harmonic filter
- Phase reactor
- High-frequency filter
- Valves consisting of IGBTs and anti-parallel diodes
- DC capacitor
- DC harmonic filter
- DC smoothing reactor
- Auxiliary equipment
The losses of each component are calculated separately. The total losses are the sum
of the separate losses of each component.
Determination of the losses of the different components is based on the available
standards. These standards apply to LCC HVDC stations however. As is stated in [4], it
is recommended that a new working group is established to formalise the evaluation of
power losses in VSC transmission.
5.2 Losses in Converter Station
5.2.1 Valve Losses
These losses amount to the greatest part of the total losses, so they are studied here in
detail. VSC HVDC uses IGBTs (Insulated Gate Bipolar Transistors). The IGBT has a
low forward voltage drop and can handle high currents. The IGBT has a voltage-
controlled capacitive gate, as in a MOSFET device [6]. HVDC Light uses a modular
concept for the installation of VSC HVDC [5]. The IGBTs are also of a modular design.
To increase the power handling, IGBT chips and diode chips are connected in parallel in
a sub-module. A StakPak IGBT has two, four or six sub-modules [50]. The number of
sub-modules is selected based on the current rating of the application. An overview of
the different HVDC Light modules is presented in Table 5-1.
Table 5-1: The HVDC Light modular concept
For further details concerning the different modules, see [5]. The valve losses are
studied in detail for the M5 configuration with a two-level converter. Some M5
configurations in existence use a three-level converter topology instead [11], [12].
However, the two-level converter describes the same basic principles concerning losses
and gives a good approximation of the losses occurring in a converter station [51]. For a
more detailed description of the losses in a two-level converter in comparison to the
losses in a three-level converter, the user is referred to [52] and [53]. Switching and
conduction losses are treated here. Losses in gate units and snubber circuits are
neglected. Gate units are used to turn the IGBT on and off and snubber circuits are
used to control the voltage over the device.
A linear model for semiconductor losses in voltage sourced converters based on IGBTs
is derived in [54] and in [55]. The losses are divided into switching losses as well as
conduction losses.
5.2.1.1 Switching Losses
The switching losses Pswitch of a two-level VSC with sinusoidal AC line current and with
IGBT switching devices are given by the following linearised equation [54], [55]:
1 U iL
Pswitch = 6.N IGBT valve . . f s .( EON , I + EOFF , I + EOFF , D ). IGBT . 5-1
U ref iref
where:
f s : the switching frequency
EON , I and EOFF , I : the turn-on and the turn-off energies of the IGBT respectively
EOFF , D : the turn-off energy of the diode due to reverse recovery charge current
U IGBT : the voltage over one IGBT; to obtain the voltage over one IGBT, the number of
series connected IGBTs has to be known (see formula 5-7). For the M5 configuration
with a DC voltage of 150 kV, a voltage over one IGBT of 0.965 kV is found.
U ref : reference voltage equal to the blocking state voltage of the IGBT occurring before
the corresponding commutation; see Table 5-2.
iL : the peak value of the AC line current assumed to be sinusoidal; with the data for the
M5 configuration in [5] (373 MVA rating) and in [14] (195 kV AC phase-phase), an iL of
1562 A is found.
iref : reference current equal to the on-state current after commutation; see Table 5-2.
The diode has negligible turn-on losses, since it turns on as soon as a forward voltage
appears. The turn-off loss due to diode recovery, however, is not negligible [4].
Obviously it follows from the formula that the switching loss is dependent on the
switching frequency and increases linearly with switching frequency.
The values of turn-on losses, turn-off losses, reference voltage and reference current are
provided by data sheets. For the HVDC Light M5 configuration, a switching frequency
of 23 times the fundamental frequency is used, so fs = 1150 Hz for a fundamental
frequency of 50 Hz [9]. Switching data for the 4 sub-module IGBT StakPak is found in
Table 5-2 (for a junction temperature of Tj = 125 C) [56].
Collector-Emitter voltage (max) UCE 2500 V
IGBT SSOA USSOA,I 1500 V
Diode SSOA USSOA,D 1500 V
Collector current IC 1300 A
Turn-off loss IGBT EOFF,I 2.15 J
Turn-on loss IGBT EON,I 1.75 J
Turn-off loss diode EOFF,D 1.00 J
Table 5-2: Switching data for the M5 configuration

In the table SSOA stands for Switching Safe Operating Area. The margin between the
maximum and the rated SSOA voltage is due to voltage spikes caused by diode reverse
recovery currents [51].
5.2.1.2 Conduction Losses
Conduction losses are derived from [55] and [57]. Generation of the desired AC output
voltage is achieved by comparing the desired reference waveform (modulating signal
with frequency fm) with a high-frequency triangular carrier wave (with frequency fc) as
depicted schematically in Figure 5-2.
Figure 5-2: Sinusoidal PWM: carrier waveform and desired waveform (upper part) and
the AC line-to-neutral voltage (lower part)
The modulation function M(t) in case of carrier based sinusoidal PWM [58] is a sine
wave. With the knowledge of the relevant modulation function the conduction loss
Pcond , I of a single semiconductor IGBT is expressed as:

U CE ,0 .iL 1 + M (t ) rCE .iL 2 1 + M (t )
Pcond , I = . sin(t ). .d t + . sin 2 (t ). .d t 5-2
2. 0
2 2. 0 2
where:
U CE ,0 : the IGBTs threshold voltage
rCE : the IGBTs differential resistance

It has to be noticed that the on-state voltage across the device consists of a constant
component and a component that depends linearly on current.
Similar equations hold for the conduction loss that appears in one diode:

U F ,0 .iL 1 M (t ) r .i 2 1 M (t )
Pcond , D = . sin(t ). .d t + F L . sin 2 (t ). .d t 5-3
2. 0
2 2. 0 2
where:
U F ,0 : the diodes threshold voltage
rF : the diodes differential resistance

For the sinusoidal PWM method these equations become:
U CE ,0 .iL M . r .i 2 2
Pcond , I = .(1 + .cos( )) + CE L .( + M . .cos( )) 5-4
2. 4 2. 4 3
U F ,0 .iL M . r .i 2 2
Pcond , D = .(1 .cos( )) + F L .( M . .cos( )) 5-5
2. 4 2. 4 3
where:
M: the modulation index; with the information for the M5 configuration obtained from [14]
(DC voltage of 150 kV and phase-phase AC voltage after converter of 195 kV), a
modulation index M of 1.06 is found. An approximation is made here, the data from [14]
U C peak
is not based on SPWM, the modulation index is defined here as M = .
Ud
: the displacement angle between the fundamental of the modulation function (voltage)
and the load current; in nominal operation mode the power factor of a VSC HVDC link is
a little under one [5].
For the two-level converter depicted in Figure 2-2 above the total conduction loss for the
6 valves becomes:
Pcond = 6.N IGBT valve .( Pcond , I + Pcond , D ) . 5-6
The values of U CE ,0 , rCE , U F ,0 and rF are provided by datasheets. Conduction data for
the 4 sub-module IGBT StakPak is shown in Table 5-3 (for a junction temperature of Tj =
125 C) [56]:
IGBT emitter threshold voltage UCE,0 1.14 V
IGBT slope resistance rCE 1.20 m
Diode threshold voltage UF,0 1.05 V
Diode slope resistance rF 0.65 m
Table 5-3: Conduction data for the M5 configuration
HVDC Light uses the StakPak IGBT [5]. To increase the power handling, six IGBT
chips and three diode chips are connected in parallel in a sub-module. A StakPak IGBT
has two, four or six sub-modules, which determine the current rating of the IGBT. To be
able to switch voltages higher than the rated voltage of one IGBT, several positions are
connected in series in each valve. In the two-level converter, each valve has to
withstand the voltage between the two DC terminals. A single valve for a 150 kV
module consists of around 300 series-connected IGBTs [5]. This is calculated as follow:
U d .U d max,99.9 .U d max, SSOA
N IGBT valve = 5-7
U IGBT , SSOA U miss distr
With series connected semiconductor devices, it is important that a single component

failure does not lead to a malfunction of the whole device. Therefore there are also
redundant IGBTs present [51]. Typical values for an M5 two-level configuration are
presented in Table 5-4 (provided by ABB Corporate Research Sweden):
Voltage factor Udmax,99.9 1.05
Voltage factor Udmax,SSOA 1.15
Voltage miss distribution Umiss distr 325 V
Redundant IGBTs 2
Table 5-4: Typical values to determine the number of series IGBTs in a valve
A three-phase two-level converter has 6 valves. For the M5 configuration, characterised

by a DC voltage of 150 kV and a current rating of 1140 A using 4 sub-modules, the total
number of IGBTs in one converter station is equal to 44 784 and the total number of
diodes is equal to 22 392 (311 StakPak IGBTs per valve). The previous numbers are
for one side of the DC link, either rectifier or inverter. With the previous numerical
values, the total valve losses can be found, as illustrated in Table 5-5.
total losses = Pswitch + Pcond 5-8
M5 1 valve 1 converter = 6 valves

Switching losses 431 kW 2.59 MW
Conduction losses 398 kW 2.39 MW
Total losses 4.98 MW
Table 5-5: Losses in converter valves
The result of 4.98 MW is 1.34% of the MVA rating of the VSC HVDC link, which is a very
reasonable result for the simplifications assumed. Hold in mind that this is only the
converter loss for one converter (either rectifier or inverter).
5.2.1.3 Two-Level versus Three-Level Converter Topology
In the formulas one has to cope with the different VSC HVDC converter topologies.
Formulas 5-1 and 5-6 are derived for a two-level converter as depicted in Figure 2-2.
Another topology used in VSC HVDC transmission is the three-level converter. Both
converter topologies and associated single phase AC voltage output are discussed in
paragraph 2.3. Concerning losses, there is also a difference between the two
topologies. A two-level converter for a high DC voltage needs a lot of series-connected
IGBTs. With the high switching frequencies used, switching losses are relatively high.
For the three-level topology in comparison to the two-level topology, each valve is only
switched half the number of times and at half the voltage. Consequently, the switching
losses for a three-level converter tend to be about half that of a two-level converter [4],
[53]. Conduction losses are also lower because of the lower forward-voltage drop [52].
The three-level topology generates a smoother voltage with fewer harmonics, lowering
losses in other components of the converter station.
With increasing level, the switching frequency of the valve drops, lowering switching
losses. The blocking voltage of the valve also decreases with increasing number of
levels. The valve blocking voltage for an n-level topology is the total DC voltage divided
by (n-1).
The evolution of converter losses is illustrated in Figure 5-3.
Figure 5-3: Evolution of converter losses
An example extracted from [4] illustrates this evolution:

switching converter
frequency [Hz] loss [%]
two-level VSC 1950 3
three-level VSC 1260 1.8
LCC HVDC 50 0.8
Table 5-6: Comparison VSC HVDC with LCC HVDC
The study performed in [53] for a 300 MW VSC HVDC gives the results listed in Table
5-7. The two-level topology is used as a reference.
capital cost losses converter size

two-level 1 1 1
three-level NPC 1.05 0.70 1.45
four-level FC 1.10 0.53 2.30
Table 5-7: Comparison of different converter topologies
5.2.1.4 Harmonic Spectra Generated by the VSC
An important parameter in the loss calculation of the other components of a VSC HVDC
link is the harmonic spectrum generated by the converter. The harmonic spectrum is
dependent on the switching frequency and the PWM pattern used (sinusoidal PWM,
sinusoidal PWM with 3rd harmonic injection or harmonic cancellation PWM, which is also
called optimal PWM) [5].
Main phase-to-neutral voltage harmonics generated by a two-level converter in
fc
comparison to a three-level converter are given in Table 5-8 [59] where mf = :
fm
harmonic number harmonic number

mf-2 2.mf-5
group 1 mf 2.mf-3
mf+2 2.mf-1
group 2
2.mf-5 2.mf+1
2.mf-3 2.mf+3
2.mf-1 2.mf+5
group 2
2.mf+1 4.mf-5
2.mf+3 4.mf-3
2.mf+5 4.mf-1
group 4
3.mf-2 4.mf+1
group 3 3.mf 4.mf+3
3.mf+2 4.mf+5
Table 5-8: Main phase-to-neutral voltage harmonics for a two- (left) and three-level
(right) converter
The neutral point is the midpoint of the DC capacitors.

If the triangular wave shape (the carrier) frequency is an odd integer multiple of the
fundamental frequency, the phase-to-neutral output voltage does not contain even order
harmonics. In a three-phase bridge circuit, all of the triple harmonics are eliminated in
the phase-to-phase voltages. Also, if the triangular wave shape frequency is a multiple
of 3, the harmonics of the order of the triangular wave shape frequency are cancelled in
the phase-to-phase and phase-to-neutral voltages (three-phase converter considered)
[4].
By increasing the switching frequency, the harmonic distortion at the AC system
connection is moved to higher orders, which are easier to filter [60]. However,
increasing the switching frequency also increases losses (see formula 5-1). Therefore,
an optimization has to take place in order to balance the harmonic level, on the one
hand, and the capital cost, power losses, footprint, etc., on the other [4].
Typical switching frequencies used in VSC HVDC found in literature are 21 times the
fundamental frequency [4], 27 times the fundamental frequency [61] and 40 times the
fundamental frequency [60].
A typical harmonic spectrum of the converter terminal phase-to-ground voltage for a
VSC operated with a sinusoidal PWM with 3rd harmonic injection and a switching
frequency of 21 times the fundamental frequency is depicted in Enclosure 13.6. 3rd
harmonic injection is when a third harmonic is added to the fundamental frequency
modulating signal to increase the power rating of the converter. A harmonic spectrum of
a converter using optimal PWM is also presented in Enclosure 13.6.
The spectrum consists of positive and negative sequence components and zero
sequence components. The power transformer prohibits the zero sequence components
from flowing into the AC network [5].
Losses in other VSC HVDC components are the consequence of harmonic currents.
Harmonic currents originate from the harmonic voltages coming from the converter and
the system impedance. Harmonic current spectra are very specific for each application
of VSC HVDC and as a consequence the best way to obtain harmonic current spectra is
to perform measurements after the commissioning of the VSC HVDC installation of

interest.
5.2.1.5 Future Developments in Valve Technology Concerning Losses
The last couple of years, VSC converter losses are decreasing and the perspective for
the future is that converter losses keep decreasing. A few factors are the reason for this
decrease:
New pulse width modulation technique;
New semi-conductor devices.
Instead of using the conventional sinusoidal PWM (carrier-modulated method) or
sinusoidal PWM with 3rd harmonic injection, some VSC HVDC converters use the
harmonic cancellation PWM. The Estlink HVDC Light project uses a two-level
converter topology with harmonic cancellation PWM [9]. This type of PWM is also called
optimal PWM (OPWM) by ABB or Selective Harmonic Elimination Modulation (SHEM)
[4]. With this method it is possible to eliminate specific harmonics. The SHEM-method
is based on pre-calculated switching angles. Sinusoidal carrier-modulated PWM is
illustrated in Figure 5-4. The upper part of the figure shows the triangular carrier wave
and the sinusoidal reference voltages. The lower part of the figure shows the phase-to-
neutral voltage for phase a. The switching angles are depicted as i.
Figure 5-4: Sinusoidal PWM; control signals (upper) and phase-to-neutral voltage (lower)
In this example [4], a switching frequency of 9 times the fundamental frequency results in
9 (2M+1) switching times per half cycle. With M number of independent switching
angles per quarter period, there are M degrees of freedom. One of these degrees of
freedom can be used to control the fundamental component, leaving the other M-1
degrees of freedom to eliminate M-1 selected harmonics. In Figure 5-5, M=2 and the
remaining degree of freedom can be used to eliminate one harmonic.
From the Estlink HVDC Light project it follows that the losses of a three-level sinusoidal
PWM configuration are the same as a two-level OPWM configuration.
Figure 5-5: Selective harmonic elimination PWM; control signals (upper three) and
phase-to-neutral voltage (lower)
Another aspect contributing to the decrease of converter losses in the future is the
development of a new PressPack IGBT [62]. This new generation significantly reduces
the conduction losses while maintaining low switching losses. The semiconductor
technology is based on the Soft-Punch-Through (SPT) concept, which reduces the
thickness of the device. This thickness reduction plays an important role in the reduction
of the losses of the IGBT. This SPT concept is combined with an enhanced planar cell
design and the overall technology is called SPT+. Conduction losses are reduced by
25%. Switching operations result in smoother waveforms with a short tail and are thus
also reduced. Diode reverse recovery currents are also softened, resulting in lower
losses. Another advantage is that this technology offers the possibility of operating the
device with significantly lower gate-resistance values than those required by
conventional technologies. This results in shorter delay times during device turn-off,
which not only lowers the turn-off losses but also improves the current sharing between
individual IGBT chips in the module. Besides lower losses, SPT+ technology achieves
the same extreme ruggedness during switching and short-circuit conditions and offers
the same low electromagnetic interference levels as current SPT technology.
Another promising development contributing to a decrease in valve losses is a new
semiconductor material instead of silicon: silicon carbide [63]. SiC has a breakdown
strength which is about ten times higher than the value for silicon, resulting in much
lower losses for SiC-based devices. However, SiC power semiconductors are not yet
available today because of manufacturing difficulties [64].
5.2.2 Converter Transformer Losses
There is no difference between a power transformer in an AC grid and a transformer for

HVDC applications. However, currents coming from a VSC substation contain
harmonics. These harmonics introduce extra losses in the converter transformer. Like
every transformer, the losses of a converter transformer consist of no-load losses and
load losses.
5.2.2.1 No-Load Losses
The no-load losses are also referred to as the iron or core or magnetizing losses. These
losses arise from the energy required to maintain the continuously varying magnetic flux
in the core. They are essentially invariant with the load on the transformer. For the
determination of the no-load losses, there is no difference between a VSC HVDC
converter transformer and an AC transformer. Typical no-load losses for a transformer
are 0.05% of rated MVA (data from Sdra Lnken HVDC Light project).
5.2.2.2 Load Losses
The load losses are also referred to as the copper or winding losses. The load losses in
a transformer are conventionally subdivided into loss as measured with DC (I2R loss)
and, in addition, eddy losses in windings and connections, and stray losses in
conductive structural parts of the transformer [65]. Stray losses occur due to the leakage
field of windings. The stray flux impinging on conducting parts gives rise to eddy
currents in them. Load losses for converter transformers are due to:
- the resistance of the windings;
- the harmonic content of the currents through the windings;
- eddy losses in the windings;
- stray losses in structural parts.
In normal service, the converter transformer load current is non-sinusoidal. Thus when
transformed into a Fourier series, it shows a considerable amount of harmonics. This
non-sinusoidal current raises the eddy loss and stray flux loss, and significantly
increases the total loss calculated or measured with purely sinusoidal current. The
actual service total load losses are calculated based on a given harmonic spectrum for
the load current. According to [65], eddy and stray losses are assumed to be
proportional to the square of the current. Following formulas are applied:
Pwinding eddy losses I 2 . f 2 5-9
Pstray losses I 2 . f 0,8 5-10
Based on a given harmonic spectrum, the total service load loss can be calculated as
follow [66], [67]:
2 2 2 0,8
100
I f 100
I f
Ptotal load loss = I LN 2
R + PWE1. n . n + PSE1. n . n 5-11
n =1 I1 f1 n =1 I1 f1
where:
n: harmonic number
100
I LN = I
n =1
n
2
: the rms value of the non-sinusoidal line current of the transformer
R: DC resistance of the windings

PWE1 : winding eddy losses at fundamental frequency
PSE1 : stray losses in structural parts at fundamental frequency
I1 : rms current at fundamental frequency
I n : rms current at harmonic n
Calculations evaluated until the 100th harmonic should be sufficient. It has to be noticed
that the converter transformer is located after the harmonic filter.
Load losses of VSC HVDC transformers are typically 0.25% (data from Sdra Lnken
HVDC Light project). With power transformers having efficiencies of 99.8% or higher
[68], the difference in load losses between a conventional AC transformer and a
transformer for VSC HVDC applications is about 0.10%.
5.2.2.3 Total Losses
The total losses are the sum of the magnetizing losses and the current-dependent (load)
losses.
The converter transformer used in the load-flow model of the M5 module of ABBs VSC
HVDC has the characteristics listed in Table 5-9 [14]:
AC grid voltage Converter side voltage MVA R X

400 kV 195 kV 396 MVA 0 pu 0.12 pu
Table 5-9: HVDC Light M5 converter transformer characteristics
An important remark is that converter transformers for VSC HVDC applications are
always equipped with a tap changer [5]. The tap changer allows the amplitude of the
filter bus voltage to be varied. This allows better control of active and reactive power.
Tap changing operations influence the filter bus voltage and as a consequence the
operating range in the capability chart is broadened, allowing a larger operating range of
active and reactive power.
The placing of the transformer in VSC HVDC applications differs from the placing of the
transformer in LCC HVDC applications. In an LCC HVDC substation, the converter
transformer is located immediately after the valves. To reduce harmonics in LCC HVDC,
a 12-pulse converter topology is often used. A 12-pulse thyristor bridge means that a
transformer with a tertiary winding is needed. With VSC HVDC, the converter
transformer is located after the harmonic filter. The VSC HVDC converter transformer is
thus less loaded with harmonics compared to the LCC converter transformer.
5.2.3 AC Harmonic Filter Losses
These filters are designed to divert the harmonic currents produced by the converter to
the ground. The VSC HVDC converter has a switching frequency of 2 kHz according to
[60]. That is 40 times faster compared to a phase-commutated converter operated at 50
Hz. Since there are only high frequency harmonics, shunt filtering is therefore relatively
small compared to the converter rating. It is sufficient to use a high pass-filter and no
tuned filters are needed. The filters consist of a capacitance C, an inductance L and a
resistance R, as illustrated in Figure 5-6.
Figure 5-6: High-pass filter used in VSC HVDC
The high-pass filter is a filter where the L and R elements are connected in parallel. This
connection results in a wide-band filter having an impedance at high frequencies limited
by the resistance. The capacity of the filter (MVAr) is about 15% of the rating of the VSC
HVDC link (MVA) [14].
The quality factor of the high-pass filter is given by the following formula [61]:
R
Q filter = 5-12
L.2 . f n
where f n is the harmonic frequency for which the filter is tuned.
In [67], the filter losses of the different components are defined as follow:
5.2.3.1 Capacitor Losses
The loss in a capacitor is due to its Equivalent Series Resistance (ESR). The filter
capacitor loss is given by:
U cap 2
Pcap = 5-13
ESR
where U cap is the fundamental frequency voltage across the capacitor.

Because of the low power factor of capacitors, the losses due to harmonic currents can
be considered to be very small and should be neglected [69].
5.2.3.2 Reactor Losses
The filter reactor loss should be calculated considering fundamental and harmonic
currents in the filter reactor. A reactor always consists of an inductance in series with a
resistance (ESR). The filter reactor loss is given by the following formula:
( I Ln )
2
100
. X Ln
Preactor = 5-14
n =1 QLn
where
n: harmonic number
I Ln : current through reactor at nth harmonic
X Ln : reactance at nth harmonic, X Ln = n. X L1
QLn : quality factor of reactor at nth harmonic
5.2.3.3 Resistor Losses
The filter resistor losses are simply given by:

100
Presistor = R. I Rn 2 5-15
n =1
where I Rn is the rms current of the nth harmonic through the resistor.
5.2.3.4 Total Losses
The total losses are given by the sum of the losses in the different components of the AC
filter.
An easier way to calculate all the losses in a three-phase harmonic filter at once is
presented in [61]. The loss of the filter is given by formula:
U LL 2 n 2 n
Pfilter = . 2 . 2 .Q filter 5-16
X C1 n 1 n 1
R
with Q filter =
L.2 . f n
where:
n: harmonic number
ULL: line-to-line voltage
XC1: capacitor reactance at fundamental frequency
XL1: reactor reactance at fundamental frequency

The quality factor for an AC harmonic filter used in VSC HVDC applications is typically
between 0.5 and 5 [70], although [61] uses a quality factor of 15.
5.2.4 High-Frequency Blocking Filter Losses
The PWM switching pattern of the valves results in a high dv/dt, up to 100 kV/s [53].
Therefore, a high-frequency filter is used to limit the transformers exposure to a high
dv/dt. A radio-interference (RI) filter capacitor is connected between the AC bus and
earth and an AC line filter reactor is installed. Determination of the losses in this filter is
analogous to the determination of the losses in the AC harmonic filter.
5.2.5 Phase Reactor Losses
A series reactor is necessary to separate the AC fundamental frequency from the raw
PWM waveform. The series reactors consist of air core devices. They are housed in an
aluminium enclosure to eliminate the high frequency disturbances from the PWM
process. Since the PWM pattern can be changed almost instantly, rapid control of both
active and reactive power can be achieved independently of each other. Changing the
fundamental frequency voltage phase angle across the series reactor controls the active
power, whereas, changing the fundamental frequency voltage magnitude across the
series reactor controls the reactive power. The phase reactor can be represented by a
reactance in series with a resistance (ESR). Power is dissipated in the resistance and
this accounts for the losses in the phase reactor. For the loss determination, 5-14
applies using the appropriate values for the phase reactor.
The value of the phase reactor in the load-flow model of the M5 module is XL1 = 0.16025
pu [14]. A typical quality factor of a phase reactor at fundamental frequency is QL1=200
(value provided by ABB Corporate Research Sweden). This gives an ESR of 8.0125 x
10-4 pu at fundamental frequency.
5.2.6 DC Capacitor Losses
The DC capacitor serves as energy storage. The capacitor also reduces the harmonics
ripple on the direct voltage and determines the dynamics of the system. Losses in the
DC capacitor are given by:
Ud 2
PDC cap = 5-17
ESR
where:
U d : the DC side voltage over the capacitor
ESR: equivalent series resistance of capacitor
Losses due to the harmonic current in the capacitor are very small because of the low
power factor and shall be neglected.
5.2.7 DC Smoothing Reactor Losses
DC smoothing reactors are used in HVDC converter stations to reduce the ripple in the
direct current and to limit converter DC side overcurrents during transient conditions.
The current through the smoothing reactor is direct current with superimposed
harmonics, mainly characteristic harmonics. Small amounts of non-characteristic
harmonics may also occur, particularly when the AC system is unbalanced [69]. The
losses due to direct current are given by:
PDC smoothing reactor = RDC smoothing reactor I DC 2 5-18
The loss due to the harmonic content of the DC current is determined as in 5-14 using
the appropriate values for the DC smoothing reactor.
One should bear in mind that the DC smoothing reactor is located after the DC filter, in
case there is a DC filter.
5.2.8 DC Harmonic Filter Losses
The DC capacitor and the DC smoothing reactor already provide suppression of

harmonics. In some cases however it is necessary to include a DC harmonic filter. This
is the case if the DC cable route shares the same right of way or runs close by telephone
wires, railroad signalling wires or similar. In these cases there is a possibility of
exposure to harmonic interference from the cable. If additional filtering is required, the
remedy is to add a filter on the DC side, consisting of a single-tuned filter, as illustrated
in Figure 5-7.
Figure 5-7: Single-tuned filter used in VSC HVDC
DC filters are normally not used in VSC HVDC transmission links [71], so they are not
studied here in detail.
5.2.9 Auxiliary and Other Equipment Losses
The auxiliary system includes the auxiliary power, valve cooling, air conditioning system,
and fire protection system The total station auxiliary losses should be determined by
appropriate measurement. The total station auxiliary peak losses are approximately
0.09% of the converter station rating [69].
The losses caused by the remaining equipment, such as surge arresters, instrument
transformers, switchgear, etc., shall be neglected. They are negligible in comparison
with the main equipment losses discussed [67].
5.3 Converter Station Equivalent Loss Model
Previous analysis described the calculation methods for the losses of the different
components of a VSC HVDC converter substation. It is very difficult however to obtain
detailed data about those individual losses because they are very specific for each
application of VSC HVDC. Losses also vary significantly with the operating point of the
DC link (standby, 50% of full load, full load). A generalized loss model is used in the
following calculations. This loss model is used throughout the report. The loss model
takes every component of a VSC HVDC converter substation into account.
In this section the converter station equivalent loss model used in the load flow
calculation is described. The loss model is obtained from ABB Corporate Research
Sweden and is based on data from the Sdra Lnken HVDC Light link, which is rated
at 600 MW and 300 kV. The VSC HVDC link is supposed to operate at a power factor
of about 1. The three-phase losses consist of a constant part and a part dependent on
the current through the phase reactor Ir. Analyzing the losses of the Sdra Lnken
HVDC Light link, leads to the results in Table 5-10. The Sdra Lnken project is a
HVDC Light project that ABB has planned for the future.
Constant losses
Incl. filter, excl. transformer 5 MW 0.83%
Transformer load losses 1.26 MW 0.21%
Transformer no load losses 0.36 MW 0.06%
Current dependent losses Nominal condition
Linear 3 x 600V x Ir 0.30% (rect) / 0.28% (inv)
Square (rectifier) 3 x 0.66 x Ir 2 0.32%
Square (inverter) 3 x 1 x Ir 2 0.43%
Total nominal losses (rect) 10.34 MW 1.72%
Total nominal losses (inv) 10.86 MW 1.81%
Table 5-10: Analysis of the Sdra Lnken HVDC Light project losses
The nominal phase reactor currents are 990 A for the rectifier and 926 A for the inverter.
Thus, the total losses in nominal operating state are 10.34 MW for the rectifier and 10.86
MW for the inverter. The phase reactor current is determined by the active and reactive
power that is injected/absorbed by the converter and is calculated as follow:
Pconv 2 + Qconv 2
Ir = 5-19
3.U AC , phase phase
The converter station losses in function of the phase reactor current are illustrated in
Figure 5-8 for the rectifier station. The efficiency of the rectifier station is also depicted in
Figure 5-8. It is clear that it is more efficient to operate the VSC HVDC around its
nominal working point. When operating at a power that is too low, VSC HVDC is
sometimes disconnected from the grid because the losses are too high.
Calculation of the losses when the converters are in reactive power control mode is
straightforward. Calculating the converter losses when the converter is in the AC
voltage control operation mode is more difficult, because it is not known in advance how
much reactive power is injected/absorbed by the converter to hold the voltage of the
associated AC bus to a constant value. In a first calculation step, losses are calculated
with an initial value (guess) of the reactive power injected/absorbed. The load flow is
performed using this value. After the load flow calculation, it is checked whether one of
the converters is in AC voltage control mode. If this is the case, the load flow is run
again with the value of the reactive power obtained from the first load flow calculation.
Calculating the losses in the rectifier when the desired active and reactive powers are
known is straightforward. However, for the inverter, the converter losses and thus the
phase reactor current are not known in advance. An iterative method is therefore used
in the calculations. In the first iteration step, the same efficiency as for the rectifier is
used for the determination of the phase reactor current of the inverter station. With this
value of the phase reactor current, losses are calculated. This value of inverter losses is
used in a second iteration step to determine a more precise value of phase reactor
current and inverter losses.
11
Losses of converter station [MW]
10
6
0 100 200 300 400 500 600 700 800 900 1000
Phase reactor current [A]
100
Efficiency of converter station [%]
80
60
40
20
0
0 100 200 300 400 500 600 700 800 900 1000
Phase reactor current [A]
Figure 5-8: Losses (up) and efficiency (down) of the rectifier station of the Sdra Lnken
HVDC Light project in function of the phase reactor current
The reason why inverter losses are greater than rectifier losses is because the inverter
uses the IGBTs more frequently, while the rectifier uses the diodes more frequently.
IGBTs have a higher loss than diodes.
5.4 Losses of DC Cable
VSC HVDC has a bipole configuration; the DC circuit is not connected to ground.
Therefore, always two cables are needed. The cable losses depend on the resistance of
the DC cable used. The different configurations of the HVDC Light modules are based
on a DC voltage of 80 kV, 150 kV or 300 kV. The chosen DC voltages are in line
with the ABB High Voltage Cable product range, i.e. 80 kV, 150 kV and 300 kV [72].
HVDC Light uses underground XLPE cables. Both land cables and submarine cables
are possible [73]. The cables are operated in bipolar mode, one cable with positive
polarity and one cable with negative polarity. Additional data can be found in [5] and in
[72].
The losses in the cable for a bipole configuration are represented as:
Pcable = 2.Rcable .I DC 2 5-20
where Rcable is the cable resistance in one cable.
When comparing an AC overhead line/cable versus a DC cable, the losses of the DC

cable per km are smaller than the AC OHL/cable [20]. DC cables do not suffer from skin
or proximity effects. There are no dielectric losses and no metallic screen or armour
losses with DC cables. An AC transmission system also requires 3 lines in contrast to a
DC transmission system, which only needs two lines. However, when the losses of the
whole DC installation (incl. rectifier and inverter) are taken into account, the losses of the
VSC HVDC installation increase significantly. Compared to AC OHL/cable, DC cables
also have other advantages over long distances:
- no transposing of the cables is necessary;
- no reactive power compensation is necessary.
Reactive power compensation is needed because AC overhead lines consume reactive
power. AC cables on the other hand generate reactive power. In the dimensioning of
the AC cable one has to cope with the fact that they do not only transfer the desired
active current but also a capacitive charging current [5], [20]. The capacitive charging
current is proportional to the length and the voltage of the AC cable. DC cables dont
have a capacitive charging current, the whole capacity of the cable can be used to
transfer the desired active power.
All the previous arguments lead to the conclusion that there exists a break-even distance
between an AC overhead line and a VSC HVDC link where the losses of the VSC HVDC
link become smaller than the losses of a comparable AC overhead line [20]. This is
illustrated in Figure 5-9 [74]. In the figure, an AC overhead line is compared to LCC
HVDC and VSC HVDC. Because the losses of LCC HVDC are smaller than VSC
HVDC, LCC HVDC has a lower break-even distance based on losses than VSC HVDC.
Figure 5-9: Line losses as a function of the length for an 800 MW power flow in a 1000
MW AC, LCC HVDC and VSC HVDC system
Each component of a VSC HVDC substation has been highlighted. Determination of the
losses of each component is based on existing standards. The standards are based on
an LCC HVDC substation however. There is a need for future work in the field of loss
determination of VSC HVDC substations. Losses in the converter valves amount to the
highest share of total station losses. However, these losses will probably decrease in
the near future. A converter station loss model has been described. The converter
station loss model is based on ABBs HVDC Light Sdra Lnken project. In this model,
converter station losses vary with active and reactive power exchanged between the
converter bus and the associated AC bus. The load flow calculation described in
chapter 3 includes this loss model.
6 IMPACT OF VSC HVDC ON LOSSES IN MESHED NETWORKS
The power flow controlling capabilities of VSC HVDC can have a positive effect on the
(losses of the) surrounding network. Control of the active power has an influence on the
power flow in the other lines of the surrounding network. Control of the reactive power
has an influence on the voltage of the grid. Control of reactive power can also reduce
the reactive power flows in the network. Loss reduction with LCC HVDC has already
been verified with the Fenno-Skan LCC HVDC project. Loss minimization with VSC
HVDC is investigated in the following chapter. The basic principles that are the cause of
the reduction in network losses are described in this chapter.
6.1 Potential Advantage of VSC HVDC in Meshed Networks: Redistribution of Power
Inserting a VSC HVDC link in a network and controlling the active power through it has
an influence on the power flows in the neighborhood of the VSC HVDC link. The VSC
HVDC link can take over a part of the flows from the surrounding lines and lighten the
load on these lines. The VSC HVDC link can also prevent loop flows and as such
lighten the load on the lines affected by the loop flow. An example of a loop flow is when
a generator wants to deliver power to a load, but instead of directly supplying the load,
the power flows via other parts of the network, unnecessary loading the other
surrounding lines in the network. Information on the positive influence of FACTS devices
in networks and their optimal positioning can be found in [75]. Advantages of VSC
HVDC, and more general FACTS, on power flows in networks are:
- Prevention of loop flows.
- More economic operation of the grid. This operation mode allows more capacity of
more economic generators to be used instead of utilizing a lot of expensive
generation. When the cheap generators are connected to the load centre via a weak
link, a VSC HVDC installation can be used to replace that weak link (or installing
BTB VSC HVDC in the weak link) and force a lot of active power from the cheap
generation units to the load centre.
- Increasing power flow transfer capability. The VSC HVDC link can inject reactive
power at its terminals and thereby can reduce the reactive power flow on the
surrounding lines, thereby allowing more active power flow through them.
- Unloading a selected line. A VSC HVDC link placed in the neighborhood of heavily
loaded lines lightens the load on these lines. The VSC HVDC takes over a part of
the active power flow in that part of the network. However, one has to take the
following into account. The losses are given as Ploss = RI 2 . This means that lines
which are nearly fully loaded contribute more to system losses. Using a VSC HVDC
link, power can be shifted from congested lines to the HVDC link, possibly lowering
total system losses.
- Directing flows between regions. This is comparable to the prevention of loop flows.
In this situation, the power between two utilities is controlled by a VSC HVDC link.
This can be an important factor with the liberalization of the electricity market, where
power flows are driven by the market price, while power flows in reality are
determined by Kirchoffs laws. A VSC HVDC link can control the exchanged active
power and can prevent the wheeling of power through other utilities.
Although minimization of network losses is maybe not the most important concern of the
transmission system operator, VSC HVDC has an influence on the losses in the network.
Controlling the active power through a (VSC) HVDC link has a redistribution of power
flows in the surrounding network as a consequence [25], [26]. Loss reduction with VSC
HVDC has not been realized yet. However, simulations with a positive outcome on total
grid losses are performed in [76]. This project investigates a new investment in the
Dutch grid, either an AC overhead line or a VSC HVDC link. It is shown that if VSC
HVDC is operated to minimize total system losses, it is possible that the losses in the
network with a VSC HVDC link can be lower than when an AC overhead line is installed.
For further details and specific data used, see [76].
With the older LCC HVDC, total network loss reduction has been realized. Total grid
losses are considerably reduced in the Fenno-Skan LCC HVDC project (see Figure 6-1)
[77].
Figure 6-1: The Fenno-Skan LCC HVDC project

The main part of the Nordel synchronous power system (400 kV) forms a network which
starts in southern Finland, and then goes northwards through Finland and into Sweden
north of the Gulf of Bothnia. It then continues southwards through Sweden with
connections into Norway, and finally terminates on Zeeland in Denmark.
Figure 6-1 shows that a connection between south-western Finland and east central
Sweden would be a natural link in the system, since it reduces the electrical distance
between these two areas from 1500 km to 200 km. Such a link would form a link
between major generation and load areas in both countries. There are nuclear power
stations, Forsmark in Sweden and Olkiluoto in Finland, close to the terminals of the link.
The link is used for power exchanges in both directions based on long term power
contracts as well as temporary power deliveries. The link also transfers power from
northern Sweden through Finland to the Stockholm area. This mode of operation
reduces the total system losses by up to 40 MW. This is because the 400 kV lines from
northern Sweden are very heavily loaded during peak demand conditions and during
periods with large water flow in the Swedish rivers, while the northern part of the Finnish
400 kV grid is relatively lightly loaded in some operating conditions. A redistribution of
the power flow pattern therefore results in substantial loss reductions.
6.2 Advantage of VSC HVDC in Meshed Networks: Boosting the Voltage
Reactive power control is an advantage that VSC HVDC has over LCC HVDC. The
reactive power controllability of VSC HVDC allows automatic control of the AC voltage.
This makes it possible to operate the surrounding grid at higher voltages and as a
consequence losses in the surrounding network are reduced [26]. Higher voltages mean
lower currents and thus lower RI2 losses.
6.3 Proper Active Power Flow Control of the VSC HVDC Link
Sending too little or too much active power through the VSC HVDC link in a meshed
network causes a rise in total system losses. This is due to extra induced loop flows in
the network. The loop flows that arise here are different from the ones described earlier.
Loop flows arise here because the network can not cope with this high amount of active
power over the VSC HVDC link. To provide the 180 MW of active power at the rectifier
end, power flows back from the inverter end to the rectifier end. These loop flows load
the surrounding lines in the network. The following example illustrates this principle.
The network in Figure 6-2 is extracted from [32]. Data of this simple 5 bus network can
be found in Table 3-1. Figure 6-3 is when an M4 HVDC Light link (190 MVA, 150 kV
rating) is included between buses 1 and 2 with an excessive active power target of 180
MW. It is clear that new unwanted loop flows arise due to the bad active power flow
control of the VSC HVDC link. System losses also increase drastically: from 6.12 MW in
the original network to 18.88 MW in the network with VSC HVDC.
42 40 19 19
1 3 4
24
28 7
89
28
24
87 7
55 53
2 5
Figure 6-2: Original 5 bus network (active power flows in MW)
36 39 24 24
1 3 4
56
180 8
60
170 58 8
62
2 70 68 5
Figure 6-3: 5 bus network with VSC HVDC link between buses 1 and 2 and excessive
active power target of VSC HVDC link (active power flows in MW)
The advantages of VSC HVDC in meshed networks have been highlighted. Active and
reactive power flow controllability of VSC HVDC can reduce network losses. When an
AC line in a network is replaced by a VSC HVDC link, active power flow over the DC link
is not determined by Kirchoffs laws. Transferring the right amount of active power over
the DC link reduces the load on the surrounding AC lines and reduces losses in the
network. Simulations in the next chapter will try to find this right amount of active power
through the DC link. Sending too much or too little active power over the DC link
introduces loop flows in the network and associated higher network losses. Reactive
power injection by the VSC HVDC terminals decreases reactive power flows in the
network. More capacity is thus available for the transport of active power. With the
precise and fast reactive power control of VSC HVDC, the voltages in the network can
be maintained at higher voltages, reducing network losses.
7 RESULTS OF INCORPORATING VSC HVDC IN MESHED NETWORKS
Different simulation cases are treated in this chapter. In 7.1 a new link, either AC or
VSC HVDC, is introduced in the IEEE 14 bus network. The best result concerning loss
minimization is investigated. In paragraph 7.2 an AC line is replaced with a VSC HVDC
link. The optimal loading (optimal active and reactive powers) of the DC link is searched,
with the purpose of loss minimization. Paragraph 7.3 treats the inclusion of a VSC
HVDC link a larger network: the IEEE 118 bus network. AC lines are replaced by VSC
HVDC links with the goal to reduce total network losses. In a final paragraph, the load
on the networks is increased and the behavior of the VSC HVDC links under this
increased loading condition is investigated. In this chapter, whenever there is referred to
the optimal active power through the DC link, the active power entering the rectifier
converter is meant.
7.1 Inserting a New Link in a Meshed Network: AC versus VSC HVDC
Transmission of electricity is characterised by generation and load centres. Electric

power is transported from generators to users via transmission lines. The losses of the
overall network can be decreased by introducing new transmission lines in an existing
network. This new transmission line lightens the load on the old transmission lines and
helps to avoid congestion of the network. A network often used in power flow analysis is
the IEEE 14 bus system. Network data can be found in Enclosure 13.7. The IEEE 14
Bus Test Case represents a portion of the American Electric Power System (in the
Midwestern US). The lower part of the network is characterised by generation centres
while the upper part is characterised by load centres. Power flow data of the IEEE 14
bus system can be found in [78].
The network consists of five generators and three of them are synchronous
compensators. Synchronous compensators are PV generators with P equal to zero that
support the voltage of the grid by injecting reactive power. The result of the load flow of
the original network is shown in Figure 7-1 (only generation part of network). The line
flows in the network are the active powers leaving their corresponding buses.
The total line losses (RI2) of the original network are 13.39 MW. Most of the power from
the lower part of the network has to flow to the loads in the upper part. Unloading of the
existing transmission lines lowers the power transport through them and consequently
lowers associated losses. The power flow through the new transmission line introduces
new losses.
Figure 7-1: Power flows in original network (line flows in MW)
7.1.1 Inserting a New AC Line
An AC overhead line is introduced in the IEEE 14 bus network. The new line is inserted
between bus 1 in the generator zone and bus 6 in the load zone. The line is a 69 kV line
and a transformer is placed at the end of the line (69 kV 13.8 kV). Transformer losses
are however not taken into account in this simulation. Base units used are an apparent
power of 100 MVA and a voltage of 69 kV. The transformer has a reactance of 0.25 pu,
which is comparable to the transformers between buses 5-6 and buses 4-7. The AC
overhead transmission line has the following characteristics (Table 7-1):
From To Voltage R X Length R X
Bus 1 Bus 6 69 kV 0.055 pu 0.20 pu 20 km 2.62 9.52
Table 7-1: Properties of AC overhead line introduced in IEEE 14 bus system
A base apparent power of 100 MVA and a base voltage of 69 kV, give a base
Vb 2
impedance of Z b = equal to 47.61 . The values of R = 0.055 pu and X = 0.20
MVAb
pu are chosen in accordance to the other AC line parameters used in the IEEE 14 bus
system. With typical values for AC overhead transmission lines of R = 0.13-0.14 /km
and X = 0.48 /km (60 Hz) [79], this leads to a line length of approximately 20 km.
The result of the load flow including the new AC overhead line is shown in Figure 7-2.
The line flows in the network are the active powers leaving their corresponding buses.
Figure 7-2: Power flows in network with new AC overhead line (line flows in MW)
The generation part of the network is lightened due to the inclusion of the new AC
overhead line. The losses in the new network are 11.14 MW and are clearly diminished
in comparison to the original case. Installing new overhead lines is a very effective
solution for the transmission system operator. However, the AC line has no power flow
control capabilities and the distribution of the power flows is determined by Kirchoff's
law.
7.1.2 Inserting a New VSC HVDC Link
Subsequently, a VSC HVDC link is introduced into the network. The link is also inserted
between buses 1 and 6. VSC HVDC has an advantage in comparison to a simple AC
overhead line: the VSC HVDC link can hold the voltage at a bus to which it is connected
constant. Thus bus 6 can be changed from a PV-bus to a PQ-bus. The VSC HVDC link
takes over the function of the synchronous compensator and holds the voltage at bus 6
to a value of 1.07 pu. Thus the rectifier operates in the reactive power flow control
mode with a value of 0 MVAr, because the slack bus holds the voltage at bus 1 constant,
and the inverter operates in AC voltage control mode, holding the voltage at bus 6
constant. For the load flow calculation, the HVDC Light configuration M1 is used (101
MVA, 80 kV rating). The characteristics of M1 can be found in [5] and [14]. For a land
cable suitable for the M1 configuration, a resistance of 0.0605 /km is found [5]. For a
cable length of 20 km, this gives a total resistance of 1.21 . The active power flow
through the DC link is varied between 10 and 100 MW. During the simulation, the
steady-state working points always were inside the operating area defined by the
capability chart found in [5]. Converter upper and lower voltage limits used were 1.1 pu
and 0.9 pu respectively. Converter transformer and phase reactor values are found in
[14]. A transformation to the same base power as the rest of the network is necessary
however:
X phase reactor X transfo
X total 100 MVA = .100 MVA + .100 MVA
101 MVA 107 MVA
Results are illustrated in Figure 7-3. Minimization of total network losses takes place
when the VSC HVDC link is transferring an active power of 87 MW. The total network
losses are then equal to 12 MW. This is lower than the original base case but higher
than in the situation where an AC overhead line is inserted. A break-down of VSC
HVDC losses at the optimal active power of 87 MW over the DC link is as follow: 1.61
MW losses in the rectifier station, 1.70 MW losses in the inverter station and 0.34 MW
losses for both DC cables. This leads to a total VSC HVDC loss of 3.65 MW.
However, incorporating a VSC HVDC link the system has several advantages over a
simple AC line. First of all, the synchronous compensator at bus 6 can be omitted.
From Figure 7-3 it follows also that the AC network is loaded more lightly in comparison
to the situation with a new AC line when the power through the VSC HVDC link is more
than about 30 MW. This is also approximately the value at which it is more favourable to
insert a VSC HVDC link than keeping the original network, from a losses point of view.
16
14
12
Losses [MW]
10
8
VSC HVDC loss
AC network loss (no VSC HVDC loss)
6 Total active power loss
Base case
New AC line
4
2
10 20 30 40 50 60 70 80 90 100
Figure 7-3: Losses in the IEEE 14 bus network while varying active power through the
VSC HVDC link
7.1.3 Conclusion
Inserting a new VSC HVDC link has its advantages, but for a small meshed network like
the IEEE 14 bus system, total system losses are smaller with a new AC overhead line.
VSC HVDCs high converter losses are the main disadvantage concerning system
losses.
7.2 Replacing an AC Overhead Line in a Meshed Network by a VSC HVDC Cable
The transmission operator can also choose to replace an old line by a VSC HVDC link.
Notice in the IEEE 14 bus network that there is a heavy load of 94.20 MW in bus 3. To
supply this load, power flows from bus 2 to bus 3 but also from bus 4 to bus 3. As a
consequence, the AC lines between buses 2 and 4 and between 5 and 4 are loaded
extra to supply the load in bus 3. This loop flow can be avoided by installing a VSC
HVDC link between buses 2 and 3. The power to supply the load at bus 3 is then forced
to flow from lines 1-2 and lines 2-3. As a consequence the other lines in the network are
unloaded, possibly minimizing total system losses.
The result for a HVDC Light M1 configuration (101 MVA, 80 kV rating) is presented in
Figure 7-4 (upper). The same values for cable length, reactances and limits as in
paragraph 7.1 apply. The synchronous compensator at bus 3 is dropped and the
inverter tries to hold the voltage at bus 3 to a value of 1.01 pu.
A minimum of total network losses is reached at the maximum of 100 MW and amounts
to 13.49 MW. A break-down of the VSC HVDC losses at 100 MW gives the following
result: 1.73 MW of rectifier losses, 1.85 MW of inverter losses and 0.46 MW losses for
both DC cables. This amounts to a total of 4.04 MW for the VSC HVDC losses.
As a consequence, an M4 configuration of higher rating (190 MVA, 150 kV rating) is
studied. The resistance for this configuration is the same [5]. The reactance has to be
adjusted:
X phase reactor X transfo
X total 100 MVA = .100 MVA + .100 MVA
190 MVA 201 MVA
The result for a HVDC Light M4 configuration is presented in Figure 7-4 (lower).
As noticed with the M1 configuration, the losses reach a minimum when a power of
around 100 MW is sent through the DC cable, a loss minimum of 14.74 MW at a power
flow through the DC link of 111 MW to be precise. At this optimal power over the DC
link, the break-down of the VSC HVDC losses is as follow: 2.63 MW rectifier losses, 2.69
MW inverter losses and 0.16 MW losses for both DC cables. This leads to a total of 5.48
MW for the VSC HVDC losses. The M4 HVDC Light has a DC voltage of 150 kV in
contrast to the M1 HVDC Light which has a DC voltage of 80 kV. As a consequence,
the losses in the DC cable are much higher for the M1 configuration. The reason why
the losses around 100 MW are greater with the M4 configuration is because the VSC
HVDC losses are higher for the M4 rating of 190 MVA than the VSC HVDC losses for
the M1 rating of 101 MVA. The constant (no-load) losses of the VSC HVDC link have a
great impact on the results.
Figure 7-4: Network losses when replacing an AC line by an HVDC Light M1

configuration (upper) and by an HVDC Light M4 configuration (lower) in the IEEE 14
bus system
In the network under study, replacing an AC line by a VSC HVDC link is not really a
favorable adjustment. With the M1 configuration, the total grid losses are about the
same as in the case with the AC overhead line. But VSC HVDC has several advantages
over the original AC line, one of them being the fact that the synchronous compensator
at bus 3 can be omitted. The inverter of the VSC HVDC takes over the role of holding
the voltage at bus 3 constant to 1.01 pu.
To obtain more favorable results, a larger network has to be studied so that the influence
of the VSC HVDC link can affect a larger area of the surrounding AC network. The
effect of VSC HVDC on the New-England 39 bus system on total system losses has
been investigated in [74]. Results were however not in favor of VSC HVDC. In the
following paragraph, the IEEE 118 bus system is studied.
7.3 Effect of VSC HVDC on a Larger Network: the IEEE 118 Bus Network
The effect of VSC HVDC is now studied on a larger network. Because the converter
losses of a VSC HVDC link are considerable, influencing a larger part of a grid has a
greater probability to have a positive effect on total network losses. For the simulations,
the IEEE 118 bus network is used. The IEEE 118 Bus Test Case represents a portion of
the American Electric Power System (in the Midwestern US). Data of the IEEE 118 bus
network can be found in [78]. The network consists of 118 buses, 54 generators, 99
loads, 186 branches and 9 transformers. Total generation and load are about 4.3 GW
each. Total initial losses in the network are 132.86 MW. For the placement of the VSC
HVDC link, the following is taken into consideration:
- The link is inserted in a meshed part of the network to have an effect on surrounding
network parts. For example, it has no use inserting a link between an isolated
generator and the network. The same applies to a link between the network and an
isolated load.
- The link replaces a branch which plays an important role in the initial case: a branch
which transports a considerable amount of power. In the original case, this branch
also contributes to a large part of total system losses.
7.3.1 VSC HVDC Link between Buses 25 and 27
The original power flow over this branch is 144 MW. Losses in this branch amount to 6.4
MW. The original branch is a 138 kV branch of length 65 km. Bus 25 is a generator with
connections to buses 23, 26 and 27. Bus 27 is a synchronous compensator with
connections to buses 25, 28, 32 and 115. An M4 HVDC Light link is installed with
active power flow controlled from bus 25 to bus 27 and active power varied between 50
MW and 190 MW. During the simulations, capability chart limits and converter voltage
limits were controlled and fulfilled. The VSC HVDC tries to hold to voltage at bus 27 to a
value of 0.97 pu (desired value), the synchronous compensator is dropped. Results are
presented in Figure 7-5 (upper).
An optimum in total network losses in found when 186 MW is sent through the VSC
HVDC link. The minimum total network losses are then 132.37 MW. The M4
configuration seems a good choice in this case. Losses are somewhat lower than in the
original case and VSC HVDC also takes over the function of the synchronous
compensator at bus 27 to hold the voltage at that bus constant to a value of 0.97 pu. A
break-down of VSC HVDC losses at 186 MW through the link is as follow: 3.23 MW for
the rectifier, 3.44 MW for the inverter and 1.45 MW for both DC cables. This leads to a
total VSC HVDC loss of 8.12 MW. However, when the TSO considers an upgrade of his
transmission network, other parameters instead of loss minimization have to be taken
into account. One of them is the cost of VSC HVDC in comparison to other
technologies. That subject is out of the scope of this study however.
Figure 7-5: Total network losses while varying active power through an VSC HVDC link
between buses 25 and 27 (upper) and between buses 69 and 75 (lower) in the IEEE 118
bus system
7.3.2 VSC HVDC Link between Buses 69 and 75
The original power flow over this branch is 110 MW. Losses in this branch amount to 4.9
MW. The original branch is a 138 kV branch of length 50 km. Bus 69 is a generator
connected to buses 47, 49, 68, 70, 75 and 77. Bus 75 is a load bus connected to buses
69, 70, 74, 77 and 118. An M4 HVDC Light is installed with active power flow
controlled from bus 69 to bus 75 and active power varied between 50 MW and 190 MW.
The voltage at bus 75 is held constant to a value of 0.97 pu (desired value). During the
simulations, capability chart limits and converter voltage limits were controlled and
fulfilled. Results are presented in Figure 7-5 (lower).
An optimum in total network losses is found when 175 MW is sent through the VSC
HVDC link. The minimum total network losses are then 132.43 MW. At the optimal
power over the DC link of 175 MW, the losses in the VSC HVDC link are as follow: 3.13
MW rectifier losses, 3.32 MW inverter losses and 0.98 MW losses for both DC cables.
7.3.3 VSC HVDC Links between Buses 25-27 and 69-75 and Their Interaction
In the following simulation, there are 2 links inserted in the IEEE 118 bus network. One
link is placed between buses 25 and 27 and another one between buses 69 and 75.
The effect of both links on total system losses is studied. Both links can cooperate and
reduce total system losses or they can counteract each other and raise total system
losses. The system is not linear. The simulation is performed varying active power
through both VSC HVDC links from 50 MW to 190 MW. Results are presented in Figure
7-6.
148
146
144
Total network losses [MW]
142
140
138
136
134
132
130
Po 50
we
r th
ro ug
h VS 100
CH 50
VD
C link W]
100 75 [M
be
t we 150 es 69 &
en ee n bus
bu 150 k betw
ses DC li n
25 S C HV
&2 200 ugh V
200 r thro
7[
MW Powe
]
Figure 7-6: Effect of two VSC HVDC links on total system losses
From the figure it can be concluded that the two links work well together in the
neighborhood of their nominal operation condition (190 MVA for the M4 configuration).
An optimum in total network losses is achieved when 186 MW is sent through the link
between buses 25 and 27 and 174 MW is sent through the link between buses 69 and
75. The optimum is then a total network loss of 131.93 MW. The optimal power flows
over the VSC HVDC links stay approximately the same as in the case where only one
VSC HVDC link was inserted in the network. In this particular case, both VSC HVDC
links have little influence on each other when both links are working near nominal
operating condition. At optimal operating conditions, both VSC HVDC links have a
positive effect on the system and total network losses are smaller compared to the
individual situations.
A break-down of VSC HVDC losses in the optimal operating point is given in Table 7-2:
link optimum [MW] rectifier loss [MW] inverter loss [MW] DC cable loss [MW] total [MW]
25-27 186 3.23 3.44 1.45 8.12
69-75 174 3.12 3.31 0.97 7.40
Table 7-2: Break-down of VSC HVDC converter loss
It is clear that the influence of VSC HVDC on loss reduction in the IEEE 118 bus network
is very small.
7.3.4 Conclusion
To conclude this paragraph, the following remark is in place. The IEEE 14 bus and 118
bus systems are characterized by a great amount of voltage control devices
(synchronous compensators). They provide the reactive power necessary to hold the
voltage at the associated bus to a constant value. The network is therefore
characterized by a small amount of reactive power flow through the lines. A lot of the
line capacity is thus available for active power transfer. The advantages of VSC HVDC
on the other hand are greater in a network with fewer voltage control devices. VSC
HVDC then takes over the function of the synchronous compensators (if there are any)
in the neighborhood of the link and the DC link has more capacity for active power
transfer in comparison to the original AC line, which was transporting a considerable
amount of reactive power.
VSC HVDC has a greater influence on a heavily loaded network. This is also one of the
reasons why a transmission system operator wants to install VSC HVDC: to lighten the
load on his network. As a consequence, in the following simulations, previous networks
are loaded extra and then the effect of VSC HVDC on total system losses is greater. In
heavily loaded networks, the voltages are also lower. VSC HVDC can also solve this
problem and operate the grid at higher voltages, lowering total system losses.
7.4 Effect of VSC HVDC on Heavily Loaded Network
7.4.1 IEEE 14 Bus System
7.4.1.1 Original Network
Subsequently, the IEEE 14 bus network is studied under heavy load conditions.
Increasing system load causes an increase in active and reactive power flows in the
network. The voltages of the buses also decrease. The synchronous compensators in
the network have to inject more reactive power to hold the voltage of the associated bus
constant. However, each generator has a limited maximum reactive power that can be
injected. If this limit is reached, the generator injects this maximum amount of reactive
power, but then the voltage of the associated bus is no longer held constant and drops.
In a load flow calculation, this is equivalent to a PV-bus being changed into a PQ-bus. In
the IEEE 14 bus network, generator limits are listed in Table 7-3. In the previous
calculations, these limits were never reached.
Generator bus Max Q [MVAr] Min Q [MVAr]

Bus 1 100 -100
Bus 2 50 -40
Bus 3 40 0
Bus 6 24 -6
Bus 8 24 -6
Table 7-3: Generator limits in IEEE 14 bus system
Loading of the system causes generators to reach their reactive power limit. In this
scenario, when system load is increased, the active and reactive loads at each bus are
increased by a certain percentage of the original load at that bus. The active power of
PV-generators is also increased by the same percentage of the original active power of
that generator. When loading is increased to 30% of original system load, generators 2,
3, 6 and 8 have reached their reactive power limit and generator 1 is injecting 11 MVAr.
Total system losses amount then to a value of 23.86 MW.
Generator bus voltages and reactive powers at 30% load increase are shown in Table
7-4.
Generator bus Desired voltage [pu] Actual voltage [pu] Injected Q [MVAr]
Bus 1 1.06 1.06 11.35
Bus 2 1.045 1.029 50 (limit)
Bus 3 1.01 0.983 40 (limit)
Bus 6 1.07 1.043 24 (limit)
Bus 8 1.09 1.069 24 (limit)
Table 7-4: Voltages at generator buses at 30% system load increase
7.4.1.2 Including M4 VSC HVDC Link: No Load Increase
In the original IEEE 14 bus network it is noticed that the most important lines are the
lines connecting buses 1-2 and buses 1-5. Although bus 5 is closest to the load area,
most of the power is transmitted via the line connecting buses 1 and 2, twice as much
power to be precise. This is due to the high resistance of the line connecting buses 1
and 5 in comparison to the resistance of the line connecting buses 1 and 2. Replacing
line 1-5 by a VSC HVDC link and controlling power through the VSC HVDC links seems
a logical solution. Line 1-5 is a line of length 22 km. The VSC HVDC link is an M4
HVDC Light configuration and the inverter tries to hold the voltage at bus 5 constant to
a value of 1.02 pu (desired value). The effect of inserting a VSC HVDC link between
buses 1 and 5 on total network losses is illustrated in Figure 7-7 (upper). An optimum is
reached when 154 MW is sent over the VSC HVDC link. Total network losses are then
12.40 MW. Capability chart limits and converter voltage limits were fulfilled during the
simulation.
Figure 7-7: Losses in function of power through the VSC HVDC link at 0% system load
increase (upper) and at 30% system load increase (lower)
7.4.1.3 Including M4 VSC HVDC Link: 30% Load Increase
When increasing system load, the previous situation gets even worse. Not only line 1-2
gets overloaded but also the lines connecting bus 2 to the load part of the network. The
voltage at bus 5 also drops drastically when increasing system load. At 30% load
increase, the voltage at bus 5 in the original IEEE 14 bus system is 0.993 pu. Results
for the heaviest loaded situation (30% system load increase) are illustrated in Figure 7-7
(lower).
An optimum in total system losses occurs when the maximum active power of 190 MW is
sent via the VSC HVDC link. The total system losses are then 17.87 MW, which is one
fourth less than with the original network at a 30% system load increase. In this
situation, only the generator at bus 3 has reached its maximum reactive power limit.
System voltages are given in Table 7-5.
Bus Voltage original network Inj. Q original network Voltage with VSC HVDC Inj. Q with VSC HVDC
30% increase [pu] 30% increase [MVAr] 30% increase [pu] 30% increase [MVAr]
1 1.06 11.35 1.06 -7.55
2 1.029 50 (limit) 1.045 45.97
3 0.983 40 (limit) 1.005 40 (limit)
5 0.993 - 1.020 -
6 1.043 24 (limit) 1.070 22.80
8 1.069 24 (limit) 1.090 22.07
Table 7-5: Voltages at 30% system load increase
The VSC HVDC link is able to hold the voltage at bus 5 to a value of 1.02. System
voltages are higher with the inclusion of the VSC HVDC link, which has a positive effect
on the network losses. A break-down of the VSC HVDC losses in both load cases is
given in Table 7-6.
load optimal power over rectifier losses inverter losses DC cable total VSC HVDC
increase DC link [MW] [MW] [MW] losses [MW] losses [MW]
0% 154 2.95 3.11 0.33 6.39
30% 190 3.27 3.50 0.50 7.27
Table 7-6: Break-down of VSC HVDC losses
7.4.1.4 Including M4 VSC HVDC Link: From 0% to 30% Load Increase
System load is varied between 0% (original system) and 30% and active power through
the VSC HVDC link is varied between 75 MW and 190 MW. There are also 3 different
loss models used:
- Sdra Lnken loss model,
- a 15% reduction of this loss model and
- a 30% reduction of this loss model.
Converter station losses amount to the highest share of a VSC HVDC link loss.
Reducing these losses has a considerable impact on the results.
Capability chart limits and converter voltage limits were controlled and fulfilled during the
simulation. Results are presented in Figure 7-8.
Obviously, the upper plane is for the Sdra Lnken loss model, the middle plane is for a
15% reduction of this loss model and the lower plane is for a 30% reduction of this loss
model.
28
26
24
22
20
18
16
14
12
30
10 20 ]
60 a se [%
80 100 a d in cre
120 10 lo
Power throu gh 140 ystem
VSC 160
HVDC link [M 180 0 erce nt s
W] 200 P
Figure 7-8: Losses in function of percent system load increase and power through VSC
HVDC link
For each percent of system load increase, the minimum total network loss is searched
together with the associated optimal active power through the VSC HVDC link. Results
are presented in Figure 7-9. The upper part shows a comparison of the losses in the
original network and the losses in the network with VSC HVDC and the three different
loss models used. For each percent of system load increase, the value of the losses
with VSC HVDC is an optimum; for each percent of system load increase, the optimal
active power flow over the DC link is given in the lower part of the figure. Increasing
system load causes an increase in total network losses. After an increase of system
load of about 20%, it is more advantageous from a losses point of view to operate the
VSC HVDC link at nominal operating condition, that is, an active power over the link of
190 MW. How lower the converter losses, how faster the VSC HVDC link starts
operating in its nominal operating condition.
In this situation, a VSC HVDC link always performs better from a losses point of view
than the original AC network. However, the positive influence of VSC HVDC on network
losses gets larger with increased system load. The fact that the transmission system
operator gets more benefits from the VSC HVDC link when the network is heavily loaded
is illustrated in Figure 7-10.
Figure 7-9: Total network losses with and without VSC HVDC (upper) and optimal active
power over VSC HVDC link (lower)
Percent loss reduction is calculated as follow:

lossno VSC HVDC lossVSC HVDC
%loss reduction = 7-1
lossno VSC HVDC
At higher system loading, the VSC HVDC has a greater positive influence on the
surrounding network.
7.4.1.5 Including VSC HVDC Link of Higher Rating: No Load Increase
When an M4 HVDC Light configuration is inserted in the IEEE 14 bus network, with an
increase of system load by 20% or more, the optimum active power over the link is the
maximum active power of 190 MW. So it may seem more advantageous to insert a VSC
HVDC configuration of higher capacity, for example the M3 configuration (304 MVA, 80
kV rating) or the M5 configuration which has a higher DC voltage and thus lower DC
cable losses than the M3 configuration (373 MVA, 150 kV rating). Inserting these links
in the original IEEE 14 bus system and varying active power over the VSC HVDC link
results in Figure 7-11. Losses are higher with the inclusion of VSC HVDC in comparison
to the original network. For the M3 configuration, a minimum of 14.48 MW total network
losses is found when 162 MW is sent through the link. For the M5 configuration, a
minimum of 15.76 MW total network losses is found when 166 MW is sent through the
link.
35
30
Percent total network loss reduction [%]
25
20
15
Sodra Lanken loss model

15% converter loss reduction
10
5
0 5 10 15 20 25 30
Percent system load increase [%]
Figure 7-10: Percent loss reduction with VSC HVDC in comparison to original system in
function of percent load increase
20
19 HVDC Light M3
HVDC Light M5
18 Original network (0% load increase)
17
16
Losses [MW]
15
14
13
12
11
10
80 100 120 140 160 180 200 220 240
Figure 7-11: Inserting an M3 or an M5 HVDC Light configuration

7.4.1.6 Including VSC HVDC Link of Higher Rating: 30% Load Increase
When the situation gets worse and the system load is increased, losses with the M3 and
M5 configurations remain higher than with the M4 configuration. This is illustrated in
Figure 7-12. At 30% load increase, the optimum with the M4 configuration is situated at
the maximum of 190 MW. With the M3 configuration, the optimum is situated at a power
through the link of 214 MW. With the M5 configuration, the optimal power over the DC
link is 220 MW. Optima are 17.87 MW, 19.49 MW and 20.57 MW respectively. No-load
losses of a VSC HVDC configuration play an important role in this simulation. They are
the cause the losses of the M3 and the M5 HVDC Light are higher than the M4 HVDC
Light.
A break-down of the M3 and the M5 HVDC Light losses is given in Table 7-7.
Although the resistance for the M5 configuration is higher than the resistance of the M3
configuration: 0.029 /km for the M5 and 0.013 /km for the M3 (compare with 0.061
/km for the M4); because the M5 configuration has a higher DC voltage, the DC cable
losses are smaller than with the M3 configuration. On the other hand, the no-load
converter station losses of the M5 configuration are higher than the no-load losses of the
M3 configuration. Overall, losses with M5 are greater than losses with M3.
0% (M3) 162 4.10 4.20 0.29 8.59
0% (M5) 166 4.83 4.90 0.17 9.9
30% (M3) 214 4.47 4.64 0.51 9.62
30% (M5) 220 5.18 5.31 0.31 10.8
Table 7-7: Break-down of the HVDC Light losses for the M3 and M5 configuration
30
M4 HVDC Light
M3 HVDC Light
28
M5 HVDC Light
Original network (30% load increase)
26
Losses [MW]
24
22
20
18
16
100 120 140 160 180 200 220 240 260 280 300
Figure 7-12: Losses in function of power through VSC HVDC link at 30% load increase
7.4.2 IEEE 118 Bus System
Subsequently, the IEEE 118 bus system is studied more in detail. The best result
concerning loss minimization in this network thus far has been obtained by inserting 2
VSC HVDC links and controlling active power through them. This behavior is now
studied under increasing loading conditions of the network. Network losses without load
increase are 132.48 MW and network losses at 30% load increase amount to 224.50
MW. Concerning reactive power limits of PV-generators and synchronous
compensators; in the original system, generator 103 has reached its upper limit while
generators 19, 32, 34, 92 and 105 have reached their lower limit. In the system without
VSC HVDC and at 30% load increase, generators 1, 12, 15, 56, 70, 74, 76, 77, 92, 103
and 104 have reached their upper reactive power limits.
When two VSC HVDC links are inserted, the synchronous compensator at bus 27 is
dropped and the inverter tries to hold the voltage to a value of 0.97 pu. The other VSC
HVDC link also tries to hold the voltage at bus 75 to a value of 0.97 pu. The result for a
30% load increase of the IEEE 118 bus network incorporating two VSC HVDC links and
using the three different converter loss models as explained before is illustrated in Figure
7-13. The upper plane represents the Sdra Lnken loss model, the middle plane a
15% reduction of this model and the lower plane a 30% reduction of this model. The
optimum of minimal total network losses is situated at the nominal operating condition of
both VSC HVDCs, being 190 MW of active power sent through the DC link. Minimal
total system losses at a 30% load increase are 217.08 MW for the Sdra Lnken
converter loss model. At this operating condition, the following generators have reached
their limits: 1, 12, 15, 56, 76, 92, 103 and 104. This shows the beneficial influence of
VSC HVDC on the surrounding network. In contrast to the case with no VSC HVDC,
synchronous compensators 70, 74 and 77 did not reach their limit yet in the case with
VSC HVDC. A summary is given in Table 7-8. The beneficial influence of VSC HVDC
on the voltage profiles is very small.
Synchronous Voltage no VSC Inj. Q no VSC Voltage with VSC Inj. Q with VSC
compensator HVDC [pu] HVDC [MVAr] HVDC [pu] HVDC [MVAr]
70 0.982 32 (limit) 0.984 24.51
74 0.950 9 (limit) 0.958 2.99
77 1.005 70 (limit) 1.006 63.15
Table 7-8: Voltages and reactive power injections with and without VSC HVDC at 30%
load increase
220
219
218
217
216
215
214
213
170 ]
[MW
75
212 175 and
69
170 us es
180 n b
175
Power th wee
rou gh V
SC HVD 180 185 link b et
C link b etw C
een bus 185 VD
es 25 and 27 190 190 SC H
[MW] V
gh
r u
o
er th
Pow
Figure 7-13: Effect of two VSC HVDC links on total system losses at 30% load increase
For each percent of load increase, the optimum active power flows over both VSC HVDC
links are searched. These optima result in a minimum total network loss for that percent
of load increase. The result is presented in Figure 7-14.
240
Loss in original AC system

220 Sodra Lanken loss model
200
180
160
140
120
0 5 10 15 20 25 30
Figure 7-14: Total network losses with and without VSC HVDC in function of percent
system load increase
The optimal power flows associated with these losses occur at the values presented in
Figure 7-15. Optimal power flows over the DC link between buses 25 and 27 are given
in the upper part, while optimal power flows between buses 69 and 75 are given in the
lower part of the figure. In the simulation, the steps in load increase are 3% and the
steps in power over the DC link are 5 MW. Concerning the active power flow over link
25-27, the optimal power in the 15% and in the 30% loss reduction cases is 190 MW for
every percent of load increase.
Figure 7-15: Optimal active power over VSC HVDC link 25-27 (upper) and over VSC
HVDC link 69-75 (lower)
For the Sdra Lnken converter loss model, the link 25-27 reaches its nominal operating
point after a load increase of 3%, while link 69-75 reaches its nominal operating point
after a load increase of 9%. Again the VSC HVDC is more profitable for the TSO with
increasing system load. This is illustrated in Figure 7-16. Using the Sdra Lnken loss
model, a loss reduction of 3.3% is reached at a maximum load increase of 30%. The
effect is less pronounced than in the IEEE 14 bus system however.
5.5
Percent total network loss reduction [%] 5
4.5
3.5
2.5
2
Sodra Lanken loss model
1.5 15% converter loss reduction
1
0.5
0 5 10 15 20 25 30
Figure 7-16: Percent loss reduction with VSC HVDC in comparison to original system in
function of percent load increase
A break-down of VSC HVDC losses for the Sdra Lnken converter loss model in the
0% and 30% load increase cases is given in Table 7-9.
0% (25-27) 185 3.22 3.43 1.43 8.08
0% (69-75) 175 3.13 3.31 0.98 7.42
30% (25-27) 190 3.27 3.51 1.51 8.29
30% (69-75) 190 3.27 3.50 1.16 7.93
Table 7-9: Break-down of VSC HVDC losses
Two networks are investigated in this chapter: the IEEE 14 bus network and the IEEE
118 bus network. In both networks, when an AC line is replaced by a VSC HVDC link,
losses are minimized. Reactive power is injected by the converters to hold the voltage at
the associated AC buses to a certain value. Active power is varied from zero to the
maximum rating of the VSC HVDC link. An optimal active power over the DC link,
corresponding to a minimum in total network losses, is searched. Subsequently, the
network load is increased from 0% to a maximum load increase of 30%. From the
simulations it follows that loss reduction with VSC HVDC is larger in heavily loaded
networks. In the simulation cases, the beneficial effect of VSC HVDC was more
pronounced in the smaller IEEE 14 bus network than in the IEEE 118 bus network.
8 ECONOMICAL ANALYSIS
Previous analysis shows it is possible to reduce network losses in a meshed network

when an existing AC overhead line is replaced by a VSC HVDC link. In this chapter
some benefits of VSC HVDC are monetized and an economical analysis is performed.
In this analysis, the TSO expects a load increase in his network and he has two choices.
He can either keep the AC OHL or he can replace it with a VSC HVDC link to minimize
total network losses. Other cases are obviously possible. For example, when the AC
OHL has to be replaced because it is outdated or when extra transmission capacity
needs to be installed, the TSO has to invest in a new transmission technology. A new
OHL is nearly impossible to install because of environmental impact and public protest.
In Europe, new OHL are virtually impossible to install. The TSO can also opt for AC
underground cables. An extensive economic analysis has to be performed here, which
is out of the scope of this thesis. Only the economical feasibility of VSC HVDC on loss
reduction in meshed networks is studied here. As a consequence, it is assumed that the
existing AC OHL can cope with the load increase. No extra capacity is installed. Also,
the existing generators provide the extra power to feed the loads and system losses. In
the liberalized electricity market, generation and transmission are unbundled. It is the
TSO who invests in a VSC HVDC link. Lower losses in his network mean that the TSO
has to buy fewer generating capacity from the generators. In the case of an AC OHL,
higher losses mean that the generating company has to provide more generating
capacity [/MW] and that more fuel is consumed [/MWh]. Because loss reduction in
this case is small compared to installed generating capacity, the existing generators
provide the extra capacity. However, when load and associated losses increase
significantly, it is possible that new generating capacity has to be purchased. In that
case, the price of generating capacity [/MW] has to be taken into account.
8.1 Data Used in Economical Analysis
Installing VSC HVDC in the networks under study reduces network losses. When
keeping the AC OHL, network losses are higher and as a consequence more fuel is
consumed by the generators. The TSO has to buy more generating capacity. It is
assumed that extra power is delivered by the marginal power plant: the Open-Cycle Gas
Turbine (OCGT). The cost of generating electricity is then assumed to be 0.04 /kWh
[81]. The TSO has to buy this extra electricity in the case of the AC OHL to supply the
losses.
8.1.1 AC Data
When installing a VSC HVDC link instead of keeping the AC OHL, the AC OHL line has
to be decommissioned. Appropriate cost data of this decommissioning is not readily
available and is thus not used in the analysis.
Maintenance costs of an AC OHL are assumed to be 1000 /km/year [82].
8.1.2 VSC HVDC Data
The VSC HVDC link consists of 2 converter stations and the underground DC cable.
The cost of one converter station (including transformer, filters and other equipment) is
taken to be 40 M [83].
The cost of the DC underground cable is 0.65 M/km [84]. The installation cost of the
DC cable amounts to 0.03 M/km [85].
The maintenance cost of the VSC HVDC link is valued at 0.5% of the capital cost of VSC
HVDC per year [86].
8.1.3 Ancillary Services Data
VSC HVDC has many advantages over the existing AC OHL [85]:
- reactive power support of the surrounding network;
- invisible technology;
- smaller right-of-way;
- higher value of surrounding properties;
- lower audible noise;
- no changing magnetic fields;
- increased power system stability;
- possibly black-start capability;
- lower number of faults than the AC OHL;
- VSC HVDC acts as a firewall against grid disturbances [21].
Both VSC converter stations have the ability to control active and reactive power
independently. Reactive power is used to support the voltage. The TSO is paid for this
service. Reactive power payments are summarized in Table 8-1:
Table 8-1: Reactive power payments
These rates are for the Irish national grid [87]. It is assumed that there is 50 MVAr of
capacity available for each converter. As can be concluded from the technical analysis,
the VSC HVDC links work in their nominal operating point (190 MVA for the M4 HVDC
Light configuration) most of the time. This means that there is not much capacity left
for reactive power support. If the VSC HVDC link works off-nominal, losses increase
again. When 50 MVAr is available for reactive power support, this means that there is
still 1902 502 = 183.3 MW over for active power transfer. Results are thus not
influenced much. It is assumed that this capacity of 50 MVAr is used for 1000 hours per
year.
The other benefits of VSC HVDC are difficult to monetize. Also, it is not the TSO who
benefits from these advantages of VSC HVDC over an AC OHL. Studies performed in
the UK and in Finland have proven that the values of properties decrease when those
properties are closer to an AC OHL [85]. It is tried here to monetize this benefit of VSC
HVDC. It is assumed that every 500 meters along the OHL there is [85]:
- one property 500 meters away with an 8% reduction in value,
- two properties 1000 meters away with a 4% reduction in value,
- three properties 2000 meters away with a 2% reduction in value.
Each property is valued at 100 000 . Although this does not affect the TSO directly,
previous values are taken into account in the financial analysis. One reason for this is
the fact that the OHL belongs to the TSO and he is responsible for the decrease in
property value. One can consider it as some kind of penalty to the TSO for the visual
impact of an OHL.
In the economic analysis, two scenarios are investigated, as explained in the following
paragraphs. In the Discounted Cash Flow (DCF) calculation, interest rate is taken as 5%
and tax rate is taken as 40%. The DCF is performed in Excel.
8.2 First Scenario: Annual Load Increase
In this first scenario, the TSO expects an annual load increase of 1% each year in his
network for the following 30 years. Each year, load is increased by 1% of the original
load in year 0. 30 years is typically also the lifetime of the VSC HVDC installation. At
the end of the 30 years, system loading is increased by 30%. This is also the maximum
value used in the simulations in the technical part of the thesis. Simulations are
performed for three converter loss models: the Sdra Lnken (SL) loss model, a 15%
reduction of this model and a 30% reduction of this model.
8.2.1 IEEE 14 Bus Test System
The financial analysis is performed as depicted in Table 8-2 and in Table 8-3. For
example, in the first year there is a 1% load increase. The TSO saves 1.12 MW in this
case during 8760 hours a year. This results in an energy saving of 9811.2 MWh in the
first year. In the case of an AC OHL, the TSO has to buy this extra generating capacity
at the price of 0.04 /kWh, which results in a saving using VSC HVDC of 0.392448 M in
the first year. This is done for the total life span of 30 years.
percent load energy financial

year power saving
increase saving saving
# % MW MWh/year M/year
0 0 0,99 8672,4 0,346896
1 1 1,12 9811,2 0,392448
2 2 1,25 10950 0,438
3 3 1,38 12088,8 0,483552
4 4 1,52 13315,2 0,532608
5 5 1,66 14541,6 0,581664
6 6 1,8 15768 0,63072
7 7 1,94 16994,4 0,679776
8 8 2,08 18220,8 0,728832
9 9 2,22 19447,2 0,777888
10 10 2,37 20761,2 0,830448
11 11 2,52 22075,2 0,883008
12 12 2,67 23389,2 0,935568
13 13 2,82 24703,2 0,988128
14 14 2,98 26104,8 1,044192
15 15 3,13 27418,8 1,096752
16 16 3,29 28820,4 1,152816
17 17 3,46 30309,6 1,212384
18 18 3,62 31711,2 1,268448
19 19 3,78 33112,8 1,324512
20 20 3,96 34689,6 1,387584
21 21 4,14 36266,4 1,450656
22 22 4,32 37843,2 1,513728
23 23 4,51 39507,6 1,580304
24 24 4,71 41259,6 1,650384
25 25 4,92 43099,2 1,723968
26 26 5,13 44938,8 1,797552
27 27 5,34 46778,4 1,871136
28 28 5,55 48618 1,94472
29 29 5,77 50545,2 2,021808
30 30 6 52560 2,1024
Table 8-2: Financial saving due to energy saving for the IEEE 14 bus system in the first
scenario
The next step is a discounted cash flow to assess which of the two options is the most
beneficial from an economical point of view. When the TSO chooses for the VSC HVDC
link, costs are:
- the investment cost of converters and DC cable;
- the maintenance of the VSC HVDC which is more expensive than the maintenance
of the AC OHL.
Benefits when choosing for the VSC HVDC option are:
- the values of the properties which go up in price;
- the saving in energy due to the lower network losses (Table 8-2);
- reactive power to support the voltage.
The Excel worksheet is also capable of incorporating the investment in new generating
capacity. If losses amount to a high value when the system load increases, it is possible
that new generating capacity needs to be installed. In a liberalized market however, it is
not the TSO who pays for this extra capacity.
0 1 2 3
DCF (SL)
M M M M
net investment cost -94,96
depreciation -3,165333 -3,165333 -3,165333
decommissioning AC 0
property value 0,968
energy saving 0,392448 0,438 0,483552
maintenance AC 0,022 0,022 0,022
maintenance HVDC -0,4748 -0,4748 -0,4748
voltage support 0,505544 0,505544 0,505544
profit before tax -2,720141 -2,674589 -2,629037
tax -1,088057 -1,069836 -1,051615
profit after tax -1,632085 -1,604754 -1,577422
depreciation 3,165333 3,165333 3,165333
net cash flow -93,992 1,533249 1,56058 1,587911
discounted cash flow -93,992 1,460237 1,415492 1,371697
TOTAL -65,1365
info cost [M]

number of converters 2 80
length link [km] 22 14,96
total [M] 94,96
Table 8-3: Discounted Cash Flow for the IEEE 14 bus system in the first scenario
Previous simulations are also performed for the other converter loss models. Results of
the discounted cash flows are presented in Figure 8-1.
-54
SL 15% 30%
-56
Net Present Value [Meuro]
-58
-60
-62
-64
-66
Converter loss model
Figure 8-1: Result of DCF for the IEEE 14 bus system in the first scenario
From the results it is obvious that it is not feasible from an economical point of view to
replace the AC OHL by a VSC HVDC link to minimize network losses. Sensitivity of the
results concerning converter loss is 3 M profit per 15% converter loss reduction.
Sensitivity of the results concerning converter cost is 17 M profit per 25% converter
cost reduction. Sensitivity to electricity price is 4.5 M profit per 50% electricity price
augmentation.
As mentioned earlier, there are other advantages of VSC HVDC over an AC OHL, but it
is difficult to monetize these advantages.
Results of the IEEE 118 bus system are presented in Figure 8-2. The link between
buses 25-27 is 65 km long and the link between buses 69-75 is 50 km long. The
difference in net present value is mainly due to the different investment costs of the DC
cable.
-60
SL 15% 30%
-65
link 25-27
-70
link 69-75
-75
-80
-85
-90
-95
Figure 8-2: Result of DCF for the IEEE 118 bus system in the first scenario
Sensitivity to converter loss in this case is 3.5 M per 15% converter loss reduction.
Sensitivity of the results concerning converter cost is 17 M per 25% converter cost
reduction. Sensitivity of the results concerning electricity price is 3 M per 50%
electricity price augmentation.
For the IEEE 118 bus system with two VSC HVDC links, previous results can be
superpositioned.
8.3 Second Scenario: Each Year the Same Loading Profile
In this scenario, the TSO expects a sudden load increase in his network starting from
year one. For example, a new load centre can be installed in his network. The load is
varied each year according to a certain loading pattern between 0% and 30% load
increase. Each year is characterized by this same loading pattern. The loading pattern
between 0% and 30% is based on data of the Belgian TSO Elia (www.elia.be). 4 days
spread over one year where chosen and from that data the loading pattern was
determined. The original networks available in Matpower were taken as the 0% load
increase case. The annual loading pattern is depicted in Figure 8-3.
For example, a 10% load increase happens 7.29% of the time in one year, or 638.6
hours in one year. In this scenario, the energy savings are the same in each year. This
is in contrast to the first scenario, where energy savings increase each year.
8
7
6
Probability [%]
5
4
3
2
1
0
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
System load increase [%]
Figure 8-3: Annual loading profile between 0% and 30% load increase
Results for this system are presented in Figure 8-4. Sensitivity to converter loss is 3
M per 15% converter loss decrease. Sensitivity to converter cost is 17 M per 25%
converter cost decrease. Sensitivity to electricity price is 4.5 M profit per 50%
electricity price augmentation. Results are comparable to the results obtained with the
first scenario.
-54
SL 15% 30%
-56
-58
-60
-62
-64
-66
Figure 8-4: Result of DCF for the IEEE 14 bus system in the second scenario
Results for this system are presented in Figure 8-5.

Results are comparable to this case in the first scenario. Sensitivity to converter loss in
this case is 3.5 M per 15% converter loss reduction. Sensitivity of the results
concerning converter cost is 17 M per 25% converter cost reduction. Sensitivity of
the results concerning electricity price is 3 M per 50% electricity price augmentation.
-60
SL 15% 30%
-65

link 25-27
-70
link 69-75
-75
-80
-85
-90
-95
Figure 8-5: Result of DCF for the IEEE 118 bus system in the second scenario
For the IEEE 118 bus system with two VSC HVDC links, previous results can be
superpositioned.
8.4 Conclusion
A final remark is in place. In the previous simulations the TSO had the option to either
keep the original AC OHL or to install a VSC HVDC link. A more realistic case may be if
the TSO opted to replace the AC OHL with a new OHL; one of higher capacity for
example. He has the advantage that the AC towers to support the OHL are already
installed. In the simulations, one has to cope with the characteristics of the new AC OHL
which have an impact on network losses. The cost of this new AC OHL also has to be
investigated. It is also uncertain if the TSO has the possibility to replace the old AC line.
Public protest can lead to the conclusion that the TSO is unable to replace the OHL.
The only possibility in that case is undergrounding, either AC underground cables or DC
underground cables. These economic analyses are out of the scope of this thesis
however.
The economical benefit of VSC HVDC on loss reduction in meshed networks is
investigated. From the results of the discounted cash flows, it is clear that it is not
feasible for the TSO to replace an AC line in his network with a VSC HVDC link to
reduce network losses for the networks under study. Because converter losses will
decrease in the future, a sensitivity analysis with respect to converter losses is studied.
The price of generating electricity is also an uncertain variable; sensitivity with respect to
the electricity price is also investigated. A final sensitivity is investigated with respect to
the converter cost, although this cost will probably not decrease in the future. Even with
more beneficial values of converter losses, converter cost and electricity price, the
investment cost in VSC HVDC technology can not be paid back via the profit in the
network losses.
9 CONCLUSIONS
Several developments take place in the transmission grids. First of all, there is the
liberalization of the electricity markets. Transmission networks also have to cope with
the large scale introduction of renewable energy, especially wind energy. A final aspect
is the public protest against overhead lines. Undergrounding of transmission lines can
become reality in the future. New technologies, specifically VSC HVDC, can play an
important role in the future transmission networks. The main advantages are its high
controllability of active and reactive power and the use of underground cables, which
make this technology invisible.
In a first part of the thesis, the working principles and technical aspects of VSC HVDC
are outlined. These working principles are implemented in the load flow calculations to
carry out the simulations in the following chapters.
When inserted in a transmission grid, VSC HVDC has several advantages. Two main
advantages of VSC HVDC in meshed networks have been investigated in this thesis.
The first avantage of VSC HVDC is the improvement of the static voltage-power
relationship. The nose point of the PV-curve is the steady-state voltage stability limit.
The transmission grid should never operate near this PML. Improvement of voltage
profiles is realized with VSC HVDC: voltages are boosted compared to the original case
and the PML is shifted to the right in the PV-plane.
In the following part the feasibility of VSC HVDC on loss reduction in meshed networks
has been studied. Aspects of VSC HVDC and their advantages in meshed grids are
discussed. The effects of these advantages on network losses are verified through
simulations. Simulations are performed on two networks: the IEEE 14 bus and the IEEE
118 bus network. Networks are studied in their original state and in a state of increased
loading. Simulation results show that VSC HVDC becomes more beneficial with
increased system loading. Concerning losses, VSC HVDC is a good option for inclusion
in a transmission network. The three main effects of VSC HVDC are the unloading of
the surrounding AC lines, the higher voltages in the neighborhood of the VSC HVDC link
and the upper reactive power limits of the generators that are not reached so quickly.
With the many changes in the power systems environment lately, TSOs are looking for
alternative technologies to reinforce their grid. Loss minimization plays an important role
for a TSO. From the foregoing work, it is clear that network losses can be reduced with
VSC HVDC.
An economical analysis has also been performed. In both networks under study,
investment costs in VSC HVDC technology could not be paid back. From a financial
point of view, the advantages of VSC HVDC in meshed grids do not outweigh the costs.
High converter losses are the main disadvantage of VSC HVDC. Future developments
can lower these losses. A combination of a demand for more flexibility in the grid and
lower VSC losses in the future will result in an increased attention in VSC HVDC.
10 FUTURE WORK
Topics not covered in this thesis and that can be treated in future work are:
The inclusion of multi-terminal VSC HVDC in meshed networks. Instead of a DC
link connected by two converters, a DC network can be included in an AC
network. Multi-terminal VSC HVDC consists of three or more converters
connected via HVDC cables. The effect of multi-terminal VSC HVDC on system
losses can be investigated.
Dynamic effects in meshed grids. This thesis describes a steady-state situation
between the VSC HVDC link and the AC network. With a dynamic model of VSC
HVDC provided by ABB in PSS/E, it is possible to simulate dynamic effects. The
tripping of an important line or generator in the network and the proper control of
VSC HVDC to support the grid can be investigated.
The feasibility of VSC HVDC in distribution networks. Distribution networks are
normally not meshed but radial. The inclusion of VSC HVDC in radial networks
can be investigated from a technical and/or economical point of view.
A more thorough economical analysis can be performed. The advantages and
costs of different technologies can be investigated: new AC OHL, FACTS device,
AC underground cable, LCC HVDC or VSC HVDC. Each technology has its
advantages and disadvantages in a specific situation. A detailed economical
analysis of a specific case can help the TSO decide which technology is best in
his grid.
11 SUMMARY
The scene of electric power systems is changing drastically. Causes of this evolution
are the liberalization of the electricity markets and the increase in renewable energy.
Uncertainty in the transmission grid and congestion of the lines raise the need for new
technology. HVDC can play an important role in this evolution, especially VSC HVDC.
VSC HVDC consists of two voltage source converter stations either directly connected or
via a DC cable. Using pulse width modulation, the VSC builds a sinusoidal voltage
waveform from the DC voltage. Advantages of VSC HVDC are its high active and
reactive power flow controllability. Active power is controlled via the voltage phase angle
over the phase reactor and reactive power is controlled via the voltage amplitude over
the phase reactor. VSC HVDC can operate in all four quadrants in the power plane.
Instant power flow reversal is also possible. Disadvantages are its high cost and its high
losses, mainly due to the voltage source converters. VSC HVDC also has some
advantages over the older LCC HVDC technology. LCC converter stations always
consume reactive power. Power flow reversal with LCC HVDC is slower compared to
VSC HVDC. On the other hand, converter station losses with LCC HVDC are lower than
VSC HVDC converter station losses.
In a following step, an existing load flow program is extended to cope with the inclusion
of one or more VSC HVDC links. A load flow calculation results in the steady-state
solution of the network. Using generation, network and load properties, voltages and
power flows in the network are obtained using the Newton-Raphson method. The
Jacobean matrix, which is the relationship between voltage and power mismatches, is
extended with the VSC HVDC variables. There are two extra equations for every
converter station. The VSC converters can operate either in reactive power flow control
mode, injecting or absorbing a predefined amount of reactive power, or in AC voltage
control mode, holding the voltage at the bus connected to the associated converter
constant. The number of iterations increases in the extended load flow program
compared to the original program.
The inclusion of VSC HVDC in a meshed network has a positive effect on the voltage
profiles of the load buses. Increasing the load at a bus leads to a decrease in the
associated bus voltage. This relationship is graphically illustrated using a PV- or nose-
curve, which is the steady-state relationship between bus power and bus voltage. For a
certain (maximum) load power, a saddle-node bifurcation is reached: a further increase
in load admittance lowers the power consumed by the load. The voltage keeps
dropping. This point of maximum loading is called the nose-point or critical point. A
study performed by ABB investigates the effect of a VSC HVDC link operated in parallel
with an AC line in a simple two-bus network. When the VSC HVDC link is operated
correctly, the new situation results in higher voltages for the same load powers and a
shift of the point of maximum loadability to the right in the power-voltage-plane.
Investigating this simple case with the own VSC HVDC load flow model leads to the
same results. The study is expanded to small meshed networks. For a simple five-bus
network, PV-curves are generated together with a voltage security indicator, which
defines the distance that the system is from the point of maximum loading. Inserting a
VSC HVDC link in parallel between two load buses, leads again to the same beneficial
conclusions: higher voltages for the same load powers and a shift of the critical point to
the right in the power-voltage-plane.
The losses of an overall VSC HVDC station are the sum of the losses of the different
components. The determination of the loss in every component is given in standards.
These standards apply to LCC HVDC converter stations however. More research has to
be performed to determine the losses occurring in a VSC HVDC installation. Harmonics
are a very important influencing factor in the loss calculations. To obtain data on
harmonics, field measurements are the best solution. Valve losses are the main
contributor to total VSC HVDC losses. Valve losses can be divided in switching losses
and in conduction losses. Switching losses increase linearly with the switching
frequency. Conduction losses depend on the threshold voltage and the resistance of the
IGBTs and the diodes. Converter losses can be decreased using high-level VSC
topologies (two-level vs. three-level) and/or by using optimal pulse width modulation,
which cancels certain harmonics. Future developments in semiconductor devices are
also taking place. A loss model that takes every component of a VSC HVDC station into
account is provided by ABB. This loss model determines the converter station losses in
function of the active and reactive power exchanged between the converter and the AC
network. Losses can be broken down in three components: no-load losses, losses that
depend linearly on the phase reactor current and losses that depend quadratically on the
phase reactor current.
Inserting a VSC HVDC link in a meshed network can possibly lead to lower network
losses. The inclusion of VSC HVDC increases losses but the proper control of active
and reactive power can lower network losses. An optimum has to be found, being a
minimum in total network losses. Transferring active power over the DC link reduces the
load on the lines surrounding the VSC HVDC link. Due to the non-linearity of the system
( Ploss = RI 2 ), heavily loaded lines contribute more to system losses than lightly loaded
lines. VSC HVDC will thus perform better when it is inserted in a heavily loaded
network. Reactive power can be injected to maintain the voltage at the associated AC
bus to a desired value. Higher voltages lead to lower currents and thus lower losses.
Loss reduction with LCC HVDC has already been realized with the Fenno-Skan LCC
HVDC link. There is no realization yet of loss reduction in meshed grids using VSC
HVDC.
Simulations are performed in the IEEE 14 bus system and in the IEEE 118 bus system.
An AC overhead line is replaced by a VSC HVDC link. The VSC HVDC link is placed in
a meshed part of the grid and replaces an important AC line. The optimal loading of
the VSC HVDC link is searched, this is an optimal active power through the DC link that
minimizes total network losses. Reactive power is used to control the voltage at the
VSC terminals to a desired value. Both the IEEE 14 and IEEE 118 networks are studied
in their original state and in a state of increased loading. For the IEEE 14 bus system
the best result, from a losses point of view, is achieved when replacing the AC line
between buses 1 and 5. When inserting a VSC HVDC link the generators dont reach
their reactive power limits so quickly and the overall results are better voltage profiles.
For each percent of system load increase, the optimal active power over the DC link
shifts to higher values. At higher system loading, VSC HVDC has a greater positive
influence. In the IEEE 118 bus system, the best result is obtained when the AC
overhead lines between buses 25 and 27 and between buses 69 and 75 are replaced.
The same conclusions are valid here as in the IEEE 14 bus network, although the effects
are less pronounced.
From the simulations it follows that it is feasible from a technical point of view to reduce
losses in a meshed network when an AC overhead line is replaced by a VSC HVDC link.
In a final section the economical feasibility is investigated. This is done using a
discounted cash flow over a period of 30 years, which is typically the lifetime of a VSC
HVDC installation. Benefits of VSC HVDC that are monetized are the lower losses
compared to the original AC line, the availability/utilization of reactive power and the
increased value of the properties (VSC HVDC is an invisible technology). Two scenarios
are investigated. In the first scenario, the TSO expects a load increase of 1% of the
original load each year. In a second scenario, the TSO expects a sudden load increase,
but load varies each year between 0% and 30% load increase according to a loading
profile. Results are comparable for both scenarios however. In each case (2 scenarios
and 2 networks: IEEE 14 and IEEE 118) the result of the discounted cash flow is
negative: the benefits of VSC HVDC do not outweigh the costs. From an economical
point of view, it is better for the TSO to keep the original AC line. Even with more
beneficial values of converter losses, converter cost and electricity price, the investment
cost in VSC HVDC technology can not be paid back via the profit in the network losses.
12 REFERENCES
[1] B. Normark, E. K. Nielsen, Advanced power electronics for cable connection of offshore
wind, ABB, presented at Copenhagen Offshore Wind 2005.
[2] S. Cole, D. Van Hertem, L. Meeus, R. Belmans, Technical developments for the future
transmission grid, IEEE, International Conference on Future Power Systems 2005.
[Online]. Available: http://www.esat.kuleuven.be/electa/publications/fulltexts/pub_1503.pdf
[3] D. Van Hertem, J. Verboomen, R. Belmans, W. L. Kling, Power flow controlling devices:
An overview of their working principles and their application range, IEEE, International
Conference on Future Power Systems 2005. [Online]. Available:
http://www.esat.kuleuven.be/electa/publications/fulltexts/pub_1501.pdf
[4] B. Andersen, VSC Transmission, Cigr Working Group B4.37, April 2005.
[5] P. Haugland, Its time to connect: Technical description of HVDC Light technology,
ABB, 2006.
[6] G. Asplund, K. Eriksson, K. Svensson, DC transmission based on Voltage Source

Converters, Cigr, Colloquium in South Africa 1997.
[7] ABB, Gotland HVDC Light Project: Ns Bcks, Sweden. [Online]. Available:
www.abb.com/hvdc.
[8] ABB, Directlink HVDC Light Project: New South Wales and Queensland. [Online].
Available: www.abb.com/hvdc.
[9] L. Ronstrm, M. L. Hoffstein, R. Pajo, M. Lahtinen, The Estlink HVDC Light

transmission system, Cigr, Security and Reliability of Electric Power Systems, June 18-
20, 2007, Estonia.
[10] B. D. Railing, G. Moreau, L. Ronstrm et al., Cross Sound cable project: Second
generation VSC technology for HVDC, Cigr, B4-102, Session, 2004.
[11] ABB, Cross Sound Cable Interconnector: Connecticut and Long Island, USA. [Online].
[12] ABB, Murraylink HVDC Light Interconnection: Victoria South Australia. [Online].
[13] L. Zhang, L. Harnefors, P. Rey, Power System Reliability and Transfer Capability
Improvement by VSC-HVDC (HVDC Light), Cigr, Security and Reliability of Electric
Power Systems, June 18-20, 2007, Estonia.
[14] ABB, User guide for the PSS/E implementation of the HVDC Light model version 1.1,
2006.
[15] G. Tang, Damping subsynchronous resonance oscillations using a VSC HVDC back-to-
back system, Thesis, University of Sasketchewan, 2006.
[16] S. G. Johansson, G. Asplund, E. Jansson, R. Rudervall, Power system stability benefits
with VSC DC-transmission systems, Cigr, B4-204, Session, 2004.
[17] ABB, Capability chart for an HVDC Light. [Online]. Available: www.abb.com/hvdc.
[18] J. E. Skog, K. Koreman, B. Pjrvi, T. Worzik, T. ndersrd, The NorNed HVDC cable
link: A power transmission highway between Norway and the Netherlands, ABB.
[19] ABB, HVDC Light Strategy: First swing stability. [Online]. Available:
www.abb.com/hvdc.
[20] D. Van Hertem, S. Cole, R. Belmans, High Voltage Direct Current (HVDC) Technology,
Electa, 2007.
[21] L. Carlsson, HVDC, a firewall against disturbances in high-voltage grids, ABB, ABB
Review 3/2005, pp. 42-46, 2005.
[22] ABB, Caprivi Link Interconnector. [Online]. Available: www.abb.com.
[23] P. F. de Toledo, Feasibility of HVDC for city infeed, Licentiate Thesis, KTH Sweden,
2003.
[24] P. F. de Toledo, G. Asplund, E. Jansson, Aspects on infeed of multiple HVDC into one
AC network, ABB Power System, HVDC Division, Ludvika, Sweden.
[25] ABB, HVDC Classic transmission losses. [Online]. Available: www.abb.com.
[26] ABB, HVDC Light transmission losses. [Online]. Available: www.abb.com.
[27] ABB, The Itaipu HVDC transmission. [Online]. Available: www.abb.com.

[28] ABB, Application descriptions for HVDC Light. [Online]. Available: www.abb.com.
[29] M. Hyttinen, J. O. Lamell, T. F. Nestli, New application of voltage source converter (VSC)
HVDC to be installed on the gas platform Troll A, Cigr, B4-210, Session, 2004.
[30] G. Asplund, K. Eriksson, O. Tollerz, HVDC Light, a tool for electric power transmission
to distant loads, presented at VI Sepope Conference, Salvador, Brazil, May 1998.
[31] B. M. Weedy & B. J. Cory, Electric Power Systems, Wiley, fourth edition, 1998.
[32] E. Acha, C. R. F. Esquivel, H. A. Perez, C. A. Camacho, FACTS: Modelling and
simulation in power networks, Wiley, first edition, 2004.
[33] C. A. Camacho, O. L. Tortelli, E. Acha and C. R. F. Esquivel, Inclusion of a high voltage

DC-voltage source converter model in a Newton-Raphson power flow algorithm, IEE,
Proceedings on Generation, Transmission and Distribution, Vol. 150, Issue 6, pp. 691-
696, 2003.
[34] L. Gengyin et al., Power flow calculation of power systems incorporating VSC-HVDC,
IEEE, International Conference on Power System Technology, Vol. 2, pp. 1562-1566,
2004.
[35] X. P. Zhang, Multiterminal voltage-sourced converter-based HVDC models for power flow
analysis, IEEE, Transactions on Power Systems, Vol. 19, Issue 4, pp. 1877-1884, 2004.
[36] G. Radman, R. S. Raje, Power flow model/calculation for power systems with multiple
FACTS controllers, ScienceDirect, Electric Power Systems Research, Vol. 77, Issue 12,
pp. 1521-1531, Oct. 2007.
[37] R. D. Zimmerman, Matpower. [Online]. Available:
http://www.pserc.cornell.edu/matpower/.
[38] T. Van Cutsem, Voltage Instability: Phenomena, Countermeasures, and Analysis

Methods, IEEE, Proceedings of the IEEE, Vol. 88, Issue 2, pp. 208-227, 2000.
[39] IEEE Working Group, Voltage Stability of Power Systems: Concepts, Analytical Tools,
and Industry Experience, IEEE/PES 90TH0358-2-PWR, 1990.
[40] P. W. Sauer, Voltage Collapse Animation (DC), PSERC, 2003. [Online]. Available:
http://www.pserc.wisc.edu/Sauer_DCVoltage_Collapse_PSERC_Sep16.ppt.
[41] S. Johansson, F. Sjgren, Voltage collapse in power systems: The influence of generator
current limiter, on-load tap-changers and load dynamics, Licentiate Thesis, Chalmers
University of Technology, Sweden, 1995.
[42] F. Milano, PSAT Documentation. [Online]. Available:
http://www.power.uwaterloo.ca/~fmilano/downloads.htm.
[43] G. Coath, M. A. Dabbagh, Effect of steady-state wind turbine generator models on power
flow convergence and voltage stability limit, AUPEC Conference, 2005.
[44] D. A. Alves et al., Continuation Fast Decoupled Power Flow with Secant Predictor, IEEE,
Transactions on Power Systems, Vol. 18, Issue 3, pp. 1078-1085, 2003.
[45] V. Ajjarapu, C. Christy, The continuation power flow: a tool for steady state voltage
stability analysis, IEEE, Transactions on Power Systems, Vol. 7, Issue 1, pp. 416-423,
Feb. 1992.
[46] C. A. Caizares, A. C. Z. de Souza, V. H. Quintana, Comparison of Performance Indices

for Detection of Proximity to Voltage Collapse, IEEE, Transactions on Power Systems,
Vol. 11, Issue 3, pp. 1441-1450, 1996.
[47] ABB, HVDC Light Strategy: Voltage Stability. [Online]. Available: www.abb.com/hvdc.
[48] http://abblibrary.abb.com/global/scot/scot221.nsf/veritydisplay/8eb29b8966164a8bc1256fd
a004decc3/$File/pv_plot_HVDC_Light.m.
[49] S. Dey, C. K. Chanda, A. Chakrabarti, Development of a global voltage security indicator

(VSI) and role of SVC on it in longitudinal power supply system, ScieneDirect, Electric
Power Systems Research, Vol. 68, Issue 1, pp. 1-9, Jan. 2004.
[50] ABB, StakPak IGBT Press-packs: A new Packaging Concept for High Power Electronics,
2004.
[51] S. Meier, S. Norrga, H. P. Nee, New voltage source converter topology for HVDC grid
connection of offshore wind farms, KTH Sweden.
[52] M. Ikonen, O. Laakkonen, M. Kettunen, Two-level and three-level converter comparison
in wind power application, LUT Finland.
[53] B. R. Andersen, L. Xu, P. J. Horton and P. Cartwright, Topologies for VSC transmission,
IEEE, Power Engineering Journal, Vol. 16, Issue 3, pp. 142-150, June 2002.
[54] M. H. Bierhoff, F. W. Fuchs, Semiconductor losses in voltage source and current source
IGBT converters based on analytical derivation, IEEE, 35th Annual Power Electronics
Specialists Conference, Vol. 4, pp. 2836-2842, 2004.
[55] Semikron Power Electronics. [Online]. Available:

http://www.semikron.com/internet/webcms/objects/applica_help/e/3_2_1_3.pdf.
[56] ABB, IGBT-Presspack Module 5SNA 130025H0004 Common StakPak, ABB Switzerland
Ltd, Semiconductors.
[57] L. K. Mestha, P. D. Evans, Analysis of on-state losses in PWM inverters, IEE,
Proceedings on Electric Power Applications, Vol. 136, Issue 4, pp. 189-195, July 1989.
[58] A. M. Gole, Sinusoidal Pulse Width Modulation, 2000. [Online]. Available:
http://encon.fke.utm.my/nikd/SEM4413/spwm.pdf.
[59] O. A. Sdaba et al., Voltage harmonics generated by 3-level converters using PWM
natural sampling, IEEE, 32nd Power Electronics Specialists Conference, Vol. 3, pp. 1561-
1565, 2001.
[60] ABB, Converter reactor and AC-filter for HVDC Light. [Online]. Available:
www.abb.com.
[61] Mathworks, SimPowerSystems VSC HVDC Matlab-Simulink Model. [Online]. Available:
www.mathworks.com.
[62] M. Rahimo, U. Schlapbach, A. Kopta, R. Schnell, S. Linder, SPT+, the next generation of
low-loss HV-IGBTs, PCIM Power Electronics Conference, Nuremberg, 2005.
[63] R. Chokhawala, B. Danielsson, L. ngquist, Power Semiconductors in Transmission and
Distribution Applications, IEEE, Proceedings of the 13th International Symposium on
Power Semiconductor Devices and ICs, pp. 3-10, 2001.
[64] K. Bergman, Silicon carbide, the power semiconductor material of the future, ABB, ABB
Review 1/1996, pp. 37-43, 1996.
[65] IEC, Convertor transformers - Part 1: Transformers for industrial applications,
International Standard 61378-1, IEC, 1997.
[66] IEC, Convertor transformers - Part 2: Transformers for HVDC applications, International
Standard 61378-2, IEC, 2001.
[67] IEC, Determination of power losses in high-voltage direct current (HVDC) converter
stations, International Standard 61803, IEC, 1999.
[68] T. C. Elliott, K. Chen, R. C. Swanekamp, Standard handbook of powerplant engineering,
McGraw-Hill Professional, second edition, 1997.
[69] IEEE Standard, IEEE Recommended practice for determination of power losses in high-
voltage direct-current (HVDC) converter stations, IEEE, 1991.
[70] C. Du, The control of VSC-HVDC and its use for large industrial power systems,
Licentiate Thesis, Chalmers University of Technology, Sweden, 2003.
[71] ABB, Differences between HVDC Light and classical HVDC. [Online]. Available:
www.abb.com.
[72] ABB, HVDC Cables: submarine and land power cables, ABBs high voltage cable unit in
Sweden, 2006.
[73] ABB, HVDC Light Cables. [Online]. Available: www.abb.com.
[74] D. Van Hertem, J. Verboomen, S. Cole, W. Kling, R. Belmans, Influence of phase shifting
transformers and HVDC on power system losses, IEEE, Power Engineering Society
General Meeting, 2007.
[75] D. J. Gotham, G. T. Heydt, Power flow control and power flow studies for systems with
FACTS devices, IEEE, Transactions on Power Systems, Vol. 13, Issue 1, pp. 60-65,
1998.
[76] S. Cole, D. Van Hertem, I. Pardon, R. Belmans, Randstad HVDC, Electa, 2006.
[Online]. Available: http://www.esat.kuleuven.be/electa/vsc-
hvdc/reports/Randstad_HVDC.pdf.
[77] ABB, Finland - Sweden submarine interconnection - Fenno-Skan. [Online]. Available:
www.abb.com/hvdc.
[78] [Online]. Available: http://www.ee.washington.edu/research/pstca/.
[79] E. Lakervi and E. J. Holmes, Electricity Distribution Network Design, IEE Power
Engineering, Series 21, Peter Peregrinus Ltd., second edition, 1996.
[80] V. G. Agelidis, G. D. Demetriades, N. Florentzou, Recent advances in high-voltage direct-
current power transmission systems, IEEE, International Conference on Industrial
Technology, 2006.
[81] RAENG, The cost of generating electricity, Royal Academy of Engineering, UK.
[82] REVOLT, REVOLT news 08/11/2004. [Online]. Available:
http://www.revolt.co.uk/news174.html.
[83] Transpower, North Auckland and Northland grid upgrade project Attachment E:
Assessment of options, September 2007. [Online]. Available:
http://www.electricitycommission.govt.nz/pdfs/opdev/transmis/gup/naan/Part-III-
Attachment-E-Assesment-of-Options.pdf.
[84] TREC, Getting the power to market would entail construction of high-voltage direct
current power lines. [Online]. Available:
http://www.trec.net.au/content/gettingthepowertomarket.html.
[85] D. Ravemark, B. Normark, Light and invisible; Underground transmission with HVDC
Light, ABB, ABB Review 4/2005, pp. 25-29, 2005.
[86] TGS, Evaluation of HVDC Light as an alternative for the Vancouver Island Transmission
Reinforcement (VITR) project, 2005. [Online]. Available:
http://www.bctc.com/NR/rdonlyres/EFE53841-4E00-4F3E-BB7A-
C7FEF49444B7/0/VITRAppQ.pdf.
[87] ESB National Grid, Proposed reactive power payment rates for 2006, 2005. [Online].
Available:
http://www.eirgrid.com/EirgridPortal/uploads/Regulation%20and%20Pricing/ESBNG%20Pr
oposed%20Reactive%20Power%20Payment%20Rates%20for%202006.pdf.
13 ENCLOSURES
13.1 Worldwide VSC HVDC Projects and Their Basic Parameters [80]
13.2 Description of the VSC HVDC Matpower Code in Matlab
The different functions are treated here. Each function is specified by its input and
output parameters and a description of the purpose of the function.
13.2.1 The Function runpf()
inputs description outputs description
input data of load flow program; MVAbase base MVA
casename contains base MVA, bus, generator, bus bus data
branch and hvdc data gen generator data
branch branch data
hvdc hvdc data
flag to indicate if Newton-
success
Raphson was successful
the number of iterations that the
iterations
Newton-Rhapson needed
With the information from the casename file, the load flow is run. This function is the
main function of the program. All the other functions are called in this function. The
output contains the result of the load flow calculation. Bus voltages are found in the bus
matrix, power flows are found in the branch matrix, generated active and reactive powers
by the generators are found in the gen matrix. Generator reactive power limits and
converter voltage limits are checked.
First of all, it is checked whether the hvdc matrix contains data. If this is the case, the
adjusted functions are run. On the other hand, if the hvdc matrix is empty, the original
load flow calculation as programmed in Matpower is run.
13.2.2 The Function HVDC_losses()
hvdc hvdc data rating of VSC HVDC link, according
HVDC_MVA_base
to M-matrix
Ptotal_rect losses in rectifier converter station
Plink losses in DC cables
Ptotal_inv losses in inverter converter station
Ploss_total total losses in VSC HVDC link
Losses are calculated based on ABBs Sdra Lnken project.

13.2.3 The Function HVDC_Limits_Converters()
Psend active power over DC link flag to indicate if rectifiers
fullfilled_rect
reactive power injected by operating point is within limits
Qr
rectifier flag to indicate if inverters
fullfilled_inv
reactive power injected by operating point is within limits
Qi
inverter
Ploss_total total losses in DC link
HVDC_MVA_base rating of VSC HVDC
To check if a converter is operating within its limits, the capabity chart from [5] is used.
13.2.4 The Function HVDC_newtonpf()

Ybus admittance matrix V complex voltages of buses
Sbus complex bus power injections flag to indicate if Newton-Raphson
success
V0 initial voltages was successful
ref index of slack bus the number of iterations that the
iterations
pv indices of pv buses Newton-Rhapson needed
pq indices of pq buses Vvr1 rectifier converter voltage amplitude
mpopt options vector Vvr2 inverter converter voltage amplitude
hvdc hvdc data Tvr1 rectifier converter voltage angle
baseMVA base MVA Tvr2 inverter converter voltage angle
bus_rows number of buses
Ploss_total total loss in HVDC link
The Newton-Raphson method is performed here to obtain the bus voltages and the
voltages of the converters. The input options vector mpopt contains information about
the maximum number of iterations, the convergence tolerance, the amount of progress
to be printed
13.2.5 The Function HVDC_CalculatePowers()
PCAL calculated active powers PCAL updated calculated active powers
QCAL calculated reactive powers QCAL updated calculated reactive powers
Vm bus voltage amplitudes active power between AC bus and
PHVDC
Va bus voltage angles converter bus
HVDCsend HVDC sending bus reactive power between AC bus and
QHVDC
HVDCrec HVDC receiving bus converter bus
Xvr1 rectifiers reactance
Vvr1 rectifiers voltage amplitude
Tvr1 rectifiers voltage angle
Xvr2 inverters reactance
Vvr2 inverters voltage amplitude
Tvr2 inverters voltage angle
hvdc_rows number of VSC HVDC links
The calculated active and reactive powers are updated with the active and reactive
power flows from the VSC HVDC terminals respectively.
13.2.6 The Function HVDC_Mismatches()

vector with mismatches in active updated vector with mismatches in
F F
and reactive power flow equations active and reactive power flow
Qsp1 reactive power injected by rectifier
Qsp2 reactive power injected by inverter
Psp active power over DC link
active power between AC bus and
PHVDC
converter bus
reactive power between AC bus
QHVDC
and converter bus
hvdc_rows number of hvdc links
bus_rows number of buses
baseMVA base MVA
Ploss_total total losses in hvdc link
The matrix with power mismatches is updated with the power flow mismatches between
the converter buses and their associated AC buses.
13.2.7 The Function Convert_Jacobean()
J Jacobean J updated Jacobean
There is a difference in structure between the Jacobean used in Matpower and the
Jacobean used in E. Acha [32]. The Matpower Jacobean is transformed to the
Jacobean from [32]. The difference in Jacobean structure for a three bus network is as
follow:
13.2.8 The Function HVDC_Jacobean()

bus_rows number of buses J extended Jacobean
J Jacobean
Vm bus voltage amplitudes
Va bus voltage angles
hvdc_rows number of HVDC links
HVDCsend HVDC sending bus
HVDCrec HVDC receiving bus
Xvr1 rectifiers reactance
Xvr2 inverters reactance
Rcd resistance of DC cable
flag to indicate rectifiers voltage
VSta1
control mode
flag to indicate inverters voltage
VSta2
control mode
flag to indicate rectifiers
QSta1
reactive power control mode
flag to indicate inverters
QSta2
reactive power control mode
The Jacobean is extended with extra columns and rows to cope with the inclusion of one
or more VSC HVDC links in the network. There are 2 mismatch equations for every
converter station.
There is however again a small difference between the Matpower code and the code in
V
[32]. In the Matpower code mismatch equations, is used instead of simply V .
V
As a consequence, following factors are introduced in the Jacobean elements:
1 1
, and similar factors for the inverter side equations.
Vm(HVDCsend(ii)) Vvr1(ii)
13.2.9 The Function HVDC_Updating()

bus_rows number of buses Vm updated bus voltage amplitudes
mismatches of voltage amplitudes Vvr1 updated rectifiers voltage amplitude
dx
and angles Tvr1 updated rectifiers voltage angle
Vm bus voltage amplitudes Vvr2 updated inverters voltage amplitude
Va bus voltage angles Tvr2 updated inverters voltage angle
hvdc_rows number of HVDC links
HVDCsend HVDC sending bus
HVDCrec HVDC receiving bus
flag to indicate rectifiers reactive
QSta1
power control mode
flag to indicate inverters reactive
QSta2
power control mode
flag to indicate rectifiers voltage
VSta1
control mode
flag to indicate inverters voltage
VSta2
control mode
TarVol1 target voltage of rectifiers AC bus
TarVol2 target voltage of inverters AC bus
The voltages of the rectifier and inverter are updated. If one ore more converters
operate in AC voltage control mode, the associated bus voltage is also updated.
13.2.10 The Function HVDC_Limits()
hvdc_rows number of HVDC links Vvr1 rectifiers voltage amplitude
Vvr1 rectifiers voltage amplitude Vvr2 inverters voltage amplitude
VvrLo lower converter voltage limit
VvrHi upper converter voltage limit
The converter voltage limits are checked and if they are exceeded, the converter voltage
is held at this limit.
13.2.11 The Function HVDC_PQpower()
bus_rows number of buses complex power between rectifier
HVDC_PQsend
V bus voltages converter and associated AC bus
hvdc_rows number of HVDC links complex power between inverter
HVDC_PQrec
hvdc hvdc data converter and associated AC bus
Power flows between the converters and the AC network are calculated.
13.2.12 The Function updategen()

gen generator data gen updated generator data
complex power between rectifier
HVDC_PQsend
converter and associated AC bus
complex power between inverter
HVDC_PQrec
converter and associated AC bus
ref index of slack bus
refbus index of slack bus
PV indices of pv buses
hvdc hvdc data
If the VSC HVDC link is connected to the slack bus, active and reactive powers of the
slack bus are updated with the powers from the VSC HVDC terminals. The same
applies to PV-buses; in that case only the reactive powers are updated.
13.2.13 The Function HVDC_printpf()
baseMVA base MVA - -
bus bus data updated with voltages
generator data updated with active (slack)
gen
and reactive (slack & pv) powers
branch branch data updated with line powers
flag to indicate if Newton-Raphson was
success
successful
et estimated calculation time
mpopt options vector
complex power between rectifier converter
HVDC_PQsend
and associated AC bus
complex power between inverter converter
HVDC_PQrec
and associated AC bus
The output is printed.

13.3 Parallel Operation of an AC Link and a VSC HVDC Link
In the simulations there are 8 unknown parameters: the bus voltages U1 and U2, the
voltage angles t and 1, the active power over the AC line Pac, the active (Pg) and
reactive (Qg) powers injected by the generator, the reactive power over the AC line
leaving bus 1 Q1ac. As a consequence, 8 equations have to be formulated. The same
definitions as in Figure 4-5 are used. The 8 equations to be solved are:
U1.U 2
Pac = sin(1 2 ) 13-1
XL
U t .U1
Pg = sin( t 1 ) 13-2
Xt
U12 U t .U1
Qg = + cos( t 1 ) 13-3
Xt Xt
U12 U1.U 2
Q1ac = cos(1 2 ) 13-4
XL XL
Qg + B.U12 = Q1dc + Q1ac 13-5
Pg = Pdc + Pac 13-6
Pdc + Pac = PL 13-7
U 2 2 U1.U 2
Q2 dc + ( + cos(1 2 )) + B.U 2 2 = 0 13-8
XL XL
Parameters are chosen as follow:

U base = 1 pu , Sbase = 1 pu
X L = 0.5 pu
X t = 0.2 pu
U t = 1 pu
B = 0.2 pu
I conv = 0.08 pu , U conv = 1 pu , Sconv = 0.08 pu
2 = 0 rad (reference).
The Matlab file can be found in [48].
13.4 Derivation of a Voltage Security Indicator
Real and reactive power losses of a two-bus system are given by:
R( PS 2 + QS 2 )
PL = 13-9
US2
X ( PS 2 + QS 2 )
QL = 13-10
US2
where US is the sending end voltage.

Insertion of 12-9 and 12-10 in 4-17 and 4-18 leads to:
( PS PR ) X + RQR
QS = 13-11
R
Inserting 12-11 in 4-17 leads to the following quadratic equation:

( R 2 + X 2 ) PS 2 (2 X 2 PR 2 XRQR + R) PS + ( X 2 PR 2 + R 2QR 2 2 RXPR QR + RPR ) = 0 13-12
Solving this equation for PS and setting the determinant greater than zero leads to 4-19.
Analyzing the reactive powers leads to the same conclusion.
13.5 Reactive Powers Injected by Converters during Generation of PV-Curves
Figure 13-1: Reactive powers injected by converters

13.6 Harmonic Spectra
Figure 13-2: Harmonic content of converter terminal phase-to-ground voltage using

sinusoidal PWM with 3rd harmonic injection. The blue dotted line shows the fundamental
frequency voltage component of the converter terminal to ground voltage [5]
Figure 13-3: Harmonic content of converter terminal phase-to-ground voltage using

harmonic cancellation PWM. The blue dotted line shows the fundamental frequency
voltage component of the converter terminal to ground voltage [5]
13.7 IEEE 14 Bus Network
Figure 13-4: The IEEE 14 Bus Test Case
Bus number Voltage [kV]

1-2-3-4-5 69
6-7-9-10-11-12-13-14 13.8
8 18
Table 13-1: Bus voltages of the IEEE 14 bus system

Bus Bus type P gen [MW] Q gen [MVAr] P load [MW] Q load [MVAr]
1 Slack - - - -
2 PV 40 - 21.70 12.70
3 PV 0 - 94.20 19
4 PQ 47.80 -3.90
5 PQ 7.60 1.60
6 PV 0 - 0 -
7 PQ 11.20 7.50
8 PV 0 - 0 -
9 PQ 29.50 16.60
10 PQ 9 5.80
11 PQ 3.50 1.80
12 PQ 6.10 1.60
13 PQ 13.50 5.80
14 PQ 14.90 5
259 MW 73.50 MVAr
Table 13-2: Generation and load data of the IEEE 14 bus system

VSC HVDC in Meshed Networks (Master Thesis Report)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

VSC HVDC in Meshed Networks (Master Thesis Report)

Uploaded by

Copyright:

Available Formats

Master Thesis Report

Doc. title Revision Page

VSC HVDC in meshed networks 01 1/118

Doc. title Revision Page

VSC HVDC in meshed networks 01 2/118

Doc. title Revision Page

VSC HVDC in meshed networks 01 3/118

10 FUTURE WORK .............................................................................................................................. 96

12 REFERENCES .............................................................................................................................. 100

13 ENCLOSURES .............................................................................................................................. 105

Doc. title Revision Page

VSC HVDC in meshed networks 01 4/118

Doc. title Revision Page

VSC HVDC in meshed networks 01 5/118

This report does not describe the following aspects:

This report has the following structure.

Doc. title Revision Page

VSC HVDC in meshed networks 01 6/118

2 ASPECTS OF VSC HVDC

2.1 The Role of VSC HVDC in the Future Transmission Grid

Doc. title Revision Page

VSC HVDC in meshed networks 01 7/118

2.2 Outlook of VSC HVDC Link

Figure 2-1: A VSC HVDC transmission scheme

Doc. title Revision Page

VSC HVDC in meshed networks 01 8/118

2.3 Converter Topology

The operational behavior of a converter is similar to a generator; the VSC delivers a

Doc. title Revision Page

VSC HVDC in meshed networks 01 9/118

Another configuration used in VSC HVDC transmission is the three-phase three-level

Figure 2-4: Three-phase three-level NPC converter and associated AC voltage

Doc. title Revision Page

VSC HVDC in meshed networks 01 10/118

2.4 Active and Reactive Power Flow Control

Figure 2-6: Circuit diagram of a VSC HVDC converter station

Doc. title Revision Page

VSC HVDC in meshed networks 01 11/118

with UC and Ud defined as in Figure 2-6.

Doc. title Revision Page

VSC HVDC in meshed networks 01 12/118

2.5 Capability Chart

Figure 2-7: Capability chart of VSC HVDC

Doc. title Revision Page

VSC HVDC in meshed networks 01 13/118

2.6 Four Quadrant Operation

2.7 LCC HVDC versus VSC HVDC

Doc. title Revision Page

VSC HVDC in meshed networks 01 14/118

Figure 2-9: LCC HVDC (left) versus VSC HVDC (right)

Doc. title Revision Page

VSC HVDC in meshed networks 01 15/118

Table 2-1: Comparison between different transmission topologies

Doc. title Revision Page

VSC HVDC in meshed networks 01 16/118

2.8 Applications of VSC HVDC

Situations where HVDC is preferred over AC are the following [20]:

Doc. title Revision Page

VSC HVDC in meshed networks 01 17/118

Doc. title Revision Page

VSC HVDC in meshed networks 01 18/118

3 VSC HVDC LOAD FLOW MODEL

3.1 Load Flow Calculation [31]